KINETIC THEORY 
 
 OF 
 
 ENGINEERING STRUCTURES 
 
 DEALING WITH 
 
 STRESSES, DEFORMATIONS AND WORK 
 
 FOR THE USE OF 
 
 STUDENTS AND PRACTITIONERS IN CIVIL ENGINEERING 
 
 BY 
 
 DAVID A. MOLITOR, B.C.E., C.E. 
 
 < 
 
 AUTHOR "HYDRAULICS OF RIVERS, WEIRS AND SLUICES," ETC.; MEM. AM. soc. c. E.; MEM. DETROIT 
 
 ENGINEERING SOC.; MEM. SOC. FOR THE PROM. ENG. EDU.J MEM. A. A. A. S., ETC.; 
 
 FORMERLY DESIGNING ENGINEER ISTHMIAN CANAL COMMISSION; 
 
 FROFESSOr, IN CIVIL ENGINEERING, CORNELL UNIVERSITY. 
 
 McGRAW-HILL BOOK COMPANY 
 
 239 WEST 39TH STREET, NEW YORK 
 
 6 BOUVERIE STREET, LONDON, E.G. 
 
 1911 
 
Copyright, 1910, 
 
 BY 
 DAVID A. MOLITOR 
 
 THE SCIENTIFIC PHES9 
 ROBERT ORUMMONO AND COMP 
 BROOKLYN. N. V. 
 
PREFACE 
 
 " ALL that mankind has done, thought, gained, or been; it is lying in magic preserva- 
 tion in the pages of books"; and so in presenting this work for publication, the author 
 hopes to preserve the results of many years of painstaking labors, study and exped- 
 ience, to civil engineering, his chosen profession. 
 
 The field of usefulness which it is proposed to supply is that of an advanced treatise 
 on stresses and deformations in engineering structures, for the use of students and prac- 
 titioners in civil engineering and particularly for specialists engaged in the design of 
 so-called " higher structures " requiring more than the ordinary methods of statics for 
 their analysis. 
 
 The reader is supposed to possess a thorough knowledge of higher mathematics, 
 including the calculus; also, of the elements of statics, embracing the composition 
 and resolution of forces, force and equilibrium polygons, etc. All these are usually 
 given in the two first years of a Civil Engineering course in connection with mathema- 
 tics and mechanics. 
 
 Details pertaining to the design of bridge members and connections are not dealt 
 with. 
 
 The present volume might advantageously be employed as a text-book in bridge 
 stresses with the aim of giving a more thorough training in the analysis of stresses and 
 deformations by economizing somewhat on the time at present devoted to detailing 
 and shop practice, which is not really justified in a four-year course of theoretical 
 preparation. 
 
 In the opinion of the author this would be highly desirable, because details and 
 shop practice can best be learned in the shop, where experience is the teacher. It 
 devolves on the college to give to its students that thorough theoretical training which 
 they cannot readily acquire in practice, a feature which must ultimately distinguish 
 the college-bred engineer from the purely practical man. 
 
 The present work, which is the gradual evolution of years of labor, thus represents 
 an effort to place before the profession a treatise on the analysis of engineering struc- 
 tures which is based on the most advanced theories and researches of the present time. 
 
 To enhance the educational value of the book, each subject is prefaced by a few 
 brief historical remarks and all important theorems bear the names of their originators, 
 a practice which has been shamefully neglected by many modern writers. 
 
 The author has made free use of any and all literature bearing on the various sub- 
 jects treated and hereby expresses his grateful acknowledgment to all who have contrib- 
 uted to the sum total of our present knowledge. Prominent among these should be 
 
vi PREFACE 
 
 mentioned Hooke, Bernoulli, Lagrange, Coulomb, Navier, Lame, de Saint- Venant, 
 Clapeyron, and Menabrea, as the founders of the theories of elasticity and virtual work. 
 
 To Clerk Maxwell, Castigliano, Mohr, Fraenkel, Culmann, Winkler, Mueller-Breslau, 
 Engesser, Weyrauch, Manderla, Asimont, Landsberg, Melan, Zimmerman, Ritter, Land, 
 and Mehrtens, we are indebted for advancing these theories to their present state of per- 
 fection and usefulness. Professor Otto Mohr enjoys the high distinction of being the 
 foremost originator of novel principles in the analysis of engineering structures. 
 
 A bibliography of the works used by the author will be found at the end of this 
 volume. 
 
 The author has avoided a sharp differentiation between analytic and graphic methods 
 and uses both without discrimination, aiming always to secure the most practical solu- 
 tion for any problem in hand. 
 
 While this work was produced with very great care and diligence, hoping to eliminate 
 typographical and other errors, it is certain that some have escaped detection. The 
 author, therefore, invites the kind indulgence of the reader and will be truly thankful 
 for having his or the publishers' attention called to any errors that may be discovered. 
 
 Finally, as the love liberated in our work is a true manifestation of character, so 
 may the present production reflect the character of, the man, the engineer, the author. 
 
 ITHACA, N. Y., September 9, 1910. 
 
DEFINITIONS OF TERMS USED THROUGHOUT THIS WORK 
 
 FREQUENTLY some of the letters are employed in a special connection other than here 
 defined, but in such cases the definitions locally given will prevent misunderstandings. 
 
 ART. 3. p = number of pin points of any frame. 
 m = number of members of any frame. 
 n = number of redundant conditions in any structure. 
 n' = number of external redundants in any structure. 
 n" = number of internal redundants in any structure. 
 
 e = number of elements or simple frames in a structure. 
 Sr = number of necessary reaction conditions for any structure. 
 2 = the summation sign. 
 ART. 4. S = total stress in any member of a frame resulting from any loads P or other 
 
 causes and + for tension. 
 
 & = the stress produced in any member by applied loads P. 
 /= the unit stress in any member, + for tension. 
 
 Z = length of any member or span of a structure for the condition of no stress. 
 JZ = change in length of a member due to any cause, positive for elongation. 
 F cross-section of any member, prismatic in form. 
 t=& uniform change in temperature in degrees, + for rise. 
 = coefficient of expansion per degree temperature. 
 E= Young's modulus of elasticity for direct stress. 
 a = the angle which a member makes with the x axis of coordinates. 
 /? = the angle which a member makes with the y axis of coordinates. 
 m = any point of a structure chosen for illustration. 
 p = l/EF = the extensibility of a member employed for brevity. 
 .R = any reaction force, or condition, of a structure resulting from loads P. 
 .ft = any reaction force, or condition, of a structure resulting from loads P. 
 Jr = any displacement of the point of application of a force R in the direction 
 
 of this force. 
 
 P = any external load or force applied to a structure. 
 P = a load or force identical in position and point of application with the force 
 
 P but having a different intensity. 
 ART. 5. J = any displacement of a pin point. 
 
 <? = the displacement of the point of application of any force P in the direction 
 of this force. 
 
viii DEFINITIONS OF TERMS USED THROUGHOUT THIS WORK 
 
 A = actual work = force times distance through which the force moves. 
 A e = ihe positive or externally applied work of deformation of any structure. 
 .A t - = the negative or internally overcome work of deformation of any structure. 
 W = a virtual work of deformation. 
 
 W a virtual work of deformation produced by a conventional case of loading. 
 ART. 7. X a , Xb, X c , etc., are the stresses in redundant members or reactions A, B, C, etc. 
 <, = the stress in any member of a principal system, due entirely to loads P 
 when all redundant conditions X are removed. This is known as con- 
 dition X = 0. 
 
 S a = ihe stress in any member of a principal system, due entirely to the conven- 
 tional loading X a =l, applied to the principal system. Condition X n =l. 
 Sb. S c , etc., are similarly defined for conventional loadings Xb=l; X c =l, etc. 
 St = the stress in any member of a principal system produced by a uniform change 
 
 in temperature. 
 
 R t = a reaction produced by a uniform change in temperature. 
 M t = & moment produced by a uniform change in temperature. 
 R , R a , R b , etc., have definitions analogous to those given for S , S a , Sb, etc. 
 M 0f M a , M b , etc., are moments similarly defined for the conventional loadii^s. 
 da, $b, $c are changes in the lengths of redundant members or supports X a , X bf 
 
 X e , respectively. 
 
 L represents the length of span of a composite structure. 
 
 ART. 8. dma = the displacement of the point of application m of any load P m , in the 
 direction of this load, when the principal system is loaded only with 
 X a = 1 . Condition X a = 1 . 
 Smb = a similar displacement of the point m for the conventional loading X b = 1 . 
 
 Condition Xb=l. 
 
 <5 TOC = the same for condition X c = 1. 
 
 daa = the change in length of the member a for condition X a l. 
 (5 j> = the change in length of the member a for condition Xb=l. 
 d&a = the change in length of the member b for condition X a =l, etc. 
 d at = the change in length of the member a resulting from a change in temperature 
 of t, when the principal system is not otherwise loaded, hence X = 0, P = 0. 
 dbt, $ct are similarly defined for members b and c respectively, 
 ART. 13. f x , f y , / 2 = the unit normal stresses in the directions X, Y, and Z, respectively , 
 
 for any isotropic body. 
 
 T XV and r x:s = the unit tangential stresses in the YZ plane. 
 r vx and r vz = the unit tangential stresses in the XZ plane. 
 T ZX and T zv = the unit tangential stresses in the XY plane. 
 G = the modulus of tangential stress. 
 
 m = the Poisson number or ratio of lateral to longitudinal deformation. 
 ART. 14. N = & normal thrust on any section. 
 R = any oblique thrust. 
 Q = a tangential force or shear. 
 I = the moment of inertia of anv cross-section. 
 
DEFINITIONS OF TERMS USED THROUGHOUT THIS WORK ix 
 
 t/ = an ordinate usually the distance from the neutral axis to the extreme fiber 
 
 of a cross-section. 
 h = height of a section. 
 6 = base or breadth of a section. 
 
 /? = coefficient of shearing strain, Eq. (ML) and Eq. (14M). 
 ART. 16. w = weight of a moving body. 
 
 H = height of fall. Later used to represent the horizontal thrust of an arch, 
 
 also, the pole distance of a force polygon. 
 v = velocity at instant of impact. 
 g = acceleration due to gravity. 
 ART. 17. jj = any ordinate to an influence line. 
 d = a panel length. 
 
 p = uniform moving load per unit of length. 
 ART. 18. m = a load point, being any one of the many possible points of application of 
 
 a certain moving load. 
 n = the point for which art influence line is drawn. Also, the location of a section 
 
 under consideration. 
 ART. 20. i = the load divide for positive and negative effects. For arches the load 
 
 divide is sometimes designated by d. 
 ART. 21. r = the lever arm of a member in the equation S = M/r. 
 
 S a = ihe stress in any member S due to a reaction unity at A, when the load 
 
 producing this reaction is to the right of the section. 
 Si, = the stress in the same member due to a reaction unity at B, when the load 
 
 producing this reaction is to the left of the section. 
 
 ART. 28. e = the kernel point for the extreme fiber of the extrados of a solid web arch section. 
 
 i = the kernel point for the extreme fiber of the intrados of the same arch section. 
 
 ART. 35. w = an elastic load, representing an abstract number without any particular unit 
 
 of measure. 
 
 ART. 45. <S = the stress in any member of a statically indeterminate frame for the con- 
 dition J = and P=l. 
 M = a moment for the same conditions. 
 A = the reaction at A for the same conditions. 
 
 /x = an influence line factor by. which all ordinates must be multiplied to obtain 
 the actual influences. 
 
CONTENTS 
 
 PAGE 
 
 PREFACE v 
 
 DEFINITIONS OF TERMS vii 
 
 INTRODUCTION 1 
 
 CHAPTER I 
 REACTIONS AND REDUNDANT CONDITIONS 
 
 1. Reaction Conditions 5 
 
 2. Structural Redundancy 6 
 
 3. Tests for Structural Redundancy 7 
 
 CHAPTER II 
 THEOREMS, LAWS, AND FORMULAE FOR FRAMED STRUCTURES 
 
 4. Elastic Deformations, Fundamental Equations 11 
 
 5. General Work Equations for any Frame 14 
 
 6. Displacements of Points. Statically Determinate Structures 18 
 
 7. Indeterminate Structures, by Mohr's Work Equation 19 
 
 8. Indeterminate Structures, by Maxwell's Law 24 
 
 9. Prof. Maxwell s Theorem (1864) 27 
 
 10. Theorems Relating to Work of Deformation 29 
 
 (a) Menabrea's law (1858). Theorem of least work 29 
 
 (b) Castigliano's law (1879). Derivative of the work equation 30 
 
 11. Temperature Stresses follow Menabrea's and Castigliano's Laws 31 
 
 12. Stresses Due to Abutment Displacements 32 
 
 CHAPTER III 
 
 THEOREMS, LAWS, AND FORMULAE FOR ISOTROPIC BODIES 
 
 1 3- General Work Equations 34 
 
 14. Work of Deformation due to Shearing Stress 38 
 
 15. Work of Deformation for any Indeterminate Straight Beam 41 
 
 16. Work of Deformation due to Dynamic Impact 43 
 
 xi 
 
xii CONTENTS 
 
 CHAPTER IV 
 INFLUENCE LINES AND AREAS FOR STATICALLY DETERMINATE STRUCTURES 
 
 PAGE 
 
 1 7. Introductory 46 
 
 18. Influence Lines for Direct Loading 4<S 
 
 19. Influence Lines for Indirect Loading 50 
 
 20. The Load Divide for a Truss 51 
 
 21. Stress Influence Lines for Truss Members 52 
 
 22. Reaction Summation Influence Lines 54 
 
 23. Positions of a Moving Train for Maximum and Minimum Moments 57 
 
 24. Positions of a Moving Train for Maximum and Minimum Shears 62 
 
 CHAPTER V 
 SPECIAL APPLICATIONS OF INFLUENCE LINES TO STATICALLY DETERMINATE STRUCTURES 
 
 25. Double Intersection Trusses 65 
 
 26. Cantilever Bridges 67 
 
 27. Three-hinged Framed Arches 72 
 
 28. Three-hinged, Solid Web and Masonry Arches 78 
 
 29. Skew Plate Girder for Curved Double Track 80 
 
 30. Trusses with Sub-divided Panels S3 
 
 CHAPTER VI 
 DISTORTION OF A STATICALLY DETERMINATE FRAME BY GRAPHICS 
 
 31 . Introductory 87 
 
 32. Distortions due to Changes in the Lengths of the Members 87 
 
 33. Rotation of a Rigid Frame about a Fixed Point 89 
 
 34. Special Applications of Williot-Mohr Diagrams 91 
 
 CHAPTER VII 
 DEFLECTION POLYGONS OF STATICALLY DETERMINATE STRUCTURES BY ANALYTICS AND GRAPHICS 
 
 35. Introductory 9!) 
 
 36. First Method, according to Prof. Mohr 100 
 
 (a) Influence on the deflections due to chord members 100 
 
 (6) Influence on the deflections due to web members . 102 
 
 (c) The deflection polygon for the loaded chord 104 
 
 37. Second Method, according to Prof. Land , 107 
 
 (a) The w loads in terms of the unit stresses in the members . . 107 
 
 (6) To find the changes in the angles of a triangle 1 07 
 
 (c) To evaluate the elastic loads w in terms of the angle changes 109 
 
 38. Horizontal Displacements 114 
 
 39. Deflections of Solid Web Beams .. . 117 
 
CONTENTS xiii 
 
 CHAPTER VIII 
 DEFLECTION INFLUENCE LINES FOR STATICALLY DETERMINATE STRUCTURES 
 
 PAGE 
 
 40. Influence Lines for Elastic Displacements 122 
 
 41. Special Applications of Displacement Influence Lines 124 
 
 (a) Cantilever beam, deflection polygon 124 
 
 (b) Simple truss, displacement of any pin point in any direction 125 
 
 (c) Three-hinged arch, rotation between two lines 125 
 
 CHAPTER IX 
 INFLUENCE LINES FOR STATICALLY INDETERMINATE STRUCTURES 
 
 42. Influence Lines for One- Redundant Condition 127 
 
 (a) When the redundant condition is internal 128 
 
 (b) When the redundant condition is external 128 
 
 43. Influence Lines for Two Redundant Conditions 130 
 
 44. Simplification of Influence Lines for Multiple Redundancy 132 
 
 (a) Structures having two redundant conditions 133 
 
 (b) Structures having three redundant conditions 135 
 
 45. Stress Influence Lines for Structures Involving Redundancy 137 
 
 (a) Multiple redundancy 137 
 
 (6) One redundant condition 139 
 
 CHAPTER X 
 SPECIAL APPLICATIONS OF INFLUENCE LINES TO STATICALLY INDETERMINATE STRUCTURES 
 
 46. Simple Beam with One End Fixed and Other End Supported 140 
 
 47. Plate Girder on Three Supports 143 
 
 48. Truss on Three Supports 148 
 
 49. Two-hinged Solid Web Arch or Arched Rib 153 
 
 50. Two-hinged Spandrel Braced Arch 166 
 
 51. Two-hinged Arch with Cantilever Side Spans 176 
 
 52. Fixed Framed Arches 178 
 
 CHAFFER XI 
 DESIGN OF STATICALLY INDETERMINATE STRUCTURES 
 
 53. Methods for Preliminary Designing 194 
 
 54. On the Choice of the Redundant Conditions 202 
 
 55. Solution of the General Case of Redundancy 203 
 
 56. Effect of Temperature on Indeterminate Structures 206 
 
 57. Effect of Shop Lengths and Abutment Displacements 208 
 
xiv CONTENTS 
 
 CHAPTER XII 
 
 STRESSES IN STATICALLY DETERMINATE STRUCTURES 
 
 PAGE 
 
 58. Dead Load Stresses 210 
 
 (a) General considerations 210 
 
 (6) Aug. Hitter's methods of moments (1860) 211 
 
 (c) The method of stress diagrams 213 
 
 59. Live Load Stresses 218 
 
 (a) The critical positions of a train of moving loads 218 
 
 (b) The method of influence lines 218 
 
 (c) Discussion of methods in common use 219 
 
 (d) The author's method 210 
 
 7 CHAPTER XIII 
 
 SECONDARY STRESSES 
 
 60. The Nature of Secondary Stresses -.-. . . 226 
 
 61. Secondary Stresses in the Plane of a Truss due to Riveted Connections 227 
 
 62. Secondary Stresses in Riveted Cross Frames of Trusses 244 
 
 (a) Dead and live load effects 244 
 
 (6) Wind effects 247 
 
 63. Secondary Stresses due to Various Causes 251 
 
 (a) Bending stresses in the members due to their own weight 251 
 
 (6) Secondary stresses in pin-connected structures 253 
 
 (c) The effect of eccentric connections 255 
 
 (d) Effect of temperature and erroneous shop lengths 256 
 
 64. Additional Stresses due to Dynamic Influences 256 
 
 (a) Forces acting longitudinally on a structure 256 
 
 (6) Forces acting transversely to a structure 258 
 
 (c) Forces acting vertically on a structure 262 
 
 (d) Fatigue of the material 265 
 
 (e) Unusual load effects 266 
 
 65. Concluding Remarks 207 
 
 (a) Features in design intended to diminish secondary and additional stresses 267 
 
 (b) Final combination of stresses as a basis for designing 268 
 
 CHAPTER XIV 
 MITERING LOCK GATES 
 
 66. The Nature of the Problem 271 
 
 67. The Theory of the Analysis 273 
 
 68. Example : Upper Gate, Erie Canal Lock No. 22 279 
 
CONTENTS xv 
 
 CHAPTER XV 
 
 FIXED MASONRY ARCHES 
 
 PAGE 
 
 69. General Considerations 298 
 
 70. Modern Methods of Construction 301 
 
 71. Preliminary Designs 304 
 
 72. Determination of the Redundant Conditions by the Theory of Elasticity 309 
 
 73. Co-ordinate Axes and Elastic Loads 314 
 
 74. Influence Lines for X a , Xb, X c , and Mm 317 
 
 75. Temperature Stresses 318 
 
 76. Stresses on any Normal Arch Section 320 
 
 77. Maximum Stresses and Critical Sections 322 
 
 78. Resultant Polygons 324 
 
 79. Example : 150-foot Concrete Arch 326 
 
 (a) Given data arid preliminary design 326 
 
 (6) Analysis of the preliminary design by the theory of elasticity, treating the structure 
 
 as symmetric 331 
 
 (c) Analysis as under (fr), treating the structure as unsymmetric 346 
 
 BIBLIOGRAPHY 
 
 Theory of Elasticity 359 
 
 Graphostatics 359 
 
 General Treatises .' , 360 
 
 Secondary and Additional Stresses 360 
 
 Special Treatises on Arches 361 
 
KINETIC THEORY 
 
 OF 
 
 ENGINEERING STRUCTURES 
 
 INTRODUCTION 
 
 The work of deformation constitutes the basis for the solution of all those problems 
 which involve the elastic properties of the material and which are not susceptible to 
 analysis by the methods of pure statics. 
 
 Work is the product of a constant force into the displacement of its point of applica- 
 tion in the direction of the force. 
 
 When a prismatic body of length I and cross-section F is either lengthened or 
 shortened a small amount M within the elastic limit, by a force S gradually produced 
 without impact, then a certain amount of work or kinetic energy is stored up in the 
 body. In the instant when the force ceases to alter the length a condition of static 
 equilibrium is established and the total kinetic energy stored then assumes the form 
 of potential energy or energy of position. W T hen the force is gradually released and 
 the body is allowed to resume its original length, the potential energy is liberated. 
 
 The work of deformation or applied kinetic energy may be thus expressed: 
 
 The principle of virtual work. When a rigid body is set in motion by any group 
 of forces, then the kinetic energy imparted is equal to the algebraic sum of all the work 
 performed by the several applied forces. 
 
 A real displacement can never result in zero work,' because motion cannot exist 
 without the presence of a force, and the combination of force and motion constitutes 
 work. 
 
 However, for a state of equilibrium wherein the applied forces do not actually alter 
 the motion, a certain equilibrium proposition may be formulated by considering possible 
 displacements that might be produced but may or may not be real. Such displacements 
 are called virtual, and any work resulting therefrom is called virtual work. 
 
2 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 The fundamental principle of statics may, therefore, be stated thus: A body at 
 rest remains at rest when for any infinitesimally small but possible displacement the 
 summation of the work of the applied forces is zero. 
 
 Letting P represent any force and d any possible virtual displacement of its point 
 of application in the direction of the force, then, according to Lagrange (1788), the law 
 of virtual work is represented by the following equation: 
 
 ............. (2) 
 
 The law of virtual velocities follows from the law of virtual work by introducing the 
 element of time, giving 
 
 ........ ..... (3) 
 
 The relations thus existing between the possible conditions of equilibrium of a rigid 
 body and the virtual displacements resulting from any cause, lead to important geometric 
 relations which are generally applicable to all branches of technical mechanics. 
 
 Clapeyron's law (1833), which covers the entire work of the external forces and 
 internal stresses acting on any frame or body in equilibrium, is derived, from Eq. (1), 
 by summing all the partial effects for an entire structure. Then, by the " doctrine of the 
 conservation of energy," the applied work A e must equal A; the work overcome, giving 
 
 ......... (4) 
 
 wherein the loads P, stresses S, displacements d projected onto the directions of the 
 loads, and the changes Al in the lengths of the members, all result from a simultaneous 
 condition of loading for a case in equilibrium. 
 
 Proceeding from Clapeyron's law. Professor Clerk Maxwell, in 1864, derives his 
 important law of reciprocal displacements and Professor Otto Mohr, in 1874, evolved his 
 two work equations which together with Maxwell's law, afford a complete solution for 
 all problems relating to frames and isotropic bodies. 
 
 Chapter I deals with the general relations existing between a frame or beam, arid 
 its supports, furnishing the definitions and necessary criteria for distinguishing between 
 statically determinate and statically indeterminate structures or those involving redundant 
 members or redundant reaction conditions. 
 
 Chapter II then presents the derivation and discussion of all the general laws per- 
 taining to framed structures and these are applied to isotropic bodies in Chapter III. 
 
 Chapters IV, V, and VIII are devoted to influence lines for determinate structures, 
 while Chapters VI and VII treat of the various problems relating to distortions and 
 deflection polygons of frames and solid web structures. 
 
 Chapters IX and X apply the various theories and methods to the analysis of 
 statically indeterminate structures chiefly by the use of influence lines. 
 
 In Chapter XI, the general question of designing a statically indeterminate structure 
 is then taken up and Chapter XII deals with stresses in statically determinate structures, 
 
INTRODUCTION 3 
 
 taking advantage of some of the more advanced ideas presented in the earlier chapters. 
 Special attention is directed to Art. 59, dealing with stresses due to concentrated live 
 loads. 
 
 Chapter XIII gives a brief treatment of secondary stresses and the two last chapters 
 are devoted to the problems of mitering lock gates and fixed masonry arches. 
 
 It is more than likely that the use of the word kinetic in the title of the present 
 volume will be questioned by some. However, after long and careful consideration 
 the author concluded that he was justified in employing the term without really depart- 
 ing from the best modern dictionary definitions. It is true that in hydraulics the term 
 kinetic energy is exclusively confined to represent the static equivalent h=v 2 /2g of a 
 dynamic head. In mechanics generally, the expression is almost limited to the term 
 Wv 2 /2g=mv 2 /2, representing the potential or kinetic energy of amoving mass. Other- 
 wise the term is not met with. 
 
 While the author employs the term kinetic in connection with engineering structures 
 which are, generally considered, at rest, it is precisely with the same idea above expressed 
 in the mechanical sense. That is, a bridge or other structure composed of elastic mate- 
 rial ceases to be at rest the moment it is acted upon by external forces. The distinction 
 between the previous and present uses of the term is, therefore, only relative, being 
 here applied in a different field and to a special class of bodies, where the displacements 
 are small and follow certain empiric laws depending on the elastic properties of the 
 material. Strictly speaking, then, we have no license to regard an elastic structure as 
 statically at rest only when it is neither stressed nor distorted. 
 
 The equation of work is thus applied to a structure in the sense that the latter is 
 a mechanical contrivance or machine in motion, instead of an inelastic body at rest. 
 It is true that this motion prevails only while a change in the static balance is taking 
 place, as when loads are added to, or removed from, the structure, but this may apply to 
 any mechanism of interrupted activity. 
 
 Therefore, while we speak of a structure as in static equilibrium, 'we may also speak 
 of it as in a state of dynamic equilibrium, a state which the structure assumes in the instant 
 that the superimposed loads do not produce any further elastic deformations. The 
 same would be true when all the loads are entirely removed, in which case the dynamic 
 equilibrium returns to the special state of static equilibrium. This is merely a broader 
 viewpoint of the subject, embracing at once all of the conditions as they really exist 
 in a frame sustaining loads. The term kinetic thus appears quite appropriate to the 
 use and application here made. 
 
 For the same structure the magnitude of the deformation of the deflection is a 
 direct function of the loads, the unit stresses which they produce, and the elasticity of 
 the material. 
 
 On the other hand, the stiffness of a given structure is independent of the magnitudes 
 of the unit stresses, but depends entirely on the ability of the structure to resist stress 
 and this in turn is a function of the elasticity of the material and the geometric shape 
 of the structure. 
 
 Theoretically speaking, and disregarding economy, a structure of maximum stiff- 
 ness is one of infinite height compared to its span, resulting in zero stress; and con- 
 
4 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 versely, a structure of minimum stiffness is one of zero height involving infinite 
 stress. 
 
 Maximum stress never occurs simultaneously in the chords and web of any structure. 
 The maximum total load produces maximum chord stresses, but only a small fraction 
 of the maximum web stress. Hence, the influence of the web on the total deflection 
 of a bridge is always small as compared with that produced by the chords and is fre- 
 quently negligible. 
 
 Throughout the present treatise, exepting Chapter XIII, all members of a frame 
 are supposedly connected at their ends by frictionless pins, and no account is taken of 
 the so-called secondary stresses occasioned in the members by friction on these pins or 
 by the rigidity of riveted connections. Chapter XIII deals separately with the secondary 
 stresses induced by riveted connections and friction on pins, etc. 
 
 Unless otherwise specifically stated, a clockwise moment of forces on the left of a 
 section is regarded as positive, so also an upward shear on the left side of a section. 
 Tensile stress and elongation are positive, while compressive stress and contraction are 
 negative. Work is positive when the direction of a force coincides with the direction 
 of the displacement of its point of application, independently of the algebraic signs of 
 the forces or stresses and the displacements. 
 
 The nomenclature is uniform throughout, though in a few instances slight departures 
 from American custom became necessary owing to conflicts encountered in combining 
 so many subjects in one book. The definitions of terms are always given in appropriate 
 places and also in summary form following the table of contents. 
 
 Equations and figures are designated by letters and numbers facilitating the matter 
 of locating cross-references. The number refers to the article and the letter to the 
 position of the equation, figure or table in that article. The article number appears in 
 the heading of every page. 
 
ART. 1. REACTION CONDITIONS 
 
 The general purpose of engineering structures, of the class here considered, is to 
 carry loads over otherwise impassable distances to certain fixed points. 
 
 The distances are called the spans, and the fixed points, to which the loads are 
 finally transmitted, are called the supports or reactions. 
 
 It is thus clear that the sum total of the superimposed loads, together with the 
 weight of the structure, must exactly equal the sum of the vertical reaction forces when 
 equilibrium exists. 
 
 The nature and purpose of a structure determine the manner in which it must be 
 supported to insure stability under all circumstances that are likely to occur. 
 
 This can always be accomplished by one of three types of supports known, respec- 
 tively, as movable, hinged, and fixed. They are illustrated in Figs. IA, IB, and Ic. 
 
 FIG. IA. 
 
 FIG. IB. 
 
 FIG. Ic. 
 
 Fig. IA shows the type of support known as movable, which can exert a single 
 upward reaction only. Any lateral force applied to such a support would result in 
 horizontal displacement provided the roller is frictionless. 
 
 Fig. IB shows a hinged support and can resist horizontal and vertical forces or 
 any resultant of these. 
 
 Fig. Ic represents a fixed support which may be made to resist any direct forces 
 and a bending moment. 
 
 Statically considered, the first type satisfies one stability condition, the second 
 type offers two such conditions, and the third type supplies three stability conditions. 
 These conditions may be represented by forces as shown in Figs. ID, IE, and IF. 
 
 Let r represent a single vertical or horizontal force to be known as a reaction con- 
 dition. Then the three types of supports will involve, respectively, one, two, and three 
 such conditions. Each reaction condition may be represented by a short link, or member, 
 which may transmit direct stress only; that is pure tension or compression. 
 
 5 
 
KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. I 
 
 Hence, for a single truss on two supports, with one end movable and the other 
 end hinged, the number of reaction conditions is Sr=3. Also, for a two-hinged arch, 
 with both ends on hinged supports, 2r=4; and for an arch without hinged ends, com- 
 monly called a fixed arch, 2r=6. 
 
 FIG. ID. 
 
 FIG. IE. 
 
 r 
 FIG. IF. 
 
 ART. 2. STRUCTURAL REDUNDANCY 
 
 Any structure, statically considered, may be classed as determinate or indeterminate, 
 depending on the manner in which it is supported, and the arrangement of the members 
 employed to form the structure. 
 
 A structure is externally indeterminate when it involves more than three reaction 
 conditions for a single truss element or frame. This implies that the reactions are 
 counted with the external forces. 
 
 A structure is internally indeterminate when it includes more members within its 
 frame than are required for internal stability according to the conditions for static equilib- 
 rium, regardless of the reactions. 
 
 Hence external and internal redundancy may exist simultaneously. The terms 
 redundant and indeterminate, as here used, are synonymous with impossible, only in so 
 far as the laws of pure statics are concerned. However, the modifying term static may 
 sometimes be omitted for the sake of brevity. Thus indeterminate will always mean 
 statically indeterminate. 
 
 The conditions for static equilibrium applied to any structure as a whole are repre- 
 sented by the following equations : 
 
 27=0, and 2M=0, ...... (2A) 
 
 wherein 2# is the sum of all the horizontal components of the external forces, includ- 
 ing the reactions; 2F is the sum of all the vertical components of these forces and reac- 
 tions; and SM is the sum of the moments of these forces and reactions about any point 
 in the plane of the structure. The sums are always algebraic sums. 
 
 These condition equations apply equally to all forces (externally applied and internal 
 stresses) active about any pin point of any structure. 
 
 When the external forces are not thus balanced among themselves and with the 
 internal stresses, then a structure is said to be statically under determinate or to be unstable. 
 
 A structure is, therefore, externally determinate when the three condition equations 
 for static equilibrium suffice to determine the reactions. 
 
ART. 3 REACTIONS AND REDUNDANT CONDITIONS 7 
 
 A structure is internally determinate when its members are arranged to form triangles 
 in such manner that the successive removal of pairs of members uniting in a point, 
 finally reduces the structure to a single triangle. The triangle is thus seen to be the 
 primary element of all determinate systems. 
 
 A structure is in every sense determinate when all reactions and stresses can be expressed 
 purely as functions of the externally applied loads, unaffected by all temperature changes 
 or temperature inequalities, and small reaction displacements. 
 
 ART. 3. TESTS FOR STRUCTURAL REDUNDANCY 
 
 Let p= number of pin points of any given frame. 
 m= number of members in the frame. 
 n= total number of redundant conditions. 
 ri = number of external redundant conditions. 
 n" = number of internal redundant conditions. 
 
 e= number of elements or simple frames in a structure. 
 2r= number of necessary reaction conditions. 
 Then for any statically determinate structure 
 
 which condition must be satisfied in any event, and when satisfied, the structure is 
 always statically determinate but not always stable. This exception will be discussed 
 below. 
 
 When 2p>m + Zr, ........... (SB) 
 
 then the structure is not in stable equilibrium and may be called statically insufficient. 
 When 2p<w + Sr then the structure involves redundant conditions or members. 
 the total number of which is given by the equation, 
 
 n=m + Zr 2p, ........... (3c) 
 
 wherein a negative n would indicate static insufficiency. 
 
 The number of necessary reaction conditions r for any structure depends on the 
 number of frame elements or simple frames of which the structure may be composed. 
 Hence in order to distinguish whether the n redundant conditions are external, internal, 
 or both, it will be necessary to determine one or the other of these by a separate 
 criterion. This criterion is readily established for the reactions. Thus in Art. 1, Sr=3 
 was found to express the number of required reaction conditions for a simple truss on 
 two supports. 
 
 A composite structure, which may be composed of any number e of simple truss 
 frames, each one of which is directly supported by piers at one or two points, will always 
 have more than three reaction conditions. The requisite number of reaction conditions 
 for any statically determinate structure is 
 
 r=3+e-l=e+2. . . (3D) 
 
8 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. I 
 
 The number of external redundant conditions n', for any given structure of e elements, 
 is thus: 
 
 w' = 2r-e 2, (3E) 
 
 and hence the number of internal redundant conditions must be 
 
 n =nn. 
 
 OF) 
 
 The number of elements entering into the makeup of any structure is thus an important 
 consideration in establishing whether the redundant conditions, indicated by Eq. (3c), 
 are external or internal. In general, a truss element may be denned as a frame or girder 
 which in itself constitutes a simple truss or beam. The illustrations given below will 
 serve to amplify this definition. 
 
 The exception to Eq. (3A), when a structure is apparently determinate but unstable, 
 will now be discussed. 
 
 FIG. SA. 
 
 FIG. 3s. 
 
 The three-hinged arch, in Fig. SA, is in every sense a determinate, stable structure 
 for which the horizontal thrust H =Pl/2f. All the reactions A, B, and H are finite 
 quantities and the stresses in the several members are also finite. This will always 
 remain true except when the middle ordinate /=0, whence // = sc . 
 
 This special case is illustrated in Fig. SB, where ACB is a straight line and the load 
 P cannot produce stress without causing a slight rotation of the two elements I and II 
 about the points A and B. Hence equilibrium does not exist until rotation through 
 a small arc ds has taken place and the stress along AC and CB is then infinite. Struc- 
 tures involving this peculiar feature of the well-known " toggle joint " are regarded tem- 
 
 porarily unstable, and equilibrium in them 
 can be established only after certain 
 infinitesimally small displacements have 
 taken place. 
 
 Another striking example of this kind 
 is shown in Fig. 3c, where the supports A 
 and C are hinged, and the support B is 
 movable, and hence 2r=5. Also w = 19 
 and p = \2, making n=w + Hr 2p=0. 
 
 This structure is made up of three 
 elements I, II and III, each one of which 
 may be regarded as a simple frame in direct contact with a support. Hence e=3 and 
 
 This proves the structure to be internally as well as externally statically determinate, 
 
ART. 3 
 
 REACTIONS AND REDUNDANT CONDITIONS 
 
 9 
 
 though a condition of temporary instability exists because the point k is common to 
 the paths of the three elements in the first instant of rotation about their respective 
 centers of rotation at A, B, and C. 
 
 FIG. 3o. 
 
 FIG. 3o. 
 
 FIG. 3K. 
 
 FIG. SQ 
 
 FIG. SE. 
 
 FIG. 3F. 
 
 FIG. 3n. 
 
 FIG. 3j. 
 
 FIG. SL. 
 
 FIG. 3n. 
 
 The point k may thus be regarded as an imaginary crown hinge for a three-hinged 
 arch AkC. Hence a load P, acting at D, will cause a small rotation of the three elements 
 before equilibrium is established and then the stresses become infinite. 
 
10 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. I 
 
 The only loads which will not cause rotation are those passing through k, but for 
 any of these loads, there will result an infinite number of conditions of equilibrium and 
 as many different systems of stresses. 
 
 The following illustrations are presented for the purpose of showing various applica- 
 tions of Eqs. (3c) and (SE). Figs. 3D to 3M are all examples of simple truss elements 
 for which e = l. Fig. SN has six such elements counting the two suspenders cd, which 
 must always be included when the elements alone are regarded. For Fig. 3o, e=2 and 
 for Fig. 3p, e=8. 
 
 It should also be noted that Eq. (3c) is applicable to any structure as a whole, 
 whether composed of one or many elements. This equation will apply when the individual 
 members and pin points are counted and also when the elements and their connecting 
 points only are considered. 
 
 For example, in Fig. SN, there are 56 members, 32 pin points, and 8 reaction con- 
 ditions, making n = Sr+ra 2p=8+56 2X32=0, whence the structure is statically 
 determinate. Now when the elements alone are considered, then e=m--=Q, p=7 and 
 2r=8, and Eq. (3c) again gives ri = 5>+ra-2/>=8+6-2X7=0. Eq. (3E) gives 
 n' = 2r e2=86 2=0. Hence, both equations may serve to test the external redun- 
 dancy, when the elements alone are considered, but internal redundancy requires count- 
 ing all members and pin points in the entire structure. It is thus important to understand 
 the exact interpretation of these two test equations. 
 
 As a general rule it is best to count intersecting web members as four members 
 instead of two, though the result is usually identical. Thus in Fig. SK, n = 2r+m 2p = 
 34-1720=3+2124=0. But in this example the structure is not composed of 
 a succession of triangles and when pairs of members, meeting in a point, are removed, 
 the structure will not reduce to a single triangle, showing that it is not stable. 
 
 Fig. 3cj is another example of the toggle joint when the members at c are not 
 connected. 
 
 The following table was arranged to illustrate all these points with reference to the 
 several cases represented by Figs. 3D to 3n. 
 
 Fig. 
 
 JY 
 
 m 
 
 p 
 
 e 
 
 n = ~ r + -m 2p 
 
 n 1 = r-e-2 
 
 Remarks. 
 
 3D 
 
 3 
 
 1 
 
 2 
 
 1 
 
 3+ 1- 4 = 
 
 3-1-2=0 
 
 Determinate. 
 
 3E 
 
 6 
 
 1 
 
 2 
 
 1 
 
 6+ 1- 4 = 3 
 
 6-1-2=3 
 
 3 times ext. ind. 
 
 3r 
 
 6 
 
 1 
 
 2 
 
 1 
 
 6+ 1- 4 = 3 
 
 6-1-2=3 
 
 3 times ext. ind. 
 
 3G 
 
 3 
 
 8 
 
 5 
 
 1 
 
 3+ 8-10=1 
 
 3-1-2=0 
 
 Once int. ind. 
 
 3H 
 
 4 
 
 21 
 
 12 
 
 1 
 
 4+21-24 = 1 
 
 4-1-2=1 
 
 Once ext. ind. 
 
 3j 
 
 6 
 
 17 
 
 10 
 
 1 
 
 6+17-20 = 3 
 
 6-1-2=3 
 
 Thrice ext. ind. 
 
 3K 
 
 3 
 
 21 
 
 12 
 
 1 
 
 3+21-24 = 
 
 3-1-2=0 
 
 Not stable. 
 
 3L 
 
 5 
 
 45 
 
 24 
 
 1 
 
 5+45-48 = 2 
 
 5-1-2=2 
 
 Twice ext. ind. 
 
 3M 
 
 3 
 
 33 
 
 18 
 
 1 
 
 3+33-36 = 
 
 3-1-2=0 
 
 Determinate 
 
 3N 
 
 8 
 
 56 
 
 32 
 
 6 
 
 8+56-64 = 
 
 8-6-2=0 
 
 Determinate. 
 
 3o 
 
 4 
 
 10 
 
 7 
 
 2 
 
 4+10-14 = 
 
 4-2-2=0 
 
 Determinate. 
 
 3p 
 
 10 
 
 16 
 
 13 
 
 8 
 
 10+16-26 = 
 
 10-8-2 = 
 
 Determinate. 
 
 3<> 
 
 3 
 
 9 
 
 6 
 
 1 
 
 3+ 9-12 = 
 
 3-1-2=0 
 
 Case of infinite stress. 
 
 3R 
 
 8 
 
 3 
 
 4 
 
 3 
 
 8+ 3- 8 = 3 
 
 8-3-2=3 
 
 Thrice ext. ind. 
 
 
 
 
 
 
 
 
 
CHAPTER II 
 
 THEOREMS, LAWS, AND FORMULAE FOR FRAMED STRUCTURES 
 
 ART. 4. ELASTIC DEFORMATIONS, FUNDAMENTAL EQUATIONS 
 
 Elastic Deformations 
 
 Let S = total stress in any member of a frame resulting from any cause, or causes, 
 
 designating tension by + . 
 I = length of any member when its S = 0. 
 Al= change in I due to stress S, + for elongation. 
 F = cross-section of any member, prismatic in form. 
 t = a uniform change in temperature in degrees, + for rise. 
 = coefficient of expansion per degree of temperature. 
 E = modulus of elasticity. 
 fS/Fumt stress in any member. 
 JZ/Z=relative elongation. 
 l/FE =|0=the extensibility, frequently employed for brevity. 
 
 Then according to Hooke's law and within the elastic limit of the material, 
 
 EF 
 
 E 
 
 from which 
 
 SI 
 
 (4A) 
 
 This equation represents the elastic deformation 
 for any member of any frame and is a fundamental 
 elasticity condition. 
 
 Referring now to Fig. 4A, let AB be any member 
 of any frame which, as a result of elastic distortions of 
 the frame, is made to undergo displacements Ji at A 
 and J 2 at B and a change in its original length of Al. 
 The new length of the member is then I + M and its new 
 position may be shown as A\Bi. 
 
 Since the displacements Ji and A% are very small 
 compared with the length AB, the two lines AB and 
 A\BI are assumed parallel for the present purpose. 
 The member and its displacements are referred to rectangular axes in Fig. 4A. Then 
 
 FIG. 4A. 
 
 11 
 
12 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. II 
 
 which, differentiated, gives 
 
 2ldl=2(x b x n ) (dx b -dx a ) +2(y b -y a ) (dy b -dy a ) . 
 
 Also from the figure x b x a =l cos a. and y b y a ^l cos /9, which values substituted 
 in the previous equation, dividing through by 21, and replacing the differentials by 
 small finite increments A, gives 
 
 Al = (Ax b - Jx a ) cosa + (4y b - Ay a } cosp = -- + etl, ..... (4s) 
 
 tir 
 
 which is a fundamental elasticity condition for any frame. 
 
 Hence, every frame will involve twice as many of the unknown, subscript-bear- 
 ing displacements Ax and Ay, as there are pin points, while there will be as many equa- 
 tions of the form Eq. (4e), as there are members in the frame. 
 
 It will now be shown that any frame, whether involving redundancy or not, is capable 
 of anatysis by proving the following: For any frame in stable equilibrium, there are as 
 many possible condition equations as there are unknown quantities, provided the elasticity 
 conditions Eq. (4s) are included. This will be true of a statically determinate frame 
 without including the elasticity conditions. 
 
 The ultimate analysis of any frame includes the determination of the reaction forces 
 or conditions; the stresses in all the members; and the deformation of the frame as a 
 result of these stresses. The deformation is considered solved when the displacements 
 Ax and Ay, of all the pin points, are found. 
 
 It is assumed that for any structure under consideration, the externally applied loads 
 P, the changes in temperature t and the abutment displacements Ar (if any exist) are 
 all known and that wherever movable connections or roller bearings occur, these are 
 frictioriless. 
 
 Let S\, 82, 83, etc., be the stresses in the several members meeting in some particular 
 
 pin point m. 
 ct\, (Xz, 3, etc., be the angles which these members make with the x axis of 
 
 coordiantes. 
 
 /?i, /?2, /?3, etc., be such angles with the y axis. 
 P x =the sum of the components parallel to the x axis, of all the external forces 
 
 P acting on the point m. 
 
 P v =the sum of the components parallel to the y axis, of all these forces P. 
 p, m and 2r as previously defined in Art. 3. Then 
 
 (4c) 
 cos ,5=0 
 
 because for a state of equilibrium, thje sum of the horizontal and vertical components of 
 all the forces acting on any one pin point must respectively equal zero. 
 
 Hence, for every pin point of a frame, two equations of the form of Eqs. (4c) may 
 be written, expressing equilibrium of the internal and external forces for that pr'- 
 
ART. 4 THEOREMS, LAWS, AND FORMULAE FOR FRAMED STRUCTURES 13 
 
 Also, one equation of the form of Eq. (4s) may be written for each member of a frame. 
 Besides these, there will be one equilibrium equation for each reaction condition. 
 
 Therefore, every stable frame affords 2r+m+2p condition equations of the first 
 degree, involving as many unknowns as there are equations, thus: 
 
 2p equations of the form (4c) involving ra unknowns S. 
 m equations of the form (4B) involving 2p unknowns Ax and Ay. 
 Zr equations for the reactions involving Zr unknowns R. 
 
 Total 2p-fra + Zr equations, involving 2p+m + r unknowns. 
 
 It follows then that the stresses S, the reactions R of known direction, and the Ax 
 and Ay projections of the pin-point displacements, may all be represented as linear func- 
 tions of the horizontal and vertical components of all the external forces P, plus similar 
 functions of assumed temperature changes t, plus linear functions of the abutment 
 displacements Ar, Jr 2 , Ar 3 , etc. 
 
 Thus any unknown function of any frame may be expressed by an equation of the 
 
 form 
 
 Z=f(P l ,P 2 , P 3 , etc.)+/i(0+/ 2 (M,M, ^ 3 , etc.), ..... (4o) 
 
 in which the coefficients are independent of the values P, t and Ar, but depend on the lengths 
 and directions of the members also on E, e and the manner in which the frame is supported. 
 The law of the summation of similar partial effects. In Eq. (4o), every set of causes 
 or conditions P, t and Ar, produces a partial value Z r for Z and the ultimate total value 
 of Z=Z' +Z" +Z'", etc., is the sum of all the partial values or effects resulting from the 
 respective sets of independent causes or conditions. Thus each effect, such as a stress, 
 whether due to loads, temperature or abutment displacements, may be ascertained or invest- 
 igated by itself and the sum total effect Z will then be the sum of the several similar partial 
 effects. This is the law of the summation of similar partial effects resulting from various 
 causes and is fundamental to the analysis of all structures involving redundancy. 
 
 The law of proportionality between cause and effect. Eq. (4o) , being true for any set 
 of effects, would remain true for any multiple of these effects. Hence, if the loads are 
 doubled the resulting stresses will likewise be doubled, etc. Therefore, the law of pro- 
 portionality holds true between the causes and their effects. Thus, if a set of loads P 
 produces stresses S in the members of a frame then another_set of parallel loads P, 
 acting at the same points, would produce stresses S which are P/P times as great as^the 
 stresses S. Or, if a single load unity, acting on a point m of any frame produces reactions 
 Ri, and stresses Si, then a load P m acting at the same point will produce reactions 
 R = P m Ri, and stresses S =P m Si. 
 
14 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. II 
 
 ART. 5. GENERAL WORK EQUATIONS FOR ANY FRAME 
 
 The general work equation, Clapeyron's law. A loaded frame is a machine in equilibrium 
 and within the limits of proportional elasticity its deformation varies directly with the 
 magnitude of the superimposed loads. 
 
 The question of deformation does not enter into statics and hence a general and 
 comprehensive treatment of the elastic frame must necessarily involve the principles 
 of mechanics. , 
 
 For example, given a frame with definite loading and supports, constituting a sys- 
 tem in external equilibrium. The frame in turn will undergo certain deformations which 
 will steadily increase in proportion to the internal stresses created in the members by 
 the external forces as they are gradually applied. The final deformation will occur 
 in the instant when the external forces are exactly balanced by the internal stresses, pass- 
 ing into the condition known as static equilibrium. 
 
 While this deformation is taking place the applied loads travel through certain 
 distances which are the paths or displacements of the points of application of the loads. 
 Thus a certain quantity of positive work of deformation is performed by the external forces. 
 
 The internal stresses in the members must accommodate themselves to the deformed 
 condition of the frame and in resisting this action must produce negative work of deformation. 
 
 In the instant when static equilibrium is established between the loads and stresses 
 the positive and negative work of deformation produced in the same interval of time 
 must exactly balance. 
 
 Let A e =the positive or externally applied work of deformation. 
 .4;= the negative or internally overcome work of deformation. 
 P=any externally applied loads including the reactions. 
 $=the stress in. any member due to the loads P. 
 M ihe change in length of any member due to the stress S. 
 =the displacement of the point of application of any force P measured in 
 the direction of this force. 
 
 A positive amount of work is always produced when the force and its displacement act 
 in the same direction. 
 
 The product %Pd represents the actual work produced by a force gradually applied 
 and increasing from its initial zero value to a certain end value P, thus exerting only 
 its average intensity during the entire time of traversing the path to perform this work. 
 Hence, for all the external forces acting on any frame, the total positive external work 
 of deformation would be A e =^Pd. 
 
 Similarly %SAl represents the actual work in any member subjected to a gradually 
 increasing stress of end value S, with an average intensity S/2 during the entire time 
 while producing a change in its length of JZ. Hence for all internal stresses in any frame 
 the total negative work of deformation would be A t -=^2$JZ. 
 
 By the " doctrine of the conservation of energy," the applied work must equal the 
 work overcome, hence 
 
 (5A) 
 or 
 
ART. 5 THEOREMS, LAWS, AND FORMULAE FOR FRAMED STRUCTURES 15 
 
 which is Clapeyron's law (1833), and may be stated as follows: For any frame of constant 
 temperature, and acted on by loads which are gradually applied, the actual work produced 
 during deformation is independent of the manner in which these loads are created and is 
 always half as great as the work otherwise produced by forces retaining their full end values 
 during the entire act of deformation. 
 
 By regarding any elastic body as composed of an infinite multiplicity of members, 
 it is readily seen that Clapeyron's law applies equally to frames and solid web elastic 
 structures when properly supported. Clapeyron's law finds extensive application in 
 problems dealing with deformations of framed structures. 
 
 Cases involving dynamic impact would imply a certain amount of kinetic energy 
 in excess of the applied work of deformation. That is, the applied forces have some 
 initial value, greater than zero, while the stresses are still zero. This would not be repre- 
 sented by the work Eq. (OA), wherein A e Ai=Q, but would give rise to the following 
 equation : 
 
 Ae-Ai-jPf, ..... . ..... (OB) 
 
 indicating a state of accelerated motion instead of one of equilibrium. For problems 
 under this heading see Art. 16. 
 
 The law of virtual work will now be considered. It is a general law, permitting of 
 more varied application than does Clapeyron's law. 
 
 The law of virtual work was first enunciated by Galilei and Stevin, and in its more 
 general form by Joh. Bernoulli. Lagrange (1788) reduced the law to an algebraic expres- 
 sion for which the following derivation may be given. 
 
 According to the general law expressed by Eqs. (4c). all forces (and stresses) acting 
 on a pin point of any structure in equilibrium will have components parallel to any 
 axis and the sum of such a set of parallel components must be zero. Calling o: the angle 
 which any force or stress P makes with the axis chosen, then for any pin point 
 
 2P cosa=0. 
 
 Now if a displacement J, parallel to the axis, be arbitrarily assigned to this pin point 
 and assuming equilibrium to continue, then the product of J with the sum of the com- 
 ponents must still be zero. This is equivalent to multiplying the above equation by 
 J to obtain 
 
 2(Pcosa)J=0. 
 
 But the displacements A cos a=d are the projections of the displacements J on the 
 directions of the several forces, hence 
 
 0, ............. (5c) 
 
 wherein P signifies that the forces are in every sense independent of the displacements d. 
 This is the law of virtual work, and is applicable to any point or group of points and 
 hence to any frame or group of bodies. The displacements d may be any possible dis- 
 placements whether or not subject to the law of elastic deformation, provided equilibrium 
 exists. 
 
KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. II 
 
 Stated in words this law implies that for any point, frame or body acted on by loads 
 P in an established state of equilibrium, the sum total work performed by these forces, in 
 moving over any small arbitrary but possible displacements o, must always be equal to zero. 
 
 If the time element is introduced into Eq. (oc) giving SPo/f =0, the law of virtual 
 velocities is obtained. 
 
 Professor Otto Mohr's work equations. Proceeding from Eq. (5c), Professor Mohr, in 
 1874, developed two equations which furnish the most comprehensive laws for framed struc- 
 tures. These equations are now derived. 
 
 Given any frame carrying the arbitrarily assumed loads PI, P 2 , P 3 , etc., which 
 in combination with temperature changes produce stresses Si, *S 2 , 83, etc., in the several 
 members of the frame and thus constituting a system of forces in static equilibrium. 
 
 All the pin points of the frame are now supposed to be subjected to small, arbitrarily 
 assigned displacements A\, J 2 , J 3 , etc., which may have been produced by some other 
 system of actual loads PI, P 2 , P 3 , etc., entirely independent of the loads P. The J's 
 will naturally vary in amount and direction for each pin point. 
 
 According to the la\vs for static equilibrium, the sum of the components, taken in 
 any fixed direction, of all forces acting on any pin point of a frame must equal zero. 
 
 Choosing for the fixed direction the displacement J lr 
 then for any particular pin point, Fig. SA, the follow- 
 ing equation may be written: 
 
 cos 
 
 which multiplied through by 
 JiP x cos 61 + Ji 
 
 5 cos 0=0; 
 i gives 
 cos 0=0. 
 
 But from the figure, A\ cos #1 =<?i, which value sub- 
 stituted in the above equation gives 
 
 0=0. 
 
 (on) 
 
 FIG. SA. 
 
 Eq. (on) represents the virtual work of all forces 
 acting on the pin point 1, in accordance with Eq. (5c). 
 One such Eq. (5n) may be written for each pin 
 point and the sum of all these equations would furnish 
 the total virtual work for the whole frame. 
 
 In this sum equation, each load P will occur only once, while each stress S will occu 
 twice, being involved once for each end of the same member. 
 The terms containing the forces P will give the sum 
 
 The sum terms involving the same stress S may be found for each member as 
 
 cos 
 
AHT. 5 THEOREMS, LAWS, AND FORMULA FOR FRAMED STRUCTURES 17 
 
 since the two displacements J cos </> for the two ends of the same member give the sum 
 Jl, or actual change in the length of such member. 
 
 Hence, the terms containing the stresses S will give the sum 
 
 where the minus sign is due to the fact that the stresses applied to pin points as external 
 forces are always opposite in direction with their respective JZ's. 
 The final sum equation for all pin points then becomes 
 
 or 
 
 2P3 = 2SM, ............ (5s) 
 
 which may be said to represent a condition of elastic equilibrium as distinguished from 
 static equilibrium. The law thus expresses the equality between the external and internal 
 virtual work of deformation for real displacements and arbitrary cases of loading, provided 
 elastic equilibrium exists. 
 
 The external forces P necessarily include the reactions due to the applied loads P. 
 However, the work of the reactions is always zero when the abutments are immovable. 
 In the general case involving abutment displacements Jr in the directions of the reac- 
 tions R produced by loads P, Eq. (OE) may be written 
 
 .... (OF) 
 
 It should be repeated that the displacements d, Ar and Al are actual and mutually 
 dependent on the same causes, such as the actual loads P and temperature changes. 
 The arbitrary loads P, reactions R and stresses S form a system in elastic equilibrium 
 which is independent of the actual loads P, reactions R and stresses S, and hence inde- 
 pendent of the actual displacements. 
 
 In the special case where the forces P, S and R become identical with the forces 
 P, S and R, Eq. (5r) becomes 
 
 v _ v - y S2 * 
 EF' 
 
 which is in accordance with Clapeyron's law. 
 
 Eq. (5r) is the fundamental law of framed structures and includes all conditions 
 of equilibrium of the external forces; of the external forces and internal stresses; and 
 of all relations existing between the stresses and the distortions of any frame. 
 
 Professor Mohr's second work equation serves the purpose of determining any displace- 
 ment d m produced by any case of loading. It follows from Eq. (5r) by allowing all the 
 arbitrary loads to vanish and substituting therefor a single load unity and the stresses 
 Si and reactions RI resulting from such unit load. The new work equation then becomes 
 
is 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. II 
 
 wherein the unit loading, henceforth called the conventional loading, may be a single 
 force unity or a moment equal to unity. In the latter case the o will represent a rotation 
 measured in arc. By dividing Eq. (5n) by unity the stresses Si become abstract num- 
 bers and the equation will read 
 
 The applications of this law, universally known as Mohr's icork equation, will be shown 
 in the following articles 6 and 7. The work represented by the conventional loading will 
 generally be designated by W. 
 
 ART. 6. DISPLACEMENTS OF POINTS. STATICALLY DETERMINATE STRUCTURES 
 BY PROFESSOR MOHR'S WORK EQUATION 
 
 The relative displacement 8 m between any two pin points mi and m, of any frame Fig. 
 GA, when Al is given for each member of the frame, may be found by applying Eq. (5i) . 
 
 If the abutments undergo known displacements Ar, as a result of the given actual 
 loading, this effect on the required d m must be included. 
 
 Likewise temperature displacements may be considered, but the problem will first 
 be treated by neglecting this effect. 
 
 Now assume two unit forces, applied at the points mi and m respectively, and, acting 
 in opposite directions along the line tn\ < m. The directions of these unit forces should be so 
 
 chosen as to make the conventional work l-o m a 
 positive quantity. This means that if d m is an 
 elongation, then the unit forces must be so 
 applied as to elongate the distance m\m. This 
 rule will be universally applied to determinate 
 structures as well as to indeterminate for 
 redundant members or conditions. A negative 
 result will indicate an erroneous assumption 
 in the direction of the unit conventional 
 load. 
 
 The stresses Si and abutment reactions R i 
 resulting from the two unit loads acting at mi 
 and m are now determined from a Maxwell dia- 
 gram or by computation. 
 
 The work Eq. (5n) applied to this assumed or conventional loading then gives, after 
 substitution of values Al=Sl/EF, excluding temperature effect, 
 
 FIG. GA. 
 
 (GA) 
 
 The stresses S are those produced in the structure by any real case of loading, as loads 
 P, temperature changes or abutment displacements, giving rise to the actual changes 
 Al in the lengths of the members. 
 
ART. 7 THEOREMS, LAWS, AND FORMULAE FOR FRAMED STRUCTURES 19 
 
 Eq. (6A) then offers a solution to the above problem, because all the quantities except 
 o, n are known or can be found from the given data. 
 
 If temperature effects are to be included then Eq. (4A) gives the value for JZ, thus 
 furnishing the final equation 
 
 ~+eU)-ZR l Jr ..... \ . . (6 B ) 
 &V / 
 
 In a similar manner the relative displacement of any pair of points, the deflection 
 of any point, or the angular change between any pair of lines of any frame, resulting from 
 given changes M of all the members of the frame, may be found by assigning such con- 
 ventional loads as will produce the conventional work 1 d on the frame at the point or 
 points in question. See also Art. 9 for other examples of this class. 
 
 ART. 7. INDETERMINATE STRUCTURES BY PROFESSOR MOHR'S WORK EQUATION 
 
 A truss, according to Chapter I, may include more than the statically necessary 
 number of members or reaction conditions, and is then called statically indeterminate. 
 It is now proposed to show the manner in which Mohr's work equation may be employed 
 to find the stresses and reactions in such a structure, loaded by any system of loads 
 P concentrated at the several pin points of the frame. 
 
 The structure must be so constituted that when all redundant members and reaction 
 forces are removed, the remaining frame will represent a statically determinate structure, 
 including the necessary conditions for proper support. This determinate frame will 
 always be called the principal frame or system.. 
 
 Now let X a , Xi, X c , etc., represent the stresses produced by the applied loads in 
 any redundant members or supports, as the case may demand. When these stresses 
 are applied to the principal system, together with the external loads, the resulting stresses 
 in all the principal members will be identical with those produced in these same mem- 
 bers by the original loading of the whole indeterminate frame. This will also be true 
 of the deformations. 
 
 It follows then that the loads P applied to the indeterminate frame, and the loads 
 P and X applied to the principal frame produce identical stresses and deformations in the 
 principal members. Also, the deformations are in each case definitely fixed by the elastic 
 changes Al of the necessary members of the principal system. Hence, for applied loads 
 only (without temperature effects) the work of the stresses in the principal frame is 
 always greater when a redundant member is omitted. The work which a redundant 
 member can do must necessarily lessen the work otherwise required of the principal 
 system. 
 
 If, then, the elastic displacements of the several pin points representing the points 
 of application of the redundant forces X, are found for the principal system, these dis- 
 placements will suffice to solve one elasticity equation for each unknown X. This may 
 be done in two ways: (1) By applying Mohr's work equation to the indeterminate sys- 
 tem, and (2) by finding the requisite elastic displacements of the principal system from 
 Maxwell's law. The solution by Mohr's work equation will now be given. 
 
20 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. II 
 
 The following definitions of terms will be strictly adhered to in all succeeding 
 discussions. 
 
 Let S =the stress in any member of the principal system. 
 
 >S> =the stress in this member due to the loads P when the several redundants 
 
 X and t are all zero, to be known as condition X=Q. 
 a =the stress in this member of the principal system when no load other than 
 
 X a = 1 is active. Condition X a = 1 . 
 
 Sb=ihe same when no load other than Xi = l is active on the principal system. 
 /S c =the same when no load other than X c = l is active on the principal system. 
 j=the stress in the same member caused by a uniform change in temperature. 
 R t =& reaction produced by a uniform change in temperature. 
 M t = a moment produced by a uniform change in temperature. 
 X a , Xb, X CJ etc., the stresses in the redundant members or reactions. 
 R , R a , Rb R c , etc., are denned like the S's with like subscripts, but represent 
 
 reaction forces instead of stresses. 
 da, <>b, $c, are changes in the lengths of the redundant members X a , X b , X c , 
 
 respectively. 
 
 The forces X a , Xb, X c , etc., may be the stresses in redundant members or they may 
 be redundant reactions as in the case of fixed arches or continuous girders, etc. 
 
 Each of the above cases of conventional loading will be known as conditions. Thus 
 condition X =0 will mean that all the redundant conditions are removed, while condi- 
 tion X a = l will signify that this force alone is applied to the principal system and all 
 other X's, S t , and P's are removed. 
 
 The stress S, in any member of a frame involving redundant conditions, is a linear 
 function of the loads P, X a , Xb, X c , etc., all treated as external forces applied to the 
 principal system. This follows because all conditions of equilibrium are represented 
 by linear equations. 
 
 Hence, according to the law of the summation of effects expressed by Eq. (4o) , the 
 general equation for stress in any member of a truss involving redundancy, would have 
 the form 
 
 wherein ax =the stress in any member due to the external force X a acting alone on the 
 principal system. Similarly 5 6j .=the stress in the same member due to the external 
 force Xb acting alone on the principal system, etc. 
 
 But by the law of proportionality S ax =S a X a , Sb x =SbXb and S cx =S c X Ci etc. Hence, 
 the general equation for any case of redundancy may be written : 
 for the stress in any principal member 
 
 /S=*So S a X a SbXb S C X C , etc., +/S< 
 
 + 
 
 for any reaction of the principal system 
 
 R=R R a X a Rb Xb R c X c , etc., -\-Rt ...... (7 A) 
 
 and for any moment on the principal system 
 
 M=M -M a X a -M h X b -M c X c , etc.,+M 
 
ART. 7 THEOREMS, LAWS, AND FORMULAE FOR FRAMED STRUCTURES 21 
 
 In these equations the quantities S , R and M are all linear functions of the 
 externally applied loads P, while all the stresses, reactions, and moments bearing 
 subscripts a, b, c, etc., are constants due to conventional loadings and are absolutely 
 independent of all external loads P and X. 
 
 The work of any member or reaction is obtained from Eqs. (7 A) by multiplying both 
 sides of these equations by the deformation which each sustains. Thus, neglecting 
 temperature effects, the work equations are 
 
 SM=[S -S a X a -S b X b -S c X c , etc.]JZ 
 RJr=[R -R a X a -R b X b -R c X c , etc.]Jr 
 
 The negative signs in Eqs. (7 A) and (7e) , indicate that the quantities involving the 
 redundants X must always be of opposite sign to the stress S resulting from the loads 
 P, for reasons above stated in this article. 
 
 Hence the increment of work performed by the redundants X, as shown by Eqs. (7s), 
 is always negative with respect to the work performed by the stress S because the forces 
 X are classed with the external forces. 
 
 The conventional unit loadings are always applied in the opposite direction to the 
 forces they replace, and this in turn puts these (negative) unit forces in the same direc- 
 tion as the displacements d which they accompany. Hence the work 1 o is always 
 positive. 
 
 This is exactly the same as the case illustrated in Art. 2, where the unit load was 
 independent of all redundant conditions. 
 
 Therefore, whenever in the following a conventional unit load represents a redundant 
 condition, the direction of this unit load must be taken in a direction opposite to the force 
 representing the stress in such member. This is equivalent to saying that when a 
 member X elongates under stress, then the unit load X = l must be so applied as to move 
 the adjacent pin points apart, and the converse when the member shortens under 
 stress. 
 
 Whenever the direction of a redundant force cannot be correctly foreseen, then some 
 assumption is made which, if it be erroneous, will result in a negative value for such 
 redundant X. 
 
 The following illustration is given to make these points clear and to avoid confu- 
 sion of ideas in all future problems. 
 
 Figures 7A to 7E represent an indeterminate structure involving three redundant 
 conditions as may be verified by applying Eq. (3c) to the problem in Fig. 7A. 
 
 The truss is supported by hinged bearings at A and D and by columns at B and C, 
 all of which may be subjected to certain displacements which may be estimated from 
 the conditions of the problem. 
 
 Figure 7s, then, represents the principal system carrying loads P, when all redundant 
 conditions are removed and shows condition X =0. The other figures represent, respec- 
 tively, the conventional loadings X a = 1 , X b = 1 and X c = 1 , the three X's being the redundant 
 conditions. Note here the directions which the unit forces take. They are negative 
 
22 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. II 
 
 with respect to the forces they replace, but in the same directions as the displacements 
 which their points of application undergo as a result of the loads P. 
 
 The reaction displacements are assumed to be downward by amounts A A, AB, AC, 
 and AD for these supports respectively, and it is also assumed that A and D move apart 
 by an amount AL. Temperature effects will be taken up later being omitted for the 
 present. 
 
 GIVEN CONDITIOM 
 
 CONDITION X=O. 
 
 FIG. 7A. 
 
 CONDITION Xf I. 
 
 FIG. 7c. 
 
 FIG. 7o. 
 
 CONDITION X c -l. 
 
 FIG. ?E. 
 
 The various reactions in the following table, given for the four conventional load- 
 ings, are now found for the principal system, which in this case is a three-hinged arch. 
 
 For Condition. 
 
 A 
 
 B 
 
 C 
 
 D 
 
 H* 
 
 V f\ 
 
 2P(L-e) 
 
 D (\ 
 
 
 2Pe 
 
 A n L-P(L-2e) 
 
 
 Ao L 
 
 .DO U 
 
 o vJ 
 
 D L 
 
 H 2h 
 
 ,, Uh-J:} 
 
 A-a 
 
 V, 1 
 
 ^o = 
 
 l(L-d) 
 
 B a = v 
 R, n 
 
 a=0 
 c', n 
 
 L> a \J 
 
 D h - ld 
 
 fla- 
 il 
 
 H h ld - 
 
 Afc 1 
 
 V 1 
 
 A b L 
 A 1 ' d 
 
 Dfe U 
 
 Bn 
 
 <-& i 
 
 L 
 D KL-rf) 
 
 Hb 2h 
 l.d 
 
 A c 1 
 
 Ac ~ L 
 
 c (J 
 
 c U 
 
 Dc L 
 
 Hc " 2h 
 
 * Obtained by taking moment equation about the middle hinge O, in a clockwise direction. 
 
ART. 7 THEOREMS, LAWS, AND FORMULA FOR FRAMED STRUCTURES 23 
 
 Hence from Eq. (7A) for reactions, these values give: 
 
 -e L-d d 
 
 A=A - 
 
 =0 
 
 C=0 
 
 and 
 
 h k d d 
 
 H = H H a X a H b X b H C X C =H X a ylXb FJT 
 
 S & & a X a &bXi) S C X C 
 
 . . . (7c) 
 
 The work equation (GA), when applied to each of the three redundant forces X a , X b 
 and X c and observing that the conventional work 1-d in each case is positive, gives 
 
 = + 1 AL 
 - + 1 AB 
 
 (7D) 
 
 The values for ^R Ar, in each of the Eqs. (?D) , may also be written out from the above 
 tabulated quantities for the three conventional loadings X a = 1 , Xb = l and X c = I . They 
 are: 
 
 h 
 
 A b AA-D b AD-H b AL 
 L d . . d .^ d 
 
 -A C AA-D C AD-H C AL 
 
 d . L d . d . 
 
 (7B) 
 
 It should be observed that when the direction of a force is opposite to the direc- 
 tion of the displacement over which the force travels, then the product which represents 
 work is always negative. This is the case with all the work quantities in Eqs. (7E) . Hence, 
 when these are substituted into Eqs. (7o) they become positive. 
 
 The quantities 2>S a Al, HSbAl and ^S C AI still remain to be evaluated. Eq. (4A) gives 
 M=Sp + ett, where S=S S a X a -S b X b -S c X c from Eq. (7 A). Then by substitution 
 
 M=[S -S a X a -S b X b -S c X c ]p+dl, 
 
 which gives the final value 
 
 . . . (7c) 
 
24 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. IT 
 
 Similar expressions follow for 2S&JZ and 2$ C JZ, and these substituted in Eqs. (7o) 
 give for the values d: 
 
 d a = 
 
 . . (7n) 
 d c = 
 
 wherein the summations, except those of the reactions, include only the members of 
 the principal system. 
 
 Now d a , 5, and d c being the changes in the lengths of the three redundants, these may 
 be evaluated from Eq. (4A) in terms of the lengths, areas and stresses of these members 
 (or reactions) and become 
 
 -A- , , , . 
 
 (73) 
 
 Hence all the terms in Eq. (7n) are now known except the three redundant forces 
 X a , X b and X c , and having three elasticity equations involving only these unknowns, 
 the latter may be found by solving Eqs. (7n) for simultaneous values of the X's, with 
 the aid of Eqs. (?E) and (7j) . 
 
 It is thus seen that the abutment displacements are of vital importance in determin- 
 ing the magnitudes of the stresses in any statically indeterminate structure. When- 
 ever these displacements cannot be determined with any degree of certainty, or when 
 small displacements indicate large resulting stresses, then such structures should not 
 be built. This applies particularly to fixed arches and continuous garders. 
 
 Eqs. (7n), based on Professor Mohr's work equation, thus furnish a means for 
 the analysis of any statically indeterminate structure. 
 
 ART. 8. INDETERMINATE STRUCTURES BY MAXWELL'S LAW 
 
 The application of Maxwell's law to the same analysis will now be given by express- 
 ing the summations in Eqs. (7n) in terms of the elastic deformations of the structure 
 instead of the stresses in the members. 
 
 Leto ma =the displacement of the point of application m of any load P, n , in the 
 direction of this load, when the principal svstem is loaded with only 
 
 Xa = l. 
 
 d mb =fi similar displacement of the same point m for the conventional load- 
 ing X 6 = l. 
 
 d nlc =ihe same for condition X c = l. 
 
 daa =the change in length of the member a for condition X a = l. 
 d^ =the change in length of the same member a for condition X b = l. 
 doc =the similar change for condition X c = \. 
 dba = the change in length of the member b for condition X a = 1 . 
 dbb =the change in length of the member 6 for condition Xb = l. 
 
ART. 8 THEOREMS, LAWS, AND FORMULA FOR FRAMED STRUCTURES 25 
 
 dbc = the change in length of the member 6 for condition X c = l. 
 d ca =the change in length of the member c for condition X a = l. 
 d c t, =the change in length of the member c for condition X& = 1. 
 d cc =the change in length of the member c for condition X c l. 
 o al =the change in length of the member a resulting from a change in tempera- 
 ture t, when the principal system is otherwise not loaded, hence 
 
 x=o, p=o. 
 
 Ob<=the same for the member b. 
 d ct =the same for the member c. 
 
 For the sake of simplicity, the abutments will be assumed immovable, thus making 
 all Jr=0. This part of the previous problem remains unchanged. 
 
 The work Eq. (OE) for the condition X 0, (when the loads P only are acting on 
 the principal system and producing the stresses S ) now becomes for the three cases of 
 displacements 
 
 > 
 
 Similarly applying the work Eq. (GA) to the conditions X a = l, Xb = l and X e = l, 
 the various displacements become 
 
 1 d aa = 2S a M u ; l-3 ab = 2S a M b and 1 3 ac = 2S a M c ; 
 and by inserting values for Al aj 4l b and Al c from Eq. (4A) then 
 
 and similarly for conditions X b = 1 and X c = 1 
 
 2 S c S a p =d c a] S S c Si,p =d c b', 2 S c S c p = d c 
 
 (SB) 
 
 Finally the work Eq. (6B), for temperature displacements for each of the conven- 
 tional unit loadings, gives 
 
 SS etf - 1 3 at ; 2S b dl = l-3 bt and SS c stf = 1 d ct ...... (8c) 
 
 Substituting all these values from Eqs. (SA) , (SB) , and (8c) into Eqs. (?H) , the follow- 
 ing important equations are obtained : 
 
 Xcdac S# a Jr +d at 
 
 Ob = ZPmdmb X a dba X b dbbX c db c T>Rb 
 d c = 2P m 3 mc -X a d m -X h d cb -X c 3 cc - 2R c 
 
 (SD) 
 
 It should be noted that for the quantities on the right-hand side of Eqs. (So), the 
 single subscripts and the second one of the double subscripts always refer to the con- 
 
26 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. II 
 
 ventional loadings X a = l, X b = l and X c = l, while the first of the double subscripts always 
 refers to a point or member of the structure. 
 
 According to Maxwell's law (proven in Art. 9) the following equalities exist in Eqs. (So) : 
 
 $ab = dba', dac=^ca and d bc = O cb . 
 
 This means that the order of the subscripts may be interchanged at will without 
 altering the equations, thus greatly simplifying the solution of problems. 
 
 All of the displacements d, in Eqs. (So), may now be determined by four Williot- 
 Mohr displacement diagrams (see Chapter VI), drawn respectively for the conditions 
 X a = l, X b = l, X c = l and t = l acting on the principal system. Hence, the three redun- 
 dant conditions X are again found by solving Eqs. (SD) for simultaneous values. 
 
 It is thus seen that any indeterminate structure may be analyzed in either of two 
 ways, by Eqs. (7n) or (SD), according to the choice of the designer or the nature of the 
 problem. 
 
 The redundants X being found by either of the above methods, all the stresses S 
 and reactions R may now be determined from Eqs. (7 A) . In these the X's are the redun- 
 dant forces from Eqs. (7n) or (So) and the stresses S , S a , S b and S c are those found for 
 the conventional loadings on the principal system and are independent of the values 
 X and of each other. 
 
 The summations in all the previous equations include only the members of the 
 principal frame. However, an equation of the form (7n) may be written to cover all 
 members, including the redundant, and this form is frequently very useful. 
 
 The work Eqs. (7o) when made to cover all members of an indeterminate frame 
 become 
 
 S# c Jr = SS c JZ ..... . (SE) 
 
 These values inserted into Eq. (7o), and others of that form, give 
 
 + 2S a ea 1 
 
 (8r) 
 
 2# c Ar = 2S c S p -X a ^S c Saf) -X b ZS c S bf ) - X c ^S 
 
 wherein the summations extend over all the members including the redundant, and all 
 the terms retain their previous significance. Therefore, when X a = l, then S =0, 
 S a = l=X a , S b =0, and S c =0. 
 
 The first terms of the right-hand side of each of the Eqs. (8r), according to Eqs. 
 (SA), may be expressed in terms of the external loads as follows: 
 
 2iS S a p= 2P m $ mo ; ^S S b f>~^P m d mb ', 2)S *ST C jO=lIP m ^ 7nc , . . . (SG) 
 
 and may be evaluated from Williot-Mohr displacement diagrams, Chapter VI. All the 
 other summations may be determined once for all either from Maxwell stress diagrams, 
 or in terms of displacements by employing Eqs. (SB) . 
 
 Problems of the kind treated in Art. 6 can now be solved for any indeterminate 
 structure by following precisely the same method there indicated except that the redundants 
 
ART. o THEOREMS, LAWS, AND FORMULA FOR FRAMED STRUCTURES 27 
 
 must first be found by one of the two methods just given, finally employing Eq. (?A), 
 to ascertain the stresses. These stresses being found, the elastic changes in all 
 members are computed. Then by applying a conventional unit load to the point for 
 which the displacement is desired, determine the stresses in the principal system for 
 this unit load and proceed as before by using Eq. (6A). 
 
 It may be well to mention in closing this subject that there is usually a wide latitude 
 n the choice of the redundant conditions. Thus, in Fig. ?A, one abutment might have 
 >een placed on rollers, thus making the principal system a simple truss on two supports^ 
 Then X a would have become the horizontal thrust instead of the stress in a member 
 
 ART. 9. PROF. MAXWELL'S THEOREM (1864) 
 
 This theorem, known as the law of reciprocal displacements, and previously men- 
 tioned in discussing Eqs. (8 D ), establishes the mutual relation between the elastic dis- 
 placements of a pair of points, or a pair of lines, whenever these displacements result 
 from simultaneous conditions of loading ;_ provided that the arrangement of the membe 
 remains unchanged and the supports are immovable. 
 
 For the sake of simplicity it is assumed that the frame is in a condition of no stress 
 and that there are no temperature changes. 
 
 Clapeyron's law then applies and the equation for actual work becomes 
 
 wherein the P's are concentrated loads, and the 3's are the deflections of the points on 
 which the loads act in the respective directions of these loads. 
 
 FIG. QA. 
 
 FIG. OB. 
 
 Each of the products \P m d m may now be regarded as a summation of work pro- 
 duced by some group of loads such that the work of 
 the group is exactly identical with the work rep- 
 resented by %Pmd m . Figs. QA, 9s, and 9c will 
 illustrate just what is meant by such groups of loads. 
 
 If in Fig. OA, two equal and opposite . forces 
 P m are applied at the points m l and m, then the 
 resulting o m represents the relative displacement 
 between these two points. We call this case 
 the loading of a pair of points corresponding 
 to the case illustrated in Fig. GA, where the loads P are each equal to unity. 
 
 FIG. 9c. 
 
28 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. IT 
 
 Fig. OB shows a pair of equal and opposite forces perpendicular to a line ik and 
 producing a couple of moment 1-Pm. This represents the loading of a line. The result- 
 ing d m represents the angular change in the line ik expressed in arc. This case may be 
 designated as the loading of a line m and, whenP = l, it becomes the unit loading of this 
 line. 
 
 Fig. 9c represents the loading of a pair of lines m^ and m. The d m of this group 
 of loads is the relative angular change between the two lines, or it is the change in the 
 angle 6, which results from the conventional loading. 
 
 The three groups of loadings are typical, and any particular load applied to a frame 
 may become one load of such a group, but not of two or more groups simultaneously. 
 
 For brevity, any such group of loads will be known as a loading and the corresponding 
 o will be the path or elastic displacement of the loading. 
 
 The several d's are always linear functions of the applied loads P and can be expressed 
 as follows: 
 
 d bm P m ; \ (9B> 
 
 O m ~ I>M r a ~r rnb* b ~r O mln L n 
 
 wherein the d's with the double subscripts are independent of the loads P, and for 
 example o am represents the special value of d a when P m = 1 and all the other loads P 
 are zero, all in accordance with the nomenclature employed in Art. 8. 
 
 Now apply loads P m to any frame, producing stresses S m in the several members, 
 and changes in their lengths Jl m =S m l/EF. 
 
 Likewise for a system of loads P n , producing stresses S n , and changes Al n = S n l/EF. 
 
 Let d' mn = the value of d m , for the point m, when certain loads P n only are active. 
 Also, let d' nm = the special value of d n for the point nj when a certain set of loads P m only 
 are active. Then note that d' mn = d mn when 2P n =l and o' nm = d nm when 2P TO =1. 
 
 The work equation for the loads P m , stresses S m and displacements d' mn and Al n due 
 the loads P n only, is according to Eq. (OE) 
 
 S I 
 
 2jPmO mn = ^o wz Jt n == Lo^-pp. 
 
 Similarly 
 
 S I 
 
 TiP n d' nm = 2}S n dl m ~ ^^n^W> 
 
 therefore, 
 
 (9c) 
 
 which is Betti's law (1872) and which extends Maxwell's law to the summation of all 
 members of a structure. 
 
 If instead of systems of loads P m and P n as above, only the unit loads P m = 1 and 
 P n = 1 are successively applied to the frame, then Eq. (9c) becomes 
 
 d mn = 8 nm , ............. (9D) 
 
 which is Maxwell's law (1864) and may be stated thus: 
 
ART. 10 THEOREMS, LAWS, AND FORMULA FOR FRAMED STRUCTURES 29 
 
 1. The relative displacement d mn of a pair of points TOI and TO, resulting from a 
 unit loading of another pair of points n\ and n, is equal to the relative displacement o nm 
 of the pair of points n\ and n, caused by a unit loading of the pair of points m\ and TO. 
 
 2. The same equality exists between the relative angular changes of two pairs of 
 lines, successively loaded with a unit loading. 
 
 3. The same equality also exists between the linear change in a pair of points result- 
 ing from a unit loading of a pair of lines, and the angular change (expressed in arc) of 
 the pair of lines, resulting from a unit loading of the pair of points. 
 
 The practical value of Maxwell's law when applied to redundant conditions was shown 
 in Art. 8. Its application to displacement influence lines is given in Chapter VIII. 
 
 ART. 10. THEOREMS RELATING TO WORK OF DEFORMATION 
 
 (a) Menabrea's law (1858), or theorem of least work. Given a statically indeterminate 
 framework in an initial condition of no stress for which the temperature is known and 
 remains constant; also, assuming that the supports are either rigidly fixed or permit 
 frictionless movements, such that 2.RJr = for the entire structure, then the changes 
 in the lengths of the members are expressed by the formula M = Sl/EF, and the work 
 equations for the several statically indeterminate quantities X a , Xf,, X c , etc., are, accord- 
 ing to Eqs. (SB) , 
 
 EF 
 
 etc., 
 
 the value S in which is given by Eq. (7A.) as 
 
 S = S S a X a SbXf, S C X C , etc. 
 The partial differentiation of S with respect to X a gives 
 
 likewise for the other X's, 
 
 (10B) 
 
 (lOc) 
 
 The actual work of deformation for the entire frame, including redundant members 
 as external forces, becomes by Clapeyron's law or Eq. (5A) 
 
 (10D) 
 
30 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. II 
 
 which when differentiated gives 
 
 _ 
 
 and by dividing through by ~dX a this becomes, after substitution of values from Eqs. (lOc) 
 
 - 
 
 - 
 
 EF EF ' 
 
 which by Eq. (10A) must equal zero. Hence, 
 
 , . .. . dA . 9 A , . 
 
 = and similarlv ^-^-=0 and ^^=0, .... (10E) 
 
 which proves that the redundant or indeterminate conditions, reduce the actual work of defor- 
 mation of the frame to a minimum. This is the theorem of least work. 
 
 (b) Castigliano's law (1879), or derivative of the work equation. This law deals 
 with the displacement of the point of application of a force. 
 
 Any external load P m acting on a framework will, by Clapeyron's law, produce the 
 actual applied work of deformation 
 
 A=$P m d m , ......... . . (10F) 
 
 when d m is the displacement of the point of application of P m in the direction of P m 
 and within range of proportionality, or within the elastic limit of the material. 
 
 Also, the actual internal work of deformation, for the entire frame, becomes, accord- 
 ing to Eq. (10D) 
 
 The partial differential derivative of A with respect to any certain external load 
 P m is 
 
 'S 2 l\ .^ s dS_ J_ 
 
 3P m EF' 
 
 wherein 3*S/3P OT is the derivative of the stress S in any member of a determinate or 
 indeterminate structure. 
 
 The general expression for S, from Eq. (lOs), by substituting for S its equivalent 
 value in terms of external loads becomes 
 
 , etc., +S m P m -S X a -S b X b -S c X c , etc., . . . (10j) 
 
 in which Si is the stress in the member S for PI =1, while S a is that stress when X a = l, 
 etc.; finally S is the stress in any particular member caused by the combined effects 
 of all P's and Z's. 
 
 The several values Si to S m are thus independent of the loads P and X. Also, the 
 loads P and X are independent of each other. Hence, S may be partially differentiated 
 with respect to P m or X and Eq. (10j) when so treated gives 
 
 --=S m and '' == ~ s > etc ........ ( 10K ) 
 
ART. 11 THEOREMS, LAWS, AND FORMULA FOR FRAMED STRUCTURES 31 
 Substituting this value in Eq. (lOn) then 
 
 but, according to Mohr's law, Eq. (GA), when abutment displacements are zero, then 
 
 (10M) 
 
 In Eqs. (IOL) and (10M) the summation extends to all the members, including the 
 redundant conditions, and hence 
 
 7)4 
 
 Ld m =-, ....... ........ (ION) 
 
 which, expressed in words, means that the path o m of a load P m is equal to the partial 
 differential derivitave of the actual work of the frame with respect to P m . 
 
 Castigliano's law thus expressed is equivalent to Mohr's work equation, that is, 
 leading to the identical result by a more circuitous process. 
 
 In similar manner the same law may be deduced for girders with solid webs. 
 
 Mohr's law thus offers the most direct method of finding any displacement which 
 may result from any specific cause. Castigliano's law will lead to the same conclusions 
 by a somewhat less direct method. 
 
 ART. 11. TEMPERATURE STRESSES FOLLOW MENABREA'S AND CASTIGLIANO'S 
 
 LAWS 
 
 When a structure is subjected to a uniform change in temperature t then, from 
 Eq. (4A), 
 
 This value of M when substituted into Eq. (10D) for Sl/EF, gives 
 
 ; ........ (llA) 
 
 and the differential of A with respect to S is 
 
 (llB) 
 
 Similarly when the temperature effect is introduced into Eqs. (10A) then for immovable 
 
 abutments as before 
 
 y c cf/ 
 
 2S^l==+8 a <&-Q ......... (He) 
 
32 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. II 
 
 Also, Eqs. (lOc) apply equally when temperature effects are included, hence by divid- 
 ing Eq. (lie) by 3A" a , and substituting values from Eqs. (lOc), then Eq. (11s) becomes 
 
 i 
 
 9 A 
 
 a ~ EF 
 But, by Eq. (lie) this expression is equal to zero, hence, as before 
 
 "dA 3A 
 
 o^r- = and similarly ~^- =0 etc., ...... (HE) 
 
 oX a oXb 
 
 which proves Menabrea's law applicable to the general case of stress from temperature 
 and applied loads, provided the abutments are rigid and immovable. 
 
 The introduction of the temperature factor SefcS/ from Eq. (!!A) into Eqs. (10o), 
 to (ION) will suffice to prove that Castigliano's law also applies to the general case includ- 
 ing temperature effects. It is not deemed necessary to repeat the transformations here. 
 
 ART. 12. STRESSES DUE TO ABUTMENT DISPLACEMENTS 
 
 Abutment displacements produce stresses which follow Menabrea's and Castigliano's 
 laws. 
 
 This is readily seen when it is considered that the supporting elements may always 
 be replaced by linked members which take up these displacements and undergo elastic 
 deformations. The immovable supports are then outside of these connecting links as 
 illustrated in Figs. ID, IE and IF, and the links are counted with the structural members. 
 If then the abutment displacements are defined as distortions Jr in these several 
 links, which are in every way equal to the former displacements, the resulting effect on 
 the structure remains unchanged. 
 
 Therefore, when the distortions Jr must be considered in the computation of the 
 X's and d's, it is only necessary to extend the work equations to include these links. It 
 is also clear that these links may have any desired sections, lengths or values of E, as may 
 be required for metal or masonry supports. 
 
 Thus, let r be the length of a supporting link; 
 R the load which the link carries; 
 Ar the change in length of this link; 
 F r the cross-section of this link; 
 e r the coefficient of expansion ; 
 t the change in temperature from the normal; 
 E r the modulus of elasticity for any material. 
 then from Eq. (4 A) 
 
 The general work Eq. (11 A) then becomes 
 
 V027 
 
ART. 12 THEOREMS, LAWS, AND FORMULA FOR FRAMED STRUCTURES 33 
 
 wherein Ar may be regarded as a constant and Menabrea's and Castigliano's laws again 
 apply. 
 
 For the case of one external indeterminate, R becomes X a and Ar becomes o a , then 
 from Eq. (12n) 
 
 By differentiating for 8 with respect to X a and imposing the condition for minimum 
 this becomes 
 
 s 3S 3 n9 v 
 
 :t + da ~ 
 
 Also, S=S S a X a and its differential is 
 
 Substituting the value from Eq. (12o) into Eq. (12c) and solving for X a then the 
 latter becomes 
 
 2S S a ~ + 2dlS a -3 a 
 Xa= ~' ........ (12E) 
 
 EF 
 
 which might be used as the expression for the horizontal thrust of a two-hinged arch 
 where X a is the redundant thrust. 
 
CHAPTER III 
 
 THEOREMS, LAWS, AND FORMULA FOR ISOTROPIC SOLIDS 
 
 ART. 13. GENERAL WORK EQUATIONS 
 
 In the previous chapters no consideration was given to solid web structures or other 
 isotropic solid bodies only in so far as was necessary in demonstrating the general laws 
 and theorems of framed structures. 
 
 While it is true that the foregoing discussion is generally applicable to all isotropic 
 solid bodies which are supported in any of the ways given in Chapter I, and stressed 
 Avithin the elastic limit by externally applied loads, yet the formula? previously given 
 will require some modifications to better adapt them to solid web and other massive 
 and homogeneous structures. Also, the introduction of shearing stress now enters as 
 a further complication. 
 
 As was previously mentioned, all structures which are purely isotropic an involve 
 only external redundancy. It is, therefore, desirable to take up this subject to the 
 
 extent, at least, of giving the special formu- 
 lae applicable to any solid in which the 
 physical properties of the material are pre- 
 sumably uniform in any and all directions. 
 Such bodies are called isotropic solids. 
 
 It will scarcely be necessary to prove 
 at length all of the theorems and laws pre- 
 viously given for frames, but a passing refer- 
 ence at the proper time will be deemed suf- 
 ficient proof of their general acceptance. 
 
 The elastic deformation resulting from 
 given stresses in an infinitesimally small 
 particle of a body is a certain and definitely 
 measurable quantity and is dependent only 
 on the magnitude of the given stresses and 
 the physical properties of the material. 
 
 If then Fig. 13A represents a small parallelepiped of mass, referred to axes X, Y 
 and Z, and having dimensions dx, dy and dz, coincident with these axes respectively, 
 then the stress acting at the corner m is determined in magnitude and direction by 
 nine components, viz., three normal stresses and six tangential stresses acting in the three 
 coordinate planes and positive in the directions of the arrows. 
 
 FIG. ISA. 
 
ART. 13 THEOREMS, LAWS, AND FORMULA FOR ISOTROPIC SOLIDS 35 
 
 Let f x , fy and f z be the unit normal stresses (tension or compression) in the direc- 
 tions X, Y and Z, respectively. Also, let - xy and r xz be the unit tangential stresses in 
 the YZ plane; r yx and r yz those in the XZ plane; and - zx and ~ zy those in the XY plane. 
 The first subscript indicating the normal stress to which they belong, and the second 
 subscript referring to the axis. 
 
 The two unit tangential stresses intersecting in any one point are equal, in each 
 case, otherwise the body would rotate about its center of gravity, which is the intersec- 
 tion of the three normal unit stresses. Also, because the three normal stresses and 
 four tangential stresses, ~, fz , r xz , r, y and T ZX , cannot have any moment about such a gravity 
 axis OZ', hence, the moment of the two remaining stresses -c yj . and r xy must equal zero. 
 But, the latter having equal lever arms must be equal to each other. 
 Therefore, 
 
 The resulting stress at the point m is then determined by the three normal stresses 
 f x ,fy,fz and the three tangential stresses r x , T U , and T Z in Eqs. (13A), and these in turn 
 determine the deformation of the parallelepiped. 
 
 Let Adx, M,y and Adz represent small elastic changes which the lengths dx, dy and 
 dz undergo, then the forces represented by the unit normal stresses / will produce the 
 following virtual work : 
 
 =f x dydzJdx+f y dzdxJdy+f z dxdyJdz = f x + 
 
 Also, let j- x , fy and f z be the tangents of small angular distortions of the 
 parallelepiped such that +f x is an increase in the angle between the Y and Z axes; f v 
 the change in the angle between the X and Z axes, etc. Then the virtual work of the 
 tangential forces is found thus : For the distortion of the YZ plane, the path is f x dy and 
 the force is r x dxdz, and since the vertical force r^jjan do no work in this direction then 
 the work of the tangential stress for the plane XZ becomes r x dxdz^r x dy, and similarly 
 for the other two planes, whence the total work becomes 
 
 x ...... (13c) 
 
 The expressions Adxjdx, 4dy/dy, and 4dz/dz in Eqs. (13B) are rates of elongation, 
 or coefficients, for which the values a x , a v and a z may be substituted. Then by making 
 dxdydz =dV, and combining Eqs. (13s) and (13c), the total internal virtual work is obtained. 
 Also, since by Clapeyron's law, this must be equal to the virtual work of the externally 
 applied forces, then the fundamental work equation for isotropic bodies becomes: 
 
 W = 
 
 This equation is applicable to any case of related displacements and elastic deforma- 
 tions o dr, a, and 7-, so long as these are small in comparison with the dimensions of the 
 structure. The external forces must include dead loads and all frictional resistances 
 which may be active at points of support. 
 
36. 
 
 KIXETIC THEORY OF ENGINEERING STRUCTURES CHAP. Ill 
 
 As was previously shown for indeterminate frames, so also the various elements 
 in Eq. (13D) may now be expressed MS linear functions of the applied loads and any 
 redundant conditions, thus: 
 
 R=R R a X a R b X b , etc. 
 
 J x = J xo J xa-A-a J xb-^-bj CtC. 
 fy = fyofyaX a ~fyb^b, etc. 
 fz =^fzo fiaX a fzbXb, etc. 
 ?x = T ro ~raX a T xb X. b , etc. 
 
 etc., etc. 
 
 Also, the virtual work of the reactions for each of the conventional loadings becomes, 
 (see Eqs. (?E)): 
 
 (13E) 
 
 xa f 
 
 (13P) 
 
 f 
 
 2R b Ar =J [f x b( 
 
 etc. 
 Then by writing Eq. (13D) for a load P a = l the following is obtained: 
 
 -'ZR a <lr. . . (13c) 
 
 from which problems of the kind described in Art. 6 may be solved. 
 
 Since the redundants X in Eqs. (13E) may be treated as independent variables, 
 their differentiation furnishes 
 
 _ f .- t 
 
 a ' 
 
 "dX a ' 3Y 
 
 which by substitution into Eqs. (13r) give 
 
 etc., and 
 
 ~ ~ r 
 
 etc. 
 
 ' ' (13H > 
 
 Similar expressions result for 2/S^Jr, etc. 
 
 The general work equation is now found by inserting for the actual distortions a 
 and f their values in terms of stresses and the moduli of elasticity for direct and tan- 
 gential stress. 
 
 The length dx subjected to the unit stress f x and a rise in temperature of t degrees 
 will be changed by an amount 
 
ART. 13 THEOREMS, LAWS, AND FORMULA FOR ISOTROPIC SOLIDS 37 
 
 The two other unit stresses/, and f z will diminish the effect on a x by an amount 
 
 /*+/. 
 
 mK ' 
 
 wherein m is the " Poisson number " (1829) which is given as 3.33 for structural steel 
 and 3 5 for hi-h steel. Professor F. E. Turneaure gives 0.1 to 0.125 for concrete, 
 quantity m is also denned as the ratio of lateral to longitudinal deformation an 
 determined by experiment. 
 
 Hence, a x and similarly a y and a,, also, T *> Y y and r* may be evaluated as follows. 
 
 m 
 
 and 
 
 and 
 
 (13J) 
 
 and 
 
 mE 
 
 wherein G is the modulus of tangential stress or shear and has the value G 2(jn + 1) ' 
 
 while m is the " Poisson number "just given. 
 
 Inserting the values given by Eqs. (18,) into Eq. (13D), the fo.lowmg equat.on 
 obtained for the actual work of an isotropic body : 
 
 . . (13K) 
 
 \lso by substititingthe values from Eqs. (13i) into Eq. (13n), reducing and integrat- 
 in , tPS^/AT., etc., the following simple form is obtamed by msertmg the 
 value from Eq. (13KJ, when temperature effect is negled 
 
 VR^^- and similarly S^Jr = ^, etc. - - (13L) 
 
 OA a 
 
 When the abutments are immovable and no temperature effects exist then 
 
 -0 and -0, etc., 
 
 MCT ab<)ve fed 
 P m , or redundant X m : ^ 
 
KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. Ill 
 
 The value of A in all the Eqs. (13L) to (13N) is that given by Eq. (13K). 
 
 \Yhen the abutments are immovable the last term of Eq. (13N), becomes zero and 
 Castigliano's law is again established. 
 
 Proceeding from Eq. (13o), and writing same for two related conditions of unit 
 loading, it is easily shown that Maxwell's law likewise applies here. 
 
 ART. 14. WORK OF DEFORMATION DUE TO SHEARING STRESS 
 
 For a beam of any section and any loading applied in the vertical plane YY, Figs. 
 
 and 14n, the resultant of the external forces on one side of any section A A may be 
 represented by a force R which may in turn be resolved into a normal force N and a 
 tangential force Q acting at the point of application of R on the section A A. 
 
 T,, 
 
 FIG. 14A. 
 
 Let M be the static moment of the normal force N about the Z axis and I z the moment 
 of inertia of the section about this same axis. 
 
 Then the unit stress/ at any point of the section is by Navier's law: 
 
 N My 
 f = J+~r (14.A) 
 
 * Z 
 
 If now this stress undergoes a small change df due to the differential change in R 
 for a neighboring section, distant dx from the first, then the shear on the area 2zidx 
 becomes 
 
 2T J * l dx=f(f+df)dF-ffdF=JdfdF. 
 
 where r x is the unit shear on the differential area d-F. 
 
 Assuming N constant in Eq. (HA) , and treating M as a variable, then by differentiation 
 
 (14c) 
 
 
AKT. 14 THEOREMS, LAWS, AND FORMULAE FOR ISOTROPIC SOLIDS 39 
 
 Also, since the shear is the differential of the bending moment, then 
 
 dM=Qdx (14D) 
 
 Combining Eqs. (14B), (14c) and (14o), then 
 
 CydMdF Qdx C , 
 
 1- x z-idx= I I ydF. 04E) 
 
 J L z l z J 
 
 and since J ydF is the static moment M s of the cross-section, this integration may be 
 considered performed to obtain 
 
 Return now to Fig. 14s, take any section pq perpendicular to the ?/ axis and let Q 
 produce a unit shear r in any point i of this section. The line of this shear intersects the 
 ?/ axis in a point H, which is determined by the tangent pH. This is a pure assumption, 
 but from the nature of the case it is the most rational assumption to make. 
 
 The shear r may now be resolved into components T U and T Z , while r x , for this sectional 
 plane, must be zero. However, r y now has the same value as previously found for - x , 
 and hence 
 
 *-!&->' <-> 
 
 Again, from Fig. MB, tan0 = = - from which r g = -r y . But tan d = , therefore, 
 
 O7 i f) P P 
 
 / * \ j. n ft A \ 
 
 T a = Tj,( itan (14H) 
 
 \ z i/ 
 
 For 0=0, T 2 =0, which is the case for any surface point, the tangent to which is parallel 
 to the y axis. 
 
 From Eqs. (14r) and (14o) it follows that for a given loading and section, the shear- 
 ing stresses depend only on M 8 and attain a maximum when the section coincides with 
 a gravity axis parallel to the neutral axis. In practical cases it is usually sufficient to 
 consider the shearing stress of a section as extending only over a unit length of the 
 beam as given by Eq. (14F). The actual work produced by shearing stress alone is 
 then found from Eq. (13K) as 
 
 By substituting the value of r, from Eq. (14n), into Eq. (14i), the latter becomes 
 
 /f 
 
 (14K) 
 
40 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. Ill 
 
 and by substituting for Ty its value from Eq. (14c), then Eq. (14K) may be solved for any 
 special section. This gives in general for any section 
 
 A=> 
 
 and 
 
 and 
 
 
 (14L) 
 
 In Eqs. (14L) ft is an involved function of the shape of the section, called by German 
 authors the " distribution number " of the section, or " coefficient of shearing strain." 
 
 This number @ is 1.2 for any solid square or rectangular section, and 1.111 for a solid 
 circular or elliptic section. Other values will be given after illustrating the computation 
 of this number by the use of Eq. (14L). 
 
 bh 3 
 For a rectangle of height h and breadth b = 2zi, then 1,. = -^; F = bh; and dF = bdy, 
 
 \m 
 
 hence, 
 
 Q 
 
 
 and 
 
 For any I section, Professor Mehrtens finds the general formula for /? as follows: 
 {3 = M\T 2 dF = j2\--(8b 3 l2b 2 t t 3 + 6bt 2 )+(-^-\ -+t' 2 a) , . . (14ivi) 
 
 wherein i= - and the special dimensions here used are shown in Fig. 14c. 
 
 Professor L. von Tetmajer, in his " Elastizi- 
 taets und Festigkeitslehre," 1905, p. 49, gives a 
 table of values for /? for I beams ranging from 
 3i"-4 Ibs. to 19"-95 Ibs., 'for which /3=2.39 
 to 2.03 respectively. Also for riveted girders 
 made of f" webs, 4 L's 3&"X3&"Xf" each, 
 and 2 plates 8f" xf" on each chord. Then for 
 
 _!.. __, , 
 
 -Y 
 
 2.71 
 
 23$" 
 
 2.49 
 
 27 A" 
 2.35 
 
 The value is independent of the unit of length and is, therefore, the same for 
 metric and U. S. measures. 
 
ART. 1.) THEOREMS, LAWS, AND FORMULAE FOR ISOTROPIC SOLIDS 41 
 
 ART. 16. WORK OF DEFORMATION FOR ANY INDETERMINATE STRAIGHT 
 
 BEAM 
 
 Direct ^tress. For axial or direct stress only, Eq. (13K) gives 
 
 (ISA) 
 Also, by observing that a x =f x /E, then Eq. (13n) becomes for direct stress alone, 
 
 - ........ < 15B > 
 
 Now ( f x dF=N, the normal direct stress, and, as dV =dFdx, then from Eq. (15A) 
 
 C&HV- C N f* dx - C N f* F dx_ CN*dx 
 J 2E a[ 'I ~2E~~J 2EF J 2EF ' 
 
 and similarly from Eq. (15B) 
 
 ff x df x dV_ CN-df x dx_^ (N-df x Fdx_ fNZNdx ( 
 
 J E-dX a J EdX a J EF3X a J EFdXa 
 
 The temperature effects in Eqs. (15A) and (15B) will now be determined on the 
 supposition that the effect is not uniform, but as shown on Fig. 14A, where 
 
 fo=the change in temperature from normal at the gravity axis of the section; 
 J=the difference in temperature of the two extreme fibers; 
 
 h = height of the section; 
 
 t/=any ordinate; 
 
 t=the temperature above normal, of any point of the section then 
 
 (15B) 
 
 Substituting this value for t into the temperature element of Eq. (15A), the latter 
 becomes 
 
 But ff x dF=N and r/ x ?/dF=M=moment of resistance of the section, therefore, 
 
 dx ......... (15F) 
 
 By differentiating Eq. (15?), and dividing through by 3X a , the value of the tem- 
 perature element for Eq. (15u), is obtained thus: 
 
 n . . 
 
 (15G) 
 
42 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP, in 
 
 The bending moment. From Eq. (HA), f x =My/I g and I z ==fy 2 dF. With these 
 values and neglecting the temperature element, Eq. (!OA) becomes 
 
 By differentiating Eq. (Ion) and dividing through by "dX a then 
 
 'M 
 
 The general work equation, including all effects, may now be written out by collect- 
 ing the several Eqs. (14L), (15A), (15c), (15n), and (lor) into the following, represent- 
 ing the total actual work for any isotropic body by 
 
 r r M 
 
 Similarly the following differential expression of Eq. (!OK) is obtained from Eqs. 
 (14L), (15s), (15D), (15.i) and (15o) or directly by differentiating Eq. (!OK) and divid- 
 ing through by ~dX a , thus: 
 
 - (loL) 
 
 Equation (15L) being true for any force X a is, of course, true for a force P m , hence 
 by substitution of the latter value and then inserting the value 3A/3P TO . thus obtained 
 from Eq. (15L) into Eq. (13N) the following equation for elastic deflection is obtained: 
 
 c Q ^Q , C t ^N 
 
 j GFZP^ dx+ J t 3p- 
 
 3M 
 
 The last three equations (15K), (15i.) and (15M) are the three fundamentals from which 
 all cases of redundancy for isotropic bodies can be solved. There are always as many 
 of these equations as there are redundant conditions X. 
 
 In each of these the first term expresses the effect due to direct or normal stress; the 
 second term that due to pure bending; the third term that due to shearing stress; the fourth 
 term gives the effect due to a uniform rise in temperature to; and the fifth term that due to a 
 difference At in the temperature of opposite extreme fibers. 
 
ART. 16 THEOREMS, LAWS, AND FORMULA FOR ISOTROPIC SOLIDS 43 
 
 The object in presenting these rather long fundamental equations is not to make 
 the subject appear complicated, but rather with a view to showing once for all 
 combined effect from all causes, thus permitting the easy choice in combining any effects 
 or in omitting such as may seem negligible in any specific problem. 
 
 Equation (15K) may also be derived from Eq. (5n) for virtual work and offers very 
 useful applications. 
 
 Thus assuming a column subjected to a thrust N ttt and a bending moment M m 
 the virtual work on the column from Eq. (OH) becomes 
 
 where A r rt and M a are due to the usual conventional loadings. But 
 
 A^te nd . MJx 
 
 lioncG 
 
 W . C l N m N a dx , r lM rnM a dx 
 
 W = \-d m 
 
 EF 
 
 r l M m M a dx 
 J El 
 
 When the virtual work becomes the actual work A then Eq. (15N) becomes identical 
 with Eq. (15K) term for term. 
 
 ART. 16. WORK OF DEFORMATION DUE TO DYNAMIC IMPACT 
 
 Problems involving the deflection or strength of a structure subjected to impact, 
 are frequently met with, and their treatment is here considered as properly belonging 
 to the subject of the present chapter. 
 Let LO = weight of a moving body; 
 H = height of a fall; 
 v= velocity at instant of impact; 
 g = acceleration due to gravity ; 
 a=any elastic displacement produced by the moving body in some structure. 
 
 Then the work expended by the moving mass is represented, either in terms of 
 velocity or height of fall as follows: 
 
 wherein v*/2g represents the velocity height or height through which a body falls in 
 acquiring a velocity v. When the body moves with a velocity v along a horizontal path, 
 thenv 2 /2g will be the height to which the body would raise itself in order to expend its 
 energy and come to a state of rest. 
 
44 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. Ill 
 
 The work thus produced by kinetic energy, when it is instantly taken up by a 
 quiescent body or structure, is twice as great as the work of the same moving body 
 gradually applied, and hence produces twice the internal work in the body struck as 
 would result from an equivalent static load. 
 
 Problem 1. The weight w falls on top of a column, stressing the column normally, 
 what sectional area is required in order that the unit compressive stress does not exceed /? 
 
 From Eq. (!OK) the work due to direct stress is given by the first term, where N 
 is the total static load, and to balance the dynamic energy, twice this amount is taken; 
 hence 
 
 (16B) 
 
 and since o = ~ this gives for F, 
 
 m 
 
 Problem 2. A weight w falls from some height striking a beam resting on two sup- 
 ports. The weight strikes the center of the beam with a velocity v; what will be the 
 stress / in the extreme fiber for a given beam section? Shear and bending resistances 
 are to be considered. 
 
 From Eq. (!OK) the actual internal work is 
 
 Let P=an equivalent static load producing the same stress / in the given beam. 
 Then, for any point of the beam of depth h and span I, 
 
 P My Plh 8/7 
 
 -x; also /=--* = __ or P = -^- ..... (16s) 
 
 Hence 
 
 i 
 
 , 
 
 + JM 2 dx_ P 2 rt 2 _ P2/3 _2f 2 ll 
 
 i_ ~2ET~4EI Jo X WEI~3Eh 2 
 
 (16r) 
 
 p 
 
 Also, for Q = the second term of Eq. (16D) becomes 
 
 =dx = _ 
 2GF 4GFjo SGF GFlh 2 ....... 
 
 2 
 
 The sum of Eqs. (16p) and (16c) gives A, according to Eq. (16D), and this must 
 equal a}V*/2g. Therefore, 
 
AKT. 16 THEOREMS, LAWS, AND FORMULA FOR ISOTROPIC SOLIDS 45 
 
 Her e G=modulus of shear = 1 =0.385 E for structural steel (see Eqs. (13j). 
 
 The coefficient of shearing strain /? is given by Eqs. (14L) and (14M). 
 
 By applying Eq. (16n) to a steel beam 2.5X2.5 inches by 78 inches long, it is found 
 that when shear is included, the stress / is about f per cent smaller than when this term 
 is omitted, showing that the internal resistance due to shear is very small and usually 
 negligible. ' This is also true when computing deflections. 
 
 ' In the same manner all problems involving impact may be solved. 
 
 Suppose a ship weighing w tons, and moving with a velocity of v feet per second 
 were to collide with a fixed structure, then the work which the ship is capable of expend- 
 in- is represented by a>v*/2g ft.tons. This work may be expended in injuring the 
 structure or the ship itself, a condition depending on the relative strength of the two 
 bodies, which must be ascertained form the design and construction of each. 
 
CHAPTER IV 
 INFLUENCE LINES AND AREAS FOR STATICALLY DETERMINATE STRUCTURES 
 
 ART. 17. INTRODUCTORY 
 
 Professors MOHR and WINKLER, in 1868, published simultaneously the first treatises 
 on influence lines describing, at that early date, practically all the uses and applications 
 of these lines known at the present time. Professor J. Weyrauch, in 1873, introduces the 
 name influence line not used by Mohr and Winkler in their earlier work. Professor Mohr, 
 in a series of articles published from 1870 to 1877, was the first to apply influence lines 
 to deflections and to redundant conditions. 
 
 Professor Geo. F. Swain, in 1887, gave the first treatise on the subject in English. 
 
 An influence line is the graphic representation of some particular effect produced, 
 at a certain point of a structure, by a single moving load occupying, successively, all possible 
 positions over the entire span. The effect may be the shear, the bending moment or 
 the deflection at any certain point of the structure; it may also represent any reaction 
 force or the stress in any member. The single moving load is usually taken equal to 
 unity, though in certain special cases it may be desirable to use any load P. 
 
 An influence line represents a certain effect for a certain point or member of a 
 structure and for any position of a moving load, while a shear or moment diagram 
 represents effects due to a single position of the load or loads for all points of a structure. 
 
 The ordinate to any influence line is thus an influence number or factor, usually 
 designated by TJ when stresses are dealt with and sometimes by o when deflections are 
 under consideration. 
 
 As a matter of convention, all positive influence line ordinates will be laid off down- 
 ward from the axis of abscissae. 
 
 A load point is any particular one of the many possible positions of the moving 
 load. 
 
 A summation influence line is one which gives the total effect at some particular point 
 due to a train of concentrated loads. The actual loads are here employed and each 
 influence ordinate is made to represent the summation of influences of the same kind 
 for a certain position of the train of loads. Usually this position is taken so that the first 
 load is over the point for which the influence line is constructed. 
 
 An influence area is the area included between the influence line, the axis of abseissce 
 and the two end ordinates. 
 
 The maximum effect is always produced when the load point coincides with the 
 maximum ordinate of the influence area. Hence influence lines are eminently suited 
 
 46 
 
ART . 17 INFLUENCE LINES AND AREAS 47 
 
 to the solution of all problems relating to position of moving loads for maximum and 
 minimum effects and stresses. 
 
 The load divide is such a load point, the ordinates to either side of which have 
 opposite signs. Hence, the influence ordinate at a load divide passes through zero and 
 the influence line intersects the axis of abscissae at such load divide. 
 
 It is assumed that if a unit load produces some effect 77 at a certain point of a given 
 structure, then a load P will produce the effect PTJ at this same point, so long as the material 
 is not stressed beyond the elastic limit. This follows from the law of proportionality 
 stated at the end of Art. 4. 
 
 Hence, the total effect Z produced by a system of moving loads, PI, P 2 , etc., will 
 be the sum of the effects Pi] of all the loads and the maximum and minimum value of Z 
 will be determined by the position of the moving loads. 
 
 The total effect is thus expressed by the following equation when the case of loading 
 is simultaneous and within working limits : 
 
 (17A) 
 
 Therefore, having given the influence line for a certain effect on some structure, then 
 the total effect, due to any system of loading, is easily found by a summation of the prod- 
 ucts of loads into corresponding influence ordinates or numbers for any desired positions 
 
 of the loads. 
 
 For a uniform moving load p per foot of length, Eq. (!?A) becomes 
 
 wherein the integral represents the area of the influence polygon between the end ordi- 
 nates of the uniform load. 
 
 The equation of any influence line may be written out by expressing the desired 
 function for a particular point in question in terms of a moving load unity acting at any 
 variable distance x from one end of the structure taken as origin. 
 
 Since influence lines represent all possible effects it is readily seen that maximum 
 and minimum stresses may be found from the same lines. Therefore, only one half of 
 a symmetric structure requires analysis. The left half is usually treated, as a matter 
 
 of conventional uniformity. 
 
 Direct and indirect loading. In the above it was assumed that the loads were directly 
 applied to the beam, which is rather the excep- 
 tion. Usually they are taken up by the floor 
 system and then transferred to the panel points 
 as load concentrations. The former case of 
 loading will be known as direct loading and the 
 latter as indirect loading. 
 
 The influence line between two successive panel 
 
 points is always a straight line, regardless of the riG 1?A 
 
 system of loading or the particular influence. 
 
 Let Fig. 17A represent two successive floor beams of any truss and the loadP = 
 
IS KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. IV 
 
 is transferred to the points a and b by the floor stringer ab. The load effects PI and 
 Po, which are carried to the panel points a and b, are 
 
 - and 
 
 Now the total influence of P l and P 2 , on the structure as a whole, must be exactly 
 equal to that produced by the resultant P. 
 
 But the influence line a'b' '', extending over the panel ab, must give, by Eq. (!?A), 
 
 , 17 x 
 
 from which 
 
 fd-: 
 
 For any influence line, >?i, r) 2 and d are constants, hence Eq. (17c) represents a 
 straight line, which was to be proven. 
 
 ART. 18. INFLUENCE LINES FOR DIRECT LOADING 
 
 The several influence lines for reactions, shears, moments and deflections for a beam 
 with direct loading will now be given. They will always be known by the names indicated 
 in Fig. ISA. Thus the A line is the influence line for the reaction A. 
 
 The influence line for a reaction is of the first importance, because the other lines 
 are generally derived from the reaction influence line. This is really seen from the 
 circumstance that for the unloaded portion of a beam or truss, the shear is always 
 equal to the end reaction and the moment is equal to this reaction into the distance from 
 the section in question to the end of the beam. 
 
 Hence, in drawing influence lines it is always best to consider the unloaded portion 
 of the span to the right or left of the section as the case may be, beca-use the only external 
 force on that side of the section will then be the reaction. 
 
 In the following the moving load will always be assumed as coming on the span from 
 the right end and the effect produced on the left half of the span only, is considered. 
 
 (a) Reaction influence lines A and B. Using the dimensions indicated in Fig. ISA, 
 the two reactions become 
 
 Px' Px 
 
 A= f- and B= ~T .......... 
 
 For x' and x variable, both expressions represent equations of straight lines which 
 are easily plotted. When P = l and x' =1 then A=l and for x' =0, A =0. Hence, the 
 A line is drawn by laying off a distance unity down from A and joining this point with B. 
 The reaction A for any load P acting at the load point m is then A m =Prj m . 
 
 In like manner the B line is found, and the corresponding reaction B m for a load 
 P at m becomes B m =Pj)' m , and i? m + fy' = l for every point of the span. 
 
ART. IS 
 
 INFLUENCE LINES AND AREAS 
 
 (b) Shear influence lines. The shear Q n for a certain point n to the left of the load 
 point m is always equal to the reaction A. But for a load point to the left of n (not 
 shown) the shear is Q n = A P = B, because P =A +B. 
 
 The influence line for Q n is thus derived from the influence lines for A and B as 
 indicated in Fig. ISA, and consists of the polygon An'n"R. 
 
 The point n becomes the load divide for shears at n and hence there will always be 
 a maximum and a minimum value for Q n depending on whether the positive or the 
 negative influence area is fully loaded. The shear due 
 to any single load P must change sign in passing the 
 point n. 
 
 (c) Moment influence lines. The moment for 
 the point n is now found when the moving load is 
 to the right or left of the section at n. Then for 
 
 P.I 
 
 m 
 
 x>a, M n =Aa; and for 
 
 M n =B(la). 
 
 Thus the moment influence line is also derived 
 from the reaction lines because the ordinates of the 
 latter, when multiplied by a or (I a) as the case may 
 be, give the ordinates to the moment line. Hence 
 the bounding lines of the moment influence line are 
 easily found. 
 
 Since A and B are both unity, then the ordinate 
 A A' --=a and the ordinate BB'=la and the lines 
 AB' and A'B inclose the required moment influence 
 line. Also, the ordinate at n is the ordinate of the 
 intersection n' between the two bounding lines. Hence, 
 if BE' should fall off the drawing, then the line ~AB' 
 may be drawn from A to n' without finding B' '. 
 
 In either case the moment influence line is then 
 the polygon An'B with all ordinates positive so long 
 as the point n is not outside the span, a case which 
 will be given later. The maximum ordinate is always 
 under n and has the value l-a(l a)/l, which offers 
 
 still another construction for this influence line. The middle ordinate of the AB' 
 \me=(l-a) which furnishes a convenient construction for this line when B' falls off 
 the drawing. 
 
 It is clearly seen that if a single load is placed at n over the maximum ordinate, 
 then a maximum moment is produced. 
 
 (d) Deflection influence lines. The deflection for a point n, produced by a single 
 load at any point m, is given in terms of the moment of inertia of the beam section and 
 the modulus of elasticity, as follows: 
 
 FIG. ISA 
 
 QEIl 
 
 -* 
 
 a*) P. 
 
 (18s) 
 
50 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. IV 
 
 This is a cubic equation and when plotted for all values of x and making P = l, the 
 influence line for deflections is obtained. 
 
 In general, any deflection polygon, drawn for a load unity acting on a fixed point 
 n of any structure, is the deflection influence line for the point n of that structure. This 
 follows from Maxwell's law. 
 
 As the subject of deflection influence lines is fully treated in Chapter VIII, no 
 further consideration is given to it now. 
 
 B 
 
 ART. 19. INFLUENCE LINES FOR INDIRECT LOADING 
 
 In the present case the loads are applied to a stringer and floor-beam system and are 
 thus transferred to the beam at certain panel points as load concentrations. The 
 
 beam is then indirectly loaded and since an influence 
 line was shown to be a straight line between panel 
 points the present case will require some slight 
 modification of the influence lines just found for 
 direct loading. See Fig. 19A. 
 
 However, all that was said for cases of direct 
 loading applies here and the modifications made 
 necessary by the indirect loading occur only in the 
 panel wherein the point n is located. As before, n is 
 the point for which the influence line is drawn and m 
 is any one of the possible load points for the moving 
 loadP = l. 
 
 (a) Reaction influence lines remain the same 
 whether the loading is direct or indirect, since the point 
 for which the influence is sought is alwavs at A or B, 
 
 - 1 / 
 
 which are also panel points. 
 
 Furthermore, the reaction influence lines being 
 straight over the length of the span will always be 
 straight between successive panel points. 
 
 (b) Shear influence lines. The Q line, outside of 
 the panel ce and containing the point n, remains the 
 same as for direct loading. But since the influence 
 line within a panel must be a straight line, therefore, 
 the points c' and e' must determine the influence line 
 for the panel ce regardless of the location of the 
 
 point n so long as this point is within the panel ce. 
 
 Hence, the polygon Ac'e'B is the shear influence line for the panel ~ce and the point 
 i is the load divide for this panel. The limiting values for shear are thus the same for 
 any point n of the same panel. 
 
 (c) Moment influence lines. Here again the influence line within the panel ~ce is 
 all that requires modification in the case of indirect loading. 
 
 FIG. 19x. 
 
ART. 20 
 
 INFLUENCE LINES AND AREAS 
 
 51 
 
 Hence for the same reasons just given to determine the modified Q line, the M line 
 for the point n is now the polygon Ac'e'B, instead of the triangle An'B for direct loading. 
 The point n is the center of moments. 
 
 ART. 20. THE LOAD DIVIDE FOR A TRUSS 
 
 In constructing influence lines for truss web members it frequently happens that 
 one or the other end ordinate falls outside the limits of the drawing. 
 
 There is a very simple way of locating the load divide i and thus facilitating the 
 construction of any influence line. The method is given first and the proof follows: 
 
 Let Fig. 20A represent a truss arranged for bottom chord loading, but no loads 
 are shown. Required to find the load divide i for the panel fg necessary to determine 
 the stress in the diagonal eg cut by the section tt. 
 
 FIG. 20A. 
 
 Construction. Prolong the unloaded chord member ek in the panel fg until it inter- 
 sects the verticals through the end reactions in a and 6. Then the intersection i of 
 the two lines af and bg will be the required load divide. 
 
 Proof. Suppose the unit moving load is now located in the vertical through i. 
 Then if i is the desired load divide, the unit load for the load point ii will produce zero 
 stress in the member eg and the influence line ordinate for the point i must be zero. 
 
 Let F and G be the panel concentrations in the points / and g due to the load P = 1 
 at i. 
 
 Then the polygon abfg may be regarded as the equilibrium polygon for the forces 
 A, B, F and G. Also, the resultant R of all forces on one side of the section tt must pass 
 through the intersection n of the two included sides of the equilibrium polygon. 
 
 But, the point n is the intersection of the two chords fg and ek, which point is also 
 the center of moments for the diagonal eg. 
 
 Hence the moment of this resultant R about n must be zero and cannot produce stress 
 in the member eg, thus proving that the load P = 1 is actually located in the load divide. 
 
 When the top chord is the loaded chord, a similar construction furnishes the point 
 i', as the load divide for the member eg, according to the same proof above given. 
 
 This construction applies to any structure even when one or both chords are straight. 
 
 When the center of moments n falls inside the span then there is no load divide and 
 all loads on the entire span will produce the same kind of stress in any particular web 
 member. 
 
52 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. IV 
 
 ART. 21. STRESS INFLUENCE LINES FOR TRUSS MEMBERS 
 
 According to Professor Ritter's well-known method of sections given in Art. 58, 
 the stress in any member of a determinate truss may be expressed thus : 
 
 (21 A) 
 
 where M is the moment of all the external forces, including the reactions, on one side 
 of the section and r is the lever arm, of the member in question, measured from a center 
 of moments determined by the intersection of the two other members cut by the section. 
 
 Hence, the influence line for the stress in an}^ member is the influence line for the 
 corresponding moment for this member, the ordinates of which are divided by r. 
 
 For any moment influence line, the ordinate at any point represents the moment due 
 to a unit load at that point and, therefore, from Eq. (2lA), i)=M=rS. 
 
 But the end ordinates for a moment influence line were formerly found to be a and 
 a', therefore, the end ordinates for a stress influence line must be 1/r times as great, 
 whence 
 
 o' 
 
 AA'=-, and BB"=- (21u) 
 
 r' r 
 
 Now l-a/r=$ =the stress in a member S due to a reaction unity at A, when the 
 load producing this reaction is to the right of the section through S. 
 
 Also I a' /r =${,= the stress in this same member S due to a reaction unity at B 
 when the load producing this reaction is to the left of the section. 
 
 Hence, the end ordinates for any stress influence line are equal to the unit stresses 
 S a and Sb produced in the member by reactions unity at A and B respectively. 
 
 Since these stresses are readily determined either by Ritter's method or by a Max- 
 well diagram, all stress influence lines for the members in any determinate truss are 
 easily found. 
 
 In Fig. 2lA, the stress influence lines are drawn for a general case of truss design. 
 The top chord is the loaded chord and the influence lines are drawn for the three mem- 
 bers L, U and D of the panel cut by the section it. 
 
 te =the stress in the member L for a reaction unity at A. This is evaluated by 
 Ritter's method in terms of the lever arms ai and r t . 
 
 Sib=ihe stress in the member L for a reaction unity at B and is expressed in terms 
 of the lever arms a'i and r t . 
 
 The stress influence line for the bottom chord L, is drawn in accordance with the 
 method illustrated in Fig. 19A, observing, however, that the end ordinates now become 
 the stresses S ta and Sn, respectively. 
 
 The center of moments for this member being at E, which is vertically over E', 
 the stress influence line for the member L becomes a triangle AE'B, determined by the 
 end ordinates A A' = to and BB' =#. 
 
 The influence line for the top chord, J7 is drawn in exactly the same manner, but in 
 this instance the resulting triangle AF'B does not offer a straight line over the panel 
 
ART. 21 
 
 INFLUENCE LINES AND AREAS 
 
 53 
 
 Here F is 
 
 PJG and hence the line E'G' must be drawn to complete the influence line, 
 the center of moments for the member U. 
 
 The influence line for the diagonal D is similarly drawn and the load divide %' ', found 
 in the lower diagram, is seen to coincide with the point i found on the truss diagram 
 by the method previously given. Also the intersection 0', between the limiting rays 
 
 is verticallv under the center of moments for the 
 
 AE' and G'B of the D 
 member D. 
 
 line. 
 
 ISTRESS INFLUENCE LINE FOR L. 
 
 STRESS |NFJLUE!NCE LINE FOR. u." 
 
 ' ' " 
 
 STRESS INKLUEJNCE LINE FOR o 
 
 It should be observed that in all three influence lines, Fig. 21 A, the limiting rays intersect 
 on the vertical through the center of moments f or the particular member. This then serves 
 as a check on the diagram. 
 
 Regarding the sign of the influence area the following rule should be observed: 
 When the center of moments is located between the supports A and B then all ordinates of 
 that particular stress influence line will be of the same sign; when the center of moments 
 is off the span, then there exists a load divide and there will be positive and negative influence 
 areas giving rise to stresses of opposite signs. 
 
54 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP, iv 
 
 The criterion for position of a moving load for maximum and minimum stresses in 
 any member is thus clearly shown by influence lines and the maximum effect of any load 
 is always produced when that load is situated over the maximum influence ordinate. 
 This subject is more fully treated in another article.. 
 
 It is sometimes necessary to construct influence lines without the divisor r, and 
 dividing the final stress thus found by r. Either method is employed as circumstances 
 may demand, though it is always better to avoid the divisor and follow the method given 
 in Fig. 21 A, unless there is some very good reason for doing otherwise. 
 
 Attention is also called to the fact that moment and shear influence lines are not 
 required when the stress influence lines are used. The former were given to render the 
 treatment complete and may be employed to find stresses, but it is usually preferable 
 to draw the stress influence lines directly. The final result of an analysis will be more 
 satisfactory when this is done. 
 
 Influence lines are seldom used for finding dead load stresses, though such a procedure 
 may sometimes be warranted, and then the following points must be observed : For 
 direct loading and for beams, it makes no difference whether the influence lines be 
 used for dead or live loads, but in dealing with framed structures an error might be 
 committed because then the influence lines are always drawn for a certain loaded 
 chord while the dead loads act along both chords. 
 
 This is easily remedied by drawing the influence lines for both chords loaded and 
 applying the dead loads to the proper lines. These influence lines are identical except 
 in the panel containing the member in question. Hence by observing this circum- 
 stance, dead load stresses may be found from the same influence lines without difficulty. 
 
 ART. 22. REACTION SUMMATION INFLUENCE LINES 
 
 Most summation influence lines become rather complicated and, therefore, their 
 use is practically restricted to a few cases for which the construction is simple. 
 
 In general, any ordinate to a summation influence line is expressed by Eq. (17A), as 
 
 and it is readily seen that when P and ij are both variables, such a line would ordinarily 
 require much labor for its determination. 
 
 However, the summation influence line for an end reaction of a simple truss is 
 very easily constructed and serves a most valuable purpose in finding the end shears 
 for a system of moving loads. This influence line will be called simply the sum A line 
 to distinguish it from the ordinary A line in Figs. ISA and 19A. 
 
 The general usefulness, of the sum A line originated by Professor Winkler, will be shown 
 later; suffice it to say here that it affords a ready means of finding stresses in the web 
 members of any truss, because, as previously shown, the shears and moments for any 
 point of a truss are easily found when the end reaction for the particular loading is known. 
 
 The sum A line for a system of concentrated wheel loads will now be demonstrated 
 using but five loads for simplicity, though the method applies to any number of loads. 
 
ART. 22 
 
 INFLUENCE LINES AND AREAS 
 
 55 
 
 Given the train of moving loads PI, P 2 , etc., in kips of 1000 Ibs., and spaced as 
 shown in Fig. 22A, to construct the sum A line for a span of 25 ft. 
 
 Since a reaction influence line is the same both for direct and indirect loading, the 
 sum A line is independent of the number and location of panel points. 
 
 For standard position of loads on a truss, the train is always assumed as coming 
 on the 'span from the right end and moving toward the left, but the above loads are 
 placed in exactly the reverse order, with the first load PI over the support B. The 
 reason for this will appear later. 
 
 The loads are applied consecutively from PI up on the vertical through A, using 
 any convenient scale. 
 
 r 
 
 POSITJON OF LOADS roft CONST *GcT ION or SUM A unr 
 
 B 
 
 FIG. 22A. 
 
 The several rays drawn from B to the respective load points c, d, e, etc., form 
 a force polygon with pole distance H =1. The sum A line is then drawn as the equilibrium 
 polygon for this force polygon and the loads P, by making 1 -2 \\Bd, 2 -3 \\Bc, 3 -4 \\Bf, 
 etc. 
 
 It will now be shown that the equilibrium polygon so obtained is really the sum- 
 mation influence line for the end reaction A . 
 
 Referring again to Fig. 22A, the line Be is easily recognized as the influence line 
 for A due to a moving load PI. Alsp^the influence line for A when the moving load is 
 P 2 , may be represented by the line Bd when Be is the axis of the absciss. Similarly 
 ~Be represents the A line for P 3 , and so on for any other loads. The summation of 
 all odinrates should then represent the desired summation influence line. 
 
 For the train coming on the span from right to left, the ordinate TJ, under the 
 forward load PI, always represents the reaction A. Thus when the train has advanced 
 
56 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. IV 
 
 to the point n, Fig. 22A, the ordinate v? represents the sum of the ordinates t, and 
 T}=A n =U+t 3 +t 2 +ti\vhkb. is readily seen by inspection. The ordinates t are the con- 
 tributions to A from the several loads on the span. It is now seen why the train 
 was first placed in reverse position for the construction of the equilibrium 
 polygon. 
 
 Hence the sum A line for direct or indirect loading is an equilibrium polygon drawn 
 for a pole distance equal to the length of span and any system of train loads placed in 
 reverse order on the span with the forward load at B. 
 
 The reaction A for any position of loading is then the ordinate to the sum A line 
 measured under the forward load when the train approaches from right to left. The 
 greatest possible reaction A will always be the end ordinate A5, Fig. 22A. 
 
 When extraordinary accuracy is required the end ordinate A 5 can be readily checked 
 by computation. 
 
 The sum A line for uniformly distributed loads may be constructed in precisely the 
 same manner as above illustrated for concentrated loads, merely by laying off the load 
 line Ag to represent the total load on the span =pl and dividing this into some con- 
 venient number of equal parts. The greater this number the more accurate will be the 
 result, and the polygon finally becomes a parabola. 
 
 The application of the sum A line, to finding the shear at any point of a beam or truss, 
 will now be presented. 
 
 The shear at any point n of a beam, for a case of direct loading, is equal to the 
 reaction A minus the loads between A and n. Hence, the sum A line gives a complete 
 solution for this case. The subtraction of any loads to the left of n can be performed 
 graphically on the diagram, Fig. 22 A, by taking^ off the proper ordinate less the loads 
 between A and n. 
 
 For indirect loading, the shear at n is equal to the end reaction A, provided the 
 forward load is exactly over the panel point n. But when the load extends over into 
 the panel, then the shear is equal to the end reaction A, minus the panel reaction a at 
 the left hand pin point of the loaded panel. See Fig. 22ra. 
 
 The values of o for all possible positions of loads in a single panel may be found 
 by drawing a sum a line noe for one panel, using as many of the loads from the forward 
 end of the train as may be placed into one panel. If this auxiliary sum a line is drawn 
 on tracing paper, it will also serve_ a ready means for finding the position of the train 
 for maximum shear in any panel mn. 
 
 The method of finding the sum a line is exactly the same as for the sum A line only 
 the pole distance for the former is the panel length d. 
 
 When the forward wheel of the train is at n, then the shear is the ordinate rj n . 
 When the forward wheel is at r the shear in the panel mn is Q r =rcre=ec where re = -n a 
 the panel reaction a for loads PI and P 2 in the panel and load P 3 at n. Similarlv when 
 the forward wheel is at s and the second wheel is at n the shear Q 8 =s/" os. 
 
 The position of the train for max.Q in the panel mn is then easily found by selecting 
 such a point s for which the ordinate O/=T? max. This maximum jj is easily found with 
 a pair of dividers and will always occur when some one of the forward wheels is at the 
 panel n. 
 
INFLUENCE LINES AND AREAS 
 
 57 
 
 If the sum a line is drawn on tracing cloth, it may be superimposed on any panel 
 of the sum A line and TJ max. can be quickly determined. 
 
 It sometimes happens that two positions of the train give the same maximum shear, 
 in which case either may be used. In Fig. 22e this would be true if oe\\cf then ~ec=of 
 and both Q r and Q s would thus be equal. 
 
 Load potitiori for Qm<. in panel m-n 
 
 FIG. 22s. 
 
 ART. 23. POSITIONS OF A MOVING TRAIN FOR MAXIMUM AND MINIMUM 
 
 MOMENTS 
 
 (a) Point of greatest moment for direct loading. Given a span of certain length 
 I, loaded with a train of loads, PI, Pa, PS, etc., to find the point n which is subjected 
 to the greatest bending moment. It is assumed that all loads remain on the span. 
 See Fig. 23 A. 
 
 Let R = resultant of all loads P on the span. 
 RI = resultant of all loads to the left of n. 
 R 2 = resultant of all loads to the right of n. 
 A and B are the end reactions for R. 
 n =the point of maximum moments to be found. 
 
 Then the moment about n is 
 
 Rx 
 
 (22A) 
 
58 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. IV 
 
 By differentiating and equating to zero, the abscissa for the point of maximum 
 moments is found thus : 
 
 dM R n a I r9 ., x 
 
 -=- r (l-2x-a) =0 or x+-=-^ ....... UOB) 
 
 ax t z z 
 
 Eq. (23B) signifies that the maximum moment under any system of loads occurs for 
 a point n when the center of gravity of the system and the point n are equidistant f rom 
 the center of the beam. 
 
 Introducing the value of x from Eq. (23 B) into Eq. (23A) the value of max. M is 
 obtained as 
 
 max.M^R^-R^. . . ........ (23c) 
 
 l> 
 
 In any special case it is necessary to find the particular load P n which must act 
 at the point n to produce a real maximum. The moment influence line clearly indicates 
 that one of the loads must fall at n to obtain the maximum moment. 
 
 FIG. 23A. 
 
 Usually the point n falls very near the center of the beam and in most practical 
 problems it will suffice to place that load at n which falls nearest the center of gravity 
 of the sj'stem of loads. 
 
 When dealing with floor stringers it frequently happens that the stringer is long 
 enough to carry three wheel loads but that the max. M so found is less than when only 
 two of the loads are on the stringer. A few of these' special cases are here given. 
 
 Case I. When there is only one load P on the span, then x=l/2 and a=0. 
 Hence Eq. (23c) gives 
 
 M Pl 
 
 max.M= (23D ) 
 
 Case H. When there are two equal loads P on the span, distant e from each other, 
 then R=2P and |=|, hence z=--|=_- 1 which gives, from Eq. (23c), 
 
 2P/Z 
 max. M =:-[ 
 
 Equating Eqs. (23o) and (23c) to find when one or two loads give equal maxima, 
 this results in the condition e=0.5858Z. Hence, when e>0.5858Z, then one load gives 
 a greater moment than two loads, all loads being of same magnitude. 
 
ART. 23 
 
 INFLUENCE LINES AND AREAS 
 
 59 
 
 Case III. When there are three equal loads P on the span, distant e from each 
 other, then R =3P and x =1/2 making a =-0. Also Ri=P and a t =e, hence 
 
 ,, 3PI n n x 
 
 max. M=. Pe=P( e}. 
 
 4 \4 
 
 (23F) 
 
 From Eqs. (23E) and (23r) it is found that when e>0.4494Z, then two loads give 
 a greater max. M than do three loads, all loads being equal. 
 
 Case IV. For four equal loads on the span may. M occurs under the second load 
 and Eq. (23c) gives 
 
 max.M=P(l+-2e] ........ (23o) 
 
 Here again it is found that when e>0.2679Z, then three loads give a greater max- 
 imum than four loads. 
 
 (b) Critical loading for maximum moments. Direct loading. In this case the 
 moment influence line for any point n is a triangle and the discussion is, therefore, 
 applicable to any beam or truss for which the moment influence line is a triangle. See 
 Fig. 23B. 
 
 :i 
 
 n-~lD..O 01 n h n 
 
 FIG. 23B. 
 
 Let R = 2P =the total load, or resultant of loads on the span. 
 
 RI =the resultant of all loads between A and n for a certain position of a train 
 
 of loads. 
 RZ= the resultant of all loads between n and B for the same position of the 
 
 train of loads. 
 
 Calling rji and rj 2 the influence ordinates for the load points of RI and R 2 respec- 
 tively, then the moment for the point n may be written as 
 
 2RiXi tan ai +R 2 x 2 tan 2 . . . . . . (23n) 
 
 Suppose now that the train is moved to the right by a small distance dx, producing 
 the change +dx in Xi and the change dx in x 2 , then the change in M n is dM n and the 
 differential coefficient thus obtained is equated to zero for maximum and minumum, 
 thus: 
 
 dM n 
 
 dx 
 
 =Ri tan ai R% tan a 2 =0. 
 
 (23j) 
 
60 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. IV 
 
 But since tan i =^-. and tan a 2 = , Eq. (23j) gives, by substitution, 
 a i a-2 
 
 ^1=??. (23K) 
 
 ai a 2 ' 
 By observing that a l +a 2 =Z and ^i +R 2 =R, then by composition, Eq. (23K) gives: 
 
 _ 
 
 a\ -f 02 a\ I 
 
 Either Eq. (23K) or Eq. (23L) may serve as a criterion for the critical position of 
 the train load on the span. The load P n , falling at n, is divided equally between 1^ 
 and R-Z. 
 
 In the first instance the criterion would be satisfied when the average unit loads on 
 both sides of the point n are equal, a condition which is absolutely fulfilled by a uniformly 
 distributed load. 
 
 According to Eq. (23L) the criterion for max. M would be that the average unit load 
 to the left of the point n must equal the average unit load on the whole span. 
 
 Also, since the maximum ordinate y is under the load point n, no position of loads 
 can give a real maximum unless one of the loads is directly over n. Hence it is readily 
 understood that there are usually two or more maxima for any point n, depending on 
 which load is placed over this point. Therefore, so long as the criterion is fulfilled, all 
 positions giving such a max. M n must be used to determine the real maximum. 
 
 In general for any moment influence line, Fig. 23s, the moment will be maximum 
 for a point n when a load P is applied at n and when the angle ^ of the influence line 
 exceeds 180. The moment at n will be minimum only when there is a load at some 
 point B for which 0'<180, that is where the angle between two successive elements 
 of the influence line is convex upward. Hence all angles in a moment influence line 
 represent critical load points and correspond to maxima or minima accordingly as these 
 angles are convex downward or upward, respectively. 
 
 (c) Critical loading for maximum moments. Indirect loading. For any point w, 
 other than a panel point, the moment influence line is a quadrilateral, hence all cases 
 for which this is true are included here. 
 
 Let .RI the resultant of all loads acting between A and c. 
 R% =the resultant of all loads acting in the panel ce. 
 R s =the resultant of all loads acting between e and B. 
 
 Other notation, as shown in Fig. 23c, which represnts a beam or girder with four 
 panels, each of length d. 
 
 From Eq. (23H) the general expression for an influence line of any number of sides 
 may be written as 
 
 M = 2Rx tan a, ........... (23 M ) 
 
 and the differential coefficient, equated to zero for maximum and minimum, becomes 
 
 -- = 2R tan a =0 ...... ... . . (23u) 
 
ART. 23 
 
 INFLUENCE LINES AND AREAS 
 
 61 
 
 The various values of R tan a are now found from the special case in hand, using 
 the notation and illustration in Fig. 23c. Three such terms are here required and these 
 are evaluated as follows: 
 
 jj V Tj c a\ d\ fje a 2 d 2 
 
 tana-i=--; tanas-; also ~~^~' 
 
 a\ a 2 TJ a\ T} a 2 
 
 from which tana 2 = 
 
 With #1, /2 2 and # 3 and these values of on, 2 and 3 , and observing that for dx 
 to the right the value R 3 tan 3 must be negative, then Eq. (23N) becomes 
 
 _ 
 dx 
 
 _ 
 d 
 
 (23o) 
 
 FIG. 23c. 
 which may be reduced to give the criterion 
 
 (23p) 
 
 Observing that R=Ri+R 2 +Rs as before; also that d=d l +d 2 and that l=( 
 then by composition of Eq. (23?) and substitution of these values the second form of 
 
 criterion is obtained as 
 
 d\ 
 
 _!l = :? (23Q) 
 
 a\ I 
 
 Eq (23r) expresses the criterion for max. M n when the average unit loads on both 
 sides of the point n are equal, provided the resultant R 2 is distributed between R, and R* in 
 
 the proportion di : d 2 . 
 
 Eq (23q) stipulates that the average unit load over the distance 01 must be equal to 
 the average unit load over the whole span, provided the portion R^/d is added to the 
 
 resultant R\. . . 
 
 The question of finding which particular load must be placed at either c < 
 fulfill either of the above criteria may be solved, by trial, as for the case of direct loading. 
 
62 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. IV 
 
 ART. 24. POSITIONS OF A MOVING TRAIN FOR MAXIMUM AND MINIMUM 
 
 SHEARS 
 
 (a) Critical loading for maximum and minimum shear. Direct loading. The shear 
 influence line is then as shown in Fig. ISA and the max. +Q n is obviously produced by 
 a full loading of the entire span between B and the load divide n. The min.Q n is 
 produced by a full loading over the distance An, which corresponds to the negative 
 influence area. This is absolutely correct for uniformly distributed loads. 
 
 For a concentrated load system, the train must cover the portion of span between 
 n and B, with the first load just to the right of n so that its influence ordinate is nn" without 
 any negative effect from the ordinate nn' . However, if the first load is small compared 
 with the second load, it may necessitate placing the second wheel at n to attain the 
 real max. +Q n =A Pi. 
 
 Hence, it is clear that either PI or P 2 must be just to the right of n to obtain max. +Q, 
 and whichever gives the greater value is then the real maximum. 
 
 (b) Critical loading for maximum and minimum shear. Indirect loading. The 
 influence line for shear, Fig. 19A, or the stress influence line for a web member, Fig. 21 A, 
 are both included here. Hence the criterion for position of loads for a max. Q will also 
 serve for maximum stress in a web member. 
 
 For uniformly distributed loads the load divide, as found in Fig. 20A, will always 
 establish the point at which the head end of the load must be located for max. +Q or 
 min. Q. 
 
 When dealing with concentrated load systems the position for maximum and 
 minimum shear is most advantageously found by graphics, as illustrated in Fig. 22e, 
 where the magnitude of these shears is at once determined, together with the critical 
 position of loads. 
 
 When analytic methods are employed, the following criterion may serve a useful 
 purpose. 
 
 Let Fig. 24A represent a truss with bottom chord loaded and i is the load divide 
 for the diagonal D. The train of loads covers the span from B to i, making RI the resultant 
 of the loads in the panel CF and R 2 the resultant of the loads between B and F. The 
 line A'C'F'B' is the stress influence line for the member >. 
 
 Then the stress in the member D is 
 
 (24A) 
 
 By shifting the train dx to the right, both Xi and x are diminished by this amount 
 dx. Hence the stress now becomes 
 
ART. 24 INFLUENCE LINES AND AREAS 63 
 
 Subtracting Eq. (24s) from Eq. (24A) the differential change in S D is obtained 
 
 dS =S D -S D ' =r t \ -^, +^===. | . (24c) 
 
 thus 
 
 From Eq. (24 c) the value dS/dx is found and equated to zero for maximum value 
 
 of S D . This gives 
 
 RI R 2 Ri+R 2 Ri R 
 
 T,=l^^ or by composition __=_=_, . . . (24D ) 
 
 where R is the resultant of all loads between B and i. This furnishes the criterion, 
 provided no additional loads have entered the panel CF and no other loads have come 
 on the span in making the shift. 
 
 FIG. 24A. 
 
 It is evident from the influence line, Fig. 24A, that any loads to the left of i would 
 produce a negative influence on SD, so that the minimum value of SD may be found 
 by loading the left portion Ai of the span, and the criterion then becomes 
 
 RI R 
 
 This criterion expresses exactly the same condition as previously found for moments 
 in Eq. (23L), but the span now becomes the length a\ or a 2 , as the case may be, cover- 
 ing only that portion of the stress influence line D, which has the same sign. 
 
 As previously found, by the graphic method, it is always necessary to place a 
 load at the panel point F for max. D and one at C for min. D, allowing as many loads 
 inside the panel CF as may be required to satisfy the criterion. 
 
64 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. IV 
 
 The above discussion of the various criteria for position of moving loads to produce 
 maximum and minimum stresses in any member of a statically determinate structure 
 is ample demonstration of the high value and practical use of influence lines. 
 
 It was not deemed advisable to extend this discussion to cover the various types 
 of special and composite structures, as the influence lines themselves, when used for the 
 determination of stresses, will give complete and comprehensive answers in all cases 
 without special analytic treatment. However, when such is desirable, the methods 
 previously employed will indicate the procedure to be followed for any particular 
 influence line. 
 
 The subject of influence lines for statically indeterminate structures and for deflec- 
 tions will be treated after presenting the general subject of distortions and deflection 
 polygons of framed structures. 
 
CHAPTER V 
 
 SPECIAL APPLICATIONS OF INFLUENCE LINES TO STATICALLY DETERMINATE 
 
 STRUCTURES 
 
 ART. 25. DOUBLE INTERSECTION TRUSSES 
 
 Usually trusses of this type are analyzed as two separate systems, dividing the 
 loads between them in such manner as may seem most probable. 
 
 The stresses found for the separate systems are then combined for the double system 
 wherever the same member forms part of each system, provided simultaneous cases of 
 loading were used. 
 
 The method of influence lines has advantages especially in the distribution of 
 simultaneous load effects, and the combined stress in a member belonging to both systems 
 is at once found for the same position of loads. 
 
 Double intersection trusses should always be designed with an even number of 
 panels, thus retaining symmetry of both systems with respect to the center of the span. 
 Otherwise the loading carried by one system in the left half span must go to the other 
 system in the right half span, which is not desirable. 
 
 Fig. 25A represents a double intersection truss for which stress influence lines will 
 now be drawn. 
 
 Bottom chord member DE. A section tt, passed through the panel DE, cuts the 
 diagonals FE and GK, hence the chord DE forms part of two panels for which there 
 are two centers of moments, F and G. Therefore, each system gives rise to a stress 
 influence line for the member DE, shown in Fig. 25A, by the two dotted triangles A'C'B' 
 and A'D'B'. The top chord members are regarded straight between alternate pin points 
 for each system in question. 
 
 Loads acting at panel points are supposed to be carried by the system to which 
 each such point belongs except at the panel where the load is divided equally between 
 the two systems. 
 
 Hence a load at D goes entirely to one system and its influence is represented by the 
 ordinate DD' ', of the influence line A'D'B'; while a load at E is carried by the other sys- 
 tem and its influence ordinate becomes EE' in the influence line A'C'B'. This same 
 reasoning applies to all panel points. 
 
 Since influence lines between . successive panel points must be straight lines, the 
 actual influence line for DE is represented by the shaded area A'O'C'D'E'K' to B' . 
 The point 0' must lie midway between the two ordinates for because at this point 
 the load is carried equally by both systems. This is not the case at M, where the entire 
 load goes to one system. 
 
 65 
 
60 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. V 
 
 Top chord member LF. The influence line for DE is like the influence line for LF, 
 except that the end ordinates are slightly different. Here the lever arms are measured 
 normally to the members LF and FG. 
 
 Diagonal member FE. The method of reasoning just applied to chord members 
 answers in every sense for the web members. 
 
 The influence line for FE is constructed as indicated in the figure. The stresses 
 in FE, due first to a unit load at A and then at B, are found for the system to which FE 
 belongs. 
 
 FIG. 25A. 
 
 Since loads at the panel points of the other system cannot cause stress in FE, the 
 influence line ordinates for these points must be zero. This means that the other influence 
 line is the base line A'B'. Observing again that the required influence line must be 
 straight between successive panel points the final line becomes A'O'C'D'E'K' to B'. 
 The point 0' , as before, is midway between the ordinates under 0. 
 
 The influence line indicates clearly that when loads are spaced 2d apart, then the 
 entire web stress goes into one system, making the design verj'- uneconomical. This 
 would easily happen for 50 ft. locomotives and 25 ft. panels. 
 
ART. 26 INFLUENCE LINES FOR STATICALLY DETERMINATE STRUCTURES 07 
 
 ART. 26. CANTILEVER BRIDGES 
 
 Any statically determinate structure extending over more than two supports (piers 
 or abutments), when continuous over r2 of these supports, may be regarded as a 
 cantilever. If r is the total number of supports, then such a structure must be provided 
 with r2 intermediate hinges such that the bending moments at these hinged points 
 are zero for any system of loads. 
 
 The reaction and typical stress influence lines for such a cantilever will now be 
 drawn. 
 
 Reaction influence lines for all simple structures are independent of the panels and 
 may, therefore, be illustrated on a simple cantilever beam, Fig. 26A. 
 
 FIG. 26A. 
 
 The illustration represents an anchor arm A B with a downward reaction or anchorage 
 at A and an upward pier support at B. The span EF is a simple beam on two supports 
 E and F, with an overhanging cantilever arm at_each_ end. _A similar arm projects out 
 from the anchor arm, making three cantilevers, BC, DE and FG, in the whole system. 
 
 The two spans CI) and GH are known as intermediate spans and are supported on 
 hinged ends which permit of horizontal displacements. Hence, there is only one roller 
 bearing at E, and the anchorage at A also admits of slight horizontal motion. 
 
 The whole system is, therefore, determinate and can have no temperature stresses, 
 neither will slight displacements of the supports cause any internal stresses. 
 
 The reaction influence lines thus involve no principles other than those already 
 elucidated for simple beams on two supports. 
 
 The central span WF is a suitable beginning, as it rests on two supports R 3 and #4 
 with a roller bearing at E. Hence the influence lines for these two reactions are drawn 
 precisely as for a simple beam between the points E and F. But the two arms at the 
 ends of this span must also be considered. 
 
 Regarding first the influence line for R 3 , it is seen that a load at F has zero effect, 
 
68 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. V 
 
 and as the load proceeds to the left its effect on R% is uniformly increasing and this 
 continues to be the case until the load reaches the point D, where the effect becomes 
 a maximum. 
 
 The same is true for a load between F and G, hence the ordinate unity at E' 
 determines the RS influence line from D^_io G' . A load at C or at H produces zero 
 effect on R s and loads outside the span CH have no influence, hence the line C'D'G'H' 
 is the influence line for R s . Further proof will scarcely be needed, though the several 
 moment equations for the separate spans CD and GH will readily supply this. 
 
 Similarly the line C"D"G"H" represents the influence line for R$, based on the 
 unit ordinate at F". 
 
 The influence line for Rs is the same as for a simple beam GH. 
 
 Precisely the same reasoning applies to the reactions Ri&nd R-2, the influence lines 
 for which are shown to the left in Fig. 26A. It should be noted with regard to RI that 
 the negative influence area being greater than the positive, provision must be made 
 for a downward reaction. When the upward reaction RI, due to live load from A to 
 B, is greater than the downward reaction R'i, due to dead load from A to D, then the 
 fastening at A must be designed to take stress in both up and down directions. 
 
 Shear influence lines. These are readily deduced from the reaction influence lines, 
 as may be seen by comparing Fig. 26s with Fig. 26A. Four typical panels are assumed. 
 One each for the central span EF, the cantilever arms BC and FG, and the anchor 
 arm A B, respectively. 
 
 FIG. 26s. 
 
 For the panel ef the shear influence line within the span is precisely as for any 
 simple beam on two supports, while outside the span these lines are continued to the 
 intersections D' and G' with the verticals through the hinges D and G, respectively. 
 The shear influence line for the panel ef is thus found as C'D'e'f'G'H' . 
 
 For the panel ab the same construction is applied as for a span on two supports 
 A and B followed by a cantilever to the right of B exactly as for the previous case. 
 The shear influence line for the panel ab is the polygon A'a'b'B'C"D" '. 
 
 When the panel is located in one of the cantilever arms as for gh, then the circum- 
 stances are slightly different, because the shear in any cantilever arm is always equal 
 to the sum of the applied forces between the section and the outer end of the arm. 
 
ART. 26 INFLUENCE LINES FOR STATICALLY DETERMINATE STRUCTURES 69 
 
 Accordingly the shear must be composed of the end reaction from the intermediate 
 
 Hence the influence line 
 It should be noted that 
 
 truss GH and the constant effect of loads between h and G. 
 
 for the panel gh is easily found to be the polygon g'h'G"H' . 
 
 loads to the left of g, even on the next span EF, can have no effect on the shear in the 
 
 panel gh. 
 
 The shear influence line for the panel cd is similarly found to be the polygon c'd'C"'D" '. 
 
 Any panel located within either of the intermediate spans CD and GH, is treated 
 precisely as for a span on two supports, since no load outside these spans can produce 
 any effect on these trusses. 
 
 Moment influence lines. These offer very little difficulty and follow directly from 
 the methods previously given. 
 
 In Fig. 26c the anchor arm was omitted, as the portion of the structure from C 
 to H affords all the illustration required. 
 
 Here, again, the central span EF is treated exactly like a simple beam and the 
 negative effects of the adjoining cantilever arms are found by prolonging the lines so 
 
 FIG. 26c. 
 
 obtained to the intersections D' and G' with the verticals through the hinged points. 
 The moment influence line for the point n, in the panel EF, thus becomes C'D'E'e'f'F'G'H' , 
 and it is readily seen that a load on any portion of the structure CH affects the moment 
 at n. 
 
 For the section t, in the panel gh, the moment is the reaction at G into the lever 
 arm a 3 , hence for a unit load, the ordinate at G must be l-a 3 , and observing that the 
 influence line must be straight over the panel gh, the required moment influence line 
 for the section t becomes F"g'h'G"H". 
 
 Stress influence lines. The general method for stress influence lines, illustrated in 
 Fig. 21 A, serves every purpose in connection with cantilevers. 
 
 In every determinate structure the stress in any member may be expressed as 
 S=M/r and from this it follows that the stress influence line for any truss member 
 is similar to the moment influence line drawn for the center of moments pertaining to the 
 member. The difference between the stress and moment influence lines is only the 
 constant divisor r applied to the ordinates of the latter to obtain the ordinates for the 
 stress influence line. 
 
 For convenience, the central span CE, Fig. 26o, will be treated first, because this 
 
70 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. V 
 
 portion of the structure is exactly like a simple truss on two supports, and all the rules 
 previously given apply here. 
 
 Regarding the web members it should be observed that the center of moments may 
 fall inside or outside the limits of the span. In the former case there will be no load 
 divide within the span. Both cases are illustrated in Fig. 26D, where the stress influence 
 lines for members U, L, D, and D' are presented. 
 
 The designations regarding the stresses in these members for unit reactions at C 
 and E, will be as previously adopted in describing Fig. 2lA, viz., S uc =ihe stress in the 
 member U for a reaction unity acting at C. 
 
 FIG. 26o. 
 
 Since these stresses are always necessary in constructing influence lines, it is best 
 to find them for all members of the structure before proceeding to draw any influence 
 .lines. These stresses are very easily found either by slide rule, using Ritter's moment 
 method, or by drawing a Maxwell diagram. 
 
 Stress influence line for U, central span. This being a top chord member, and in 
 compression, its influence line is negative between C and E, which is also recognized by 
 the stresses S^, and S ue found for this member for the reactions unity, acting first 
 at C and then at E. 
 
 These stresses are laid off from A'B', Fig. 26D, upward and on the verticals through 
 
ART. 20 INFLUENCE LINES FOR STATICALLY DETERMINATE STRUCTURES 71 
 
 C and E, respectively. The stress influence line is then easily drawn, observing that 
 the intersection n' must fall under the center of moments n for the member U. Hence 
 this is a check and may also be used as a shorter method of finding the U line, when 
 only S llc is known. Between the panel points m and o the influence line must be straight, 
 hence m'o' is drawn. 
 
 To complete the influence line over the side spans, the U line, just found, is continued 
 to intersections B" and F" with the verticals through the hinged points B and F and the 
 final U line is then found to be A'B"m'o'F"G' . The portions below the base line A'G' 
 are positive and the upper portion is negative. The positions of loads to produce max. U 
 and min. U are thus clearly indicated and no other criterion is required. 
 
 Stress influence line L, central span. This is constructed in precisely the same manner 
 as the U line, but L being in tension, as found for C = l, the stress +Si c is laid off down- 
 ward under C. The point m" . vertically under the center of moments m for the member 
 L, is thus found arid may serve to complete the L line without using the stress Si e . 
 The L line is then obtained as A'B'm"F'G' and is positive over the span CE and negative 
 over the two cantilever spans AC and EG. 
 
 Stress influence line D, central span. The center of moments for this member is at 
 i, which is within the supports, and hence this member has no load divide. The stress 
 influence line is constructed exactly as was done for the U line, by laying off the positive 
 stresses S dc and S de to obtain the two limiting rays, with intersection i' under the point i. 
 
 However, since the rate of increase in the stress D must be uniform for a load advanc- 
 ing from E to o, the ray E' r7 i' must be continued to o ly finally drawing the line m^i to 
 complete the polygon. 
 
 Over the two side spans the D line is drawn as for the two previous cases, giving 
 the complete line A'"B f "C f "mio l E'"F"'G"'. 
 
 Stress influence line D', central span. The center of moments for D' is outside the 
 span at k. hence the influence line must show a load divide. The stresses + S'd c and S'd e 
 serve to construct the stress influence line for D' which is in all other respects like the 
 D line. The limiting rays intersect in k' and the load divide is at q. 
 
 Stress influence line U, cantilever arm, Fig. 26B. The effect of a moving load, coming 
 on the span from left to right, begins when the load is just to the right of A and increases 
 uniformly until the load reaches B. From B toward m the effect is uniformly decreas- 
 ing and becomes zero when the load is over n, which is the center of momentsjor U. 
 The stress in U for a load P = l at B is-S ub , and laying this off from the_base A'o', the 
 negative U line is easily drawn. For the effect due to loads in the panel mo draw m'o' to 
 complete the influence line. Loads to the right of o or to the left of A have no influ- 
 ence on U. 
 
 Stress influence line L, cantilever arm. The center of moments for L is at o and 
 the stress in L for a load P = l at B is+S lb . Hence the influence line becomes A"B"o" 
 
 and is positive. 
 
 Stress influence line D', cantilever arm. The center of moments for this diagonal 
 is at i, which is inside the cantilever arm. The rule for load divide is now reversed 
 because loads on opposite sides of i produce opposite stresses in D', and a load at i cannot 
 produce any stress. Hence i is the load divide for D'. 
 
72 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. V 
 
 The stress in D' due to a unit load at F is S' d f, hence making the ordinate under 
 F equal to this stress the influence line s'i'F'G' is obtained. 
 
 Since loads to the left of h cannot affect D', and since the influence line is a straight 
 line over a panel, the influence line is completed by drawing h's'. 
 
 Stress influence line D, cantilever arm. The center of moments now falls off the 
 arm at k and hence all loads produce the same kind of stress in D. Lay off the stress 
 +S df on the vertical through F and complete the influence line as shown in Fig. 26E, 
 obtaining s"v'F"G" for the D line. 
 
 The intermediate spans A B and FG are simple trusses on two supports and require 
 no further consideration. 
 
 FIG. 26E. 
 
 ART. 27. THREE-HINGED FRAMED ARCHES 
 
 In taking up this subject it is necessary to show briefly the general relations between 
 the reactions and external loads. 
 
 Let Fig. 27A represent any three-hinged arch ACB, which may be framed or solid. 
 The single load P may be applied at any point m. 
 
 For all positions of P between A and C, the reaction R% will always have the 
 direction BC and the two reactions RI and R% must be components of P. Similarly 
 for all positions of the load P between C and B, these reactions will have the directions 
 AC and eB, respectively. Hence for any position of the load P, the reactions at A and 
 B are found by resolving P into two components as shown, and the points of application 
 of all possible positions of P will lie on the prolongation of either AC or BC. 
 
 If the reactions R and R 2 be resolved, each into a vertical component and one 
 along AB, the two last-named components H' will be equal and opposite, while the two 
 vertical components A and B will be precisely the same as for a simple beam on two 
 supports A and B, because the forces acting at C are in equilibrium among themselves. 
 
 Therefore, 
 
 A+B=P; 
 
 Pa 2 
 A -, 
 
 and B= = . 
 
 (27A) 
 
AKT. 27 INFLUENCE LINES FOR STATICALLY DETERMINATE STRUCTURES 73 
 
 The horizontal component of H' is known as the horizontal thrust of the arch and 
 is found from 
 
 H=H'cosa (27B) 
 
 If the reactions R\ and R 2 had been decomposed into components _4/, B' and H, 
 then the former would no longer be the reactions for a simple beam AB. However, 
 when a =0 and the end supports are on a horizontal line, then the vertical end reactions 
 are always as for a beam on two supports. 
 
 FIG. 27A. 
 
 Taking moments about the hinge C, when P alone is acting, then for condition of 
 
 equilibrium at C 
 
 M=Al l -H'y=Al l -H'fcosa=Q, 
 
 and substituting values from Eqs. (2?A> and (27s) this gives 
 
 Hi or H= J-J-. . . 
 
 
 (27c) 
 
 When P = l is applied at C the value H c becomes the ordinate vertically under C 
 for the influence line for horizontal thrust. 
 Hence, 
 
 ,-i, (27D) 
 
 -_^ and when Zi=Z 2 =o, then 
 
 making the H influence line a triangle with middle ordinate T? C under C, as given by 
 
 Eq. (27D). 
 
 The influence lines for the vertical end reactions are found exactly as for a simple 
 
 beam AB. 
 
 The stress influence line for any member of a three-hinged framed arch will now be 
 
 developed. 
 
 When the structure is symmetric, the half arch only requires analysis, but for 
 unsymmetric arches the entire structure must be treated, though the method remains 
 the same as would appear from the previous discussion. 
 
KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. V 
 
 In Fig. 2?B, the half arch is shown with a load P producing reactions R\ and R 2 , 
 and since the latter must be transmitted to the point C, the right half of the span may 
 be considered removed and its effect replaced by R 2 acting at C. The left half of the 
 span is then held in equilibrium by the two reactions R : and R 2) the latter applied at C. 
 
 The reaction R\ is now resolved into A and K and the reaction R 2 is similarly 
 replaced by a vertical reaction C and a reaction K which is equal arid opposite to the 
 first K. 
 
 FIG. 27s. 
 
 If the forces K were zero, then the half span would represent a simple truss on two 
 supports, with vertical reactions A and C and the influence line for any member could 
 be found by the method previously given. Thus, the influence line for any member- 
 is easily drawn when the stresses, first for A=l, and then for 5 = 1 are known. Such 
 an influence line is drawn for the diagonal D, and is shown as the polygon A'E'F'C' ',. 
 Fig. 2?B, where the stress D a , due to A = 1 is positive, and the stress D c , resulting from 
 C = l, is negative. In each case the half span is supposed to be held first at C and then 
 at A in order to find these stresses. 
 
ART. 27 INFLUENCE LINES FOR STATICALLY DETERMINATE STRUCTURES 75 
 
 The influence line thus found, must now be corrected so as to include the effect of 
 the force K. 
 
 Since K=Hsec{3, therefore K is proportional to H and the stress Dk, produced 
 by the force K, must likewise be proportional to K and H. Hence, the influence line 
 for the stress D k , is similar to a horizontal thrust line and must be a triangle whose 
 middle ordinate under the hinge C is equal to the stress D k , produced by a unit load 
 applied to the whole span at the point C. 
 
 To find this stress D k for a unit load at C, substitute P--=l into Eqs. (27 A) and 
 (27c) and obtain 
 
 A=i and # = l'i- 
 
 J 
 
 If the stress in D is found first for A = 1/2 and then for // =l/4f } separately, then 
 the combined effect will be the stress D k =%D a +D H , which is much more convenient 
 than to find D k directly for the diagonal force K, The stress D H is the stress in D 
 for the above force H = l-l/4f applied horizontally at A and supposing the half arch 
 rigidly held at C. 
 
 The stress D a was previously found, and hence the required stress D^ is readily 
 obtained from 
 
 (27B) 
 
 In the present case D^ was found to be negative and the influence line for the 
 member D due to the effect K is drawn as the triangle A"C"B" ' . 
 
 The final D line is now obtained by combining the two influence lines just found- 
 This is done in the last figure and gives the polygon A'"E"F"C'"B" r . The stress +D a 
 is laid off down from A'" and the stresses D c and D^ being negative, are laid off 
 upward from the base line A'"B'" and being of the same sign they are added. The 
 polygon is then drawn as indicated in the figure and as a check, the point n" must fall 
 vertically under the 'center of moments n for the member D. Still another check is 
 found from the circumstance that a load at i produces zero stress in the member D and 
 hence the influence line must intersect the base line vertically under i in the point i'. 
 This last mentioned condition is always true even though the point i falls below C. 
 
 The point i is the intersection between BC and An, where n is the center of moments 
 for D. This offers a very valuable condition and obviates the necessity of finding the 
 stress D c , whenever the center of moments n can be located within limits of the drawing. 
 
 It is thus seen that when n and i are conveniently located, which is always the case 
 for chord members, then the influence line can be determined when D a alone is given. 
 However, as a check, it is always desirable to know D^, so that when the stresses in 
 all the members are to be determined by influence lines, it is advisable to find S a and 
 Sh for each member before proceeding to draw the lines. This can be done by drawing 
 two Maxwell diagrams, one for A = 1 and the other for H=l/4f. A slide-rule computa- 
 tion will also answer if the lever arms are carefully determined. 
 
 In Fig. 27s the loads were supposedly applied to the top chord. Had they been 
 applied to the bottom chord then the panel line E"F" would have become G"H". 
 
76 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. V 
 
 Typical stress influence lines will now be given for the three-hinged, framed arch, 
 Fig. 27c, treating the top chord as the loaded chord. 
 
 The example will be taken up in such manner as to indicate a complete solution 
 for all stresses in the several members by the method of influence lines. 
 
 ~?B 
 
 B' 
 
 I UU INFLUENCE LINE. 
 
 "^luiii]|p|i||]^ 
 
 90 LIN&THV 
 
 The first step will be to determine the stresses S a and S k for all the members and 
 this will be done analytically by Ritter's method of moments, being very careful about 
 the sign of each stress. 
 
 The same stresses could also be found in a very convenient way by drawing two 
 Maxwell diagrams, one for A~l, supposing the half truss ~AC fixed at C; and the other 
 
ART. 27 INFLUENCE LINES FOR STATICALLY DETERMINATE STRUCTURES 77 
 
 for a load P = l acting vertically downward at C and producing the horizontal thrust 
 // = l-Z/4/= 0.833, which is again applied at A, supposing the half truss fixed at C. 
 
 The following table contains the stresses S a and S a thus obtained and the stresses 
 S=S a +S are then easil found. 
 
 MEMBER. 
 
 STRESSES. 
 
 MEMBER. 
 
 STRESSES. 
 
 -Sa 
 
 SH 
 
 8t 
 
 Sa 
 
 S H 
 
 Sh 
 
 U,A 
 
 -1.00 
 
 + 1.02 
 
 + 0.52 
 
 U& 
 
 -0.45 
 
 + 0.52 
 
 + 0.30 
 
 f7,L, 
 
 + 0.93 
 
 -0.57 
 
 -0.11 
 
 iW 
 
 -1.37 
 
 + 1.08 
 
 + 0.40 
 
 u& 
 
 -1.29 
 
 + 0.79 
 
 + 0.15 
 
 U 3 U t 
 
 -3.11 
 
 + 1.97 
 
 + 0.42 
 
 U 2 L 2 
 
 + 1.32 
 
 -0.85 
 
 -0.19 
 
 AL t 
 
 0.00 
 
 -1.44 
 
 -1.44 
 
 U 3 L 2 
 
 -1.65 
 
 + 0.77 
 
 -0.05 
 
 L,L. 
 
 + 0.85 
 
 -1.87 
 
 -1.45 
 
 U 3 L 3 
 
 + 2.19 
 
 -1.15 
 
 -0.06 
 
 L 2 L 3 
 
 + 2.08 
 
 -2.38 
 
 -1.34 
 
 U t L 3 
 
 -2.45 
 
 + 0.96 
 
 -0.26 
 
 L 3 C 
 
 + 4.40 
 
 -3.35 
 
 -1.15 
 
 u t c 
 
 + 3.05 
 
 -1.19 
 
 + 0.33 
 
 + for tension and for compression. 
 
 The values in this table will suffice for the construction of all the stress influence 
 lines, but only those for (7 2 t/ 3 , L 2 L 3 , U 2 L 2 and / 4 L 3 will be drawn, as shown in Fig. 
 27c. 
 
 Stress influence line U 2 U 3 . The stress S a = 1.37 is laid off, to a convenient scale, 
 upward from A' and the stress ^=+0.40 is laid off downward from A'B', vertically 
 under C. The lines EC' and C'B' are then drawn. 
 
 The center of moments for U 2 U 3 is at L 2 , hence find Z/ 2 vertically under L 2 and 
 Finally since the influence line must be straight over the loaded panel 
 
 As a check observe 
 
 draw A'L 2 '. 
 UjU~ 3 , draw U 2 'U S 
 
 to complete the influence line A'U 2 UzC'B'. 
 
 that the load divide i/ falls vertically under i', which latter is the intersection of AL 2 
 with BC produced. Still another check may be had by resolving a unit load at U 2 
 into components parallel to U 2 U S and U 2 L 2 as was done to the right of the truss diagram. 
 The stress thus found for U 2 U 3 must equal the ordinate U 2 'F. Lastly, the ordinate 
 S c is the stress in U 2 U 3 for a load unity at C when the half truss AC is fixed at A 
 and the half span CB is removed. 
 
 Stress influence line L 2 L 3 . The method is precisely as for the previous case, being 
 careful to observe the signs of the stresses S a and S k in laying them off from the base 
 line A 7 /?'. The point U' 3 , vertically under the center of moments for this member, 
 again serves to complete the stress influence line, with all the checks just mentioned 
 for the upper chord member. 
 
 Stress influence lines U 2 L 2 and U 4 L 3 . The same method again applies and the two 
 lines illustrate the effects of the signs of the stresses S a and S k . The center of moments 
 n, for the member U 2 L 2 now falls outside the span, but the same relations hold good 
 as before. Similarly for the center HI of the member U 4 L 3 . In this last case the load 
 divide t' 3 falls below the middle hinge C and is no longer a real load divide, as the influence 
 line for TT^Ls indicates, but all other relations continue to exist as before. 
 
KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. V 
 
 ART. 28. THREE-HINGED, SOLID-WEB AND MASONRY ARCHES 
 
 Fig. 28A illustrates a three-hinged, solid-web arch, which may be built of masonry 
 or steel, according to circumstances. If of masonry, then the section will be rectangular, 
 and if of steel, then the section will resemble that of a plate girder. 
 
 Whenever the depth of the section is small compared with the radius of the arch 
 center line, then the effects of curvature are very minute and the stresses in the arch 
 rib may always be found by the well known beam formula of Navier. This condition 
 always prevails in bridges and all larger arches, but when dealing with small arched 
 culverts and structures of that class, then the effects of curvature may require special 
 consideration. 
 
 For any girder section then, the stress on the extreme fiber is derived from 
 
 N 
 
 wherein M is the bending moment, A" the normal axial thrust, F the cross-section, / 
 the moment of inertia of the section and ?/ the distance from the center of gravity of 
 the section to the extreme fiber. M k is the total bending moment of the external 
 forces about one of the outer kernel points e or i (Fig. 28A) and W is the moment of 
 resistance of the section. 
 
 Calling r the radius of gyration of the section, then 
 
 7 Fr 2 r 2 I 
 
 W=-= =Fk where k=^-~ ...... (2SB) 
 
 y y y Fy 
 
 The quantity k is the distance from the gravity axis to a kernel point and the negative 
 sign indicates that it is measured in the opposite direction from y for all sections. 
 
 Hence, the influence lines for the kernel moments M e and Mi will serve to find the 
 stresses f e and /; for the extreme fibers of extrados and intrados of any particular section 
 tl, by using the multiplier /y. = l/W = l/Fk. For unsymmetric sections the two kernel 
 distances are not equal so that {i e = l/Fk e for extrados and fi.i = \/Fki for intrados. 
 
 For each section the stresses on the extreme fibers thus become 
 
 ''-7S, and fi =~K ..... ' ' (28c) 
 
 where M e =K ac e and Mi=K ac i, and K ac is the end reaction at A for a load unity at C 
 found by resolving the vertical reaction A along AC and AB. See also Art. 49 on two- 
 hinged arches. 
 
 The moment influence lines M e and M; are now found by the method previously 
 outlined for framed arches. 
 
 Each of these moments becomes zero when the unit moving load passes through 
 a load divide d. Hence the load divides are easily located for each case. 
 
 Also since moments and not stresses are now considered, the end ordinate for the 
 influence line at A is equal to x e or Xi as the case may be. 
 
ART. 28 INFLUENCE LINES FOR STATICALLY DETERMINATE STRUCTURES 79 
 
 Hence to draw the influence line Mi, lay off Xi at A and project the point d down 
 to d', draw Gd'C' and after projecting down the center of moments i to find i', finally 
 complete the influence line by drawing A'i' and C'B'. Now find as a check, that 
 ~C'~D = x'i. 
 
 \ 
 
 B' 
 
 FIG. 28A. 
 
 The Af t - influence line becomes the/; influence line with a factor m=* 
 
 The influence line M e is found in precisely the same manner. 
 
 In both cases the ordinate C'C" is equal to M t or M e as given by Eq. (28c) for a 
 
 unit load at C. The M e influence line becomes the / influence line with a factor 
 
80 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP, v 
 
 The shear influence line Q OT is found by a similar process. Here the center of 
 moments, for a web member at m, would be practically at an infinite distance because 
 the chords are parallel or nearly so. Hence the line AdAj.t will intersect the chord 
 C/I in the load divide d, which in this case is only imaginary, but serves the purpose 
 as before. 
 
 The equation for shear along the section it is easily found to be 
 
 and this gives, for a unit reaction at A, when the half arch is fixed at C, the end ordinate 
 A'G = l-cos <f>. Also since the shear is constant between A and m, the limiting lines 
 A'm' and Gd' must be parallel. 
 
 Hence the shear influence line is easily constructed by drawing GC'd' , and then 
 completing the polygon by drawing A'm' || Gd' and joining C' with B' ' . 
 
 Since the loading is directly applied the line Wira', vertically under m, completes 
 the Q 7> , line. 
 
 Finally the ordinate C'C" =H[ - - ) wherein 7/ = l--^rfor a unit load at C. 
 
 \ cos a ) fl 
 
 When the loads are indirectly applied, then the condition that the influence line 
 must be straight between load or panel points must be carefully observed. 
 
 For masonry arches the section is always rectangular and 7 =/>D 3 /12, hence k = r' 2 /y = 
 ^F^Z), where D is the thickness of the arch ring. The points e and i thus become the 
 middle third points of the arch section and Eqs. (28c) become 
 
 , GM e , 
 
 f*~~TF and ^ = 
 
 from which the stresses in a masonry arch section are readily found in terms of the 
 moments about the kernel points of the section. The stress influence lines thus become 
 moment influence lines with a factor 6/D 2 for each section. 
 
 ART. 29. SKEW PLATE GIRDER FOR CURVED DOUBLE TRACK 
 
 In general, influence lines for skew bridges are drawn in all respects like those for 
 any simple truss or beam with the exception that at one end of the span the load always 
 leaves the span before it is entirely outside of the limits of the longer truss. This merely 
 calls for a slight correction in the end panel of the influence line as ordinarily drawn. 
 
 At the opposite end of such truss some load comes on the end floor beam before 
 the moving load is on the truss, but since the end floor beam transmits its load directly 
 to the bridge support and not to the truss, this last named circumstance does not affect 
 the influence line. These points will be brought out in connection with the following 
 problem. 
 
 It may seem inappropriate to apply influence lines to plate girders, but the practical 
 designer who has dealt with problems of skew plate girders on curved double track 
 will readily appreciate the advantages of the method there presented. 
 
KT. 29 INFLUENCE LINES FOR STATICALLY DETERMINATE STRUCTURES 81 
 
 MOMENT INFLUENCE LINE " ~ -- 
 POINT 7, GIRDER NO.I.TRACK NO.2. 
 
 FIG. 29x. 
 
82 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP.) 
 
 This method was first suggested and used by Mr. T. B. Moenniche, C. E., in designing 
 a skew double-track girder of 125 ft. span for the Virginian Ry. Co., in 1906. Thi 
 author is indebted to Mr. Moenniche for the solution which follows. 
 
 The principal advantage in the present application of influence lines consists ii 
 reducing the solution of this rather complicated problem to a method instead of a time 
 consuming process of experimentation. 
 
 The principle of the method may be outlined as follows: The floor-beam reaction! 
 are first determined for a unit load situated on track Xo. 1, and then for a unit load 01 
 track No. 2, which requires a careful computation of the coordinates or intersectioi 
 points of each track on each floor beam. These floor beam reactions are then used a! 
 influence or reduction factors for the ordinates to any particular influence line, becaus< 
 the factors represent the proportion of a total load which can reach either girder a 
 any panel point. This will give two influence lines for each panel point of a girder 
 one separately for each track. 
 
 In Fig. 29A, the upper diagram represents the structure in plan and the two sue 
 ceeding figures show the floor beam reactions graphically. The last two diagram: 
 are the moment influence lines for point 7, girder No. 1, for tracks No. 1 and 2, respectively 
 
 The two reaction diagrams were drawn merely as a convenient way of showing 
 the figures and in practice they would serve the same purpose, though they might b( 
 dispensed with entirely. The reactions appearing as ordinates are plotted to sonn 
 convenient scale making the sum of two reactions for each floor beam equal to unity 
 except at the ends where the floor beams do not connect with both girders. 
 
 The moment influence lines for point 7 are drawn in the usual manner by laying 
 off a 2 vertically down from B and drawing A'T and 7'B' '. Since the load on track \o 
 1, when coming on from the right, begins to affect girder No. 1 as soon as it passes the 
 abutment at E, therefore, the distance #10 must be treated as a panel arid the influence 
 line is drawn straight from 10 to B'. 
 
 The ordinates d of the influence line are now successively reduced by multiplying 
 them with the corresponding ordinates or factors m of the reaction diagram to obtaii 
 the real ordinates y of the required moment influence line. Thus ^6 =< Vn 6 and similar!} 
 for the influence line point 7, girder 1, track No. 2, the ordinate Jjg=0 8 w 8 . 
 
 With the use of these two influence lines, represented by shaded areas, the actual 
 moment for point 7 may be found for any case of loading for each track and the sun: 
 of the two gives the combined max. M 7 . 
 
 Shear influence lines may be drawn in a similar manner, though they would scarcely 
 serve a useful purpose except when dealing with a truss. In the above example onl} 
 the reaction influence lines would be useful to determine the end shears. 
 
AKT. :-o INFLUENCE LINES FOR STATICALLY DETERMINATE STRUCTURES S3 
 
 ART. 30. TRUSSES WITH SUBDIVIDED PANELS 
 
 From a point of economy, large trusses must have long panels, but with the increase 
 in panel length the stringers of the floor system soon attain such proportions as to 
 impose a practical limitation on the panel length. 
 
 To overcome this difficulty, modern practice has developed a new type of truss 
 especially adapted to long spans, wherein the stringers are supported by floor beams 
 placed at the half-panel points. These intermediate floor beams are carried at the ends 
 by a secondary set of truss members, which latter serve to carry the loads, thus locally 
 applied, to the main truss system. This has given rise to the truss with subdivided 
 panels. 
 
 In the analysis of these truss types, it is not always a simple matter to determine 
 the criterion for maximum stresses in the members, and it is believed that the following 
 treatment by influence lines will serve to clear up the doubts usually encountered. 
 
 Fig. 30A shows a truss of 200 ft. span with four different forms of subdivided 
 panels. These are all combined in the one structure, though in practice only one form 
 should be used in the same structure excepting in the first panels ALiL 2 , where the 
 several compression members shown offer distinct advantages in stiffening the ends 
 of the truss. Otherwise the panel L 2 L 3 , Case I, is the best and most economical type 
 for adoption, as will appear later. 
 
 In general, all primary members in the structure will receive stresses which are 
 at least equal to those which would exist if the secondary members were all removed 
 and in addition to these primary stresses nearly all of the members, excepting the 
 verticals, will receive increased stresses due to local loads from the secondary members. 
 This local loading is transmitted to the truss through the vertical M Z K which never 
 acts as a primary member and gets no stress other than that due to a full panel load, 
 besides the dead weight of bridge floor and bottom chords. 
 
 It is the load, locally transferred to the points M , which enters as a disturbing 
 element, and the four cases here treated will illustrate about all the practical points. 
 
 To make the four cases directly comparable, the same panel L 2 L 3 was retained, 
 but the diagonals were successively rearranged. The stress influence line for the 
 primary stress in U 2 L S remains the same in each case, hence it is drawn only once 
 and is then modified according to local conditions in the panel L 2 L 3 for each case. 
 
 The Diagonal Member U ,L ,. 
 
 Case I. The stress influence line for U 2 L 3 , when acting only as a principal truss 
 member, is drawn by laying off the end ordinate A' A" = +0.568 which is the stress 
 U 2 L 3 for a reaction unity at A. The ordinate B'B" (not shown) is 2.790 and is 
 the stress in the same member for a unit reaction at B. By drawing A" ' B' and A'B" 
 the points U' 2 and L' 3 are found and the line t/ 2 'L 3 ' completes the influence line 
 A f U 2 L s 'B' with load divide at i. 
 
 This gives the maximum stress for M 3 L 3 but not for U 2 M Zl because a load at K 
 would produce additional stress in the latter member without affecting the former. 
 
 Hence to find the influence line for U 2 M 3 , it is necessary to determine the stress 
 
84 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 i STRESS INFLUENCE LINE M 3 L 3 . 
 
 FIG. 30A. 
 
ART. :so INFLUENCE LINES FOR STATICALLY DETERMINATE STRUCTURES 85 
 
 in this member due to a local load unity at M 3 which gives the total ordinate 
 under M 3 . 
 
 A load at M z will be carried entirely by U 2 M 3 and M^U~ 3 , hence the stress due to 
 a unit load is readily found by resolving the unit load into components parallel to these 
 two members as indicated by the triangle of forces in the figure for Case I, drawn to half 
 the scale of the influence line ordinates. The stress in U 2 M 3 is thus found to be 
 +0.5, which is laid off from m to complete the influence line A'U 2 'M 3 L 3 B' for the 
 member U 2 M 3 . It is seen that the point M' 3 does not fall on the line B'A" as usually 
 assumed by authors on this subject. The error from this assumption is more apparent 
 in Case II, and is not always on the safe side. 
 
 Case II. The influence line for the primary stress in U 2 L 3 is as before, and again 
 gives maximum for the lower portion M 3 L 3 . However, a unit load at M 3 is now carried 
 to the members U 2 M 3 and L 2 M 3 , and these stresses are again found by a triangle of 
 forces drawn in the figure for Case II. 
 
 The stress thus produced in U 2 M 3 is now +0.62, from which the new influence 
 line for U 2 M 3 is readily obtained as A'U 2 'M 3 L 3 B'. 
 
 Case III. Here the primary stress holds good for the upper portion U 2 M 3 and 
 the stress in the lower portion M 3 L 3 is diminished by the compression due to local 
 loading at M 3 . A unit load at M 3 is now carried by M 3 L 3 and M 3 U 3 and the triangle 
 of forces again gives the ordinate iM 3 , which is here negative. 
 
 Case IV. This is like Case III, but a local load at M 3 is now carried by L 2 M 3 
 and M 3 L S equally. Hence the stress in M 3 L 3 due to a unit load at M 3 is found as 
 0.62, which gives the ordinate iM 3 . 
 
 Members L 2 M 3 and M 3 U 3 , whenever they occur alone, can act only as secondary 
 members to carry the portion of load locally applied at M 3 . When the panel is subject 
 to counter stress as in panels L 4 L 5 and L 5 Le, then the local stress must be combined 
 with the counter stress when a counter U^MQ is employed. However, it is usually 
 preferable to omit the counter member and design the main diagonal to take both the 
 direct and counter stresses. 
 
 The verticals U 2 L 2 , U 3 L 3 , etc., may be treated entirely as primary members assuming 
 the panels to be L 2 L 3} L 3 L 4 , etc. 
 
 The vertical UjLi is merely a suspender for the double panel AL 2 and its maximum 
 stress will occur when the span is fully loaded from A to L 2 . 
 
 The member M 6 V, and others of similar character often employed in large bridges, 
 are entirely secondary and perform no work as truss members. The sole purpose of 
 these members is to stiffen certain long compression members. 
 
 Bottom chord member L 2 L 3 . The influence line for this member is easily drawn. 
 The center of moments is at U 2 and the end ordinate at A is the stress in the member 
 for a unit reaction at A. The influence line thus becomes a triangle A'L 2 'B'. 
 
 Case I. With the diagonals all as tension members the local loads at the middle 
 points cannot affect the bottom chord stresses and the influence line just described 
 gives maximum stress. 
 
 Case II. Here the bottom chord stress is increased by the local load transmitted 
 from M 3 to L 2 . This is shown in Fig. 30A. For a unit load at M 3 the stress in the 
 
86 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. V 
 
 bottom chord is the horizontal component in L 2 M 3 found in the small force diagram 
 for Case II. This stress, equal to +0.36 i _must then be the additional plus ordinate 
 under M 3 and the stress influence line for L 2 L 3 thus becomes A'L 2 K'L 3 B'. as shown. 
 
 Case III is exactly like Case II, and the horizontal component of M 3 L 3 due to a 
 unit load at M 3 will be the positive ordinate to be added to the ordinary influence line 
 under K. 
 
 Case IV. Here the two members L 2 M 3 and M 3 L 3 will transmit the unit load 
 from M 3 after the manner of a three-hinged arch. The local stress in the oottom 
 
 L 2 L 3 20 
 
 chord then becomes ===== = = +0.358 for a unit load at M 3 . 
 
 4M 3 K 4X13.95 
 
 Hence the positive ordinate under K must be increased by 0.358 to determine the 
 proper influence ordinate under K. 
 
 Top chord member U 2 U 3 has four cases as just described for the bottom chord. 
 
 Case I. Here the local load at M 3 produces a compression in the top chord like 
 a suspension s} r stem. The amount of this compression for a unit load at M 3 is readily 
 found from the Eq. (27o) given for three-hinged arches and is H' =l\L 2 /fd cos a, where 
 li +I 2 =d=pa,nel length, see the truss diagram Fig. 30A. 
 
 This will give the negative stress ordinate to be added to the negative stress 
 ordinate under M 3 for the final influence line. 
 
 Cases II and III are similar and here the local effect in the top chord is the com- 
 ponent in U 2 M 3 or M 3 U 3 parallel to U 2 U 3 . 
 
 Case IV does not influence the top chord for local load at M 3 . It is similar to 
 Case I for bottom chord. 
 
 It is thus seen that the best criterion for trusses of this type is obtained from 
 influence lines and the ease and clearness with which any case is solvable speaks greatly 
 in favor of this method of analysis, where critical positions of loads and stresses are all 
 given from the one solution. 
 
 In conclusion it may be remarked that the type of panel illustrated in Case I is in 
 all respects preferable to the others. The type Case II deserves second choice, and 
 the others, especially Case IV, should be avoided whenever possible. 
 
 The panel L\L 2 is in a certain sense indeterminate and should never be used except 
 in end panels as here illustrated, where the several posts tend to steady the shock of 
 trains coming on the structure. 
 
CHAPTER VI 
 DISTORTION OF A STATICALLY DETERMINATE FRAME BY GRAPHICS 
 
 ART. 31. INTRODUCTORY 
 
 The displacement between any two points or pairs of lines of any determinate 
 frame may be found analytically as shown in Arts. 6 and 9. However, many problems, 
 to be considered later, require a complete solution for every pin point of a structure, 
 and then the analytic method would become quite laborious. 
 
 The graphic solution here given includes the two following problems: (1) To find 
 the distortion, or change in form, of a frame resulting from changes in the lengths of its 
 members, first published in 1877 by the French engineer Williot; and (2) to find the 
 effect on this distortion produced by a yielding of the supports, or reaction displacements. 
 This latter contribution came from Professor Otto Mohr in 1887, and without it the Williot 
 diagram had little real value. Hence it is proper to call the complete diagram a Williot- 
 Mohr displacement diagram, contrary to the position taken by Professor Mueller- 
 Breslau, who erroneously credits the entire subject to Williot. 
 
 Having given the changes in the lengths of all the members of a structure, by Eq. 
 (4fi), when the stresses and cross-sections are known, let it be required to find the hori- 
 zontal and vertical displacements of all the pin points. 
 
 The Williot diagram, to be presented first, offers a partial solution of this problem, 
 which is completed by Mohr's rotation diagram. 
 
 ART. 32. DISTORTIONS DUE TO CHANGES IN THE LENGTHS OF THE 
 MEMBERS BY A WILLIOT DIAGRAM 
 
 Since any determinate frame is formed by a succession of triangles, this elementary 
 frame is first examined. The process may then be extended to include any number of 
 such triangles or a complete frame. 
 
 In Fig. 32A suppose the point c to be connected with the points a and 6 by members 
 1 and 2, and that the lengths of these members also undergo changes Al and + J2, 
 respectively, while the points a and b move to new positions a' and &'. It is now 
 required to find the direction and amount of displacement resulting for the point c. 
 
 Both members are first moved parallel to themselves until they occupy the positions 
 a'c 2 and b'c i} respectively. The length of member 1 is now shortened by the amount 
 Jl because this is a negative change; similarly the member 2 is lengthened by the 
 amount J2, giving the new lengths a'c 4 and b'c 3 , respectively. 
 
 87 
 
8S KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. VI 
 
 With a' and 6' as centers and corrected lengths a'c 4 and b'c 3 as radii, describe arcs 
 which may intersect in a point c' representing the new position of the point c. But 
 because Al and Al are always very minute compared with the lengths of the members 
 to which they belong, the arcs c 4 c' and c 3 c' are replaced by their tangents, drawn respec- 
 tively perpendicular to the original directions of the members. The diagram shows 
 these displacements several hundred^ times larger than they actually are compared 
 with the real lengths of the members ac and be. 
 
 It is clear that the point c' might have been determined in a separate figure, 
 without including the members themselves, by dealing exclusively with the directions 
 and changes in the lengths of these members. Thus, in Fig. 32B, the Jl's are plotted 
 on a very large scale, affording greater accuracy in the results. 
 
 If the point is regarded as a fixed point or pole, and if from this pole the dis- 
 placements of the points a and 6 are drawn, making Oa' =aa' and Ob' =66', both in 
 magnitude and direction; also, applying at a' the shortening Jl in the length of the 
 member 1; and at b' the lengthening +J2 of the member 2; then the perpendiculars 
 
 FIG. 32B. 
 
 erected at the extremities of A\ and J2 will intersect in c' , which is the new position 
 of c. The displacement of c is then represented in direction and amount by the ray Oc'. 
 
 In laying off the values of Al, in Fig. 32s, the following rule must be observed with 
 regard to signs: If a' is regarded as fixed, then Al representing the shortening of ac, 
 it follows that c moves in the direction from c toward a, and hence Jl must be drawn 
 from a' in the direction of c to a. Likewise, if 6' is regarded fixed, then since +J2 
 represents a lengthening of be, it follows that c moves away from 6 and hence +J2 
 must be drawn from 6' in the direction of 6 to c. 
 
 Any succession of triangles, as in Fig. 32c, may be treated in the same manner. 
 It is necessary, however, to assume that one of the members, as ab, retains its direction, 
 and that some point of ab, as a, remains fixed. The changes Al in the several members 
 being given, the construction is carried out as follows : 
 
 The point 0, Fig. 32c, is the assumed pole, and since the point a is considered 
 fixed it must coincide with 0. Also, since the member ab does not change in direction, 
 then Jl is drawn parallel to ab and, being negative, it must be applied in the direction 
 of 6 toward a. The displacement of the point 6 is thus given by Ob'. 
 
ART. 33 DISTORTION OF A STATICALLY DETERMINATE FRAME 89 
 
 The point c recedes from a by an amount +J3, and approaches b by an amount 
 J2. If, therefore, J2 be drawn from b' in the direction of c toward b and parallel 
 to cb, and if J3 be drawn from or a', in the direction of a toward c and parallel to 
 ac, then the intersection c' of the perpendiculars, to the extremities of J2 and J3, will 
 determine the new position of c. The displacement of c is thus given, both indirection 
 and amount, by a ray Oc' not drawn in the figure. 
 
 The point d is joined to a and c by the members ad and dc, Fig. 32c. Its displace- 
 ment Od' is now found, in Fig. 32o, by drawing +J4 parallel to ad from and in the 
 direction a to d; also by drawing J5 parallel to erf from c' in the direction d to c. 
 The intersection d' , of the two perpendiculars to the extremities of J4 and J5, is the 
 new position of d. Finally the new position e' of the point e is similarly found by 
 applying the same method. 
 
 I 
 \ 
 
 \ 
 \ 
 \ 
 \ 
 
 FIG. 32o. 
 
 Fig. 32o, whose polar rays Ob', Oc' , Od' , and Oe' give the displacements of the several 
 points b, c, d and e, both in direction and magnitude, is called the "Williot displace- 
 ment diagram," for the frame abcde, giving credit to the name of its originator. 
 
 ART. 33. ROTATION OF A RIGID FRAME ABOUT A FIXED POINT PROFESSOR 
 
 MOHR'S ROTATION DIAGRAM 
 
 In the Williot diagram it was assumed that the member ab did not alter its position, 
 but merely changed its length. However, this is not always the case and more frequently 
 this member ab is also subjected to a rotation about some instantaneous center. In 
 the latter event, the elastic displacements, given by the Williot diagram, will require 
 further changes to make them comply with the initial change in the position of the 
 member ab. It is supposed, for the present, that the frame has undergone its elastic 
 deformation and has passed into a rigid state. 
 
 The solution of this phase of the problem was first proposed by Professor Mohr and 
 is based on the following theorem in mechanics: The motion of a rigid body, at any 
 given instant, may be defined as a rotation about a certain point called the instantaneous 
 center of rotation, and the direction of motion of any point of this body, at the instant in 
 
90 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. VI 
 
 question, will be perpendicular to the line joining such point with the instantaneous center 
 of rotation. 
 
 Thus, if abcde, Fig. 33A, represents a rigid frame rotating about the instantaneous 
 center P, each arrow, perpendicular to its respective ray, will then represent the direction 
 of motion of the point to which the ray is drawn. 
 
 Now, if from any pole 0, Fig. 33B, a series of radial rays be drawn respectively 
 parallel to the arrows representing the motions of the several points, then these rays 
 Oa", Ob", Oc", Od", and Oe" will in turn represent, both in magnitude and direction, 
 the several displacements of the several points, a b, c, d and e. The figure a"b"c"d"e" 
 will be the new position of the rigid frame abcde. 
 
 For, a"0-LaP; b"0-LbP; c"0-LcP; etc., because the direction of motion of each 
 such point of the rigid frame was perpendicular to the ray joining this point with the 
 instantaneous center of rotation. 
 
 Also. a"0:b"0:c"0, eic.=aP:bP:cP, etc., because the displacements of the several 
 points a, b, c, etc., are respectively proportional to their velocities, and these in turn 
 
 FIG. 33A. 
 
 FIG. 33n. 
 
 are proportional to the distances of the several points from the instantaneous center 
 of rotation. 
 
 From these conditions it. follows that: 
 
 (a) If the points a", b", c" , etc., of the diagram Fig. 33B, are joined by straight 
 lines, so that for every member ab of the frame, there will be a corresponding line a"b" 
 in the displacement diagram, then these latter lines will produce a figure a"b"c"d"e" 
 which will be similar to the rigid frame. 
 
 (6) Also, the line joining any two points of the rigid frame, as ab, will be perpen- 
 dicular to the corresponding line a"b" of the displacement diagram. 
 
 Hence, if it is possible to determine the position of any two of the points a", b", 
 c", etc., of the displacement diagram, then the figure a"b"c"d"e" can be drawn by- 
 similarity with the figure abcde. The method of determining two such points a" ', b" 
 will be illustrated in the examples. 
 
 Therefore, the secondary displacements, due to rotation of the rigid frame, are 
 found by inserting in the Williot displacement diagram a figure a"b"c"d"e" similar 
 to the frame abcde, and having its sides respectively perpendicular to the sides of the 
 
ART. M DISTORTION OF A STATICALLY DETERMINATE FRAME 91 
 
 latter frame. The original displacements Ob', Oc', etc., are then combined with the 
 displacements Ob", Oc", etc., due to rotation to produce the resultant displacements 
 V'J', c r '7, etc. 
 
 The complete solution for the displacement of the pin points of any framed 
 structure, definitely supported, may then be outlined in the three following steps: 
 
 (a) Assume any member and a point on the axis of this member as fixed and 
 construct a Williot displacement diagram. 
 
 (b) Then assume the frame as rigid and subjected to a rotation such as will 
 conform to the actual conditions of the supports. 
 
 (c) The displacements of the several pin points may then be found, in magnitude 
 and direction, by scaling the distances between the m" and m' points. The direction 
 of any displacement will always be from m" toward m' '. 
 
 The complete diagram, combining the Williot displacement diagram with the Mohr 
 rotation diagram, will hereafter be known as a Williot-Mohr diagram. 
 
 The analytic solution of deflection problems is conducted with the aid of Professor 
 Mohr's work, Eq. (GA), and may be advantageously employed when only one point of 
 the structure is to be treated, or when great accuracy is required in the solution. 
 
 This method is fully discussed in Art. 6 and, in combination with Maxwell's theorem 
 given in Art. 9, offers the most accurate analytic solution of a great variety of problems 
 dealing with displacements of points and lines. 
 
 However, all problems requiring a complete solution for all the pin points of a 
 structure, may be solved by a Williot-Mohr displacement diagram when extreme 
 accuracy is not necessary. 
 
 Attention is called to the fact that the value for E, in all deflection problems, 
 should be chosen low rather than high, because there is always a slight amount of lost 
 motion and permanent set, which follows the first loading of a new structure and which 
 is not truly an elastic deformation. This can be compensated for in the calculations 
 by assuming a small E. 
 
 As a general rule, the changes Al, in the lengths of the members, are best figured 
 on gross sections rather than on net sections. Experience indicates that this gives 
 results nearer the truth. Also the stresses in the members must be simultaneous and 
 not maximum. 
 
 ART. 34. SPECIAL APPLICATIONS OF WILLIOT-MOHR DIAGRAMS 
 
 (1) Distortion of a simple truss. Fig. 34A has been solved in three ways, for the 
 purpose of illustrating that the displacements are not affected by the choice of the 
 member and point which are regarded as fixed in position, and also in order to show 
 how to select the most convenient of the possible forms of solution. 
 
 The truss is supported on rollers at e and is fixed at a. The extensions (+) and 
 contractions ( ) of the members are assumed to be as marked upon them. 
 
 The solution given in diagram 6 is made under the supposition that the direction 
 of the member af and the point a remain fixed. 
 
 According to the method outlined in Art. 32, a displacement diagram is constructed 
 
92 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. VI 
 
 x DOUBLE *C/UE. / 
 
 y 
 
 /(d)\ 
 
 OF DISPLACE MCNTS. 
 
 i , , , i , . . , i ,,,,11 
 
 o i 2! 
 
 FIG. 34A. 
 
ART. 34 DISTORTION OF A STATICALLY DETERMINATE FRAME J3 
 
 with pole at 0. Since a is now supposed to remain fixed, the point a' coincides with 
 0, and the displacements of the various points 6, c, d, e and / are obtained by scaling 
 the distances Ob 7 , Oc r , Od' . . .Of. 
 
 In reality, however, the direction of the member uf does not remain fixed, for the 
 member revolves about the point a. Therefore, the displacements just found must 
 be combined with those displacements 06", Oc", Od", etc., which are caused by 
 revolving the rigid frame, or truss, abge about the point a until the point e shows a 
 resulting displacement, e'e" ', parallel to the direction of motion of the roller bed at e. 
 
 In other words, the resultant movement of the point e will be horizontal when 
 the roller bed is horizontal. If the bed were inclined, as em, diagram a, then the point 
 e would move parallel to em. 
 
 The figure a"b"c"d"e"f"g"h", similar to the figure of the truss, can then be con- 
 structed, making the sides respectively perpendicular to the members of the truss. 
 This figure is definite and can be drawn when the points a" and e" are found. The 
 point a" will coincide with a' or 0, since it remains fixed and a"e" will be perpendicular 
 to ae. Also e'e" will be parallel to the roller bed, which is horizontal in the present 
 case, but may be in any direction, as e'm" ', for a skew-back. 
 
 The actual displacements of the points 6, c, d, etc., will then be given in direction 
 and in amount by the distances &"&', c"c', d"d f , etc., but the horizontal and vertical 
 projections of these displacements are more generally desired and may be scaled from 
 the drawing. 
 
 The deflection polygon of the bottom chord is found graphically by projecting 
 the points a', a", 6', b", etc., on to the verticals through the panel points a, 6, etc. 
 The points a", 6", c", etc., will be projected in a'", 6"', c'", etc., and will form a straight 
 line a'"e'" . The points 6', c' , d' , etc., will fall in 6 , c , d , etc., and the ordinates 6 6"', 
 C(,c'" , etc., will give the vertical deflections of the panel points 6, c, etc. 
 
 In diagram c the direction cb is assumed and the point c is fixed. All that has 
 been said regarding the first solution applies here. It will be seen that this solution 
 gives exactly the same displacements as previously found, while it occupies only about 
 half the space of the first diagram because the member be has a lesser angular motion 
 than the member af. 
 
 The third solution, diagram d, is the simplest of all, and its diagram covers the 
 least area; for the member eg, which is now assumed as fixed, j-eally has no angular 
 motion, but simply drops vertically. The relative displacement b'g' of any two points, 
 6 and g, may be scaled off directly. Although this displacement diagram was drawn 
 on the assumption that the line eg and the point c remained fixed, nevertheless any 
 two points may be directly compared. Hence, any point may be chosen as the origin 
 of coordinates from which to scale the horizontal and vertical displacements of all other 
 points relatively to this origin. Naturally the displacements desired would be those 
 with respect to the point a, for this is the fixed point, giving for the point e a horizontal 
 movement from a' to e'. All other points move to the right and down from their 
 original positions by amounts which may be scaled from the diagram, taking a' for the 
 origin. 
 
 Hence it may be concluded that if a truss contains a member which will move 
 
94 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. VI 
 
 parallel to itself, or which has no angular motion, then this member is the best one to 
 choose as the fixed member in constructing the displacement diagram. The rotation 
 displacements are thereby eliminated. 
 
 (2) A French roof truss. Fig. 34s. In this example the stresses were found 
 from a Maxwell diagram, and the corresponding changes in the lengths of the members 
 were obtained by means of the formula Jl=Sl/EF, taking E at 25,600,000 pounds per 
 square inch for wrought iron. Changes of length, due to temperature, are neglected. 
 The values of Al being very small, ten times these values were taken. All the data 
 necessary in the solution of this truss are given in the following table: 
 
 TABLE OF THE VALUES OF S, F, I, AND 10JZ IN FIG. 34n 
 
 Member. 
 
 Stress, 
 Lbs. 
 
 S 
 
 Area, 
 Sq. in. 
 F 
 
 Length, 
 Inches, 
 I 
 
 10UO, 
 Inches. 
 
 Member. 
 
 Stress, 
 Lbs. 
 S 
 
 Area, 
 
 Sq. in. 
 F 
 
 Length, 
 inches. 
 I 
 
 10(40, 
 Inches. 
 
 1 
 
 -21,800 
 
 6.82 
 
 83.53 
 
 -0.106 
 
 15 
 
 -20,970 
 
 6.82 
 
 83.53 
 
 -0.102 
 
 2 
 
 -20.040 
 
 6.82 
 
 83.53 
 
 -0.094 
 
 16 
 
 -19,210 
 
 6.82 
 
 83 . 53 
 
 -0.091 
 
 3 
 
 - 18,300 
 
 6.82 
 
 83.53 
 
 -0.087 
 
 17 
 
 -17,470 
 
 6.82 
 
 83.53 
 
 -0.083 
 
 4 
 
 -16.540 
 
 6.82 
 
 83.53 
 
 -0.079 
 
 18 
 
 -15,710 
 
 6.82 
 
 83.53 
 
 -0.075 
 
 5 
 
 + 19,730 
 
 2.02 
 
 96.53 
 
 + 0.370 
 
 19 
 
 + 15,330 
 
 2.02 
 
 96.53 
 
 + 0.288 
 
 6 
 
 + 16,430 
 
 2.02 
 
 96.53 
 
 + 0.307 
 
 20 
 
 + 13,570 
 
 2.02 
 
 96.53 
 
 + 0.256 
 
 7 
 
 + 9,040 
 
 1.86 
 
 100 . 08 
 
 + 0.189 
 
 
 
 
 
 
 8 
 
 + 9,260 
 
 1.86 
 
 96.53 
 
 + 0.189 
 
 22 
 
 + 6,200 
 
 1.86 
 
 96.53 
 
 + 0.126 
 
 9 
 
 + 12,560 
 
 1.86 
 
 96.53 
 
 + 0.256 
 
 23 
 
 + 7,960 
 
 1.86 
 
 96 . 53 
 
 + 0.161 
 
 10 
 
 - 3,300 
 
 1.40 
 
 48.07 
 
 -0.043 
 
 24 
 
 - 1,680 
 
 1.40 
 
 48.07 
 
 -0.024 
 
 11 
 
 + 3,300 
 
 0.78 
 
 96.53 
 
 + 0.162 
 
 25 
 
 + 1,680 
 
 0.78 
 
 96.53 
 
 + 0.087 
 
 12 
 
 - 6,600 
 
 2.48 
 
 96.53 
 
 -0.102 
 
 26 
 
 - 3,520 
 
 2.48 
 
 96.53 
 
 -0.055 
 
 13 
 
 + 3,300 
 
 0.78 
 
 96.53 
 
 + 0.162 
 
 27 
 
 + 1,680 
 
 0.78 
 
 96.53 
 
 + 0.087 
 
 14 
 
 - 3.300 
 
 1.40 
 
 48.07 
 
 -0.043 
 
 28^ 
 
 - 1,680 
 
 1.40 
 
 48.07 
 
 -0.024 
 
 The truss is composed of the two frames aeh and sqh, designated as I and II. 
 These are connected by the member as and by the pin at h. 
 
 The displacement diagram of frame I is first drawn, assuming as fixed the direction 
 of any member, as ab, and the position of a point, as a, of this member. The point a' 
 then coincides with the pole 0, and the displacement Ob' , of the point b will be equal 
 to J12. The points c' , d', e*, f, g' and h' are then found as directed in Art. 32. 
 
 The displacement diagram of frame II is next drawn, assuming as fixed the direction 
 of the member sk and the position of the point s. 
 
 Having thus determined the points k' , I', m', q' , ri ', p' and h' in diagram II, the 
 relative changes in the positions of the points h, s and q, parallel to the straight lines 
 joiningjthe points h and s, h and q, and s and q, may be found. These changes are called 
 Ahs, Ahq and Asq. 
 
 Diagram I may now be completed by inserting the values Ahs, Ahq and Asq previously 
 found from diagram II. The point s' is found from Al and Ahs, and the point q' is 
 found from Ahq and Asq. Since q moves on a horizontal roller bed and since e is fixed, 
 the figure c"d"b"g"h"f"a"s"q" can be drawn as in the preceding problem. This figure 
 gives the displacements of all the points of frame I, and those of the points s and q. 
 
ART. 34 DISTORTION OF A STATICALLY DETERMINATE FRAME 
 
 The displacement diagram of frame II may now be completed by transferring the 
 displacements of the points h and q from diagram I into diagram II, thus determiniaf 
 
 Lj_ 
 loo 
 
 \ DIAGRAM I. 
 
 \ N 
 
 DISPLACEMENTS. 
 
 i 
 
 DIAGRAM H. 
 
 too zoo 
 
 FIG. 34s. 
 
 the points q" and h" from which the figure q"l"s"ri'h"m"k"p" can be drawn. 
 
 As a check, it should be remembered that the line q"h" in diagram II must be 
 
96 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP, vi 
 
 perpendicular to the line hq, and that the displacement s's" of the point s must be the 
 same, both in direction and in amount, in both diagrams. 
 
 Diagram II might have been dispensed with in the present case, as the values dhs t 
 Ahq and Asq might have been found directly by summing the J/'s. However, the use 
 of diagram II is general, and it becomes necessary when the points h, p, k, m and q, 
 or h n and s, or /, s and q, are not in straight lines, as in the case of a curved top chord. 
 
 Measurements from diagram I show that the point g undergoes the greatest 
 displacement, having a horizontal movement of 1.93/10=0.193 inch to the right, 
 a vertical downward movement of 2.33/10 =0.233 inch, or a resultant movement of 
 oV 7 =3.02/10 = 0.302 inch. The horizontal movement of the point q is = q'q" =2.61/10 = 
 
 FIG. 34c. 
 
 0.261 inch. The displacements given by the diagram are here divided by 10, because, 
 as already stated, the changes Al were originally taken ten times too large. 
 
 (3) A three-hinged arch. Fig. 34c. It is required to draw the displacement 
 diagram. 
 
 Independent diagrams for each of the elastic frames, I and II, are first drawn by 
 assuming as fixed the direction of, and a point on, some member of each frame, as was 
 done in the previous problem. 
 
 The frames I and II are then regarded as rigid, and each frame is supposed to 
 revolve in such a way as to satisfy the conditions imposed by the supports. 
 
 The displacement diagrams are omitted and only the second part of the problem 
 is solved. 
 
ART. 34 DISTORTION OF A STATICALLY DETERMINATE YRA ME 97 
 
 Supposing that the points a' and c' in Fig. 34c, diagram I, and the points b' and 
 c', diagram II, have been found from the corresponding displacement diagrams (not 
 shown) , let it be required to complete the solution by adding Mohr's rotation diagram. 
 
 The following conditions must exist between the figures a"c" and b"c" , which are 
 to be similar to the frames I and II respectively: 
 
 a. The displacement of the point a is zero, hence a" will coincide with a'. 
 
 b. Similarly b" will coincide with b'. 
 
 c. The line a"c", in diagram I, must be perpendicular to the line ac. 
 
 d. The line b"c" , in diagram II, must be perpendicular to the line be. 
 
 e. Both diagrams, I and II, must give the same displacement c'c" for the point c. 
 
 Hence a"c" is drawn through a', perpendicular to ac, also b"c" through b' perpen- 
 dicular to be. Now, in diagram I, c'n is drawn parallel to ac, intersecting a"c" in n, 
 and the projection of the required displacement c'c", parallel to ac, is thus obtained. 
 Likewise, in diagram II, c'm is drawn parallel to cb, intersecting b"c" in m, giving c'm, 
 the projection of c'c" parallel to cb. 
 
 Diagram I may now be completed by transferring c'm from diagram II and erecting 
 a perpendicular to c'm at m. This perpendicular will intersect na"c" in c", and the 
 figure a"c" can now be drawn by similarity with the frame I, since the members in the 
 two figures are respectively perpendicular. 
 
 In like manner diagram II may be completed by transferring c'n from diagram I 
 and drawing nc" perpendicular to c'n, thus determining c". As a check, the displace- 
 ments c'c" must be equal and parallel in the two diagrams. 
 
 (4) A cantilever bridge, similar in principle to the Memphis bridge, is represented 
 in Fig. 34D. 
 
 It is assumed, as in the two preceding problems, that separate displacement 
 diagrams have been made for each of the elastic frames, I, II, III, and IV, and of these, 
 the diagrams for I and III can be completely solved, as was done with the example in 
 Fig. 34A, since each is a determinate framed structure on two supports. 
 
 It is here deemed necessary only to complete the rotation diagrams of the frames 
 II and IV. The displacements of these frames depend on those of the points c and m 
 and upon that of the point s, respectively. 
 
 Let it be assumed that the points d' and n' have been determined for the elastic 
 frame II as in diagram II, and let it be required to find the figure d"n" which shall be 
 similar to frame II. 
 
 The members cd and ran are elongated by Jl and J2 respectively, and their elonga- 
 tions must be applied to the^points d' and n' in diagram II, giving the points c' and m'. 
 Now if the displacements ~c'c" and m'm" be drawn as found by diagrams I and III (not 
 shown), the points c" and ra" are obtained. These represent the original positions of 
 the points c and ra. The point n" must have originally occupied the same relative 
 level with ra", also d" must have been_at the same relative level with c". Lastly, the 
 line <Fn" must be perpendicular to dn. Hence, if the distance x, or the horizontal 
 distance between d' and d", can be found, then the figure d"n" becomes determinable 
 and the diagram can be completed. If it is granted that the point w will always be 
 midway between the points c and ra, this problem becomes definite, and the value x 
 
98 
 
 KIXETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. VI 
 
 may be found from the figure, thus: x=^(e~ g +f) , which is (+) when measured to 
 the right. 
 
 Now assuming the displacement diagram of the frame IV drawn, and the points 
 v' and t' determined, as shown in diagram IV, required to find the figure v"t" . 
 
 The point v is a fixed support, hence v" must coincide with v' ; also the direction of 
 v"t" must be perpendicular to vt. To find t", apply J3 from t' to s', and then transfer 
 the displacement sV from diagram III (not shown) into diagram IV, thus giving the 
 
 IT 
 
 
 FIG. 34o. 
 
 original position of s". The point t" must have dropped from the height s", and must, 
 therefore, be on the same horizontal line with s". Hence, t" is at the intersection of 
 tf't" and the horizontal through s", and the figure v"t", similar to the frame IV, can 
 then be drawn. 
 
 NOTE. For other problems see paper on "The Graphical Solution of the Distortion of a Framed 
 Structure" by the Author in the Journal Ass'n. Eng. Societies, June, 1894. 
 
CHAPTER VII 
 
 DEFLECTION POLYGONS OF STATICALLY DETERMINATE STRUCTURES BY 
 
 ANALYTICS AND GRAPHICS 
 
 ART. 35. INTRODUCTORY 
 
 The deflection polygon for any series of panel points, either of the top or bottom 
 .chord, may be found from a Williot-Mohr diagram as shown in the previous chapter. 
 
 However, when the deflections, so obtained, are intended to form the basis for stress 
 computations, then greater accuracy is actually required than that obtainable from the 
 above method of solution. This is especially true of large structures, because the error 
 committed by substituting the tangent of an arc for the arc itself is of a cumulative 
 nature, always giving excessive values. This fact has been overlooked by most authors 
 treating of indeterminate structures. 
 
 Professor Mohr, in 1S75, established a very important relation between deflection 
 polygons and equilibrium polygons drawn for simultaneous cases of loading. Accordingly 
 a deflection polygon may be derived from a moment diagram by dividing the ordinates 
 of the latter by a certain constant which depends on the geometric shape of the structure. 
 
 The elastic deformation in the angle included between two successive chord members 
 can always be expressed in terms of the coexisting changes in the lengths of the truss 
 members or in terms of the stresses in these members for any simultaneous case of 
 loading. Such an angular change may be regarded as an elastic load w, which designation 
 will always be used in the following chapters. 
 
 The several elastic loads w for all pin points of an entire chord will then suffice 
 to determine the deflection polygon for that chord precisely as the external loads P will 
 determine an equilibrium polygon from which moments and stresses in the chords may 
 be found. 
 
 Naturally such a deflection polygon will furnish displacements in one fixed direction 
 only, and this is usually all that is required. However, two such polygons, for horizontal 
 and vertical deflections, respectively, would, if drawn for the same pin points and same 
 loading, determine actual displacements. 
 
 Since the total deflection of a truss is almost entirely due to the deformation in the 
 chords, it is often admissible to neglect the web system. For arches with parallel chords 
 this is always true. 
 
 The deflection polygon is usually drawn for the loaded chord only. If the loaded 
 chord is straight, as is often the case, then the computations are greatly simplified. 
 
 In the following, two methods will be given for finding such deflection polygons. 
 In the first method the elastic loads w are derived from the changes Al of all members 
 
 99 
 
100 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP, vn 
 
 of the structure, according to Professor Mohr. The second method is a modification of 
 the first, given by Professor Land, in 1887. and expresses the w loads in terms of the unit 
 stresses of the several truss members. 
 
 ART. 36. FIRST METHOD, ACCORDING TO PROFESSOR MOHR 
 
 (a) Influence on the deflections due to chord members. For any particular case of 
 loading, the stresses and changes in the lengths of the several members are supposed 
 to be known. Neglecting the web members for the present, by regarding their changes 
 Jl all zero, the influence of a single chord member is first considered. See Fig. 36A. 
 
 The deflection y m of any lower chord point m, due to a shortening in the upper 
 chord member opposite the point m, may be found from Mohr's work equation. Let 
 a unit force act downward from m (if vertical deflections are desired) . Then from 
 Eq. (GA), l-ym^Sma^lm, when there are no abutment displacements. 
 
 Let M ma represent the static moment about the point m when only the unit load 
 at m is acting. This is easily found from either one of the reactions, resulting from 
 the unit load, multiplied by the distance from m to that reaction. Then the stress 
 S ma =M ma /r m as may be seen from Fig. 36A. 
 
 Now if \-y m be regarded as a moment M mw due to a load w hung at the point m, 
 then by proportion 
 
 W m M mw l-y 
 
 ~ = Wm = 
 
 S 
 
 ^ 
 
 The load w, thus expressed, will be known as an elastic load. 
 
 Hence, a moment diagram drawn for the elastic load w m acting at m, will give the 
 ordinate M mw =y m under the point m. Laying off w m on the load line to the right and 
 with a pole distance H=-l, to scale of forces, drawing the moment diagram A'm'B', 
 then the ordinate y m represents the deflection of the point m due to a change M m in the 
 length of the member fg opposite the point m, because 
 
 M, = 1 * =SmM m . 
 
 mu , 
 
 Now the polygon A'm'B' is also the influence line of deflections for the point m for 
 any moving load w m . Hence, the deflection y nm , at the point n, is that due to a load 
 w m at the point m, and so on for any pin point of the chords. 
 
 In like manner the deflection y n is obtained for a load w n =Al n /r n and the polygon 
 A'n'B' is then the influence line of deflections for the point n for a moving load w n . 
 The ordinate y mri} under the point m, is now the deflection at m for a load w n at n. 
 But, by Maxwell's law y^ n = y nm . 
 
 By drawing these influence lines successively for all the pin points of the "top and 
 bottom chords, the several partial effects of all the elastic loads w on any particular 
 point as m may be found, and the summation of these effects 2?/ mn will be the total 
 deflection d m for the point m resulting from the given case of actual loading. 
 
 The summation, however, is more easily performed by combining the several loads 
 w into a load line 21 w and with a pole #=!, drawing the force and equilibrium polygons. 
 
ART. 36 DEFLECTION POLYGONS BY ANALYTICS AND GRAPHICS 
 
 101 
 
 The latter is a moment polygon and also a summation deflection polygon, any ordinate 
 d m of which represents the actual deflection of the corresponding pin point m for that 
 particular case of loading and is measured to the scale of lengths. This does not 
 include the effect of the web members. 
 
 If the external loads are applied to the lower chord pin points then the true deflec- 
 tion polygon for the lower chord is the inscribed dotted polygon A'c"m"n"B' ', because 
 the deflection polygon between the successive floor beams must be a straight line. 
 
 When the loads w are multiplied by E and the scale of lengths is chosen 1 : a, then 
 by making the pole distance H =E to the scale of forces, the deflections will have the 
 scale 1 : a. When the deflections are. desired TO times actual to the scale I: a, then the 
 pole should be chosen E/m to the scale of forces. 
 
 FIG. 36A. 
 
 As is seen from the above, the elastic loads w are ratios; that is, they have no 
 dimension but depend on the unit stresses in the members and the geometric shape of 
 the structure. From this and the real value w = M/r, it follows that these elastic loads 
 may also be defined as tangents of the angular changes in the angles 6, or simply 
 w m =J6 m from Fig. 36A. 
 
 For a top chord point, the angular change Ad is positive, while for a lower chord 
 point it is negative, but in either case the angular change produces downward deflection. 
 
 For a top chord point 
 
 M . Al _ . Mp 
 
 
 and 
 
 '= . hence 
 
 and for a bottom chord point 
 -M=S and 
 
 S=- 
 
 M 
 
 hence again 
 
 Mp 
 
102 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP.VII 
 
 Therefore, for any chord point of a simple truss 
 
 ^Jl m = M m p m= M, 
 
 '' r m r 2 = +IF; 
 
 (36s) 
 
 which is the fundamental equation for the elastic loads w, due to chord members only. 
 
 (b) Influence on the deflections due to web members. In considering the effect of the 
 web members, the influence of a single diagonal is first analyzed for the case when the 
 center of moments falls outside the span, Fig. 36s. The stress S a changes sign while 
 
 V, 
 
 to- 
 
 FIG. 36B. 
 
 the moving load passes through the panel un, hence it is necessary to consider whether 
 the top or bottom chord is loaded as this determines the load divide. 
 
 Let mn be the member in question and the lower chord carries the roadbed. Required 
 to find the deflections of all the lower pin points for any given system of loads. See 
 Fig. 36s. 
 
 The influence line for the deflection of a lower chord point is determined when the 
 deflection for that point is known. Having this influence line, all the other required 
 deflections are easily found. This will now be shown. 
 
 The deflection y c of any chord point c, as previously found for a unit load at c, is 
 l-y c =S a Jl for which is found an elastic load w c = Al/r, wherein I and r refer to the 
 member mn with center of moments at o for a section pq. 
 
ART. 30 DEFLECTION POLYGONS BY ANALYTICS AND GRAPHICS 103 
 
 This elastic load produces reactions A a and B a and, as previously shown, the moment 
 M wo (f r a U tne f rces on one s id e f tne se tion ail d f r tne center of moments o) is 
 found to be M WO =S a M = l-y c - 
 
 Hence the influence line for the moment M wo , for_a moving load w = Al/r, may be 
 found. By laying off T/^cV 7 as an ordinate from a'b' under the point c to any con- 
 venient scale, and drawing a line oV 7 to intersect the vertical through o in the point o', 
 the two limiting lines of the influence area are determined. Thus, a'u'o' represents the 
 influence of B a and o r b r a/' f thejnfluence of A a . Between the panel points n and u the 
 influence line is a straight line n'u'. 
 
 Also, for any other position of this elastic load the same influence line will be found. 
 Hence the most suitable point of application for this elastic load will be the load divide 
 i' for the panel whose section is under consideration. This would make any line through 
 the load divide i' an influence line for the elastic load w. Such a line would intersect 
 the verticals through the two adjacent panel points in points n' and u' and the two 
 limiting lines ~rW and oV must be made to intersect in the point o' which is any con- 
 venient point in the vertical through o. These two limiting lines cut off the closing line 
 tfV on the verticals through A and B. Then any load w=M/r, applied at the load 
 divide i, will produce a zero deflection in the corresponding point c and a zero moment 
 M wo about the point o. 
 
 Now the elastic load w may be replaced by two such loads w n and w u , acting at the 
 points n and u, respectively. These may be determined graphically by drawing the 
 force polygon Fig. 36B, corresponding to the moment diagram a'u'n'V. 
 
 The analytic method for finding w n and w u is as follows: The position of w is 
 evidently immaterial, because the resulting loads w n and w u will be the same for any 
 position of w. If the load divide is chosen as the point of application, then the moment 
 M WO =Q and the sum of the moments produced by the loads w n and w u must likewise 
 belero. Hence, the resultant of the two components must pass through the center 
 
 of moments o. 
 
 Also, by drawing a line nh\\um, Fig. 36s, and remembering that the resultant o 
 w n and w u must equal w, then by taking moments about 71 
 
 Ai _ M no M , no r 
 
 w u Xun=--no or w u = -'==- because ==- 
 r r un ' un u 
 
 wo r . 
 Similarly by taking moments about u, and noting that j-- , t 
 
 jl . Al uo M mo _M 
 
 w n Xun=-uo or > = .-= = =---. 
 r r un r mh r 
 
 The two new elastic loads, which were substituted for w, may now be evaluated 
 from the following equations, without involving the lever arm r, thus: 
 
 Wu =* and w n (36c) 
 
104 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. Yii 
 
 In most cases this simple relation proves of valuable assistance, because r is 
 frequently outside the limits of the drawing. 
 
 Regarding the signs of the two elastic loads iv u and w n it is discerned from the above 
 that the signs must always be opposite and it is necessary to distinguish which of the 
 two is the positive load and then call the other one negative. The sign of Al alone 
 determines this without regard to the sign of the work S a Al and Professor Mohr gives 
 the following simple rule, which will always identify the positive elastic load w. 
 
 Calling all top chord members negative and all bottom chord members positive and 
 giving the proper sign to Al for the web member in question, then the positive w is found on 
 that side of the section or panel where the sign of Al coincides with that of the adjacent chord. 
 
 The same result is obtained from the force polygon, Fig. 36s, drawn for the three 
 loads w. This is also seen when the distortion in the quadrilateral ingun, due to the 
 change Al in the member mn, is considered. Assuming the top chord mg immovable, 
 then Al in mn will cause the point u to drop, hence w u is positive when the bottom 
 chord is the loaded chord. Were the top chord the loaded chord, then nn would be 
 considered immovable in applying this reasoning. 
 
 When the center of moments o falls inside the span there is no load divide, whence 
 the deflection influence line for a web member can have only a positive area. The 
 maximum stress is then due to the total span loaded, the same as for maximum moments. 
 
 However, this in no wise affects the previous discussion, but some proof of the 
 identity of the effects of the substitute loads w u and w n with that of w must be furnished 
 for this case. 
 
 Thus the elastic load w=Al/r, acting at c, is in equilibrium with the reactions A 
 and B resulting from w, and it is also the resultant of the two substitute loads w u and 
 w n . Hence the sum of the moments of the forces, A, B, w u and w n about any point o 
 in the plane of forces, must be zero. But the resultant w of w u and w n , always passes 
 through the center of moments o, hence this resultant must be equal and opposite to 
 the resultant of A and B. The method is thus general and applies equally to cases of 
 o inside and outside the span. 
 
 The elastic loads w. u and w n may also be expressed in terms of the moments of the 
 external forces, as was previously done for the chords, and the following general expres- 
 sions are thus obtained: 
 
 Al Mp Al Mp 
 
 w u = = and w n = = =F--; 
 
 . . . . . . . (36n) 
 
 U ^ r n r n r '~ '* 
 
 also 
 
 Al , Al Al Mp . 1,11 
 
 w ~= -*- = -f> hence -=- T - ...... (36 E ) 
 
 v 'n ' > it r n 
 
 This condition is easily seen directly from Fig. 36B. 
 
 (c) The deflection polygon for the loaded chord. This deflection polygon may be 
 found by applying the formulae (36B) and (36o) to any given frame. The elastic loads 
 w are now computed for each chord and each web member and all those acting at the 
 same pin point are algebraically added for total effects. 
 
 The deflection polygon is then found by combining the loads w into a force polygon 
 and drawing the moment or equilibrium polygon exactly as shown in Fig. 36A, following 
 
ART. 36 DEFLECTION POLYGONS BY ANALYTICS AND GRAPHICS 105 
 
 the description relating thereto. These moments may also be computed to obtain the 
 deflections analytically. 
 
 Before concluding this subject a few special cases will be illustrated. 
 
 (1) When there is no end post, as in Fig. 36B, then Hitter's section cuts only two 
 members 'ad and ac and some doubt may exist regarding the application of the formula 
 
 for w. 
 
 To overcome this difficulty, add a member ak\\cd and of any_convement length. 
 Then the center of moments o l will be the_center for the member ad treated as a web 
 member. The two imaginary members ak and kd are assumed rigid and Eqs. (36D> 
 again apply. The load w thus found for the point a will have no effect, and the result 
 is the same as would be found by applying Eq. (36n) to the center of moments at c 
 
 with the lever r\. 
 
 (2) When there is a vertical end post and the deflection polygon for the lower chord 
 is required, then a load applied at a, Fig. 36c, will not produce stress in the verticals 
 nor end post. The end chord section, and these members, therefore, do not influence 
 the deflection polygon of the lower chord. 
 
 t\ 
 
 FIG. 36c. 
 
 If the load is applied at the upper point c, then it is best to construct the deflection 
 polvgon by regarding the Al=Q for the two end posts. Now since the whole top chord 
 is lowered by amounts -M and -M, of these vertical end posts, it is merely necessary 
 to raise the closing line of the deflection polygon by these amounts at the two respective 
 ends. The deflection ordinates are thus corrected by the addition of the end pos 
 
 (3) Frames with vertical posts, as in Figs. 36c and 36D. 
 
 When the center of moments cannot be ascertained by a section through three 
 members, as for the member w', Fig. 36D, the elastic load for the point a' cannot 
 
 directly found. ,, 
 
 To overcome this difficulty assume the disposition in Fig. 36E, where the small 
 
 chord dl is interposed at a' and the vertical aa' is split into two posts such that the 
 
 deflection remains the same as for the original case. Then the original solution is again 
 
 applicable by passing two sections as indicated and cutting three members by each section 
 
 The vertical post is now made up of two substitute posts for which the work mu* 
 
 be equal to that of the original post, thus: 
 
 SI \ 
 
 4-2-1 
 
 ^2l EF I 
 
10G 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP, vn 
 
 Also for each substitute there will be a load w active at the center of moments o 
 and the resultant of these two loads will be w a , Fig. 860. 
 
 Hence, 
 
 Jl Al 
 
 In the ordinary case, Fig. 36c, the deflection polygons for top and bottom chords 
 are alike when the web system is neglected. When the latter is included it is best to 
 construct the deflection polygon of the chord which ends in the supports (the lower 
 chord in this instance) and the deflection of the other chord is then found by simply 
 correcting the first polygon by the Al of the verticals, being careful to regard the sign. 
 
 (4) Trusses with parallel chords become very much simplified because the height 
 of the truss and the inclination of the web members are then uniform, Fig. 36F. 
 
 FIG. 36F. 
 
 Hence for all chord members, Eq. (36s) gives 
 
 Jl 
 
 (36F) 
 
 For the web members, the center of moments is at infinity, making the elastic loads 
 equal and opposite for the same section. Each pair thus constitutes a moment which 
 is balanced by an infinitesimally small force acting at an infinite distance. For this 
 and other reasons it is better to determine the elastic loads for the web members directly 
 from the shear. Calling the shear Q, Eq. (36o) then gives 
 
 (36o) 
 
 Here r u =r n for each panel and <f> is the angle of the diagonal with the vertical. 
 The figure shows the case where the w loads are desired for the bottom chord panel points. 
 
 When <j6 is constant for all web members then the factor l/Er becomes constant 
 and may be applied to the scale of the elastic loads w, which are then l/Er times natural 
 size. Thus by making the pole distance Er/l times larger, the resulting deflection 
 polygon for web members will be unchanged. 
 
 (5) Composite structures, like three-hinged arches and cantilever systems, may 
 likewise be treated by the above method. 
 
 111 
 
 Al 
 
 SI 
 
 Ql ] 
 
 
 Wu 
 
 r u 
 
 EFr u 
 
 EFr u cos (f) 
 
 
 
 Jl 
 
 SI 
 
 Ql 
 
 
 T w n 
 
 r- n 
 
 EFr n 
 
 EFr n cos < 
 
 
ART. 87 DEFLECTION POLYGONS BY ANALYTICS AND GRAPHICS 
 
 107 
 
 For any positon of the moving elastic load w and any panel, the same work equation 
 applies and also Eq. (36e) . 
 
 Hence, with due regard to signs, the influence area of a deflection y is again the 
 influence area for the moment M^ and also the influence area of the particular member 
 under consideration. It is only necessary to observe the signs of the loads w and the 
 corresponding position of the closing line of the deflection polygon. 
 
 The w loads which result from the chord members will have the same sign as the 
 moment influence line for that panel. When the moment changes sign in any par- 
 ticular panel, as might be the case in cantilever systems, then the w loads of the adjacent 
 pin points must be of opposite signs. 
 
 The w load resulting from the diagonals will always be of opposite signs according 
 to the rule above given, no matter whether the center of moments for such diagonals 
 falls inside or outside the span. 
 
 (6) For indeterminate systems the foregoing method is made applicable by removing 
 the redundant members or reaction conditions and replacing these by external forces 
 X acting on the principal system in the manner described in Art. 7. 
 
 The deflection of an indeterminate system under certain loads P is the same as the 
 deflection of the statically determinate frame subjected to the loads P and X and the 
 work Eq. (GA) is in like manner applicable to this principal system, so loaded. 
 
 An example showing the application of the above method may be found in Table 50s. 
 
 ART. 37. SECOND METHOD, ACCORDING TO PROFESSOR LAND 
 
 (a) The w loads in terms of the unit stresses in the members. 
 
 This method is sometimes preferable to the former when the deflection polygon of 
 any succession of members is desired without regard to the remaining members of the 
 structure. The first method is, however, preferable when a more general solution is 
 required. 
 
 Here, any succession of members or rods is treated purely as a kinematic chain " 
 involving only the angular changes occurring 
 in the angles included between successive 
 members and the changes Al in the lengths of these 
 members. 
 
 The problem resolves itself into two parts: 
 (a) to find the changes in the angles of any 
 triangle resulting from changes M in the lengths 
 of the sides, and (6) to evaluate the loads w in 
 terms of these changes and finally to construct the 
 deflection polygon. 
 
 (b) To find the changes in the angles of a 
 triangle. 
 
 Let a, /? and- 7- be the three angles of a triangle ABC and/ a ,/ 6 and/ c the unit stresses 
 respectively in the three members opposite these angles, as given in Fig. 37A. Also let 
 
108 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. VII 
 
 Ja, AB and Af be the changes in the three angles resulting from the changes Jl a , Jl b 
 and Al c in the lengths of the three sides. 
 
 Then from the figure, sin a\ =x/l c and sin 2 =y/lb, and by differentiation and 
 treating all quantities as variables, 
 
 cos 
 
 l c AxxAl c l b Jy-yJl b 
 
 and cos a^Ja<>= -. 
 
 r * /.* 
 
 Adding these expressions and solving for Aa\ + J2 = J gives 
 
 Jxx 
 
 Jl b 
 
 / 7 
 
 ?- (37A) 
 
 But J c cos i ={, cos 2 =/", hence Eq. (3?A) becomes 
 
 .. (37B) 
 
 ry b 
 Now from Fig. 3?A, 
 
 Jx_Jy_f f Mc_fc. Mb_fb. ^_ cot/? . and y._ cotr 
 T~Y F J Z c ~' Z 6 ~' r~ *<*> r~ tr ' 
 
 which values substituted into Eq. (3?B) give 
 
 ^a = (/ a -/ 6 )cot r + (/ a -/ c )cot^. . _. , "."- \ . . (37c) 
 
 The values EJ{3 and EAj- can be derived in the same way to obtain the following: 
 E Aa - (f a -f b ) cot r - (f c -f a ) cot B 1 
 
 = (f b -f c ] cota-(/ a -/ 6 ) cot r \. . ."../. . (37n) 
 = (f e -fJ cot B-(f h -ft cot a 
 
 (3?E) 
 
 wherein tensile stresses / are positive and compressive stresses are negative. 
 
 Since the sum of the three angles of a triangle is always constant, therefore, 
 
 whence the third value is easily found after having computed any two from Eqs. (37n). 
 
 The above formula are extensively used in Art. 61 dealing with secondary stresses. 
 
 When temperature changes are to be included then the unit stresses / must be 
 corrected by dE. A positive t produces elongation in all members, hence dE must 
 then be positively applied to all values/. The contrary is true for t. 
 
 The change in any peripheral angle ^ of a frame is easily found by summing the 
 changes Aa occurring in the several angles a whose apices meet in that angle 0. Then 
 ^ = Sa and J^ = 2Ja, making 
 
 EI>Aa ...... .... (37F) 
 
ART. 37 DEFLECTION POLYGONS BY ANALYTICS AND GRAPHICS 
 
 109 
 
 using the minus sign when 0=360 Ha which is the case when the vertices of the 
 angles a are on the opposite side of the kinematic chain from the peripheral angle <p. 
 The values EJa are obtained from Eqs. (3?D). 
 
 The products of the form (f a fb) cot 7- in Eq. (37c) may be graphically found 
 from a large scale truss diagram asjndicated in Fig. 3?A. The quantity (f a fb) is laid 
 off perpendicular to either AC or BC so as to include the angle ?. The base, to scale 
 of forces /,' will then be the desired quantity. The sign is easily found by inspection 
 of the given data and depends on the sign of (f a f b ) and the sign of the cot f. 
 
 (c) To evaluate the elastic loads w in terms of the angle changes J and the changes 
 M in the lengths of the members, and finally to construct the deflection polygon for the 
 kinematic chain. 
 
 In Fig. 37 R such a chain of chord members is shown and the J</ and Al are now 
 supposed to be given. The system is referred to coordinate axes (x, y). 
 
 A' 
 
 FIG. 37s. 
 
 Now let /l=the angle which any member makes with a line parallel to the x axis 
 and through the right-hand end of the member. See Fig. 37s regarding the sign of A 
 for different members. 
 
 OA, 81, 3 2 are the displacements of the points A, 1, 2, etc., respectively, measured 
 from the closing line A'B' and parallel to the given y axis. 
 
 w\, w 2 , w 3 are the elastic loads, the moment polygon for which represents the 
 
 ' These loads are now to be found 
 
 The 
 
 deflection polygon A'l'2'3' with closing line A'B'. 
 
 and then the polygon is easily constructed as for the first method above given, 
 loads w, parallel to the y axis, are applied, as before, in the direction in which the 
 deflections are to be measured. 
 
 In the figure rF|| 'A T B 7 \\Wc, hence it follows that ce^(o 2 diK- and ac = d 2 d 3 
 and by addition of these equations, ae is obtained as 
 
 )-T + (^2 03)- 
 d 2 
 
 (37o) 
 
1 10 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. VII 
 
 But since the sides of_the deflection polygon are respectively parallel to the rays 
 
 y-f f> nij j. 
 
 of the force polygon, then -r-=-fr. This value of ae inserted in Eq. (37c) gives 
 
 ds ti 
 
 Also from Fig. 37s, 
 
 .Vi 2/2 =fe sin 4 :. 
 
 The differential of Eq. (37j) gives 
 
 Ji/i Ay 2 =A1 2 sin X 2 +1 2 cos k 2 d& 2 =i d 2 (37iv) 
 
 which must equal o\ o 2 because 'both expressions represent the deflection of the point 
 2 with respect to the point 1. 
 
 Also from the figure d 2 =1 2 cos X 2 which divided into Eq. (37K) gives 
 
 and in like manner is found for + ^ 
 
 , 
 
 -- -j =-;- tan 
 "3 h 
 
 Making H = 1 and substituting these values into Eqs. (37n) then 
 
 ~ tan A 2 +-r-^ tan A 3 ....... (37L) 
 
 But 180->(2 + >l3 = 2 then - 
 
 Also -r^=Tf and -7-^=^, whence Eq. (37i.) becomes 
 1 2 & /a Jb 
 
 f 2 t&n A 2 +/ 3 tan 4 ....... (3?M) 
 
 The general expression for any pin point n is then 
 
 Ew n = E4 tyn -f n tan X n +/ + 1 tan 4 . + 1 ..... (3?N) 
 
 in which Ed<f> n is given by Eq. (37r) for top chords, using the sign for bottom chords. 
 
 For any negative angle ^ (see Fig. 37s) tan A also becomes negative. Similarly 
 the signs of the unit stresses / must be considered in substituting values in Eqs. (37r) 
 and (37N). The equations as they stand are written for positive values in all cases. 
 
 // the elastic loads w are multiplied by E and the force polygon be constructed to any 
 
ART. 37 DEFLECTION POLYGONS BY ANALYTICS AND GRAPHICS 111 
 
 convenient scale of forces with a pole H=E to scale of forces, when the scale of the truss 
 diagram is l:a, then the deflections will be actual to the scale I: a. But if the deflections 
 should appear m times actual when measured with the scale 1 : a, then the pole distance must 
 be made E/m to the scale of forces. 
 
 Finally the closing line A 'B' is found from the reaction displacements. If these 
 are zero, then dA and d B are zero and the closing line must join the ends of the equilibrium 
 polygon. Otherwise $A and d B must be ascertained from the conditions of the supports. 
 
 When the web system does not include vertical members then it is preferable to 
 choose these web members for the kinematic chain rather than a chain of chords, because 
 the deflection polygon will then furnish the deflections of all the pin points instead of 
 only those belonging to one chord. 
 
 When the deflections in one direction only are required, then the above polygon 
 affords a complete solution, but for a general determination of displacements in the 
 plane of the truss, a second deflection polygon would be required, giving deflections 
 perpendicular to the direction chosen for the first polygon. In the above illustration 
 this would be parallel to the x axis and the resultant displacement of any point would 
 be the resultant of the horizontal and vertical displacements of that point. 
 
 The displacements in the x direction could be computed from the same Eq. (37isr) 
 by substituting cot X for +tan A in each case, but a better solution is given below. 
 
 The method fails when any angle ^=90, hence no member of the kinematic chain 
 should be parallel to the direction of the deflections. If this occurs then a pair of 
 substitute members must be employed as was done in Fig. 36B. For this reason also, 
 the w loads from Eq. (37x) cannot usually be employed for finding horizontal displace- 
 ments as per Art. 38. 
 
 For a straight chord, the Eqs. (37F) and (37*) give the following simple formula 
 for w n 
 
 ......... (37o) 
 
 Example. The above method will now be illustrated by the following example 
 of a two-hinged framed arch as shown in Fig. 37c. The same problem is also solved 
 in Art. 50, using the method given in Art. 36. 
 
 The unit stresses and angle functions appear in the truss diagram of Fig. 37c arid 
 the nomenclature used in Eqs. (37o) as indicated on a small triangle, is successively 
 applied to all the triangles of the frame. 
 
 The deflection polygons are drawn first for the bottom chord and then for the top 
 chord, both for the conventional loading X a =l, and the computations of the Ew loads 
 are given in detail for the bottom chord only, because these represent the general case 
 necessitating the use of Eqs. (37D), Eq. (37r) and Eq. (3?N). For the top chord, the 
 computations do not require the use of Eq. (3?N) and the resulting values of the loads 
 Ew only are given without the details of the figuring. 
 
 In the present problem, the method would not apply to the kinematic chain 
 f/0^1^1^2^2, etc., because the verticals are parallel to the direction of the vertical 
 deflections. Hence, the chords are treated separately, thus avoiding all vertical members 
 in the chains. 
 
112 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. VI] 
 
 15* 
 
 15' 
 U f g -O.OI07 
 
 3 The ^ ig" r "bo'v* +he members reprlcserrl unit SfreSSesfl 
 
 Kip. 
 
 BOTTOM CHORD DEFLECTION POLYGON 
 
 All deflections af*- times actual 'in f>et, when meaeUred + 
 
 scale of lengths . The figures in ft. 'are E time* actual 
 
 I I 
 
 I I 
 
 I I 
 
 I 
 I 
 I 
 
 LENGTHS' 
 
 DEFLECTIONS. 
 
 FIG. 37c. 
 
AKT. 157 DEFLECTION POLYGONS BY ANALYTICS AND GRAPHICS 113 
 
 The elastic load Ew Q for the bottom chord has no effect on the deflection of this 
 chord and is, therefore, neglected. 
 
 Computation of the elastic loads Ew for the bottom chord panel points. 
 
 For panel point L i : 
 
 By Eqs. (3?D), #Ji =(-0.0257 -0.0503)3.059 - (0.0255+0.0257)0.454 = -0.2557 
 
 =0.0 (0.0503+0.0107)0.890 = -0.0543 
 
 = (0.0625-0.0347)0.1125 -(-0.0342-0.0625)1.394 ==+0.1379 
 
 ByEq. (37F), -EAfa =EI>Aa = -0.1721 
 By Eq. (37N), Ewi =0.1721 -0.0255x0.660+0.0347X0.5596= +0.175 
 
 For panel point L 2 : 
 
 By Eqs. (37D), EActi =(-0.0342-0.0625)1.394 - (0.0347+0.0342)0.5596= -0.1733 
 
 EAa% =0.0 (0.0625 +0.0352)0.7173 = -0.0701 
 
 EJci 3 = (0.0791-0.0493)0.4192 -(-0.0498-0.0791)0.963 ==+0.1366 
 
 By Eq. (37r), -EJfa =EI>Aa = -0.1068 
 By Eq. (37N), Ew 2 =0.1068 -0.0347X0.5596 +0.0493x0.4312= +0.109 
 
 For panel point L :; : 
 
 By Eqs. (37D>, EActi= (-0.0498 -0.0791)0.963 - (0.0493+0.0498)0.431 =-0.1668 
 
 #J 2 =0.0 (0.0791+0.0736)1.038 =-0.1585 
 
 = (0.1065-0.0687)0.821 -(-0.0657-0.1065)0.649 ==+0.1428 
 
 By Eq. (37r), -EAfa =EI>Aa = -0.1825 
 By Eq. (37N), Ew 3 =0.1825-0.0493 X0.4312 +0.0687 X0.3147 = +0.183 
 
 For panel point L 4 : 
 
 By Eqs. (37o), EAon =(-0.0657-0.1065)0.6487- (0.0687+0.0657)0.3147 = -0.1540 
 =0.0 (0.1065+0.1000)1.542 =-0.3184 
 
 = (0.0889-0.0990)1.409 -(-0.0579-0.0889)0.462 ==+0.0536 
 
 By Eq. (37r), -EA$*=ET>Aa = -0.4188 
 ByEq. (37N), Ew =0.4188 -0.0687X0.3147 +0.0990X0.1865= +0.416 
 
 For panel point L- : 
 
 By Eqs. (37D), EAa^ =(-0.0579-0.0889)0.462 - (0.0990+0.0579)0.1865= -0.0971 
 
 EJa 2 =0.0 (0.0889+0.1081)2.164 = -0.4263 
 
 = (0.0541-0.1207)2.111 -(-0.0270-0.0541)0.400 -0.1082 
 
 By Eq. (37r), -EA^=E^Aa = -0.6316 
 ByEq. (37ar), Ew 5 =0.6316 -0.0990X0. 1865 +0.1207 X0.0621 = +0.621 
 
 For panel point L, ; : 
 
 By Eqs. (37D),jBJi= (-0.0270 -0.0541)0.400 - (0.1207+0.0270)0.062 =-0.0416 
 JJa 2 ==0.0 (0.0541+0.1290)2.500 = -0.4578 
 
 By Eq. (37p), -i^J0 6 =i^SJa= -0.4994 
 ByEq. (37u), ^Ewe =0.4994 -0.1207 X0.0621 = +0.492 
 
114 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. VII 
 
 The minus sign in Eq. (37r) applies here because the angles a are on the side of 
 the chord opposite to the angle <p. 
 
 The Ew elastic loads are now combined into a force polygon and the equilibrium 
 polygon drawn for these loads acting through the several lower chord panel points will 
 then represent the deflection polygon for the lower chord. 
 
 If the pole distance H were made equal to E then the deflections would be actual 
 to the scale of lengths according to the rule above given. However, this would be too 
 small a scale for accurate results and therefore, the pole distance was made equal to 4w 
 units, giving deflections E/4 times actual to the scale of lengths of the drawing. By 
 constructing a scale four times smaller than the scale of lengths, the deflection ordinates 
 E times actual may be scaled directly from the drawing, and these values are written 
 on the ordinates. 
 
 The same problem is solved by the graphic method in Fig. 50s and again by the 
 method of Art. 36 in Table 50B. 
 
 Using the Ew loads computed for the top chord panel points the top chord deflection 
 polygon is obtained in precisely the same manner and these loads and the deflections 
 found for the top chord are comparable with the values in Art. 50 just referred to. 
 
 The numerical values of the Ew loads for the top chord panel points are as follows: 
 
 #w = -0.139 Ew 2 = +0.117 Ew 4 = +0.415 %Ew 6 = +0.479 
 
 Ewi = +0.098 Ew 3 = +0.203 Ew 5 = +0.608 
 
 The details for the computation of the load Ew% are given for illustration. Thus : 
 
 By Eqs. (37o), EAcn=( 0.0625+0.0352)0.7173 -(-0.0498-0.0625)1.394= +0.2266 
 = ( 0.0493+0.0498)0.431 -( 0.0791 -0.0493)0.419= +0.0302 
 = (-0.0657 -0.0791)0.963 -0.0 -0.1394 
 
 By Eqs. (37r) and (37o) Ew 2 =EA$ 2 =#Z Jo: = +0.1174 
 
 which follows for a straight chord when the angles a and ^ are 'on the same side of the 
 chord. 
 
 ART. 38. HORIZONTAL DISPLACEMENTS 
 
 To find the displacements in a horizontal direction when the deflection polygon 
 for vertical deflections is given. 
 
 It is readily seen, from the previous description, that the elastic loads w are indepen- 
 dent of the direction of the deflections because in every instance their value is dependent 
 on the shape of the truss and the changes in the lengths of the members, regardless of 
 how these were produced. See Eqs. (36s) and (36c). Hence, the same elastic loads 
 may be employed to find deflections in any direction. 
 
 The force polygon, used for the deflection polygon for vertical deflections, will thus 
 be the same for any other direction of deflections when revolved through an angle 
 equal to the angle included between these two directions. When horizontal and vertical 
 
ART. 38 DEFLECTION POLYGONS BY ANALYTICS AND GRAPHICS . 
 
 115 
 
 deflections are under consideration this angle is 90 and the same force polygon may 
 readily serve both purposes, since the rays for horizontal deflections are then perpen- 
 dicular to the rays drawn for the vertical deflections. 
 
 In Fig. 38A, let the succession of members A to B represent the bottom or tension 
 chord of some structure, fixed at A and supported by a roller bearing on a horizontal 
 plane at B. Also given the deflection polygon A'c'd'e'f'B' to find the deflection polygon 
 A"c"d"e"f"B" for horizontal deflections. The loads w are all multiplied by E and the 
 deflections are m times natural. 
 
 The left-hand end of the chord havirig no horizontal motion, it is most convenient 
 to call all horizontal deflections positive to the right. 
 
 The pin points are horizontally projected on to the closing line A"e , which latter 
 is perpendicular to the closing line A'B'. The polygon A"c"d"e"f"B" is then drawn 
 
 *>-dL^ 
 
 FIG. 38A. 
 
 by making the sides perpendicular to the respective rays of the force polygon and 
 maintaining the same order of succession in the pin points formerly used in constructing 
 the deflection polygon for vertical deflections. The horizontal abscissa B Q B" is then 
 the horizontal displacement of the point B. Similarly the abscissa cfod" is the horizontal 
 displacement of the point d, etc. 
 
 For any horizontal member, as ef, the horizontal displacement of the point / with 
 respect to e must naturally be the M for that member. Hence since_the scale of deflec- 
 tions is m times actual to the scale of lengths the displacement e"f" =mAl to the scale 
 of lengths used. 
 
 If the point B is made to roll on some inclined plane ik instead of the horizontal, 
 then a new closing line A7% must be determined for the two deflection polygons, as 
 follows: draw a line 'B^' \\~ik and erect a perpendicular to AB prolonged, and passing 
 through B". The intersection of these two lines gives k" and the vertical ordinate 3 B 
 represents the vertical displacement of the support B. 
 
116 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. VII 
 
 Hence, making B'k' =d B the closing line A'k' may be drawn. Also a line A"k J_ A'k' 
 becomes the required closing line for the horizontal deflection polygon. 
 
 The same solution will apply to deflections in any direction other than 90 with 
 the vertical. Also, when the deflection polygon for vertical deflections of all points of 
 a truss is given, as in Fig. 36A, then the horizontal deflections of all pin points are found 
 precisely as above by adhering strictly to the sequence in which these points occur on 
 the given deflection polygon. 
 
 The change d AB in the length of the chord AB, Fig. 38A, may easily be found by 
 taking the loads w parallel to AB. In this case the displacement d^B becomes the 
 intercept on AB produced and included between the extreme rays of the equilibrium 
 polygon. Hence this displacement is equal to 
 
 ,ww 
 
 (38A) 
 
 wherein y is the ordinate of any pin point measured normally to A B. 
 This same result may also be found by computation from 
 
 (38u) 
 
 which follows for a pole distance of unity. 
 
 For the case given in Fig. 37fi, where the angle changes A$ and the changes in the 
 lengths of the members are given, the total effect on the length of span AB then becomes 
 for n members, 
 
 ^ B = Si"~V0 + SrJZ cos (;-), .--.. ,>.;-. . (38c) 
 
 where a is the angle which the span AB makes with the horizontal. 
 
 Example. The lengthening d^B for the span AB due to the loading X a = l in Fig. 
 37o is computed from Eq. (38c) , using the values E Al as given in Fig. 50A. The value 
 \Ed AB was found to check the value determined graphically in Fig. 50s. The 
 values Ed<{> are those above computed for the bottom chord panel points and a =0 
 because the span is symmetric. 
 
 Point. 
 
 E4$ 
 
 Feet. 
 
 EyJ<l> 
 
 E4l Fig. 50A. 
 Feet. 
 
 cos >l 
 
 KJl co* X 
 
 A 
 
 0.1721 
 
 6.188 
 
 1.0649 
 
 . 286 
 
 0.834 
 
 0.2385 
 
 L, 
 
 0.1068 
 
 14.58 
 
 1.5571 
 
 0.597 
 
 0.873 
 
 0.5212 
 
 L 3 
 
 0.1825 
 
 21.05 
 
 3.8416 
 
 0.805 
 
 0.918 
 
 0.7390 
 
 L< 
 
 0.4188 
 
 25.77 
 
 10.7925 
 
 1.080 
 
 0.954 
 
 1.0303 
 
 L, 
 
 0.6316 
 
 28.568 
 
 18.0435 
 
 1.511 
 
 0.983 
 
 1.4853 
 
 L, 
 
 = 0.4994 
 
 29.50 
 
 14 . 7323 
 
 1.814 
 
 0.998 
 
 1.8104 
 
 = 50.0319 
 
 5.8247ft. 
 
 These values substituted into Eq. (38c) give for a lengthening E times actual, 
 4B =iE27/^ + A#2J/cos >l= 50.03 19 +5.8247 =55.8566 ft., where -29000 kips per 
 sq. inch. 
 
r. 39 DEFLECTION POLYGONS BY ANALYTICS AND GRAPHICS 117 
 
 ART. 39. DEFLECTION OF SOLID WEB BEAMS 
 (a) Deflection due to moments. The differential equation of the elastic curve is 
 
 dx2 = -E~r ' ' ' ( 39A > 
 
 | Eq. (36s) gives the elastic loads w, for only one chord, in terms of moment and 
 truss dimensions as 
 
 Mp Ml 
 
 7/1 = L = 
 
 r 2 EFr 2 ' 
 
 Considering both chords, each of area F, and neglecting the web effect, which is 
 usually quite small, then for a unit length of girder 
 
 EFr 2 El ' 
 
 Hence Eq. (39A) gives directly the elastic loads w per unit length of girder as 
 
 d 2 v M 
 
 (39s) 
 
 An equilibrium polygon drawn for these loads, with pole distance equal to unity, 
 gives deflections to the scale of lengths chosen for the drawing. 
 
 By treating the moments M per unit length of girder, as loads which now become 
 El times too large, and constructing an equilibrium polygon for these loads M with a 
 pole distance H=El, the same deflection polygon is again obtained, giving deflections 
 to the scale of lengths of the drawing. In other words the moment polygon for any 
 case of loading becomes the load line for the deflection polygon corresponding to such 
 case of loading. 
 
 Deflections are usually drawn several times actual size, in which case the pole 
 must be divided by as many times the scale of the drawing. Thus, if the scale of lengths 
 is 1 :n and the deflections should appear twice actual, then the pole H =EI/2n. 
 
 When the value / is variable, the pole distance is made to vary as a function of 7, 
 as illustrated in the example below. See also Art. 47. 
 
 (b) Deflection due to shear. From Eq. (15M) the deflection of a straight beam 
 when acted on by shearing forces only, is 
 
 sn ^f) 
 
 (39c) 
 
 Disregarding all unnecessary refinements, since shear deflections are small compared 
 to the total deflection, tire value /? may be taken as constant and Q will be assumed 
 uniformly distributed over the web plate of sectional area FI =^ 
 
118 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. VII 
 
 For a single concentrated load 'dQ/'dP m = 1 and Q =R =the end reaction, hence 
 
 3 / 30 , 8 / 
 
 Taking (? =0.3332? and Fi=F/3 and assuming an average value for ,3=2.5 for| 
 ordinary girder sections, then approximately 
 
 (c) Deflection due to combined shear and moments. Ordinarily it will suffice to 
 figure the deflection due to shear for a point at or near the point of maximum moments 
 and to determine the percentage which this d m is of the moment deflection ordinate 
 at the same point. All other moment deflection ordinates may then be increased in 
 the same proportion to obtain the deflection polygon for combined shear and moment 
 effects. 
 
 For a uniformly loaded beam or constant / and neglecting shear effect, the deflection 
 becomes 
 
 1 r 
 = EIJ 
 
 When the depth of a girder is constant, but the moment of inertia varies so as to 
 maintain a constant stress on the extreme fiber at all sections, then the ratio M/EI 
 becomes constant and the deflection for such a case would be 
 
 -if: 
 
 where I m is the moment of inertia at the section of maximum stress. 
 
 When the moment of inertia varies abruptly the deflection may be expected to 
 fall between the two above values, making the numerical coefficient close to 1/9. 
 Taking in the shear effect, the approximate formula for maximum deflection, of a beam 
 of variable /, becomes 
 
 % '"max^ | ""*max fon^\ 
 
 "-*"' ' ' 
 
 However, for general cases of loading and variable 7 the graphic solution, above 
 given under Art. 39A, should be employed. 
 
 (d) Example. Given a plate girder of variable section and uniformly varying water 
 load as shown in Fig. 39A, required to find its deflection polygon. The girder is designed 
 as a double web beam and normally occupies a vertical position so that there are no 
 dead load stresses. 
 
 These girders are spaced 9.175 ft. center to center so that each will carry a full 
 water load P=31.25X9.175(?-a) 2 giving rise to the end reactions A and B, and moments 
 M, all as indicated in the diagrams, Fig. 39A. 
 
DEFLECTION POLYGONS BY ANALYTICS AND GRAPHICS 
 
 119 
 
 FIG. 39A. 
 
120 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. VII 
 
 The moments of inertia were computed for the various sections on gross areas, and 
 are plotted in the upper diagram. 
 
 The bending moments were also computed, using the formula 
 
 (39H) 
 
 and the results were plotted in millions of foot-pounds in the second diagram. 
 
 The w loads were then taken as the moments M over certain convenient lengths 
 Ax, instead of unit lengths according to the formula (39s) , whence 
 
 The lengths Ax are usually chosen with respect to the girder sections, so that / is 
 constant over each length Ax. Where the depth of section is variable the mean value 
 of / is taken for each length Ax. 
 
 Since M and Ax are both expressed in feet, while E and 7 are both for inches, the 
 factor 12X12 = 144 is introduced into Eq. (39 j) to establish a true ratio between like 
 units. Also, M and E are both expressed in millions of pounds, making the unit 
 weight one million pounds. This furnishes the true values for w in terms of the given 
 
 data as 
 
 144MAx MAx MAx ,_. . 
 
 --' (39K) 
 
 144 
 
 wherein E =28 is chosen low rather than high. 
 
 Now the scale of lengths was taken as 1 : 120, and if the deflections are to appear 
 
 twice actual size then the pole must be made equal to 
 
 (* 
 
 OO T 
 
 The pole distances are thus a constant function of 7 and may be found for all 
 values of 7. 
 
 The force polygon is now easily drawn to any convenient scale of forces, using the 
 same scale for the loads w and the pole distances 77. This scale has no influence on 
 the deflection polygon, but merely affects the size of the force polygon. 
 
 The deflection polygon is then constructed as the equilibrium polygon of the w 
 loads with the various pole distances, and the ordi nates included between this polygon 
 and the closing line A'B' will represent the deflections to twice natural scale. Had 
 the pole distances been taken twice as long, then the deflections would have been actual. 
 
 The deflection due to shear, at the point of maximum moments, is found from 
 
 5x537X12 
 Eq. (39n) as o m = = ; ^-p-=0.09S inch which is 10.5 per cent of the maxi- 
 
 mum deflection due to bending alone. 
 
ART. 39 DEFLECTION POLYGONS BY ANALYTICS AND GRAPHICS 121 
 
 The total deflection of the girder is therefore 10.5 per cent greater than the 
 amounts given on the deflection polygon of Fig. 39A. The maximum deflection due 
 to combined shear and bending effect is thus 0.93 +0.098 = 1 .028 inches, for the point 7. 
 
 The approximate formula (39o) gives for this same point 
 
 A 5.37X12X56.5 2 X12 2 . 
 07 = 9X28X134000 ' + 
 
 See Art. 47 for another method of dealing with variable moment of inertia, by 
 treating the quantities MAx/I as elastic loads and making the pole distance equal to E. 
 
CHAPTER VIII 
 DISPLACEMENT INFLUENCE LINES FOR STATICALLY DETERMINATE STRUCTURE! 
 
 ART. 40. INFLUENCE LINES FOR ELASTIC DISPLACEMENTS 
 
 Deflection influence lines could not receive adequate treatment in Chapter T\ 
 because these depended on a knowledge of deflection polygons. The, latter were taker 
 up in Chapters VI and VII. The subject is now treated from the most general aspec 
 covering the influence line for any elastic displacement. 
 
 Proceeding from Maxwell's law, Professor Mohr, in 1875, first developed displacemen' 
 influence lines. The application results from a consideration of two conventiona 
 loadings, of unit work each, applied to any frame so that one of the loadings represent; 
 the desired influence at some point n, and the other loading is a vertical moving loac 
 P = l applied at any load point m of the loaded chord. This includes all cases of con- 
 ventional loadings given in Art. 9, besides the simpler problems pertaining to displace 
 ments of points. 
 
 As applied to any simple beam or truss, the following theorem is now established 
 A deflection polygon, drawn for a load unity acting in a fixed direction on some detiniti 
 point n of any frame, is the deflection influence line for the deflection of the point n, in thi 
 flxed direction for any system of parallel moving loads applied to the loaded chord. Wher 
 the system of moving loads does not consist of parallel loads, then, of course, no influence 
 line is possible. 
 
 To prove this theorem, the simple beam, Fig. 40A, is loaded with a vertical loac 
 P n = l applied at the point n. A deflection polygon, drawn for this case of loading 
 will then be the deflection influence line for the point n, according to the following 
 demonstration. The method of drawing the deflection polygon is given in Art. 39. 
 
 The moment diagram is first drawn for the conventional loading P n = l kip, appliec 
 at n and thus making the maximum ordinate under n equal to M n = l-aa'/l=4.8 kip 
 ft. The symbol M is used to indicate a moment due to a conventional unit load. 
 
 This diagram is divided into suitable sections (2 ft. apart) and the moment areas 
 MAx are computed and treated as elastic loads w, for which the deflection poly- 
 gon A "B" is finally drawn, all as shown in Fig. 40A. 
 
 Assume the modulus '=28000 kips per sq. inch, 7=2087, in inches, M in kip 
 feet, and Jo? in feet. Then the pole distance for the force polygon becomes H =7/144 
 when the w loads are made equal to MJx, by Eq. (39j), for deflections of actual size, 
 However, the scale of lengths of the drawing was made 1 :36 and the deflections should 
 appear 200 times actual, hence H must be made 7/200X36X144=56.35 w units. 
 
 122 
 
40 
 
 DISPLACEMENT INFLUENCE LINES 
 
 123 
 
 With a pole H= 56.35 and the w loads MAx, construct the force and equilibrium 
 polygons as shown, and the deflection polygon so found represents the deflection influence 
 ine for the point n according to the following application of Maxwell's law. 
 
 The vertical deflection of any point m, for the given load P n = l, is represented by 
 :he vertical ordinate rj m of the deflection polygon A"B", vertically under the point m. 
 For, by Maxwell's law, rj m =d mn =d nm , or in words, T? OT equals the elastic displacement 
 :>f the point n for a unit load P m acting at the point m. Hence m being any position 
 
 |g.iKip. 
 yn 
 
 24-" 
 
 SOIb. I Beam , I =2O87.z in* 
 
 ; I 
 
 _ 
 
 % 
 
 , i- fcw 
 
 7 
 
 
 rl 
 
 M DIAGRAM FOR P = I KIP. 
 
 B 
 
 Scale for len^thsland momenta. 
 
 DEFJ.ECTIO|^ INFLL|ENCE L|NE FOR; POINT n. 
 
 B 
 
 Deflections zoo times actual. 
 
 FIG. 40A. 
 
 Df the moving load point, it follows that all ordinates rj m represent deflections for the 
 point n due to a unit load at the variable point m. This makes the polygon A"B" the 
 iesired deflection influence line for the point n. 
 
 Therefore, for any train of moving loads, the deflection at the point n becomes 
 
 (40A) 
 
 Since Maxwell's law is generally applicable to any case of conventional loading, 
 whether for a single point, a pair of points or a pair of lines, as illustrated in Art. 9, 
 
124 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. VIII 
 
 therefore, the above application affords a solution for any displacement influence line 
 other than those for simple vertical deflections. However, the train of loads must 
 consist of parallel forces P, as otherwise no influence line is possible. 
 
 ART. 41. SPECIAL APPLICATIONS OF DISPLACEMENT INFLUENCE LINES 
 
 (a) Required the deflection influence line for the point n of the simple cantilever 
 beam, Fig. 4lA. The deflection polygon is drawn according to the method of Art. 31) 
 and illustrated in Art. 40, for the conventional loading P n = l at the point n. This 
 then becomes the deflection influence line for the point n. 
 
 FIG. 4lA. 
 
 The beam is anchored at A and supported by a hinged bearing at R and a roller 
 bearing at C. The point n is the hinged connection at the end of the cantilever arm 
 Bn. The following moments M and reactions R, resulting from the conventional 
 loading P n = l, are then found: 
 
 A = --^, B=- 
 
 L\ 
 M A =0, MK-- 
 
 ^ 
 
 ^ V 
 
 M n =0. 
 
 The M diagram is constructed by making the ordinate at B' equal to 1 2 , and after 
 dividing this diagram into suitable sections and computing the loads w\ to w 8 , the 
 deflection polygon is easily drawn, observing the method followed in Art. 40. 
 
 The closing line of the deflection polygon is then found to be A"B"C", which 
 completes the influence line. The point B" is the intersection of the deflection line 
 
ART. 41 
 
 DISPLACEMENT INFLUENCE LINES 
 
 with the vertical through B and n"C" is a straight line. Upward deflections are negative. 
 
 The support at A is elastic and its displacement o Aa , due to the conventional 
 loading P n =-l, should be computed from Eq. (4A) including temperature effect if 
 desirable. This displacement is then applied at the point A" and the final closing 
 line Ai"B"C\" i-s thus made to include this effect. By applying any train of loads the 
 resulting actual deflection of the point n, according to Eq. (40A) becomes d n = HPi). 
 
 For any case of variable moment of inertia of the given beam the pole distance H 
 is made variable, as was done in Fig. 39A. 
 
 (b) Given a simple truss, Fig. 41 B, on two supports, to find the displacement influ- 
 ence line for any panel point n for displacements d n in the fixed direction rm 7 . The 
 loading is to consist of a system of vertical concentrated loads P acting on the bottom 
 chord. 
 
 FIG. 41s. 
 
 Apply the conventional loading P n = l in the given direction nn' and compute the 
 reactions H, A and B for this load, then determine t-he stress S\ and resulting change 
 Jl in length, for each member. Finally compute the w loads for the several panel points 
 and draw the deflection polygon for the loaded chord (here the bottom chord) using any 
 of the methods of Chapter VI, but preferably the one given in Art. 36 (c) . The direction 
 of the w loads is always taken parallel to the system of loads P, hence the displacement 
 influence ordinates jj will be vertical ordinates, measured vertically under the respective 
 loads P. This is necessary because the direction of the deflections is determined by 
 the direction assigned to the w loads. 
 
 The proof that the polygon A 'B' is the desired influence line again follows from 
 Maxwell's law, viz., for any ordinate rj m =d mn = d nm . 
 
 The displacement d n of the point n in the direction nn', caused by the system of 
 vertical loads P is then 
 
 (c) Given the three-hinged arch, Fig. 41c, to find the influence line ior the 
 angular rotation 3 n between the two lines An and Bn. The train loads are to be carried 
 bv the top chord. 
 
126 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. VIII 
 
 The conventional loading now becomes one of loading a pair of lines with a positive 
 moment equal to unity applied to each line in the direction in which the angle J would 
 increase for vertical loads acting on the top chord. 
 
 The horizontal thrust for this conventional loading is obtained by taking moments 
 about the crown n and gives H = (2-{-AV)/ < 2f and the reactions A and :i are found by 
 taking the moments of the external forces about B and A respectively. 
 
 Then for this case of conventional loading determine the stresses S\ and changes 
 Al in the lengths of the members and compute the w loads for the half span, which is 
 all that is required for a systematic structure. 
 
 FIG. 41c. 
 
 The moment diagram, drawn for the w loads, will then be the displacement influence 
 line for the change d n in the angle J. The d n for any particular set of loads is then 
 d n =2Prj, measured in arc. 
 
 The above problems will suffice to show the perfectly general solution of displace- 
 ment problems afforded by the application of Maxwell's law. These problems can also 
 be solved analytically with the aid of Mohr's work equation as indicated in Art. G. 
 
CHAPTER IX 
 INFLUENCE LINES FOR STATICALLY INDETERMINATE STRUCTURES 
 
 ART. 42. INFLUENCE LINES FOR ONE REDUNDANT CONDITION 
 
 The principle, illustrated in Chapter VIII, for drawing deflection influence lines, is 
 easily applied to the construction of the influence line for any external or internal 
 redundant condition. 
 
 Eqs. (7n) and (So) for one redundant condition and a single external force P = 1 
 become 
 
 ' 
 
 a a0 aa f- 
 
 and < 42i > 
 
 "a 1 ' "ma ~ X a O aa = X a f) a 
 
 after substituting the value d a as obtained from Eq. (4A) and neglecting temperature 
 and abutment displacements. 
 
 This gives X a in terms of displacements or stresses, whichever is preferred, as 
 
 Y - 
 
 ' 
 
 where the redundant may be external or internal. 
 
 If the single load P = l is vertical and applied to some panel point of the loaded 
 chord then Eq. (42s) represents the influence line for X a for a unit load applied at any 
 panel point m. 
 
 A deflection polygon, drawn for the conventional loading or condition X a = l, will 
 give the ordinates r} m =d ma =d am = 'ZS S a p for the displacement- of the point a, for a unit 
 load at any point m of the loaded chord. Hence the deflection polygon drawn for 
 condition X a = 1 is the X a influence line provided all the ordinates are divided by the 
 constant denominator d aa +p a = 'Sc?o + p a , The constant displacement d a a=2>S a 2 o is 
 found by computation or otherwise, and p a is a given constant depending on the length 
 and section of the redundant member. When X a is a reaction then ,o a =0 for the case 
 of immovable supports. 
 
 Accordingly the following theorem may be stated: The ordinates to the influence 
 line of any redundant X a are some constant factor p. 1 ~ (d^ +p a ) times the ordinates to 
 a deflection polygon drawn for the loaded chord, for the conventional loading X a = l (P=0), 
 applied to the principal system. 
 
 127 
 
128 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. IX 
 
 (a) When the redundant X<j is an internal condition, as illustrated in Fig. 42A, the 
 above theorem is applied as follows: If the direction of stress in the redundant can be 
 anticipated, then this should be done, otherwise an arbitrary assumption must be made 
 and, if it was erroneous, the resulting value of the redundant X a will turn out negative 
 with respect to S . In any case the conventional loading X a = l, being external work, 
 should be so applied as to produce positive work when the redundant member is removed. 
 In the illustration, the member X a will be assumed in tension so that J/ a is a positive 
 elongation. Hence, the unit forces X a = 1 must be applied so as to move the points n 
 and HI apart. The principal system is the entire frame exclusive of the member X a . 
 
 The stresses S a are computed or found from a Maxwell diagram and from these the 
 w loads and deflection polygon for the loaded chord (which was taken to be the bottom 
 chord in the illustration) are determined by one of the methods in Chapter VII. 
 
 The deflection polygon so obtained may be represented by the broken line A'n'rii'B', 
 with ordinates i) m =d ma =d a m, which, according to Eq. (42B) are (daa+pa) times the actual 
 
 influence line ordinates for the redundant stress X a . 
 system of moving loads, becomes 
 
 Hence the stress X a , for any 
 
 (42c) 
 
 (b) When the redundant Xa is an external condition, the quantity p a becomes zero 
 for the assumption d a =0 for immovable supports, and this gives rise to a slight simpli- 
 fication in the solution of problems of this class. 
 
 Eq. (42n) then becomes in general 
 
 Oaa 
 
 (42D) 
 
 The illustration of a simple beam on three supports is given, in Fig. 42B, as a case 
 of one external redundant condition X c . Any one of the three supports might be 
 treated as the redundant one, but the middle support C is here chosen. 
 
 The conventional loading X c = 1 is again chosen so as to produce a positive quantity 
 of applied work. The deflection polygon A"C"B" of the simple determinate beam AB, 
 
ART. 42 INFLUENCE LINES FOR INDETERMINATE STRUCTURES 
 
 125) 
 
 which now becomes the principal system, is found for this load X c = l exactly as was 
 done in Art. 40, Fig. 40A. The unit load is applied at C in a downward direction 
 because the deflection of this point of the principal system is also downward when the 
 support is removed. 
 
 Any ordinate rj m of this deflection polygon is equal to the deflection d mc =3 cm , which 
 is cc times the X c influence line ordinate according to Eq. (42D). Also the special 
 ordinate f) c =d cc is given by the same deflection polygon and is a constant quantity. 
 Hence the factor /* for the influence line X c is equal to l/d cc and the ratio dmc/dcc 
 determines the value of the influence line ordinate for any point m. 
 
 k 
 
 im 
 
 FIG. 42s. 
 
 It is thus seen that for the case of a single external redundant X c the magnitude of 
 the moment M M =lilz/l and the pole distance H of the deflection polygon, do not affect 
 the influence line ordinates T im =d me /8 ee , though they do affect each of the ordinates $ mc 
 and S cc separately. 
 
 For any system of moving loads the required redundant reaction becomes 
 
 (42E) 
 
130 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. IX 
 
 ART. 43. INFLUENCE LINES FOR TWO REDUNDANT CONDITIONS 
 
 When there are two redundants, then two deflection polygons may be drawn for 
 the loaded chord of the principal system ; one for condition X a 1 and one for condition 
 X\)-=\. These polygons are then combined into an influence area for X a the ordinates 
 to which, times a certain factor /*, will give the influence line ordinates for the X a 
 influence line. Another combination similarly made for X b will give ordinates which 
 when multiplied by a factor fib, will give the ordinates for the'Afc influence line. The 
 analytic solution for these influence lines is derived from Eqs. (SD). 
 
 Thus for two external redundants with d a and 05 both zero and neglecting temper- 
 ature and reaction displacements, Eqs. (80) give, for a single load P, 
 
 = X a O aa 
 
 ........ . . (43A) 
 
 These equations solved for X a and A^ and observing that ^=^06 by Maxwell's 
 law, give 
 
 __ 
 ma 
 
 * -p 
 
 _ _ bb p _i}ma r -p 
 
 a * {ta^ma^ 
 
 * * Oab T)a 
 
 aa o6"F~ 
 066 
 
 - 
 
 mb ma * 
 
 wherein dma and dmb are the variable ordinates to the two deflection polygons X a = 1 
 and Afe = l, respectively, for any panel point m. The other quantities are all constants 
 given by the same deflection polygons. 
 
 These equations for a load P l, are the equations for the influence lines X a and 
 Xb. Their numerators then give variable ordinates which may be called r) ma and i) m b 
 while the constant denominators ij a and y b may be used as influence factors fta = l/T) a 
 and fjt b = l/r)b for the two influence lines X a and Xb respectively. 
 
 For any train of moving loads the redundant reactions are expressed by 
 
 ma } 
 
 \ ........... (43c) 
 
 inb 
 
 The problem of a simple beam, Fig. 43A, continuous over four supports, is chosen 
 to illustrate this case. 
 
 The redundant supports are considered removed and the principal system then 
 becomes a simple beam on two supports. In the present case the supports X a and A" 
 are treated as redundants, although any two of the four reactions could be thus regarded. 
 
ART. 43 INFLUENCE LINES FOR INDETERMINATE STRUCTURES 
 
 131 
 
 The moment diagrams for the two conventional loadings X a = l and X h =--l are now 
 drawn and from these the deflection polygons are constructed precisely as illustrated 
 in Fig. 40A. The two poles H are made equal and as a check on the drawing <? a6 'must 
 equal o& a . When the structure is a frame, then the deflection polygons for the loaded 
 chord are drawn as described in Art. 37. 
 
 The deflection polygons so obtained are the influence lines for the deflections d n 
 
 FIG. 43A. 
 
 and #6 of the two redundant supports. For immovable supports these are zero and 
 Eqs. (43A) apply to any case of loading. 
 
 To obtain the influence areas for the redundants, the two deflection polygons are 
 combined into two influence areas by computing the ordinates from Eqs. (43s) as follows: 
 
 b^ 
 Obb 
 
132 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. IX 
 
 The factors /*, and fj. b are the reciprocals of the denominators of Eqs. (43B) all the 
 values for which are given on the two deflection polygons. The ordinates rj ma give, 
 when plotted, the shaded area C'D', which is the influence area for X a . The ordinate 
 T) a , under the support X a , gives the influence factor // a = l/V In similar manner the 
 ordinates rj mb furnish the shaded area C"D", which is the influence area for Xb with 
 the factor /j. b = I/fa, obtained from the ordinate rj& under the support X b . 
 
 A purely graphic construction of the influence area X a is given by Mueller-Breslau as 
 follows: Draw the d a deflection polygon for the w a loads, using any convenient pole H. 
 Then for the w^ loads draw an equilibrium polygon passing through the three points 
 C', B' and D' '. This is perhaps the best and most convenient solution and is also 
 illustrated in Fig. 43A. 
 
 The d a and db deflection polygons are first drawn as above described with any 
 assumed pole distance which may be the same for each polygon. Then draw the d b 
 line through the three points C', B' and D', and also draw the d a line through the three 
 points C", A" and D". The shaded areas thus formed represent the X a and X b influence 
 areas. 
 
 To find the new poles 0' in the two force polygons, which are necessary in drawing 
 an equilibrium polygon through three given points, proceed as follows: In the d b force 
 polygon, draw OK\\C"D" and OKf^C^W 7 thus determining the points K and K'. 
 Then draw KO'\\C'D' and K'0'\\C'B' and the new point 0', thus found by the inter- 
 section of K'O' and KO' ', will be the required pole for drawing the db line through C', 
 B' and D'. 
 
 The pole 0' in the d a force polygon is found in a similar way as indicated in the 
 figure, using the points F and F'. The new pole 0' ' , found by the intersection of FO' 
 and F'O', serves to draw the d a line through the points C", A" and D". 
 
 This construction is based on the fact that the influence of a load at B on X a must 
 be zero, likewise for a load at A on X b . 
 
 The above methods are not practicable for more than two redundants, and the 
 following method of simplification is given for multiple redundancy. 
 
 ART. 44. SIMPLIFICATION OF INFLUENCE LINES FOR MULTIPLE REDUNDANCY 
 
 In many problems of this class it is possible to so choose the redundant forces 
 X that they will have a common point of application. Then for certain directions 
 of the X'8 the d coefficients bearing different subscripts and appearing as factors of 
 the X's in Eqs. (So) may be reduced to zero. Whenever this is possible then 
 v a o- ba=--0, dac=dca-0, ^bc=^^d>=0', etc. ; and the following simple equations are 
 obtained 1 . 
 
 9* X^ 7~> S ~V & \^ Z? Af* 1 J} 1 
 
 O b = Ztr'mOmb AfcOfcfc Ltti b ar + 0bt 
 
 (44A) 
 
ART. 44 INFLUENCE LINES FOR INDETERMINATE STRUCTURES 133 
 
 involving only one redundant X in each equation. The same assumptions may be 
 applied to Eqs. (7n). 
 
 This method of analysis was first introduced by Professor Mueller-Breslau, and 
 serves a most valuable purpose, especially when applied to fixed arches. 
 
 As will be shown presently, this disposition of the redundants is easily made 
 when their number does not exceed three. Beyond this number the analysis leads 
 to many complications. Fortunately, the important cases of redundancy are nearly 
 always limited to from one to three redundants, and then there is no difficulty in 
 following the scheme here proposed. 
 
 The solution of Eqs. (So) for simultaneous values of the X's is thus avoided and 
 the other chances for serious errors are greatly lessened. 
 
 It is usually expedient to treat temperature stresses and abutment displacements 
 separately from the primary stresses, and then Eqs. (44A) become for load effects only 
 and immovable abutments, 
 
 Z2->imma -\r ^*m mb /it \ 
 
 a = ^ ; x b = 5 ; etc (44s) 
 
 *!=!--; *6i=l-rJ etc (44c) 
 
 Oaa Obb 
 
 By placing 2.R a Jr=0, d a also becomes zero, likewise for S-fi^Jr and d b , a circum- 
 stance which should be noted in writing Eqs. (44s) and Eqs. (44c). 
 
 In all of the following investigations, the X's will represent either a single force or 
 a couple applied to a principal system. The d's in the first case will then represent 
 linear displacements and in the second case they will be angular distortions expressed 
 in arc. Thus, if a redundant X a is applied to the point a then ^ is the displacement 
 of the point a in the direction of X a for a force X a = 1. When X a is a moment or couple, 
 then daa becomes the angular rotation of a rigid principal system as produced by a 
 couple X a --=l. The displacements d a , d b , etc., and the conventional loadings X a = l', 
 X b = l, etc., are always positive in the opposite sense in which the redundants X a , X b , 
 etc., are supposed to act. 
 
 It should also be noted that in all cases here considered, the points a and b are 
 coincident and these designations are retained merely to distinguish the particular forces 
 from each other. 
 
 The Eqs. (44A) to (44c) are equally applicable to graphic and analytic solutions, 
 but the latter method is useful only when there are not more than three redundant 
 conditions, and great accuracy is desired. The graphic method is less laborious and 
 leads to a more comprehensive presentation. 
 
 The above will now be applied to general cases of two and three redundants. 
 
 (a) Structures having two redundant conditions. Here the two redundant forces 
 can always be applied at the same point and the case can be solved by applying Eqs. 
 (44s) and (44c), provided the directions of X a and X b are so chosen that #&, =&=(). 
 
134 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. IX 
 
 To accomplish this, assume the direction of X a as may seem most convenient, and 
 find the displacement a\a of the point a for a load X a = l, Fig. 44A. Now take X b 
 acting at the same point a and at right angles with aid. 
 
 Then ^=0, because it represents the displacement of the point 6 in the direction 
 of Xj, when only the force X a = l is active. Therefore, dab also becomes zero 
 by Maxwell's law. If then the displacement 6j6 of X b is found for the load Xb = l, 
 this in turn must be perpendicular to X a because d ab is zero. This always fur- 
 nishes a valuable check on the solu- 
 tion when the graphic method is 
 employed. 
 
 The example of a two-hinged 
 arch with a column support at the 
 center, Figs. 44B to 44D, is used to 
 illustrate the application of the 
 method to two redundants. 
 
 Fig. 44B shows the given struc- 
 ture with hinged supports at A, B 
 and C, thus involving two external 
 redundants. By removing the sup- 
 port at A the structure becomes 
 determinate, involving a simple truss 
 on supports B and C, and an over- 
 hanging cantilever AC'. The redun- 
 
 FIG. 44A. 
 
 dant forces X a and X b are then applied to the principal system at A and are treated 
 as external forces. 
 
 The first redundant X a is assumed to act horizontally and the second redundant X b 
 is applied at the same point and making the angle 6 with X a , such that d a b=dba=0. 
 
 Fig. 44c represents the conventional loading X a = l and a Williot-Mohr displace- 
 ment diagram, drawn for this loading, will furnish the displacement aa t for the 
 point A and the displacement mm a for the point m. Hence d ao becomes the pro- 
 jection of act! on the direction of X a and <?*, is the projection of mm a on the direction 
 of the force P m . 
 
ART. 44 INFLUENCE LINES FOR INDETERMINATE STRUCTURES 135 
 
 Similarly in Fig. 44D, another displacement diagram _drawn for the conventional 
 loading X b = l and acting at 90 with the displacement aal, gives the displacements 
 661 and mm b from which the values du, and dmb are found by projecting the displace- 
 ments on the directions of their respective forces X b and P m . 
 
 The angle 6 is thus graphically determined, and as a check, the displacement 66 t 
 must be perpendicular to X a . 
 
 Thus having the constants d m and d bb and the displacements d ma and d mb for any 
 pin point m, the influence produced by any load P m , according to Eqs. (44s) becomes 
 
 Zi mUma j v PmOmb /A4 -. 
 
 a= and X b = ^ , ........ (44D) 
 
 d aa O bb 
 
 where dma is negative with respect to the force P m , as may be seen from Fig. 44c. 
 The values X a and X b for any set of loads become 
 
 Z^imflma , _j v" ^^mPmb / AA \ 
 
 a = 5 - and X b = r - ......... (44E) 
 
 and the several values d ma and dmh are found from the two displacement diagrams 
 above described. 
 
 When the load P = 1 is vertical then the values in Eqs. (44D) represent influence 
 line ordinates for any point m for the redundants X a and X b . The same displacement 
 diagrams will furnish all values d ma and d mb for finding all the influence line ordinates 
 for both redundants. 
 
 For a uniform rise in temperature of t, the point A will move horizontally an 
 amount etl and the projections of this displacement on the directions of X a and X b 
 will then be respectively d at = etl and d bt = etl cos 0, whence 
 
 1 M 1 V 1 & COS 
 
 ^l-r- and X bt = l 
 
 (b) Structures having three redundant conditions. The most prevalent case of 
 three redundants is a fixed arch, and hence this style of structure is chosen, as an 
 illustration, see Figs. 44E to 44j. 
 
 Fig. 44E represents any general arch with fixed supports. It is transformed into 
 a determinate structure by removing one fixed support and converting the frame 
 into a cantilever arm to be treated as the principal system. 
 
 If, now, this support is replaced by three redundant conditions, a moment l-X a 
 and forces X b and X c , all acting on the rigid disc AJBO, Fig. 44r, the stresses in the 
 structure will remain exactly the same as in the original fixed_condition. 
 
 Then d a will be the angular rotation of the rigid disc ABO about some pole 0; d b 
 will be the linear displacement of the point in the direction of X b ; and o c will be that 
 
136 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. IX 
 
 displacement in the direction of X c . These displacements are supposed to be so chosen 
 that the moment X a and the forces Xb and X c , all applied to the rigid disc ABO, will 
 
 exert precisely the same effect on the principal 
 system as was previously produced by the re- 
 dundancy. 
 
 The pole is taken as the instantaneous center 
 of rotation of the disc, when acted upon by a 
 moment X a = l. If this pole is chosen as the point 
 of application for the redundants Xb and X c , then 
 dba=Q and ca =0. From this it follows that the 
 angular displacements <5 a & and 3 ac produced by the 
 loadings X b = l and X c = l acting on the disc, must 
 likewise be zero if the drawing or computation is 
 correct. Finally, if Xb is arbitrarily chosen in a 
 vertical direction and X c is taken perpendicular to 
 the displacement which the point b (pole 0) under- 
 goes as a result of the loading .X& = 1, then 
 
 PRINCIPAL SYSTEM. 
 
 CONDITION Xl 
 
 (J) 
 
 FlGS. 44E, F, G, H, J. 
 
 Hence this disposition of the redundant con- 
 ditions will cause all the quantities d bearing double 
 subscripts of different letters, to become zero, and 
 the simplified Eqs. (44A) will now apply. 
 
 When the abutments are considered immov- 
 able, d a , db, and d c become zero and Eqs. (44B) and 
 (44c) will suffice to solve the fixed arch. For 
 vertical loads these same equations furnish the 
 ordinates for the X a , X b and X c influence lines 
 by inserting a single load P m = 1 . 
 
 In proceeding to a solution it is best to apply 
 a unit moment e-l/e, Fig. 44c, to the points A and 
 B, representing the conventional loading X a = l 
 acting on the principal system. A Williot-Mohr 
 displacement diagram, drawn for this condition, will 
 furnish the displacements A^A and B t B of the two 
 points A and B. The pole 0, being the instan- 
 taneous center of rotation of the rigid disc, is 
 found by the intersection of two lines 
 
 and OB-LBiB. Then X b is applied at in a verti- 
 cal direction and X c is made active at and in a direction perpendicular to the 
 displacement found for from the displacement diagram drawn ior the condition 
 X b =*l. See Figs. 44n and 44j for the conventional loadings Xj, = l and X e = l. 
 
ART. 45 INFLUENCE LINES FOR INDETERMINATE STRUCTURES 137 
 
 ART. 45. STRESS INFLUENCE LINES FOR STRUCTURES INVOLVING 
 
 REDUNDANCY 
 
 (a) Multiple redundancy. The previous articles 42 to 44 show how influence lines 
 for redundant conditions may be determined. In the present article it is proposed to 
 present the methods of drawing stress influence lines for the members of any struc- 
 ture involving redundant conditions. 
 
 Th? general equation for the stress in any member of the principal system of a 
 statically indeterminate structure is Eq. (7 A). This equation, if written for a single 
 external load P m = l, will represent^ the stress influence line ordinate, under the point 
 ra, for the member S. Hence, if represents the stress in the member for condition 
 X =0 and P m = 1, then Eq. (7 A) gives the desired influence ordinate for any point ra as 
 
 i) m =S=S -S a X a -S b X b -S e X e ,etc., ...... (45A) 
 
 where rj m is the algebraic sum of the partial influences S , X a , X b , X c , etc., on the stress 
 of a certain member S, due to a load P m = 1 acting at the panel point ra. In other 
 words, a separate influence line may be drawn for each term of Eq. (4oA) and the 
 algebraic sum of the several influence ordinates under a certain load point m becomes 
 the stress influence ordinate y m for that particular load point. Such an influence line 
 may be regarded as a summation influence line of partial stress effects instead of a 
 summation load effect as originally implied by the definition in Art. 17. 
 
 Accordingly, for any system of external parallel loads, the total stress in any 
 member is represented by 
 
 ....... (45u) 
 
 As may be seen from Eq. (45A) such a stress influence line will always necessitate 
 drawing as many influence lines as there are redundants, plus one for the S stress. 
 However, the influence lines for the redundants remain unchanged for all members of 
 the same structure and hence need be drawn only once. Also the stressesjS a , S b and 
 S c are constants for a given member but vary for different members. The S influence 
 line is the same as the stress influence line for any member of a determinate frame and 
 will be different for each member. See Chapter IV. 
 
 Therefore, it is advisable to draw the influence lines for the redundants without 
 the S a , S b , S c , etc., factors and then draw the ~So stress influence line for any particular 
 member S. Finally insert for the X's the values for any set of ordinates, as for point 
 k, and these combined with the stresses S a , S b , S c in Eq. (4oA) give the required ordinate 
 rj k for the stress influence line S. The several ordinates rj for all panel points furnish 
 the required influence line. However, much of this work is done graphically as illus- 
 trated in Fig. 45A. 
 
 The example is a two-hinged framed arch with a column supporting the crown, thus 
 involving two external redundant conditions X a and X b as indicated in Fig. 45A. The 
 loads are vertical and applied to the top chord. Required to find the stress influence 
 line for the chord ik. 
 
138 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. IX 
 
 The influence lines for X a and X b are drawn supposedly by the method given in 
 Art. 43. The ordinary stress influence line for the chord ik is then drawn as indicated 
 and called the S line, meaning that it is the influence line for condition X = with 
 P = l. This line A'i'k'B' incloses a negative area representing compressive stress in 
 the top chord of a simple truss AB. 
 
 A 
 
 B 
 
 FIG. 45A. 
 
 The influence lines X a and X b are both positive and the stresses S a and Sb, for 
 the member ik, are found to be negative. The ordinate TJ^ for the panel point k is 
 found from Eq. (45A) as 
 
 ....". T . . ,. (45c) 
 
 giving a negative residual for y k . 
 
 In like manner all ordinates for the S influence line may be found and plotted to 
 give the shaded area A 'B' , which is the influence area for the chord ik. 
 
 The multiplications S a r) a , etc., can be performed graphically by laying off angles 
 corresponding to tana=S a and tan ft=S b as indicated in the figure. The products 
 S a rj a and S b y b are then taken off directly for any ordinate, added or subtracted graphically 
 and the sum S a y a +S b T) b is then applied to the ordinate S in the proper direction. The 
 angles a and /? are constant for the same member but vary for different members. 
 
ART. 45 INFLUENCE LINES FOR INDETERMINATE STRUCTURES 
 
 139 
 
 The stresses S a =tan a and Sb =tan /? are found from Maxwell diagrams or by computation. 
 
 It is thus seen that all loads between A and C produce compression in the member 
 ik, while all loads between C and B will produce tension. 
 
 The stress resulting from any system of loads is then 
 
 etc.=2Pi?, 
 
 (45D) 
 
 observing the sign of rj for each particular point. 
 
 Since the large majority of practical problems of redundancy do not, and should 
 not, involve more than one redundant condition, the general case is here treated in 
 less comprehensive manner. For a special method of deriving the influence line for a 
 web member from those of two adjacent chords, see Art. 52. 
 
 (b) One redundant condition according to Eq. (45A) gives rise to the simple stress 
 equation 
 
 /"cT \ 
 
 (45E) 
 
 *jm *^ ^o *^&^*-a *^a I ct -^-ct 
 V>a 
 
 which makes it possible to represent the S influence area as the area inclosed between 
 the X a influence line and the S /S a influence line, which latter is drawn with as much 
 ease as the simple S line. The resulting S 
 influence area will then require a factor S a 
 applied to all its ordinates. 
 
 These influence areas may be drawn in 
 either of two ways: First, by plotting both 
 the X a and the S /S a lines from a common 
 straight base, observing the rule that areas 
 above this base are negative, see Fig. 45u. 
 The S line ordinates TJ will then_ be measured 
 between the X a line and the S /S a line and 
 these ordinates will be positive when measured 
 down from the X a line. Second, by first 
 plotting the X a line from a straight base and 
 
 then constructing the S area by applying the +S /S a ordinates y' down from the X a 
 line, see Fig. 45c. In this case the +X a ordinates are best plotted above the base so 
 that the 4- TJ ordinates of the S area will appear below the base in the customary way. 
 
 The first method is more generally used, though the second leads to a very interesting 
 property by which the S line may be derived from the X a line when one point of the S 
 line and its zero points are known. 
 
 This property is shown in Fig. 45c, where it is seen that over any distance, as m'B', 
 for which the ~S~ /S a line is straight, the corresponding elements of the S line and the 
 X a line will intersect in points b', b" , &'"_which are in a straight line parallel to P and 
 passing through the end zero point of the S /S a line. 
 
 Further details and applications are not given here, as these will be illustrated in 
 connection with specific problems in Chapter X. 
 
CHAPTER X 
 
 SPECIAL APPLICATIONS OF INFLUENCE LINES TO STATICALLY INDETERMINATE 
 
 STRUCTURES 
 
 ART. 46. SIMPLE BEAM WITH ONE END FIXED AND OTHER END SUPPORTED 
 
 The solution by influence lines is illustrated by Figs. 46A, using the general method 
 developed in Art. 45B. 
 
 Applying the criterion of Eq. (3c) to this structure, where m = l, 2r=4 and 2p=4, 
 gives ra + 2r-2p = l. Hence there is one redundant condition and by Eq. (SB) this 
 condition must be external. 
 
 Accordingly, the reaction X a , acting at the expansion end B, is chosen as the 
 redundant support, reducing the beam to a simple cantilever arm as the principal 
 determinate system. 
 
 The problem is considered solved when the shear and moment influence lines are 
 found for any point n of the beam. The case of direct loading is assumed. 
 
 The X a influence line is represented by Eq. (42o) as 
 
 i ^ 
 
 1 . Ont, , . 
 
 where dma is the variable influence line ordinate at any point m with [1 = 1/8^ is the 
 constant factor, when d ma and daa may be acutal or measured to the same scale. 
 
 The equations for moment and shear at any point n and for the vertical reaction 
 A , are given by Eqs. (?A) written in the form of Eq. (45E) as 
 
 ~J A Y -- 4 Y \ 
 
 ** A a-A-a ~ -^rtl ~r~ -A-a 
 \Aa 
 
 (46B) 
 
 wherein M is the moment about the point n produced in the principal system by a 
 load unity acting at any point m of the structure. M a is the moment at the point n 
 due to the conventional loading X a = l. Similar definitions follow for Q~ , Q*, also A^ 
 and A a . 
 
 140 
 
ART. 46 SPECIAL APPLICATIONS OF INFLUENCE LINES 141 
 
 The following solution by influence lines is based on these Eqs. (46B), when the 
 influence line for X a is known or found from Eq. (46A) . 
 
 The ordinates d ma are ordinates of a deflection polygon drawn for the conventional 
 loading X a = l kip, being a load unity acting downward at the point B when the beam 
 is treated as a cantilever. This deflection polygon is drawn as described in Art. 40, 
 and as shown in Fig. 46Aa. 
 
 The M moment diagram for the conventional loading X a = l, is drawn by making 
 the end ordinate A^C' =1 and drawing C^B'. This end ordinate may be measured to 
 any convenient scale and was drawn to half the scale of lengths, making all the moment 
 ordinates to half scale of lengths. The figured ordinates represent moments, in kip feet, 
 and are scaled at points Jz=2 ft. apart. 
 
 The areas w =Jfdx are now computed and treated as elastic loads for the construc- 
 tion of the force polygon (a) and the deflection polygon (b) when the pole distance H=EI 
 Takino- E =28000 kips and 7=2087 in. 4 and reducing Ax to inches and M to kip 
 inches then the pole would be . When the scale of lengths is 1:60 then, 
 
 , ,, . ,, TT 28000X2087 O7n ^ nnita 
 
 for deflections 25 times actual, this pole becomes #=744x60x25 =270 - 
 
 When / is variable the pole is made variable as was done in Fig. 39A, or the w loads may 
 
 be divided by / as is done in Art. 47. 
 
 The force and equilibrium polygons are now drawn, using any convement_scale for 
 the w forces and the same scale for H . This gives the deflection polygon A'C', Fig. 
 46Ab with ordinates 25 times actual. Thus the actual end ordinate ^ = 1.96/25 inches. 
 This 'same polygon is also the X a influence line with a factor p = l/i) B irrespective of 
 the scale of ordinates, since all ordinates rj m and TJ B are measured with the same scale. 
 To obtain the X a influence line with a factory 1, divide all the ordinates of Fig. 46Ab 
 by the end ordinate , Bf whence the curve A'C', in Fig. 46AC, is obtained with the end 
 ordinate WC' = l, drawn to any convenient scale of ordinates. The redundant react 
 X can thus be found for any case of concentrated loads P and is X a = ^Pr l /t) B . 
 
 To obtain the M n influence area for moments about the point n, the X a influence 
 line is now combined with the M /M a influence line, with a factor p = M a according t 
 
 q ' In Fio- b the line ^C 7 represents the W influence line provided the end ordinate 
 WC' is made equal to z, but this end ordinate is actually , fl> hence the linem question 
 must have a factor p=z/y B . But since M a =z, therefore, the same line n C becomes 
 the W /M a influence line with a factor p-lfa which is the same as the factor for 
 the X a line. Hence the area IVC 7 represents the M n influence area with a fad 
 
 / ' = ^irnifarly / ?n 5 'Fi g . c, the area A^V represents the M n influence area with the factor 
 a=M a =z because ,' fl was here made equal to unity. This was done to show the two 
 solutions and to illustrate the fact that the second step of drawing the Z influence line 
 with a = 1 is really unnecessary and adds nothing except to simplify the final factor. 
 
 The algebraic sign of the M n area is derived from_Eq. (46B) since the larger _ X 
 area (which is positive) is subtracted from the positive M /M a area, leaving a negat 
 area as the remainder. 
 
142 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 Oi- 
 
 o/Q, LINE. 
 
 C' 
 
 FIG. 46A. 
 
ART. 47 SPECIAL APPLICATIONS OF INFLUENCE LINES 143 
 
 The moment area M m , for moments about a point m, is indicated to show a case 
 where a portion of the area below the X a line is positive, creating a load divide at the 
 point i. 
 
 The moment area MA, for moments about the abutment A, is also shown. It is 
 the area between the line A'C' and the X a line, which is entirely a positive area. 
 
 In all of these moment influence areas, the moments obtained will be expressed in 
 the same units as the applied loads P and the ordinate 2. Thus for P in Ibs. and z in 
 feet the moments will be ft. Ibs. The scale of the ordinates y is immaterial and is 
 eliminated in the factor p, Thus if Fig. b is used, the moment about n for any train 
 of loads would be 
 
 ii;-- S 
 
 and using Fig. c it would be 
 
 and the units would depend solely on those chosen for z and the loads P. 
 
 To obtain the Q n influence area for shear at the point n, the X a influence line is 
 now combined with the Q /Q a line, with a factor /*=Q a = l, according to Eq. (46e), 
 see Fig. 46Ad. 
 
 For a load unity to the right of n, the shear in the principal system would be 
 Q) = l and for X a = l the shear Q a = l, hence Q /Q a = ^ so that the broken line A'n'DC' 
 represents the Q /Q a influence line with a factor unity, and the Q n influence area is the 
 shaded area with a factor /*=Q a = l, with the portion below the X a line positive, as 
 before. Were Fig. b used, the factor would be / w=Q a /'?B == l/'?s- 
 
 The point n is always a load divide for Q H . 
 
 The end reaction A for any system of loads is simply HP X a , from Eq. (46B), 
 because A = DP and A a = l. 
 
 A uniform live load covering the whole span would give X a =3pl/8, as may be found 
 from Eq. (15j). The graphic diagram, Fig. 46Ab, gave X a =7Al2p=0.372pl 
 
 ART. 47. PLATE GIRDER ON THREE SUPPORTS 
 
 The example chosen is a general case of unequal spans and variable moment of 
 inertia as illustrated in Figs. 4?A. All dimensions in inches, and the loads are directly 
 applied to the girder. 
 
 According to the criterion of Eq. (3c) this structure involves one external redundant 
 condition. Any one of the three supports may be taken as the redundant condition, 
 but it is generally most convenient to assume the middle support at C and thus obtain 
 the principal system as a simple beam on two supports A and B. 
 
 The Xc influence line for the redundant support is again given by Eq. (42n) as 
 
 (47A) 
 
 where S mc is the variable influence line ordinate at any point m with the constant factor 
 fi = \/d ect and d mc and d cc must be measured to the same scale. The ordinates d mc are 
 
144 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP, x 
 
 the ordinates of a deflection polygon drawn for the conventional loading X c = \ kip, 
 acting downward at C on the principal system or simple beam AB. 
 
 This deflection polygon is the equilibrium polygon drawn for a series of elastic loads* 
 w = MJx/I when the pole H=E. The variable moment of inertia is not taken care of] 
 in the manner previously given, by varying the pole distance, but the alternative 
 method of involving / in the w loads is here employed. The moments of inertia are 
 given for the various sections of the beam in Fig. 4?Aa. The M diagram for X c = 
 1000 Ibs. is drawn in Fig. 47Ab, with an ordinate under the point C equal to 1000 l^lo/l = 
 175,000 in. Ibs. 
 
 The several M ordinates are scaled from the diagram or computed, and these 
 values divided by the corresponding 7's furnish the figures for plotting the ~kZ/7 diagram^ 
 Thus the ordinate at C is 1 75000 H- 34982 =5. 00. The w loads are then computed aa 
 the areas of the M/l diagram using the horizontal distances between the ordinates in 
 inches. These loads are applied at the centers of gravity of the respective areas. 
 
 The w loads are now combined into a force polygon, Fig. 47AC, with pola 
 H =El 100 XI 20 -=2333 w units, all drawn to the same scale, but this scale may be taken 
 as any convenient one without regard to the deflections. The particular pole will then] 
 give deflections one hundred times actual, because the scale of lengths for the girder 
 span was chosen 1:120 and the pole was made 100X120 = 12000 times too small. 
 I/ =28,000,000 Ibs. per sq. inch. The resulting equilibrium polygon 'A'C'B', Fig. 47ACJ 
 is the deflection polygon for the load X C = IQOO Ibs. with ordinates 100 times actual 
 and it is also the X c influence line with a factor 1/c where c is the ordinate under C. 
 measured to the same scale as the other influence ordinates. 
 
 It is clear from Eq. (4?A) that the scale of the influence ordinates is immaterial so; 
 long as the same scale is used for the ordinate c, but when actual deflections are sought 
 then a natural scale, as inches, must be employed. 
 
 The reaction X c for any case of concentrated loads may then be found from Fig. 
 47AC, by multiplying the loads by the d ordinates and dividing the sum of these products 
 by c, thus X c = ^Pd/c, where c and the d's are measured to the same convenient scale. 
 
 The A and B influence areas are easily found when the X c line is given. Thus in 
 Fig. 47 AC, the line B'C'A" represents the A /A a influence line with some factor, and 
 the shaded area is the A influence area with a factor I/TJA to be proven. 
 
 For this proof the end ordinate A' A" is evaluated according to Eq. (46s), and the 
 factor required for this ordinate will be the factor for all ordinates TJ of the A influence 
 area. 
 
 For the point A, A = l and A c = l-l 2 /l and X c has the factor 1/c. Hence ~A f A c = 
 1/1 2 and the end ordinate should be 
 
 i 2 \i x c -\ i 2 \d I 
 = ~ = '' ...... (7B) 
 
 where the Tactor is l 2 /cl. 
 
 But = - = , hence the factor is simplv \fr\A. 
 
 d C T) A ' 
 
ART. 47 
 
 SPECIAL APPLICATIONS OF INFLUENCE LINES 
 
 145 
 
 k 9i 
 
 rlW 
 
 
 P 
 n t 5 
 
 1= fZ.U 
 6 
 
 7 
 
 a 3 
 
 10 
 
 u 
 
 i 
 i 
 
 1 
 
 1 
 
 3 
 
 S.94 
 
 *u= 
 
 FIG. 47x. 
 
146 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP.X 
 
 . 
 Similarly \/r IB is the factor for the B influence area indicated in Fig. 47 AC as the 
 
 area included between the X c line and the line A'C'B". 
 
 Hence the reactions A and B, for any train of concentrated loads, become 
 
 A= 2Pr y and B= 2Pr i '. (47 c) 
 
 f)A "T iB .-. 
 
 The algebraic signs of the areas are determined as before. All areas below the X c 
 line are positive and those above this line are negative. The ordinates may be measured 
 with any scale so long as the same scale is used for all. 
 
 The M influence area is a portion of the A area for all points between A and (7, 
 while for points between C and B the M area is a portion of the B area. See Fig. 
 4?Ad. 
 
 Thus for the point 4 the M 4 influence area is shown as the shaded area between 
 the X c line and the broken line A'-i'B'. The factor is M C /C=X/TJ A , where x is the 
 ordinate of the point 4. When x is expressed in feet the resulting moment is 
 in ft. Ibs. 
 
 Similarly the M-j influence area is indicated by dotted lines and the factor is 
 M c /c=x/r) B . 
 
 The usual directions as to algebraic signs and scales apply as before. 
 
 The Q influence area for shear is shown, in Fig. 4?Ae, for the point 4 as the shaded 
 area with a factor Q C /C = \/T)A the same as for the A influence are.a. The line A'4 7 is 
 parallel to B'C'4". 
 
 Similarly the Q 7 influence area is indicated by dotted lines and has a factor Q c /c = 
 I/T)B' B'7' is parallel to A'C'7". The signs are again determined with reference to 
 the areas above and below the X c line. 
 
 All of these influence areas have a load divide at the point C' and the Q areas have 
 two load divides each. This determines the positions of train loads for positive and 
 negative effects, and to obtain maximum effects, the heaviest loads should be placed 
 over the maximum ordinates. 
 
 Temperature effects. A uniform change in temperature will expand the girder 
 equally in all directions and will produce a slight lifting of the ends A and B equal to 
 d(72- 12) =0.0105 inch, for =25 F. and =0.000007. When the three supports are 
 on the same level then no stress will be produced by uniform temperature changes. 
 
 However, when the sun shines down on such a structure experience teaches that 
 the top flange is heated much more than the bottom flange and this difference in tem- 
 perature may become quite considerable and will cause the girder to assume a curved 
 position, convex upward, when the top flange is warmer than the bottom. 
 
 The maximum difference in temperature between the two flanges is bound to remain 
 somewhat problematic, but observations indicate that differences of 30 F. are quite 
 common. In the present example Jf=+20 F. will be assumed and on this basis and 
 that the temperature varies uniformly between the two flanges, a set of w t elastic loads 
 is now computed from the formula w t = -sJUx/h = -0.00014 dx/h, where Ax is the 
 horizontal distance between sections and h is the depth of girder, both in inches. 
 
AKT. 47 SPECIAL APPLICATIONS OF INFLUENCE LINES 
 
 The following table gives the w t loads: 
 
 147 
 
 Point. 
 
 dx 
 Inches. 
 
 h 
 
 Inches. 
 
 l 
 
 10,000u> ( 
 
 Point. 
 
 dx 
 Inches. 
 
 h 
 Inches. 
 
 w t 
 
 10,000u>, 
 
 1 
 
 60 
 
 19.5 
 
 0.00043 
 
 4.3 
 
 7 
 
 60 
 
 72 
 
 0.000117 
 
 1.17 
 
 ') 
 
 60 
 
 34.5 
 
 0.00024 
 
 2.4 
 
 8 
 
 60 
 
 72 
 
 0.000117 
 
 1.17 
 
 3 
 
 60 
 
 49.5 
 
 0.00017 
 
 1.7 
 
 9 
 
 45 
 
 64.5 
 
 0.000098 
 
 0.98 
 
 4 
 
 60 
 
 64.5 
 
 0.000115 
 
 1.15 
 
 10 
 
 45 
 
 49.5 
 
 0.000127 
 
 1.27 
 
 5 
 
 90 
 
 72 
 
 0.000175 
 
 1.75 
 
 11 
 
 45 
 
 34.5 
 
 0.000183 
 
 1.83 
 
 6 
 
 90 
 
 72 
 
 0.000175 
 
 1.75 
 
 12 
 
 45 
 
 19.5 
 
 0.000323 
 
 3.23 
 
 An equilibrium polygon drawn for these w t loads with a pole distance unity, would 
 represent the deflection curve of the beam due to M. However, for convenience it is 
 better to multiply these small loads w t by some factor as 10,000 and then make the 
 pole distance 10,000 instead of unity. 
 
 The scale of lengths for the girder was made 1 : 120 and if the ordinates are to appear 
 say 5 times actual, the pole distance should be made equal to H = 10,000/5X120 = 
 16.67, using the scale chosen for the w t loads. 
 
 Fig. 47Af shows the force and equilibrium polygons for the w t loads, with 
 ordinates five times actual. Since these w t loads are all negative for +At, the 
 3 ct deflection curve was drawn above the closing line A 'B'. The ordinate 
 under the point C represents 5d ct and this must be increased by the small 
 ordinate 5x0.0105, previously found for the uniform rise in temperature to obtain 
 total effect. 
 
 According to Eq. (44c), the redundant reaction produced by this temperature 
 effect is X ct =d ct /d ee . The deflection polygon for X e = l kip, gives 100 cc =0".93, 
 whence + cc =0".0093, and the d ct deflection polygon gives -5d c< =0".74+0.05=0".79, 
 making 3 ct = 0'M6. 
 
 Hence X ct =d et /d ce = -0.16/0.0093 = -17.2 kips, which indicates a downward reaction 
 at C to maintain the girder on the middle support. As the entire girder has an approxi- 
 mate weight of 7440 Ibs. the above temperature effect would lift the girder off the center 
 support and cause the beam to carry itself on two supports, producing a compression 
 of about 3100 Ibs. per sq.in. on the extreme fiber. This temperature stress would not, 
 however, be fully developed unless the span is loaded down in contact with the center 
 support. 
 
 The dead and live load stresses must finally be combined with the temperature 
 effect to obtain the real stress at any point. 
 
 The application of these influence areas is illustrated by placing a single load P n at 
 any point n of the span and showing the values of the several functions. The same 
 process is followed for each load of a train of loads and the total effect of the train is 
 the sum of the individual load effects. 
 
 The ordinates r? under the point n, in each of the influence areas, are all scaled to 
 the same scale to which 7j_ 4 and rj B are measured. This may be any convenient scale, 
 which is universally used for all ordinates of the drawing. 
 
148 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. X 
 
 The values of the various functions for the load P n are 
 
 Y fnOn _ ' -p _ 
 
 A c = -- /" 
 
 C 57. 
 
 B = _M = _-M- PB = - .207P n ; 
 ijjs 15.94 
 
 +-B=P, which checks to 1 percent. 
 
 n =49.4P n in.lbs. for P n Ibs. 
 
 X ct l 2 a 17200X300X194 '. 
 
 M 4t J 2 - = - 72Q =1,390,330 in.lbs. 
 
 ART. 48. TRUSS ON THREE SUPPORTS 
 
 Figs. 48A illustrate the analysis by influence lines of a truss on three supports. 
 The bottom chord is the loaded chord and the lengths and cross-sections of the members 
 are written over the members in Fig. a, while the stress SA, produced by a unit reaction 
 at A on the simple span AB, are written below the members. The stress diagram, 
 from which the SA stresses are found, is shown in Fig. b. The solution is similar to 
 that given in Art. 47. 
 
 As in the previous problem, Art. 47, the present structure involves one external 
 redundant condition, which is again taken as the middle support C, reducing the truss 
 to a principal system on two supports, A and B. 
 
 The X c influence line is derived from Eq. (42o) as in the previous problem and 
 becomes 
 
 (48A) 
 
 where the variable influence line ordinate d mc is the ordinate for a deflection polygon, 
 drawn for the loaded chord and for the conventional loading X c = 1 kip acting downward 
 at C on the prinicpal system or simple truss AB. The influence line factor ju = l/ cc . 
 The elastic loads for this deflection poylgon are found from Eqs. (36s) as : 
 
 M c I 
 
 - (48B) 
 
 for each pin point of the top and bottom chords. The effect of the web members is 
 usually neglected as being insignificant, but when it is desired to include their effect, 
 Eqs. (36c) will give the w loads due to the web members and these are added to the 
 
ART. 48 
 
 SPECIAL APPLICATIONS OF INFLUENCE LINES 
 
 149 
 
 chord loads found from Eq. (48R). See Art. 36, for a complete discussion of this method, 
 and Art. 50 for a complete problem. 
 
 A Williot-Mohr diagram might also be employed to obtain this deflection polygon. 
 See also the example in Art. 50, Fig. 50B. 
 
 In the present example the web system will be_ neglected and the w loads arc found 
 for the chords, using Eq. (48 B). The moments M are those produced by the conven- 
 tional loading X c = 1000 Ibs. and are represented in the moment diagram, Fig. b, expressed 
 in inch-lbs. The ordinate at the point C will be P//4 =528,000 in. Ibs. 
 
 Using these moments and the values for /, F and r given, in Fig. a, for each member, 
 the following computation of the w forces is made: 
 
 w 
 
 Member. 
 
 A/c 
 Inch-lbs. 
 
 I 
 Inch. 
 
 F 
 Square Inches. 
 
 r 
 Inch. 
 
 Md 
 Fr* 
 
 1 
 
 L L 2 
 
 127,125 
 
 508.5 
 
 22.7 
 
 384 
 
 14.8 
 
 2 
 
 t/!^ 
 
 254,250 
 
 509.0 
 
 42.0 
 
 384 
 
 20.8 
 
 3 
 
 L 2 L 4 
 
 381,375 
 
 508.5 
 
 22.7 
 
 384 
 
 57.7 
 
 4 
 
 C7 3 f7 5 
 
 528,000 
 
 299.3 
 
 44.7 
 
 434.9 
 
 2X18.7 
 
 5 
 
 L 4 L 6 
 
 381,375 
 
 508.5 
 
 22.7 
 
 384 
 
 57.7 
 
 6 
 
 7 5 t/ 7 
 
 254,250 
 
 509.0 
 
 42.0 
 
 384 
 
 20.8 
 
 7 
 
 L 6 L 8 
 
 127,125 
 
 508.5 
 
 22.7 
 
 384 
 
 14.8 
 
 
 
 
 
 * 
 
 z. 
 
 224.0 
 
 The modulus #=29,000,000 /6s. per xq. inch, was not embodied in the tabulated 
 values, hence the w loads are E times too large and the deflections would be natural size 
 for a pole H=E. But the scale of lengths of the drawing was 1:300, and wishing to 
 make the deflections 400 times actual the pole must be made equal to E/3QQ X400 =241. 1 'w 
 units. 
 
 The force and equilibrium polygons, Fig. c, are then drawn, using any convenient 
 scale for w forces. The X c influence line is thus found as the polygon A CB, with a 
 factor ft = l/c.' The actual deflection of the point C is d ee = l". 07/400=0". 0027, 
 obtained by measuring the ordinate c in inches. All influence ordinates are measured 
 with the same scale, which may, however, be any convenient scale, because the factor 
 JJL makes the quotients d mc /d cc independent of an absolute scale. For this same reason 
 the M diagram might have been drawn with any middle ordinate as unity, though 
 when the actual deflection d cc is wanted for temperature investigations, the method 
 here given is less confusing. 
 
 The A influence area is now found by combining the X c influence line with the 
 ordinary A line for the span AB, in such manner as to make the ordinate at C equal 
 to zero and then finding the factor u which reduces the ordinate at A to unity. 
 
 The line A'CB, Fig. c, is the ordinary A influence line with a factor I/IJA, and 
 when this is combined with the X c influence line, the areas included between the two lines 
 will represent the A influence area with a factor ^ = !/TJA =0.0187. 
 
 Since it is more difficult to redraw the X c line so as to make the ordinate c = l, 
 than it is to apply the factor p, the latter method is decidedly preferable to the one 
 frequently given. 
 
150 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. X 
 
 t i ? 1 
 
 I A INFLUENCE AREA , |fX/rk=+0.|Ql87 
 
 INFLUECE AREA , 
 
 Upper chord 
 I Lower chord 
 
 Tension is + , compreision is 
 
ART. 48 SPECIAL APPLICATIONS OF INFLUENCE LINES 151 
 
 The point C is a load divide for both reactions A and B, and in the present case 
 the same influence area will serve for both these reactions, because l\ =1 2 . When the 
 spans are unequal, then the B influence area is included between the X c line and the 
 line AC prolonged to the vertical through B and the factor then becomes ,B = I/>?,B. 
 
 The influence area for a chord member L 2 L 4 is obtained by combining the ordinary 
 S stress influence line of the member for the principal system with the A" c influence 
 line, Fig. d. 
 
 A load at C can produce no stress in any member of the indeterminate truss, hence 
 the ordinate under C, for every stress influence line, must be zero. Therefore, the line 
 BCA' must be common to all stress influence lines of the members between A and C, 
 web members included. The point C must be a load divide for all members. 
 
 The center of moments for the chord L 2 L 4 is at C7 3 , hence the line A3, Fig. d, 
 completes the S stress influence line with a factor SA/T)A, where SA is the stress in the 
 chord L 2 L 4 for a unit reaction at A. 
 
 The area included between the X c polygon and the S influence line is thus the 
 required influence area for the chord L 2 L 4 with the factor JJ.=SA/^A- The area below 
 the X c polygon is positive as usual. The stress *SU = +1.99, hence the factor 
 /i =+1.99/53.6 =+0.0371. 
 
 If the vertical L 3 t/ 3 were absent, then the S line would have to be straight over 
 the panel L 2 Z/ 4 and a line 2C would be necessary to complete the S influence line. 
 With the member L 3 t/ 3 acting, the chord stress L 2 L 4 is, therefore, greater than when 
 the vertical is omitted. 
 
 The S influence lines and their factors for all chord members are shown by dotted 
 lines in Fig. d, and the final influence areas are always positive below the X c line. The 
 factors ft are negative for the top chords because SA is then negative. 
 
 The stress in the member L 2 L 4 , for any train of moving loads, is expressed by 
 
 (48c) 
 
 where each ij is the ordinate of the shaded influence area vertically under its lespective 
 load P and S A is given in the same units as the loads. 
 
 This influence area is also the M 3 influence area with a factor fi=x 3 /i)A and gives 
 moments in foot units when x 3 is measured in feet as indicated. 
 
 The scale of the influence ordinates is immaterial so long as all ordinates are 
 measured with the same scale. This is apparent from Eq. (48A). 
 
 The influence area for a web member L 2 U 3 is found from exactly similar considerations 
 as those shown to exist for the chords. The shaded area in Fig. e is the influence area 
 for the web member L 2 U 3 . The S lines for the other web members are indicated by 
 dotted 'lines. 
 
 The \me-BCA', Fig. e, is again one of the limiting lines of the S influence line for 
 each web member, and when the chords are parallel in the panel containing the particular 
 web member, then the other -limiting line A2' will be parallel to BC and the line 2'3 
 completes the S line for the member L 2 U S . 
 
 The load divide i for this member, considering the principal system, must fall 
 
152 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP, x 
 
 vertically over the point i' where the line 2'3 intersects the base line AH. This serves 
 as a check or when the chords are not parallel, as in the panel L 3 L 4 , then it may be 
 used to complete the S line as indicated for the member t/ 3 Z/ 4 with the load divide t. 
 
 In case the chords are not parallel and the center of moments for the diagonal 
 falls outside the drawing, then the ordinate 753 as for the member U S L 4 , may be com- 
 puted from the formula 
 
 r/ 3 x 3 +a 3 Jx 3 +a 3 \ /63.6+91\. Q 
 
 u-ir or ^ 3= t^> 4 = r^r-r 3 - 6 < ^ 
 
 where x 3 is the distance L L 3 from A to the panel containing the web member, and a s 
 is the distance from A to the center of moments of the member, both in the same units, 
 as feet. 
 
 The factor {J.=SA/T)A again applies and the sign of SA determines the sign for //. 
 
 Stresses due to temperature. When the three supports A, B and C are on the same 
 level then a uniform change in temperature produces no stresses in the structure. 
 However, when the sun illuminates the bridge from above, there is usually a difference 
 in the temperature of the two chords. Assuming this difference as Jf=20 F. as in 
 Art. 47, then the bottom chord would be cooler than the top chord by this amount and 
 the changes in the lengths of the bottom chord members would be eMl and the w t 
 elastic loads would be 
 
 (48 E ) 
 
 These loads being extremely small, it is well to multiply them by say 10,000 and 
 then using a pole distance of 10,000, the d cf deflections would be natural size. However. 
 the scale of lengths on the drawing was 1 : 300 and for deflections five times natural the 
 pole becomes 
 
 10000 rr _ 
 = 5X300 * units. 
 
 The loads w t are figured for the panel points 1, 3, 5 and 7, taking =0.000007, 
 J*=20 F. and r=384 inches. The d ct deflection polygon is shown in Fig. d, and has 
 negative deflections, five times actual. Hence o ct =0".95/5 =0".19 and d cc was previously 
 found from Fig. e, as 0".0027, hence from Eq. (44c) 
 
 This value is in kips because d cc is the actual deflection for one kip, and 
 X ct being negative produces the same effect on the principal system as a load X ct hung 
 at the point C. The reaction at A, due to a load of 70.4 kips at C, would then 
 be A t =70.4Z 2 /(Zi +Z 2 ) = +35.2 kips. 
 
 The stresses in the members are then S t =S A A t =35.2 S A kips. The stresses 
 S A , due to a unit reaction at A, are already found in Fig. b, hence the tem- 
 perature stresses are readily determined. Thus for the member U 3 U 4 , S A = -2A2, 
 hence S t = -2.42 X35.2 = -85.2 kips. The negative sign indicates compression. 
 
ART. 49 
 
 SPECIAL APPLICATIONS OF INFLUENCE LINES 
 
 153 
 
 ART. 49. TWO-HINGED SOLID WEB ARCH OR ARCHED RIB 
 
 This and the two-hinged framed arch are perhaps the most common structures 
 involving external redundancy which are met with in practice. They will receive 
 special attention here as deserving a prominent place among commendable structures. 
 The external redundancy may be said to offer less objection here than in any other 
 class of structures. 
 
 The present theory will be developed in its most general application to unsym- 
 metric arches and will be equally applicable to masonry, concrete or steel arched ribs. 
 
 Fig. 49A represents an unsymmetric two-hinged arched rib of any cross-section and 
 the lettered dimensions are in general the same as those previously employed for three- 
 hinged arches in Art. 28. 
 
 The arch thrust along AB is treated as the external redundant condition. When 
 it is removed, by replacing the hinged support at A by a roller bearing, the simple beam 
 AB (though curved) on two supports, then becomes the principal system. 
 
 FIG. 49A. 
 
 According to Eqs. (7 A), the following values for the reactions, thrusts and moment 
 for any point ra may be written: 
 
 A =A A a X a , where 
 B=B -B a X a , where 
 
 and A a =-l-sina 
 
 
 and 
 
 = l- since 
 
 M m =M -M a X a , where M =A x-P(a' -x') and M a = l-ycosa 
 
 N m = N - N a X a , where N = Q sin <f> ; N a = - 1 cos (<f> -a) 
 
 T m = T -T a X a , where T =Q cos<f> = (A - X p) cos<; 7 7 = l-sin(0-a) 
 
 (49A) 
 
154 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. X 
 
 in which A and B are the vertical end reactions; M m is the moment of the external 
 forces about any axial point m; R m is the resultant thrust, on the normal section mm', 
 produced by all the external forces acting on one side of this section; N m is the com- 
 ponent of R m , normal to the section mm'; and T m is the tangential component of R m 
 in the section mm'. Q is the vertical shear at any point m produced by all the loads 
 P acting on a simple beam AB. The conventional loading X a = \ is applied as indicated 
 in the opposite direction of X a . 
 
 Eqs. (49A), with the special values inserted, thus become 
 
 SPa' . 
 
 A = , -- 1- X a sm a 
 
 v . 
 X a sm a 
 
 I 
 
 M m =M X a y cos a 
 N m =Q sin </> +X a cos (< -a) 
 T m =Q cos < X a sin (<f> -a) J 
 
 / 
 
 SPa 
 
 A,, sm a 
 
 (49s) 
 
 where M and Q have the values in Eqs. (49A). 
 From Fig. 49A, 
 
 M m =N m v; R m =VN m 2 + T n ?; H=X a cosa. ^ . . . 
 
 In the above equations, all terms except X a are derived from the static conditions 
 and the solution becomes possible when X a is determined. 
 
 The X a influence line for the redundant haunch thrust. When the haunches are on 
 the same level then this thrust X a becomes the horizontal thrust H. 
 
 The equation for X a is given by Eq. (42o) as 
 
 where d ma is the vertical deflection ordinate of any point m of a deflection polygon, 
 drawn for the loaded chord and for the conventional loading X a = l applied to the 
 principal system, and o^ is the change in the length AB, due to X a = \ acting on 
 the principal system. 
 
 Since dma and d^ are not parallel displacements, they cannot be obtained from 
 the same deflection polygon as in the previous examples, Arts. 46^8. This circum- 
 stance necessitates the construction of an extra displacement polygon for daa, if the 
 graphic solution is strictly followed, or of finding d^a by computation, which is usually 
 advisable. 
 
 The X a influence line thus becomes the d TOa deflection polygon with a factor p. = l/aa- 
 Should this deflection polygon be constructed for a pole distance H =daa then the factor 
 
 p-1. 
 
 According to Art. 40, the dma deflection polygon is the equilibrium polygon drawn 
 for a set of elastic loads w with a pole distance H = 1. These elastic loads represent 
 partial areas of a moment diagram drawn for the conventional loading X a = l. 
 
ART. 49 SPECIAL APPLICATIONS OF INFLUENCE LINES 155 
 
 The moment produced by X a = 1 about any axial point m is 
 
 M, n = 1 y cos a .......... (49E) 
 
 and the elastic loads, by Eq. (39i) become 
 
 M Au Axy cos a 
 
 ^-JT--J -- -r, 
 
 El El cos (j> ' 
 
 (49F) 
 
 where Au is the width of a partial moment area measured along the axial line, and Ax 
 is the horizontal distance between the vertical moment ordinates, making Au=4x/cos<j>. 
 
 The deflection d aa may be obtained from a deflection polygon drawn for the same 
 set of w loads by allowing these loads to act in a direction parallel to X a , as per Art. 38. 
 
 According to Eq. (38B) the deflection of any point in any direction, may be expressed 
 as the sum of the moments of all w loads on one side of the point, when the direction of 
 the w loads is taken parallel to the required direction of the deflection. 
 
 Hence 
 
 a daa = ^T ^(J/ COS ) 
 
 (49o) 
 
 which affords a purely analytic solution for computing the X a influence line ordinates 
 and also the d m displacement. 
 
 Since the w loads are best found by computation, it is a comparatively easy task 
 also to compute the values wy cos a and thus obtain d aa = I>wy cos a, which is the 
 required pole distance for the d ma deflection polygon. Hence according to Eq. (49n) 
 
 i 
 and the X a influence line is an equilibrium polygon drawn for the elastic loads w with 
 
 a pole distance H = Swn/ cos a. 
 
 Since E enters into w, and hence into each term of Eqs. (49o), as a constant factor, 
 which cancels in the numerator and denominator of Eq. (49n), it would be proper to 
 multiply Eq. (49r) by E and thus make the elastic loads equal to Ew without affecting 
 the ordinates r? a from Eq. (49n). Should it be desirable to obtain values of d ma from 
 the X a influence line, then the jj a ordinates must be divided by E and multiplied by 
 
 cos a. 
 
 . 
 
 For very flat arches, the effect of the axial thrust should be considered in the 
 determination of X a , by allowing for the quantity d f m produced by N a , Eq. (49A), on 
 the displacement daa- 
 
 Thus for N a = -I cos (< -a) and Au = Jar/cos <j>, then from Eq. (15.v), 
 
 2 /j 
 
 - 
 
156 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. X 
 
 which is the effect due to N a only, and this added to the displacement previously found 
 for moments only, gives the total displacement 
 
 2 
 
 (49HH) 
 
 Hence, when axial thrust is to be considered in Eq. (49n), then the value 3 aa should 
 be computed from Eq. (49HH) and used as the pole distance H in drawing the equilibrium 
 polygon for the elastic loads w. When the loads w are taken E times actual, then the 
 E in the second term of Eq. (49HH) is omitted. 
 
 Temperature stress. For any change in temperature above or below the normal, 
 X at is expressed by Eq. (44c) as 
 
 X at =m^, ..... V . '.I/ .. . (49i) 
 
 Oaa 
 
 wherein d al is the change in the distance AB due to any change in temperature of 
 acting on the principal system. 
 
 When the structure retains a uniform temperature then the change t will be uniform 
 in all directions so that the shape of the structure will always be similar to its normal 
 geometric shape. Hence, the distance AB will change as for any case of linear expan- 
 
 sion and 
 
 dl 0.000488/ 
 
 for e =0.0000065 per 1F., and Z = 75. 
 
 When the change in temperature is not uniform, as when the upper flange is +Jl c 
 warmer than the lower flange, then d at must be found from a deflection polygon drawn 
 for a set of w t elastic loads computed from the formula w t = edtJu/D, where D is the 
 depth of section and Au is the length between sections measured along the axis. The 
 pole distance, if made equal to unity, will give the actual o at , and if the pole be made 
 equal to d aa = 'Zwy cos a as for X a , then the displacement found will be X nt measured 
 parallel to AB. The w t loads must likewise be taken parallel to AB in drawing the force 
 and equilibrium polygons since d at is a displacement along AB. 
 
 X at will be positive for +t, while for + At the w t loads are negative and X at would 
 also be negative. 
 
 In any case, when X at is determined, then the quantities M mt , N mt , and T mt are 
 found from Eqs. (49fi) by omitting all the terms involving the effects of the loads P, 
 thus: 
 
 M mt = ~X at y cos a =N mt v 
 
 N m t= X at cos((f>-a) 
 T mt = -X at sin (< -a) 
 
 Abutment displacements. When the abutments undergo displacements Ar in the 
 directions of the reaction forces R, then, for the case of external redundancy where 
 d a =0, Eq. (8D) gives for P=0, 
 
ART. 49 SPECIAL APPLICATIONS OF INFLUENCE LINES 157 
 
 where R a has the several values A a =-l-sina, B a = \- sin a, and X a = l, while Jr 
 represents the vertical displacements A A, AB and a change M in the length of span I. 
 Hence the redundant X ar , due only to abutment displacements, is 
 
 S/g Jr _ A a 4 
 
 Xar ix 
 
 or 
 
 JB sin a A A sin a + M sec a. 
 
 X-ar = 5~~' ' " " 
 
 When AA=AB, which is usually the case, then 
 
 . (49L) 
 
 showing the effect due to an elongation M, in the span I, to be precisely the same in 
 character as that due to a uniform fall in temperaturegiven by Eq. (49i). 
 
 For a two-hinged arch with a tension member AB, and one expansion bearing, the 
 problem becomes one involving an internal redundant member X a and Eq. (42s) is 
 then used vvith the term p a in the denominator. The solution would otherwise remain 
 
 unchanged. . 
 
 Stresses on any arch section. Neglecting curvature, which is always permissible, 
 the stress on the extreme fiber of any arch section mm' is given by Navier's law as 
 
 N My (4j)M) 
 
 J~F I ' 
 
 where N is the normal thrust on the section, M=Nv is the bending moment about the 
 gravity axis F is the area of the section, y is the distance from the gravity axis to the 
 extreme fiber and I=Fr* is the moment of inertia of the section about the gravity axis. 
 
 Fig 49B gives a graphic representation of the distribution of stress on any arch 
 section mm' in accordance with Eq. (49M). The center of gravity is at c and the uniform 
 axial stress f m =N/F is the stress ordinate at c, while the stresses on the extreme fibers 
 /. and /,-, are applied as ordinates giving the line tt' as representing the uniform , 
 tribution of stress over the section. . 
 
 The point e is the kernel point for the extrados while * is the kernel point for the mtrados. 
 These points are determined by the distances k.=i*/e and k^fi/i, where r is the radius 
 of gyration of the section and e and i are the respective distances to the extreme f 
 the section measured from the gravity axis c. 
 
 When N and * are given, the line tt' is easily constructed as indicated by prolong- 
 ing i?_to 6 to find t, and by prolonging ?ito a to find f. The stress /. is then the inter- 
 cept bn while the stress f t is the intercept an. 
 
158 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. X 
 
 Using e and i as subscripts, referring respectively to extrados and intrados, then 
 for M=Nv and I^Fr 2 , Eq. (49M) gives 
 
 and 
 
 (4*0 
 
 The moments about the kernel points e and i are evaluated from the figure as 
 
 N(v+k e )=M e and N(v-kd=Mi ........ (49o) 
 
 FIG. 49B. 
 For r 2 /e=k e and r' 2 /i=k{, Eqs. (49N) become 
 
 - _i 4. = _ - _j? 
 
 F\ + k e ) F\ k e I Fk e 
 
 F \ ki/ ''F \ ki i Fki 
 Also solving Eqs. (49o) for N and v, then 
 
 M e Mi 
 
 N=, -r and v = 
 
 (49?) 
 
 Me -Mi 
 
 (49 Q ) 
 
 Hence Eqs. (49p) and (49cj) furnish a complete solution for the stresses f e and /, 
 on the extreme fibers, the normal thrust N and its distance v from the gravity axis, 
 for any unsymmetric section tt' of an arch ring, in terms of M et Mi, k e and &,. 
 
ART. 49 
 
 SPECIAL APPLICATIONS OF INFLUENCE LINES 
 
 159 
 
 When the section is symmetric about a horizontal gravity axis, which is usually the 
 case, then these equations give, for e=i=D/2, or the half depth between extreme fibers, 
 
 _- __ _ 
 = ,-i _______ 
 
 M e Mi 
 
 =s ~ and i== 
 
 Me-Mi 
 
 1. T "~ , Ctllvl 
 
 i) M e +Mi 
 
 (49R) 
 
 2k M e -Mi 2N 
 
 For a rectangular section of unit thickness and of depth D, making e=i=D/2. 
 
 fc _* x> F==1 . D} /=i^- 3 
 
 M e +Mi 
 
 N=- 
 
 and ?'= : 
 
 2N 
 
 (49s) 
 
 In all the above formulae M e and Mi have opposite signs when N acts between the 
 two kernel points e and i. When v>ki, both moments are negative and when v is negative 
 and larger than k e , both moments are positive. The figure shows v to be positive 
 when measured from the gravity axis at c toward the extrados. The stresses f e and 
 fi take their signs from M e and Mi respectively, and compression is regarded as a negative 
 stress. 
 
 The stresses on any arch section mm' may, therefore, be found for any simultaneous 
 position of a moving train of loads when the influence areas for M et M t and T m have 
 been drawn. 
 
 The resultant polygon for any simultaneous case of loading may also be drawn by 
 plotting the offsets v from the gravity axis of each section examined, observing that 
 +v is toward the extrados and v is toward the intrados. 
 
 The method of combining the X a influence line with ordinary moment and shear 
 influence lines to obtain the M e , Mi and T m influence areas for any section mm' will 
 now be illustrated. 
 
 Kernel moment influence areas. The moment equations for the kernel points accord- 
 ing to Eq. (49B) are 
 
 ,_ \M oe ! [ M oe 1 , 
 
 =M ae \ -a-, X n \=y e cos a \- X a \ 
 
 lM ae J [j/ecos a 
 
 ' 
 
 (49r) 
 
 . 
 
 The intercept of the M oe influence line on the vertical through A is simply the 
 ordinate x e of the point e. Similarly for the M oi influence line this intercept is the ordinate 
 
160 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. 3 
 
 Xi of the point i. Hence the positive intercepts on the vertical through A, of th 
 M oe /y e cos a and of the M^/yi cos a influence lines, are easily computed when the coordi- 
 nates of the kernel points e and i are given, provided the X a influence line was drawn 
 for a factor /* = !. 
 
 The coordinates of the kernel points are derived from the coordinates (x m y m ) of the 
 axial point for the same section, when the kernel distances k and the angle are known 
 for that section. From Fig. 49s, 
 
 , 
 x e = x m +k e sm<f>, y e = y m -k e cos(j> } 
 
 I, V 49u) 
 
 X{ =x m -ki sin 0, yi =y m +k { cos ^ J 
 
 which apply to all points from A to the crown, and from there to the abutment B the 
 signs of the last terms are reversed. By using the I x ordinates the intercepts on 
 the vertical through B are found. 
 
 One intercept only need be computed as the two limiting rays of the M /y cos a. 
 line must intersect on the vertical through the center of moments. See Fig. 49c. 
 
 By constructing the X a influence area for // = !, which may always be done by making 
 the pole distance H = 2wy cos a, and since both the X a and the M /y cos a lines are pos- 
 itive, then by applying them below the closing line A'W, the area included between the 
 two lines will represent the M m influence area. The portion below the X a line will 
 be the positive area because X a is subtractive in Eqs. (49-r). The entire M m influence 
 area has a factor p. =y cos a =M a . 
 
 The T m influence area for tangential force on any section mm'. From Eqs. (49u) 
 the following equation for T m is obtained. 
 
 From this the end ordinate at A, for the T /T a Yme, becomes 1- cos <^ sin (0 a) 
 because the end ordinate for the Q line is unity. In this case the end ordinate at B is 
 numerically the same but negative. For axial points to the right of the crown the end 
 ordinate at B is positive and the one at A is negative. Other details are illustrated 
 in connection with the example. The final T m influence area has a factor /i=sin (<-a). 
 
 Example. A two-hinged arched rib, modeled after the Chagrin River Bridge near 
 Bentleyville, O., was selected to illustrate the application of the previous theory. 
 The structure was made unsymmetric by shortening the span at the right-hand end as 
 shown in Figs. 49c. The clear span thus became 164 ft. and the rise 27.89 ft. This 
 bridge was designed to carry a live load of 24 kips per truss per panel. The arch section 
 is composed of a f-inch web plate and 4-6" X6" X 11/16" angles, with 2-14" X7/16" flange 
 plates on each flange. Besides these there is one 14" X|" plate on each flange extend- 
 ing from sections 2 to 5 and 7 to 10. The sections are all symmetric about the gravity 
 axis and all general dimensions are given on the drawing and in Table 49A. 
 
 The bridge consists of two steel arched ribs 27 ft. between centers and carrying a 
 total dead load of 1,058,000 pounds. The following values are assumed as a basis 
 
ART. 49 
 
 SPECIAL APPLICATIONS OF INFLUENCE LINES 
 
 161 
 
 X ordinores are ?.a times 
 I toicole of Ittjfeth ,. 
 
 M e INFLUENCE AREA. >\=y g CoiqC-18.-l 
 
 B' 
 
 FIG. 49c. 
 
162 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. X 
 
 for the analysis: #=29,000 kips per sq.in. =4,176,000 kips per sq.ft.; e -0.0000065 
 per 1F.; andZ = 75F. 
 
 The first step in the analysis is to compute the Ew loads from Eq. (49r) using the 
 tabulated dimensions in Table 49A. In the same table the values Ewy cos a are computed, 
 giving the required pole distance H=E2wycosa for constructing the X a influence line 
 in accordance with Eq. (49n). 
 
 TABLE 49A 
 Xa INFLUENCE LINE 
 
 Sec- 
 tion. 
 
 Coordinates of 
 Gravity Axis. 
 
 <*> 
 
 l 
 
 / 
 
 in. 4 
 
 / 
 
 ft. 
 
 6 = 
 1 
 
 Jx 
 
 ft. 
 
 0J.r = 
 Ax 
 
 Ew = 
 Axy cos a 
 
 Ewy cos a 
 
 X 
 
 ft. 
 
 y 
 ft. 
 
 cos <t> 
 
 cos <f> 
 
 ft, 
 
 1 COS (j> 
 
 ft. 
 
 I COS (j> 
 
 ft. 
 
 
 
 
 
 
 
 36 15' 
 
 1.240 
 
 74,180 
 
 3.578 
 
 0.3466 
 
 9.4 
 15 
 
 15 
 15 
 15 
 
 15 
 15 
 
 15 
 15 
 15 
 
 15 
 
 4.6 
 
 0.652 
 
 
 
 
 
 1 
 
 9.4 
 
 6.01 
 
 32 15 
 
 1.182 
 
 74,180 
 
 3.578 
 
 0.3260 
 
 4.954 
 
 29.8 
 
 179 
 
 2 
 
 24.4 
 
 14.12 
 
 26 00 
 
 1.113 
 
 64,710 
 76,530 
 
 3.121 
 3.691 
 
 0.3567 
 0.3017 
 
 4.939 
 
 69.7 
 
 984 
 
 3 
 
 39.4 
 
 20.30 
 
 20 00 
 
 1.064 
 
 66,170 
 
 3.191 
 
 0.3332 
 
 5.014 
 
 101.8 
 
 2065 
 
 4 
 
 54.4 
 
 24.73 
 
 14 15 
 
 1.032 
 
 56,750 
 
 2.737 
 
 0.3770 
 
 5.673 
 
 140.3 
 
 3468 
 
 5 
 
 69.4 
 
 27.24 
 
 7 30 
 
 1.009 
 
 48,160 
 40,420 
 
 2.323 
 1.949 
 
 0.4352 
 0.5176 
 
 7.198 
 
 196.0 
 
 5339 
 
 6 
 
 84.4 
 
 27.89 
 
 00 
 
 1.000 
 
 33,590 
 
 1.620 
 
 0.6173 
 
 9.011 
 
 251.3 
 
 7007 
 
 7 
 
 99.4 
 
 26.77 
 
 7 30 
 
 1.009 
 
 40,420 
 48,160 
 
 1.949 
 2.323 
 
 0.5176 
 0.4352 
 
 7.198 
 
 192.6 
 
 5156 
 
 8 
 
 114.4 
 
 23.58 
 
 14 15 
 
 1.032 
 
 56,750 
 
 2.737 
 
 0.3770 
 
 5.673 
 
 133.7 
 
 3153 
 
 9 
 
 129.4 
 
 18.58 
 
 20 00 
 
 1.064 
 
 66,170 
 
 3.191 
 
 0.3332 
 
 5.014 
 
 93.1 
 
 1730 
 
 10 
 
 144.4 
 
 11.82 
 
 26 00 
 
 1.113 
 
 76.530 
 64,710 
 
 3.691 
 3.121 
 
 0.3017 
 0.3567 
 
 4.939 
 
 58.4 
 
 690 
 
 11 
 
 159.4 
 
 3.14 
 
 32 15 
 
 1.182 
 
 74,180 
 
 3.578 
 
 0.3260 
 
 3.983 
 
 12.5 
 
 39 
 
 12 
 
 164.0 
 
 
 
 34 20 
 
 1.211 
 
 
 
 0.3260 
 
 0.0 
 
 
 
 
 
 
 
 EZwycos a = 29,810 
 
 NOTE. The x abscissae are measured horizontally from A. 
 The y ordinates are measured vertically from AB. 
 The Jj are measured horizontally between sections. 
 a=l 06', cos a = 0.9998, sin a = 0.0192. 
 
 The products QAx in Table 49A are summed by the prism oidal formula treating 
 Jx as the length of a prismoid whose middle area is 6 and whose end areas are the means 
 of the successive tabulated values of 6. This is necessary because 6 is a variable quantity. 
 
AHT. 19 SPECIAL APPLICATIONS OF INFLUENCE LINES 163 
 
 Calling 0o-i the mean between and 0i, and 0]- 2 the mean between 6\ and 2 
 etc., then the values 6 Ax become: 
 
 Q j x =^[200 +0 _!] =^[0.6932 +0.3363] =0.652; 
 o o 
 
 0! Jz = ^[0o-i +40i +0i_ 2 ] =[0.3363 + 1.3040 +0.3413] =4.954. 
 
 6 2 Ax =^[0i -2 +20 2 +20 2 ' +0 2 - 3 ] =^?[0.3413 +0.7134 +0.6034 +0.3174] =4.939. 
 
 6 z Ax = ^[0 2 -3 +403+03-4] =y [0.3 174 + 1.3328 +0.3551] =5.014, 
 etc., etc. 
 
 After collecting the tabulated data from the drawing, Fig. 49c, the remaining 
 computations are quite simple and no further comment is necessary here, as the several 
 operations are indicated in Table 49A. 
 
 The X n influence line is now drawn by combining the Ew loads into a force polygon 
 and making the pole distance equal to E2wy cos a. The resulting equilibrium polygon 
 represents the X a influence line with ordinates to the scale of lengths and /* = !, in 
 accordance with Eq. (49n). 
 
 In the drawing, the pole distance was made 1/20 of the actual length so that the 
 ordinates ?? are twenty times too large. For this reason a special scale of ordinates, 
 twenty times the scale of lengths, was constructed and used for all influence ordinates. 
 
 The scale of forces for the force polygon may be any convenient scale, so long as the 
 pole distance is measured with the same scale as the forces. 
 
 The influence areas for the kernel moments and tangential force are now drawn by com- 
 bining the M /M a lines and the T /T a line each with the X a line. According to Eqs. 
 (49x), (49u), (49v), this will require computing the kernel distances k, and the coordi- 
 nates of the kernel points and finally the end ordinates at A of all influence lines. 
 The end ordinates at B may be obtained by substituting for the abscissa? x the values 
 I x. These computations are given in Table 49s, which is self-explanatory. 
 
 In Fig. 49c the influence areas M e and Mi are constructed for section 3. The X a 
 polygon is copied from the original one by transferring down the several j) a ordinates 
 from a horizontal base' ~A r W. The M /M a lines are constructed from the end ordinates 
 A'A"=x c /y e cosa and A f A 7r '=Xi/y i cosa, using the coordinates for the two kernel, 
 points as given in Table 49u. The lines WA" and B'A'" are thus determined. 
 
 The kernel points e and i are the centers of moments and have the abscissae x e and 
 x i} as given in the table, and from these the points e' and i' are located. 
 
 The two moment influence lines are completed by drawing the lines A'e' and A'i'. 
 
 The M oe /M ae influence line is thus found to be the broken line A'e'B' and the shaded area 
 
 included between it and the X a line is the M e influence area for section 3. The fac- 
 
 tor for the ordinates i) e , when measured to the same scale as the rj a ordinates, is 
 
 i = cos a =18.41. 
 
164 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. X 
 
 TABLE 49n 
 KERNEL POINTS AND END ORDINATES FOR Al me , M mi , AND T m AREAS 
 
 Sec- 
 tion. 
 
 F 
 
 Sq. in. 
 
 D 
 
 2 
 
 in. 
 
 i 2I 
 
 * 
 
 k sin < 
 ft. 
 
 k cos <j> 
 ft. 
 
 Coordinates of kernel points, 
 Eqs. (49u). 
 
 End Ordinatea. 
 
 \-2FD 
 
 Eq. (49R) 
 
 ft. 
 
 x e 
 ft, 
 
 Ve 
 ft. 
 
 x i 
 ft. 
 
 * 
 
 ft. 
 
 x e. 
 
 x i 
 
 COS <ji 
 
 y e cos a 
 
 y^ COS a 
 
 sin (<t> a) 
 
 
 1 
 2 
 
 3 
 
 4 
 5 
 
 6 
 
 7 
 
 8 
 9 
 10 
 
 11 
 12 
 
 81.6 
 79.0 
 89.5 
 87.8 
 86.3 
 85.7 
 75.2 
 73.6 
 75.2 
 85.7 
 86.3 
 87.8 
 89.5 
 79.0 
 81.6 
 
 36.0 
 35.5 
 33.4 
 33.8 
 31.6 
 29.5 
 27.4 
 27.0 
 24.9 
 27.0 
 27.4 
 29.5 
 31.6 
 33.8 
 33.4 
 35.5 
 35.7 
 
 2.13 
 2.04 
 2.11 
 1.99 
 1.86 
 1.71 
 1.64 
 1.53 
 1.64 
 1.71 
 1.86 
 1.99 
 2.11 
 2.04 
 2.13 
 
 36 15 
 32 15 
 
 26 00 
 20 00 
 14 15 
 
 7 30 
 00 
 
 7 30 
 14 15 
 20 00 
 
 26 00 
 32 15 
 34 20 
 
 1.14 
 
 0.91 
 
 0.68 
 0.46 
 
 0.23 
 0.00 
 
 0.23 
 0.46 
 0.68 
 
 0.91 
 1.14 
 
 1.80 
 
 1.86 
 1.87 
 1.79 
 
 1.66 
 1.53 
 
 1.66 
 1.79 
 1.87 
 
 1.86 
 1.80 
 
 10.54 
 
 25.31 
 
 40.08 
 54.86 
 
 69.63 
 84.40 
 
 99.17 
 113.94 
 128 . 72 
 
 143.49 
 158.26 
 
 4.21 
 
 12.26 
 18.43 
 22.94 
 
 25.58 
 26.36 
 
 25.11 
 21.79 
 16.71 
 
 9.96 
 1.34 
 
 8.26 
 
 23.49 
 38.72 
 53.94 
 
 69.17 
 
 84.40 
 
 99.63 
 114.86 
 130.08 
 
 145.31 
 160.54 
 
 7.81 
 
 15.98 
 22.17 
 26.52 
 
 28.90 
 29.42 
 
 28.43 
 25.37 
 20.45 
 
 13.68 
 4.94 
 
 2.504 
 
 2.064 
 2 . 174f 
 
 1.058 
 
 1.470 
 1.747t 
 
 1.633 
 
 2.135 
 
 2 . 895f 
 4.307 
 
 8.884 
 55.555 
 
 8.884* 
 4.307 
 2.895 
 
 2.135 
 1.633 
 
 
 
 3.202 
 2 . 582* 
 
 2 . 869 
 2.265* 
 
 
 
 
 
 4.284 
 
 0.700 
 
 NOTE. D is measured between extreme fibers. For all dimensions in inches, k is in inches. 
 The coordinates of the axial points are given in Table 49A. 
 
 * For sections 7 to 12 the ordinates are for the B end. 
 t Values used in Fig. 49c. 
 
 Similarly the dotted line A'i'B' represents the M oi /M ai influence line and the area 
 included between it and the X a line represents the Mi influence area with factor 
 Pi=yi cos a =22.13 for section 3. 
 
 Hence for any simultaneous position of moving loads the two kernel moments for 
 section 3 are represented by the expressions 
 
 and 
 
 (49w) 
 
 where the subscripts refer to the kernel points. 
 
 The T m influence area for section 3, is found by laying off the end ordinate A' A" = 
 cos </sin ((f> a) =2.895 and drawing the line A"B'. The negative end ray A' 3' is parallel 
 to A"B', and the T /T a line is thus the broken line A'3'3"B' and the shaded area is the 
 T 3 influence area with a factor /*=sin (< a). 
 
 It will be seen that for sections near the crown the end ordinate at A becomes very 
 large and when <=, this ordinate approaches infinity, while becomes zero. Hence 
 there will always be several sections near the crown for which the T m influence area 
 must be found by making n = l and reducing the X a line accordingly. This is done 
 by using the original form of Eq. (49s) which is 
 
 T m =Q cos < X a sin (<j> a). 
 
ART. 49 SPECIAL APPLICATIONS OF INFLUENCE LINES 165 
 
 For the A ordinate Q = l, hence this ordinate is simply cos < and by reducing all 
 the T) U ordinates by multiplying them by sin (<-), the .T 5 area is found with a factor 
 
 1=1. 
 
 The rj a ordinates are easily reduced by graphics, laying off the line B'z such that its 
 deviation from the vertical is sin (0 -a) in a distance unity. The rj a ordinates pro- 
 jected over horizontally are then reduced to the small horizontal intercepts between 
 Wz and the vertical. 
 
 The T 5 influence area is thus constructed and is represented by the shaded area 
 included between the broken line A'5'5"B' and the reduced X a line. The factor p = l. 
 
 The tangential force for section 3, for any train of moving loads is expressed by 
 
 (49x) 
 
 and the tangential stress by T Z /F 3 , where rj t is any ordinate to the T 3 area under some 
 particular load P. 
 
 The stress due to a unifo 
 (49 J) whence for t = 75 F., 
 
 particular load P. 
 
 The stress due to a uniform change in temperature is found from Eqs. (49i) and 
 
 l-.a*_ etf 0.000488 1 
 
 Xat 5 it ?, 
 
 d aa COS Oi d aa COS a 
 
 Eq (49o) gives d aa = Hwycosa and Table 49A furnishes EZwij cos < =29810, hence 
 L=29810/JS ft. for X a = l kip. Making =4,176,000 kips per sq.ft., Z = 164 ft. and cos 
 a a =0.999S, then 0^-0.00714 ft. and X at = ^mfxo.9998 = n - 21 ki P s - 
 
 * Eqs. (49K) , will furnish the means of finding the values M mt , N mt , T mt , and v=M mt /N mt 
 and with these^and Eqs. (49?) the temperature stresses f e and f t may be computed for 
 
 any section mm'. 
 
 Stress due to abutment displacements. Eq. (49L) furnishes the haunch thrust X ar 
 for any change Al in the length of span. The quantity Al must be estimated from the 
 elastic properties of the abutments and is always more or less problematic, though it 
 is well to investigate the probable stress which might be created by such a displacement. 
 
 In the present example it is assumed that for the maximum case of loading the 
 haunches will spread an amount JZ=0.03 ft, then for d aa =0.0071< ft, X a - kip, 
 and cos a =0.9998, Eq. (49L) gives 
 
 y & - 03 = -4.20 kips. 
 
 ar = ~<LTc^ 0.00714X0.9998 
 
 The stresses /, and /,- may then be found in precisely the same manner as was 
 just described for the case of temperature stresses. 
 
 The stresses due to temperature and abutment displacements should be separately 
 investigated for several typical sections, especially the crown section, so as to 
 thldesigner to judge of the relative importance which these may have in comparison 
 with the combined dead and live load stresses. 
 
 The live load stresses for section 3 will now be found for a single jloac P acting at 3, 
 tion 4, merely to illustrate the use of the influence areas shown in Pig. 4 
 
166 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP X 
 
 M e3 =Me4P = 18.41 X0.45P = 8.285P. 
 M =i*P =22.13 X0.17P =3.762P. 
 
 = -6.84P per sq.ft. = -0.0474P per sq.in. 
 
 M,- 3.762P 
 
 = LOOP; T 3 =p t i)t*P =0.324 X0.94P =0.305P. 
 
 The resultant K A , of ^1 and JT a , is found graphically to be 1.19 P, which, for a single 
 load to the right of section 3, should check R 3 , as it does. 
 
 ART. 50. TWO-HINGED SPANDREL BRACED ARCH OR FRAMED ARCH 
 
 The analysis of the two-hinged frame arch is accomplished in the same general manner 
 followed in the previous article, where the corresponding solid web arch was treated, 
 only that the frame is usually simpler. 
 
 The general Eqs. (49s) for the reactions A, B, and- moment M m apply equally to an 
 unsymmetric framed arch, and the influence line for the redundant haunch thrust X a 
 is determined precisely as in the previous problem except that the w loads are found by 
 the method given in Art. 36. 
 
 The stress in any member then becomes 
 
 S=S S a X a =S a \-^-X a \, ........ (50A> 
 
 according to Eq. (45E) for one external redundant condition. This equation is later 
 employed to construct the stress influence lines for the members. 
 
 The X u influence line for the haunch thrust is represented by Eq. (49n) as 
 
 Z* ' Oma _ ! ' O-ma f ~p. \ 
 
 a = K """v* 'Ja? ........ (OUB) 
 
 daa 2jWy COS a 
 
 where d^a is the vertical deflection ordinate of any point m of a deflection polygon, drawn 
 for the loaded chord and for the conventional loading X a = l applied to the principal 
 system; and d aa = 'wy cos a is the change in the span AB, due to X a = l acting on the 
 principal system. 
 
 According to Art. 40, the d ma deflection polygon is the equilibrium polygon drawn for a 
 set of elastic loads w, with a pole distance H = l, according to the method given in Art. 36c. 
 
ART. 50 SPECIAL APPLICATIONS OF INFLUENCE LINES 167 
 
 The elastic loads w are functions of the changes in the lengths of all the members 
 of the frame as given by Eqs. (36fi) and (36c). These are algebraically summed for all 
 the panel points to obtain the total loads w. Thus the elastic loads for the chords alone 
 are w c = Al/r and each web member contributes two elastic loads w u = M/r u and w n = M/r n , 
 acting at the two adjacent panel points u and n of the loaded chord. Hence when the 
 loads P are to be applied to the top chord, then the w u and w n elastic loads are com- 
 puted for the top chord panel points. The lever arms r for the chords are measured 
 as shown in Fig. 36A and the lever arms r u and r n for the web members are measured as 
 explained in Figs. 36s and 36r. The details of the computation of the w loads are illus- 
 trated in Table oOu, and Fig. 50A, in connection with a complete example. 
 
 For any braced arch with parallel chords, the w loads for the web members may 
 always be neglected. 
 
 The displacement d aa = 'Swy cos a is computed for the same w loads by taking the 
 sum of their moments about the line A B joining the haunches. The lever arms for an 
 unsymmetric span thus become the vertical ordinates of the respective pin points 
 times cos a. See also Table 50A. 
 
 The X a influence line thus becomes the equilibrium polygon drawn for the elastic 
 loads w with a pole distance o aa = ^wy cos a. 
 
 Still another method of finding this influence line consists in drawing the deflection 
 polygon for the loaded chord by means of a Williot-Mohr displacement diagram, which 
 also furnishes the value d^ from which the influence line ordinates rj a may be computed 
 and plotted. This solution is illustrated in Fig. 50B. 
 
 Since the modulus E does not affect the X a influence line it is more convenient to 
 comput-e all displacements and the w loads E times too large, thus avoiding the small 
 quantities resulting in many decimals. 
 
 Temperature Stress- Calling Al t the change in the length of any member due to any 
 temperature effect, then from Mohr's work Eq. (5n) the change d at in the length of 
 the span AB becomes 
 
 l.d at =ZS a 4l t =2S a dl .......... (50c) 
 
 The redundant thrust X at , due to any temperature effect, is found from Eq. (44c) as 
 
 . at _ a _ g 
 
 at ~~d^~ daa 'EZwycosa 
 
 For a uniform change in temperature of t, above or below a certain normal, the 
 change o^ in the length of the span 'AB, is found from Eq. (49j) as 
 
 dAB 
 
 cos a 
 
 In any case the stresses in the members are best found from S t = ^SaXat for X at , or 
 from a Maxwell diagram drawn for the external loading X at , acting on the principal system. 
 
 For a uniform change of t from the normal temperature, the resulting stresses 
 become S t = ^S a X at . The effect on the final stresses S for full loading, will then 
 be additive or S max = (S +S t .) 
 
168 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. X 
 
 Abutment displacements will affect the stresses in the members by producing a redundant 
 thrust X ar as found from Eq. (49L) when JZ is an increase in the length of span AB. 
 The stresses S r would be found from a Maxwell diagram drawn for X ar acting on the 
 principal system. They could also be found as S r = =F*S a X ar , for X ar , since *S a is the stress 
 produced in any member S by X a = 1 acting outward. 
 
 The stress influence area for any member $ is derived from the previous Eq. (50 A) 
 from which any stress influence ordinate TJ becomes 
 
 s=s a ^-x a ^=y). .:.; . . (SOP) 
 
 These influence areas are alike in principle for all members, whether chords or web 
 members, and represent the area inclosed between the X a influence line and the S /S a 
 influence line, times a factor S a . 
 
 The lSo/S a influence line is the ordinary stress influence line S for any determinate 
 frame, nultiplied by the factor l/S a . Hence the end ordinates for the S line may be 
 found in the usual way. Thus the stress SA in the member S due to the upward reaction 
 .4=1 is the end ordinate of the S line at the point A, and this ordinate divided by 
 S a becomes the required end ordinate rj A =S A /S a of the S /S a line at A. Similarly the 
 end ordinate of the S /S a line at B is -fjB=SB/S a , where SB is the stress in the member 
 due to the upward reaction B = \ acting on the principal system. 
 
 Hence if the stresses SA, SB and S a , for all the members of the principal system, are 
 found either by computation or from three Maxwell diagrams, then all the data for the 
 several stress influence lines are at hand provided the X a influence line is known. 
 
 In drawing the S /S a influence lines the following points should be observed: 1, 
 That the end bounding lines must always intersect in a point i' on the vertical through 
 the center of moments i of the particular member treated. 2, that this S /S a line 
 must be a straight line between adjacent panel points of the loaded chord. 3, that when 
 one of the end ordinates is too large to be conveniently measured, then half this ordinate 
 may be laid off at the center of the span. This is frequent!}' done as in the example 
 which follows, see Figs. oOc. 
 
 The signs of the end ordinates and of the factors fJi=S a all follow from the signs of 
 the stresses SA, SB and S a . 
 
 Deflection of any point m. Applying Mohr's work equation (Gfi) the deflection 
 d, n of any point m, becomes 
 
 (600) 
 
 SI 
 
 wherein Si is the stress in any member due to a load P = l, applied at the point m in 
 the direction of the desired deflection ; and S is the corresponding stress due to any cause 
 or actual condition of simultaneous loading for which the deflection is sought, includ- 
 ing load effects as well as temperature changes and abutment displacements. The sum- 
 mation covers all the members of the principal system. 
 
ART. 50 
 
 SPECIAL APPLICATIONS OF INFLUENCE LINES 
 
 169 
 
 For any redundant condition X a the stress S by Eq. (7 A) becomes 
 
 S=S -S a X a +S t , . (50H) 
 
 where S t = S a X at is the stress due to temperature effect. 
 
 Example. A two-hinged, riveted, spandrel-braced arch, taken from a thesis by 
 Mr. A. V. Saph, 1901, Cornell University, is used to illustrate the above method. 
 
 The arch has a span of 168.75 ft. between pin supports, a rise of 29.5 ft., and weighs 
 1,060 ,320 Ibs. without abutment shoes, making a uniform dead load of 43.2 kips per truss 
 per panel. The top chord is the loaded chord. The abutments are symmetric and, 
 therefore, a =0. 
 
 Fig. 50A shows the half span with the lengths of members in feet below the lines, 
 and the values EM, in feet, as found from Table 50A, above the lines. The various lever 
 arms used in the computation of the stresses S A , S B and S a , and of the w loads, are also 
 shown. The actual values Al in feet =EM + 29000, because the areas F were not reduced 
 to square feet. See also the example in Art. 52, where M is actual. 
 
 TABLE 50A 
 COMPUTATION OF EAl, j) A , TJ B AND E2S a 4l t . 
 
 
 Stresses in Kips. 
 
 
 
 Unit 
 
 
 End Ordinates, 
 
 Temperature Effect. 
 
 
 
 
 
 Area, 
 
 ..enscth 
 
 Stress, 
 
 S a l 
 
 Oo 
 
 
 *ES a 4lt 
 
 Member 
 
 S A 
 
 SB 
 
 S a 
 
 F 
 
 i 
 
 Oa 
 
 F 
 
 
 *EM t 
 
 
 
 
 
 
 
 
 for 
 
 for 
 
 for 
 
 
 
 * 
 
 
 SA 
 
 SB 
 
 -stlh 
 
 
 
 
 A = l 
 
 5=1 
 
 a 
 
 Sq.in. 
 
 ft. 
 
 Kips. sq. in. 
 
 ft. 
 
 / Sa 
 
 T>B Sa 
 
 ft. 
 
 Kip.ft. 
 
 Kip. ft. 
 
 uu 
 
 - 0.320 
 
 - 5.437 
 
 -0.211 
 
 19.80 
 
 15.000 
 
 -0.0107 
 
 -0.1605 
 
 1.517 
 
 25.770 
 
 197.9 
 
 
 41.8 
 
 u c/ 
 
 - 1.165 
 
 - 6.902 
 
 -0.697 
 
 19.80 
 
 15.000 
 
 -0.0352 
 
 -0.5280 
 
 1.671 
 
 9.917 
 
 197.9 
 
 
 137.9 
 
 u.u, 
 
 - 2 725 
 
 - 8.953 
 
 -1.457 
 
 19.80 
 
 15.000 
 
 -0.0736 
 
 - 1 . 1040 
 
 1.870 
 
 6.145 
 
 197.9 
 
 
 288.3 
 
 
 
 - 5 588 
 
 -11.755 
 
 -2.649 
 
 26.48 
 
 15.000 
 
 -0.1000 
 
 - 1 . 5000 
 
 2.109 
 
 4.438 
 
 197.9 
 
 
 524.2 
 
 ifu* 
 
 -10.008 
 
 -14.336 
 
 -4.121 
 
 38.11 
 
 15.000 
 
 -0.1081 
 
 -1.6215 
 
 2.429 
 
 3.479 
 
 97 9 
 
 
 815.5 
 
 
 
 u*u* 
 
 -14.062 
 
 -14.062 
 
 -4.917 
 
 38.11 
 
 15.000 
 
 -0.1290 
 
 -1.9350 
 
 2.859 
 
 2.859 
 
 197.9 
 
 
 973.1 
 
 
 T V " 
 
 01 79 
 
 
 + 1 085 
 
 42 50 
 
 11 228 
 
 + 0.0255 
 
 0.2863 
 
 -0.159 
 
 
 88.9 
 
 96.5 
 
 
 LI 
 
 366 
 
 6.231 
 
 1.388 
 
 40.00 
 
 17.190 
 
 0.0347 
 
 0.5965 
 
 0.263 
 
 4.481 
 
 136.1 
 
 188.9 
 
 
 LI* 
 
 1 269 
 
 7.516 
 
 1.848 
 
 37.50 
 
 16.335 
 
 0.0493 
 
 0.8053 
 
 0.687 
 
 4.067 
 
 129.4 
 
 239.1 
 
 
 L L 
 
 2 857 
 
 9 . 386 
 
 2.575 
 
 37 . 50 
 
 15.725 
 
 0.0687 
 
 1.0803 
 
 1.109 
 
 3.645 
 
 124.5 
 
 320.6 
 
 
 
 5 685 
 
 11.958 
 
 3.712 
 
 37.50 
 
 15.258 
 
 0.0990 
 
 1.5105 
 
 1.532 
 
 3.221 
 
 120.8 
 
 448.4 
 
 
 J* L 
 
 10 027 
 
 14 . 363 
 
 5.131 
 
 42.50 
 
 15.029 
 
 0.1207 
 
 1.8140 
 
 1.954 
 
 2.799 
 
 119.0 
 
 610.6 
 
 
 T? L 
 
 917 
 
 
 -0.605 
 
 23 . 52 
 
 35.942 
 
 -0.0257 
 
 -0.9237 
 
 1.516 
 
 
 379 . 5 
 
 
 229.6 
 
 UT 
 
 1 179 
 
 - 2.043 
 
 -0.678 
 
 19.80 
 
 29.312 
 
 -0.0342 
 
 -1.0025 
 
 1.739 
 
 3.014 
 
 309.5 
 
 
 209.8 
 
 r/ 1 /" 1 
 
 1 502 
 
 - 1.976 
 
 -0.732 
 
 14.70 
 
 20.918 
 
 -0.0498 
 
 -1.0417 
 
 2.052 
 
 2.700 
 
 220.9 
 
 
 161.7 
 
 
 U>L 3 
 
 - 1.857 
 2 042 
 
 - 1.817 
 - 1 193 
 
 -0.773 
 -0.681 
 
 11.76 
 11.76 
 
 14.450 -0.0657 
 9.730-0.0579 
 
 -0.9494 
 -0.5634 
 
 2.402 
 2.999 
 
 2.351 
 1.752 
 
 152.6 
 102.7 
 
 
 118.0 
 70.0 
 
 
 
 U L 
 
 - 1.622 
 
 + 0.109 
 
 -0.318 
 
 11.76 
 
 6.932 
 
 -0.0270 
 
 -0.1872 
 
 5.101 
 
 -0.343 
 
 73.2 
 
 
 23.3 
 
 
 77 V 5 
 
 
 
 
 
 0.0 
 
 11.76 
 
 6.000 
 
 0.0 
 
 0.0 
 
 0.0 
 
 0.0 
 
 63.4 
 
 
 0.0 
 
 
 rr r 
 
 
 
 If) 71 
 
 13 34 
 
 32 928 
 
 . 0503 
 
 1.6563 
 
 1.515 
 
 
 347.7 
 
 233.3 
 
 
 T7T 
 
 .UI/ 
 1 451 
 
 2 514 
 
 0.834 
 
 13.34 
 
 25.741 
 
 0.0625 
 
 1.6088 
 
 1.740 
 
 3.015 
 
 271.8 
 
 226.7 
 
 
 1 
 
 2.166 
 3.413 
 4.869 
 4.367 
 
 2.848 
 3.339 
 2.843 
 - 0.294 
 
 1.055 
 1.421 
 1.622 
 0.857 
 
 13.34 
 13.34 
 18.25 
 15.84 
 
 20.828 
 17.871 
 16 . 524 
 16.156 
 
 0.0791 
 0.1065 
 0.0889 
 0.0541 
 
 1.6475 
 1.9033 
 1.4690 
 0.8640 
 
 2.053 
 2.402 
 3.002 
 5.096 
 
 2.700 
 2.350 
 1.753 
 -0.343 
 
 220.0 
 188.7 
 174.4 
 170.6 
 Tot'k 
 
 232.1 
 268.1 
 282.9 
 146.2 
 3293.4 
 
 3593 . 2 
 
 
 * Where =29,000 kips instead of 29,000X144. lE^S a M t - 299.8 kip. ft. 
 
170 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. X 
 
 U ' 
 
 The f i^ure] obove-the members| represent EA] 
 other dijnen&ions are lengths in ft. 
 
 DEF-LEcflONS ARE \E TIlllES ACTUAL IN F*ET. 
 
 E= 29OOto KIPS PER SQ.IN. S 
 \ .XI 
 
 " " : r " 
 
AKT. 50 SPECIAL APPLICATIONS OF INFLUENCE LINES 171 
 
 Table 50A gives these stresses, in kips, as found by computation, also the areas F 
 of the members, being gross for compression and net for tension members. The lengths 
 I of the members, unit stresses/ and quantities EAl (for shortening) are also included 
 in this table, and finally the end ordinates of the S /S a influence lines are obtained. 
 
 The X a influence line is now found by two different methods to illustrate the applica- 
 tions frequently referred to elsewhere. 
 
 The first method is by constructing a deflection polygon of the top chord by means 
 of a Williot-Mohr displacement diagram, Fig. 50B, using the quantities EM, in feet, as 
 the changes in the lengths of the members for the conventional loading X a = l, producing 
 stresses S a . The second method is by finding this deflection polygon from the w loads. 
 
 The first method requires little description other than to say that the displacement 
 diagram Fig. 50B is drawn for displacements EAl, Table 50A, on the assumption that the 
 point LQ and the direction of the member L 6 U 6 remain fixed and since the span is symme- 
 tric about this member and the point L 6 , therefore no rotation diagram is necessary. 
 The deflection polygon of the top chord is then found by projecting the vertical deflec- 
 tions of the top chord panel points onto the verticals through these points, furnishing 
 the polygon ~A rT lT y with a closing line A'A", horizontally through A' '. These deflec- 
 tions are E times actual, in feet, measured to the scale of displacements. 
 
 The horizontal displacement \Ed aa between A and L 6 is also obtained from the same 
 diagram as the horizontal distance between A' and L' 6 . 
 
 The vertical ordinates of the deflection polygon are values of dma, and hence the 
 X a influence line ordinates y a are found by dividing the several deflection ordinates 
 Eo ma by the constant Ed m according to Eq. (50s), giving 
 
 This shows that the factor E and the scale of the ordinates do not affect X a . The 
 ordinates r] f! , for all panel points, are plotted to any convenient scale to obtain the X a 
 influence line. 
 
 Second method. The same displacements EM, in feet, are here employed to com- 
 pute the elastic loads Ew, using the lever arms, also in feet, as given on Fig. 50A. See 
 Table 50s. 
 
 In the present example all the members are included and the table indicates exactly 
 how much each member contributes to the several total Ew loads. The method of 
 Art. 35 is rigidly followed. By using displacements EM which are E times too large, 
 the w loads are also multiplied by E. 
 
 Each chord member furnishes one w load which acts at the center of moments 
 for that chord while each web member contributes two loads w u and w n acting at the two 
 adjacent panel points u and n of the loaded chord or the chord for which the deflection 
 polygon is to be drawn. 
 
 " The w c loads resulting from the chord members are always positive as found by 
 
 Eq. (36s), thus: 
 
 M EM ro , 
 
 uv = or Ew c = ......... (DUK; 
 
172 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP, x 
 
 The negative and positive loads for a web member are 
 
 Al Al EM EAl 
 
 w u =- and w n = or Ew u = and Ew n = -- . . (50L) 
 
 TU r n TU T n 
 
 as given by Eq. (36o). 
 
 The lever arm r for a chord, is the distance from the center of the moments for 
 such chord and perpendicular to the chord. The lever arms r u and r n are perpendicular 
 distances onto a web member as described in Figs. 36s, and 36F. They may be identified 
 by comparing the values in Table 50s with the dimensioned lever arms on Fig. 50A. 
 
 The rule for the signs of the two loads for any web member was stated in Art. 36, 
 and is as follows: Calling all top chord members negative and all bottom chord members 
 positive, and giving the proper sign to Al for the web member in question, then the positive 
 w load is found on that side of the panel where the sign of Al coincides mth the sign of the 
 adjacent chord. 
 
 For the panel point f/ the total load Ew (Table O()B) is made up of the positive 
 load produced by the chord L Li and one of the loads from each of the web members 
 U Li and U\L\, the signs of which are negative according to the above rule. 
 
 The w load for point A produces zero effect on the deflection and need not be 
 considered. 
 
 For the panel point U l the total load Ew\ becomes 
 
 ALL JUpLi AVjLz . JE/nZq J7 2 L 2 
 
 " ""' ~~ =auby 
 
 25.7 13.4 17.15 "' 11.7 
 
 and similarly for the other panel points of the loaded (top) chord. 
 
 The bottom chord points receive only the loads produced by the top chord members 
 except at LI where the member U L , being the end post, contributes a load Jt/ L /10'.2. 
 
 The total Ew loads acting in the same vertical are then summed for the seven panel 
 points of the half span and used in constructing the deflection polygon with a pole dis- 
 tance H =E2yw. 
 
 The y ordinates are measured vertically from the line A~B to the points of application 
 of the respective Ew loads, distinguishing between the loads acting at the top and 
 bottom panel points. 
 
 In order that the X a influence line ordinates may appear to a scale twenty times 
 as large as the scale of lengths, the pole distance was made equal to E2yw/20, see Fig. 
 50c. Note the agreement of the ordinates here found with those obtained in Fig. 50fi. 
 
 Stress influence areas. The end ordinates i) A and r) B of the S^/Sa lines are com- 
 puted in Table 50A and these serve to construct all stress influence areas for the several 
 members. 
 
 Fig. 50c shows stress influence areas for six typical members, and each one has a 
 factor S a as per Eq. (50A). The end ordinates are applied down from the closing line 
 when positive and up when negative. The signs of the influence areas are uniformly + 
 for areas below the .Y a influence line and the factor p takes the sign of S a . 
 
SPECIAL APPLICATIONS OF INFLUENCE LINES 
 
 173 
 
 
 
 P^ 
 
 o 
 
 CO 
 
 co 
 
 co 
 
 00 
 
 f^. 
 
 ,_! 
 
 a; ' 
 
 ~ "3 *** s 
 
 & II - -i 
 
 <*- 
 
 
 
 *? 
 
 
 1C 
 
 cb 
 
 T 1 
 
 
 o 
 
 <N 
 
 co 
 
 o 
 
 CD 
 
 rf 
 
 3q 
 
 S 
 
 "8 1 1 
 
 "o"2 c 
 
 V 
 
 
 
 TZ "t^ 
 
 
 O 
 
 d 
 
 d 
 
 d 
 
 d 
 
 d 
 
 d 
 
 i-H 
 
 3 s2 
 
 -t-> i*"! 
 
 "S-2" 
 
 ^ 'M O 
 
 1 *x ei 
 
 *& 
 
 
 
 O O 
 
 S 
 
 | 
 
 _j_ 
 
 
 
 
 
 ii 
 
 II 
 
 2 II 
 
 UJ C ^^ 
 
 5 CD 03 c 
 
 In 
 
 
 
 sC 
 
 *3 
 
 
 ii 
 
 II 
 
 11 
 
 || 
 
 || 
 
 ii 
 
 to 
 
 S 
 
 S~i 
 
 .5 aj 
 
 S > 
 
 Q) 
 
 
 
 tit; 
 
 
 1 
 
 ii 
 
 ii 
 
 
 
 
 S 
 
 N 
 
 OJ Sl*^ 
 
 " -i 
 
 ai "Sto 
 
 >s 
 
 
 
 * * 
 w 
 
 Tl ,- 
 
 
 g 
 
 sT 
 
 fiq 
 
 
 
 I 
 
 
 
 1 
 
 S, 
 
 BJJ 
 
 SjJl 
 
 o3 ^ 
 
 = 2^^ 
 
 o^ 
 
 
 
 .2 ~ 
 
 o Sji 
 
 
 
 4 
 
 J- 
 
 K-f 
 
 4 
 
 4 
 
 
 
 2ls 
 
 sl o 
 
 ItB^l 
 
 g g 
 
 
 
 2 
 
 ; 
 
 b" 
 
 b 
 
 ^ 
 
 b 
 
 b 
 
 b 
 
 b 
 
 
 ^ 
 
 CC C H-. 
 
 g5s c 
 
 ^ - 
 
 
 
 * 
 
 
 
 
 
 
 
 
 r+K 
 
 
 
 
 
 
 
 
 1 
 
 
 T-l 
 
 CD 
 CO 
 
 O 
 
 CO 
 
 1 
 
 C5 
 
 M 
 
 CD 
 rf 
 
 S 
 
 i-C 
 
 us 
 
 CO 
 
 CO 
 
 to 
 
 q 
 
 oo 
 
 CD 
 
 Ss 
 
 
 R3 
 
 
 ic 
 
 IN 
 
 CO 
 
 rf 
 
 06 
 
 co 
 
 T * 
 
 1C 
 
 d 
 
 d 
 
 ^ 
 
 co 
 
 CO 
 
 O5 
 
 w 
 
 * 
 
 
 1 
 
 + 
 
 
 
 
 
 
 
 
 
 
 
 
 t-_ 
 
 a 
 
 
 i-C 
 
 co 
 
 us 
 
 1C 
 
 CO 
 
 c 
 
 co 
 
 CO 
 
 CO 
 
 1C 
 
 co 
 
 1C 
 i-C 
 
 CO 
 
 CD 
 
 oo 
 
 iC 
 
 rf 
 i I 
 
 q 
 
 (M 
 
 (N 
 
 1C 
 
 00 
 IN 
 
 
 1C 
 
 O5 
 (N 
 
 1 
 
 -3- 
 
 n 
 
 i-H 
 
 a 
 
 CD 
 O 
 
 i i 
 
 o 
 o 
 
 T I 
 
 oo 
 
 CO 
 <N 
 
 co 
 
 s 
 
 T-H 
 
 CO 
 
 1C 
 (N 
 
 O 
 
 
 
 3 
 
 rf 
 TO 
 IN 
 
 co 
 
 CO 
 
 3 
 
 H* 5 
 
 4 
 
 d 
 
 d 
 
 d 
 
 d 
 
 d 
 
 d 
 
 d 
 
 d 
 
 d 
 
 d 
 
 d 
 
 d 
 
 d 
 
 
 
 
 oo 
 
 CO 
 
 rf 
 
 CO 
 
 co 
 
 CO 
 
 (N 
 
 
 
 
 
 
 
 
 
 
 
 
 CD 
 O 
 
 
 
 q 
 
 CO 
 
 q 
 
 q 
 
 
 
 
 
 
 
 
 
 
 
 
 - d 
 
 ^ d 
 
 4 1 
 
 4" 
 
 '-i 
 
 4* . 
 
 
 
 
 
 
 
 
 
 
 
 b" ii 
 
 b || 
 
 ^ B 
 
 ^ " 
 
 b || 
 
 ^ 
 
 
 
 
 
 
 
 
 
 
 
 d ic 
 
 *" ,c 
 
 
 CD ic 
 
 00 iC 
 
 |C 
 
 
 
 
 
 
 
 
 
 
 ri 
 
 H 
 
 - 1 
 
 d 
 
 d 
 
 d 
 
 
 
 
 
 
 
 
 
 
 
 V 
 
 
 f 
 
 oo 
 
 O5 
 
 o 
 
 (N 
 
 00 
 
 rf 
 
 
 1C 
 
 o 
 
 
 
 
 
 
 
 
 
 
 00 
 
 00 
 
 ffi 
 
 ^^ 
 
 1-4 
 
 
 ^* 
 
 
 
 
 
 
 -. 
 
 
 
 
 43 
 
 4* 
 
 o 
 
 eo d 
 
 S, d 
 
 ^i d 
 
 "i . 
 
 "C d 
 
 
 
 
 
 
 X 
 
 
 
 
 b ii 
 
 b II 
 
 b II 
 
 b II 
 
 ^ " 
 
 ^ ii 
 
 b II 
 
 
 
 
 
 
 5 
 
 
 
 
 ^ co >. 
 
 "^ C^ cO 
 
 2 ^ 
 
 eo ^** 
 
 00 
 
 
 C^ ^ 
 
 
 
 
 
 
 . 
 
 
 
 
 q T-H 
 
 O i-i 
 
 O i 
 
 1C I 
 
 I-H CO 
 
 
 O5 O 
 
 
 
 
 
 
 ?i 
 
 
 
 
 
 T-* 
 
 d ^ 
 
 d ^ 
 
 d ^ 
 
 d 
 
 d ^ 
 
 
 
 
 
 
 g 
 
 
 
 iC 
 C<l 
 
 oo 
 co 
 
 co 
 
 o 
 
 CO 
 1C 
 
 oo 
 
 IN 
 
 CO 
 
 
 
 
 
 
 
 
 I 
 
 ^T 
 
 OQ 
 
 
 o 
 
 q 
 
 T-H 
 
 i i 
 
 1 1 
 
 1 1 
 
 
 
 
 
 
 
 
 _C 
 
 Q 
 
 4 
 
 
 d 
 
 d 
 
 n d 
 
 * d 
 
 a O 
 
 d 
 
 
 
 
 
 
 
 
 m 
 & 
 
 O 
 
 
 tj 
 
 *H 
 
 *~7 
 
 e> 
 
 '^i. 
 
 ** 
 
 
 
 
 
 
 
 
 J2 
 
 " 
 
 
 J^ II 
 
 ^ " 
 
 > II 
 
 b |l 
 
 b || 
 
 ^ " 
 
 
 
 
 
 
 
 
 
 1 
 
 
 
 O r 
 
 ^> OOICD 
 rf 
 
 ^o " 
 
 ""3 Oi 
 
 CO O- 
 
 co "" 
 
 
 
 
 
 
 
 
 . 
 
 
 d 
 
 CO I> 
 
 
 CO jX 
 
 Ci C- 
 
 ^ oc 
 
 <*>. CC 
 
 
 
 
 
 
 
 
 c-i 
 
 
 "s 
 
 ] 1 1 
 
 I-H T 1 
 
 14 
 
 ^ *~ 
 
 
 o 
 
 
 
 
 
 
 
 
 | 
 
 
 3 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 BB 
 
 
 i 
 
 a 
 
 
 CO 
 CO 
 
 co 
 co 
 
 
 
 o 
 
 CO 
 
 CO 
 O5 
 (N 
 
 CO 
 
 >o 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 '" (N 
 
 
 
 
 
 
 
 
 
 
 
 
 "i <=> 
 
 S. d 
 ^ ii 
 
 S, d 
 r^\ ii 
 
 S, d 
 ^ 11 
 
 b (| 
 
 *~i d 
 
 b 11 
 
 
 
 
 
 
 
 * 
 
 
 
 
 ""p 
 
 KJ !l 
 
 KJ ii 
 
 ""rf V 
 
 j ^8 ^ 
 
 | CD rl 
 
 H CO CC 
 
 
 
 
 
 
 
 
 
 
 
 CD ff 
 
 = c3 
 
 CO C 
 
 
 3 2 " 
 
 i 3"= 
 
 
 
 
 
 
 
 
 
 
 oo 
 
 co 
 
 8 
 
 C3 
 
 i 
 
 co 
 
 (N 
 
 CD 
 
 
 *4 ^5 
 
 iC 
 
 rf 
 CO 
 
 (N 
 
 - 2 
 
 ^ d 
 
 
 rf 
 
 ^ d 
 
 1C 
 (N 
 
 co 
 b d 
 
 u? II 
 
 
 
 "2 
 
 4* d 
 
 >-3 d 
 
 *3 d 
 
 *s. ^ 
 
 S, 
 
 ^(* II 
 
 
 ^J il 
 
 b* II 
 
 b" II 
 
 b ii 
 
 b il 
 
 b 
 
 
 
 o 
 
 .a 
 
 ^ " 
 
 <N ? 
 
 4 ii 
 
 Pis 
 
 4" ii 
 
 : ^ o ? 
 ^ 2 e 
 
 J 00 C 
 
 3 2 ' 
 
 *-3 11 
 ?^S a 
 
 2 S c 
 
 - I' 
 J5 C, 
 
 B 
 
 ^1 t- 
 
 !> ^ 
 
 d? 
 
 4 *"1 00 IN 
 5 (N O- 
 i ^ C 
 
 1 O C^ 
 
 "^ rf u: 
 
 O rf 
 
 '"J rf 
 
 1 t T 
 
 
 N S 
 
 CO CT 
 1-H tf 
 
 S c 
 
 i-! tf 
 
 5 
 
 1 
 
 5 
 
 
 
 d ^ 
 
 3 d c 
 
 O i- 
 
 
 
 
 
 
 
 
 
 
 
 
 
 t 
 
 b 
 
 b 
 
 b" 
 
 b" 
 
 b 
 
 ^ 
 
 b"* 
 
 1 4* 
 
 4 
 
 4" 
 
 4 
 
 4" 
 
 4- 
 
 i 
 
174 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. X 
 
 FIG. 50c. 
 
ART. 50 SPECIAL APPLICATIONS OF INFLUENCE LINES 175 
 
 The method of computing stresses for any train of loads need not be repeated here 
 except to call attention to the load divides, which must be carefully observed for each 
 member in placing the loads on the span. 
 
 Temperature effects. For a general case of unequal temperatures in the several mem- 
 bers the following assumptions are made: 
 
 Let the normal . temperature be 65 F., for which the structure has no temperature 
 stresses. Then assume a case where the top chord is heated to 130 F.; the bottom 
 chord to 104 F., and the web system to 117 F. For s=0.000007, and #=29000 
 kips per sq.in. the values of Ed become 
 
 Top chord Z = 130-65 =65, Eet = 13.20; 
 WebSystem t = 117 -65 -52, #rf = 10.56; 
 Bottom chord = 104 -65 -39, Eet= 7.92. 
 
 The values E Al t = dlE and ES a Al t are computed in Table 50A, using I in feet and 
 E in kip feet. Finally the half sum E2S a 4l t is found to be 3293.4 -3593.2 = -299.8 
 kip feet. Table 50fi gives ^E^lyw=54.7Q and from Eq. (50o) for cosa = l, 
 
 - *. *. * . i * 
 
 ' == ~ = ~ 5 ' 7 PS ' 
 
 which is a thrust acting outward the same as the conventional loading X a = 
 Hence the temperature stress in any member becomes 
 
 wherein S t has the same sign as S a . 
 
 A more severe stress would be produced when the top chord is colder than the 
 bottom chord, or when the top chord has a temperature of +10 F. at the same time 
 that the bottom chord has a temperature of -16 F, a case which might occur on a clear, 
 cold day. The stresses S t produced by such a condition would have the opposite sign 
 of S a when found from +X at . 
 
 For a uniform rise of 65 above the normal, the elongation for the whole span 
 becomes, by Eq. (50E), 
 
 _.! A D 771 
 
 = 2226.6 ft., 
 
 cos a 
 and from Eq. (SOo), 
 
 Edat 2226.6 
 
 giving stresses S t = S a X ai . 
 
 For a uniform fall of 65 below the normal ^ == -2226.6 ft. and X at = -20.33 kips, 
 
 giving stresses S t =S a X at , 
 
 Compare these results with those of the example in Art. 59, for a solid web arch o 
 
 about the same dimensions. 
 
176 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. X 
 
 Abutment displacements. Assuming that as a result of yielding abutments the length 
 of span is increased an amount J/=0.03 ft. then from Eq. (49L) 
 
 0.03 X 29 ,000 _ _ . e 
 ~ 109.53 '' lpS ' 
 
 Comparing this result with the one obtained for the solid web arch in Art. 49, it 
 is clear that the framed arch is almost twice as stiff. 
 
 Deflection of the crown due to the temperature effect producing X at = 5.48 kips, 
 Table 50A, and Eq. (50M). In this case the abutments are assumed rigid making Jr=0 ? 
 and the stresses S are not included, since temperature effects alone are desired. This 
 reduces Eq. (50c) to 
 
 i /St i , 
 
 where S t = -S a X at and = - = -EAlX at ; also S l = for a load unit Y at the 
 
 r r Zi 
 
 center of a symmetric span, and dlE =EAlt. Hence 
 
 which can easily be computed for any X at using the values S A) EM and EAl t given 
 in Table 50A. 
 
 ART. 51. TWO-HINGED ARCH WITH CANTILEVER SIDE SPANS 
 
 Occasionally a structure of this type is peculiarly adapted to certain sites as the 
 one at High Bridge, Ky. Its application has, however, received adverse criticism because 
 the analysis of stresses was considered too complicated. 
 
 This objection might apply to any of the ordinary methods of analysis, but not when 
 the problem is solved by influence areas. 
 
 As a type of bridge it is commendable and deserves careful consideration whenever 
 a particular site offers a suitable profile and good foundations for the center span. 
 
 Fig. 5lA represents one of several forms which might be employed. The center 
 span is the same as the arch in the previous article with the exception of the end posts 
 which are here made vertical, thus increasing the span to 180 ft. Owing to this dif- 
 ference, the computations in Tables 50A and 50s no longer apply because the stresses 
 SA, SB and S a are now slightly changed. Otherwise the preliminary computations would 
 be precisely the same in both structures. 
 
 The difference between the two-hinged arch and the present structure with cantilever 
 side spans is entirely in the principal system which results when the redundant thrust 
 X a is removed. In the first case the principal system is a simple truss on two supports 
 A and B, while in Fig. 5lA the principal system becomes a cantilever when the hinged 
 support at A is made movable for the purpose of analysis. The supports at D and F 
 are roller bearings and the simple trusses DC and EF are hinged at C and E respectively. 
 
ART. 51 
 
 SPECIAL APPLICATIONS OF INFLUENCE LINES 
 
 177 
 
178 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. X 
 
 The two-hinged central span, as in any ordinary two-hinged arch, is made deter- 
 minate by substituting a roller bearing for the hinged bearing at A. 
 
 The XQ influence line is found as in the previous problem and the computation of the 
 Ew loads is not repeated here. However, the loads Ew and_EVt- 2 , must_now be included 
 because these loads affect the directions of the extreme rays C'A' and B'E' of the equilib- 
 rium polygon outside of the span AB while they have no bearing on 'the A" a influence 
 line within this span. 
 
 The complete X a influence line for the whole span DF is found by prolonging the 
 extreme rays of the equilibrium polygon A'B' to the points C" and E' and finally draw- 
 ing the lines D'C' and E'F'. The middle ordinate A may be computed from the force 
 polygon as follows: 
 
 X= " measured to scale of lengths, 
 
 4/2 
 
 or 
 
 measured to scale of ordinates. 
 
 The lines A'C' and B'E', found by laying off A, are respectively parallel to the 
 extreme rays of the force polygon. 
 
 Stress influence areas- These are found precisely as in Art.' 50, so far as the span 
 AB is concerned, by laying off end ordinates T,A =&A/Sa and fjB = ^B/Sa and the factor 
 becomes a=S a . In each case the two end rays so determined will intersect in a point 
 i r , which is vertically under the center of moments i of the member in q^stion. Also 
 the S /S a influence lines must be straight over the panel containing the member. 
 
 Outside of the span A'B the S /S a lines- are drawn as for any cantilever system as 
 per Art. 26. 
 
 ART. 52. FIXED FRAMED ARCHES 
 
 A framed arch with fixed ends has three external redundant conditions according 
 to Eq. (3c) and Fig. 3j, and hence requires for its analysis three elasticity equations either 
 of the form of Eqs. (7n) or Eqs. (8D) . 
 
 Temperature changes and abutment displacements come into prominence here. 
 These effects are bound to remain more or less problematic because the actual circum- 
 stances attending the construction and later life of the structure cannot be foretold 
 with any high degree of certainty. Therefore, it would seem unnecessary to attempt 
 the analysis of the load effects with any extraordinary refinement and some assumptions 
 may be made to simplify the work, provided they are on the side of safety. 
 
 It is nearly always permissible to neglect the effect of the web system in comput- 
 ing the elastic loads w. In preliminary work it is also admissible to choose a constant 
 cross-section for the chord members. 
 
 As a general criticism it might be added that fixed framed arches are not commendable 
 for the flat type, since the temperature and reaction effects produce very considerable 
 
ART. 52 
 
 SPECIAL APPLICATIONS OF INFLUENCE LINES 
 
 179 
 
 stresses which increase as the rise diminishes. Good rock foundations must be available 
 under all circumstances, otherwise the fixed arch is prohibitive. 
 
 It is always advisable to separate the computation of the load stresses from the 
 temperature and reaction displacement stresses so as to determine the relative importance 
 
 of the latter. 
 
 General relations between the external forces and the principal system. There are several 
 wavs in which the three external redundant conditions may be applied, depending on 
 the choice of the principal system. One of these was illustrated in Figs. 44E to J, where 
 the principal system was a cantilever fixed at one end. Another assumption might 
 be made by cutting the arch at the crown and creating two cantilevers fixed at the 
 respective abutments of the arch. 
 
 However, the simplest determinate structure is always a truss on two supports, 
 and in the present case that disposition will be made, as it affords the most compre- 
 hensive solution. 
 
 FIG. 52A. 
 
 All three methods have been used by Professor Mueller-Breslau and others, and the 
 final results, are identical and involve about the same amount of labor. In each case 
 the external redundant conditions may be so chosen that the application of the 
 plified Eqs. (44A) becomes possible in accordance with the discussion in Art 44 
 
 Fi 52A illustrates the relation of the external forces for a single applied load I , 
 producing reactions R, and R z , intersecting in the point C on this load. This fixes the 
 points ttl and b, on the verticals through the outer supports a and ftjnth span I. ine 
 triangle ^Cb[ thus becomes a resultant polygon with the closing line a^. 
 R l may be resolved into the vertical component A and the haunch thrust H along afr. 
 The reaction R 2 may similarly be resolved into the vertical reaction B and the 
 H' which latter is equal and opposite to the H' acting at ai. 
 
 ' The vertical reactions A and B are the same as for a simple beam o* with deter- 
 minate supports A and B . Hence A =P(l-e)+l and B = Pe/l. Also the moment, f, 
 P intTof the simple beam, equals M^-KH, wh_ere_H is the honzonta componen 
 ' and K is the vertical ordinate of the triangle a.Cb, through the point m. Hence 
 
180 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. X 
 
 K = M om /H=M om /H' cos a. Therefore, the resultant polygon a l Cb l is determined 
 when H' and the closing line a^bi are found. 
 
 The external redundant conditions are now removed by relieving the fixed ends 
 and resting the curved structure on two determinate supports at the extreme outer 
 points a and b, Fig. 52B. This simple truss of span /, provided with a hinged bearing at 
 a and a roller bearing at b, constitutes the principal system. 
 
 To each end of the principal system an imaginary rigid disk is attached as shown 
 in the figure by the two shaded triangles. The redundant conditions are now applied 
 to these disks as external forces or moments, thereby re-establishing conditions of stress 
 in the principal system which are identical with those produced in the original structure 
 while the redundant conditions were active. The two disks are not connected at 0, but 
 are free to transmit a set of forces independently to each abutment. 
 
 The equal and opposite forces Xb are chosen vertically, and the one acting upward 
 is supposed to act on the disk OA. The forces X c are equal and opposite, but of unknown 
 direction /? with the horizontal, the one acting to the right is applied to the disk OA. The 
 
 FIG. 52s. 
 
 two moments X a are also equal and opposite and act separately, one on each disk. The 
 pole is not yet determined, but will be fixed by certain geometric conditions to be 
 established later. 
 
 To determine the exact relations between these redundant forces and the principal 
 system it is preferable to discuss only the forces acting on one end of the span. In Fig. 
 52c, the left-hand abutment is shown with the redundant forces which are active at 
 that end only. A similar set, not shown, would be active at the right-hand abutment. 
 
 The structure is now referred to coordinate axes (x, y) with origin at 0. The y axis 
 is made vertical, and the x axis is coincident with the redundant X c making the angle 
 /? with the horizontal. The location of and the angle /? are still unknown. The ordi- 
 nate y m of any point m is measured vertically from the x axis, while the abscissa x m of 
 this point is measured horizontally from the y axis instead of parallel to the x axis. This 
 is more convenient Jnjihe considerations which follow. 
 
 The rigid disk aa'O connects the origin with the arch along aa' ', and to this origin 
 are applied two equal and opposite forces H' , which are equal and parallel to the original 
 haunch thrust acting at a^ The equilibrium of the principal system and of the fixed 
 
ART. 52 
 
 SPECIAL APPLICATIONS OF INFLUENCE LINES 
 
 181 
 
 arch thus remains undisturbed. In the original fixed condition the force P produced 
 the reactions RI and R2 and these were resolved into components A and H' acting at 
 ai and B and H' acting at 61. 
 
 Suppose now that all the external forces to the left of a section tt act on the principal 
 system only, and that the three forces H' and the vertical reaction A are applied to the 
 rigid disk aa'O and are thence transmitted to the principal system. Then the force H' 
 at a\ and the force H' ', in opposite direction at 0, form a couple with lever arm z cos a 
 producing a moment X a =H'z cos a. The other force H', acting at and to the right, 
 may be resolved into two components X b and X c , where X b is vertical along the y axis, 
 and X c acts along the x axis. The external forces to the left of the section and acting 
 on the principal system, are then P, A , X b , X c and a moment X a =H'z cos a =Hz . 
 Of these the two forces Xb and X c and the moment X a constitute the redundant condi- 
 tions while the forces P and A are known and all are applied to the principal system 
 
 FIG. 52c. 
 
 to the left of the section tt. A similar set of external forces acts on the principal system to 
 the right of the section, but these are not shown in Fig. 52c. 
 
 The moment of all external forces about any point m of any frame, involving thre 
 redundant conditions, is expressed by Eq. (7 A) as 
 
 M m =M om -M a X a -M b X b -M c X c 
 
 (52A) 
 
 wherein M om =A (l i -x m ) -Pd-^ the moment about m due to the load P acting on the 
 principal system. This is condition X=0. . . 
 
 M a = l=the moment about m due to the moment X a = l applied to the principal 
 
 system. Condition X a =1. . . , 
 
 M 6 = l.* m =the moment about m due to the force X b = l acting on the prmcipa 
 
 system. Condition X b = l. 
 
 M c = l.</ m cos/?=the moment about m due to the force X c = l acting on the prin- 
 cipal system. Condition X c = l. 
 
182 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP, x 
 
 Substituting these values into Eq. (52 A) gives the following fundamental moment 
 equation for fixed arches: 
 
 M m =M om l-X a x m X b y m cos8X c ...... (52s) 
 
 Before M m can be determined for any point m of the arch, the three redundants 
 X a , X b and X c must be evaluated from three simultaneous work equations of the form 
 of Eqs. (44n) , which equations may be made to apply to the present' problem by so 
 locating the (x, y} axes that d ab , d ac and d bc all become zero. 
 
 Since there are as many values for the redundants as there are panel points on the 
 arch, and for each point there will be as many values as there are positions of the mov- 
 ing loads requiring investigation, the only practical solution of the problem is by means 
 of influence lines, first for the three redundants and then for the moment M m for each 
 member of the arch. 
 
 The stress S in any member may always be found from the moment M m about 
 the center of moments m for that member and the lever arm r m measured from this moment 
 center perpendicular to the direction of the member. Hence 
 
 M m M om M a M b M t 
 
 o = - ; o = --- , ^a -- > &b , ana o c = , . . . (o2c) 
 
 I'm I'm ?"m ^m I'm 
 
 where the lever arm r m is constant for the same member. 
 
 Therefore, Eq. (52s) will furnish the stress in any member as 
 
 M 1 
 S = - = [M om -l-X a -x m X b -y m cos px e ] =S -S a X a -S b X b -S C X C , (o2c) 
 
 ' m 'm 
 
 M a I M b x m M c y m cos 8 
 
 where S a = - = ; S b = ?= ; and S c = = if - E ....... (52E) 
 
 * * - ' 4* 4" V / 
 
 ' m 'm 'm 'm 'm 'm 
 
 Location of the coordinate axes. According to the conditions imposed by the simplify- 
 ing process of Eqs. (8n), discussed in Art. 44, the coordinate axes must be so located 
 that the displacements d, bearing different double subscripts, must be made zero. 
 
 These displacements d as given by Eqs. (SB), then become 
 
 =0', 
 
 and substituting the values for the stresses from Eqs. (52E) then 
 
 ^ xl v 1 
 
 ^ 2j XW = (J 
 
 . ^xycosfll v 
 
 =04, = 1 -^f = cos P2xyw a =0 
 
 (52r) 
 
ART. 52 SPECIAL APPLICATIONS OF INFLUENCE LINES 183 
 
 where w a = l-l/EFr 2 for a moment unity, according to Eqs. (36B) and (36o), giving 
 w a =4l/r=Sl/EFr=Ml/EFr 2 and representing a certain geometric function called an 
 elastic load for some particular pin point with coordinates (x, y). 
 
 Eqs. (52r) then represent the conditions which determine the location of the 
 coordinate axes such that the simple work Eqs. (44A) become applicable. 
 
 The first two conditions (52r) imply that the origin of the (ar, ?/) coordinate axes 
 is the center of gravity of the several elastic loads of all panel points of the principal sys- 
 tem. This must be so because the moments ^lxw a and %yw a could not be zero unless the 
 resultant Hw a passes through the origin. According to the third condition, the angle 
 between the axes must be such that the centrifugal moment 2>xyw a =Q, which is true 
 when the axes are related as conjugate axes. 
 
 The origin may then be located with respect to any assumed pair of axes as the 
 (z, v) axes in Fig. 52c, where the z axis is taken vertically through the point a and the v 
 axis is any horizontal axis conveniently located say through a'. The coordinates z, i\ 
 of all the panel points, are then determined from the arch diagram and tabulated. The 
 w a elastic loads are computed from Eq. (36fl) (and Eqs. (36D) if the web members are 
 included) for each pin point. See the problem in Art. 51. Finally the moments zw a 
 and vw a are found and from these the coordinates li and z ' for the origin are obtained 
 
 from 
 
 , ^vw a , 2zw a ._ , 
 
 ''- and * ~ ..... 
 
 This fixes the y axis, which is parallel to the vertical z axis through the center of 
 gravity 0. The x axis, while passing through 0, makes some angle ft with the horizontal 
 such that I,xyw a =0, according to the last of Eqs. (52r). 
 
 The (x, y) coordinates are derived from the (z, v) coordinates when the angle /? is 
 determined. Taking /? positive when measured to the left of the origin and below the 
 horizontal, or to the right and above the horizontal as shown in Fig. 52c, then 
 
 x=li v and y=z z ' +x tan /? ...... (52H) 
 
 The angle is found by substituting the value for y from Eqs. (52n) into the 
 condition equation, giving 
 
 z f +x tan fi]w a =0; 
 
 or I>xzw a -z f Zxw a + tan p2x 2 w a =0; 
 
 and noting that 2xw a =Q, then 
 
 (52j) 
 
 The abscissa? x being known from Eqs. (52n) the values Xxzw a and Xx 2 w a are readily 
 found and tan /? is then obtained from Eq. (52j). Finally, the y ordmates are computed 
 by the second Eq. (52n) and the new axes and (x, y) coordinates are thus determined. 
 
 For symmetric arches the x axis is horizontal and tan =0, thus greatly lessening 
 the foregoing computations. 
 
184 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. X 
 
 Influence lines for X a , X&, X c and M m . The coordinate axes (x, y) having been located 
 to fulfill the requirements making ^=^=0, d ac =d ca =0 and d bc =o cb =0, the simplified 
 Eqs. (44A) become applicable to the present problem. 
 
 Since it is desirable to investigate separately the effects due to loads, temperature 
 and abutment displacements, the load effect will be considered first for immovable abut- 
 ments. Eqs. (44A) then take the simple form of Eqs. (44B), as 
 YP ft 
 
 *^*- m.u tnn. -*T -. jji^ jnu i -\r *-* * m^ jnr. / ~c\ \ 
 
 (,)2K) 
 
 and 
 
 Y p ^ 
 
 *->* m u mc 
 
 r s 
 
 ' 
 
 Eqs. (SB) give the displacements dm, w> and cc in terms of the stresses in the 
 members. Noting that p =l/EF, and substituting the values for S a , S b and S c from Eqs. 
 (52E) , then Eqs. (SB) give 
 
 EFr 2 ' 
 
 
 d bb = 
 
 'EFr 2 ' 
 
 and calling l/EFr 2 =w a ; x 2 l/EFr 2 =x 2 w a =xw b and y 2 l/EFr 2 =ij 2 w a =yw c then Eqs. (52x) 
 become for a single load unity at any point ra : 
 
 1 $mb 1 ' ' 
 
 * ' " 
 
 ' " 
 
 (52L) 
 
 Eqs. (52L) furnish the values X a , X b and X c for any position m of a single moving 
 load P TO = 1, as functions of the deflections d ma , d^ and d mc of the point m resulting from 
 the conventional loadings X a = ], X b = l, and X c = l. 
 
 The exact significance of the deflections 3 ina , dmb and d mc may be determined from 
 their values as given by Eqs. (8 A), noting that S = M om /r m for a load P^-l applied at 
 m from Eqs. (52c), also using the values given by Eqs. (52E). Hence 
 
 is> v^oo ' y "* om' 
 
 EF 
 
 J_ 
 
 b lEF 
 
 I 
 
 M om xl 
 
 =cos 
 
 (52M) 
 
 Now ^ ma is the deflection of a point m due to X a = l and P TO =0, and by Eqs. (52M) 
 it is equal to the sum of the moments M om w a of the loads w a about m. Hence d ma must 
 represent the ordinate to a moment digram drawn for the loads w a with a pole unity. 
 Also, if this moment diagram be drawn with a pole H a = 2w a , then the resulting ordinates 
 i) a will represent ordinates of the X a influence line according to Eqs. (52L). 
 
ART. 52 SPECIAL APPLICATIONS OF INFLUENCE LINES 185 
 
 Similarly the moment diagram drawn for the loads w b =xw a with a pole H b = 
 will furnish ordinates jj b for the X b influence line and a third moment diagram drawn 
 'or the loads w c =yw a with a pole H c = cos {32>yw c will give ordinates )j c for the X c influence 
 ine, noting that one of the cos /? factors cancels. 
 
 Hence the equations for the influence lines of the three redundant conditions from 
 Eqs. (52L), become 
 
 1 /^ <> 
 
 (52N) 
 
 dr, 
 
 me 
 
 TTvi TT 
 
 cos Lw c H 
 
 *-< 
 
 These influence lines remain the same for the same structure and hence are drawn 
 
 only once. 
 
 The moment influence line M m , for any point TO with coordinates x m and y m , is derived 
 from Eq. (52s) , giving any moment ordinate as 
 
 r, m =M m =l\l f > m -\X a +x m X b +y m cos8X c ]=rj om -[-n a +x m -f) b +y m cos fa], . (52o) 
 
 where T lom is any ordinate of the ordinary moment influence line M om , drawn for the point 
 TO of a simple beam on two determinate supports A and B , which is a different line for 
 each point TO. The ordinates r) a , y b and TJ C are those respectively of the X a , X b and X c 
 influence lines all under the same panel point of the truss. 
 
 Hence, the moment influence line M m for any point TO is drawn by computing the 
 ordinates -[rj a +x m rj b +y m cos fa] for _all panel points of the loaded cord and plotting 
 these ordinates negatively from the M om influence line. Positive areas thus correspond 
 to positive moments. 
 
 This M m influence line will serve to find the maximum and minimum bending 
 moments for the point TO due to any system of concentrated loads. The influence line 
 gives the load divides. 
 
 The stress influence line for any member may be derived from the moment influence 
 line drawn for the center of moments TO for that particular member. Thus if r m is the 
 lever arm for a certain member S with center of moments m, then S=M m /r m , and the 
 ordinates T?, for this stress influence line by Eq. (52o) become 
 
 (52P) 
 
 where the several 9 ordinates are as above defined and all measured under the same 
 panel point. Positive areas then give positive or tensile stress. The moment influence 
 line for the center of moments for any member may be directly used by applying the 
 factor l/r w to obtain the stress. 
 
 The rather lengthy operation of computing all these ordinates for the several stress 
 influence lines for all the members may be considerably shortened by employing the 
 
1S6 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. X 
 
 following suggestions: Thus the multiplications x m -fj b and y m cos /?j? c , may be performed 
 graphically by laying off angles whose tangents are respectively x m and y m cos /?, as illus- 
 trated in Fig. 52o, using the scale of the TJ ordinates. 
 Two such diagrams are required for each M m or S line. 
 
 If a proportional divider is at hand, then it may be 
 used to perform these multiplications by setting the divi- 
 ders first for the ratio 1 : x m and then for 1 : y m cos /?. 
 
 Aside from the above methods, the stress influence 
 line for any web member may be derived from the 
 influence lines of two adjacent chords by using the 
 following method given by Professor Mueller-Breslau. 
 
 When the top chord is the loaded chord, then the 
 FIG. 52o. adjacent members of the bottom chord are used, and 
 
 vice versa. 
 
 In Fig. 52E, assume a unit positive stress in L\ and resolve the same into com- 
 ponents 1 parallel to D\ and +2 parallel to D 2 . 
 
 FIG. 52E. 
 
 FIG. 52r. 
 
 Then assume a unit positive stress in L 2 and resolve this into components + n\ 
 parallel to D\ and /i 2 parallel to D 2 . Then for the top chord loaded and no load at 
 tr>, the stresses D\ and D 2 become 
 
 D 2 =+e 2 L l - 
 
 +L 2 
 
 (52Q) 
 
 By substituting influence ordinates for the stresses D and L this furnishes a ready 
 means of deducing a D influence line from the two influence lines of the adjacent chords. 
 Thus the DI influence area is the area between the L 2 line and the z\L\/ // t line and the 
 
ART. 52 
 
 SPECIAL APPLICATIONS OF INFLUENCE LINES 
 
 187 
 
 final Di area thus obtained will have a factor m. The ordinates to be used in the equa- 
 tions are always under the same load point. 
 
 When the bottom chord is the loaded chord and there is no load at the point n, 
 then using the stresses as indicated in Fig. 52F, the diagonals D 2 and D 3 become 
 
 (52 R) 
 
 The value of D 2 will be the same in both equations Q and R and the signs will come 
 alike by noting that the influence ordinates for the L and U lines have opposite signs and 
 hence the coefficients s and have opposite signs from those used in Eqs. (52 Q ). 
 
 The resultant polygon for any case of applied loads P may be located by finding the 
 corresponding values of X a , X b , and X c from the three influence lines for thes 
 having previously located the (a;, y) axes. 
 
 The redundants X e and X b , Fig. 52c ? were taken as the components of H , respec- 
 tively coincident with the x and y axes. The angle /?, which the x axis makes with t 
 horizontal is given by Eq. (52j) and the y axis was taken vertically, 
 izontal component H of X c is 
 
 H =X C cos /?, and from Fig. 52c, 
 tan a.=t&n (3+-jj 
 
 
 cos a 
 
 z ~ 
 
 cos a 
 
 X a = Xg 
 
 H'cosa H 
 
 H 
 
 . . (52s) 
 
 X a = 
 
 COS = 
 
 These dimensions fix the location of the closing line a 
 of the abutments also the haunch thrust H' all in terms of 
 
 on the two end verticals 
 X and X 
 
 r, 
 
 
188 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. X 
 
 A force polygon is drawn by laying off all the loads P in proper succession, dividing 
 this load line into the parts A and B and at the dividing point draw a line parallel to 
 a[bi of length equal to H' '. This determines the pole of the force polygon from whicli 
 the reactions RI and R 2 and the resultant polygon through ai and 61 are easily drawn. 
 See also the force polygon in Fig. 52A, showing how the pole is located when H' , A 
 and B are given. 
 
 Temperature stresses. For the general case of temperature effects, each member may 
 be supposed to undergo some change 4l t =etl in its length. These changes will pro- 
 duce a deformation in the frame, giving rise to external redundant forces X at , X bt and X ct , 
 which may be found from Eqs. (44c), where abutment displacements and load effects 
 are excluded. 
 
 Using the values of d a t, $bt and d ct as given by Eqs. (Sc) and then noting the sub- 
 stitutions for Saa = w a , dbb = 2>xw b and d cc = *>yw c made in Eqs. (52N) , the Eqs. (44c) become 
 
 Y _dat_'S a Jlt 
 
 -A at * vi 
 
 Oaa 
 
 _d bt _ 
 
 J\.bt ^ 
 
 COS 
 
 (52T) 
 
 where d at represents a rotation measured in arc, while d bt and d ct are the displacements 
 of the origin 0, due to the assumed temperature changes measured in the directions of the 
 y and x axes respectively. 
 
 These redundant conditions would then produce a moment M m t about any point 
 m as given from Eq. (52s), where M om =Q, thus 
 
 -M mt =X at +x m X b t+y m cos^X ct . ..... (52r) 
 
 Since the changes Al t in the lengths of the members represent a simultaneous condition 
 and the stresses S a , S b and S c were previously found in computing the w loads, it is best 
 to solve Eqs. (52T) analytically, using the denominators previously found for the pole 
 distances of the X influence lines. The moment Eq. (52u) can then be solved for any 
 moment point to obtain the stress in .a corresponding member. 
 
 For a uniform change in temperature of t from the normal, an approximate deter- 
 mination of temperature stresses may be obtained for arches of high rise. For flat arches 
 this assumption would not be permissible. When the effect of the redundants X a and 
 X b is neglected then the stresses may be found from Eqs. (52T) and (52u) as 
 
 v -- ........ (52v) 
 
 \ 
 
 The effect of abutment displacements may be investigated in a similar manner, using 
 Eqs. (44A) and omitting the terms representing the loads P and the temperature. The 
 
\BT. 52 SPECIAL APPLICATIONS OF INFLUENCE LINES 189 
 
 iisplacements Ar must then be assumed or estimated and o a , d b and d c become zero. Hence 
 the redundants and moments may be found in precisely the same manner above illustrated 
 for temperature effects, thus 
 
 Xar 
 
 **. 
 
 (52w) 
 
 where the d's in the denominators are those used in Eqs. (52T). 
 
 Example. Owing to the comparatively few fixed arches of the framed type in 
 existence, it was difficult to find a suitable example to use in illustrating the above method 
 of analysis. For this reason the present example, Figs. 52a, was taken with slight modi- 
 fication, from the one given by Professor Mehrtens in his " Statik der Baukonstruk- 
 tionen," Vol. Ill, p. 343. 
 
 The bridge has a clear span of 162 ft. and rise of 23.69 ft. The roadway is 30 ft. 
 wide and the estimated weight is about 400,000 Ibs., making 20 kips per truss per panel. 
 The uniform live load is taken at 110 Ibs. per sq.ft. of roadway or 30 kips per truss per 
 panel, for medium highway loading. The top chord is the loaded chord and the arch 
 is symmetric, making /? 0. 
 
 " The top chord panel points are on the arc of a circle whose radius is R u = 163.7 ft. 
 The chords are parallel and the radius for the bottom chord points is R t = 148.7 ft. 
 
 Ordinarily the computation might be carried out by neglecting the effect of the 
 web members, but for the sake of completeness all members will be included. 
 
 The cross-sections F, lengths I, and lever arms r of all the members, are tabulated 
 in Tables 52A and 52n, and the Ew a loads are computed by successive steps indicated by 
 the headings of the columns. The lever arms r u and r n are found as described in Figs. 
 36B and 36F In the present solution the areas were converted into sq.ft. in computing 
 Al so that the modulus E enters with its real value of 4,176,000 kips per sq.ft. instead of 
 29 000 kips per sq.in. as used in the problem of Art. 51. The stresses S a due to a moment 
 fc-1 kip ft., are easily found, as the reciprocals of the lever arms r, being careful 1 
 observe the signs of the stresses. 
 
 The arch being symmetric, the y axis is known to be the vertical axis of symmetry, 
 and the angle ft, which the x axis makes with the horizontal, becomes zero. Hence, the 
 (x y) axes are located by computing the ordinate z ' =E2zw a /EXw a -.21.927 *; thus 
 determining the distance from the v axis to the center of gravity of the w a loads, bee 
 Table 52c. The y ordinates then become y=z-z '. The (x, z) coordinates must of 
 course be determined from the arch dimensions or the equations of the chord curves. 
 Table 52c then gives the functions Ew b =Exw a , Ew c =Eyw a , Exw b and Eyw c all in 
 terms of the w a loads. The sums EXw a , EXxw b and E2yw e represent the pole distances 
 for the X a , X b and X e influence lines such that the influence ordinates will 
 when measured to the scale of lengths. 
 
190 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. X 
 
 X c INFLUENCE LIKI^-ORDINATES IOO TIMtSfO SCALE OF LENGTHS 
 
 NCE LINE-ORDIMATCSACTUALTOSCALEOFLfNGTHS. ! 
 I 
 
 FIG. 52c. 
 
SPECIAL APPLICATIONS OF INFLUENCE LINES 
 
 191 
 
 TABLE 52A 
 w a LOADS FOR CHORD MEMBERS 
 
 
 
 
 
 l 
 
 .S',, 
 
 S a l 
 
 
 
 Chord. 
 
 F 
 
 I 
 
 r 
 
 
 
 *EJl = jr 
 
 *Ew a = ^ 
 
 Panel 
 Point. 
 
 
 Sq.in. 
 
 ft. 
 
 ft. 
 
 Kips. 
 
 Kips sq.ft. 
 
 ft. 
 
 
 
 2 
 
 100 
 
 20.73 
 
 14.92 
 
 -0.0670 
 
 -0.0965 
 
 -2.002 
 
 0.1341 
 
 1 
 
 2 4 
 
 110 
 
 19.51 
 
 14.84 
 
 -0.0674 
 
 -0.0883 
 
 -1.728 
 
 0.1165 
 
 3 
 
 4 6 
 
 120 
 
 18.72 
 
 14.79 
 
 -0.0676 
 
 -0.0811 
 
 -1.512 
 
 . 1022 
 
 5 
 
 fi 8 
 
 124 
 
 18.25 
 
 14.77 
 
 -0.0677 
 
 -0.0786 
 
 -1.440 
 
 0.0975 
 
 7 
 
 8 10 
 
 130 
 
 18.03 
 
 14.79 
 
 -0.0676 
 
 -0.0737 
 
 -1.325 
 
 0.0896 
 
 9 
 
 
 82 
 
 20.59 
 
 15.14 
 
 0.0661 
 
 0.1161 
 
 2.390 
 
 0.1580 
 
 2 
 
 Q K 
 
 94 
 
 19.33 
 
 15.21 
 
 0.0657 
 
 0.1008 
 
 1.944 
 
 0.1277 
 
 4 
 
 ^ 7 
 
 100 
 
 18.55 
 
 15.25 
 
 0.0656 
 
 0.0945 
 
 1.757 
 
 0.1152 
 
 6 . 
 
 7 9 
 
 104 
 
 18 . 13 
 
 15.28 
 
 0.0654 
 
 0.0906 
 
 1.642 
 
 0.1074 
 
 8 
 
 1(9-9') 
 
 128 
 
 9.00 
 
 15.31 
 
 0.0653 
 
 0.0734 
 
 0.662 
 
 0.0432 
 
 10 
 
 * Above w a loads are for ' = 29,000X144=4,176,000 kipa per sq.ft. 
 
 TABLE 52s 
 w a LOADS FOR WEB MEMBERS 
 
 Member. 
 
 F 
 Sq.in. 
 
 I 
 ft. 
 
 r 
 
 ft. 
 
 *- 
 
 Kips. 
 
 S a 
 f-p 
 
 ips. sq.ft. 
 
 .<s ; 
 tUl-^s 
 
 r 
 
 ft. 
 
 r u 
 and 
 r n 
 
 ft. 
 
 *Edl 
 
 * Ew a = 
 
 For loaded panel 
 Points. 
 
 At 
 Panel 
 Point. 
 
 + 
 
 - 
 
 0-1 
 1-2 
 
 2-3 
 3^ 
 
 4-5 
 5-6 
 6-7 
 7-8 
 8-9 
 9-10 
 
 8 
 14 
 
 18 
 10 
 18 
 6 
 14 
 6 
 10 
 6 
 
 15.03 
 24.06 
 
 15.26 
 21.79 
 15.66 
 20.09 
 16.19 
 18.78 
 16.91 
 17.76 
 
 
 
 
 
 
 
 
 
 780.87 
 132.91 
 443.54 
 154.04 
 331.25 
 175.75 
 260.26 
 206.93 
 231.40 
 
 + 0.00128 
 0.00752 
 0.00225 
 0.00649 
 0.00302 
 0.00569 
 0.00384 
 0.00483 
 0.00432 
 
 + 0.0132 
 0.0602 
 0.0324 
 0.0519 
 0.0725 
 0.0585 
 0.0922 
 0.0696 
 0.1037 
 
 + 0.3176 
 0.9186 
 0.7060 
 0.8128 
 1.4565 
 0.9471 
 1.7315 
 1 . 1769 
 1.8417 
 
 {12.81 
 \13.05 
 /21.90 
 \18.96 
 [13.38 
 \13.65 
 J19.98 
 \17.70 
 / 13.80 
 1 14.40 
 /18.45 
 \16.62 
 /14.37 
 \15.18 
 /17.10 
 \15.84 
 J15.03 
 \16.05 
 
 0.0248 
 
 
 
 2 
 2 
 4 
 2 
 4 
 4 
 6 
 4 
 6 
 6 
 8 
 6 
 8 
 8 
 10 
 8 
 10 
 
 0.0243 
 0.0419 
 
 
 0.0485 
 0.0533 
 
 0.0517 
 0.0407 
 
 
 0.0459 
 0.1055 
 
 
 0.1011 
 0.0513 
 
 
 . 0570 
 0.1205 
 
 
 0.1141 
 0.0688 
 
 
 0.0743 
 . 1225 
 
 
 0.1148 
 
 
 
 
 * 
 
 Above w. loads are for = 29,000X144 = 4,176,000 kips per sq.it. 
 
192 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. X 
 
 TABLE 52c 
 COMPUTATION OF w LOADS AND POLE DISTANCES 
 
 
 T/it <1 
 
 
 
 
 
 
 
 
 
 
 Pin 
 
 Points. 
 
 own 
 Ew a 
 Loads. 
 
 z 
 
 Ezw a 
 
 X 
 
 y=z-z ' 
 
 Exw a 
 
 Ew c = 
 
 Eyw a 
 
 Exwb 
 
 Eyw c 
 
 Pin 
 Points. 
 
 
 
 ft. 
 
 
 ft. 
 
 ft. 
 
 
 
 
 
 
 
 
 0.0248 
 
 12.03 
 
 0.2983 
 
 90 
 
 - 9.897 
 
 2.232 
 
 -0.245 
 
 200.88 
 
 2.429 
 
 
 
 1 
 
 0.1341 
 
 0.00 
 
 0.0 
 
 81 
 
 -21.927 
 
 10.862 
 
 -2.940 
 
 879 . 83 
 
 64.474 
 
 1 
 
 2 
 
 0.1451 
 
 22.31 
 
 3.2372 
 
 72 
 
 + 0.383 
 
 10.447 
 
 + 0.056 
 
 752 . 20 
 
 0.021 
 
 2 
 
 3 
 
 0.1165 
 
 9.99 
 
 1.1638 
 
 63 
 
 -11.937 
 
 7.340 
 
 -1.391 
 
 462.39 
 
 16.601 
 
 3 
 
 4 
 
 0.1893 
 
 29.83 
 
 5.6468 
 
 54 
 
 + 7.903 
 
 10.222 
 
 + 1.496 
 
 552 . 00 
 
 11.823 
 
 4 
 
 5 
 
 0.1022 
 
 17.03 
 
 1.7405 
 
 45 
 
 - 4.897 
 
 4.599 
 
 -0.501 
 
 211.55 
 
 2.451 
 
 5 
 
 6 
 
 0.1292 
 
 34.99 
 
 4 . 5207 
 
 36 
 
 + 13.063 
 
 4.651 
 
 + 1.688 
 
 167.44 
 
 22 . 046 
 
 6 
 
 7 
 
 0.0975 
 
 21.53 
 
 2.0992 
 
 27 
 
 - 0.397 
 
 2.633 
 
 -0.039 
 
 71.08 
 
 0.015 
 
 7 
 
 8 
 
 0.1040 
 
 38.01 
 
 3.9530 
 
 18 
 
 + 16.083 
 
 1.872 
 
 + 1.673 
 
 33.70 
 
 26.900 
 
 8 
 
 9 
 
 0.0896 
 
 23.69 
 
 2 . 1226 
 
 9 
 
 + 1.763 
 
 0.806 
 
 + 0.158 
 
 7.26 
 
 . 279 
 
 9 
 
 4(10) 
 
 0.0027 
 
 39.00 
 
 0.1053 
 
 
 
 + 17.073 
 
 0.0 
 
 + 0.046 
 
 0.00 
 
 0.787 
 
 4(10) 
 
 
 B 
 
 -2w? a = 
 
 
 E 
 
 -2zw a = 
 
 
 
 E 
 
 E 
 
 E 
 
 E 
 
 
 
 
 
 
 
 
 2 
 
 2 
 
 2 
 
 2 
 
 
 
 1 . 1350 
 
 
 24 . 8874 
 
 
 
 55.664 
 
 + 5.117 
 
 3338.33 
 
 147.826 
 
 
 
 
 
 24 . 8874 
 
 
 
 
 -5.116 
 
 
 
 
 
 H o=2. 270 
 
 *o'~ 
 
 
 = 21.92 
 
 7ft. 
 
 
 
 #b = 6676.6 
 
 H c = 295.65 
 
 
 1.135 
 
 TABLE 52o 
 ORDINATES FOR MOMENT INFLUENCE LINES 
 
 
 
 
 
 M 4 Influence Line, 
 
 M 5 Influence Line. 
 
 Panel 
 
 
 
 
 rim = rjom [r)a + 54 T/b +7.9)} c ] 
 
 Tjm = rjom [rja + 45)?6 4.897 JJr] 
 
 Point. 
 
 i a 
 
 '" 
 
 / c 
 
 
 
 
 ft, 
 
 ft. 
 
 ft. 
 
 TjO 
 
 54, 6 
 
 7.9, e 
 
 rjm 
 
 no 
 
 45, b 
 
 4.897>? c 
 
 9* 
 
 2 
 
 8.2 
 
 0.078 
 
 0.100 
 
 14.4 
 
 4.21 
 
 0.79 
 
 1.20 
 
 13.5 
 
 3.51 
 
 0.49 
 
 2. 28 
 
 4 
 
 14.6 
 
 0.104 
 
 0.330 
 
 28.8 
 
 5.62 
 
 2.61 
 
 5.97 
 
 27.0 
 
 4.68 
 
 1.62 
 
 9.34 
 
 6 
 
 18.2 
 
 0.088 
 
 0.535 
 
 25.2 
 
 4.75 
 
 4.23 
 
 -1.98 
 
 31.5 
 
 3.96 
 
 2.62 
 
 11.96 
 
 8 
 
 20.0 
 
 0.048 
 
 0.647 
 
 21.6 
 
 2.59 
 
 5.11 
 
 -6.10 
 
 27.0 
 
 2.16 
 
 3.17 
 
 8.01 
 
 10 
 
 20.5 
 
 0.000 
 
 0.657 
 
 18.0 
 
 0.00 
 
 5.19 
 
 -7.69 
 
 22.5 
 
 0.00 
 
 3.22 
 
 5.22 
 
 8' 
 
 20.0 
 
 -0.048 
 
 0.647 
 
 14.4 
 
 -2.59 
 
 5.11 
 
 -8.12 
 
 18.0 
 
 -2.16 
 
 3.17 
 
 3.33 
 
 6' 
 
 18.2 
 
 -0.088 
 
 0.535 
 
 10.8 
 
 -4.75 
 
 4.23 
 
 -6.88 
 
 13.5 
 
 -3.96 
 
 2.62 
 
 1.8S 
 
 4' 
 
 14.6 
 
 -0.104 
 
 330 
 
 7.2 
 
 -5.62 
 
 2.61 
 
 -4.39 
 
 9.0 
 
 -4.68 
 
 1.62 
 
 0.70 
 
 2' 
 
 8.2 
 
 -0.078 
 
 0.100 
 
 3.6 
 
 -4.21 
 
 0.79 
 
 -1.18 
 
 4.5 
 
 -3.51 
 
 0.49 
 
 0.30 
 
 All ordinates are in feet. 
 
ART. 52 SPECIAL APPLICATIONS OF INFLUENCE LINES 193 
 
 The three influence lines for the redundants are now drawn by constructing the 
 force polygons, using the w a , w b and w c forces in the order of the loaded panel points and 
 drawing the corresponding equilibrium polygons. The scales used for the w forces arc 
 any convenient scales, noting that the pole distance must of course be laid off to the same 
 scale as the forces. The poles H b and H c were divided by one hundred, thus making 
 the tjb and rj c ordinatcs hundred times actual when measured to the scale of lengths. 
 The w c forces being of both signs it is best to plat their algebraic sums from some initial 
 point of the load line and number the points in the order of the loads, thus 0, 1,2, 3, etc., 
 to 10. The pole rays are drawn in the same order. Note the check by which the end 
 rays of the X a line intersect in a point on the y axis. 
 
 If the w a forces were made to act horizontally, then by the method given in Art. 
 38, the horizontal resultant of these forces would be obtained acting at the intersection 
 of the outer rays. This resultant EHw a would intersect the y axis in the center of gravity 
 0. However, the method of finding zj by computation is preferable, as the point 
 must be accurately located. 
 
 The moment influence line for any point as panel point 4, is now constructed by 
 computing the i? w ordinates from Eq. (52o) as illustrated in Table 52o. The coordi- 
 nates of point 4 are x =54 ft. and y=7.9 ft. and the rj a , T) b and rj c ordinates are scaled 
 from the three X influence lines. The rj ordinates are obtained from the M o4 influence 
 line which is the ordinary moment influence line for point 4 of a simple beam of span 
 Z = 180 ft. The table indicates the details of combining these several ordinates to obtain 
 the ordinates which are finally plotted (Figs. 52c) , giving the M 4 influence line. The 
 MS line is similarly found. 
 
 The M 4 influence line may be used to obtain the stress S 3 _ 5 in the chord 3-5 with 
 lever r = 15.21 ft. Since S=M/r, this same influence line becomes a stress influence 
 line with a factor l/r. 
 
 The M 5 influence line may thus be regarded as the stress influence line for the 
 chord 4-6 with a factor l/r = 1 -*- 14.79 =0.068. 
 
 Stress influence lines might also be developed from the formula 
 
 S =S S a X a S b X b S C X 
 
 C , 
 
 but this would require computing the stresses S b and S c in addition to the S a stresses 
 already found, and the stresses S A and S B required to draw the ordinary S line for any 
 member in question. 
 
 The influence lines for the web members can best be found from the chord influence 
 lines as described in the theoretical portion of this article. 
 
 It is not considered necessary to go further into this problem, as the theory is fully 
 treated and the general method is illustrated by finding a single stress influence line. 
 
 The temperature stresses are determined precisely as previously outlined. 
 
CHAPTER XI 
 DESIGN OF STATICALLY INDETERMINATE STRUCTURES 
 
 ART. 53. METHODS FOR PRELIMINARY DESIGNING 
 
 The term indeterminate, according to previous definitions given in Art. 2 and the 
 tests for identification discussed in Art. 3, is always used in the sense that the laws ot 
 pure statics fail to give a solution when applied to structures involving statically inde- 
 terminate or redundant conditions. In all structures of this class the analysis of stresses 
 can be accomplished only by resorting to the theory of elasticity. Hence, indeterminate 
 is not synonymous with impossible only in so far as the application of the laws ot statics 
 is concerned. 
 
 The complete analysis of a statically indeterminate structure, according to Eqs 
 (7 A) and (7n), involves the solution of three separate problems as follows: 
 
 1. The determination of all internal stresses So, resulting in the principal system 
 from the externally applied loads P. 
 
 2. The determination of all internal stresses St and Sr produced in the principal 
 system by the temperature changes and abutment displacements. 
 
 3. The determination of elastic deformations in the principal system produced by 
 the several stresses S a , Sb, S c , etc., S t and S r from which the redundant conditions X a , 
 Xb, X c , etc., X t and X r may be evaluated. 
 
 Eqs. (7n) and (SD) , furnish the solution of these three problems for any indeterminate 
 structure by means of considerations which are of similar character in each problem. 
 
 A statically determinate structure can always be analyzed by the methods known 
 in statics, when the live loads and the dead loads of the structure are known. Experience 
 in shop practice supplies approximate data for the weights of structures in common use, 
 but new types of structures, departing from the ordinary forms, necessitate repeated 
 approximations to determine the dead loads before a final analysis for the given live 
 loads becomes possible. 
 
 A statically indeterminate structure is likewise incapable of analyis until its dead 
 weight is known with some degree of certainty and will require a preliminary design 
 which is much more difficult to approximate than in the case of determinate structures, 
 because this involves a knowledge of the magnitude of the redundant conditions. These 
 redundant conditions in turn require that the cross-sections of the members be also 
 known. 
 
 Fortunately, the influence of a system of loads can be made to depend on relative 
 cross-sections, thus rendering an approximate solution of the redundants X possible, 
 
 194 
 
ART. 53 DESIGN OF STATICALLY INDETERMINATE STRUCTURES 195 
 
 before the actual cross-sections are definitely evaluated. This is done by making cer- 
 tain assumptions like adopting a uniform cross-section for all chord members and 
 neglecting the web members entirely, whence the variable F, involved in p=l/EF or 
 Eq. (42u) , is treated as a constant. The w loads required for constructing the deflection 
 polygon for the loaded chord for any condition X = l may then be made EF times the 
 values given by Eq. (36s) , using a pole distance of EF. The details of this process will 
 be discussed later; suffice it to say now that it becomes possible to construct an approx- 
 imate influence line for any redundant condition without knowing the final cross-sections 
 of the members. 
 
 With the aid of these approximate X influence lines the redundants may be evaluated 
 for any given live loads and some assumed dead loads. Then applying all these loads 
 simultaneously to the principal system, the first approximate values for the stresses 
 S may be determined from a Maxwell .diagram. The redundants X ar*e applied along 
 with the external forces. 
 
 These stresses S will now serve as a basis for a close approximation of the cross- 
 sections F from which our new influence lines for the redundants X may be found in the 
 usual manner as illustrated by the problems in Chapter X. A final analysis of the stresses 
 is now possible and from this the final sections are found. If the first assumption for 
 dead load was grossly in error then the first X influence lines might be used again, employ- 
 ing revised dead loads before the final analysis is made. 
 
 This process is on the whole similar to the one above cited for new types of deter- 
 minate structures, the dead loads for which are not accurately known. The additional 
 difficulty here encountered consists in evaluating the redundants, which depend both 
 on the sections and loads at the same time. Therefore, all such work is likely to be some- 
 what tedious. 
 
 The method of determining approximate values for the redundants in terms of the 
 mutual relations of the sections of the members among themselves, instead of the actual 
 sections, will now be described. 
 
 In some cases as for arches with parallel chords it is always permissible to assign 
 a constant section to all chord members and neglect the web system entirely. This 
 is then a very simple case and affords a ready solution for the X influence lines by com- 
 puting all w loads EF times too large. 
 
 When there is only one redundant condition or when there are two or three of 
 these so chosen as to comply with the conditions discussed in Art. 44 then for the above- 
 mentioned case of constant chord sections and disregarding the web system, the redundants 
 may be found analytically from Eq. (42o) whence 
 
 S , 53 , 
 
 - . . 
 
 where the stresses S must be obtained from a Maxwell diagram drawn for a maximum 
 case of combined live and assumed dead loads for the principal system. The stresses 
 
196 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XI 
 
 S a are found for the conventional loading X a = 1 from a similar stress diagram, and these 
 are independent of any actual loads. Eq. (53 A) shows that whenever any of the variables 
 common to the numerator and denominator become constant, they cease to affect the 
 value of X a . 
 
 The stress in each member of the principal system may now be found from 
 
 S=S S a X a , ... ........ (53s) 
 
 and the first approximate sections F may be evaluated from these stresses. 
 
 The approximate effect due to changes in temperature may also be included b} 
 finding X at from Eq. (44c) and Eqs. (SB) and (Sc) as 
 
 o at ZS a tl 
 
 A a< =^ = -- . , ......... (o3c ; 
 
 Oaa * 
 
 using some assumed constant value for F. 
 
 If for any reason the lengths of the chord members are all equal, then I, in Eqs, 
 (53A) and (53c) , would likewise be eliminated. 
 
 A somewhat more comprehensive method, especially when the chord sections are 
 not assumed equal, is obtained by drawing the influence lines for the redundants. This 
 method will now be outlined. 
 
 When the redundants are so chosen that the simplified Eqs. (44s) become applicable 
 then the X influence lines may be approximated by computing w loads for relative 
 sections of the members. Thus for any redundant 
 
 where F c is some constant chosen about equal to a typical chord section as described 
 below. 
 
 The w a loads from Eqs. (36s) and (36c), for condition X a = l, then become 
 
 and multiplying each by EF C , then, 
 
 ^. . (53 E ) 
 
ART. 53 DESIGN OF STATICALLY INDETERMINATE STRUCTURES 197 
 
 M a is the moment, with lever arm r, produced by the conventional loading Z a = l. 
 
 If a deflection polygon be drawn for these EF c w loads with pole distance unity, the 
 ordinates r? will represent values EF c d ma measured to the scale of lengths. Hence the 
 numerator of Eq. (53n) represents the sum of the products Prj, and the denominator is 
 
 2>S%pr = ^p/--jr=EF c 'ZM a w, according to Eq. (53E), thus making it possible to 
 
 construct any X influence line provided suitable assumptions for the values F C /F can 
 be made. 
 
 The equation of the X influence line for a load P = 1 then becomes 
 
 77T TJT ^ 77T TJ1 JJt 
 
 Zj2jr c O ma Hir c u ma 
 - = . f^F^ 
 
 7^ FW^ 1\/T iii * ^uor^ 
 
 2>S~l-pT 
 
 in which F c need not be numerically known, but the ratios F C /F, for different members 
 of the frame, must be approximated in order to compute the EF c w loads from Eqs. 
 (53E). 
 
 The value of F c should be chosen equal to that of a chord section of most frequent 
 occurrence, which would usually be nearest the crown of an arch or where the chord 
 approaches the horizontal. This will make F c /F = l for most chord members and 
 usually that assumption may be made for both top and bottom chord members. The 
 web system may be entirely neglected in the first approximation or if it is thought 
 desirable to include the web members then F C /F may be given some constant value 
 ranging from 2 to 10 for these members, depending on the character of the structure. 
 
 The F c should not be regarded as an average value of the chord sections because 
 the deflection polygon of a frame is governed largely by the chords near the center of the 
 span. Thus, if the depth of an arch is much greater at the springing than at the crown, 
 then the crown sections will govern. 
 
 From the approximate X influence lines the redundant conditions may be evaluated 
 for any case of total loading and the stresses S , S a , etc., can be found by Maxwell dia- 
 grams or otherwise. Then Eq. (53B) will furnish the stresses S from which the sections 
 of the several members are derived. The stresses S may also be obtained from a single 
 stress diagram by applying all redundants and external loads simultaneously to the prin- 
 cipal system. Temperature effect may be included in the redundants at the same time. 
 
 Another method of making a preliminary design would result from the use of a 
 Williot-Mohr displacement diagram as used in Art. 50. However, the deflection poly- 
 gons as found from the w loads are more easily obtained except for oblique loads P. 
 
 To illustrate the entire procedure of making a preliminary design for a structure 
 involving redundant conditions, the fixed arch in Art. 52 will now be investigated. 
 
 Example. The data for this arch are given in the example of Art. 52 and need not be 
 repeated here. 
 
 The EF c w a loads will be computed from Eq. (53E) making F C /F = 1 and neglecting 
 the web system. The moment M a = I and the lengths I, lever arms r and ordinates z 
 are taken from Tables 52A and 52c. 
 
 
198 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XI 
 
 TABLE 53A 
 COMPUTATION OF APPROXIMATE LOADS EF c w c 
 
 Pin 
 
 Point. 
 
 Chord. 
 
 I 
 Feet. 
 
 r 
 Feet. 
 
 1 
 
 ~2 
 
 Feet. 
 
 z 
 Feet, 
 
 EF c w a ^~ 
 
 zEF c w n 
 
 1 
 
 0-2 
 
 20.73 
 
 14.92 
 
 . 00449 
 
 0.00 
 
 0.0931 
 
 0.000 
 
 3 
 
 2-4 
 
 19.51 
 
 14.84 
 
 0.00454 
 
 9.99 
 
 0.0886 
 
 0.885 
 
 5 
 
 4-6 
 
 18.72 
 
 14.79 
 
 0.00457 
 
 17.03 
 
 0.0856 
 
 1.458 
 
 7 
 
 6-8 
 
 18.25 
 
 14.77 
 
 0.00458 
 
 21.53 
 
 0.0836 
 
 1.800 
 
 9 
 
 8-10 
 
 18.03 
 
 14.79 
 
 0.00457 
 
 23 . 69 
 
 0.0824 
 
 1.952 
 
 2 
 
 1-3 
 
 20.59 
 
 14.14 
 
 0.00500 
 
 22.31 
 
 0.1030 
 
 2.300 
 
 4 
 
 3-5 
 
 19.33 
 
 15.21 
 
 0.00432 
 
 29 . 83 
 
 0.0835 
 
 2.491 
 
 6 
 
 5-7 
 
 18.55 
 
 15.25 
 
 0.00430 
 
 34 . 99 
 
 . 0798 
 
 2.792 
 
 8 
 
 7-9 
 
 18.33 
 
 15.28 
 
 0.00428 
 
 38.01 
 
 . 0784 
 
 2.980 
 
 10 
 
 4(9-9') 
 
 9.00 
 
 15.31 
 
 0.00427 
 
 39.00 0.0384 
 
 1.498 
 
 
 
 
 
 
 
 0.8164 
 
 18.156 
 
 This makes z ' = 
 
 2zEF c w a 18.156 
 
 = 22.24 ft. from which the origin can be located 
 
 ZEF c w a 0.8164 
 
 and the ordinates y=zz ' be computed. The remaining w loads and pole distances 
 are then easily found. 
 
 TABLE 53 B 
 APPROXIMATE w LOADS AND POLE DISTANCES 
 
 
 
 Coordinates. 
 
 
 
 
 
 Pin 
 
 Point. 
 
 EF c Wa 
 
 X 
 
 y 
 
 EF c w b = 
 
 xEF c W a 
 
 EF c w c = 
 yEF c w a 
 
 xEF c Wb 
 
 yEF c Wc 
 
 
 
 Feet. 
 
 Feet. 
 
 
 
 
 
 1 
 
 0.0931 
 
 81 
 
 -22.24 
 
 7.541 
 
 -2.070 
 
 610.82 
 
 46: 04 
 
 2 
 
 0.1030 
 
 72 
 
 + 0.07 
 
 7.416 
 
 + 0.007 
 
 533.95 
 
 0.00 
 
 3 
 
 0.0886 
 
 63 
 
 -12.25 
 
 5.582 
 
 -1.085 
 
 351.67 
 
 13.29 
 
 4 
 
 0.0835 
 
 54 
 
 + 7.59 
 
 4.509 
 
 + 0.634 
 
 243.49 
 
 4.81 
 
 5 
 
 0.0856 
 
 45 
 
 - 5.21 
 
 3.852 
 
 -0.446 
 
 173.34 
 
 2.32 
 
 6 
 
 0.0798 
 
 36 
 
 + 12.75 
 
 2.873 
 
 + 1.017 
 
 103.43 
 
 12.97 
 
 7 
 
 0.0836 
 
 27 
 
 - 0.71 
 
 2.257 
 
 -0.059 
 
 60.94 
 
 0.04 
 
 8 
 
 0.0784 
 
 18 
 
 + 15.77 
 
 1.411 
 
 + 1.236 
 
 25.40 
 
 19.49 
 
 9 
 
 0.0824 
 
 9 
 
 + 1.45 
 
 0.742 
 
 + 0.120 
 
 6.68 
 
 0.17 
 
 4(10) 
 
 0.0384 
 
 
 
 + 16.76 
 
 0.000 
 
 + 0.644 
 
 0.00 
 
 10.79 
 
 Totals.. 
 
 0.8164 
 
 
 
 36 183 
 
 3 660 
 
 2109 72 
 
 109 92 
 
 
 H a = 1.633 
 
 
 
 
 + 3.658 
 
 H b = 4219. 4 
 
 tf c = 219.8 
 
 
 
 
 The three influence lines for X a , X b , and X c may now be drawn precisely as was done 
 in Figs. 52c, by using the values in Table 53B. 
 
ART. 53 DESIGN OF STATICALLY INDETERMINATE STRUCTURES 19i) 
 
 Comparing the ordinates found from these influence lines, Fig. 53A, with those pre- 
 viously obtained in Art. 52, it will be seen that the former are fairly close to the correct 
 values, but inclined to be a little large. 
 
 X,, INFLUENCE LINE - ORDtNATES ACTUAtTOSCALEOF LENGTHS- 
 
 ISO' 
 
 FlG. 53A. 
 
 The live load per truss per panel was stipulated at 30 kips and for an assumed dead 
 load of 20 kips the total load would be 50 kips. . 
 
 Using the ordinates from the influence lines in Fig. 53A and the total load of 50 kips 
 per truss per panel, the redundants X a , X b and X c are computed in the following Table 
 53c. For a symmetric loading X b =0 and need not be considered. 
 
200 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. XI 
 
 TABLE 53c 
 REDUNDANTS A' a AND X c FOR TOTAL LOADING 
 
 
 Ordinates. 
 
 
 Redundants. 
 
 
 Panel 
 
 
 Panel 
 
 
 
 
 
 
 
 Point. 
 
 
 
 
 
 
 
 
 
 f]a 
 
 TjC 
 
 
 Xa 
 
 Xc 
 
 
 
 Feet. 
 
 Feet. 
 
 Kips. 
 
 Kip- feet. 
 
 Kips. 
 
 
 2 
 
 8.4 
 
 0.086 
 
 50 
 
 420 
 
 4.3 
 
 X a = 2X3730 = 7460 k. ft. 
 
 4 
 
 14.5 
 
 0.298 
 
 50 
 
 725 
 
 14.9, 
 
 X c = 2X97.9 = 195.8 kips 
 
 6 
 
 19.0 
 
 0.520 
 
 50 
 
 950 
 
 26.0 
 
 A = iSF = 225 kips 
 
 8 
 
 21.5 
 
 0.684 
 
 50 
 
 1075 
 
 34.2 
 
 
 i(10) 
 
 22.4 
 
 0.740 
 
 25 
 
 560 
 
 18.5 
 
 
 Totals 
 
 
 
 225 
 
 3730 
 
 97.9 
 
 
 
 
 
 The redundants might be obtained for any position of the live load, but since the 
 chords are stressed to their maximum for total loading over the span, and since the web 
 system plays a rather unimportant part in structures of the class here considered, no 
 further investigation of stresses is warranted at this point. 
 
 The stresses for the total loading, including the redundants X a and X c applied as 
 external forces, are now found from a Maxwell stress diagram in Fig. 53 B, and from these 
 the preliminary cross-sections of the members are deduced. 
 
 From the sums in Table 53A, the value of z</=22.24 ft. This locates the pole 
 on the y axis. The_rigid disk, upon which the redundants act, may now be replaced by 
 two rigid members 10 and 00. For the symmetric total loading, X b =0, and hence the 
 external loads are X c applied at 0; and a moment X a applied anywhere to the disk; a 
 vertical end reaction A ; and the loads P acting at the several top chord panel points. 
 
 In order that the moment X a may be incorporated in the stress diagram, it must be 
 resolved into two equal and opposite forces acting on the rigid disk. Since these forces 
 may be applied anywhere on the disk, it is best to choose the lever arm 1/2, thus making 
 the forces V =2X a /l each, one acting upward at the support A and the other applied 
 downward at the pole 0. The forces V and X c , acting at 0, are then combined Jnto a 
 resultant Z and from this resultant the stresses in the rigid members 10 and 00 are 
 found, making it possible now to draw the stress diagram beginning at the reaction A. 
 The remainder of the work offers no particular difficulty. 
 
 It might be desirable to test out the resultant polygon with the aid of Eqs. (52s). 
 From these 2 =Z a /X c =38.10 ft. and the ordinate CD, Fig. 53s, determines the inter- 
 section of the resultants R, RI and R 2 in the point C. The closing line aibj. for symmetric 
 loading, is fully located when OD=z is given, hence the points a } and 61 are found on 
 the horizontal through D. The resultant RI is obtained by combining A and X e at a! 
 and R 2 is the resultant of B and X e at 6 t . The resultant R = 2P must be in equilibrium 
 with RI and R 2 (see Fig. 52A) , hence the three forces must intersect in a point C. 
 
 A further check is obtained by computing the ordinate CI>=Z#/4X C = 103.4 ft. 
 
 The stresses S are now tabulated and cross-sections F are determined in Table 53D, 
 using a rather low unit stress of say 9200 Ibs. per sq.in., instead of an allowable 15,000 
 
ART. 53 DESIGN OF STATICALLY INDETERMINATE STRUCTURES 201 
 
 Ibs. This then covers extra metal for joints, splices, latticing, etc., and for reduction 
 to net sections for tension members, and for l/r for columns. 
 
 o BOO KIM. 
 
 FIG. 53s. 
 
 TABLE 53D 
 PRELIMINARY STRESSES AND SECTIONS 
 
 Top Chord. 
 
 Bottom Chord. 
 
 Diagonals. 
 
 Diagonals. 
 
 Mem. 
 
 8 
 
 Kips. 
 
 F 
 
 Sq.in. 
 
 Mem. 
 
 S 
 Kips. 
 
 F 
 Sq.in. 
 
 Mem. 
 
 S 
 Kips. 
 
 F 
 Sq.in. 
 
 Mem. 
 
 S 
 Kips. 
 
 F 
 
 Sq.in. 
 
 0-2 
 2-4 
 4-6 
 6-8 
 8-10 
 
 - 920 
 - 1028 
 -1114 
 -1178 
 -1210 
 
 100 
 110 
 120 
 124 
 130 
 
 1-3 
 3-5 
 5-7 
 
 7-9 
 9-9' 
 
 748 
 846 
 930 
 970 
 988 
 
 82 
 
 94 
 100 
 104 
 128 
 
 0-1 
 2-3 
 4-5 
 
 6-7 
 8-9 
 
 - 70 
 
 + 166 
 + 160 
 + 117 
 + 95 
 
 8 
 18 
 18 
 14 
 10 
 
 1-2 
 3-1 
 5-6 
 
 7-8 
 9-10 
 
 140 
 - 90 
 - 55 
 
 + 10 
 
 + 48 
 
 14 
 10 
 6 
 6 
 6 
 
202 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XI 
 
 These cross-sections were used in the example of Art. o2, in making the complete 
 analysis. 
 
 Naturally a two-hinged arch or other structure involving one redundant only, 
 would be susceptible to the same method of design above outlined, though the labor would 
 be greatly lessened. 
 
 Foi an unsymmetric arch the angle /? would have to be determined- and the closing 
 line aibi would not be horizontal. 
 
 ART. 54. ON THE CHOICE OF THE REDUNDANT CONDITIONS 
 
 In the analysis of a structure involving redundancy, it becomes necessary to remove 
 the redundant conditions, whether external or internal, and thereby reduce the indeter- 
 minate system to a determinte principal system to which the redundant conditions are 
 applied along with the external loading. 
 
 The particular reactions or members best suited to represent the redundant con- 
 ditions designated by X a , Xj,, X c , etc., are those which reduce the given structure to the 
 simplest possible principal system. This will usually offer no difficulty. 
 
 However, many cases present themselves wherein it would be difficult to decide 
 which of several possible assumptions would represent the most judicious selection of the 
 redundants in the direction of simplifying the analysis. Therefore, a few suggestions 
 along these lines will not be out of place. 
 
 According to the general method just referred to, the elastic deformation of an 
 indeterminate structure subjected to loads P, will be exactly the same as that of its prin- 
 cipal system loaded with loads P, X a , X b , X c , etc. In each case the deformation is 
 entirely derived from the elastic changes Al in the lengths of the members of the principal 
 system. 
 
 The structural deformation is thus affected by the redundant conditions merely 
 to the extent of altering the stresses in the necessary members of the principal system. 
 
 Hence, for the same case of loading, the stresses S in the principal system are 
 always diminished by the redundant conditions provided temperature stresses and 
 abutment displacements are excluded. 
 
 Generally speaking, any member or reaction of an indeterminate structure may be 
 removed to produce the principal system so long as the latter still remains a stable, 
 determinate structure and does not become subject to infinitesimally small rotation, 
 a condition described in Art. 3, Fig. 3c. 
 
 The rule should be to select a principal system of the simplest possible form, always 
 avoiding composite structures, such as the three-hinged arch or a cantilever, whenever 
 a simple beam or truss could as well be used. 
 
 Solid web structures should always be transformed into externally determinate beams 
 by assigning the redundant conditions to the supports. 
 
 Framed structures, if externally indeterminate, should always be so transformed 
 as to remove the external conditions. The only exception to this rule might be a con- 
 tinuous girder wherein a top chord member over each intermediate pier might be treated 
 as a redundant member. 
 
ART. 55 DESIGN OF STATICALLY INDETERMINATE STRUCTURES 203 
 
 In the case of the fixed arch, always involving three redimdants, several assumptions 
 may be made. First removing one fixed end and thus reducing the arch to a simple 
 cantilever arm acted upon by a moment X a , a vertical force X b and a horizontal force X c . 
 Second, by removing two members adjacent to one end support and one member adjacent 
 to the other end support, thus forming a girder on two determinate supports acted upon 
 by three external forces replacing the three members thus removed. 
 
 Also three members may be removed at the crown converting the arch into two 
 cantilever arms with three external forces X a , X b , and X c applied to the end of each arm 
 to replace the redundant members. 
 
 Frequently all the redundant conditions (not exceeding three) may be applied at 
 one point as described in Art. 44. Then for certain assumed directions of the X's, the 
 work equations become exceedingly simple and afford excellent graphic solutions. The 
 best of these possible solutions was chosen in Art. 52, where a complete fixed arch problem 
 is solved. 
 
 Indeterminate structures, having a vertical axis of symmetry, will always have two 
 d's bearing double subscripts of like letters, which become equal. Thus d aa =d bb , which 
 condition greatly simplifies the determination of X a and X b , as illustrated in the case of 
 a girder over four supports with the two outer spans equal. See Fig. 43A. In this and 
 similar problems, when the above mentioned symmetry exists, the two influence areas 
 for X a and X b will also be equivalent but symmetrically placed. 
 
 When the redundant conditions are internal then the only way of deriving the princi- 
 pal system is to remove such redundant members and replace them by external forces A'. 
 
 Composite structures, or those composed of several determinate frames combined 
 into an indeterminate system, are best treated by assigning the redundant conditions 
 to the reactions and treat as for external redundancy. 
 
 ART. 55. SOLUTION OF THE GENERAL CASE OF REDUNDANCY 
 
 In the following it will be assumed that the preliminary design is completed and it 
 is now desired to make the final analysis for stresses due to any causes such as loads, 
 temperature and abutment displacements. 
 
 Problems of the general type are usually quite complicated, and since it is important 
 to know the stresses due to loads, to temperature changes and to yielding supports sepa- 
 rately, these should always be dealt with in this manner. This is not done merely as a 
 matter of convenience, but it is necessary to know the relation between the load and 
 temperature stresses as a criterion in judging the merits of any particular structure 
 under consideration. 
 
 In the general discussion which follows here, only the load effects will be included, 
 while the other matters will be taken up later. Hence the quantities 2#Jr and d t , S t , 
 etc., will all be neglected, assuming for the present that they are all zero. 
 
 To make the case perfectly general, no distinction will be made between external 
 and internal redundancy, hence any of the redundants may be assumed to belong to 
 either class. 
 
 For n redundant conditions, Eqs. (?A), offer a general solution for the stress S in any 
 
204 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. XI 
 
 member, the moment M about any point, the shear Q on any section or any of the 
 reactions R, all for the principal system only. The redundant conditions X, involved 
 in these equations, are treated as external forces applied to the principal system along 
 with the loads P and their reactions. These redundants may be found from Mohr's 
 work equations (?H) or (80) of which there will always be as many equations as there are 
 redundants, rendering the solution of the X's possible by solving for simultaneous values 
 of n unknowns X in n equations. 
 
 It is usually perferable to employ Eqs. (80) in terms of deflections which are 
 obtained from deflection polygons of the loaded chord, though in principle the solution 
 remains the same whether dealing with conventional stresses or conventional deflections. 
 
 In general then for n redundants, including effects due to temperature changes and 
 yielding supports, Eqs. (?A) give 
 
 (55A) 
 
 S=S S a X a SfrXb etc. S n X n 
 M=M -M a X a -M b X b - etc. -M n X n M t +M r 
 R=R - R a X a - R b X b - etc. -R n X n R t +R 
 Q=*Qo - Q a X a - Q b X b - etc. - Q n X n Q t +Q 
 
 wherein tne quantities S , M , R and Q are all linear functions of the externally applied 
 loads P acting on the principal system as a result of what is known as condition .X"=-0. 
 The quantities bearing subscripts a, b, c, etc., to n, are constants due to conventional 
 loadings X a = l, X b = l, X c = l, etc., to X n = l, while the subscript prefers to temperature 
 effects and the subscript r to effects produced by yielding supports. 
 
 The n redundants may be obtained from Eqs. (7n) in terms of the constant stresses 
 due to the conventional loadings or from Eqs. (So) in terms of certain deflection constants 
 obtained from the same conventional loadings. The latter equations are 
 
 d n = 
 
 B X a d na 
 
 etc. X^d an 
 
 e tc. X n d bn 
 
 etc. X n d cn 
 
 etc. X n d nn 
 
 +d c 
 
 (55B) 
 
 wherein the o's have the definitions given in Art. 8, and all those bearing double sub- 
 scripts of like letters are equal by Maxwell's law. Thus 
 
 =d c 
 
 =d etc. d = 
 
 etc. 
 
 meaning that the order of the subscripts is immaterial and that only half of the d's 
 need be determined, or if they are all determined, that they must check. 
 
 Hence, either one of the subscripts may be made to refer to the point of application 
 of a load, while the other subscript deals with the conventional loading or condition 
 X \. The terms RAr express the effect due to yielding supports and the d t quantities 
 
ART. 55 DESIGN OF STATICALLY INDETERMINATE STRUCTURES 205 
 
 express the effect of temperature changes, both of which will be placed equal to zero 
 for the present, though they will be considered in following articles. 
 
 The redundants X may represent indeterminate reaction conditions or they may 
 be stresses in redundant members, and in either case the displacements d a , d b , S c , etc., 
 o n are the displacements of the points of application of the respective redundants in the 
 directions of their lines of action. These values o may thus be expressed in terms of 
 the redundants themselves in accordance with Eqs. (7j). When the point of application 
 of a certain redundant is rigid, then its d =0. 
 
 For each of the n redundant conditions there will be one work equation of the form 
 of Eqs. (55B) involving deflections due to conventional loadings. These together with 
 the case of actual loading due to loads P, will constitute in all n + l cases of loading on 
 the principal system to solve for one position of a moving train of loads. 
 
 The aim, in all practical problems of this nature, consists in representing all the 
 required stresses or other functions by influence lines, thus requiring n -f 1 influence line-* 
 to determine each such stress or function. However, the n influence lines for the n redun- 
 dants will be the same for all cases and all stresses or functions of the same structure, 
 while the influence line for the load effect must be separately found for each stress, 
 moment, shear or reaction. 
 
 For influence lines the applied load becomes unity and the functions ^P m d m = 1 d m . 
 Each of the n redundants will furnish a deflection polygon for condition X = l, drawn 
 for the loaded chord of the principal system. The n deflection polygons will then 
 furnish all the conventional deflections in Eqs. (55s) for a load P = l. Also, since the 
 subscripts may be interchanged, one such deflection polgyon drawn for X A = l will fur- 
 nish all the double subscript bearing o's of the first equation. 
 
 Thus as a matter of convenience, all of the d coefficients in a single equation 
 can be determined from one deflection polygon drawn for the loaded chord of the prin- 
 cipal system. The same might also be accomplished by drawing a Williot-Mohr dis- 
 placement diagram for the principal system, though this would usually be more laborious. 
 
 Therefore, the n Eqs. (55s) can be solved successively with the aid of n deflection 
 polygons, in accordance with the following form with transposed subscripts and for a 
 
 single load P = l. 
 
 X I 
 
 y- 7 
 
 X2 
 
 X b O hc -X c d cc . . . - X ^nc = 
 
 , . . . (5oc) 
 
 where the values of d a , 3 b , S c , etc., were obtained from Eqs. (7j) and may be chosen to 
 represent anv of the possible cases of redundancy, as changes in the lengths of redundant 
 members elastic displacements of redundant supports or angular changes between pairs 
 
206 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XI 
 
 of lines according to Art. 9. When the redundants are immovable reactions, then these 
 S's become zero. 
 
 The d coefficients in Eqs. (55c) can thus be obtained from n deflection polygons 
 drawn as above described and the n equations solved for simultaneous values of the X's 
 thus furnish the values for a complete solution of Eqs. (OOA). 
 
 Eqs. (55c) are written for a single load P = l intended specifically for use with influence 
 lines, which are most valuable when dealing with concentrated load systems. 
 
 Examples involving dead loads only, or when the live load cannot occupy more than 
 one or two positions, could be solved advantageously without resorting to influence lines 
 according to the very interesting example of a lock gate in Chapter XIV. 
 
 For solid web structures and masonry arches, all the above is applicable, remember- 
 ing that in these cases the redundants must all be external. As soon as the X's are found 
 the stresses on any section are readily ascertained. 
 
 The solution of the n simultaneous equations is a matter of considerable labor and 
 may offer some difficulty. The method best suited to problems of the kind here con- 
 sidered is given in connection with the lock gate problem referred to above. See Chap- 
 ter XIV, Art. 68. 
 
 ART. 66. EFFECT OF TEMPERATURE ON INDETERMINATE STRUCTURES 
 
 Determinate systems are not materially affected by temperature changes. But it 
 is not correct to say that the temperature stresses are zero, because any structure which, 
 in consequence of changes in temperature, undergoes such deformations as to change 
 the positions of the points of application of the external forces must thereby create some 
 stresses. These will usually be small and are entirely negligible except possibly in very 
 flat suspension systems or very flat three-hinged arches where the horizontal pull or thrust 
 is a function of the middle ordinate. 
 
 For the case of indeterminate structures, wherein the redundant conditions ares 
 direct functions of elastic deformations, the general rule implies that stresses always 
 accompany changes in the lengths of structural dimensions, no matter what external 
 cause may have produced such changes. 
 
 All temperature changes in externally indeterminate structures will affect the reac- 
 tions and these in turn set up temperature stresses in the members. The only exception 
 to this case is a continuous girder with supports on the same level and subjected to abso- 
 lutely uniform temperature throughout. When such a structure is unequally heated 
 then excessive temperature stresses may result, as shown in the examples of Arts. 47; 
 and 48. 
 
 On the other hand where the redundancy is entirely internal, no temperature stresses 
 are produced so long as the whole structure retains a uniform temperature throughout. 
 When this uniformity does not exist, then temperature stresses are created, though the 
 reactions may or may not be materially affected. 
 
 Hence, in all cases of redundancy, it will be necessary to investigate the question 
 of temperature stresses in a very thorough manner, as in many instances these may 
 assume dangerous proportions. 
 
ART. so DESIGN OF STATICALLY INDETERMINATE STRUCTURES 207 
 
 For this reason also, statically determinate structures should always receive the 
 preference, other considerations being nearly equal. 
 
 Internal redundancy is less objectionable than external, and according to the best 
 modern engineering experience, no design should be made to include more than one 
 redundant condition of any kind. This is especially true of steel structures, though to 
 a lesser degree applicable to masonry arches of short spans where the poor conductivity 
 of masonry acts as a protective agency against excessive temperature deformations. 
 
 For long-span masonry arches this consideration assumes greater importance. Even 
 though monumental structures of this class have been built and are regarded with pride 
 and admiration from an esthetic standpoint, they cannot be accepted as representative 
 of the best practice when viewed from the point of modern and progressive engineering. 
 
 The general effect of temperature changes on indeterminate systems is thus to pro- 
 duce deformations and resultant stresses of the same kind as those created by externally 
 applied loads in accordance with Eqs. (55A). 
 
 Since the temperature effects may be plus or minus in character, depending on the 
 existing conditions, it is always possible to increase the stresses due to the loading by 
 choosing appropriate changes in temperature. For this reason the temperature effect 
 is applied with a plus sign in Eqs. (55A), meaning thereby that the function, whether posi- 
 tive or negative, suffers a numerical increase. Of course some assumptions will produce 
 a decrease, but these are not vital to the problem and require no consideration. 
 
 It is always desirable to determine the temperature stresses apart from all other 
 influences so as to furnish a clear conception of their relative importance to the load 
 stresses. This then offers a criterion by which to judge the feasibility of a structure 
 involving redundancy. 
 
 The temperature stresses are determined from Eqs. (7n) and (80) by dropping out 
 all terms depending on P m , S and R, thus reducing these equations to the following 
 forms, covering all members of the principal system: 
 
 d a =ea a = ^S a stt-X at ^S^p -X bt 2S a S b p . . . etc. 1 
 
 k ... (56A) 
 d b = eil b = '2S b etl-X at 2SJS b p-X bt Z,Szp . . . etc. J 
 
 or from Eqs. (80) 
 
 8a=X a tpa=8 at -X a td aa -X b t8 ab etc. ] 
 
 \ (o6B) 
 
 8 b =X bt p b =d bt -X a t8 btt -X bt 8 bb ... etc. J 
 
 In both Eqs. (56A) and (56s) the displacements d a and d b become zero for external 
 redundancy with immovable supports. 
 
 Accordingly the redundant conditions X at , X bt , etc., may be found from either set of 
 the above equations, depending on the method used in finding the load effects from Eqs. 
 (7n), or (80) since both these and Eqs. (56A) and (56u) involve the same constants. 
 
 The stresses S t are then easily found from a Maxwell stress diagram drawn for the 
 principal system with the forces X at , X bt , etc., applied as external forces. See also examples 
 in Arts. 47, 48, 49 and 50. 
 
208 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XI 
 
 Regarding the assumptions which may be made as a basis for computing temperature 
 stresses, the following conclusions are taken from some experiments on steel arches at 
 Lyons, France, and given in An. d. ponts et chaus., 1893, II, p. 438-444. 
 
 For an air temperature in the sun of 5 F. higher than the shade temperature, when 
 the latter was 90 F., the steel had an average temperature of 115 F., while the parts 
 exposed directly to the sun were heated to about 130 F. and the shaded portions indicated 
 about 104 F. At the same place the coldest winter temperature was about 15 F. 
 
 The steel was thus subjected to a mean range of 15 to +115 = 130 F. giving 
 an average of +65 F. Therefore, such a structure should be designed for a mean 
 temperature of 65 F., allowing for uniform changes of 65 F. from this mean. 
 
 The difference between maximum and minimum simultaneous temperatures in 
 the steel, amounting to 26 F. in these experiments, may cause very serious stresses in 
 certain structures like fixed arches and continuous girders over several supports. In 
 the latter case the entire top chord would elongate relatively to the bottom chord and 
 thus set up an arching effect, relieving the intermediate supports and increasing the 
 two end supports, while a uniform change in temperature would produce no stresses. 
 
 Structures over a single span would not be so seriously stressed for unequal temper- 
 atures in the two chords. In fact this assumption might work to advantage in the 
 case of arches, though similar conditions for a clear cold day might prove more severe. 
 
 The painting of steel structures of the indeterminate class thus assumes considerable 
 importance, since light colors will obviously keep down the temperature while black 
 paint will absorb heat. 
 
 Masonry structures are not so severely distorted by temperature changes, but the 
 elasticity of masonry being proportionately lower, the temperature stresses may prove 
 equally dangerous. 
 
 ART. 57. EFFECT OF SHOP LENGTHS AND ABUTMENT DISPLACEMENTS 
 
 Every structure is designed for a certain geometric figure for a condition of no 
 stress. This is the figure which the structure should present when at a mean temperature 
 and when carrying no loads of any kind. Naturally the unavoidable errors in the shop 
 lengths of the members as built and the small inaccuracies in the joints and in the loca- 
 tion of the supports during erection, preclude the practical possibility of accomplishing 
 this end. 
 
 For structures of the determinate class this will have no significance, but for indeter- 
 minate types, these errors introduce initial stresses which may assume momentous 
 proportions. 
 
 To avoid such stresses requires the utmost care during construction and erection, 
 and even then only a partially satisfactory outcome can be expected. 
 
 The difficulties attending the erection of a structure so that no member will be 
 stressed prior to the introduction of the closing link, at the proper temperature, are 
 known to every practical bridge man. 
 
 In any event the final member should be inserted at the calculated mean tem- 
 perature. If this cannot be accomplished, then the exact length which this member 
 
ART. 57 DESIGN OF STATICALLY INDETERMINATE STRUCTURES 209 
 
 should have must be ascertained by measuring the space which is actually available 
 when the structure is all in place and under no stress. Then applying proper tem- 
 perature corrections to the entire structure, the required length of the final member must 
 be ascertained before attempting to force the latter into its place. 
 
 When the end supports are not correctly placed or the entire structure has an eror 
 in length, then all the values S , S a , etc., and the redundant conditions will differ from their 
 computed values. If such displacements are in the same direction with a redundant 
 X, then the stress X is affected thereby, otherwise the principal system only receives 
 the additional stress. 
 
 Should a redundant member possess an erroneous length, then all the members 
 will receive certain initial stresses, while errors in the lengths of principal members would 
 change somewhat the geometric figure, but could not effect the stresses S , S a , etc., of the 
 principal system. 
 
 In dealing with abutment displacements it is necessary to distinguish between elastic 
 changes which the- supports may undergo as a result of loading, and permanent settlement. 
 
 An example of elastic displacement is given in Fig. 45A, where the supporting column 
 CD will be shortened by an amount d b , which can be determined and then be included in 
 the computations as for any other member of the structure. 
 
 However, should the pier at C undergo permanent settlement, or should one abut- 
 ment, as B, be pushed out horizontally or settle vertically, then serious difficulties 
 would arise unless these displacements could be determined beforehand, which is usually 
 quite impossible. 
 
 In either case the stresses occasioned by such displacements only, may be found 
 as follows, provided the displacements are known or can be estimated with reasonable 
 accuracy. 
 
 If, in Eqs. (55s), the terms involving the temperature effects be dropped, then for 
 internal redundancy, the displacements d a , d b , etc., are evaluated from Eqs. (7j), and the 
 work of the reactions is found as shown in the derivation of Eqs. (?E). 
 
 When the redundancy is purely external then the Eqs. (55s) again apply by treat- 
 ing the reactions R as the reactions of the principal system and evaluating the elastic 
 displacements d a , d b , etc., and the Ar for each reaction R, using such considerations as 
 given in Art. 12, or estimating these displacements as elastic yielding in the masonry 
 supports, etc. See also examples in Articles 49 and 50. 
 
 In any case the combined stresses from all causes are given by Eqs. (55A) and no 
 further comment is necessary here except to emphasize the inadvisability of adopting 
 externally indeterminate structures whenever immovable supports are not available. Even 
 for bed rock foundations, this will depend largely on the depth to which the masonry 
 must be carried and also on the quality of masonry used. 
 
 When steel towers or pendulum piers are employed to support an indeterminate 
 structure, then the elastic deformation of such supports must be considered in comput- 
 ing the structural stresses. 
 
CHAPTER XII 
 STRESSES IN STATICALLY DETERMINATE STRUCTURES 
 
 ART. 58. DEAD LOAD STRESSES 
 
 (a) General considerations. The purpose of taking up the question of stresses ir 
 statically determinate structures is not with any intention of covering this subject exhaust- 
 ively, but merely to present in the briefest possible space the methods which are besl 
 suited to the analysis of all ordinary structures. 
 
 It was not deemed advisable to take up the present chapter before having treated 
 influence lines and developing the general criteria for the position of loads for maximum 
 and minimum stresses in Chapter IV. The fundamental conceptions there presented 
 are very material to a broader understanding of determinate structures and hence the 
 present order of subjects was considered justifiable. 
 
 Since it is desired to treat methods of analysis and not types of structures, only 
 the general case of non-parallel chords will be considered. Whenever a structure is sim- 
 plified by introducing parallel chords and otherwise simple relations between its dimen- 
 sions, then naturally the analysis becomes less complicated, until it might be said that 
 the problem is reduced to a mere application of arithmetic. The so-called algebraic 
 methods are, therefore, passed over without further consideration. 
 
 Since the dead loads are fixed in position and magnitude, the stresses produced by 
 them in any structure must be absolutely invariable. The live loads, however, owing 
 to their shifting position, may produce a variety of conditions and stresses which may 
 tend to increase or to diminish dead load stresses. 
 
 Therefore, every member may be said to be subjected to a maximum and a minimum 
 stress corresponding to the peculiar or critical positions of the live loads. These critical 
 positions are always determined from certain tests or criteria, differing for different 
 members of the same structure according to the principles discussed in Articles 20, 23 
 and 24. 
 
 It was found there that for the chord members the maximum stresses are 
 produced for the case of maximum loading over the entire span, while the minimum 
 stresses in these members result from the minimum total load, which is the dead 
 load. 
 
 This condition is very different for the web members, where the dead load rarely 
 produces minimum stresses because different positions of partial live load may pro- 
 duce live load stresses of opposite signs. When these are combined with the dead load 
 
 210 
 
ART. 58 STRESSES IX STATICALLY DETERMINATE STRUCTURES 
 
 211 
 
 stresses to obtain total stresses, then the minimum total stresses are less than the dead 
 load stresses, while the maximum total stresses exceed the dead load stresses. 
 
 When a web member is designed to take only one kind of stress like tension, and this 
 stress is reversed by the live loads, then the panel containing such member must be 
 supplied with a counter web member or the original web member must be differently 
 designed so as to resist both tension and compression. This latter is considered the 
 best modern practice. 
 
 Hence, in order that the relation between the dead and live load stresses may be 
 clearly brought out in the analysis, it is always necessary to determine these stresses 
 separately. Also, the methods which apply to the solution of dead load stresses are usually 
 not the best adapted to finding the live load stresses, which presents another reason for 
 dealing with the general subject under separate headings. 
 
 However, each method given will be applied directly to all the members of a struc- 
 ture, showing the application both to chords and web members before proceeding 
 further. 
 
 (b) Aug. Hitter's methods of moments (1860). This method is based on the gen- 
 eral theorem of moments. Accordingly for all external forces in the same plane and 
 constituting a system in equilibrium, the sum of the static moments, about any point 
 in this plane, must equal zero. 
 
 If now such a set of forces or loads be applied to a frame, Fig. 58A, and these loads, 
 together with the two end reactions A and B, form a system in equilibrium, then if the 
 
 FIG. 58A. 
 
 frame be cut by a section tt, the stresses in the members cut, if acting as external forces, 
 must maintain equilibrium of all the external forces on either side of the section. If, 
 further, such planes can be passed which do not cut more than three members of the 
 frame, then the stresses in the members cut may be determined one at a time, by taking 
 as the center of moments the intersection of two of the members. The moments of the 
 two intersecting members thus become zero. 
 
 The following designations are used: 
 
 For member U the center of moments is n and the ever arm is r n ; 
 
 For member L the center of moments is m and the lever arm is r m ; 
 
 For member D the center of moments is i and the lever arm is r ; 
 
 41so calling M the sum of the moments of the external forces A, P, and P 2 to the 
 
212 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XII 
 
 left of the section it and using subscripts n, m and i to designate the centers about which 
 the moments are taken, then for positive moments in a clockwise direction 
 
 Ur n +M n =Q or U = ~ 
 
 " 
 
 -Lr m +M m =Q or 
 
 M. 
 
 -Dri-Mi =0 or Z>=- -1 
 
 (5SA) 
 
 where a positive stress indicates tension and a negative stress indicates compression. 
 Thus a top chord member is always in compression and a bottom chord member is always 
 in tension, while a web member may have either stress depending on its inclination with 
 respect to its center of moments. 
 
 Hence, the stress S, in any member of a determinate frame, may always be expressed 
 in terms of a moment M and its related lever arm r, by 
 
 (58B) 
 
 which is Ritter's fundamental moment equation so extensively used in the previous 
 chapters. 
 
 The centers of moments for the chord members are thus seen to be located opposite 
 the chord in question, while for a web member the center of moments may be anywhere, 
 depending upon the relative inclinations of the two chords composing the panel. The 
 lever arms should be scaled from a large scale drawing or be computed. 
 
 When the three members cut by a section intersect in the same point, the method 
 fails. Also, when the chords are parallel the lever arm for the diagonal becomes infinite. 
 Hence the stress in such a diagonal cannot be found by a direct application of Eq. 
 1.58B) without some modification. 
 
 For the case of chords which are parallel or nearly so, the stress in a, web member 
 may be found by first computing the stresses in the chords and then finding the web 
 stresses. This is done by choosing for the center of moments for a web member, any 
 convenient point of one chord and then writing a moment equation, including the other 
 chord of the panel cut among the external forces. 
 
 However, there is a simpler method of finding the web stresses in structures with 
 parallel chords and that is from the shear in the panel cut. 
 
 Thus in the center panel mq the chords are parallel and the diagonal pq is cut by 
 a section ft'. The lever arm for pq would be infinite. 
 
 Since the sum of the moments of the external forces to the left of the section and of 
 the three members cut must be zero, therefore the sum of the vertical components of 
 these forces and stresses to the left of the section must likewise be equal to zero. 
 
 The sum of the vertical components to the left of the section Ft' is called the shear Q. 
 
ART. 58 STRESSES IN STATICALLY DETERMINATE STRUCTURES 213 
 
 It is equal to the end reaction A minus the sum of the panel loads between A and the section. 
 Thus 
 
 Q=A-*lP, .......... (58c) 
 
 where the summation covers only the panel loads to the left of the section and the shear 
 is taken positive upward on the left of the section. 
 
 The chords, being horizontal, cannot have a vertical component, hence the only 
 stress in any of the three members cut, which can have a vertical component, is the stress 
 in the diagonal pq=D. The vertical component of D l is then DI cos 0, and the sum 
 of the vertical components of all forces and members cut on the left of the section W 
 becomes 
 
 or 
 
 When DI is vertical, then 0=0 and the stress becomes equal to the shear. 
 
 The above moment equation also admits of a graphic solution which will not be 
 considered here. 
 
 (c) The method of stress diagrams. The first description of these diagrams seems 
 to have come from Bow, and in 1864 Clerk-Maxwell published a paper " On Reciprocal 
 Figures and Diagrams of Forces " in which he presents a scientific treatment of the sub- 
 ject. Cremona, in 1872, discussed the geometric properties of stress diagrams, showing 
 their general usefulness in connection with graphostatics. 
 
 English and American writers, therefore, call such diagrams "Maxwell stress diagrams," 
 while in Germany and France they bear Cremona's name. 
 
 The Maxwell stress diagram, so called in the present work, serves a most valuable 
 purpose in the graphic analysis of all determinate frames, and is generally applicable 
 to all cases which are susceptible to treatment by Ritter's moment method. 
 
 A peculiar relationship exists between a frame and its stress diagram by which each 
 member of the frame is parallel to a line of the diagram and each pin point of the 
 frame is represented by a force polygon in the diagram. It is thus equally possible to 
 construct a stress diagram for a given frame or to construct a frame from a given stress 
 diagram. Hence, the term " reciprocal figures " used by Maxwell. 
 
 The closed force or funicular polygon, which constitutes the basis for the Maxwell 
 diagram, was known to Stevin, 1608, and Varignon, 1725, and marks the beginning of 
 graphics. Such a polygon may be drawn for any set of forces in equilibrium. 
 
 Since all the forces meeting in a pin point must constitute a system in equilibrium, a 
 closed force polygon may be drawn for each such point. Hence a Maxwell diagram is 
 merely a succession of closed force polygons drawn for all the pin points of a frame. 
 
 For any given frame, the directions of all forces and members are given or must be 
 assumed before proceeding to an analysis of stresses. Also, if the frame constitutes a 
 system in equilibrium, then the externally applied loads must be in static equilibrium 
 with the supports or reactions, and all these in turn must be in equilibrium with the 
 internal stresses in the members. 
 
KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XII 
 
 Any pin point in equilibrium and acted upon by any number of forces of known direc- 
 tions, may then be represented in the force polygon by a succession of forces and stresses 
 respectively equal and parallel to the forces and stresses meeting in that pin point. 
 Since all directions are known, two magnitudes may be found by inserting the unknown 
 members in such a way as to close the force polygon. 
 
 This is the fundamental principle of the Maxwell diagram. It is illustrated in 
 Fig. 58s, where the force P and stresses S\ and 82 are known and the stresses S 3 and S 4 
 are of unknown magnitudes, but of given positions. All the forces meet in the point 
 A and are supposed to be in equilibrium. 
 
 The method of drawing the stress diagram and nomenclature used in the figure is 
 precisely the same in all cases and affords an easy way of deciding the direction of 
 action of each unknown force. 
 
 The clockwise arrow indicates the order in which the forces are assembled in the 
 stress diagram. Passing around the point A in this direction the first known force reached 
 is P. Hence the letters a, b, c, d, e are supplied as shown in the angles between the 
 
 Forces Acting on A. 
 
 FlG. 58B. 
 
 Stress Diagram. 
 
 forces, such that the force P is included between a and 6. In speaking of the force ab 
 in the stress diagram, we mean a force equal and parallel to the force P and acting 
 in the given direction from a to 6 in the stress diagram. 
 
 Observing this designation, the other known forces are added in their proper order 
 in the stress diagram and made to act in the given directions a-b-c-d as found by going 
 around the point A in a clockwise direction. The stress diagram from a to d is thus 
 obtained and may now be closed by drawing a line de' \ S% through d and another line 
 ae" || 84 through a. The intersection e thus found completes the force polygon and 
 determines the stresses 3 and S both in magnitude and direction of action. Thus the 
 arrows around the force polygon must be in the same direction as indicated by the 
 initial given force P. Accordingly the force 83 acts in the direction from d to e and the 
 force S^ acts in the direction from e to a. All forces in the force polygon are laid off to 
 a certain scale by which the unknown forces are finally determined. A force or stress 
 acting away from the pin point exerts a positive or tensile stress. 
 
 It is thus seen that the unknown stresses in two of the members meeting in any 
 pin point may be determined by means of a closed force polygon. However, if a given 
 frame presents no pin point involving only two unknown forces, then the method can- 
 
ART. 58 STRESSES IN STATICALLY DETERMINATE STRUCTURES 
 
 215 
 
 not be applied except by first finding the forces in excess of two, by some other means. 
 When there are three unknown forces acting on the point, and there are no redundant 
 conditions involved, then Ritter's moment method will always furnish the necessary 
 solution for one of the unknowns, prior to drawing the Maxwell diagram. 
 
 The forces in Fig. 58B were arbitrarily assembled in a clockwise direction. A 
 counter clockwise direction might have been chosen with equal right, though the stress 
 diagram would then have occupied a symmetric position with respect to the present 
 
 case. 
 
 To illustrate the method, a simple truss, Fig. 58c, is used, 
 are each 10 kips and the reactions A =B =40 kips. 
 
 k 
 
 -43.0 
 
 B-40K. 
 
 The panel loads P 
 
 FIG. 5Sc. 
 
 The reactions must always be computed in order to supply the forces which are 
 necessary to establish a system of external forces in absolute equihbrium. The 
 diagram is then lettered and the direction of assembling the forces is chosen so as to locate 
 the stress diagram properly on the paper. 
 
 A pin point embracing only two unknown members must be fleeted for a point 
 beginning and by inspection it is seen that either abutmentsaUsfies this condition 
 theTress diagram iscommenced by drawingjhe triangle * inwhich the vertica upward 
 reaction is known and the two forces cb and tajire found, lollo^ -ound 
 triangle in the direction a-c-b, it is seen that the force cb acts toward the 
 
216 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XII 
 
 The stress in ba is found 
 
 produces negative or compressive stress in the member cb. 
 to act away from A , and is thus positive or tensile. 
 
 The construction of the diagram is now continued by assembling the forces around 
 pin point 1, beginning with the first known member be, then adding the known force 
 cg = Pi and finally closing the figure by drawing gf and/6, which are the two unknowns 
 for point 1 . 
 
 The members ub and bf being_known_the closed force polygon for pin point 2 may 
 next be drawn to find the stresses fe and ed. 
 
 -*TTZ\__ \_LI_U. _v- ' - 
 
 FIG. 58o. 
 
 The process is continued, taking the pin points in the following order: A-l-2-3-4-5-6- 
 7-8. As a final check rp= B. 
 
 AH the stresses are then scaled from the stress diagram and written on the truss 
 diagram with proper signs, for compression and + for tension. 
 
 The present example shows a truss in which the chord stresses are nearly all equal 
 and the web stresses are small. It is similar to the Pauli truss and is economical in design. 
 
 A very interesting Maxwell diagram is presented in Fig. 53e, where some of the 
 external loading consists of a moment. 
 
 It frequently happens that loads are not applied directly to the pin points, in which 
 
ART. 58 STRESSES IN STATICALLY DETERMINATE STRUCTURES 
 
 217 
 
 case certain load concentrations must be effected before proceeding to draw a stress- 
 diagram. The members so loaded will usually receive a combination of direct stress and 
 bending as illustrated by the member AC, Fig. 58D, representing a portion of a roof truss. 
 The parallel loads PI to P 5 , acting on the member AC, may be combined (graphically 
 or analytically) into a resultant R of kncfwn position and magnitude, and R may then 
 be resolved into the components R \ and R%, constituting the pin point concentrations 
 at A and C respectively. 
 
 FIG. 58e. 
 
 After all loads have been thus concentrated on the several pin points of the struc- 
 ture the total reactions may be computed in the usual manner, taking into account only 
 the loads R,, R 2 , etc., which now replace the loads P, and then the stress diagram is 
 drawn in the usual way and the direct stresses in the members are obtained. 
 
 The reactions R, and R 2 are the same as for a span d, taken perpendicular to the 
 direction of the forces, and the bending moment M, for any point k distant x from A , 
 may be found from the equilibrium polygon or by computation. 
 
 For a direct stress - S, as obtained from the stress diagram, the total thrust m the 
 member becomes N=-S-(Ri -ft -ft -ft) cos a. and the moment M-- 
 
218 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP, xil 
 
 r]H, where H is the pole distance measured to the scale of forces and T? is 
 the ordinate of the equilibrium polygon measured to the scale of lengths. 
 
 The unit stresses on the extreme fibers of the column A C then become by Eq. (49ivi) 
 
 f _, 
 
 J p j , 
 
 where y is the distance of the extreme fiber from the gravity axis of the section. 
 
 The dead weight of inclined or horizontal members would be considered in this 
 manner, only that for uniform loads the treatment is simplified. 
 
 The example given in Fig. 58E belongs to the class previously pointed out, in which 
 not less than three unknown members meet in every pin point, and hence a Maxwell 
 diagram cannot be constructed without first computing one member by Ritter's method 
 of moments. 
 
 The resultant R is found as in the previous figure, and supposing the support at .4 
 to be a roller bearing, the reaction at R\ must be vertical and RZ must pass through B 
 and d. Hence the reactions of known directions may be found from the force polygon 
 by resolving R into R \ and R%. 
 
 The stress in the member AB is then found by passing a section tt', cutting only three 
 members. Then with C as center of moments Eq. (58s) gives 
 
 AB-H-(Rr-R, l Y-(Rr-R r V-- R (r- r ' 
 C" Kl 2)Ji ( K 2]h J\ r 2 
 
 RI and H being now known the Maxwell diagram can be commenced for the pin point 
 A and continued throughout the truss. 
 
 For vertical loads the solution may be conducted analytically. 
 
 ART. 69. LIVE LOAD STRESSES 
 
 (a) The critical positions of a train of moving loads to produce maximum and minimum 
 stresses in any member of a given determinate structure must be known prior to applying 
 any method for finding the stresses themselves. This question was fully discussed in 
 Arts. 20, 23 and 24, and in the General Considerations of Art. 58. 
 
 All the methods for the analysis of stresses which follow here presuppose a knowl- 
 edge of the criteria for the positions of loads. The reader is referred to the articles just 
 mentioned without repeating this discussion here. 
 
 (b) The method of influence lines, fully treated in Chapters IV and V, may be men- 
 tioned as the most universal, answering as it does all questions relating to criteria for 
 positions of loads and magnitudes of stresses for any determinate structure. 
 
 The method requires no further explanation here except to point out the types of 
 structures which warrant its application. 
 
 In general, the more complicated the geometric figure of a structure, the greater 
 the advisability of employing the method of influence lines, since the geometric relations 
 of the truss dimensions are thereby expressed^, independently of the loads. 
 
ART. 59 STRESSES IN STATICALLY DETERMINATE STRUCTURES 219 
 
 Hence it would scarcely be advisable to apply influence lines to any truss with parallel 
 chords except in cases of complicated cantilever systems. On the other hand a saving 
 of labor might be expected in analyzing structures in which either or both chords are 
 curved. The method certainly offers obvious advantages in all cases illustrated in 
 Chapter V, especially when concentrated loads are employed. 
 
 It may be added that any structure which does not warrant the accurate analysis 
 by concentrated load systems should not be dignified to the extent of being called a 
 modern bridge. Nor should anyone not familiar with these more exact methods be 
 intrusted with the design of bridges. 
 
 It should also be emphasized that all computations of stresses should err decidedly 
 on the side of safety, since the secondary stresses produced by the friction on pin-con- 
 nected joints are frequently a very considerable quantity, to say nothing of the failure 
 to analyze properly the effect of the riveted connections. 
 
 (c) Discussion of methods in common use. In general, all methods involve the 
 solution of the following distinct problems: To find certain moments, shears and end 
 reactions resulting from certain critical positions of the train of loads and then to deter- 
 mine the maximum and mimimum stresses in a particular member in terms of these moments 
 or shears. 
 
 Both analytic and graphic solutions have been proposed to solve all of these prob- 
 lems in their various phases. Some are applicable only to trusses with parallel chords while 
 others are more general. Still others are used when the position of the train is chosen 
 so as to avoid the complications arising from the exact loadings required by the criteria 
 for maximum and minimum stresses in a certain member. 
 
 It would be outside the province of the present chapter to present all of these 
 methods in detail, though many of them are exceedingly interesting and ingenious. The 
 author would offer the general criticism that the methods in common use are too special 
 in their application, and when a more general problem is encountered no one method 
 will suffice, but several methods must be combined, thus necessitating an intimate knowl- 
 edge of many limited methods to accomplish a complete analysis. 
 
 The following method is an attempt to determine the live load stresses in each of 
 the members of any simple truss by applying the same process throughout. 
 
 (d) The author's method of determining live load stresses is based on Ritter's com- 
 prehensive and universal method of moments. The moment of the live loads on one 
 side of the section taken about the center of moments for any particular member is 
 derived from the principles of the sum A line, see Fig. 22fi. 
 
 Thus the stress in any member of any determinate truss is given by Eq. (58s) as 
 
 except for the web members of a truss with parallel chords where the stress is derived 
 from the shear according to Eq. (58o) as 
 
 ( 59B ) 
 
220 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XII 
 
 The values of M and Q for any position of a live load will now be derived and 
 evaluated first from the sum A line, and then from a single tabulation of moments. 
 
 The critical position of the loads for any particular effect on a certain member is 
 supposed to have been determined in accordance with Arts. 20, 23 and 24. 
 
 M 1 P il 11 1 
 
 1 
 
 *B 
 
 , *m - 
 
 1 lam - ., 
 
 ... | - _, 
 
 FIG. 59A. 
 
 Referring to Fig. 59A and calling MB the sum of the moments of all loads on the span 
 about the reaction B and M m the sum of the moments about any point m; also calling 
 Q m the vertical shear at the point m; then for loads 1 to q } covering a distance xq=ljr (l 
 
 !*P; (59c) 
 
 (59D) 
 
 (59E) 
 (59r) 
 
 or 
 
 M m =l am (A- 
 
 lam 
 
 (59c) 
 
 wherein A am = 3. Pe/l am may be defined as the end reaction at A of the loads 1 to m 
 covering the distance x m of a simple span Z am . 
 
 The values A and ^4 aw may be obtained from a reaction summation influence line, 
 or sum A line, drawn for the reaction A and span I, according to Art. 22, Figs. 22A and 
 22s. The ordinates of the sum A line may be computed from Eq. (59o), or constructed 
 graphically by the method given in Art. 22. 
 
 Such a sum A line is shown in Fig. 59s, drawn for a span of 200 feet and using a train 
 of Cooper's EGO loading, consisting of two locomotives followed by a uniform load of 3000 
 Ibs. per linear foot per rail. 
 
 The ordinate y x under the first load PI at x, represents the end reaction A for 
 the position of loads indicated, when the span A B is loaded for a distance 61 from B. 
 
 The reaction A am , for the same train covering the distance x m from m on the 
 span l am> may be obtained from, the same sum A line as follows: Had the sum A line 
 
ART. 59 STRESSES IN STATICALLY DETERMINATE STRUCTURES 
 
 221 
 
 been drawn for a span l am then the ordinate at x m from B' would be the required reaction 
 A am . However, this polygon was actually drawn for a span I, and hence the ordinate 
 i) xm distant x m from B' is l/l am times too small, therefore, 
 
 L 
 
 (59n) 
 
 This illustrates how a sum A line drawn for any span may be used to obtain end 
 reactions for any other span. However, in order to avoid the multiplication of all ordi- 
 
 o 
 
 . 
 
 SCALE OF LOADS. 
 
 300 KIPS. 
 
222 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XII 
 
 nates by I, it is best to draw the sum A line for a span of 1000 ft., and then derive all 
 ordinates according to Eq. (o9n) ? by which the ordinates of one A line are to those of 
 another A line in the inverse proportion of the spans for which the lines were drawn. 
 
 A portion of such a 1000 ft. sum A line is shown in Fig. 59B, from which the actual 
 reactions A and A am , for a span /=200 ft., are obtained as 
 
 and ^= ....... (59.) 
 
 I 
 
 where JJA is the ordinate distant 61 from B' and f) xm is the ordinate distant x m from B'. 
 
 Any moment M TO or chord stress is now readily found from Eq. (59o) , using the values 
 from Eq. (59i) as follows: 
 
 l a 
 
 , ...... (59j) 
 
 and the stress in any chord member having the point m for its center of moments with 
 a lever arm r m , becomes by Eqs. (59A) and (59j) 
 
 (59K) 
 
 The web stresses for trusses with non-parallel chords are derived from moments Mi 
 of the external forces about the center of moments i for any particular web member. 
 
 Calling the lever arm r t - for a web member with moment center i distant Z at - to the 
 left of A then for x m <d and A nm = ' 
 
 Mi = Al a i A nm (L a i + l an ) , 
 or M^lOOO'-^l^^+U, .... V '.. . - . (59D 
 
 when x m >d, a case which happens very rarely, indicating that some loads extend to the 
 left of n, then 
 
 *" -"-w -Anmv/ai -rl an ) ^ ^ r(l n i 
 
 **^ x 
 
 wherein the two last terms must be computed. It should be remembered that this last 
 case does not occur more than once or twice, if at all, for a whole analysis and hence 
 involves only a slight amount of work. 
 
 In any case Eq. (59A) gives the stress in a web member for non-parallel chords, as 
 
 (59x) 
 
 The web stresses for trusses with parallel chords are derived from the shear accord- 
 ing to Eq. (59E) 
 
ART. 59 STRESSES IN STATICALLY DETERMINATE STRUCTURES 223 
 
 Thus the vertical shear to the left of a section through any panel when there 
 are no loads to the left of n, making x m <d, becomes 
 
 0-1 A 1000> M 1000^ m 
 
 y m ^A-A nm = ^ ~^~, (59o) 
 
 wherein A nm is the reaction at n for the loads in the panel ^m extending over a distance 
 x m on a span d. 
 
 When x m >d, which would be the rare case for one or two loads to the left of n, then 
 
 r -2, R -3~2,, P > <w 
 
 where the two last terms would have to be computed for the loads between x and m as 
 in the previous case. 
 
 The stress in any web member of a truss with parallel chords is then found from 
 Eq. (58o) as 
 
 S=Q m secd, . . (59 Q ) 
 
 where Q m is given from Eqs. (59o) or (59p). 
 
 The above demonstration, necessitating the use of one sum A line drawn for a span 
 of 1000 ft., was used only for the development of the formula?. In practice, the sum 
 A line is dispensed with by tabulating the ordinates 1000r M for any case of standard 
 loading. 
 
 One such table (see Table 59A) , will then suffice for the computation of all chord and 
 web stresses in any simple determinate truss for the general case of moving train loads. 
 
 Table 59A thus represents the values of 1000A =1000^ =M B =the sum of the 
 moments of all loads on the 1000 ft. span about the reaction B, expressed in kip feet. 
 These moments are readily computed from Eqs. (59c), and (59i), thus 
 
 4/5 = 1000^4=-- VVp + / 3 VV, . (59R) 
 
 where l q is the distance the train is moved ahead whenever a new load P 4 is brought 
 on the span. 
 
 Eq. (59n) , holds until the uniform load reaches the point B, after which the moments 
 must be figured as for uniformly distributed loads. The table suggests the ease of 
 computation and might be extended over the whole span of 1000 feet, though for pur- 
 poses of illustration this was not considered necessary. 
 
 To facilitate interpolation, the differences in M B for one foot were added in a separate 
 column. Interpolations for distances between those given are thus readily performed 
 and these are strictly accurate up to the point where the uniform load is reached. After 
 that there is a slight error because the moment for uniformly distributed loads varies 
 as I 2 . However, the error from this source is entirely negligible. 
 
 Example. Required the maximum live load stresses in the chord U 2 U 3 , and in the 
 diagonal ~U^L Z of the 200 ft. truss shown in Fig. 59e, using Cooper's E 60 loading. 
 
224 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XII 
 
 TABLE 59A. 
 REACTIONS FOR COOPER'S E-60 LOADING, SPAN OF 1000 FEET. 
 
 Wheel 
 No. 
 
 Load 
 Lenj^th, 
 
 6, 
 
 Feet. 
 
 Total Load 
 
 S?P 
 
 Kips. 
 
 M B =IQOQT) A 
 
 Kips. 
 
 Diff. for 
 1 Foot. 
 
 Kips. 
 
 Wheel 
 No. 
 
 Load 
 Length, 
 
 &1 
 Feet. 
 
 Total Load 
 
 S?P 
 
 Kips. 
 
 -1/5=1000^ 
 Kips. 
 
 Diff. for 
 1 Foot, 
 
 Kips. 
 
 
 
 
 
 
 
 
 
 
 523.5 
 
 1 
 
 
 
 15.0 
 
 0.0 
 
 
 26 
 
 144 
 
 531 
 
 41,293.5 
 
 
 
 
 
 
 15.0 
 
 
 
 
 
 538.5 
 
 2 
 
 8 
 
 45.0 
 
 120.0 
 
 
 27 
 
 149 
 
 546 
 
 43,986.0 
 
 
 
 
 
 
 45.0 
 
 
 
 
 
 553.5 
 
 3 
 
 13 
 
 75.0 
 
 345.0 
 
 
 28 
 
 154 
 
 561 
 
 46,753.5 
 
 
 
 
 
 
 75.0 
 
 
 
 
 
 568.5 
 
 4 
 
 18 
 
 105.0 
 
 720.0 
 
 
 29 
 
 159 
 
 576 
 
 49,596.0 
 
 
 
 
 
 
 105.0 
 
 
 
 
 
 583.5 
 
 5 
 
 23 
 
 135.0 
 
 1,245.0 
 
 
 30 
 
 164 
 
 591 
 
 52,513.5 
 
 
 
 
 
 
 135.0 
 
 
 
 
 
 598.5 
 
 6 
 
 32 
 
 154.5 
 
 2,460.0 
 
 
 31 
 
 169 
 
 606 
 
 55,506 . 
 
 
 
 
 
 
 154.5 
 
 
 
 
 
 613.5 
 
 7 
 
 37 
 
 174.0 
 
 3,232.5 
 
 
 32 
 
 174 
 
 621 
 
 58,573.5 
 
 
 
 
 
 
 174.0 
 
 
 
 
 
 628.5 
 
 8 
 
 43 
 
 193.5 
 
 4,276.5 
 
 
 33 
 
 179 
 
 636 
 
 61,716.0 
 
 
 
 
 
 
 193.5 
 
 
 
 
 
 643.5 
 
 9 
 
 48 
 
 213.0 
 
 5,244.0 
 
 
 34 
 
 184 
 
 651 
 
 64,933.5 
 
 
 
 
 
 
 213.0 
 
 
 
 
 
 658.5 
 
 10 
 
 56 
 
 228.0 
 
 6,948.0 
 
 
 35 
 
 189 
 
 666 
 
 68,226.0 
 
 
 
 
 
 
 228.0 
 
 
 
 
 
 673.5 
 
 11 
 
 64 
 
 258.0 
 
 8,772.0 
 
 
 36 
 
 194 
 
 681 
 
 71,593.5 
 
 
 
 
 
 
 258.0 
 
 
 
 
 
 688.5 
 
 12 
 
 69 
 
 288.0 
 
 10,062.0 
 
 
 37 
 
 199 
 
 696 
 
 75,036.0 
 
 
 
 
 
 
 288.0 
 
 
 
 
 
 703.5 
 
 13 
 
 74 
 
 318.0 
 
 11,502.0 
 
 
 38 
 
 204 
 
 711 
 
 78,543.5 
 
 
 
 
 
 
 318.0 
 
 
 
 
 
 718.5 
 
 14 
 
 79 
 
 348.0 
 
 13,092.0 
 
 
 39 
 
 209 
 
 726 
 
 82,146.0 
 
 
 
 
 
 
 348.0 
 
 
 
 
 
 733.5 
 
 15 
 
 88 
 
 367.5 
 
 16,224.0 
 
 
 40 
 
 214 
 
 741 
 
 85,813.5 
 
 
 
 
 
 
 367.5 
 
 
 
 
 
 748.5 
 
 16 
 
 93 
 
 387.0 
 
 18,061.5 
 
 
 41 
 
 219 
 
 756 
 
 89,556.0 
 
 
 
 
 
 
 387.0 
 
 
 
 
 
 763.5 
 
 17 
 
 99 
 
 406.5 
 
 20,383 . 5 
 
 
 42 
 
 224 
 
 771 
 
 93,373.5 
 
 
 
 
 
 
 406.5 
 
 
 
 
 
 778.5 
 
 18 
 
 104 
 
 426.0 
 
 22,416.0 
 
 
 43 
 
 229 
 
 786 
 
 97,266.0 
 
 
 
 
 
 
 426.0 
 
 
 
 
 
 793.5 
 
 19 
 
 109 
 
 426.0 
 
 24,546.0 
 
 
 44 
 
 234 
 
 801 
 
 101,233.5 
 
 
 
 
 
 
 433.5 
 
 
 
 
 
 808.5 
 
 20 
 
 114 
 
 441.0 
 
 26,713.5 
 
 
 45 
 
 239 
 
 816 
 
 105,276.0 
 
 
 
 
 
 
 448.5 
 
 
 
 
 
 823.5 
 
 21 
 
 119 
 
 456.0 
 
 28,956.0 
 
 
 46 
 
 244 
 
 831 
 
 109,393.5 
 
 
 
 
 
 
 463.5 
 
 
 
 
 
 838.5 
 
 22 
 
 124 
 
 471.0 
 
 31,273.5 
 
 
 47 
 
 249 
 
 846 
 
 113,586.0 
 
 
 
 
 
 
 478.5 
 
 
 
 
 
 853.5 
 
 23 
 
 129 
 
 486.0 
 
 33,666.0 
 
 
 48 
 
 254 
 
 861 
 
 117,853.5 
 
 
 
 
 
 
 493.5 
 
 
 
 
 
 868.5 
 
 24 
 
 134 
 
 501.0 
 
 36,133.5 
 
 
 49 
 
 259 
 
 876 
 
 122,196.0 
 
 
 
 
 
 
 508.5 
 
 
 
 
 
 883.5 
 
 25 
 
 139 
 
 516.0 
 
 38,676.0 
 
 
 50 
 
 264 
 
 891 
 
 126,613.5 
 
 
 All loads in kips for one rail. 
 
ART. 59 STRESSES IN STATICALLY DETERMINATE STRUCTURES 225 
 
 For the chord U 2 U 3 , the center of moments is at m and the maximum stress occurs 
 when the span is fully loaded, making 6 t =1 = 200 ft., and x m =l am =60 ft. Also r m =34.5 ft. 
 
 Then Table 59A gives for 200 ft., 1000yj A =75036 +703.5 =75739.5, and for x m =60 ft 
 1000^ =6948 +4X228 =7860, whence by Eq. (59K), 
 
 = J- [75739.5(1^ -7S60J =430.78 kips. 
 
 For the diagonal U 2 L 3 with load divide at z w = 10.5 ft., making 6 t = 140 + 10.5 
 = 150.5 ft., Z an =40 ft., Z at -=98 ft., d=20 ft., and r^ = 132 ft., and thus x m <d, Table 59A 
 gives 1000r? A =43986 + 1.5X553.5 =44816.3 and 1000^ = 120+2.5X45=232.5. 
 
 Hence by Eqs. (59L) and (59N). 
 
 The minimum stress in C/ 2 ^3 may be found by treating the symmetric member L 7 U 8 
 in the right half of the span. 
 
 Similarly the stress in any member of the truss may be determined for any desired 
 position of the train of loads, all by means of the one Table 59A. 
 
 It is believed that the ease and simplicity of applying the above method, together 
 with its universal applicability, should commend itself to the practical designer. 
 
 Certainly it does not seem warranted to employ approximate methods when an 
 exact solution is possible without additional labor. 
 
CHAPTER XIII 
 SECONDARY STRESSES 
 
 ART. 60. THE NATURE OF SECONDARY STRESSES 
 
 In the previous chapters a framed structure was regarded as a system of individual 
 members linked together by frictionless pin connections. It was also supposed that the 
 neutral or gravity axes of all members meeting in a panel point actually intersected in 
 one point. The stresses computed on this basis are called primary stresses. 
 
 Owing to the friction, which always exists in pin-connected joints, and the rigidity 
 introduced by riveted connections, this condition is never fully realized in practice. Hence 
 certain moments or bending effects are always produced in the proximity of the connected 
 ends, which set up bending stresses of variable character and magnitude. These are 
 generally called secondary stresses as distinguished from the primary stresses. 
 
 All structures involving redundancy are subjected to elastic deformations caused 
 by the loads, temperature, and abutment displacements giving rise to what may be called 
 additional stresses. These also exist to a slight degree in determinate structures, especially 
 when their geometric shape is peculiarly subject to large load and temperature deflec- 
 tions as in the case of three-hinged arches. Still other causes, such as erroneous shop 
 lengths of members, wind pressure, and impact and brake effects of moving loads, pro- 
 duce stresses belonging to this class which do not admit of exact determination and can 
 at best be merely estimated. 
 
 The first theoretical discussions of secondary stresses were given by Asimont, 1877; 
 Manderla and Engesser, 1879, and Winkler, 1881. Graphic solutions were published by 
 Landsberg, 1885; W. Ritter, 1890; and Mohr, 1891-3. A resume of most of these is given 
 by C. R. Grimm, C. E., " Secondary Stresses in Bridge Trusses," 1908. 
 
 The facts which were revealed by these discussions gave rise to very material 
 improvements in the modern practices of designing structures. Thus, the details of con 1 
 nections at panel points and between floorbeams and truss members have been vastly 
 improved, and particular attention is now being given to a more judicious design of the 
 members themselves, especially those subjected to compressive stress. 
 
 It is not the purpose here to treat all the various methods of calculating secondary 
 stresses, but merely to present what appears to the author as the most understandable 
 and usable method. The nature of the problem is such as to preclude the possibility 
 of considering all the complexities involved and yet remain within the limits of practical 
 utility. Hence some approximations must be applied, and justly so, because many of 
 the influences are too small to warrant the labor involved in their analysis, and others 
 are beyond reasonable limits of estimation. 
 
 226 
 
ART. Gl SECONDARY STRESSES 227 
 
 The three primary causes for secondary stresses in bridge members are: the weights 
 of the members themselves producing deflections in all except vertical members and giving 
 rise to bending stresses; the absence of frictionless panel point connections; eccentric 
 connections between members meeting in a common panel point. 
 
 Riveted connections are here treated like fixed ends so that the elastic deformations 
 of the structure are supposedly taken up by flexure in the members themselves with- 
 out producing any changes in the angles at the panel points. While this assumption is 
 not absolutely true, because the bending moments cannot be resisted without producing 
 some elastic angular distortions, yet it is on the side of safety and tends to compensate 
 some other factors which are entirely neglected. 
 
 Pin-connected members, according to more recent experience, cannot be regarded 
 as free from bending moments on account of the rather excessive friction which always 
 prevails on the pins. In the case of eye-bars this may add considerable bending stress 
 owing to the slenderness of such members. This should not be construed to mean 
 that pin connections are inferior to riveted joints, but that the former do not possess 
 all the advantages usually attributed to them. The distinct superiority of pins is to be 
 found in the prevention of eccentric transmission of the direct stresses, a condition which 
 is difficult to accomplish in riveted connections. 
 
 The excessive bending stresses carried from floor-beam connections to the vertical 
 trusses can hardly.be estimated and should be avoided as far as possible by the introduc- 
 tion of flexible connections in preference to the rigid type so much used in the past. 
 
 The secondary stresses, which will now be considered, are those caused by the elastic 
 deformations taking place in the plane of the frame itself and due to the three primary 
 causes above enumerated. 
 
 ART. 61. SECONDARY STRESSES IN THE PLANE OF A TRUSS DUE TO 
 
 RIVETED CONNECTIONS 
 
 Every elastic structure must undergo certain deformation when subjected to loads. 
 If the structural members are rivet-connected at the ends, then this deformation is 
 taken up by the members themselves in the production of certain bending moments which 
 are resisted by the rigid panel connections. 
 
 Figs. 6lA and 61 B show the several possible combinations of single and double flexure 
 with compression and tension as they might occur in any structure. 
 
 The straight chord AB, in each case, represents the position of the member on the 
 supposition of frictionless pin connections, while the curved line indicates the member 
 as it would be distorted by the bending moments and axial forces resulting from the fixed 
 riveted connections. 
 
 Assuming such a member to be released at the ends by sections passed close to the 
 panel points, then the internal stresses may be replaced by external forces P a and P b . 
 Each of these may be resolved into two forces and a moment acting at the original fixed 
 ends. Thus P a is equivalent to S a , A and a moment M a while P b is replaced by S b , 11 
 and a moment M b . The conditions of equilibrium require that P a =P b and hence that, 
 
228 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIII 
 
 S a =P a cosd=S b =P b cos6; also A=B=P sin 6 and M a +M b -Al=0. Since is 
 always very small S a S b =P a -=P b , which assumption is entirely permissible. 
 
 In the case of compression members the concave side of the elastic curve is always 
 
 FIG. 6lA Tension Members. 
 
 toward the resultant P, and hence a maximum moment may occur at some point inter- 
 mediate between A and B. For tension members the convex side is toward the force, 
 and the maximum moment must occur at one end of the member. 
 
 FIG. 61s Compression Members. 
 
 In any case then, the stresses produced in a member by the bending moments M 
 and M b , are readily determined provided these moments are known. Of the several 
 
ART. 61 
 
 SECONDARY STRESSES 
 
 229 
 
 methods proposed for the evaluation of the end moments, the analytic method of Professor 
 Fr. Engesser is probably the most direct and is the one given here. 
 
 For any structure of w members there will be 2m unknown end moments to be 
 evaluated. These are derived from the angular deflections r, which the respective fixed 
 ends of the members undergo as a result of certain distortions Ja, J/?, Jf, etc., produced 
 in the angles of the frame by the work of the external forces and reactions, as per Eqs. (37n). 
 
 The first step in the development of the method is then to derive the relation between 
 the moments M a and MI, and the deflection angles r a and r b for a column on two fixed 
 supports. 
 
 The following assumptions are made in the interest of simplifying the theory for 
 practical applications. 
 
 Since in well designed members the shear, direct thrust and temperature have a very 
 slight effect on flexure, these factors are neglected. However, for poorly designed members 
 with small moments of inertia and slender dimensions, the deflections due to normal 
 thrust etc., may become very considerable and these assumptions might not hold. 
 
 It is further assumed that the axes of all the members are situated in the same 
 plane, and that these axes are centric for all members meeting in a point. Also, that the 
 external loads are all applied at the panel points, neglecting for the present the flexure 
 produced in the members by their own weight. 
 
 Since the fixed ends, Fig. 61c, do not of themselves produce any direct thrust or stress 
 in the member ~AB (not being absolutely fixed in space), therefore, the column involves 
 only two redundant conditions which are the moments M a and M b . The principal system 
 is then a beam on two supports. 
 
 FIG. 61c. 
 
 The effect of the direct stress S is not considered here because for well designed 
 members with comparatively large moments of inertia this was found by Manderla to 
 be entirely negligible. Otherwise a moment Sij would enter the general moment as 
 given by Eq. (6lA). 
 
 Eq. (7 A) now gives for A =0, A n = l/l and A b = -l/l, 
 
 A=A -A a X a -A b X b =-^+^, 
 
 and the moment M m , about any point ra, becomes 
 
 M m = 
 
 a = (X b - X a ) + X a , 
 
 (61A) 
 
230 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIII 
 
 giving 
 
 _ x , 'dM m _x 
 
 - ~ I -dX b ~T 
 
 Then according to Eq. (15L), the virtual work of the moment M a is 
 
 ~j7rr I "nXl ~-^X a + lX a -^X b + -^X a X a -rjj(2X a + X b ) . 
 
 EiL\_Zi Zi o o ^ 'QtiiL 
 
 A similar process will furnish r b and calling X a =M a and X b =M b , then for any 
 member of single curvature 
 
 and 
 
 (61n) 
 
 When the moments M a and M b act in the same direction then the elastic curve 
 will have a point of counterflexure and a similar derivation to the one just given will 
 furnish the values 
 
 and T b =-^=-ft2M b M a ) (61c) 
 
 6EI 
 
 Considering a triangle as the fundamental element of a frame, the distortions Act, 
 Aft and Af in thethree angles resulting from the changes in the lengths of the members, 
 expressed in terms of unit stresses, were determined in Art. 37. It is now necessary to 
 show the relation between these angle distortions and the- deflection angles T at each end 
 of each member forming the triangle. 
 
 FIG. 61o. 
 
 Referring to Figs. 6lD and 6lE, it is easily seen that 
 
 (6lD) 
 
 Fig. 6 ID represents an impossible condition because at least one member of the 
 three composing the triangle must undergo double curvature as shown by Fig. 61 E and 
 as required by Eq. (3?E) , whereby 
 
 0. . (61s) 
 
ART. 61 
 
 SECONDARY STRESSES 
 
 23 1 
 
 The elastic distortions Aa, J/3 and Jj- are given by Eqs. (37o), and may be considered 
 known for all angles of a frame and for one simultaneous case of loading and stress. 
 
 While Eqs. (37D) were derived on the assumption of frictionless pin connections 
 between all members, they are now used to determine the amount of distortion in each 
 angle which is prevented from taking place by virtue of the rigid riveted connections. 
 
 Two equations of the form of Eqs. (61c), may be written for each side of a triangle 
 or for each member of a frame. Using the designations indicated by the double sub- 
 scripts in Fig. 6lD, where the first refers to the apex or panel point and the second to 
 the far end of the adjacent member, then by inserting the deflection angles r into Eqs. 
 (6 ID), the following formulae are obtained for any triangle ABC'. 
 
 QEJa = 
 
 . . (6lF) 
 
 cb +r ca ) = (2M A +M bc ) + 
 
 *a lb 
 
 +M ac ) 
 
 In the Eqs. (61r), the six moments are unknown, and hence a solution is impossible 
 unless at least three of them may be independently derived from other conditions to be 
 developed later. 
 
 According to Eqs. (Glo) the sum of the three angle distortions J for any triangle, 
 must equal the sum of the six deflection angles r and by the condition of Eq. (6lE), both 
 sums must then be equal to zero, hence 
 
 (61c) 
 
 Also, for any apex angles meeting in the same panel point, the sum of the angle 
 distortions J must equal the sum of the deflection angles r, hence for any panel point 
 
 2J = 2r, ........ .... (6lH) 
 
 and therefore, if one of the deflection angles r is known, the others around that panel 
 point may be successively found by applying equations of the form of Eqs. (6lD). 
 
 For any determinate frame, there will thus be as many unknown deflection angles 
 - as there are panel points, and for p panel points there will be p equations of the form 
 of Eqs. (61n), each involving one unknown deflection angle T. 
 
 The following values 
 
 M ab l c 
 
 r J 
 
 6/ c 
 
 "* ca'b 
 
 (61 1) 
 
232 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIII 
 
 are now substituted into Eqs. (61r), as a matter of convenience, to obtain the following 
 formulae for any triangle ABC, thus 
 
 EAa=K ba +2(K ac +K ab )+K ea , . . . / ,~ : . . (61j) 
 EAp=K ab +2(K bc +K ba )+K cb , . . , .'._. . (61 K) 
 EA r =K bc +2(K cb +K ca )+K ac . . t -. \ . [. . ".. (61 L ) 
 
 Adding these three equations and imposing the condition of Eq. (6 IE), then for 
 any triangle ABC, 
 
 =Q. ....,'-., (6lM) 
 
 Assuming now that the three quantities K^, K^ and K ba are known, then the remain- 
 ing three may be obtained from the above equations as follows: 
 
 from Eq. (61J), K ca =EJa-2(K ac +K ab ) -K^, . .>. . .. (Glx) 
 
 from Eqs. (6 IK) and (<ol^) , K^^E^+K^-K^+K^, . . . ; . .' (61o) 
 from Eqs. (6lL) and (6lM), K cb =EAr+k ab +K ba -K ca . . t . . . . (61r) 
 
 By the conditions of static equilibrium, the sum of all the moments about any panel 
 point must be zero, from which p equations of the following form are obtained, using 
 the values from Eqs. (61i). Thus, for any panel point A 
 
 ~ '' 
 
 including all the members meeting in this point. See Fig. 61n and Eq. (61u) for the 
 manner of forming these equations. 
 
 The following procedure will furnish a solution for all the bending moments of a frame, 
 using Eqs. (6 IN) to (61q) : In the first triangle, evaluate three of the quantities K in 
 terms of the other three, using Eqs. (61 N, o, P). Two of the values K so found, also 
 belong to the adjacent or second triangle and a third value K in this second triangle 
 may be found from Eq. (6lQ), making three values known to find the other three by 
 applying Eqs. (6lN, o, P) to the second triangle. 
 
 This process is continued throughout the series of triangles up to the last one, where 
 two extra moment condition Eqs. (61q) become available and these suffice to deter- 
 mine the three first assumed values K. This is merely a process of successive elimination, 
 eminently suited to the present problem. 
 
 When the values K are thus found then the several moments M are obtained by sub- 
 stitution into Eqs. (61i). See the example below for symmetric structures with symmetric 
 loading. 
 
 The solution of the whole problem is possible, because for any frame with m members 
 and p panel points, there will be (ra 1) triangles furnishing f(w 1) equations of the 
 form of Eqs. (6lN, o, P) and p equations of the form of Eq. (61q) making in all |(rw 1) +p 
 
ART. Gl SECONDARY STRESSES 233 
 
 available equations from which to find 2m unknown moments M. Hence, if the prob- 
 lem is determinate, there must be as many equations as there are unknowns and 
 
 2m = |(m-l)+p, giving 2p=m+3 (61n) 
 
 which by Eq. (3x) is actually true for any determinate frame composed of triangles 
 and, therefore, the solution is possible. 
 
 It was previously pointed out that the assumptions regarding the signs of the 
 deflection angles r as illustrated in Fig. 6lD, are impossible, and that not all these angles, 
 nor the moments M in the members of a triangle can have the same algebraic sign on 
 account of the condition imposed by Eq. (6lE). Therefore, the computed results must 
 furnish deflection angles r and moments M of both signs. 
 
 However, to avoid complicated rules, it is necessary to adopt some conventional 
 or standard figure and then reverse the assumed deflections wherever the computed angles 
 T are negative. For this reason the conventional figure of distortion is assumed as indicated 
 in Fig. 61r, wherein the members in alternate triangles I, III, "V, etc., are made to 
 present a concave line outward, indicating positive moments M. 
 
 FIG. 61r. 
 
 The even triangles II, IV, VI, etc., must then represent negative moments, but 
 in order that Fig. 61r may represent the conventional assumption of all positive moments, 
 the Eqs. (3?D) are divided through by minus one when applied to the even triangles with 
 convex lines outward. 
 
 Hence, for the odd triangles with positive moments M, Eqs. (3?D) are applied in 
 
 the form 
 
 E Ad = (f a ~fb) COt r + (f a ~f c ) COt /? 1 
 
 EJp = (f b -fc) cota+(/ 6 -/a) cot r \, (61s) 
 
 E A Y - (f c -fa) cot p + (f c -f b ) cot a 1 
 
 and for the even triangles the negative moments are also made positive by using Eqs. 
 (61s) in the form 
 
 = (/6-/ a ) cot r + (fc-fa) cot/? I 
 (f c -fb) cota + (/-/ fc ) cot r , (6lT) 
 
 = (fa -fc) cot p + (f b -fc) cot a 
 
234 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XI! I 
 
 and then all moments, represented by Fig. 61r, may be considered positive. The sub- 
 scripts a, b, and c in Eqs. (61s) and (Gli), refer to the sides a, b, and c opposite the 
 respective angles a, {3, and 7-. 
 
 Since the above assumptions do not represent real or possible conditions, some of 
 the moments M, resulting from the final computation, must develop negative signs. 
 Thus, if M ac and M^ in Fig. 61o, should turn out with negative signs, then the real 
 deformations of the triangle would be as represented in Fig. 6 IG. 
 
 FIG. 61n. 
 
 In evaluating Eqs. (61cj) for the several panel points, it is necessary to observe the 
 relative signs of the moments in the several members according to the following direc- 
 tions. Thus, for any panel point, C, Fig. 61n, the clockwise moments should be treated 
 as positive and the counter clockwise moments as negative to obtain the following 
 equation : 
 
 = +M oa -M ob +M oc -M od =Q, 
 
 or in terms of the values K 
 
 +K 
 
 6/0 
 
 ion 
 
 Qlob 
 
 lob 
 
 " 
 
 'On 
 
 (61u) 
 
 The arithmetic operations required in the solution of problems are somewhat lengthy 
 but not difficult especially when the values 7 and I of the members are well rounded off, 
 which is always permissible in these computations. 
 
 The secondary stress resulting from the bending moments M =QKI/l, produced in any 
 member by the riveted connections, may now be found from Navier's law, Eq. (49M), as 
 
 My_ QKy 
 
 (61v) 
 
 where y is the distance from the gravity axis of the member to the extreme fiber and 
 f s is the unit secondary stress caused by the bending of the member due to the moment 
 Mg, or Mb applied at one end of the member. 
 
 The total maximum unit stress in the member is then ff s , where f=S/F, which 
 is the primary unit stress due to the direct loading. 
 
ART. 01 SECONDARY STRESSES 235 
 
 Thus for a compression member, the maximum total stress occurs on the side where 
 f s is negative. 
 
 The eccentricities e, Fig. 6lA or 6lB, produced at either end of any member ~AB by 
 the secondary moments, may be found from 
 
 and *!_. ...... (61W) 
 
 The secondary stresses here considered may vary in percentage of the primary stresses 
 from 5 to 100 per cent, depending on the type of truss and the details of design employed 
 for the members. 
 
 The subject is, therefore, one of considerable importance and cannot be dismissed 
 in the usual manner as taken care of by the safety factor. 
 
 The question of maximum secondary stress in any member has not yet received 
 consideration here, though it is a matter of vital importance. 
 
 It does not follow that the maximum secondary stress occurs" simultaneously with 
 the maximum primary stress for any particular member. The results of computations 
 show this to be quite true for the chords, but not for the web members, though there 
 seems to be no criterion available to indicate the position of a train of loads to produce 
 maximum secondary stress in a member. 
 
 The method of influence lines, while not impossible, becomes practically prohibitive 
 on account of the labors involved. The only reasonable treatment which suggests itself 
 is to analyze the structure first for total maximum load and then for a few typical posi- 
 tions for the moving load and from these results select the probable maximum values 
 for each member. 
 
 Since the entire truss must be completely solved for each simultaneous position 
 of the live load, this appears to be the only practicable advice to give in view of the 
 immense labors involved in such analyses. 
 
 The above method is now illustrated by the solution of a problem. 
 
 Example. A single-track, through Pratt truss, designed by Messrs. Waddell & 
 Hedrick, in 1899, for the Vera Cruz and Pacific Ry., Mexico, is selected. The truss has 
 six panels with a span of 145 feet, a depth of 30 feet, and the distance between trusses 
 is 17 feet. The equivalent uniform live load for WaddelTs class W was used. The case 
 here investigated is for maximum total load over the entire span. 
 
 The cross-sections F and the moments of inertia I are computed and tabulated 
 together with the lengths I, widths b and distances y for all the members of the truss. 
 These together with the stresses S, due to the total maximum load Fig. 61i, are given 
 in Table 61 A, where the unit stresses f=S/F and the functions 61/1 and 6y/l are com- 
 puted. Impact is not considered. 
 
 A live load of 54000 Ibs. and a dead load 20800 Ibs. per panel were assumed for 
 
 tke computation of the stresses. 
 
 While it is not necessary to know how the members were designed as long as F and 
 7 are given for each, yet it may be of some interest to the reader to have this 
 Figs. 61 J thus represent the sections of the several members. 
 
236 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIII 
 
 FIG. 61i. 
 
 B-D. 
 
 D-F. 
 
 A-C-E. 
 
 E-G. 
 
 B-C. 
 
 T*8 
 
 D-EANOF-G. 
 
 B-EAnoD-G. 
 
 II 1 
 
 FIGS. 61J. 
 
 C E 
 
 FIG. 6lK Unit Stresses and Cotangents 
 
ART. 61 
 
 SECONDARY STRESSES 
 
 237 
 
 TABLE 6lA 
 DATA FOR THE TRUSS FIG. 61i. 
 
 Mem. 
 
 F 
 Gross, 
 
 Sq.in. 
 
 7 
 
 Gross, 
 
 in.< 
 
 / 
 in. 
 
 6 
 
 In. 
 
 y 
 
 In. 
 
 s 
 
 Lbs. 
 
 f- S 
 f- F 
 
 Lbs. sq.in. 
 
 67 
 
 l 
 
 6v 
 1 
 
 AC 
 
 15.9 
 
 139 
 
 290 
 
 12.37 
 
 6.19 
 
 + 158,400 
 
 + 9960 
 
 2.88 
 
 0.128 
 
 CE 
 
 15.9 
 
 139 
 
 290 
 
 12.37 
 
 6.19 
 
 + 158,400 
 
 + 9960 
 
 2.88 
 
 0.128 
 
 EG 
 
 26.9 
 
 282 
 
 290 
 
 12.5 
 
 6.25 
 
 + 252,060 
 
 + 9370 
 
 5.84 
 
 0.129 
 
 BD 
 
 26.5 
 
 750 
 
 290 
 
 15.0 
 
 7.5 
 
 -252,060 
 
 -9510 
 
 15.52 
 
 0.155 
 
 DF 
 
 29.4 
 
 805 
 
 290 
 
 15.0 
 
 7.5 
 
 -283,500 
 
 -9640 
 
 16.66 
 
 0.155 
 
 AB 
 
 33.5 
 
 1464 
 
 462 
 
 16.25 
 
 8.34 
 
 -245,000 
 
 -7310 
 
 19.01 
 
 0.108 
 
 BE 
 
 17.6 
 
 323 
 
 462 
 
 12.0 
 
 6.0 
 
 + 156,200 
 
 + 8875 
 
 4.19 
 
 0.077 
 
 DG 
 
 17.6 
 
 323 
 
 462 
 
 12.0 
 
 6.0 
 
 + 48,400 
 
 + 2750 
 
 4.19 
 
 0.077 
 
 BC 
 
 13.7 
 
 119 
 
 360 
 
 12.37 
 
 6.19 
 
 + 68,000 
 
 + 4960 
 
 1.98 
 
 0.103 
 
 DE 
 
 14.7 
 
 288 
 
 360 
 
 12.0 
 
 6.0 
 
 - 44,200 
 
 -3010 
 
 4.80 
 
 0.100 
 
 FG 
 
 14.7 
 
 288 
 
 360 
 
 12.0 
 
 6.0 
 
 6,800 
 
 - 460 
 
 4.80 
 
 0.100 
 
 6 = width of member in the plane of the truss. 
 
 y = distance from neutral axis to extreme fiber of section. 
 
 I is taken about the gravity axis perpendicular to the plane of the truss. 
 
 The distortions in all the angles of the several truss triangles are now computed 
 from Eqs. (61s) and Eqs. (6lT), using the unit stresses/ from Table 6lA and the cotangents 
 of the angles, all as shown in Fig. 6lK. 
 
 Applying Eqs. (61s) to the odd triangles I, III and V, and Eqs. (6lT) to the 
 even triangles II and IV the following values Ed are obtained. 
 
 Triangle I. Eqs. (61s) : 
 
 EAai=( 4960-9960)0.0 +( 4960+7310)1.24 ==+15,215 
 i=( 9960 +7310)0.806 + ( 9960-4960)0.0 ==+13,920 
 i = ( -7310 -4960) 1.24 +(-73 10 -"9960) 0.806= -29,135 
 
 00 
 
 Triangle II. Eqs. (6 IT): 
 
 = ( 9960 -8875)0.806 + ( 4960- -8875)1.24 =- 3,980 
 
 = ( 4960-9960)0.0 +( 8875-9960)0.806=- 875 
 
 2 = ( 8875-4960)1.24 +( 9960-4960)0.0 =+4,855 
 
 00 
 
238 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIII 
 
 Triangle III. Eqs. (61s): 
 
 = ( -3010 +9510)0.0 +(-3010-8875)1.24 =-14,737 
 = ( -9510-8875)0.806 + ( -9510 +3010)0.0 = - 14,818 
 = ( 8875+3010)1.24 +( 8875+9510)0.806= +29,555 
 
 00 
 Triangle IV. Eqs. (6 IT) : 
 
 = ( 9370 -2750)0.806 + (-3010 -2750) 1.24 =- 1,806 
 = ( -3010 -9370)0.0 + ( 2750 -9370)0.806 = - .5,336 
 = ( 2750+3010)1.24 +( 9370+3010)0.0 =+ 7,142 
 
 00 
 Triangle V. Eqs. (61s) : 
 
 = (- 460+9640)0.0 +(- 460-2750)1.24 -- 3,980 
 = ( -9640 -2750) 0.806 + (-9640+ 460)0.0 = - 9,986 
 -( 2750+ 460)1.24 +( 2750+9640)0.806= +13,966 
 
 00 
 
 The condition Eq. (6lE) must be satisfied for each triangle as indicated above, offer- 
 ing a complete check on the numerical results. 
 
 FIG. 6lL Values EA. 
 
 The values EA are now entered on the diagram Fig. 6lL, where the conventional 
 assumptions for the moments M are indicated according to Fig. 6 IF. The moment 
 Eq. (61q), is then written out by following the directions given under Eq. (61u). 
 
 Using the values for 67/7 from Table 6lA in Eq. (61q) the following equations are 
 obtained : 
 
ART. (51 SECONDARY STRESSES 239 
 
 For panel point A, 
 
 "ob 'oc 
 
 and after inserting the values from Table 61 A, this gives 
 
 19.01^=2.88^ or K ac =Q.QQK ab . ...... (1) 
 
 For panel point B, 
 
 2 M b = - M te + M bc - M be + M bd = 0, 
 giving 
 
 - ig.Oltffc, + 1.98# 6c -4. l9K be + 15.52K bd =0, 
 or 
 
 K bd = 1.23^-0.12^ +0.27 K be ......... (2) 
 
 For panel point C, 
 
 2M c = M Ctt -M c6 +M ce =0, 
 giving 
 
 2.8&K M - 1.98^ +2.8K C9 =0, 
 or 
 
 ^ ca . . . ........ (3) 
 
 For panel point D, 
 
 2M d = -M db +M de -M dg +M df =0, 
 giving 
 
 - 15.52K db +4.SOK de -4. l9K dg + 16.66 K df =0, 
 
 or 
 
 (4) 
 
 For panel point E, 
 
 SM e = -M ec +M eb -M ed +M en =0, 
 
 giving 
 
 -2.88K ec +4.19# e6 -4.8QK ed +5.84K eg =0, 
 
 or 
 
 ... (5) 
 For panel point F, noting that for symmetrical loading Mf d =M fh , then 
 
 panel point G, where for symmetric loading AT,, =--M gi and M gd = M ghi then 
 
 or 
 
 -r /A^ 
 
 ,11? 71 r Hf . , ... 1 71/f ?IT . ffon 
 
 For 
 
 giving , , v n 
 -5MK ge +4.19X ffd -4.80^/ +4.19A,, -o.84^ fft - =0, 
 
 #^--2.433^+1.746^. (7) 
 
240 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP.XIJI 
 
 When there is no middle vertical FG, then, for symmetric loading, an equation of 
 the form Eq. (6lM) is written out for the center triangle DGH to supply the last condition. 
 
 For an unsymmetric truss or unsymmetric loading, the several equations must be 
 extended over all pin points of the structure. 
 
 Assuming now that K ab , K^ and K ac , Fig. 6lL, are known, then Eqs. (6lN, o, P) 
 will furnish the values K^, K bc and K cb for the first triangle ABC, and by applying this 
 process successively to all the triangles it becomes possible to express all the values K 
 in terms of K ab and K^ by including the moment equations (1) to (5) just evaluated. 
 The two last Eqs. (6) and (7) will then serve to find K ab and K^, as will be shown later., 
 and these values substituted back will furnish all the unknowns K. 
 
 In applying Eqs. (6lN, o, P) to the several triangles, it is decessary to use the standard 
 lettering employed in the derivation of these formulae and hence each triangle is sketched 
 in Figs. 6lM, to exemplify the process. The angle a must be so located in a triangle 
 that it is included between the two adjacent known K's, and the angle /? must be adjacent 
 to the third known K, while f is between the adjacent unknown K's. This then deter- 
 mines the vertices (A), (B) and (<?), shown in parenthesis, for the standard lettering. 
 
 FIGS. 6lM. 
 
 The designations of the angles in Fig. 6lK, for the solution of the values EA were 
 selected so as to comply with the conditions just described. The arrows in the Figs. 
 6lM indicate the values K found from a previous triangle and from one new moment 
 Eq. (6lQ) above evaluated as (1) to (5). 
 
 Eqs. (6lN, o, P) and Eqs. (1) to (5) thus give: 
 
 At panel point A, by Eq. (1) : K ac =Q.QK ab . 
 
 By Eq. (6lN), K M 
 
 By Eq. (61o), K be 
 
 By Eq. (61p), K A =EA n +K ta> +K l>a -K ea ^ -44,350 + 16.2^ 
 
 At panel point C, by Eq. (3), K ee =O.Q9K A -K M = -45,816+26.4^+2.38^. 
 
 I By Eq. (6lN), K ec =EJa 2 -2K ee -2K A -# 6c = 147,2 17 -76.6^-6.76^ 
 
 II \ By Eq. (61o), K bc =E4p 2 +K cc -K hc +K ec -=71,3Ql -41.6^-2.38^ 
 
 By Eq. (61p), K eb =EA r2 +K cb +K bc -K ec = - 157,577 +84.2# a6 +6.767^. 
 
ART -6l SECONDARY STRESSES 241 
 
 At panel point B, by Eq. (2) : ^ M = l-23X te -0.12X 6c +0.27X 6e = 15,780-10.2X o6 +0.83X 6([ . 
 
 [ ByEq. (61 N ), K db =EJa 3 -2K bd -2K be -K eb = -31,502 + 19.4X a6 -3.66X 6o 
 AIII j ByEq. (61o), K ed ~EJ i 8 3 +K M -K eb +K db = l 27,037 -75.0X a6 -9.59Xj >a 
 
 I By Eq. (61p), K de =E4 r3 +K be +K eb -K db = -25,129 +23.2X a6 +8.04X 6a . 
 
 At panel point E, by Eq. (5) : 
 
 K eg =0.493X ec -0.72X e6 +0.82X ed =290,203 - 159.9X n6 - 16.07X 6a . 
 
 I" By Eq. (6lN), K ge =EJa^-2K eg -2K (id -K de = -811,157+446.6X a6 +43.28X 6a 
 AIV I ByEq. (61o), K dg =EJ l 3 4 +K eg -K de +K ge = -501,161 +263.5X^ + 19.17X6,, 
 
 I By Eq. (61p), K gd =E4 r4 +K ed +K de -K ge =920,207 -498.4X tt6 -44.83X 6a . 
 
 An panel point D, by Eq. (4) : 
 
 K d1 =0.932X d6 -0.288X de +0.252X dff = -148,414 +77.80 K ab -O.gOX^. 
 ByEq. (6lN), K fd =EAa5-2K df -2K dg -K gd =374,963 -184.2X a6 +8.29X 6a 
 AV ByEq. (61o), K g f=E^s + K df -K ad + K fd = -703,644 +392.0X ab +52.22X fca 
 
 By Eq. (61?) , K fa =EJ r5 +K dg +K gd -K fd =58,049 -SOJX^ -33.95^. 
 
 The moment Eqs. (6) and (7) for panel points F and G now furnish independent 
 values for X/ ff and K g f and by combining these with the last two equations of AV, the 
 values K ab and X^ may be found as follows: 
 
 By Eq. (6), X /ff =6.942X /d =2,602,993 -1278.7X a6 +57.55X 6a . 
 
 By Eq. (7), ^--2.433^+1.746^-3,580,226-1056.81^-183.57^. 
 
 Substituting these values of Kf g and K g j into the last two equations of AV, then, 
 
 2, 544,954 -1228.0X aft + 91.50X^=0, 
 4,283,870 -2348.8X a6 -235.79X fea =0, 
 
 from which K ab = 1947.5 and K ba = -1676.8. 
 
 These values of K ab and X^ are now substituted into the above twenty equations, 
 furnishing all the values X as follows: 
 
 (1) K ab = 1947.5 
 
 (2) K ba =- 1676.8 
 
 (3) X ac =6.6X ab = 12,854 
 
 (4) # = 15,215- 29,602+ 1,677= -12,710 
 
242 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIII 
 
 (5) K bc = 29,135- 16,748+ 3,354= 15,741 
 
 (6) K cb =- 44,350+ 31,549- 3,354 = -16,155 
 
 (7) K cc =- 45,816+ 51,414- 3,991= 1,607 
 
 (8) K ec = 147,217-149,178 + 11,335= 9,374 
 
 (9) K be = 71,391- 81,016+ 3,991= - 5,634 
 
 (10) K eb =- 157, 577 + 163 ,979 -11, 335=- 4,933 
 
 (11) K bd = 15,780-19,864- 1,392- - 5,476 
 
 (12) Kdb=-- 31,502+ 37,781+ 6,137= 12,416 
 
 (13) K ed = 127,037-146,062 + 16,081 = - 2,944 
 
 (14) K de =- 25,129+ 45,182-13,481= 6,572 
 
 (15) K eg = 290,203-311,405+26,946= 5,744 
 
 (16) K ge = -811,157 +869,753 -72,572 = -13,976 
 
 (17) X d ,= -501,161 +513,166 -32,144 = -20,139 
 
 (18) K gd = 920,207-970,634+75,171= 24,744 
 
 (19) K df =- 148,414 + 151,515+ 1,509= 4,610 
 
 (20) K id = 374,963-358,730-13,901= 2,332 
 
 (21) K gf = -703,644 +763,420 -87,562 = -27,786 
 
 (22) K ta = 58,049- 98,738+56,927= 16,238 
 
 Eq. (6lM) now furnishes a convenient check on the above numerical results, by tak- 
 ing the sum of all the six values K for each triangle. 
 
 Thus for AI, the values (1) to (6) give 
 
 S#i =30,542 -30,542 = 
 and for All, the values (5) to (10) give 
 
 2K 2 = 26,722 -26,722= 
 and for AIII, the values (9) to (14) give 
 
 2# 3 = 18,988 -18,987= ?. 
 and for AIV, the values (13) to (18) give 
 
 2 A' 4 =37,060 -37,059 = 1 
 and for AV, the values (17) to (22) give 
 
 2^5=47,924-47,925= -1 
 
 The secondary stresses due to the bending at each end of each member are now found 
 from Eq. (61v) as given in the following Table 6lB. 
 
ART. (51 
 
 SECONDARY STRESSES 
 
 243 
 
 TABLE 6lB 
 SECONDARY STRESSES DUE TO BENDING 
 
 Mem. 
 
 End. 
 
 6y 
 I 
 in. 
 
 K 
 
 /.- 5 ? 
 
 Eq. (61v). 
 Lbs. sq.in. 
 
 '- 
 
 Lbs. 
 sq.in. 
 
 67 K 
 
 " SI 
 in. 
 
 Mem. 
 
 End. 
 
 E 
 D 
 E 
 G 
 D 
 G 
 D 
 F 
 G 
 F 
 
 6.V 
 I 
 in. 
 
 K 
 
 '.= 6J P 
 Eq. (61y), 
 Lbs. sq.in. 
 
 /-I 
 
 Lbs. 
 sq. in. 
 
 61 K 
 
 ~ SI 
 in. 
 
 AB 
 
 AC 
 EC 
 CE 
 BE 
 BD 
 
 A 
 B 
 A 
 
 C 
 B 
 C 
 C 
 E 
 B 
 E 
 B 
 D 
 
 0.103 
 
 + 1,948 
 - 1,677 
 + 12,854 
 -12,710 
 + 15,741 
 -16,155 
 + 1,607 
 + 9,374 
 - 5,634 
 - 4,933 
 - 5.476 
 + 12,416 
 
 200 
 172 
 1645 
 1627 
 1621 
 1664 
 205 
 1200 
 434 
 380 
 849 
 1924 
 
 -7310 
 -7310 
 + 9960 
 + 9960 
 + 4960 
 + 4960 
 + 9960 
 + 9960 
 + 8875 
 + 8875 
 -9510 
 -9510 
 
 0.15 
 0.13 
 0.23 
 0.23 
 0.46 
 0.47 
 0.03 
 0.17 
 0.15 
 0.13 
 0.34 
 0.72 
 
 ED 
 EG 
 DG 
 DF 
 GF 
 
 0.100 
 
 - 2,944 
 + 6,572 
 + 5,744 
 -13,976 
 -20,139 
 + 24,744 
 + 4,610 
 + 2,332 
 -27,786 
 + 16,238 
 
 294 
 657 
 741 
 1803 
 1551 
 1905 
 715 
 361 
 2779 
 1624 
 
 -3010 
 -3010 
 + 9370 
 + 9370 
 + 2750 
 + 2750 
 -9640 
 -9640 
 - 460 
 - 460 
 
 0.32 
 0.72 
 0.13 
 0.32 
 1.76 
 2.16 
 0.28 
 0.14 
 19.62 
 11.46 
 
 0.128 
 
 0.129 
 
 0.103 
 
 0.077 
 
 0.128 
 
 0. 155 
 
 0.077 
 
 0.100 
 
 0.155 
 
 
 
 The total stress on the extreme fiber is f s +/, noting that no increase was made in / for buckling 
 effect in compression members. 
 
 The actual signs of the values K, and hence also of the moments M^&KI/l, now being 
 determined, the real character of the distortions may be represented diagrammatically 
 in the following Fig. 6 IN. 
 
 For compression members the most severely stressed fibers will occur on the side 
 where /. is negative and for tension members on the side where / is positive. Thus in 
 the compression member AB, the critical stressesoccur on the upper side at A and on the 
 lower side at B, while for the tension member AC, these occur on the upper 
 and on the lower side at C. 
 
244 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIII 
 
 ART. 62. SECONDARY STRESSES IN RIVETED CROSS FRAMES OF TRUSSES 
 
 The analysis of cross frames, so far as this is possible, presents many obstacles, some 
 of which cannot be overcome owing to the variable character of the external loads. 
 
 The bending effect due to all forces acting on a frame, even when direct stresses are 
 neglected, leads to very complicated formulae of more or less questionable value, and 
 no attempt is made here to discuss the general problem. 
 
 A cross frame is usually subjected to the dead load of the bridge floor concentrated 
 on the floor beam; the live load, impact and centrifugal force applied to the floor; wind 
 pressure against the moving load and the vertical trusses; and unequal deflection of the 
 main trusses due to a variety of causes, but particularly to unequal temperature. 
 
 The magnitudes of the secondary stresses depend of course on the details of construc- 
 tion and bracing employed in any special frame, so that many different forms would 
 require investigation. However, only two principal types will be discussed and the 
 same formulae are applicable to both through and deck bridges. 
 
 In the following, the unbraced and one type of braced cross frames are considered 
 first for direct loading and then for wind effect. 
 
 (a) Dead and live load effects. Unbraced cross frame with rigid post connections. The 
 construction, shown in Fig. 62A with lettered dimensions, is analyzed by assuming 
 equality between the unknown bending moments induced in the posts by the symmetric 
 loads P, as follows: M = M 2 and M] =M S , positive as indicated when they produce 
 compression on the outer fibers of the posts. 
 
 Call M the bending moment on the floor beam due to symmetrically placed loads 
 P acting on a simple beam which rests on two supports C and D. 
 
 Also call mi the moment at any point of the post AC distant x from A ; m?. the 
 moment at any point of the strut AB; and m^ the moment at any point of the beam 
 CD. Then from Fig. 62A. 
 
 m l =M 1 +( M ~ Ml \x, m 2 =M 1 and m 3 =M +M. . . . (62A) 
 \ h / 
 
 Considering only the effect due to bending, by neglecting shear and direct thrust, 
 the virtual work of deformation for the entire frame would be by Eq. (15n) 
 
 ~ Cmfdz. .- . : . . . . . . . . (62n) 
 
 -&/3./0 
 
 ~ 
 
 -&/3 
 
 Substituting the values from Eqs. (62A) into Eq. (62p,) then 
 
 which integrated by considering everything constant except x, gives 
 
 ?^.+ * \M *b+2M f b Mdx+ P^WI, (62c) 
 
 /2 bl3\_ J Jo 
 
ART. 62 
 
 SECONDARY STRESSES 
 
 245 
 
 which is the general expression for any case of loading where M must be evaluated for 
 each case. 
 
 Now since the unknown moments M and MI must have such values as will make 
 the first differential derivative of W with respect to each, equal to zero, then after some 
 reduction 
 
 'dW 
 
 and 
 
 3M 3/i 
 
 _ ^ =f) 
 
 - " " 
 
 (62n) 
 
 ?*" m * II 
 
 -j 
 
 ,-"" 
 
 "* 
 
 .-I, I,- 
 
 T 
 
 50 ITIj Ij - 
 
 D 
 
 FIG. 62A. 
 
 
 FIG. 62B. 
 
 Solving Eqs. (62n) gives 
 
 ,. 6l\o . \ oil 
 
 = -Mn(-r^-+2 -TJ- 
 
 and 
 
 from which 
 
 (62s) 
 
 Mi = 
 
 ^i- 1 Mdx 
 
 3/i6 
 
 (62p) 
 
 -1 
 
 1/2 '-/\ W 3 
 
 For the symmetric loading shown in Fig. 62A, the integral Mdx becomes 
 
 and for a uniform load p per foot of floor beam, 
 
 fb f)b 3 
 
 JQ 
 
 (62G) 
 
246 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIII 
 
 MQ and MI may thus be found from Eqs. (62F) and (62r), and from these the moments 
 TOI, m-2 and ra 3 , for any point of the frame, are given by Eqs. (62 A). 
 
 The maximum fiber stress in the posts must be combined with the direct stress sus- 
 tained as a truss member. 
 
 The upper strut receives a compression of (MoMJ/h and a bending moment MI 
 over its entire length. 
 
 The floor beam will receive a bending moment of Mo+M and a direct stress of 
 (Mi-Mo) /h. 
 
 Example. Given the cross frame at DE, Fig. 61i, with the following dimensions: 
 h =28 ft., b -=17 ft., a = 7 ft., P -59000 Ibs., h =226 in. 4 = 0.0109 ft. 4 , 7 2 = 1134 in. 4 =0.0547 
 ft., 4 and 7 3 = 15262 in. 4 =0.736 ft. 4 The style of the frame is as shown in Fig. 62A. 
 
 Using the value from Eq. (62c) in Eq. (6.2r), and substituting the above data, then 
 
 3X00109 59000 
 __ 28X0.736* 4 (U n __ 
 
 /3X0.0109X17 W3X0.0109X17 \ - > 
 
 V 28X0.0549 )\ 28X0.736 / 
 
 and from Eq. (62s) 
 
 For t/=5.5 inches, Eq. (61v) then gives the secondary stress in the post DE. 
 
 t ,, ,Miy 1499X12X5.5 
 
 for the upper end D, f d = -^~ = =438 Ibs. per sq.in.; 
 
 /I 
 
 t ., . ,M v 3541X12X5.5 
 
 for the lower end E, f e = p- = = 1034 Ibs. per sq.m. 
 
 1\ 22u 
 
 The actual deformation of this frame is indicated in Fig. 62s. 
 
 A braced cross frame with rigid post connections Fig. 62c, will now be treated as in 
 the previous case. 
 
 Neglecting shear and direct stress as before, and dealing only with the effect due 
 to bending, then Eq. (62B) becomes 
 
 Integrating this expression as above, and then differentiating first with respect 
 to MQ and then with respect to MI and placing these derivatives equal to zero, the follow- 
 ing equations are obtained: 
 
 ~~ = A! M + 2hi M l + 2eMi - 0. 
 
ART. 62 SECONDARY STRESSES 
 
 Solving these two zero equations and noting that h=h\+e, then 
 
 247 
 
 i M 
 
 and 
 
 f 
 
 Jo 
 
 Mdx 
 
 (62n) 
 
 /*6 
 
 wherein I Mcfo is given by one of Eqs. (62a) and the moments MQ and M\ may 
 
 Jo 
 
 thus be found. 
 
 The stresses in the upper struts then become 
 
 M I M, 
 
 = -- and /S e / = -- 1 
 e e 
 
 t. 
 
 *, 
 
 I. 
 
 B 
 
 11 
 
 / 
 
 H 
 
 1L. 
 
 -m, 
 -I. 
 
 K 
 
 > .. _ 
 
 FIG. 62c. FIG. 620. 
 
 and the stresses in the post will be 
 
 and 
 
 FIG. 62E. 
 
 Example. Taking dimensions as in the previous example and making hi =19.5 ft. 
 then M =4770 ft.-lbs., and ^=1661 ft.-lbs. giving /, = 1393 Ibs. per sq.in., and f c = 
 
 Ibs. per sq.in. . 
 
 (b) Wind effects. Unbraced cross frame with rigid post connections. 
 
 wind loads on the trusses of a bridge may be carried to the abutments by means of com- 
 plete horizontal trusses in the planes of the top and bottom chords respectively, pro 
 vided the end reactions of the top chord wind system can be carried to the abutments 
 by suitable end postal bracing. In this case the intermediate cross frames suff< 
 little or no distortion and hence carry no bending effects. 
 
 However when the wind pressure along the top chord is carried down to the I 
 torn chord locally at each cross frame, then the latter must be distorted and thus resi 
 bending In this case the total wind pressure is carried to the abutments through the 
 bottom chord lateral system. The external forces on a cross frame are then as repn 
 
248 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIII 
 
 sented in Fig. 62D. The wind loads w are transmitted through the posts to the lower 
 lateral system, producing bending effects as shown in Fig. 62E. 
 
 Assuming that the wind loads w are equal and calling T the tangential stress at the 
 point of counterflexure and S the normal stress at the same point, then 
 
 H=w = 
 
 (62i) 
 
 The general work equation for the frame, considering bending effects only, is the 
 same as Eq. (62fi) in which the moments mi, m 2 and ra 3 must be evaluated for the wind 
 forces now acting. 
 
 The moments of the external forces about A, B, C, D and 0, from Fig. 62o (count- 
 ing clockwise moments positive) are respectively 
 
 M 3 =- T(h -ho) -Sb =H(h ~ho) = - 
 
 (62J) 
 
 M 2 = Th Q Sb 2wh = Hh because 
 -Sb-2w(h-ho)=0 
 The moments mi, m 2 and m^ thus become 
 
 mi=Mi + ( ^- - J x=H(x +hoh) with origin at A 
 \ h I 
 
 ( ^-r 
 
 -\x=H(hho)(-rl} with origin at A 
 
 ~with origin at C 
 
 (62K) 
 
 Substituting these values into Eq. (62s) then 
 2# 2 C h , 
 
 W =ETJ O (x+h ~v 
 
 which gives by integration, 
 
 - dx 
 
 (62L) 
 
 Since &o must be so chosen that the internal work will be a minimum, this particular 
 value may be obtained by equating to zero the differential derivative of W with respect 
 to ho, thus obtaining 
 
 ^ W h SRI QM _i_ b (i> i \ j_^ & n 
 5r- =7-(6rto 3ft) +- r (h -hi) +-j- =0, 
 
 from which 
 
 b_ Oh b' 
 
 /3 /I /2 
 
 (62M) 
 
ART. 62 
 
 249 
 
 Substituting this value of ho into Eqs. (62X), will give the moments at all points 
 of the frame and from Eqs. (62j) the stress S and the moments at the ends of the post 
 may be found. 
 
 Example. For the cross frame of the first example and a horizontal wind load of 
 Ibs. the following values are obtained: 
 
 oo 
 2S 
 
 3X28 
 (XOl09 
 
 0.0547 / 
 
 17 6X28 17_ 
 0.736 0.0109 0.0547 
 
 - 14.25 ft. 
 
 Substituting values into Eqs. (62j), then, 
 
 Mi = -H(h-ho) = -3100(28-14.25) = -42,625 ft.-lbs. 
 MO =Hho =3100 X 14.25 -44,175 ft.-lbs. 
 _ 2w(h -ho) _ 6200(28 - 14.25) 
 
 b 17 
 
 This gives for the stress at the bottom of the post AC, 
 
 = -5197 Ibs. 
 
 M y_S 
 ' /i F 
 
 44175X12X5.5 5197 , 100ftn . 
 
 ~T4~7 = 12,899 3o3 Ibs. per sq.m. 
 
 226 
 
 For a braced cross frame with rigid post connections Fig. 62r, the analysis is con- 
 ducted in precisely the same manner as in the previous case, making as before, H =w = T. 
 
 /t 
 
 Counting clockwise moments positive, then the moments of the external forces about 
 the points E, C and are respectively 
 
 M l = -T(h l -ho) = -H(h l -ho) ' 
 
 M = Tho=Hho .... , -. .. . . (62N) 
 
 -Sb-2w(h-ho)=0 
 
250 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIII 
 
 The moments at any point x from C then become: 
 
 for Cl, Mo + ( Ml ^ M -}x = H &o -*) , 
 
 i,\ 
 
 (62o) 
 
 for EA, Mi (x -h,) =H(h l -h ) (-}, 
 
 e \ e / 
 
 for CD, Mo ~x =HhJl ~ . 
 
 These values inserted into the general work Eq. (62B) give for the frame 
 
 i . 
 --- dx+-- 
 
 e 
 
 from which is obtained after integrating 
 
 i *e^^H 
 *o) q- + Q 
 
 o J o 
 
 Taking the first differential derivative with respect to /? and equating the same 
 to zero then 
 
 3fto ^i 
 from which 
 
 MM 
 0= - .............. ^ ^ 
 
 Example. Given a cross frame as per Fig. 62r, for which ft =28 ft., 7^ =19. 5 ft., 
 6 = 17 ft., e=8.5 ft., /i =226 in. 4 , 7 3 = 15262 in. 4 and w>=3100 Ibs. to find /% M and M l . 
 From Eq. (62p) 
 
 19.5(56 + 19.5) 
 
 ho= - iT><226 = 10 - 96ft -' 
 2(28+39)+^ 
 
 and from Eqs. (62N) 
 
 MI = -3100(19.5-10.96) = -26,474 ft.-lbs. 
 M =3100X19.96 =33,976 ft.-lbs. 
 
 ^=-^^0) = -6215 Ibs. 
 6 
 
 The maximum stress in the post ~AC is then 
 
 My _S^ _ 88976X12 XBJ _^18 _ 9928 _ 423 Ibs. per sq.in. 
 /i /'ac 226 14.7 
 
 The stresses in the top struts and diagonal are not so easily determined when the 
 double latticed type of bracing is used. In the present case with one diagonal AF, the 
 
ART. 03 SECONDARY STRESSES 251 
 
 stresses are determined by passing a section tt through the three members and the point 
 of counterflexure, and taking moments about A, then, 
 
 P-^T 3100(28-10.96) 
 
 _T(hhv) -eEF=Q, or EF=- ^ - = -62i5 Ibs. 
 
 o.o 
 
 The stress in AJ3 is found by taking moments about F, obtaining, 
 
 - T(hi -h ) -Sb -we +eAB =0, 
 whence 
 
 T ( hl _^) + Sb +we 3100(19.5 -10.95) -6215X17 +3100X8.5 = 
 AB= - T 8.5 
 
 Also from the vertical shear on the section 
 
 ji a =0, or TF = --^- = +^a = !3,842 Ibs. 
 
 In concluding the subject of cross frames it might be added that good designs should 
 aim at deep floor beams and slender vertical posts, as is clearly indicated by Eq. (62n) , 
 which shows that M is diminished when 7 3 is increased and increased when I r is increased. 
 
 It has been stated by Mr. Grimm, " Secondary Stresses in Bridge Trusses," p. 80, 
 that the assumption of fixed connections between floor beams and posts is not verified 
 by investigations. The author suggests an explanation for this by calling attention 
 to the fact that recent tests indicate that compression members as formerly built are 
 not nearly as stiff against buckling as was supposed and furthermore, a slight initial 
 deformation approaching the elastic curve which the stressed member might attain, 
 would almost obviate secondary stresses in the cross frames. That such deformations 
 actually exist, or might be produced in overloaded posts by a permanent set, there can 
 be little doubt, and hence it is quite easy to understand why some of the high theoretical 
 stresses do not appear to exist when the actual stresses are measured with mstrum 
 
 ART. 63. SECONDARY STRESSES DUE TO VARIOUS CAUSES 
 
 (a) Bending stresses in the members due to their own weight. The maximum bending 
 moment, in any member, resulting from its own weight, when supported on fnctionless 
 pin bearings at the ends, will occur at the center of the member and i 
 
 (63A) 
 
 where I is the length of the member, p its weight per unit of length and 6 is the angle 
 which the member makes with the horizontal. 
 
 WhenThe member is fixed at the ends and there is no direct stress the bendmg moments 
 M at the ends of the member, and M c at the center, become respe, 
 
 and 
 
252 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIII 
 
 In any case these moments produce a unit stress / on the extreme fibers of the mem- 
 bers which is given by Navier's law, Eq. (49M) as 
 
 (630 
 
 where y is the normal distance from the neutral axis to the extreme fiber. 
 
 . It is thus seen that M increases as the square of I while / is inversely proportional 
 to 7. Hence, long members of small moment of inertia may receive severe secondary 
 stress due to their own weight. 
 
 When the member is in compression or tension, the combined bending stress on the 
 extreme fiber, due to the direct stress and the uniform load, must be investigated. This 
 cannot be accomplished by algebraic summation of the separate bending effects, because 
 axial compression increases the deflection due to cross binding, while axial tension dimin- 
 ishes this deflection. The bending stress, resulting from the simultaneous loading must, 
 therefore, be found. 
 
 The analysis for riveted end connections, while very complicated, is not usually 
 necessary. For pin-connected members where the cross bending effect is always more 
 severe, the following approximate solution is given. 
 
 A horizontal compression member, with centric pin connection at each end, is shown 
 in Fig. 63A. Let M c designate the maximum bending moment which, in this case, occurs 
 at the center of the member. Calling p the weight of the member per unit of length, 
 S the axial stress, I the length of the member and 3 the deflection at the center, then 
 
 I ,'.'.-.. . . (63D) 
 
 _ = _ 8Sd 
 ~48ET~3MEI P 
 from which 
 
 Assuming the effect of the direct stress on the deflection d to be the same in character 
 as that of the uniform load p, which is not quite true, then 
 
 (63E) 
 
 (63F) 
 
 48EI -5S1 2 ' 
 This value of d substituted into Eq. (63D) gives 
 
 6pm 
 
ART. 63 SECONDARY STRESSES 253 
 
 The combined direct and bending unit stress on the extreme fiber of the member 
 then becomes by Eq. (49M) 
 
 (030) 
 
 where F is the cross-section and y is the normal distance from the neutral axis to the 
 extreme fiber of the member. 
 
 Horizontal tension members with pin-connected ends, when similarly treated, lead 
 to the following formulae: 
 
 , 6pPE7 
 " "* M <=48W+ (63H) 
 
 The combined direct and bending stress on the extreme fiber is again found by 
 
 Eq. (63o). 
 
 Example- Given an eye bar 2 X 15 in., making F =30 sq.in.; 5=600,000 Ibs.; Z=660 
 in.; 7=562.5 in. 4 ; p=8.5 Ibs. per in.; and E= 29, 000 ,000 Ibs. per sq.in. 
 
 Then from Eqs. (63n), 3 =0.4824 in.; and M e --= 173 ,450 in. Ibs. The stress on the 
 extreme fiber then becomes by Eq. (63c) 
 
 ,600,000 173,450X7.5 =2313 lbg . 
 
 J 30 562.5 
 
 showing that the bending stress alone increases the tension in the lower fiber by 11.5 
 
 per cent. 
 
 (b) Secondary stresses in pin-connected structures. In all usual stress computations 
 the question of how equilibrium is established between the stresses in the several mem- 
 bers meeting at a pin or panel point is not considered. It is thus assumed that the 
 individual members extend to the pin centers with full effective sections where the stres 
 are suddenly balanced in a point. 
 
 In reality the case is very different and no such balance in the stresses can be accom- 
 plished The nearest approach to a realization of the ideal condition is in the case of two 
 abutting compression members and then only when there are no bending stres 
 
 be transmitted. ,, , , 
 
 When all the members are connected by pins according to the usual methods of 
 construction, the stresses are transmitted past the panel points in very much the same 
 manner as for riveted connections on account of the factional resistance on the pins 
 created when the structure is distorted by superimposed loads^ The advantages uu^ 
 claimed for pin connections are not all realized in pract.ce and wh, e cent connections 
 are best made by this stvle of panel joint there may be suffic.ent fnctional resistance on 
 the pin to produce secondary bending stresses qnite equal to those occurring m rive. 
 
 is thus an erroneous idea to regard a pin-connected structure as free from secondary 
 
254 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIII 
 
 stresses. This is never so and the rnalysis may be even more complicated, though the 
 results are probably more reliable than for riveted structures. 
 
 The exact behavior of a member with pin bearings depends on many circumstances, 
 especially the diameters of the pins. When these are sufficiently large the member can- 
 not turn under load, and secondary stresses must result. These may be partially relieved 
 by the vibratory action of the live loads which would probably allow the structure to 
 adjust itself for some mean condition of loading and reduce the secondary stresses to a 
 minimum. However, temperature changes would enter as a disturbing element to pre- 
 vent such favorable action from establishing permanent relief. 
 
 The frictional resistance offered by a pin connection is now analyzed. 
 Let S =the stress in any member. 
 
 R =the total frictional resistance on the surface of contact between the pin 
 
 and the member. 
 r=the radius of the pin. 
 ^o the angle of repose, making tan ^o = ( a=the coefficient of friction between 
 
 the two metals. 
 
 c =the eccentricity of S required to resist the moment of the frictional force R. 
 e=the actual eccentricity of S required to resist the bending moment M a due 
 
 to secondary stress. 
 
 The line of stress S, Fig. 63s, is supposed to suffer a displacement c from the pin 
 center A of an amount which will make the moment Sc exactly equal to the opposing 
 moment of the frictional resistance R. 
 
 FIG. 63B. 
 
 For a frictionless pin the eccentricity c would be zero and the stress would pass through 
 the pin center. 
 
 Hence from the figure 
 
 Sc=Rr and c=r sw.p, ......... (63i) 
 
 showing that c is independent of the stress S. 
 
 When c>e no rotation will take place as a result of deformation of the truss due to 
 load effect, and when c < e the member will turn on the pin and the full amount of bend- 
 ing cannot be developed by the frictional resistance. 
 
 The secondary stresses in a pint-connected structure are found in precisely the same 
 manner as shown in Art. 61, for riveted connections, except that for all members where 
 
ART. 63 
 
 SECONDARY STRESSES 
 
 255 
 
 c<e the limiting value c will govern and the secondary stress becomes accordingly less. 
 Eq. (61w) gives the value of e=QKI/Sl, furnishing the limiting value 
 
 ^ Sic Sir sin p 
 = = ~ 
 
 and by Eq. (61v) the limiting secondary stress becomes 
 
 f _M a y_Syrsmp 
 Js j- j 
 
 (DDK) 
 
 Regarding the values of p, which are purely empiric, no good experimental data 
 seem to be available, though it is preferable to choose rather high values from 12 to 14, 
 making sin p =0.2 to 0.25. 
 
 (c) The effect of eccentric connections. It frequently happens that the line of stress 
 through a member is not coincident with its neutral axis, thus giving rise to eccentric 
 connections. 
 
 It might be said that such connections should never be tolerated, and hence their 
 effect in producing secondary stresses would not require investigation. However, there 
 may be rare cases where eccentricity in unavoidable and proper provision must then be 
 made for taking care of the bending stresses thereby produced. 
 
 These bending stresses are not merely additional stresses, but the secondary stresses 
 previously found on the assumption of centric connections will be incorect on account 
 of the changes which must take place in the deflection angles r and the angles a, (3, j-, 
 etc., between the truss members, as a consequence of the eccentricity. 
 
 Calling c a and c b the eccentricities of the stress at the two ends of a member AB, Fig. 
 63c, then the moments M a =S ab c a and M^S^ will produce deflection angles r by 
 Eqs. (61c), which become 
 
 I t^tr n if \ ^db/' 
 
 (63L) 
 
256 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP, xill 
 
 The values Er a and Er b , resulting from eccentricity, are now incorporated with their 
 proper signs into the values EJa, and EA$ of Eqs. (6lN, o, p) and the secondary stresses 
 are' then computed in the ordinary way as described in Art. 61. 
 
 Finally, the stresses f 8 on the extreme fibers must be combined with the direct stresses 
 / resulting from the bending moments SabC a and SbaC b . It should be observed that the 
 stresses f s and / frequently have opposite signs, in which case the eccentricity effect may 
 partially neutralize the secondary stress due to riveted connections. 
 
 It is sometimes important to investigate the effect due to the eccentricity of the 
 end supports where the friction of the expansion bearings may produce considerable 
 bending stress in the end members. Good designs will, however, obviate any serious 
 stresses from this source. 
 
 (d) Effect of temperature and erroneous shop lengths. The manner of dealing with 
 these effects in the case of indeterminate structures has received attention in previous 
 chapters. For determinate structures having end supports and pin. connections absolutely 
 frictionless, a small change in geometric shape, due to temperature or erroneous shop 
 lengths, would have no appreciable effect on the stresses in the members. 
 
 However, when the panel connections are not frictionless then any effect, as unequal 
 temperature, etc., which is productive of a deformation of the structure, will also create 
 secondary stresses. 
 
 The unit stresses / which result from any deformation of the structure due to any 
 changes Al in the lengths of the members, may be found from the proportion 
 
 Al f d 
 
 whence f=jE=Eet.-'. . . ._ . .' . . . (63M) 
 
 Knowing the stresses / for all the members of the structure due to any simultaneous 
 condition of temperature or loading, the values EAa and K, and the consequent secondary 
 stresses, may all be found in the manner described in Art. 61. 
 
 ART. 64. ADDITIONAL STRESSES DUE TO DYNAMIC INFLUENCES 
 
 (a) Forces acting longitudinally on a structure. The moving loads in passing over a 
 structure exert a certain horizontal impact which is transmitted through the floor system 
 to the main trusses. When the brakes are set the amount of this impact may become 
 considerable arid increases with the number of braked wheels on the span and .the mag- 
 nitude of the moving loads. The forces thus applied to the structure are called tractive 
 forces and produce certain additional stresses in the truss members. 
 
 The effect which the tractive forces exert on the trusses depends on the relative posi- 
 tion of the roadway to the points of support. For a bridge, supported on the level of the 
 loaded chord, all the additional stresses are confined to this chord and the floor system. 
 When the supports are on some other level then all the truss members will be stressed, 
 
ART. 64 
 
 SECONDARY STRESSES 
 
 2:,7 
 
 and this gives rise to a great variety of cases depending on the type of truss considered. 
 Two cases will be presented for illustration. 
 
 Let 7"= the tractive force on one rail under any wheel load P. 
 
 /=the coefficient of friction between the braked wheel and the rail, usually 
 taken between 0.15 and 0.2, and varying greatly according to weather 
 conditions. 
 
 Then the total tractive force, for n wheels on the bridge, becomes 
 
 (64A) 
 
 For a through bridge, the compression in the bottom chord members will then be 
 increased by the tractive force producing the additional stress 2 X 1 T, where x is the dis- 
 tance from the fixed end of the span to the head of the train. 
 
 All top chord and web members remain unaffected. When the train approaches 
 from the roller end the bottom chord receives compression and the stress increases uni- 
 formly from the roller end toward the fixed end. 
 
 For a deck bridge, as shown in Fig. 64A, the reactions due to a tractive force 2T 1 
 becomes: 
 
 A=jST; B=-j2T; H = VT (64 B ) 
 
 FIG. 64A. 
 The additional stresses in the members of panel 1-2 are found as follows: 
 
 For top chord, 
 
 17=*- 
 
 fi 2 
 
 h 2 
 
 I T 
 
 V" '*2/ ^X-*- 
 
 I or bottom chord E*- 
 
 M, Ax, -g(fto-e)-eSg!T 
 
 cog 
 
 For diagonal 
 For vertical 
 
 D = (A-Lsin X) sec 6 
 F=A-Lsin A 
 
 (64c) 
 
 The angles Jl and are regarded positive as shown in the figure. ^ The tractive 
 force is taken positive in the direction from right to left, supposing the tram to appn 
 the span from the roller end at B. 
 
 The maximum value for the additional stress is found for the same position o 
 train loads as is used to produce maximum primary stress in any particular member. 
 
258 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIII 
 
 These additional stresses become less important for long span, massive, structures 
 where the dead load is large in proportion to the live load. For smaller structures, the 
 chord members, adjacent to the hinged end of the span, may receive excessive additional 
 stresses requiring increased area or provision for reversal of stress. 
 
 The effect of tractive forces on the floor system is not taken up here. 
 
 When the loaded chord is on a grade, the total weight G of the combined dead and 
 live loads on the span exerts a certain component G sin a longitudinally along the struc- 
 ture and in a down grade direction, where a is the grade angle. This component is trans- 
 mitted to the hinged support and should be considered in designing this bearing, especially 
 if the support happens to be a steel tower. 
 
 (b) Forces acting transversely to a structure. They are wind pressure, lateral 
 vibrations and centrifugal force on a curved track. These will be separately discussed. 
 
 Wind pressure is the force due to wind against the total area of all structural elements 
 of both trusses, the floor system and live loads, as presented in elevation. While the 
 wind may have any direction, it is presumed that at times this direction will become 
 normal to the exposed area and thus attain maximum intensity. 
 
 Experiments have shown that the normal wind pressure may reach a value of 50 
 Ibs. per sq.ft., which amount is assumed in figuring the wind stresses in the unloaded struc- 
 ture. 
 
 Usually it would be unwise to cross a bridge when the wind has sufficient velocity 
 to produce such pressure, and hence the maximum pressure to be assumed for the loaded 
 structure need not exceed about 30 Ibs. per sq.ft., applied to the total area of the struc- 
 ture, the moving load, and the portion of the leeward truss which is not covered 
 by the moving load. While both trusses may not receive the full intensity of pressure 
 it is customary to make no deduction on this account. 
 
 The wind loads falling on the truss members are considered as concentrated loads 
 acting on the several pin points of the structure, and in estimating the stresses in 
 the web members of the horizontal wind trusses or lateral trusses, these loads are treated 
 as moving loads. The wind pressure on the live load is always considered as a moving 
 load applied to the wind truss of the loaded chord. The total maximum wind load will 
 govern for the chord stresses. 
 
 The end reactions of the wind trusses must eventually be carried to the main truss 
 supports through the end portals or sway bracing. Sometimes only one wind truss is 
 used, and then the wind loads of the other chord are carried to this wind truss by 
 sway bracing at each panel point, throwing the entire wind load on the one lateral system. 
 
 Since the center of gravity of the wind load on the moving load is always above the 
 plane of the lateral system of the loaded chord, it is necessary to consider the overturn- 
 ing effect of these loads in producing unequal vertical loading of the two main trusses. 
 Figs. 64B show five cases which are met with. In Figs, a, b and c one wind truss is pro- 
 vided and a sway brace at each panel, while in Figs, d and e there are two wind trusses. 
 
 Let Wi =the wind pressure per panel on the live load. 
 
 hi =the lever arm of w\ to the plane of the horizontal wind truss. 
 
 Wz =the wind pressure per panel on the bridge. 
 
 Ji2=the lever arm of w 2 to the plane of the horizontal wind truss. 
 
ART. 04 
 
 SECONDARY STRESSES 
 
 2o9 
 
 Then the overturning moment is 
 
 +w 2 h 2 in Figs, a and b } 
 
 fc TW i' with one 
 
 w 2 h 2 in l H ig. c 
 
 in Figs, d and e with two wind trusses 
 
 truss 
 
 . 64n) 
 
 and the vertical load V, acting down on the right-hand truss and up on the left-hand truss 
 becomes V=M/b. Depending on the direction of the wind, this leads to positive and 
 negative stresses of equal intensity in each vertical truss. Hence, V is treated as a 
 live load for the vertical truss to obtain the additional stresses due to overturning effect 
 
 rr 
 h, 
 
 i 
 
 
 
 t f 
 
 T^ 
 
 
 (a) 
 
 (c) 
 
 (d) 
 
 FIG. 64s. 
 
 of the wind on the moving live load, though it might be better to determine the stresses 
 separately for the wind loads Wi and w 2 . Calling V\ and V 2 the respective vertical wind 
 loads, pd the live load per panel and gd the dead load per panel, then for parallel chords 
 the additional wind stresses become SdVi/pd+SiV 2 /gd, where Sd is the dead load and 
 Si the live load stress in any member. 
 
 It should be noted that the case in Fig. c is most favorable, while the one in Fig. b 
 is most unfavorable in the production of loads V, and may give rise to stresses amounting 
 to as high as 30 per cent of the primary stresses in certain members. 
 
 The stresses in the wind trusses may be found by applying the methods given in 
 Chapter XII, and those occurring in the portal bracing and sway system were analyzed 
 in Art. 62s. 
 
 All of these additional stresses resulting from wind pressure in the primary members 
 must finally be combined with the primary stresses to obtain total stresses. 
 
260 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIII 
 
 The additional wind stresses increase in importance with the length of the span, 
 and may assume enormous proportions in large structures. It should be mentioned, 
 however, that the frequency in the occurrence of maximum wind stresses is extremely 
 rare, and their simultaneous combination with maximum primary stresses is very 
 improbable. 
 
 In view of these considerations it would seem proper to emplo}^ higher allowable 
 unit stresses, say such as are permissible for quiescent loading, possibly restricting the 
 total sections of the primary members more nearly to those required only for maximum 
 dead and live loads. For single-track bridges exceeding a span of 200 ft., and for double- 
 track structures of over 350 ft. .span, the wind stresses must be fully considered, while 
 for smaller structures the end posts and end chord sections should be appropriately strength- 
 ened, making perhaps little allowance in the other members. 
 
 Lateral vibrations, while they exist, do not appear to produce any serious increase in 
 the primary stresses of a bridge. This is certainly true for highway structures, and in the 
 case of railroad bridges, this effect may be reduced to insignificance by making appropriate 
 provisions to keep the moving loads steady while passing over the span. 
 
 The magnitude of the vibratory forces here considered depends on the velocity of 
 the moving load, the character and type of locomotive, the condition of the track and 
 the relative stiffness and rigidity of the floor system and main trusses. Short spans 
 and narrow structures, also skew bridges might receive considerable vibratory stress, 
 especially in the end panels. Coned wheels and loose or spread rails are very objectionable. 
 
 It is needless to say that factors of such uncertain constitution and variable nature 
 are not susceptible to analysis, and a direct measurement (even if it could be made) of 
 such additional stresses would merely indicate what might occur in certain cases, but 
 the results from different experiments could not be accurately correlated with the 
 surrounding circumstances. 
 
 The remedy thus lies in obviating, so far as possible, the conditions producing dele- 
 terious effects rather than in any attempt to estimate their magnitude and weigh down 
 the structure by allowing extra material in the required cross-sections. If this can be 
 done, then the slight additional stresses due to lateral vibrations may safely be neglected. 
 The vertical effect of impact is considered elsewhere in this article. 
 
 Centrifugal force due to curved track. This problem necessarily deals only with rail- 
 road bridges and includes the centrifugal force acting transversely to the structure and 
 producing stresses in the stringers and in the lateral system of the loaded chord; the 
 overturning effect of the centrifugal force producing stresses in the stringers, floor beams 
 and in the main trusses; and the unequal distribution of the vertical loads on the floor 
 and between the main trusses due to the eccentricity of the curved track axis with the 
 straight bridge axis. This latter problem was solved by influence lines in Art. 29, for 
 the case of a skew plate girder on a curve, and will receive no further attention 
 here. 
 
 The amount of the centrifugal force acting transversely to the bridge is given 
 by the formula 
 
 C=^-P=cP, (64B) 
 
 gR 
 
ART. (54 
 
 SECONDARY STRESSES 
 
 261 
 
 where v is the velocity of the moving train; P is any moving load on the bridge; 
 #=32.2 ft. per second, being the acceleration due to gravity; and R is the radius of 
 curvature of the track. The units are foot, pound, second. For velocities in miles 
 per hour, and curvature in degrees D, c =0.00001 \7v 2 D. 
 
 The following table gives values of c=v 2 /gR for curves of one to ten degrees and 
 velocities which may be regarded as maximum for such curves. 
 
 Degree of Curve, 
 D 
 
 v Miles per 
 Hour. 
 
 r 
 
 Degree of Curve, 
 D 
 
 r Miles per 
 Hour. 
 
 c 
 
 1 
 
 60 
 
 0.042 
 
 6 
 
 50 
 
 0.176 
 
 2 
 
 58 
 
 0.078 
 
 7 
 
 48 
 
 0.189 
 
 3 
 
 56 
 
 0.110 
 
 8 
 
 46 
 
 0.198 
 
 4 
 
 54 
 
 0.136 
 
 9 
 
 44 
 
 0.204 
 
 5 
 
 52 
 
 0.158 
 
 10 
 
 42 
 
 0.206 
 
 In computing the additional stresses S c in the main truss members resulting from 
 centrifugal force, the maximum total stress might be obtained from the train giving 
 the maximum primary stresses rather than from the train moving with the highest 
 velocity. For short spans the maximum combination is likely to occur from passenger 
 trains, while for long spans this may be true for freight trains moving at lower 
 speed. 
 
 The methods of finding the stresses in the main trusses and in the lateral system 
 of the loaded chord are precisely the same as given for wind loads. 
 
 Since the centrifugal force C is a linear function cP of the moving load, therefore 
 the horizontal forces coming on the lateral truss of the loaded chord will, at all points, 
 be c times the moving load. The same is true of the overturning effect separately 
 considered, where the line of action of C is above the plane of the lateral truss and 
 passes through the center of gravity of the load P, which 
 may be assumed at the same level as the center of pressure 
 of the wind load. 
 
 The stresses in the lateral truss of the loaded chord will 
 then be found in the ordinary way for loads cP considered 
 as moving live loads. When these loads P are uniformly 
 distributed then the stresses found for loads cp will ^be 
 cp/w times those found for the uniform moving wind 
 load iv. 
 
 The stresses in the main trusses, due to overturning 
 effect of the centrifugal force C, are analyzed in the manner 
 previously explained for wind loads. Fig. 64c represents the 
 condition at a panel point assuming the load P and the 
 centrifugal force C to be loads per panel and applied at the FIG. 64c. 
 
 center of gravitv of the moving load. 
 
 The track is eccentric with the truss and the floor beam is superelevated to suit 
 
262 KIXETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIII 
 
 the curved track. The resultant K, of P and C, intersects the plane of the lower 
 laterals at a distance e from the truss axis and may there be resolved into forces P 
 and C =cP as they would act on the main and lateral systems. 
 
 The reactions Vi and V 2 , representing the vertical loads which are transmitted 
 to the main trusses, then become 
 
 P 
 and Vi= 
 
 With these loads Fj and V^, which vary for each panel point on account of the 
 variable eccentricity e, the method given in Art. 29 may be advantageously used to 
 find the combined stresses in the main trusses by influence lines and that is the best 
 solution for this problem. It should be noted that the load Vi becomes maximum 
 for quiescent loading, while V% attains maximum for the highest velocity of the 
 train. 
 
 (c) Forces acting vertically on a structure. This embraces several causes which 
 together exert a more or less severe dynamic influence on the stringers, floor beams 
 and vertical trusses of a bridge, all included under the general term impact. 
 
 So far as these can be separately enumerated, they may be classified as follows: 
 (1) The true impact due to the instantaneous application of a moving load to a structure 
 at rest; (2) the impact resulting from the unbalanced condition of the engine drivers; 
 irregularities in the surfaces of the rails and imperfections in the rolling stock, etc.; 
 (3) elastic vibrations of the structure which tend to increase the effect of dynamic 
 loads otherwise applied to a rigid body; and (4) vertical centrifugal forces occasioned 
 by the vertical curvature (deflection and camber) of the track. 
 
 Owing to the extremely variable conditions and combinations of circumstances r 
 it is practically impossible to analyze separately the magnitudes of these several causes 
 and their resulting effects, though a brief discussion may serve to point out precau- 
 tionary measures tending to reduce the impact stresses in bridges under traffic. Some 
 very valuable tests have been conducted during late years, both in Europe and America, 
 which give a very good idea of the sum total effect produced on railroad bridges by 
 the moving loads, and from these, formulas have been deduced which make it possible 
 to provide suitable strength in designing bridges. 
 
 The additional stresses, resulting from this combination of causes, are a function 
 of the velocity and magnitude of the moving load, and hence assume minor importance 
 when dealing with highway structures. 
 
 The true impact due to the instantaneous application of live loads is really not a 
 serious source of additional stresses, because forces are transmitted through a structure 
 with a rapidity approaching the velocity of sound, while the moving train could not 
 exert its effect in so short a space of time. 
 
 Theoretically, a load instantaneously applied would produce a dynamic effect 
 twice as great as the static load, but in reality this cannot take place and there will 
 
ART. 64 SECONDARY STRESSES 263 
 
 always be a sufficient lapse of time between the application of the loads :md tin- 
 necessary elastic deformation of the structure to Obviate to a great extent this extreme. 
 dynamic impulse. The vibrations accompanying the moving loads really constitute 
 the element of danger. 
 
 The most unfavorable conditions prevail in short span structures and in the wd> 
 members subject to a rapid change of stress. 
 
 The causes enumerated under (2) and (3) are really of the most serious character, 
 and while much may be accomplished by a careful maintenance of the track and 
 rolling stock, a certain average condition will usually prevail beyond which no remedial 
 measures will be possible or feasible. 
 
 In point of design and construction of a railroad bridge the following suggestions 
 may be offered: The bridge and its approaches back for some distance should be 
 straight and when curves are absolutely unavoidable speed restrictions would seem 
 proper. The rails should be long and the joints and tie connections of the best and 
 most rigid construction, carefully maintained. The connections at the abutments 
 require the most attention, avoiding uneven settlements and loose rails. The intro- 
 duction of continuous roadbed over bridges is very desirable as affording some elasticity 
 to absorb the impact rather than to transmit the same to the structural portions of 
 the floor and trusses. A similar effect is produced by long ties over widely spaced 
 stringers. Very rigid floors and rails directly on stringers may be classed as 
 objectionable. 
 
 The greater the proportion of dead to live load and the longer the span, the less 
 will be the effect due to impact. 
 
 The avoidance of synchronous vibrations between the moving load and the 
 structure is of the utmost importance and may be practically accomplished by properly 
 stiffening all the members and by providing thorough bracing in the lateral and sway 
 systems. 
 
 The effect of vertical centrifugal force might be separately estimated, but the 
 amount is small and the attending conditions are too uncertain to warrant 
 this. 
 
 As early as 1859, Gerber proposed a general formula providing a certain reduction 
 in the allowable unit stresses to cover these several sources of dynamic impact. This 
 formula is 
 
 wherein /' is about the elastic limit of the material and S t and S d are the respective 
 live and dead load stresses in any member. 
 
 The following formula are at present in use and give the factor by whic 
 live load stress should be multiplied, so as to include the dynamic effects considered 
 under the present, heading. Calling S t the live load stress, S d the dead load stress 
 and I the strength of span for chord members, or the loaded distance producing 
 maximum stress in a web member, then <p is. given as follows: 
 
264 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP, xill 
 
 (1) C. C. Schneider, 1887, railroad bridges, 
 
 , _ 1 300 
 
 1+300 1+300 
 
 (2) C. C. Schneider, electric roads, 
 
 ,_ t 150 _l+4oO 
 1+300 ~r+300 
 
 (3) J. A. L. Waddell, railroad bridges, 
 
 , 400 _/+900 
 
 * Z+500 "Z+500 
 
 (4) J. A. L. Waddell, highway bridges, 
 
 ,_ 1 100 _Z+250 
 ^ Z + 150 ~r+150 
 
 (5) H. S. Pritchard, 1899, railroad bridges, 
 
 , Si 2S t +S d t 
 
 Y = + o . o = o , o ; for counters 0=2 
 
 (6) J. Melan, railroad bridges, 
 
 24 
 
 _ 
 +30 1+30 
 
 (7) J. Mdan, secondary roads, 
 
 (8) F. Engesser, railroad bridges, 
 
 
 (9) F. Engesser, highway bridges, 
 
 fgg _ A 2 
 
 ^ = l-50+ y 3300 ; for Z>33, = 1.50 
 
 (10)* American Railway Eng'r and Maintenance of Way Ass'n, 1910, 
 
 , _l 2 +40,000 
 * 1 2 + 20,000 
 
 Table 64A gives values of as obtained from each of the above ten formula for 
 a few even lengths I. 
 
 Only those values of are comparable which are intended for the .ame cla of 
 structures as (1), (3), (5), (6), (8) and (10). It is clearly seen that formula (lO),ives 
 the highest values for short spans I and the lowest values for long spans, noting that in 
 
 '. . . (64a) 
 
 * Where I is the span length and all members receive the same impact. 
 
ART. 64 
 
 SECONDARY STRESSES 
 
 265 
 
 this formula I represents the length of span and not the loaded distance as in formula? 
 (1) to (9). 
 
 VALUES OF 
 
 TABLE 64 A. 
 AS FOUND FROM THE VARIOUS EQS. (64c) 
 
 I Feet. 
 
 (l) 
 
 (2) 
 
 (3) 
 
 (4) 
 
 (5) 
 
 (6) 
 
 (7) 
 
 (8) 
 
 (9) 
 
 (10) 
 
 10 
 
 1.97 
 
 1.48 
 
 1.78 
 
 1.63 
 
 Cannot 
 
 1.73 
 
 1 .55 
 
 1.97 
 
 1.66 
 
 1.99 
 
 50 
 
 1.86 
 
 1.43 
 
 1.73 
 
 1.50 
 
 be 
 
 1.44 
 
 1.33 
 
 1.69 
 
 1.50 
 
 1.89 
 
 100 
 
 1.75 
 
 1.38 
 
 1.67 
 
 1.40 
 
 solved 
 
 1.32 
 
 1.24 
 
 1.67 
 
 1.50 
 
 1.67 
 
 200 
 
 1.60 
 
 1.30 
 
 1.57 
 
 1.28 
 
 without 
 
 1.24 
 
 1.18 
 
 1.67 
 
 1.50 
 
 1.33 
 
 300 
 
 1.50 
 
 1.25 
 
 1.50 
 
 1.22 
 
 knowing 
 
 1.21 
 
 1.16 
 
 1.67 
 
 1.50 
 
 1.18 
 
 400 
 
 1.43 
 
 1.21 
 
 1.44 
 
 1.18 
 
 the 
 
 1.20 
 
 1.14 
 
 1.67 
 
 1.50 
 
 1.11 
 
 500 
 
 1.38 
 
 1.19 
 
 1.40 
 
 1.15 
 
 stresses 
 
 1.18 
 
 1.13 
 
 1.67 
 
 1.50 
 
 1.07 
 
 600 
 
 1.33 
 
 1.17 
 
 1.36 
 
 1.13 
 
 
 1.18 
 
 1.13 
 
 1.67 
 
 1.50 
 
 1.05 
 
 
 
 
 
 
 
 
 l 
 
 
 
 The American Railway Engineer and Maintenance of Way Association formulae 
 is undoubtedly entitled to the greatest confidence, being based on very extensive 
 experiments which were carried out by the committee on impact tests under actual 
 conditions of traffic. The formula is not as yet officially adopted by the association. 
 
 However, no allowance was made for secondary stresses in the members and hence 
 the formula (10) may be said to include more than actual impact effect which is 
 probably true of all the Eqs. (64o). 
 
 (d) Fatigue of the material. Based on the classic experiments of Wohler (1859-1870) 
 which were continued by Professor J. Bauschinger, a formula was proposed by Launhardt 
 (1873) and later modified by Weyrauch, aiming to apply the principles of the fatigue 
 of material to the design of bridge members. 
 
 Wohler's law states that for a large number of repeated load applications, rupture 
 of a material is produced under stress which is below the ordinary (static) breaking 
 strength of that material. The conditions under which Wohler's experiments were 
 made differ so radically from those attending the actual operating conditions of bridges, 
 that it is questionable whether anyone is justified in applying the Launhardt- Weyrauch 
 formula? to bridge designs. 
 
 Wohler's load repetitions followed in quick succession and were continued without 
 interruption (several million times) until failure was produced. Bridge members are 
 subjected to a repetition of stress which is always followed by a rather long period of 
 rest, and few structures, even under the heaviest traffic, are retained in service long 
 enough to receive several million applications of the moving load. Also, a well-designed 
 bridge is never calculated for stresses approaching e ven the elastic limit, while Wohler 
 bases all his observations on stresses exceeding this limit. 
 
 In addition to these facts, modern experiments made under conditions which 
 respond quite closely with bridge practice, though limited in extent, point uniformly 
 
 con 
 
 against the existence of fatigue in bridges. 
 
 The above mentioned Launhardt-Weyrauch formulae have been extensively used 
 
266 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XI I] 
 
 in designing bridge members and are still retained in many specifications. However, 
 this practice does not seem justifiable, especially when allowance is made for impact and 
 secondary stresses generally. 
 
 The fatigue formula undoubtedly served a good purpose in the days when so many 
 important factors were neglected, but at the present time the aim should be to make 
 proper allowance for all the known stress elements and thereby reduce the " factor of 
 ignorance " to a minimum. 
 
 (e> Unusual load effects may be produced by loads of unusual magnitude applied 
 at rare intervals or by loads which occur under unusual circumstances. 
 
 It is possible that the maximum combination of moving loads, wind pressure and 
 temperature effects may take place, under which the structure might be stressed to its 
 utmost capacity, while under ordinary conditions of traffic the stresses would be far 
 below those for which the design was made. The maximum load basis for a design is 
 then an unusual load effect for the average use of a structure. 
 
 At times it may be necessary to transport some abnormal piece of freight or a train 
 of locomotives, which would exceed the loads assumed in the computations. This may 
 be done without danger even if the members are stressed quite near the elastic limit, 
 provided the design is made with some forethought. For most ordinary structures such 
 a practice might prove very disastrous. 
 
 It should be observed that, while the stress in any member for a particular position 
 of a moving load is exactly proportional to the intensity of this load, the combination of 
 maximum and minimum stresses for such member, which fixes the required cross- 
 section, might not be tlms proportional. In the case of chord members, all governing 
 stresses are proportional to the loads because the critical positions of the loads are 
 alike for all chords, that is, for maximum chord stress the whole span is fully loaded 
 while the minimum chord stress occurs for dead load only. 
 
 Hence, the chord sections are directly proportional to the total moving load, while 
 most of the web sections are not so proportional and increase more rapidly than the 
 loads. This is particularly the case with all counter web members wherein the stress 
 is the difference between that produced by the moving load and that due to the dead 
 load. It also applies to the sections of such web members as are subject to stress reversal. 
 
 This is an important item in making provision for a future increase in train loads, 
 where 25 to 50 per cent additional carrying capacity might easily be secured by providing 
 a slight increase in the counters and web members having stress reversals. Many old 
 structures would still be usable for considerable overload had such provision been made 
 in their design. 
 
 The best way to provide for such overload in view of future increase in train loads, 
 is to design the structure for the given case of loading and allowable unit stresses and 
 then to increase the sections of the counters, and web members with stress reversals, 
 for the desired overload, which of course could not exceed a reasonable amount according 
 to the limitations imposed by the chord sections. 
 
 A structure so designed might then carry say 30 per cent overload without imposing 
 unduly on certain few web members, while the ordinary structure would fail by the 
 buckling of these members long before the chords had received any serious stresses. 
 
iART. 05 SECONDARY STRESSES 267 
 
 In the same way a highway bridge may be designed to carry an occasional overload 
 W a certain heavy road roller, etc., without increasing the sections of the chords and 
 [main web members, but by providing for the extra counter stresses due to the excess load. 
 
 To the second class of unusual load effects belong derailments, collisions with drift- 
 wood during times of high water and possible settlements of piers or abutments. 
 
 The absolute prevention of such occurrences may be classified with the impossible. 
 However, the ultimate destruction of a structure when they do occur may be safeguarded 
 by applying certain precautionary measures which should never be overlooked. 
 
 Thus the bridge floor should be made sufficiently strong to allow a derailed train 
 to pass over without breaking through, and the wheels should be guided between guard 
 rails or by other means to prevent collision with the main truss members, necessitating 
 certain clearances for through bridges to accomplish this. In the case of deck bridges 
 t may be advisable to utilize the top chord as a barrier or protection to prevent the train 
 'rom leaving the structure. 
 
 The design of the floor should, therefore, aim at the use of solid web, riveted 
 girders and braces in preference to built-up frames, and the soft, ductile varieties of 
 steel are more desirable than the harder, brittle varieties. 
 
 In cases where high water may reach the bottom chord, slender eye-bar members 
 ire decidedly objectionable. 
 
 ART. 65. CONCLUDING REMARKS 
 
 (a) Features in design intended to diminish secondary and additional stresses. The 
 
 'ollowing suggestions should be given careful consideration: 
 
 Skew structures and bridges on curves should never be built except in very rare 
 cases where no other disposition is possible. These types should be regarded as measures 
 )f last resort. 
 
 The axes of all members of a truss should be in the same plane and should intersect 
 .n a point at all connections. 
 
 Special attention should be directed toward a careful design of all connections with 
 i view of reducing the secondary stresses. Thick gusset plates and large diameter rivets 
 materially reduce the sizes of these plates and are thus desirable from this aspect. 
 
 The widths of members, in the plane of the truss, should not be chosen larger than 
 s absolutely required to secure proper connections and stiffness against buckling -in 
 compression members. It may thus be desirable to taper compression members from 
 ihe center toward both ends. 
 
 The cross-sections of members should be so chosen that the material is concentrated 
 as far from the neutral axis as possible, thus securing the largest moments of inertia 
 for the smallest over all dimensions. The cross form is thus the least advantageous, 
 while a square box form is most desirable. In rare instances the cross form may be 
 acceptable, when the width of a member is determined by other conditions. Secondary 
 stresses occurring simultaneously in the plane of the truss and in a cross frame are additive 
 in members of the box type while in the cross form they are not additive, a consideratK 
 which may become important at times. 
 
268 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIII 
 
 The secondary stresses may be reduced to a certain degree by shortening all tension 
 members and lengthening all compression members by amounts Al, determined for the 
 case of maximum total load. This practice is generally followed and provides a camber 
 in the unloaded structure which exactly disappears under the full load when the structure 
 assumes its true geometric shape. 
 
 The plane of the lateral system should coincide with that of the chords and the 
 plane of the floor should be as close as possible to that of the lateral system. 
 
 The web members of the lateral systems should present considerable stiffness against 
 buckling, making l/r not greater than 140 for all these members, whether in compression 
 or tension. 
 
 The end portals, or cross frames, should be heavy to carry the wind reactions and 
 dynamic effects to the supports. 
 
 Floor beams should be made deep to reduce secondary stresses in the cross frames- 
 When this cannot be done then flexible connections should be provided between the 
 truss members and the beams, a type which is best suited to large bridges, while riveted 
 connections are desirable for small spans. 
 
 The stringers should be made heavy and continuous and should be designed tot 
 transmit the tractive forces to the panel points of the loaded chord instead of to the floor 
 beams by inserting proper tie members between the stringers. Long ties over widely 
 spaced stringers tend to relieve impact vibrations. 
 
 The use of pin connections is not admissible for bridges of less than 200 ft. span, 
 and even in larger structures the advantages to be gained may not be so easily demon- 
 strated. Some benefits may be derived from pins for the web system but certainly not 
 for the compression chord. 
 
 According to prevalent practice in designing pins, the diameters are usually so large 
 that the friction produces secondary stresses quite equal to those resulting from riveted; 
 connections. Perhaps this condition might be remedied. 
 
 Also pin-connected columns are not as stiff as those with riveted ends, though the 
 great convenience offered by pin connections during erection and the saving of material 
 and other advantages in point of design, speak greatly in their favor. 
 
 The advisability of using the higher classes of steel for the main truss members of 
 large structures and employing soft steel for the floor system was previously mentioned. ' 
 
 In choosing between different styles of trusses, those of the statically determinate 
 class should always receive preference, other things being equal. The primary stresses 
 will usually be less than in similar indeterminate systems, especially when temperature 
 stresses are included. Yet the deformations and secondary stresses may be less and 
 frequently the connections may be simpler for the indeterminate types. 
 
 The use of light-colored paints is advisable to reduce temperature effects, especially 
 in structures involving redundant conditions. 
 
 (b) Final combination of stresses as a basis for designing. If the several stresses 
 resulting in a bridge member from all causes could be absolutely known, then there is nqj 
 good reason why a design should not be based on a unit stress of say four-fifths the elastic 
 limit of the material and still allow ample leeway for some deterioration and unusual effects. 
 
 The low unit stresses of one-third to one-half the elastic limit, generally employed 
 
ART. 65 SECONDARY STRESSES 269 
 
 are intended to cover, more or less blindly, the unknown stresses, on the presumption 
 that these are in a way proportional to the known primary stresses. In reality, few, 
 if any, of the secondary and additional stresses are directly related to the primary 
 stresses, but are produced by totally different causes. 
 
 Our knowledge of the properties of material, while of a purely empiric nature, 
 undoubtedly approaches the truth as closely as the knowable accuracy of the moving 
 loads would require. The behavior of full-size bridge members has also been investigated 
 to an extent which should make it possible to design such members with a reasonable 
 expectancy of developing a strength commensurate with that observed on test specimens. 
 This was perhaps not possible in the past, but should be expected when the results of the 
 elaborate tests of the American Society for Testing Materials become available. 
 
 If the stresses in the members of a structure are determined with similar exactness 
 and combined into totals representing the maximum and minimum stresses for each 
 member as a result of all known material causes, and on a basis of equivalent quiescent 
 loads, then there is no valid argument why the safe unit working stress / should not be 
 taken at least equal to two-thirds the elastic limit of the material. 
 
 If also the main members and counters, subjected to reversals of stress, are designed 
 for a 30 per cent overload, then such a structure should still be usable even for a 30 
 per cent increase in the assumed moving loads. 
 
 Furthermore, since the combined effect of the maximum values of all the known 
 primary, secondary and additional stresses is one of very rare occurrence, this provides 
 still greater safety and subjects the structure to a less severe average usage than that 
 intended by the maximum combination of all loads. 
 
 Assuming then that the following stresses have been computed, or approximately 
 estimated when no other means is afforded, then the required area for any member may 
 be determined from the formulae given below. 
 
 Let Sd =the stress in any member due to dead load. 
 Sz=max. stress in this member due to moving load. 
 S'i=mm. stress in this member due to moving load. 
 5,0= the max. wind stress due to Wi and w 2 , Art. 64b. 
 S r =the stress due to tractive forces, Art. 64a. 
 S t =the temperature stress when redundant conditions exist. 
 S c =the stress due to centrifugal force from curved track. Art. 64b. 
 ^=the impact coefficient or moving load factor Eq. (64c). 
 Z=the length of a member in inches. 
 r=the least radius of gyration in inches. 
 
 /=the allowable unit stress = two-thirds the elastic limit of the material. 
 M 8 =the bending moment produced by rigidity of joints, weight of the member, 
 eccentricity, etc., representing the total secondary stress, Arts. 61, 62, 63. 
 
 Then the required area F for any tension member becomes: 
 
 t-8m4-8*+&l+-rar. ( 65A ^ 
 
270 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIII 
 
 Whenever the area My/fr 2 can be added to the area given by the first term of Eq. 
 (65A) to obtain F without materially altering y and r, then it may be done, otherwise 
 a new section must be chosen to satisfy the equation. Also, that combination of the 
 stresses in the parenthesis which gives maximum must be used. 
 
 Eq. (65A) will also apply to any compression member when / is reduced for buckling 
 effect according to a modified form of Mr. T. H. Johnson's formula, for values of Z/r< 120: 
 
 For hinged ends and soft steel of 30,000 Ibs. per sq.in. elastic limit, 
 
 /= 20,000 -88^- 1 
 
 T 
 
 I ..... ..... (65B) 
 for flat ends, / = 20,000 -7 1 1 - \ 
 
 For hinged ends and medium steel of 35,000 Ibs. per sq.in. elastic limit, 
 
 /=24,000-105- 
 
 T 
 
 (65c) 
 
 for flat ends, /=24,000- 85- 
 
 When S'i is opposite in sign from Sd and Si, then the member must be capable of 
 carrying an additional stress l.3</>'S'i in the place of <l>Si, using the maximum combina- 
 tion of the same sign as S'i in Eq. (65A). In this case, which is one of stress reversal, 
 the factor 1.3 should also be applied to the stress </>Si. The impact factors and <// 
 being determined by one of the formula? in Eqs. (64c) for the load lengths producing 
 the respective stresses Si and S'i. However, this should not be construed to mean that 
 the area must be designed to carry the total loads LSflS'i+US^Si as if simultaneously 
 applied. 
 
 Aside from the additional 30 per cent in the live load stresses, members with stress 
 reversals may not require any increase in area, though the design should always be made 
 so as to provide for the two kinds of sitress. 
 
CHAPTER XIV 
 MITERING LOCK GATES 
 
 ART. 66. THE NATURE OF THE PROBLEM 
 
 A mitering lock gate is in principle a three-hinged arch, so placed that the hinges 
 have vertical axes, and the arch elements are acted upon normally by horizontal water 
 loads. Each half arch is called a gate leaf and the hinges are replaced by compression 
 joints adjacent to the quoin ends and miter post, see Figs. 6?A and 68A. 
 
 Each leaf is designed to swing about a vertical axis at the quoin end, thus permitting 
 the gate to be opened when the pressures on up- and downstream sides are equalized, 
 a condition which prevails when the gate is swung in air or when the water level is equal 
 on both sides. 
 
 The fastening on the top of the quoin post is called anchorage, and is usually in 
 tension while the gate is open. At -the same time the entire weight of a leaf is supported 
 on a pintle located at the foot of the quoin post. See Figs. 68A. 
 
 The maximum stresses on the anchorage and the pintle are encountered when the 
 gate is swung in air, while the maximum stresses in the structural elements of the gate 
 occur when the gate is closed against a full head of water on the upstream side with no 
 water on the downstream side, which is the case when the lock chamber is empty. 
 
 The ordinary working condition of a pair of gate leaves is less severe, being due to 
 a full head of water on the upstream side and a counter pressure on the downstream 
 side produced by the water inside the lock arid sufficient to float the largest vessels for 
 which the lock is intended. These two water levels are usually called the upper and 
 lower pool levels. 
 
 In order that a closed gate may act as a barrier or structural element against a 
 water head, the bottom horizontal girder is made to bear against a horizontal miter sill 
 extending over the full length of each leaf. The amount of pressure exerted by each 
 leaf against this miter sill is somewhat problematic, depending on the adjustment of 
 the sill to the gate. Usually this adjustment is so made that the pressure is just sufficient 
 to make a water-tight joint, though it may approach zero or may be made to take the 
 entire reaction due to the full head of water. 
 
 Hence, a variety of conditions may occur, ranging from zero sill pressure to the full 
 hydrostatic' head. Usually these two extreme assumptions are made and the resulting 
 stresses in the gate are found for each case. The actual condition during operation 
 may then be assumed at some intermediate state of sill contact which can only 1 
 surmised but never absolutely known. 
 
272 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIV 
 
 The structural elements of a single gate leaf consist of a series of horizontal girders 
 connected vertically by a sheathing or skin plate and a number of vertical intermediate 
 braces or stiffening girders. The horizontal girders of both leaves thus constitute a 
 series of three-hinged arches transmitting the water pressure through the quoin hinges 
 to the lock walls. The vertical system of stiffeners may be regarded as a set of trusses 
 connected with the horizontal arches and thus constituting a series of vertical continuous 
 girders, the supports of which are the elastic horizontal girders. 
 
 If the horizontal girders were designed to carry exactly their proportionate water 
 loads and at the same time be allowed to deflect in proportion to these loads, then for 
 no contact between the miter sill and the vertical stiffeners, the entire work would fall 
 on the horizontal girders and the vertical stiffeners would do no work and would thus 
 be superfluous. 
 
 However, for practical reasons, the top horizontal girders are always designed much 
 stronger than is actually required for the water load and also the other girders are not 
 all different, owing to commercial sizes of rolled shapes and the saving in manufacture 
 resulting from duplication of like parts. Also, some sill pressure must be provided to 
 accomplish water tightness of the gate. Therefore, the vertical system of stiffeners 
 must necessarily do a certain amount of work in distributing these inequalities between 
 the horizontal girders and the vertical stiffening system. The reactions, which the 
 horizontal girders thus present to each vertical girder, may be considered as redundant 
 reactions of a vertical continuous girder on n flexible horizontal supports, where n is 
 the number of horizontal arch girders. 
 
 The vertical girders are not very definite structural elements, being made up of a 
 rather disconnected system of web bracing between the horizontal arches. The total 
 vertical stiffness of a gate is thus produced by the combined effect of the sheathing 
 plates, the quoin and miter posts and the intermediate vertical girders. If these were 
 all combined in one vertical system so as to represent one vertical girder for a < gate leaf, 
 then the average stiffener per foot of gate could be approximated by dealing with an 
 average vertical section of the gate leaf. Such a hypothetical vertical girder could then 
 be treated as a continuous girder, on n horizontal supports consisting of n three-hinged 
 arches. 
 
 The problem would thus be to find the magnitudes of these n redundant supports 
 Xi to X n of the vertical girder for a certain water load (usually the maximum) together 
 with the elastic displacements d\ to d n of the n supports. 
 
 Since the contact at the miter sill cannot be absolutely adjusted, it is best to inves- 
 tigate two cases; one with full contact against the miter sill, giving the maximum sill 
 pressure; and the other where the miter sill pressure is just zero and there is theoretically 
 no contact. 
 
 It is certain that all possible cases must be included between these two extremes. 
 However, in practice it is customary to design the bottom horizontal girder to carry 
 the total water load without sill contact and to design the sill amply strong to carry the 
 same full load, but to adjust the gate so that the sill pressure is only sufficient to make 
 a water-tight joint. 
 
 The author has proposed details of construction obviating such duplication of 
 
ART. 67 MITERING LOCK GATES 
 
 273 
 
 strength in the gate and miter sill by the introduction of a flexible sill connection whereby 
 the water pressure against the sill can never exceed a certain allowable quantity. 
 
 ART. 67. THE THEORY OF THE ANALYSIS 
 
 The principal system. The problem is analyzed according to the solution of the 
 general case of redundancy presented in Art. 55, using the graphic method for all deflection 
 polygons. Since there are only two cases of loading requiring investigation it is best to 
 employ the general Eqs. (55A) and (55B) necessitating two solutions, each involving n 
 equations of n unknown quantities X. 
 
 The effect of temperature will be neglected in the present, though it could be 
 determined in accordance with Art. 56. 
 
 To make the problem perfectly definite, it is first necessary to decide on the principal 
 system with its determinate reactions, and then to apply the redundant forces along with 
 the external loads following the general scheme illustrated in Figs. 7 A to ?E, Art. 7. 
 
 The case of full contact at miter sill is illustrated in Figs. 67A and B. 
 
 Fig. 67A shows a lock gate in plan, with the water loads p, per linear foot, applied 
 to the upstream surface. The resultants R, of the loads p on each leaf, give rise to 
 arch thrusts R' and R" at the quoin ends and a horizontal thrust H acting between the 
 miter posts. The gate is supposed to undergo elastic deformations, as indicated by the 
 dotted deflection curves. The lock walls take up the arch thrusts. 
 
 Each horizontal girder receives a total load P which is proportional to the depth 
 of water below the surface. These loads are easily computed for all the horizontal girders 
 and are assumed to be known. 
 
 Fig. 67fl shows a projected elevation of the gate with the several loads P applied 
 to the respective horizontal girders. The bottom girder rests against the miter sill over 
 its entire length. 
 
 The vertical girder AB represents an element one foot thick lengthwise of the gate 
 leaf and may be regarded as a plate girder whose chords are portions of the sheathing 
 plate and all material making up the average vertical stiffness of the leaf, while its web 
 may be neglected. 
 
 This vertical girder is now treated as the principal system and the problem consists 
 of determining the stresses in, and the deflection of, this girder. 
 
 The determinate supports for this vertical girder (or principal system) are the sill 
 reaction B, which is immovable in the present case, and the reaction A offered by the 
 top horizontal arch girder. All the other horizontal girders 2 to n (shown by dotted 
 lines) are now considered removed and the redundant reactions X which they present 
 to the vertical girder are shown as external forces Xi to X n . 
 
 The principal system is thus the vertical giider on two determinate supports A 
 and B, with A elastic and B immovable. The external loads on the principal system 
 are the water loads P and the redundant reactions X. 
 
 The yielding condition of the support A may be obviated by introducing another 
 redundant reaction Xi and including the top horizontal arch as a part of the principal 
 system with immovable supports at the quoin ends D and E. The reaction A is then 
 
274 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIV 
 
 
 -^^- :: '^' 
 
 " ] 
 
 T 
 
 
 'I J 
 
 ii L CI 
 
 1 | 
 
 1 
 
 A\ 
 
 i 
 
 1 
 
 ! 
 
 FIGS. 67A, B Principal System, Full Loading, for Case of Full Contact at Miter Sill. 
 
 <-'_" 
 
 &** 
 
 FIG. 67c Principal System, Full Loading, for Case without Contact at Miter Sill. 
 
AET. 67 
 
 MITERING LOCK GATES 
 
 275 
 
 regarded as the resultant of the immovable reactions A' and A", and the deflection di of 
 the point of application of X lf can then be evaluated in terms of X it making 3 l =di ^X^ 
 The deflection curve of the vertical girder is indicated by a heavy dotted line. 
 
 FIG. 67o Principal System, Condition X=0. For Case of Full Contact at Miter Sill. 
 
 The case without contact at the miter sill is illustrated in Fig. 67c, where the principal 
 system consists of the vertical girder A B supported on two yielding supports. Herwe, 
 by taking in the top and bottom horizontal arches as part of the principal system the 
 fixed reactions become the four points D, E, F and G, while the elastic supports A and 
 
 FIG. 67E-Principal System, Condition X,= l. For Case of Full Contact at Miter Sill. 
 
 B undergo certain deflections ^ and 9 , which are evaluated in terms of the redundants 
 A'i and X 6 The reaction B is the resultant of the arch thrusts B' and B . 
 
 This case presents no new difficulty except to add one redundant condition X n 
 
276 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIV 
 
 and another water load P n to those involved in the previous case. No further con- 
 sideration is, therefore, given to the case without sill contact. 
 
 The conventional loadings of the principal system are illustrated in Figs. 67o and E, 
 for the case of full contact at the miter sill. 
 
 In Fig. 6?D the condition X =0 is represented. This embraces the total water 
 loads on the principal system and the determinate reactions A and B resulting there- 
 from, when all the redundants X are removed. 
 
 Fig. 6?E shows, for example, the condition X 3 = l, where all loads are removed from 
 the principal system except a unit load applied in the place of ^3 and acting in a direction 
 opposite to the direction assumed for the redundant reaction X 3 . Similarly all other 
 conventional loadings from Xi =1 to X n = \ are applied one at a time. It was not deemed 
 necessary to show them all by figures. 
 
 The external forces. Eqs. (55A) may now be applied to the principal system as 
 above illustrated, choosing the case of n horizontal arches with, or without, miter sill 
 contact. The only distinction between the two cases is that X n =0 for full contact, 
 otherwise the general formula apply to both cases. 
 
 r 
 
 * 
 
 p, 
 
 5 
 
 5 
 
 * 
 
 6 
 
 *, 
 
 p 
 
 P 
 
 
 
 
 
 
 
 
 
 
 I 
 
 x, 
 
 i 
 
 3 
 
 
 
 4 
 
 x. 
 
 ' 5 
 
 m 
 
 . 
 
 m* 
 
 , 
 
 1 
 
 
 
 X 
 
 ^rv 
 
 n 
 X 
 
 
 
 
 
 
 
 V 
 
 "m 
 
 
 
 
 
 
 
 
 1 
 
 
 
 
 
 
 
 
 
 u 
 
 
 
 
 
 
 
 v 
 
 "3 
 
 
 
 
 
 
 
 
 V, . 
 
 2 
 
 
 
 
 B 
 
 FIG. 67F. 
 
 Let Fig. 67r represent a general case of n redundants, showing the principal system 
 on two determinate supports A and B. Also, let m be any panel point about w r hich 
 the moment M m may be desired to determine the stresses in the chords of the vertical 
 girder. 
 
 Eqs. (55A) for n redundants and moments about m become: 
 
 A=A -A l X l -A 2 X 2 -A 3 X 3 
 B=B B\X\ B 2 X 2 B^Xs 
 m = M mo -M ml X! -M m2 X 2 -M m3 X 3 
 
 A m X m . . A n X n 
 
 B m X m . . B n X n 
 
 (67A) 
 
 The lettered constants in Eqs. (67A) are evaluated as follows: 
 
ART. 67 
 
 MITERING LOCK GATES 
 
 277 
 
 Condition. 
 
 X m =l 
 
 A -- 
 
 _!'A* 
 
 AS= /*7 
 
 Am = ~h~ 
 
 1-hm+i 
 
 Am+i = ^ 
 
 A = 
 
 = 
 
 / ^2\ 
 
 '. = hm I 1 7" ) 
 
 V hS 
 
 --^( l -h.) 
 
 M 
 
 The tabular values substituted into Eqs. (67 A) give after considerable reduction: 
 
 . . . . (67B) 
 
 The last of Eqs. (67s) for m = l and h m =h lf gives M A =0; and for m=n and 
 h m =h n =0, gives M fi =0, which results must follow from the conditions of equilibrium. 
 
 The same expression for M m may also be obtained by taking the moments about 
 m of all the external forces to either side of a section through m. 
 
 The redundant reactions X. The general work Eqs. (55n), when the effect of tempera- 
 ture is neglected, will furnish n equations for finding n redundant reactions X, as follows: 
 
 etc., 
 
 etc., etc., 
 
 +X 2 d n? . +X 3 3 n3 
 
 2_4+etC.,+^"n^2n 
 
 etc., etc., etc., to 
 
 (67c) 
 
 wherein the subscript m has all values successively from 1 to In all ^es of double 
 subscripts the first one always indicates the point of application of the conventional 
 
278 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIV 
 
 load while the second one refers to the location of the deflection. It should also be 
 remembered that all deflections d, with like double subscripts, are always equal; thus 
 ^2-6=^6-2 and d m5 = 5m , etc. 
 
 The double subscript bearing d's are found from deflection polygons drawn for the 
 principal system, and for each case of conventional loading, requiring n such deflection 
 polygons for n redundants. Thus, the several double-subscript-bearing deflection ordinates 
 3 of the first of Eqs. (67c) are all obtained from a deflection polygon drawn for condition 
 Xi =1 acting on the principal system. 
 
 The deflections 8 l} d 2 , <5 3 to 3 n are the actual displacements of the points 1 to n, of the 
 principal system. Since these points are the points of support offered by the horizontal 
 arch girders to the vertical girder, they represent the deflections really produced in the 
 arch girders by the redundant reactions ,X\ to X n respectively. Hence di, 3 2 , 3 to d n 
 cannot be deternined until the redundants Xi to X n are known, though they may be 
 expressed in terms of the latter without increasing the number of unknowns. 
 
 Let di =the deflection of horizontal girder I at the point of support of A"i and due 
 to the conventional load X\=l. 
 
 d-2. =the deflection of horizontal girder II at the point of support of X% and due to the 
 conventional load ^2 = 1. Similarly for ^3 to d n . When several horizontal girders are 
 of like sections then their deflections d become equal, thus obviating the necessity for 
 drawing n such deflection polygons. 
 
 When the conventional deflections d\, d%, d% to d n are known, then the actual deflec- 
 tions di, 82, 3 to d n , due to the redundant reactions Xi to X n , become (by the law of pro- 
 portionality between cause and effect) : 
 
 fli-di^i; 8 2 =d 2 X 2 ; 8 3 =d 3 X 3 ; etc., to 8 n =d n X n . . . .' (6?D) 
 
 It should be noted that in the particular case here considered, the deflection d\ =d\-\ 
 because the vertical girder is not deflected by the load X\=\ while the total displacement 
 of the principal system at point 1 is that due to the deflection of the horizontal arch, and 
 the vertical girder merely moves with the .arch. Similarly when there is no sill contact 
 then d n =8 nn . These conditions make it possible to draw the deflection polygons for 
 X\ = 1 and X n = 1 for the vertical girder. 
 
 The conventional loading for the principal system (being the vertical girder of 1 
 ft. thickness measured lengthwise of the horizontal girder) is taken as a unit force of say 
 one kip = 1000 Ibs. The corresponding conventional loading for the horizontal arch 
 girders should then be one kip per foot of wetted length of the arch. Whatever loads are 
 used, this ratio between the conventional loads of the horizontal and vertical systems 
 must be maintained. 
 
 The values from Eqs. (6?D) are substituted into Eqs. (67c), which latter are then 
 solved by any process of elimination to obtain the n redundants X. 
 
 When these redundants X are thus evaluated for any particular case of water loads 
 PI to P n , then the functions A, B, M m and di to d n may all be obtained by substitutions 
 in Eqs. (67B) and (67o) , and the problem is considered solved. 
 
ART. 08 
 
 MITERING LOCK GATES 
 
 279 
 
 The moment M m , will in turn furnish the chord stress in the vertical girder for the 
 center of moments m. The reaction B will represent the pressure against the miter sill 
 which must be considered in designing this sill. 
 
 ART 68. EXAMPLE-UPPER GATE, ERIE CANAL LOCK NO. 
 
 22 
 
 The example originally chosen for analysis was one of the huge Panama Canal lock 
 gates with sixteen horizontal arches. After completing the drawings and computations 
 it was deemed advisable to use a smaller gate for illustrative purposes as the theory would 
 be less obscured by the rather extensive tabulated computations. Accordingly one of 
 the small 1909 steel gates of the Erie Canal was used. 
 
 Figs. 68A show one leaf of this gate in elevation and two sections. There are six 
 horizontal girders and the upstream side is covered by a f-inch sheathing plate, while the 
 downstream side is open. 
 
 The material is soft steel with a modulus of elasticity #=29,000 kips per sq.in. The 
 quoin and miter-post cushions are made of white oak which when saturated with water 
 becomes quite soft. According to the best available data, the modulus EI, for such 
 material, may be taken at about 30 kips per sq.in. 
 
 The cross-sections of the horizontal girders are shown in Figs. 68B, from which it is 
 seen that girders II to V are exactly alike, so that there are only two different types of 
 horizontal girders requiring analysis. 
 
 Since the sheathing plate is continuous between the girders, a certain portion of this 
 plate will act with the flange material of each horizontal girder. It is difficult to say 
 just how much of the sheathing will actually act in that manner, but in the present analysis 
 a width of sixteen inches is thus included with the upstream flanges of the horizontal 
 girders except for girder I, where the sheathing stops at the girder web, and only eight 
 inches are included here. 
 
 The cross-sections F and moments of inertia I, of these horizontal girder sections, 
 are computed from Figs. 68B and given in Table 68A. 
 
 TABLE 68A 
 VALUES OF F AND I FOR THE HORIZONTAL GIRDERS. 
 
 At Points 
 
 Girder I. 
 
 Girders II to V. 
 
 F 
 sq.in. 
 
 7 
 in. 1 
 
 F 
 
 sq.in. 
 
 / 
 
 in.< 
 
 2 and 10 
 
 3 to 9 
 
 26.45 
 26.91 
 
 3740.6 
 4120.1 
 
 30.96 
 31.30 
 
 4054.6 
 4466.0 
 
 Fig. 68c shows the average vertical girder and the water loads for the gate. 
 This girder is made up of the material which gives vertical stiffness to the gate and 
 includes the sheathing plate on the upstream surface and such plates and angles as 
 
280 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIV 
 
 GENERAL DRAWING OF MITERING LOCK GATE 
 
 DOWNSTREAM ELEVATION. 
 
 wood 
 
 Cushio 
 
 DOWNSTREAM SIDE OPEN. 
 
 SECTION ON GIRDER IV. 
 
 FIG. 68A. 
 
 SECTIONS OF HORIZONTAL GIRDERS. 
 GIRDER I. , GIRDERS IITO,V. 
 
 Z9" 
 
 POINTS ZfclO. 
 
 AT POINTS 3T09. 
 
 FIGS. 68s. 
 
 29 
 
 AT POINTS 2 It io. 
 
 AT POINT* 3 TO 9. 
 
ART 
 
 MITERING LOCK GATES 
 
 281 
 
 extend vertically over the downstream surface. The total web mav be considered 
 
 ^ f the leaf 
 
 The total volume of this material is then divided by the length of the gate leaf 
 (24.3 ft.) giving the average section of the vertical girder per horizontal foot of the gate 
 
 r-roo.5 
 
 i -30.3 * 
 ATCRSURTAC _ 
 
 MITER SILL. 
 
 The loads p are th water pressures in kips per foot- of horizontal arch: 
 The loads P= 27 parrthe total pressures for the horirontal girders. 
 
 FIG. 68c. 
 
 is shown in Fig. 68c. The vertical girder is thus considered as an average strip of the 
 ;ate, one foot thick lengthwise of the leaf. 
 
 The water loads p are the actual pressures (in kips) per horizontal foot of gate and 
 xtend vertically from center to center of the panels between the arches. The loads 
 *=-27p are the total pressures for the horizontal girders of wetted length 27 ft. Hence 
 ili-P would be the total water pressure on one gate leaf, while ?p would be the total 
 3ad on the vertical girder with an exposed surface of 1 ho. 
 
282 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIV 
 
 The deflections d of the horizontal arches are now found for the conventional loading 
 according to the previous article. This loading should be one kip per horizontal foot 
 of wetted length of the arch, corresponding to a conventional loading of one kip for the 
 vertical girder one foot wide horizontally. However, in the present problem a conven- 
 tional load of q = l/l -=1/27 =0.037 kip will be used for the horizontal girders, giving 
 deflections l/l times too small. That is, the conventional load should have been I kips 
 on each leaf, uniformly distributed, but instead of this one kip was distributed over each 
 leaf of 27 feet wetted length. 
 
 The deflection of a three-hinged arch under a uniformly distributed load is made 
 up of a direct cross bending of the beam or half arch, and a deflection of the arch crown 
 due to axial compression, as shown in Fig. 68D. 
 
 Let N av be the average axial thrust in the half arch of length I and cross section F. 
 Also, let d c be the deflection at the crown normally to DC due to a shortening Jl in the 
 length / as a result of the compression N av . 
 
 Then 
 
 i * V fl ft j 
 
 = EF and 
 
 __ 
 
 tan a EF tan a ' 
 
 (68A) 
 
 FIG. 680. 
 
 The effect of d c on the arch is merely to lower the crown to C' , leaving the beams 
 DC and CE otherwise unchanged and producing the dotted deflection line DC'E. With 
 this must be combined the deflection due to bending and shear in the beam, finally 
 producing the deflection curve DC'E, with ordinates d measured normally to the girder 
 DC. The bending deflection will be determined from the bending moments, by the 
 method given in Art. 39 for solid web beams. 
 
 Before proceeding to this it may be well to say a few words regarding the particular 
 deflection d which is to be used in the solution of the problem, since this point may 
 involve a considerable element of doubt. 
 
 In the first place the deflections must be determined in the direction normally to 
 the girder DC, because that is the direction of action of the water loads P and of the 
 redundant reactions X. Ordinarily the deflections for any three-hinged arch would be 
 taken normally to the span DE. 
 
 The principal system was chosen as an element of average vertical cross-section 
 of the gate and the deflection properly belonging to such a hypothetical vertical girder 
 
ART. 08 
 
 MITERING LOCK GATES 
 
 283 
 
 would necessarily be an average deflection. Hence, the value of d which is actually 
 used is obtained by dividing the area DCC', of the deflection curve DC*, by the length 
 / of the leaf. This average deflection d av , is shown in Fig. 68D under the assumed vertical 
 girder. See also Figs. 68r. 
 
 The bending moments for the various points of a horizontal girder, due to the load 
 9 -=0.037 kip per foot of wetted length, are now determined by drawing a resultant 
 polygon through the girder and then taking the moments of the resultant normal thrust 
 about the center of gravity of the section, as shown in Fig. 68E. 
 
 The loads </i to q<n are first combined into a force polygon furnishing the resultant 
 R and the reactions R' and H, which latter can be determined when the points D and C 
 through which the resultant polygon shall pass, are fixed. The point C is taken in the 
 center of the joint and the point D is taken so that the resultant R' is normal to the 
 hollow quoin. The resultant polygon is 'then easily drawn and the offsets v, between 
 it and the gravity axis of the girder, are thus found. The axial thrusts N are then scaled 
 from the force polygon and from these and the eccentricities v the moments M =Nv are 
 computed in Table 68s . 
 
 The elastic loads w=MAx/l are also determined and from these the required 
 deflection polygon is drawn. 
 
 TABLE 68s 
 ELASTIC LOADS w FOR HORIZONTAL GIRDERS. 
 
 Pt. 
 
 TV 
 Kips. 
 
 
 inches. 
 
 M = Nv 
 Kip in. 
 
 Horizontal Girder I. 
 
 Horizontal Girders II to V. 
 
 / 
 in. 4 
 
 M 
 
 I 
 
 Ax 
 Inches 
 
 M4x 
 
 ""7" 
 
 / 
 in.< 
 
 M 
 I 
 
 4x 
 Inches. 
 
 Mix 
 __ 
 
 D, 1 
 
 1.38 
 
 0.0 
 
 0.00 
 
 
 0.0 
 
 
 
 
 0.0 
 
 
 
 
 
 
 
 
 
 40.00 
 
 0.0126 
 
 
 
 40.00 
 
 0.0116 
 
 2 
 
 1.396 
 
 1.7 
 
 2.373 
 
 3740.6 
 
 0.00063 
 
 24.00 
 
 0.0377 
 
 4054.6 
 
 0.00058 
 
 24.00 
 
 0.0348 
 
 3 
 
 1.40 
 
 7.4 
 
 10.360 
 
 4120.1 
 
 0.00251 
 
 32.64 
 
 0.1142 
 
 4466.0 
 
 0.00232 
 
 32.64 
 
 0.1054 
 
 4 
 
 1.40 
 
 13.2 
 
 18.480 
 
 4120.1 
 
 0.00449 
 
 32.64 
 
 0.1643 
 
 4466.0 
 
 0.00414 
 
 32.64 
 
 0.1514 
 
 5 
 
 1.40 
 
 16.4 
 
 22.960 
 
 4120.1 
 
 0.00558 
 
 32.64 
 
 0.1877 
 
 4466.0 
 
 0.00514 
 
 32.64 
 
 0.1728 
 
 6 
 
 1.40 
 
 17.4 
 
 24.360 
 
 4120.1 
 
 0.00591 
 
 32.64 
 
 0.1845 
 
 4466.0 
 
 0.00545 
 
 32.64 
 
 0.1697 
 
 7 
 
 1.40 
 
 15.8 
 
 22 . 120 
 
 4120.1 
 
 0.00537 
 
 32.64 
 
 0.1542 
 
 4466.0 
 
 0.00495 
 
 32.64 
 
 0.1421 
 
 8 
 
 1.40 
 
 12.0 
 
 16.800 
 
 4120.1 
 
 0.00408 
 
 32.64 
 
 1000 
 
 4466.0 
 
 0.00376 
 
 32.64 
 
 0.0920 
 
 9 
 
 1.40 
 
 6.0 
 
 8.394 
 
 4120.1 
 
 0.00204 
 
 24.00 
 
 0.0266 
 
 4466.0 
 
 0.00188 
 
 24.00 
 
 0.0246 
 
 10 
 
 1.39o 
 
 0.5 
 
 0.698 
 
 3740.6 
 
 0.00018 
 
 33.00 
 
 0.0030 
 
 4054.6 
 
 0.00017 
 
 
 33.00 
 
 0.0028 
 
 C, 11 
 
 1.37o 
 
 0.0 
 
 0.0 
 
 
 0.0 
 
 316.84 
 
 0.9848 
 
 
 
 316.84 
 
 0.9072 
 
284 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIV 
 
 0.5 
 
 I KIR 
 
 FIG. 68E. 
 
ART. 68 MITERING LOCK GATES 285 
 
 There are two different kinds of horizontal girders, thus making two such deflection 
 polygons decessary, one for Girder I and one for Girders II to V. The sections are given 
 in Figs. 68s and Table 68A. The increments Ax are measured along the span Z)C such 
 that 2Ja;=L All dimensions are now taken in inches and loads in kips. 
 
 According to Art. 39, an equilibrium polygon drawn for the elastic loads w with a 
 pole distance equal to "=29,000 kips, will represent the deflection polygon for the 
 moments M, with ordinates measured to the scale of lengths of the drawing. 
 
 In Figs. 68r, the pole distances were made equal to H= E/20 ,000 = 1.4 5, giving 
 deflections 20,000 times actual to the scale of lengths. 
 
 The moment deflection polygon is constructed by plotting the values M/I, from 
 Table 68s, as ordinates at the points 1 to 11, using any convenient scale, then finding 
 the centers of gravity g of the several trapezoidal areas of this M/I polygon and applying 
 the loads Wi to WIQ at these centers. The equilibrium polygon drawn for the w forces 
 is then the moment deflection polygon, with deflections measured parallel to the direction 
 taken for the w loads, in this case perpendicular to the chord of the girder. 
 
 The scale for the loads and the force polygon is immaterial so long as it is convenient 
 and the pole distance H must be laid off to the scale of loads. The pole may be 
 anywhere to suit a convenient closing line D'C'. 
 
 The actual moment deflection of girder I at point 6 is thus 
 
 -0.00188 ind, 
 
 The actual deflection due to shear at the same point 6, which is the point of maximum 
 moment, is by Eq. (39D) , 
 
 5X24 ' 36 
 
 = 
 6 
 
 _ 
 2EFi 2X29000X30X0.370 
 
 =0.00019 inch, 
 
 where Fj is the area of the web plate and M 6 is taken from Table 68s. 
 
 The deflection for point 6, due to shear alone, is thus about 10 per cent of the 
 deflection for the same point and due to moments. Hence, all the deflection ordinates 
 are increased 10 per cent to obtain the combined moment and shear deflection polygon. 
 
 The crown deflection d c , Eq. (68A), due to rib shortening by axial thrust, is now 
 
 found. . . 
 
 Since a portion of the gate, at the quoin and miter posts, is made of oak, which i: 
 probably saturated when the gate is in operation, it is necessary to make allowance for 
 this condition in computing Al. 
 
 For an average thrust JV- 1.4 kips, and a net Z-= 292 inches, F -26.91 sq.m., and 
 =29 000 kips per sq.in. for the steel part of the gate and Z'=24.8 inches, F'=24 
 =864 sq.m-, and E l =30 kips per sq.in. for saturated oak, then for tan a =0.367. 
 Eq. (68 A) gives for the steel and oak portions 
 
 1.4 X292 1.4X24.8 =0.00499 inch, 
 
 de = 29000 X26.91 X0.36V ^30 X 864 X0.367 
 
286 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIV 
 
 DEFLECTION POLYGONS OF HORIZONTAL GIRDERS 
 For Conventional Loading of One Kip Uniformly Distributed Over Wetted Length 
 
 _* 5. <JL T .AN T. . POAV G ON . 
 
 --- r .W>TTD* LO.CTH '- 17 FT. "T ' ~ ' --- 
 
 j 
 DEFLECTION ^OLYGON FOR GIRDER I. 
 
 M/I POLYGON FOR GIRDERS II TO V. 
 
 DEFLECTION "POLYftOIH FOR GIRDERS II TOY 
 
 h^a**C' 
 
 All deflections are 20,000 times actual 
 measured to the scale of lengths. 
 
 FIGS. 68F. 
 
ART. 68 
 
 MITERING LOCK GATES 
 
 and 20,000 times actual in feet becomes 
 
 ^=0.00499(^5^) =8.32 feet. 
 \ iz / 
 
 This crown deflection d c =8.32 ft. is now incorporated in the deflection polygon 
 for Girder I,J^ig. 68r, by laying off the ordniate C 7 C 77 =d c and then drawing the "final 
 closing line D'C" to complete the deflection polygon. The ordinates written in this 
 diagram represent deflections 20,000 times actual in feet. 
 
 In similar manner the deflection polygon for girders II to V is drawn. 
 
 9/ &) v 1 2 
 
 The actual moment deflection at point 6 is now "' =0.00171 inch, and the 
 
 20,000 
 
 actual deflection due to shear at the same point is 
 
 &MR 5x2436 
 
 d * =2EF\ = 2 X29000 X 30 X0.375 =0 - 00019 inch > 
 
 being about 11 per cent of that due to moments. Hence, the deflection ordinates are 
 increased 11 per cent to obtain the combined moment and shear deflection polygon. 
 
 The crown deflection d c is found as before, and all numerical values are the same 
 except the cross-section F of the girder, which is now 31.3 sq.in. Hence 
 
 1.4X292 1.4X24.8 . ' 
 
 c 29000 X 31.3 X 0.367 + 30X864X0.367 
 
 or 20,000 times actual in feet gives d c =S.l5 feet. 
 
 This value of d c is now used to complete the deflection polygon for girders II to V. 
 
 The average deflection d av for each girder is now computed from the ordinates taken 
 from the deflection polygons Figs. 68F, and the distances Az between these ordinates. The 
 values Az were taken in inches and the areas finally divided by the length Z in inches, 
 which eliminates the unit of length. The computations are given in the following table : 
 
 TABLE 68c 
 DEFLECTIONS d av FOR THE HORIZONTAL GIRDERS. 
 
 4z 
 
 Horizontal Girder I. 
 
 Horizontal Girders II to V. 
 
 
 d* 
 
 Areas 
 
 d* 
 
 Areas 
 
 Inches. 
 
 Feet. 
 
 dJz. 
 
 Feet. 
 
 dJj. 
 
 26.5 
 
 + 1 . 58 
 
 20.94 
 
 + 1 . 50 
 
 19.88 
 
 27.5 
 
 1.58+3.12 
 
 64.63 
 
 1.50+2.96 
 
 61.33 
 
 27.5 
 
 3.12+4.64 
 
 106.70 
 
 2.96+4.38 
 
 100.93 
 
 31.2 
 
 4.64 + 6.10 
 
 167.54 
 
 4.38+5.80 
 
 158.81 
 
 32.5 
 
 6.10+7.30 
 
 217.75 
 
 5.80+6.92 
 
 206.70 
 
 32.4 
 
 7.30+8.14 
 
 250.13 
 
 6.92+7.76 
 
 237.82 
 
 32.3 
 
 8.14 + 8.52 
 
 269.06 
 
 7.76+8.20 
 
 257.75 
 
 31.8 
 
 8.52 + 8.68 
 
 273.48 
 
 3.20+8.35 
 
 263.15 
 
 27.0 
 
 8.68+8.62 
 
 233.55 
 
 8.35+8.30 
 
 224.78 
 
 26.3 
 
 8.62 + 8.48 
 
 224.86 
 
 8.30+8.22 
 
 217.24 
 
 21.8 
 
 8.48+8.32 
 
 183.12 
 
 8.22 + 8.15 
 
 178.43 
 
 316.8 
 
 Tntn.1 
 
 2011.76 
 
 Total 
 
 1926.82 
 
 
 
 The deflection ordinates d are 20.000 times actual in feet. 
 
288 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIV 
 
 This gives for horizontal girder I, 
 
 2011.76 .. 6.35X12 
 
 d av = 3168 =6.35 feet, or actually 2000Q -- 0.00381 inch, 
 
 and for girders II to V, 
 
 89 fi 
 
 - =6-082 feet, or actually - =0.00365 inch, 
 
 which values will be used in Eqs. (67o) to obtain the -deflections d^ to 5 , observing 
 that d'a v for arches II, III, IV and V is the same quantity. 
 
 Attention is called to a slight oversight on Figs. 68F, where the deflection ordinates 
 should have been measured at the numbered points 2 to 10, instead of on the lines of 
 the w loads. The effect on the results is scarcely appreciable and therefore, the error was 
 not corrected. 
 
 The conventional deflections d of the principal system are determined graphically in Figs. 
 68c and 68n, for the case of full contact at the miter sill. This requires drawing a deflection 
 polygon for the principal system for each of the five cases of conventional loading. 
 
 The conventional loading for the horizontal arches was made equal to 1 kip, uni- 
 formly distributed over the wetted length of a gate leaf, being 1/1 = 1/27 =0.037 kip 
 per foot of arch. Hence the same load \/l =0.037 kip must be employed as the con- 
 ventional load for the principal system. 
 
 The deflection of the principal system is made up of the deflection of the vertical 
 girder and the elastic displacements of the two determinate supports A and B. For 
 full contact at the miter sill the B support is assumed as being immovable and the A 
 support is subject to the elastic displacement of the top horizontal arch caused by a con- 
 ventional load X = 1/1 acting on the principal system. The elastic displacement of the 
 point A is determined from the deflection d av just found for the top horizontal arch. 
 
 For the condition X\ l/l, there are no moments produced in any part of the vertical 
 girder and hence the deflection polygon for this case of loading would show no bending 
 effect of the vertical girder, but merely an elastic displacement of the A support which 
 was previously found to be d\-\=d av for the top horizontal girder. Hence, this con- 
 ventional deflection polygon is easily drawn and becomes a triangle aa'b, Fig. 680, wherein 
 the ordinate aa'=d av -= di_i for the top horizontal girder. This deflection line also 
 furnishes the values d l _ 2 , ^1-3, ^1-4 and <?j_s, which are respectively equal to d 2 -i, <V-i, 
 ^4_i and d$_i. The latter are the end ordinates at A of the several deflection polygons 
 to be drawn for the conditions X 2 = l/l to Xr, = \/l. 
 
 Hence, the deflections of the principal system are easily found when the deflection 
 lines of the vertical girder are once drawn for each of the several conventional loadings. 
 
 The deflection lines for the vertical girder are drawn, again using the method for 
 plate girders, Art. 39. However, no allowance is made for shear because the web is too 
 insignificant. 
 
 A set of w loads is now computed for each of the loadings X z --=l/l to X 5 = l/l, and 
 observing that the moment of inertia 7=971.6 in. 4 , for the vertical girder, is constant, 
 it is preferable to make these loads equal to MAx for a pole distance H =EI, which would 
 give actual deflections to the scale of lengths. 
 
AUT. 68 
 
 MITERING LOCK GATES 
 
 289 
 
 For simplicity in determining the moments M and the w loads, the conventional 
 loads X = l are used, giving moments 27 times too large. The deflections for the loads 
 A 7 " = 1/7 are then obtained from 
 
 MAx 
 
 =^Tfr, using a pole distance 
 
 Zt Ji/l 
 
 H =27EI =27 X29000 X971.6 =760,762,800, 
 
 where the dimensions are kips and inches, giving actual deflections in inches to the scale 
 of lengths. Instead, however, the pole was made #=27#//30,000 =25,360 for deflec- 
 tions 30,000 times actual to the scale of lengths, and w loads 27 times actual. 
 
 
 w loads for X 2 = 1 Kip. 
 
 w loads for X 3 1 Kip. 
 
 Ar 
 
 
 
 
 M for -X" 2 = l Kip. 
 
 w-MAx 
 
 M for X 3 = l Kip. 
 
 w = MAx 
 
 inches. 
 
 Kips and inches. 
 
 
 Kips and inches. 
 
 ' 
 
 
 A 2 = 0.2341 B. = 0.7659 
 
 
 4 3 = 0.4305 B 3 = 0.5695 
 
 
 
 M t = 0.7659 X 0.00= 0.0 
 
 
 M! = O.O 
 
 
 48.00 
 
 
 w,= 882.0 
 
 
 u>!= 655.2 
 
 
 M 2 = 0.2341X157.0 =36.75 
 
 
 M 2 = 0.5695 X 48.00= 27.3 
 
 
 40.25 
 
 
 w 2 = 1289.6 
 
 
 w 2 = 1561.7 
 
 
 M 3 = 0.2341 X 1 16.75 = 27.33 
 
 
 M 3 = 0.5695 X 88.25= 50.3 
 
 
 40.25 
 
 
 w> 3 = 910.5 
 
 
 w 3 = 1674.4 
 
 
 MI = 0.2341 X 76.5 =17.91 
 
 
 vl/ 4 = 0.4305= 76.50= 32.9 
 
 
 40.25 
 
 
 w 4 = 539.4 
 
 
 w t = 976.1 
 
 
 M 5 = 0.2341 X 36.25= 8.49 
 
 
 M, = 0.4305 X 36.25= 15.6 
 
 
 36.25 
 
 
 u> 5 = 153.9 
 
 
 w> s = 282.8 
 
 
 M 6 = 0.2341X 0.0 = 0.0 
 
 SM> =3775.4 
 
 M 8 = 0.0 
 
 ~Lw = 5150.2 
 
 
 w loads for X 4 = 1 Kip. 
 
 w loads for X b = 1 Kip. 
 
 Ax 
 inches. 
 
 M for X 4 = l Kip. 
 Kips and inches. 
 
 w = MAx 
 
 MforX 5 =l Kip. 
 Kips and inches. 
 
 -** 
 
 
 A 4 = 0.627 4 = 0.373 
 
 
 /1 5 = 0.823 B & = 0.177 
 
 
 48 00 
 
 M, = 0.0 
 
 Wl = 429.6 
 
 1 
 
 u>,= 204.0 
 
 
 M 2 = 0.373 X 48.0 =17.9 
 
 w 2 = 1022.4 
 
 M 2 = 0.177 X 48.00= 8.5 
 
 w; 2 = 485.0 
 
 
 M 3 = 0.373 X 88.25X32.9 
 
 w 3 = 1626.1 
 
 M 3 = 0.177 X 88.25 = 15.6 
 
 M> 3 = 770.8 
 
 40.25 
 36.25 
 
 M 4 = 0.373 X 128.5 =47.9 
 M 5 = 0.627 X 36.25 = 22.7 
 
 w> 4 = 1420.8 
 M> 5 = 411.4 
 
 M 4 = 0.177 X 128.5 =22.7 
 M 5 = 0.177 X 168.75 = 29.9 
 
 w t = 1058.6 
 
 tw s = 541.9 
 Su> = 3060.3 
 
290 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIV 
 
 The several conventional deflection polygons, for loadings X^l/l =0.037 kip, are now 
 drawn as illustrated in Figs. 68c and 68n. The factor 30,000 was introduced to obtain dia- 
 grams of convenient size to the scale of the drawing, and it is worthy of comment that the 
 highest desirable accuracy may be obtained from small drawings by an appropriate choice 
 of the scale and this factor. The scale employed for the force polygon is entirely immate- 
 rial so long as it is convenient, noting that the w loads and the pole must be measured 
 to the same scale. The pole distance remains the same for all the diagrams and is 25,360 
 w units. 
 
 The moment diagrams were drawn for the conventional loadings X = 1 to correspond 
 with the moments in Table 68D, and serve merely to locate the centers of gravity gi, <7 2 , 
 etc., which are the points of application for the w loads. 
 
 The value d av =0.00381 inch, as previously found for the top horizontal arch, now 
 furnishes <5i_ 1 =30,000 d av = 114.3 inches, from which the deflection line for X^=l/l is 
 drawn. This diagram then furnishes the deflections #1 -2=^2-1, di_ 3 -=d s _i, ^ 1 _ 4 =o 4 _ 1 
 and ^1-5=^5-1, which are necessary in fixing the closing lines of the several other deflec- 
 tion polygons. 
 
 With this explanation the drawings should be readily understandable and will 
 furnish all the deflections d required in solving Eqs. (67c) . 
 
 The problem is solved completely for the case of full contact at the miter sill, and 
 the method of deriving the deflections for the case of no sill contact is shown in Figs. 68H. 
 
 When there is no sill contact then the bottom support of the vertical girder becomes 
 the bottom horizontal arch in the same manner as the principal system was previously 
 f ormed for the top support at A , requiring now the determination of d' av for the bottom 
 arch (for a loading q =0.037 k. per foot of arch) as was done for the other arches. This 
 d'^ then gives <5e -6=30,000 d' av for the principal system and furnishes the means of draw- 
 ing the deflection polygon for condition X 6 = l/l from which ^0-5=^5-6, ^6-4=^4-6, 
 e~tc., are obtained. These deflections then serve to complete the conventional deflection 
 polygons for the case of no contact at the miter sill in the manner shown for the two 
 conditions X 4 = l/l and X 5 = l/l in Figs. 68n. Hence, all the work previously accom- 
 plished in solving the case for full contact is also usable for the case of no contact, though 
 the final deflections d are different in the two problems. 
 
 The formation of the final equations for the solution of the redundants X is now pos- 
 sible by introducing the numerical values for all the deflections d into Eqs. (67c) , remem- 
 bering that these were all taken 30,000 times actual, which in no wise interferes with their 
 use in the equations. 
 
 The numerical terms 27> TO l7n , Zp m d 2m , etc., must first be computed for the actual 
 water loads p per foot of the horizontal arches. These loads are given on the diagram 
 Fig. 68c and represent the actual water pressures on the girders for vertical depths extending 
 from center to center of the respective spaces between arches. Thus the load pi is the 
 water pressure per horizontal foot of arch and over a depth extending from the water 
 surface down to a line 24 inches below the top arch. The load p% is the pressure per 
 horizontal foot of the arch included between the depths 38 inches and 82.12 inches below 
 the surface, and so on for each girder. 
 
 The double-subscript-bearing deflections are taken from the deflection polygons in 
 
ART. 08 MITERING LOCK GATES 
 
 DEFLECTION POLYGONS FOR THE PRINCIPAL SYSTEM 
 
 FOR FULL CONTACT AT MITER . 
 THE CONVENTIONAL LOAD *1/1= 1/27O.O37 KIP CON. 
 
 291 
 
 * 2 MOMENT DjIAGRAM FORjX^=IKIR 
 
 All deflections are 30.000 times actual measured to the scale of lengths. 
 FlGS. 68G. 
 
292 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIV 
 
 DEFLECTION POLYGONS FOR THE PRINCIPAL SYSTEM 
 
 FOR FULL CONTACT AT MITER. 
 THE CONVENTIONAL LOAD = 1/1= 1/27= O.037 KIP CONCENTRATED. 
 
 
 . . ._!. .._, 
 
 v 
 
 upstream chord. 
 
 
 
 
 n 
 
 
 Gravity axis. 
 
 t 
 
 
 
 tn~ 
 
 
 
 4>" 
 
 
 1 
 
 i 
 
 2 
 
 3 
 
 4 r 
 
 5 
 
 
 L AX=4ft^ 
 
 I W-'iS J 
 
 4o'.'z5 > 
 
 ' 4o'/25 > 
 
 < 36.25 
 
 
 
 
 
 
 
 
 
 
 
 
 
 MOMENT DIAGRAM FOR|X = I 
 
 I 21 
 
 
 
 __jPi 
 oi'o, 
 
 
 nit 
 
 
 r 
 
 
 i 
 
 
 i 
 
 MOMENT DIAGRAM ^0^X5= I KIP 
 
 I DEFLECTION POLYGON FOR X ft 1/l- 0.037 K- NO MITER SILL CONTACT. 
 
 SCALES 
 
 BO 
 LOADS. 
 
 1000 O 5000 10,000 \ 
 
 All deflections ire 30,000 times actual measured to the scale of lengths. 
 
 FlGS. 68H. 
 
ART. 68 
 
 MITERING LOCK GATES 
 
 293 
 
 Figs. 68a to H, using the values 30,000 times actual, in inches. The deflections di to S 5 
 are obtained according to Eqs. (67o), using 30,000 times the deflection d av , found for the 
 horizontal arches. 
 
 All these numerical values are now entered in Table 68E, and the sum products 
 SpwAm are computed, giving all required data for writing out the general equations 
 for the redundants. 
 
 TABLE 68E 
 COMPUTATION OF THE TERMS K = Zp m dam FOR EQS. (67c) 
 
 
 
 
 
 
 
 
 
 
 
 
 
 3 = davX 
 
 Pt. 
 
 Pm 
 
 s im 
 
 Pm "im 
 
 *m 
 
 Pm 3 tm 
 
 S 3m 
 
 Pm 3 vn 
 
 3 4 m 
 
 Pm 3 <m 
 
 s sm 
 
 Pm*sm 
 
 Eq.(67 D ) 
 
 m 
 
 Kips. 
 
 
 
 
 
 
 
 
 
 
 
 
 1 
 
 0.314 
 
 114.3 
 
 35.890 
 
 87.6 
 
 27.506 
 
 65.2 
 
 20.473 
 
 42.8 
 
 13.439 
 
 20.2 
 
 6.343 
 
 114.3Xi 
 
 9 
 
 1.150 
 
 87.6 
 
 99.590 
 
 69.6 
 
 80.040 
 
 53.8 
 
 61.870 
 
 37.0 
 
 42.550 
 
 18.4 
 
 21.160 
 
 109. 5Xi 
 
 3 
 
 1.783 
 
 65.2 
 
 116.252 
 
 53.8 
 
 95.925 
 
 43.6 
 
 77 . 739 
 
 30.7 
 
 54.738 
 
 15.4 
 
 27.458 
 
 109. 5Xj 
 
 4 
 
 2.475 
 
 42.8 
 
 105.930 
 
 37.0 
 
 91.575 
 
 30.7 
 
 75.983 
 
 22.0 
 
 54.450 
 
 11.0 
 
 27.225 
 
 109.5X4 
 
 6 
 
 3.015 
 
 20.2 
 
 60.903 
 
 18.4 
 
 55.476 
 
 15.4 
 
 46.431 
 
 11.0 
 
 33 . 165 
 
 5.6 
 
 16.884 
 
 109. 5X. 
 
 
 
 2Pm im = 
 
 418.565 
 
 *PA = 
 
 350.522 
 
 *WW- 
 
 282.496 
 
 *Pm*m = 
 
 198.342 
 
 ***,- 
 
 99.070 
 
 
 The summations cover all the redundants from 1 to 5 and the subscript m has all successive values from 1 to 5. All 
 the values d are 30,000 times actual in inches. 
 
 FINAL EQUATIONS FOR THE REDUNDANTS 
 
 Eq. (1) Xi^.j 
 Eq. (2) X^,, 
 
 Eq. (3) X^s-i 
 etc. 
 
 <? 2 _ 2 + X 3 <5 2 - 
 
 (68B) 
 
 etc. 
 
 After substituting the numerical values from Table 68E, the following equations 
 are obtained : 
 
 (1) 114.3Xi +87.6X2+&5.2X 3 +42.8X4+20.2X 5 + H4.aX 1 =418.56 
 
 (2) 87.6X! +69.6X 2 +53.8X 3 +37.0X 4 + 18.4X + 109.5X 2 =350.52 
 
 (3) 65.2X! +53.8X2 +43.6X3 +30.7X 4 + 15.4X 5 + 109.5X 3 =282.50 [ . (68c) 
 
 (4) 42.8Xi +37.0X 2 +30.7X 3 +22.0X4 + 1 l.OXs + 109.5X4 = 198.34 
 
 (5) 20.2X! +18.4X2 + 15.4X3 + H.OX4+ 5.6X5 + 109.5X5= 99.07] 
 
 The terms are now collected in the above equations and the solution is carried out 
 in the following tabular forms: 
 
29t 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES, CHAP. XIV 
 TABLE 68F SOLUTION OF THE EQUATIONS 
 
 Operations. 
 
 Coefficients of the X'a. 
 
 Numerical 
 Term 
 K. 
 
 Xi 
 
 Xi 
 
 A'.i 
 
 Xi 
 
 X, 
 
 Eq. (1).. 
 
 228.6 
 
 87.6 
 
 65.2 
 
 42.8 
 
 20.2 
 
 418.56 
 
 
 Eq. (2) 
 
 87.6 
 
 179.1 
 33.55 
 
 53.8 
 24.97 
 
 37.0 
 16.39 
 
 18.4 
 7.74 
 
 350.52 
 160.31 
 
 Jg-frEq- (l)] = 0.383[Eq. (1)] 
 
 Eq. (2),=Eq. (2)-0.383[Eq. (1)] 
 
 145.55 
 
 28.83 
 
 20.61 
 
 10.66 
 
 190.21 
 
 Eq. (3).. 65.2 
 
 53.8 
 
 153.1 
 
 30.7 
 
 15.4 
 
 282.50 
 
 65 2 
 [Eq. (l)] = 0.285[Eq. (1)] 
 
 24.97 
 
 18.58 
 
 12.20 
 
 5.76 
 
 119.29 
 
 Eq. (3)! = Eq. (3)-0.285[Eq. (1)] 
 
 28.83 
 
 134.52 
 
 18.50 ' 
 
 9.64 
 
 163.21 
 
 9C OQ 
 
 ^ [Eq. (2) 1 ] = 0.198[Eq. (2),] 
 
 
 5.71 
 
 4.08 
 
 2.11 
 
 37.66 
 
 Eq. (3) 2 = Eq. (3), -0.198[Eq. (2),] 
 
 
 128.81 
 
 14.42 
 
 7.53 
 
 125.55 
 
 Eq. (4).. 42.8 
 
 37.0 
 
 30.7 
 
 131.5 
 
 11.0 
 
 198.34 
 
 42 8 
 [Eq. (l)] = 0.187[Eq. (1)] 
 
 16.39 
 
 12.20 
 
 8.00 
 
 3.78 
 
 78.27 
 
 Eq. (4), = Eq. (4)-0.187[Eq(l)J 
 
 20.61 
 
 18.50 
 
 123.50 
 
 7.22 
 
 120.07 
 
 f)f\ f*~t 
 
 ^- 5 [Eq. (2)J = 0.142[Eq. (2),] 
 
 
 4.08 
 
 2.92 
 
 1.51 
 
 27.01 
 
 Eq. (4) 2 = Eq. (4) 1 -0.142[Eq. (2)J 
 14 42 
 
 
 14.42 
 
 120.58 
 
 5.71 
 
 93.06 
 
 12881 " q ' ( 3 )2-l- - 112 L Ec l- ( 3 )2-l 
 
 
 
 .62 
 
 .o4 
 
 14 .Do 
 
 Eq. (4), = Eq. (4) 2 -0.112[Eq. (3)J 
 
 
 
 118.96 
 
 4.87 
 
 79.00 
 
 Eq. (5) 20.2 
 
 18.4 
 
 15.4 
 
 11.0 
 
 115.1 
 
 99.07 
 
 20.2 
 228^ [Eq. (l)] = 0.0884[Eq. (1)] 
 
 7.74 
 
 5.76 
 
 3.78 
 
 - 1.78 
 
 37.00 
 
 Eq. (5) 1 = Eq. (5)-0.088[Eq. (1)] 
 
 10.66 
 
 9.64 
 
 7.22 
 
 113.32 
 
 62.07 
 
 1 f\ f\f\ 
 
 
 
 
 
 
 145.55 ^ q ' (^iJ- - 0732 !^ !- ( 2 )J 
 
 
 . 11 
 
 .51 
 
 .7o 
 
 id. yz 
 
 Eq. (5) 2 = Eq. (5),-0.073[Eq. (2)J 
 
 
 7.53 
 
 5.71 
 
 112.54 
 
 48.15 
 
 
 
 
 
 
 
 12881^ q ' ^J- - 0585 ^- ( 3 ):J 
 
 
 
 .o4' 
 
 
 .o4 
 
 Eq. (5) 3 = Eq. (5) 2 -0.058[Eq. (3)J 
 
 
 
 4.87 
 
 112.10 
 
 40.81 
 
 
 
 
 
 
 
 118%^ q ' ^ 3 " q ' ^ ^ 
 
 
 
 
 
 
 Eq. (5) 4 = Eq. (5) 3 -0.04irEq. (4)3] 
 
 
 
 
 111.90 
 
 37.58 
 
 37.58 
 
ART. 68 
 
 MITERING LOCK GATES 
 
 295 
 
 The solution of the equations is completely given in Table 68r, which also illustrates 
 the method. The successive operations are expressed in the first column in unmis- 
 takable language, and the order of performing the computations is to work down the 
 table first entering the given equations, then obtaining the products / [Eq. (1)] down 
 through each sub-table, and finally making the subtractions. At this point apply the 
 following checks : Note that the symmetric values must be the same and that the terms 
 included between the double-ruled lines and those in the columns headed K are not 
 checked. Therefore, check the latter terms by repeating the numerical work and check 
 the symmetric terms by inspection. Then proceed with the work of obtaining the 
 products/" [Eq. (2)i] and repeat the same programme above outlined. 
 
 The final Eq. (5) 4 gives the value of ^5 = 0.3358 kip per horizontal foot of horizontal 
 girder V. The other redundants are then determined from the equations in the last 
 sub-table of Table 68F by successive substitutions as indicated in the following Table 680. 
 
 Terms. 
 
 .sr, 
 
 X, 
 
 X3 
 
 Xt 
 
 X* 
 
 K- 
 
 99.07 
 
 62.07 
 
 48.15 
 
 40.81 
 
 37.58 
 
 x t c> 
 
 -38.65 
 
 -38.02 
 
 -37.79 
 
 -37.64 
 
 
 
 
 
 
 
 37.58 
 
 x t c t 
 
 - 7.16 
 
 - 4.70 
 
 - 3.72 
 
 -37.64 
 
 
 x 3 c s 
 
 -13.58 
 
 - 8.50 
 
 -41.51 
 
 3.17 
 
 = 0.3358 
 
 
 
 
 
 x 3 ' 17 
 
 
 .AjCj 
 
 18.73 
 
 51.22 
 
 .64 
 
 4 ~4 87 
 
 
 
 
 
 6 64 
 
 
 
 
 -78.12 
 
 10.85 
 
 
 = 0.6509 
 
 
 .A 
 
 
 
 
 7.53 
 
 
 
 
 
 10 85 
 
 
 
 
 
 20.95 
 
 
 = 0.8818 
 
 
 
 
 
 2 10.66 
 
 
 
 
 
 20.95 
 
 = 1.0178 
 
 
 
 
 
 1 20.2 
 
 
 
 
 
 
 = 1.0371ki 
 
 ps per horizontal 
 
 "oot of gate. 
 
 
 
 C = the coefficients from the last of Tables 68F. 
 
 The redundants X are thus expressed in kips per horizontal foot of gate, representing 
 the average intensity of these forces for the average vertical stiffness of a gate leaf. 
 
 The reactions A and B and the moments M for this average vertical girder, may now 
 be found from Eqs. (67e) employing the values in the following table. 
 
296 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XIV 
 
 TABLE 68n 
 REACTIONS AND MOMENTS, VERTICAL GIRDER 
 
 Point. 
 
 p 
 Kips per ft. 
 
 X 
 Kips per ft. 
 
 p-X 
 Kips per ft. 
 
 h 
 Inches. 
 
 hi -h 
 Inches. 
 
 (p-X)h 
 Kip-inches. 
 
 (p-X)(hi-h) 
 Kip-inches. 
 
 1 
 
 0.314 
 
 1.0371 
 
 -0.7231 
 
 205.0 
 
 0.0 
 
 -148.236 
 
 0.0 
 
 2 
 
 1.150 
 
 1.0178 
 
 + 0.1322 
 
 157.0 
 
 48.0 
 
 + 20.724 
 
 6.336 
 
 3 
 
 1.783 
 
 0.8818 
 
 0.9012 
 
 116.75 
 
 88.25 
 
 105.192 
 
 79.513 
 
 4 
 
 2.475 
 
 0.6509 
 
 1.8241 
 
 76.5 
 
 128.5 
 
 139.536 
 
 234 . 380 
 
 5 
 
 3.015 
 
 0.3358 
 
 2.6792 
 
 36.25 
 
 168.75 
 
 97.120 
 
 452.115 
 
 Totals 
 
 8.737 
 
 3 . 9234 
 
 
 
 
 + 214.336 
 
 772.344 
 
 Eqs. (67fi) then give for A, B and M 3 
 
 779 
 -h) =i_' 
 
 =3.767 kips. 
 
 \ 
 
 341.848 = 196.057 kip-inches, 
 
 t , My 196.057X23.4 __,_ . 
 giving for f=j= -- =- ~ - = 4.727 kips per sq.m. on the most extreme fiber of 
 
 J- \j i J. .O 
 
 the vertical girder at the panel 3. 
 
 In like manner the stress may be found for any other point of the vertical girder, 
 using the results given in Table 68n. 
 
 B+PQ represents the pressure which the miter sill must sustain on the assumption 
 of full contact. 
 
 Finally the static condition 
 
 ././ -. . . . . . . (680) 
 
 must be fulfilled by the above numerical results. 
 
 The deflections of the vertical girder are now readily found from Eqs. (67c) and the 
 redundants X which are now known. 
 
 The deflections d av of the horizontal girders were found for a load of 1 kip uniformly 
 distributed over the wetted length of 27 ft. Hence, for a load of 27.X" uniformly dis- 
 tributed over the same length, the deflection would be 27d av X. Accordingly the 
 deflections d to d$ become 
 
 <?! =27d av Xi =27 X0.003S1 X 1.037 =0.107 in. 
 
 d 2 =27d'a V X 2 =27 X0.00365 X 1.018 =0.100 in. 
 
 3 3 =27d v X 3 =27 X0.00365 X0.882 =0.087 in. 
 
 <? 4 =27d'a V X 4 =27 X 0.00365 X 0.651 =0.064 in. 
 
 d 5 =27d' m ,X 5 =27 X0.00365 X0.336 =0.033 in. 
 
ART. 68 MITERING LOCK GATES 297 
 
 An approximate idea of the deflection curve at any vertical ra other than the 
 average, may be obtained by assuming that the same proportionately exists between 
 the deflections d av and d m as between di and d m , etc., where the point ra designates any 
 vertical section of the gate. Thus in Fig. 68r, for a vertical girder at point 8, the 
 deflection d 8 =8.68 ft. for the top horizontal arch while d av for that arch was 6.35 ft., 
 giving a ratio d 8 /d av = 8.68/6.35 = 1.367. For the arches II to V the ratio between 
 similar ordinates becomes dg/d ap = 8.35/6. 082 = 1.373. 
 
 Hence, in considering a vertical girder at point 8, where the deflections in the 
 horizontal arches are about 37 per cent greater than the average, the effect on all the 
 conventional deflections would be to increase them approximately by the same constant 
 percentage. This would be equivalent to multiplying Eqs. (68c) by a constant 1.37 
 and since the water loads p remain unchanged the resulting redundants X would not 
 be materially different than for the average vertical girder. 
 
 However, the deflections di to 5 would be increased throughout by the same 37 percent. 
 
 Since the ratio d m /d av is not necessarily constant, though nearly so in the present 
 example, this proposition cannot be accepted for more than it is worth, and approximate 
 results only can be expected. 
 
 Where a rigid determination is desirable, it would be necessary to repeat the whole 
 computation for the actual deflections of another principal system with vertical girder 
 at point 8. This would not involve very much labor, however, beyond solving Eqs. 
 (68c) for a new set of numerical coefficients. 
 
 A summary of the steps employed in the solution of the above problem -is given in 
 closing this subject. 
 
 Horizontal arches. Draw a moment deflection polygon for each of the different 
 arches, using a conventional load of one kip uniformly distributed over the wetted length 
 of each leaf. These deflections are then combined with the arch deformation for axial 
 thrust or rib shortening for the same loading, to obtain the total displacements of the 
 several points of the arches with respect to the fixed quoin post supports. 
 
 These total displacements are then averaged for one gate leaf by dividing the area 
 of the displacement polygon by the straight length of the girder, giving the deflection 
 d av for each arch. Then d=d av X, and for the top arch d av =d l . li while for the bottom 
 arch d' av =d nn when there is no sill contact and d' av =0 when full sill contact is assumed. 
 
 The principal system. Draw conventional deflection polygons for the principal 
 system, using conventional loads of 1 /I kips for each condition X = l/l where I is the 
 wetted length of one gate leaf as above. This gives the conventional deflections of all 
 panel points of the vertical girder for each condition of loading, remembering that the 
 vertical girder represents a strip of gate one foot thick and of average cross-section. 
 
 The water loads p per horizontal foot of gate for each arch, are then used as the 
 actual loads in computing the numerical terms Sp m 8 am for each condition X-l/l. _ 
 resulting values X are then the redundant reactions of the average vertical gir 
 against the horizontal arches per linear foot of the latter. 
 
 When the vertical girder is a truss frame instead of a plate girder, then the method 
 of deflections given in Art. 36 must be used, neglecting the effect of the web member* 
 as the nature of the problem does not warrant the extreme accuracy impl* 
 web system is considered. 
 
CHAPTER XV 
 FIXED MASONRY ARCHES 
 
 ART. 69. GENERAL CONSIDERATIONS 
 
 The fixed arch is statically indeterminate in the third degree, or, as commonly 
 expressed, it involves three redundant conditions, provided the arch ring may be treated 
 as an elastic body. 
 
 According to the theory of elastic deformation, these redundant conditions are 
 determined from the elastic properties of the material, while for the older " line of 
 thrust " method the redundants are merely approximated by assigning certain assumed 
 conditions to be fulfilled by the line of thrust. 
 
 A line of thrust is defined as an equilibrium polygon for any case of simultaneous 
 loads acting on the arch ring. Hence, for any section, the adjacent elements of the 
 line of thrust will represent the resultant of all the external forces on either side of the 
 section, including the reactions.. From this it follows that the sum of the moments of 
 all external forces about any point on the line of thrust, or resultant polygon, must always 
 be zero. 
 
 Therefore, if the external redundant reactions were known, then the real resultant 
 polygon could be drawn, but since the redundants can be found only from the theory 
 of elasticity, the line of thrust method fails to give a direct solution. 
 
 A resultant polygon may always be drawn through three given points, as will 
 be shown later. Hence, by successive trials, a resultant polygon may be approximated 
 so as to fulfill certain requirements which are now discussed. The line of thrust is 
 henceforth called the resultant polygon, this being a more appropriate designation. 
 
 The theory of elasticity fixes the position of the resultant polygon in terms of the modu- 
 lus E of the material and the elastic deformation of the arch ring. According to the 
 older method an infinite number of resultant polygons may be constructed for the same 
 arch and same loading, and it becomes a question to decide which of all possible lines 
 is the most probable. 
 
 Hagen (1844 and 1862), according to his " theory of the most favorable distribution 
 of stress," defines the most probable resultant polygon as the one for which the vertical 
 projections of the minimum distances, between the sides of this polygon and the bound- 
 ing lines of the arch ring, become equal. 
 
 Culmann (1866) advanced the theory that of all the possible resultant polygons, 
 the most probable one must approach the arch center line in such manner as to reduce 
 the stresses in the critical sections to a minimum. Carvallo (1853) and Durand-Claye 
 (1867) adopted this theory with slight modifications. 
 
ART. 69 FIXED MASONRY ARCHES 299 
 
 The theory of elastic deformation was introduced by Navier in 1826, by his analysis 
 of the stresses on an arch section, in which he assumes a combined thrust and bending 
 moment distributed over the section. According to Navier, the stress on the extreme 
 fiber of any arch section becomes 
 
 f _N My_N M 
 
 J ~^ ......... 
 
 where N is the normal thrust on the section, M the bending moment, F the area, y the 
 distance from the neutral axis to the extreme fiber and / and W the moments of inertia 
 and of resistance, respectively, of the cross-section. The formula when applied to curved 
 beams, or arches, is not strictly accurate, but when the radius of curvature is large in com- 
 parison with the depth of the arch ring, a condition which always obtains in ordinary 
 arches, then the error due to omitting curvature is extremely small and is never considered. 
 
 The experiments of Bauschinger (1834-93), Koepcke (1877), and the Austrian Society 
 of Engineers and Architects (1895), have substantiated the correctness of the theory of 
 elasticity within knowable limits of the elastic properties of building materials. 
 
 Nearly all modern writers on this subject adopt Navier's law as the basis of investiga- 
 tion. The various arch theories by Winkler (1867), Mohr (1870), Belpaire (1877), J. 
 Weyrauch (1878), F. Engesser (1880), Mueller-Breslau (1880), and those of American 
 writers, differ principally in manner of presenting the subject. Each author has con- 
 tributed something in the direction of rendering the theory of elasticity more usable, and 
 this is especially desirable even at the present time. 
 
 Professor Winkler was the first to prove that the most favorable resultant polygon in an 
 unsymmetric arch must intersect the arch center line in at least three points, while -for 
 symmetric arches four such intersections are necessary. He also, proved that according 
 to the theory of elasticity the most probable resultant polygon is the one for which the 
 I'esidual departures from the arch center line become a minimum, according to the method 
 of least squares. This then furnishes the real criterion with which to judge an arch design 
 in connection with certain other limitations to be discussed later. 
 
 The common graphic solution is nothing more than the application of the principles 
 of the three-hinged arch to the fixed arch by assuming the location of the hinged points, 
 at the crown and springing. The location of these hinged points may be altered until 
 such a resultant polygon is found as will approach the arch center line in the manner 
 required by the theory of elasticity. 
 
 The most probable resultant polygon, when found, must remain within the middle 
 third of the ring, when tensile stress is not permissible. This follows from Eq. (69A). 
 Hence it is clear that the most economic masonry arch is one for which the center line 
 coincides most closely with the resultant polygon drawn for the case of average or half 
 total live load, provided the arch ring has sufficient thickness at all points to prevent 
 excessive unit stresses due to unsymmetric loading. This idea will be followec 
 treating of methods for preliminary designs. 
 
 Formerlv an arch problem was considered solved, when, for a given span, rise and 
 loading a resultant polygon could be found which, for the critical position of the live 
 load, remained entirely within the middle third of the ring. However, this is far from 
 
300 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XV 
 
 constituting an acceptable solution, as will be shown in the following. The fact that most 
 arches have stood so well is no defense of this method of designing, especially when it 
 is understood that the usual factor of safety employed in such designs ranges between 
 ten and sixty, rarely going below twenty. 
 
 Furthermore, solid spandrel arches are usually designed with very high safety 
 without considering the additional carrying capacity of the masonry in the spandrel, 
 which is often sufficient to sustain the loads without regard to the ring. 
 
 The author adds the criticism that few fixed arches exist which are not disfigured by 
 unsightly cracks and this, in conjunction with the very wasteful expenditure of mate- 
 rial necessitated by low unit stresses, constitutes a fertile field for doubt regarding the 
 expediency and justification of building fixed arches in this progressive age. 
 
 While it is freely admitted that these masonry structures, many of which are monu- 
 mental in character, have stood well, and will undoubtedly continue to stand, every 
 engineer must agree that a cracked arch is neither an achievement nor 'the ultimate aim 
 of his ambition, to say nothing of the dissatisfaction which such a blemish constitutes 
 to both engineer and owner. 
 
 Practically, the objections to fixed masonry arches may seem trivial and mostly 
 a matter of sentiment, while theoretically they are important. The remedy for all these 
 objections is to introduce hinged joints during construction or preferably for all time. 
 This has been practiced in Germany since 1880, and has given excellent satisfaction, 
 though many fixed arches are still being built. This will continue to be the case so long 
 as we do not come to recognize that " new truths are better than old errors." * 
 
 The theory ot elasticity is most applicable to arches of small depth of ring and high 
 rise, and the analysis of stresses will be more accurate the nearer the shape of the arch 
 center line coincides with the resultant polygon for average loading. 
 
 This is true when dealing with the arch ring alone, as the ring is the only portion of 
 an arch which could be expected to follow the laws of elastic deformation. Hence, in order 
 that the theory shall be at all applicable, the relation between the arch ring and its span- 
 drel filling must be such as to allow the ring perfect freedom to undergo elastic 
 deformations. 
 
 The style of structure which best satisfies this requirement is one where the arch 
 ring supports the roadway on a succession of piers, producing what is called an open 
 spandrel filling. Even here the roadway should be provided with expansion at one end. 
 
 Any other style of spandrel will introduce further redundants which are wholly 
 beyond analysis, because the spandrel is then subjected to bending and direct stress 
 the samfe as the ring 1 . 
 
 Therefore, to apply the theory of elasticity or in fact any theory, to a solid spandrel 
 fixed arch is, in the author's opinion, a mere waste of time which should be expended in 
 a more profitable pursuit. In the following, nothing but open spandrel arch bridges 
 are dealt with. 
 
 While the theory of elasticity is Undoubtedly the only trustworthy basis for the 
 
 * See paper by the author on Three-Hinged Masonry Arches, Trans. Am. Soc. C. E., Vol. XL, 1898, 
 p. 31. 
 
ART - 70 FIXED MASONRY ARCHES 301 
 
 analysis of fixed arches, and should always be employed in designing structures of any 
 importance, it does not follow that the older methods of investigation, and many of the 
 empiric rules in common use, are all worthless. On the contrary, the preliminary design 
 can be carried out most efficiently by the aid of these simpler methods, and the final 
 design should then be tested by the application of the more exact method of the 
 theory of elasticity. 
 
 The theory of fixed arches has reached a status of perfection quite in keeping with 
 the nature of the problem, the still existing uncertainty being a function of the material 
 and other circumstances, depending on the rigidity of the abutment foundations, condi- 
 tions of erection, etc., all of which can never be definitely known nor be entirely eliminated 
 in the best designs. 
 
 When the loading is not known with considerable accuracy, as for arches, sustain- 
 ing high earth banks, the design must necessarily remain more or less indefinite, though 
 the elastic property of the superimposed earth assists greatly in distributing pressures 
 and rendering conditions more favorable. Also in dealing with small arches, such as cul- 
 verts and road crossings, the resultant polygon theory will probably continue to be the 
 only method of approximate analysis. 
 
 ART. 70. MODERN METHODS OF CONSTRUCTION 
 
 Many difficulties are encountered in the construction of fixed masonry arches, owing 
 particularly to insufficient elasticity in masonry. The natural deformations in the arch, 
 caused by shrinkage of the masonry or concrete, due to the setting process, stress and 
 temperature, usually cause cracks, which, while rarely of a serious character, are reasons 
 for discouragement to the engineer, who has probably applied eveTy known precaution 
 to prevent their occurrence. 
 
 So long as there are no abutment displacements after completion of an arch, the 
 above difficulties may be fairly well controlled. However, when an arch was designed 
 for certain allowable unit stress on the extreme fiber at the critical points, then, if the 
 structure is to be regarded safe for all time, the original stress must not be exceeded even 
 when the material increases in strength (as by setting of the cemjent) or when the arch 
 or its abutments undergo slight elastic or permanent displacements. For this there 
 is absolutely no assurance, though it is an essential necessity in the general assumption. 
 
 If an arch ring could be built in such manner that its resultant polygon would pass 
 through the center of the ring at the crown and springing points at the time of releasing 
 the falsework, a large proportion of the redundant stress could be prevented. In other 
 words, the bending moments at the critical points would then be almost zero for sym- 
 metric loading, reserving the strength for the unsymmetric live load and other contingencies 
 affecting the shape of the arch ring. 
 
 By the ordinary process of construction, the arch is commenced at the abutments 
 and the load is gradually applied to the falsework, thus distorting the latter and causing 
 bending stresses in the completed portion of the ring. The final settling of the ring, 
 produced by the contraction of the mortar or concrete, as a result of compression and 
 shrinkage during setting, even when the abutments remain immovable, are sufficient 
 
302 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XV 
 
 to create serious initial stresses, thus destroying to a large extent the usefulness of the 
 structure and frequently cracking the arch. 
 
 Several methods have come into use by which these initial arch stresses may be more 
 or less completely prevented. One of the oldest of these consisted in setting the arch 
 stones on the entire falsework, spacing the joints by small strips of wood, and lastly 
 filling all the joints simultaneously with mortar. This is still a very good programme for 
 construction of small arches, but not so well adapted to large structures, owing to the 
 excessive stresses produced in the falsework, frequently causing considerable settlement 
 before the ring can be closed. However, the falsework could be wedged up just previous 
 to filling the joints. 
 
 Another method, frequently employed in the construction of brick and concrete 
 arches, is to close a complete ring adjacent to the falsework, and after allowing this 
 course to set firmly, the remainder of the arch is completed, thus carrying the load by 
 the first ring rather than by the falsework. By this means the settlement during 
 construction becomes very slight, but, as is readily seen, the intrados is excessively 
 stressed, a condition which cannot be averted and which is highly objectionable. 
 
 A programme of construction frequently followed on large arches is to commence work 
 simultaneously at two, four or six points of the ring and closing at three, five or seven 
 points, respectively. The larger the span the greater the number of points of commence- 
 ment. 
 
 The most modern method consists of the introduction of temporary flexible or 
 hinged joints at the crown and springing and following the last-mentioned programme 
 of construction. These flexible joints are made of stone with curved roller-like surfaces; 
 or iron blocks may be used. By this means the resultant polygon becomes determinate 
 during the period of construction. After the falsework is removed and the structure 
 has assumed a normal condition of stress, these open joints are grouted with cement 
 mortar. 
 
 Sheet lead or lead blocks have also been used for this purpose, allowing sufficient 
 surface of contact on the arch center line to carry the pressure without causing the lead 
 to flow. A safe allowable pressure for lead is from 1000 to 1500 Ibs. per sq. in. 
 
 The author would recommend an alloy of lead with from 5 to 10 per cent of copper 
 added, thus permitting an allowable unit stress of about 2500 Ibs. per sq. in. The 
 last mentioned method gives very good results in preventing cracks and excessive 
 stresses, during construction, but subsequent abutment displacements or other changes 
 cannot be compensated and still remain as a serious objection to this class of structure. 
 Eventual settlements in foundation masonry, contraction of mortar or concrete due 
 to setting and drying out; compression due to loading, elastic deformations caused by 
 temperature and load effects all these influences are ever present to create stresses 
 which, in spite of all precautions during construction, may attain dangerous proportions 
 and make it utterly impossible to estimate the ultimate strength of a fixed masonry 
 arch. 
 
 However, it should be remembered that, for masonry arches, the live load is generally 
 a small fraction of the dead load, and for this reason an arch which is sufficiently strong 
 to sustain its own weight permanently will carry temporary live loads with perfect safety. 
 
ART. 70 FIXED MASONRY ARCHES 
 
 Masonry arches possess the redeeming feature that when stressed to the breaking 
 limit the masonry generally chips near the surface, thus relieving the stress and allowing 
 the resultant polygon to return to a more favorable position. In this manner an arch 
 which has become distorted may readjust itself to a new condition of stress. 
 
 Hence fixed masonry arches, in view of their past history, cannot be condemned 
 on account of insufficient safety which they afford, but because masonry is not a suitable 
 material from the criterion of inability to withstand elastic deformations well. A fixed 
 plate girder arch would be a far more rational structure than a fixed masonry arch, though 
 the former would require frequent painting and even then might deteriorate and thus 
 outlive its usefulness earlier than the masonry structure. But from an engineering 
 standpoint the plate girder would probably be more satisfactory because it would not 
 open up cracks and thus bring discredit to the designer and builder. 
 
 In this connection it may be of interest to recall the recommendations proposed by 
 the Austrian Society of Engineers and Architects, which, if strictly carried out, would 
 limit the application of fixed masonry arches to comparatively short spans with bed 
 rock foundations. These recommendations require the fulfillment of the following 
 conditions : 
 
 1. The abutments must be perfectly rigid. 
 
 2. The falsework must retain its form during the period of construction of the 
 arch ring. 
 
 3. The masonry must be of the best quality. 
 
 4. The construction of the arch ring must be most carefully conducted. 
 
 5. The falsework must not be released until the cement has thoroughly set. 
 
 6. When the falsework is finally released, it must be done gradually and uniformly. 
 
 7. To this the author adds that the arch ring should be closed at the lowest possible 
 temperature. 
 
 The use of sand jacks or sand pots, which was introduced in 1854 during the construc- 
 tion of the Austerlitz Bridge, in Paris, offers a very novel and efficient means of releasing 
 falsework. 
 
 The necessity of these recommendations is clearly understood after what has been 
 said regarding the theory of fixed masonry arches and its practical applications and 
 limitations. 
 
 The two first conditions can be realized only when rock foundations are available. 
 The other requirements can generally be fulfilled by exercising proper care and by 
 permitting nothing other than first-class materials and workmanship. 
 
 From the above review of the subject and actuated by many years of practical 
 experience, the author ventures the following opinion as to the general advisability of 
 adopting the fixed type of masonry arch for any particular case in hand: If the span 
 i is moderate and the rise h comparatively high, such that l/h will be between two and 
 four, then for bed rock foundation and open spandrels the fixed masonry arch 'may be 
 chosen as an appropriate type of structure. The design for such a bridge should be 
 based on a factor of safety of at least ten, allowing no tension unless carried by steel 
 reinforcement, and after the preliminary design is completed it should receive a final 
 analysis according to the theory of elasticity. 
 
304 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XV 
 
 When all this is followed by a most careful and approved system of construction, 
 it may frequently happen that the results will not prove entirely satisfactory to the 
 engineer. However, such a structure will be certain to last well, to cost little or nothing 
 for maintenance, and to be safe practically for all time, even though small cracks may 
 appear at any time after removing the falsework. 
 
 On the other hand, when the spandrel filling must for some reason be made solid. 
 and bed rock foundations are not available, also when the span is great and the rise 
 small, a fixed masonry arch should never be built. A hinged arch is the only rational 
 solution for such a case in the light of modern engineering experience, and that type 
 can be employed only when the spandrels can be so arranged as to allow of proper 
 expansion or rotation at the hinged points. 
 
 All this applies strictly to bridges and riot to small arched culverts and floors where 
 the hinged type is in many ways impracticable and the fixed type, preferably of reinforced 
 concrete, has given good satisfaction. 
 
 It is thus seen that the successful construction of a large masonry arch may justly 
 be considered a masterpiece of engineering achievement and requires far more intimate 
 knowledge and experience than is required for a steel structure of the same span. 
 
 Masonry and concrete are materials which are solely adapted to carrying compressive 
 stress. Hence, it is rational to reinforce this material to better resist tensile stress by 
 the introduction of steel, but this should be confined within proper limits and with the 
 sole aim of developing thereby the otherwise latent capacity of the material to do work 
 in compression. 
 
 When, therefore, steel is used to reinforce concrete in compression, the result is 
 bound to be wasteful, unprecedented, and a monument to engineering misconception. 
 Designs of this kind should be discountenanced, both from an aesthetic as well as from 
 an engineering standpoint, because they imply a misapplication of the true purpose of 
 both masonry and steel. 
 
 ART. 71. PRELIMINARY DESIGNS 
 
 On the supposition that a fixed masonry arch is feasible only when the foundations 
 are rigid, it is fair to assume that the fixed conditions will remain unchanged even 
 though the transmission of stress may call forth very slight elastic changes in the abut- 
 ments. 
 
 In any problem the given data are usually the live load to be carried, the clear span 
 and the clear rise of the intrados and the allowable unit stress. The elevation of the 
 roadway is also known. 
 
 The preliminary design would then include a reasonably accurate determination of 
 the thickness and shape of the arch ring at all points over the span. 
 
 To accomplish this, it is necessary first to decide on the general outlines of the 
 bridge, particularly on the design of the spandrel filling and the roadway, which should be 
 completely detailed before proceeding to the design of the arch ring. By this means 
 a large portion of the dead load can be accurately estimated, leaving the bare arch ring 
 as the only variable portion of the dead load. 
 
ART - 71 FIXED MASONRY ARCHES 
 
 305 
 
 Since the dead load for masonry arches is usually large compared with the live 
 load this procedure is clearly indicated. 
 
 The next step should be to determine the axis of the arch ring so that it will 
 coincide with the resultant polygon for an average load, which latter is taken as the 
 dead load plus half the uniformly distributed live load. In preliminary designs the live 
 load, whatever it may be, is always taken as uniformly distributed. 
 
 The shape of the arch center line thus obtained will be the one yielding the most 
 economic sections for the unsymmetric maximum and minimum live loads. 
 
 However, this center line cannot be directly found because the resultant polygon 
 upon which it depends must be determined before the dimensions of the arch ring and 
 its weight are definitely known. 
 
 Hence, this part of the problem must be solved by successive approximations and 
 trials. We must seek to find such an arch ring for which the resultant polygon will 
 coincide approximately with the center line. 
 
 With the aid of approximate formulae, the thicknesses of the arch ring at the crown 
 and abutments are determined, and from these the radius of curvature at the crown is 
 approximated to find the first arch center line. The weight of this arch ring is now 
 estimated and combined with the weight of roadway arid half live load, and a resultant 
 polygon is then drawn through the centers of the crown and springing sections. This 
 then gives a criterion as to the modifications required to make a second approximation 
 for a correct center line. This process is continued until a center line is found which 
 coincides closely with the resultant polygon. 
 
 The maximum stress for unsymmetric loading is now estimated from another 
 resultant polygon drawn for the critical quarter points and this then furnishes the 
 criterion for the thickness of the arch ring. If this stress is sufficiently close to the 
 allowable stress, the preliminary design may be accepted and the final rigid analysis 
 may then be undertaken. Otherwise further alterations in the dimensions must be 
 made until the preliminary design is sufficiently close to the assigned requirements. 
 
 The approximate formulae and method of conducting this investigation will now 
 be given, and this will constitute a complete solution as far as is possible by the use of 
 the older, resultant polygon method. 
 
 Fig. 7lA represents a symmetric arch ring of unit thickness. If the arch were 
 slightly unsymmetric, the present solution might answer for the symmetric portion and 
 the resultant polygon could afterward be extended into the unsymmetric portion which 
 was omitted in the first analysis. However, this could not be done in the final rigid 
 analysis by the theory of elasticity and should never be considered for badly unsymmetric 
 structures. 
 
 All pressures and loads will be expressed in terms of cubic feet of masonry per square 
 foot of surface, so that all areas on the drawing icpresent true relations of loads from 
 which actual weights are obtained by multiplying the areas by the weight of one cubic 
 foot of masonry. This is the most convenient unit for graphic solutions, as it does away 
 with estimating the weight of the arch ring and spandrel filling. 
 
 The following data are supposed to be given : The system of loading; width of roadway ; 
 span and rise of intrados; and the allowable unit stresses of the masonry or concrete. , 
 
306 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XV 
 
 The first step should be to sketch a general outline of the structure in sufficient 
 detail to design the roadway and the spandrel filling. This will be considered completed 
 and will not be discussed here. 
 
 If the bridge is to be for a single track railway for Cooper's E 60 loading, span 150 
 ft., rise 50 ft., and width of arch ring 13 ft., then the uniformly distributed load will be 
 1070 pounds per square foot, and taking the weight of one cubic foot of concrete equal 
 to ?- = 140 Ibs., then the load in terms of cubic feet of masonry would be 1070-:- 140=7.64 
 cubic feet = p. This includes 67 per cent for impact. 
 
 The floor, ballast, ties, rails, etc., complete, will weigh 710 Ibs. per sq. ft. according 
 to a design previously made, and this gives the dead load of floor q =710/140 =5.07 
 cu.ft. of masonry. Since the floor contains considerable steel, its actual thickness is 
 not q, but in the example chosen it was only 4 ft. 
 
 The preliminary design of the arch ring is now undertaken, referring to Fig. 7lA, 
 for the lettered dimensions. 
 
 LIVE LOAD 
 
 r~- ^ r 
 
 ^T 
 ^^ d 
 
 -~r. v > .r- 
 
 ^, \ 1 _,-" 
 
 ~~*T 
 
 
 1 
 
 
 FIG. 7lA. 
 
 The thickness of the arch ring at the crown is approximated from the formula? of 
 A. Tolkmitt,* which are rational but approximate, and, therefore, furnish much better 
 results than purely empiric formula?.. 
 
 Thus for a uniform live load p/2 over the whole span, and an allowable unit stress 
 0.4/, which is supposed to be 0.4 as great as for the unsymmetric critical loading, the 
 crown thickness should be 
 
 4+T5)' 
 
 t for feet or meters, (71 A) 
 
 where the allowable unit stress / is likewise expressed in cu. ft. of masonry, as was done 
 for the loads. 
 
 * Leitfaden fuer das Entwerfen Gewoelbter Bruecken, 1895. 
 
ART - 71 FIXED MASONRY ARCHES 307 
 
 For the unsymmetric live load, extending over half the span to the center, the 
 following value of Z) is obtained : 
 
 where G = $(q + ?-+*} f for feet or meters, (71s) 
 
 \ 2i Au/ j 
 
 When the value from Eq. (7 IB) exceeds that given by Eq. (71 A) then the larger 
 dimension should be adopted. 
 
 It is clearly seen that the full allowable unit stress / could not be taken in the 
 above formulae, because this average case of loading will stress the arch ring only about 
 half as much as when the resultant polygon passes through the middle third points. In 
 all arch designs where tensile stresses are prohibitive, this condition will govern. 
 
 The next step is to choose a preliminary shape for the intrados such that, for the 
 case of dead load plus half live load over the whole span, the resultant polygon will 
 exactly coincide with the arch center line. 
 
 At this point all theory fails and the judgment and experience of the engineer must 
 guide in making a suitable first approximation. However, the following practical 
 suggestions will serve a valuable purpose. 
 
 Theoretically the intrados can never be a true circle nor a true parabola for any arch 
 which is made to follow the resultant polygon, as above required, for economic reasons. 
 
 The resultant polygon becomes a parabola when the total load is uniform- per foot 
 of arch. This can never happen in a real arch, even if the spandrel filling were neglected, 
 because the stresses in the arch ring increase from the crown toward the abutments, 
 thus necessitating a variable thickness of ring, increasing with the stress. 
 
 On the other hand the resultant polygon becomes a circle when the superimposed 
 load at the springing points becomes infinite, a condition which is never attainable. 
 
 Hence an arch of economic design and shape to fit the resultant polygon for average 
 loading will have an intrados which ^neither a circle nor a parabola, but a curve 
 lying between these two. 
 
 Having thus established the limits between which the true intrados must be situated 
 and knowing from experience that the true line falls nearer to the mean of the two curves 
 than to either one of them, for all open spandrel arches, it becomes an easy matter to 
 approximate the intrados. 
 
 Hence, for open spandrels with roadway supported on piers or columns, the intrados 
 may safely be taken half way between the circle and the parabola, both drawn through 
 the three given points of the intrados. 
 
 The radius of the circle is given by the formula 
 
 and the parabola passing through the same three points is 
 
 which gives the coordinates y =ho/l and x =/ /4 for the quarter points, see Fig. 7U. 
 
308 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XV 
 
 From this a mean point at each quarter span is readily determined and a three- 
 center curve is then made to pass between these two and the given crown and abutment 
 points. 
 
 When the spandrel filling is solid, then the intrados approaches nearer to the circle 
 and for a plain arch ring without any spandrel loading, the true intrados approaches 
 the parabola. Between these limits the designer will soon be able to make very close 
 approximations even for the first trial. 
 
 To draw the extrados, no good rules can be given, though a fair approximation can 
 be made by applying the formula for the thickness of any point of an arch ring in terms 
 of the thickness DO at the crown, as follows: 
 
 = V / .- . . . /;.;. (71B) 
 
 cos 9 
 
 For flat arches this gives good approximations, but for semicircular arches the 
 formula should not be used for angles greater than 30 from the crown. Even for 
 <=30 the values are a little large and for (/> =90, D = GO , which is clearly impossible. 
 
 However, the 30 from the crown include the most important portion of the arch 
 ring, so that the remainder of the extrados can be approximated with sufficient accuracy 
 for the first trial. 
 
 Another method is to sweep an arc with radius r'=r+Z)o through the center of the 
 crown joint, which will be a good approximation for a center line between crown and 
 quarter points. 
 
 It is not prudent to spend more time in preliminary guessing, but with the dimensions 
 above given proceed at once to the graphic analysis. 
 
 For this purpose the arch should be drawn on a scale of at least 1 : 100, using a scale 
 1 ft. =3000 cu. ft. of masonry for the scale of forces. For small structures larger scales 
 may be employed. 
 
 The arch ring is now divided up into suitable sections corresponding with the panel 
 lengths chosen for the spandrel piers and the areas of these sections and of the piers are 
 then computed from the drawing. 
 
 Having thus found all the areas, including the uniform floor load q per ft. of bridge, 
 we may proceed to draw the resultant polygon for a live load p/2 extending over the 
 whole span. This polygon should be passed through the center points of the crown 
 and abutment joints and if it coincides quite closely with the assumed arch center line, 
 then the investigation may be continued for the unsymmetric loading, otherwise the 
 shape of the arch ring must be corrected and the first resultant polygon must be recon- 
 structed. A second trial nearly always gives acceptable results, and sometimes the first 
 trial is sufficiently close. For symmetric loading the resultant polygon is drawn for 
 the half span only. 
 
 With the dead loads previously found and a live load covering half the span and 
 extending past the crown to the load divide i for the critical point m, see Fig. 7lA, now 
 pass a resultant polygon through the points a, b, and n. The point a is the outer third 
 point of the arch section at A, the point b is the inner third point at B and the point n 
 is about Do/8 above the center line at the crown. This is the first approximation. 
 
ART. 72 FIXED MASONRY ARCHES 309 
 
 If this resultant polygon remains within the middle third at every point and crosses 
 the axial line in at least three points, then the design is acceptable. If the polygon 
 does not touch the middle third point at the two critical sections m and s, Fig. 7lA, 
 then the ring is too thick. If the polygon goes outside the middle third at one of these 
 sections and remains inside at the other, then the point n should be shifted and a new 
 polygon should be passed through a, b, and n. However, if the resultant polygon 
 passes outside the middle third at both critical sections, then the arch ring must be 
 made thicker and the investigation is then repeated. 
 
 Lastly the stresses at the four sections a, m, n, and b must not exceed the allowable 
 unit stress as given by Eq. (69A). 
 
 When all these requirements are fulfilled, the crown section should be tested for 
 full live load over the entire span, which is easily done from the loads already found for 
 the half span, since this is a case of symmetric loading. 
 
 This then constitutes a complete design according to the ordinary graphic method, 
 and the arch ring so found should now be subjected to a rigid analysis according to the 
 theory of elasticity which follows. 
 
 The above outline will be exemplified more fully when presenting a problem at the 
 close of this chapter. 
 
 The criterion for position of moving loads to produce maximum and minimum stress 
 is discussed in another <place. 
 
 ART. 72. 
 
 Introductory. A fixed, solid web arch must be treated as a structure with three 
 external redundant conditions according to Eq. (3c) and Fig. SF. Hence there is very 
 little difference between a fixed framed arch and a fixed solid web or masonry arch, 
 since the complications arising from external redundancy are alike in both cases while 
 the internal stresses must be found by the methods peculiar to frames and isotropic 
 bodies, respectively. 
 
 The method of treating the external redundant conditions involves the use of 
 deflection polygons for the determinate principal system, and in this connection a 
 further difference exists in the manner of determining the elastic loads for the two types 
 of fixed arches. 
 
 The elastic loads w for a framed structure are easily expressed in terms of the 
 geometric relations existing between the members and the angles composing the frame. 
 For isotropic bodies the elastic loads are functions of the cross-sections obtained by 
 integration and expressed in terms of moments. 
 
 Hence, with due regard to these differences, \vhich are easily distinguished, both 
 framed and solid web, fixed arches may be analyzed in precisely the same manner. 
 However, the general subject of solid web and masonry arches involves many consider- 
 ations not met with in framed arches. For this reason the solid web arch will be treated 
 in full detail in the following articles. 
 
 Regarding the reliability of the method, which is based on the theory of elasticity, 
 
310 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. XV 
 
 the author would add here that for steel arches, either framed or solid web, the method 
 is entitled to full confidence, while for stone and concrete it is reliable when the design 
 is based on the condition that tensile stress does not occur. In attempting to design 
 in concrete for equal tensile and compressive stress by reinforcing with steel, the method 
 must be regarded as more or less approximate, since it involves the modulus E of the 
 material, which is quite uncertain owing to the heterogenous character of masonry and 
 concrete generally, and especially when the latter is combined with steel. The intro- 
 ductory remarks of Art. 52 are especially applicable here. 
 
 General relations of the external forces to the principal system. The considerations 
 under this heading, in Art. 52, are again applied to the analysis of the solid web arch, 
 and the principal system is chosen as a determinate beam on two supports. 
 
 FIG. 72A. 
 
 Fig. 72A shows the principal system for an unsymmetric arch with the redundant 
 conditions acting as external forces on the left-hand abutment. Similar redundants 
 act on the right-hand abutment, and these are equal and opposite to the set shown for 
 the left-hand end. See Figs. 52n and 52c. 
 
 The arch ring aa'bb', loaded with a single load P, is referred to coordinate axes 
 (x, y) choosing the y axis vertical and the x axis making some angle /9 with the horizontal. 
 The origin is supposed to be known and is fixed by certain geometric relations to be 
 established later. 
 
 The ordinate y m , of any point ra of the gravity axis or axial line of the arch ring, 
 is measured vertically Irom the x axis, while the abscissa x m> of this point, is measured 
 horizontally from the y axis instead of parallel to the x axis. This is a mere matter of 
 convenience. 
 
 The span AB=l is taken as the line joining the intersections between the axial line 
 
ART. 72 FIXED MASONRY ARCHES 311 
 
 and the verticals through a'\ and b'\. The support at a\ is assumed as hinged while 
 the one at b\ is made movable for the principal system. 
 
 The single load P then produces reactions RI and RZ, intersecting in the point C 
 and passing respectively through the' unknown points a^ and 61 on the verticals through 
 the supports a\ and b\. The triangle aiC&i thus becomes a resultant polygon with 
 the closing line a\b\. The reaction RI may be resolved into the vertical component A 
 and the haunch thrust H' along a\b\. The reaction R% may be similarly resolved into 
 the vertical reaction B and the thrust H', which latter is equal and opposite to H' 
 acting at a\. 
 
 The vertical reactions A and B are the same as for a simple beam of span I on 
 determinate supports. Hence, 
 
 A =j(l-e) and ' B =*. . ....... (72 A ) 
 
 Also, the moment for any point m of the simple beam, equals 
 
 -\f i\f 
 
 osa or tf--i--, .... (72 B ) 
 
 where K is the ordinate ig through m, and H is the horizontal component of H' . 
 
 Therefore, the resultant polygon a\Cb\ becomes fixed whenever H' and the closing 
 line a\b\ are found. 
 
 The origin is connected to the arch along aa' by the rigid disc aa'O and to this 
 origin are applied two equal and opposite forces H f , which are equal and parallel to the 
 original hanuch thrust acting at a\. The equilibrium of the principal system and of 
 the fixed arch thus remains undisturbed. 
 
 Suppose now that all the external forces to the left of a section tt' act on the 
 principal system only, and that the three forces H' and the vertical reaction A are 
 applied to the rigid disc and are thence transmitted to the principal system. Then the 
 force H' at a] and the opposing force H' acting at 0, form a couple with lever arm 
 z cosa, producing a moment X a =H'z cosa. The other force H', acting at and to 
 the right, may be resolved into two components X b and X c , where X b is vertical along 
 the y axis, and X c acts along the x axis. The external forces to the left of the section 
 and applied to the principal system are then P, A , X b , X c , and a moment 
 Xa=H'z cos a = Hz . Of these the two forces X b and X c and the moment Xa constitute 
 the redundant conditions while the forces P and A are known, and all are applied to 
 the principal system to the left of the section tt'. A similar set of external forces (not 
 shown) acts on the principal system to the right of this section. 
 
 The moment of all the external forces abou'c any point m of any structure, involving 
 three external redundant conditions, is given by Eq\ (7 A) as 
 
 M m ^M om -M a X a -M b X b -M c X c , ....... (72c) 
 
 wherein 
 
 M om =-A (li x m )Pd=the moment about m due to the load P acting on the 
 principal system, This is condition X=0. 
 
312 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XV 
 
 M a = l is the moment about m due to the moment X u = l applied to the principal 
 
 system. Condition X a 1 . 
 Mb = 1 x m is the moment about m due to the force X b = 1 acting on the principal 
 
 system. Condition X b = 1 . 
 M c = l-y m cos /? is the moment about m due to the force X c = l acting on the 
 
 principal system. Condition 'X c = 1 . 
 
 Substituting these values into Eq. (72c), gives the following fundamental moment 
 equation for fixed arches: 
 
 M m =M om -l-X a -x m X b -y m cospX e , . (72D) 
 
 Before M m can be determined for any point m of the arch, the three redundants 
 X a , Xb, and X c must be evaluated from three simultaneous work equations of the form 
 of Eqs. (44s) which may be made to apply to the present problem by so locating the 
 (x, y) axes that d ab =d ba =Q, ^=^=0 and d bc =d cb =0. 
 
 The Redundant conditions for a single load P 1 then become 
 
 ^w>f " "wZ> " "me 
 
 for fixed abutments and omitting temperature effects. 
 
 Neglecting the term involving the axial thrust N in Eq. (15N) the deflections in 
 Eqs. (72E) may be expressed as virtual work in terms of the moments M om , M a = l, 
 M b =x and M e =y cos/?, as follows: 
 
 ~M om du C Mldu Cdu 
 
 _ fM om M a du _ ( 
 dma ~J~~~ET ~J 
 
 rM om M b du fM om xdu C 
 
 8* =) - - Er -J -^7; fa - J 
 
 El 
 
 j~. r ~\/r1j,. r-.vj,. 
 
 (72r) 
 
 El 
 
 CM om M c du CM om y cos 8du C M 2 c du A/ 2 cos 2 3du 
 
 O mc =J ^J E j -I ^ = J E J~ = J gf- 
 
 Introducing the elastic loads 
 
 du du du 
 
 v7 =w a, x~T^T =w b, and y-j^f =w c (72c) 
 
 Hit til ' hi 
 
 into Eqs. (72r), then substituting these values into Eqs. (72E) and replacing the inte- 
 grations by summations, then 
 
 V _m _ - om b -^ _ mc _ j omc . . 
 
 ~~~ ; Ac ~~ 
 
 The modulus E, being involved in both numerator and denominator, of Eqs. (72n), 
 has no effect on the redundants except when temperature and reaction displacements 
 are included. 
 
 If the axial thrust is to be considered, then the deflections due to this effect alone 
 
ART. 72 
 
 FIXED MASONRY ARCHES 
 
 313 
 
 and for N a =Q, A\ = 1 sin (f> and A' c = 1 cos </>, according to Eq. (15N), become for 
 du=dx/cos (f): 
 
 /A T ~ / fl ' 2 rk A 
 
 -fij- = J E p CO^A 
 
 fN c 2 du _ fcos 2 fi du _ C l cos < Ax 
 CC= J ~Ej r= J EY Jo ~EF~ 
 
 (72j) 
 
 These values added for total effect includin axial thrust in the determination of 
 
 the redundants will change Eqs. (72n) to the following: 
 
 X a =- 
 
 I,M om w c 
 
 sn 
 
 (72K) 
 
 EF cos (f> 
 
 
 Eqs. (72n) and (72x) being written for a single load P = l represent the equations 
 of the influence lines of the redundants. 
 
 If axial thrust is to be considered in Eqs. (72K) it will be well to compute the two 
 functions in the denominators and thus include them in the pole distances when drawing 
 the Xb and the X c influence lines. 
 
 It should be pointed out that in general, axial thrust becomes important for very 
 flat arches, in which case the function S sin 2 <f)Ax/EF cos $ becomes very small, while 
 cos <f>Jx/EF becomes large. Hence it is never necessary to include the thrust function 
 in determining X^, while for X c it may take on considerable proportions. 
 
 Regarding the curvature of an arch, which is involved in the integral J du and which 
 
 could be fully considered should the necessity arise, it must be remembered that the 
 depth of the arch ring is. almost without exception, very small compared with the 
 radius of curvature of the axial line, and no appreciable error is committed when this 
 effect is entirely neglected. Some comparative analyses have been worked out by 
 various investigators, which show quite conclusively that this error is smaller than the 
 knowable accuracy of the strength of materials, to say nothing about temperature 
 stresses and abutment displacements, which may be estimated but never become known. 
 
 For these reasons, which are considered ample, the question of axial curvature will 
 not be treated here. 
 
 The location of the coordinate axes was to be so chosen that the displacements 
 o<j6 = ^ar"=^c ;== 0, thus permitting the use of the simplified Eqs. (44s) instead of the 
 involved form of Eqs. (8b) . 
 
 According to Eq. (15is T ) and noting that N a =0, these displacements become 
 
 fM a M b du Cxdu v 
 
 ab== J TH~ = J ~w = ^ x ' 
 
 CM a M c du_ (ydu 
 dac ~J El "JET ^ y 
 
 CM b M c du fxydu v 
 bc = J ~KL~ = J FT = ^ x 
 
 , o = 2w c =0 
 w=0 
 
314 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. XV 
 
 If the coordinate axes are located so as to comply with the conditions of Eqs. (72L) 
 then the origin of these axes must be the center of gravity of the elastic loads w a because 
 %xw a =Q and 'i/w a =Q. Also the axes must be conjugate in order that the centrifugal 
 moment ^lxyw a 0. 
 
 The manner of computing the elastic loads and then of finding the positions of the 
 axes with respect to any unsymmetric arch ring follows here. 
 
 ART. 73. COORDINATE AXES AND ELASTIC LOADS 
 
 Unsymmetric arches. The elastic loads w a depend on the geometric shape of the 
 arch sections while the loads Wkxw a and w c =yw a are seen to be functions of the w a 
 loads and of the coordinates of the axial points of the arch. Hence the w a loads may 
 be found for any arch ring of given sections while the w^ and w c loads require for their 
 determination a careful location of the coordinate axes in compliance with the conditions 
 given by Eqs. (72b) . 
 
 The first step is then to evaluate the w a loads for finite distances Ju, which is done 
 
 by dividing up the arch ring into an 
 even number of lengths and 
 puting El for each section, 
 
 W a = I w a = I -=-7 are found by summa- 
 J Jo Jiil 
 
 tion according to Simpson's rule. This will 
 answer all practical purposes and will be 
 the more accurate the shorter the lengths 
 Au or the greater the number of sections 
 taken. 
 
 The manner in which this is done 
 will now be illustrated, as the accuracy 
 of the analysis depends almost entirely on the values W a . See Fig. 73 A. 
 
 Instead of dividing the axial line into small lengths Ju, it is preferable to make 
 these spaces equal horizontally, so that Ju = Ax/cos (f>, where < is the angle which the 
 normal section makes with the vertical. The Eqs. (72c) then become 
 
 Au Ax xAx , yAx 
 
 after com- 
 the values 
 
 Ix'^ix* 
 
 T 
 I 
 
 X , ^ 
 
 T N * 
 
 M 
 
 i ' i - 
 
 I't-M 
 
 'I 
 
 i 
 
 If 
 
 B 
 +\ 
 
 *** 
 
 i 
 
 x^x^x-*.x4 
 
 *-" 
 
 
 
 i - 
 
 
 
 FIG. 73A. 
 
 Wa El El 
 
 w b =xw a = 
 
 El cos 
 
 and 
 
 El cos 
 
 > [ (73 A) 
 
 Each half of the arch is now spaced off into an even number of spaces Ax, beginning 
 at the crown. When the spaces Au appear to get too large they may be shortened 
 toward the haunches of the arch, but it is best to keep as many as possible of the Ax alike. 
 
 Now tabulate the depths D of the arch ring and angles < for the axial points from 
 A to B and figure the reciprocal values of El cos <j> for each. If the section is one of 
 a steel girder the real moment of inertia is to be used while for a masonry arch it is best 
 to take a section of width unity and depth D making I=--D 3 /12. 
 
 Simpson's rule is now employed to sum the values w a in order to obtain finite 
 
 r/* I x / 
 
 w=== ~~~~ for the 
 
ART. 73 
 
 FIXED MASONRY ARCHES 
 
 315 
 
 The quantities 1/EIcoscf) are treated as parallel ordinates to some irregular curve 
 and the volume of length I, which is divided into an even number n of equal spaces Ax-, 
 is then iven from 
 
 W a =-^-[w +4wi+2w 2 +4 : W3+2w 4 to 4u' n _i 
 
 o 
 
 (73s) 
 
 One such compression can be written for any even number of equal spaces Ax, Ax', 
 or Ax". 
 
 To find W a =\w a for each of the sections, Simpson's rule for three quantities is 
 then applied. This is the well-known prismoidal formula for volume and is used here 
 to figure volumes between successive mean areas taking the originally computed values 
 w for the middle terms. 
 
 Since the loads W a are wanted for the points m themselves and not between these, 
 we interpolate mean values between consecutive pairs of computed values and then 
 figure the sums W a , using the computed W a as the middle term in the formula. 
 
 Thus the formula, 
 
 Ax 
 w a3 =-^-[w 2 +4:W 3 +w 4 ], (73c) 
 
 as here applied, becomes 
 
 Ax \w 2 +w s , w 3 +WA~\ 
 
 w ^=-^- -V--+4w 3 +- M, (730) 
 
 6 [ 2 2 J 
 
 when the interpolated means are used. This gives a much better check on Eq. (73s) 
 than could be obtained by the use of Eq. (73c). 
 
 Beginning at A, and making - I .=w _i, - -=wi_ 2 , etc., the several values 
 
 
 
 of W a are as follows: 
 
 Ax' 
 i'Wo = ^-\2wo +WQ _il = half load for an end section 
 
 D 
 W - 
 
 6 
 
 W a2 = - _ 
 
 2 
 
 Ax 
 
 a3 = [MJ 2 _ 3 
 
 Ax 
 
 etc., 
 
 w ^x 
 W a s=- Q - 
 
 etc., 
 
 etc. 
 
 Ax" 
 
 _ 8 +2lV 8 ] +[2Wg +W> 8 _9 
 
 etc. 
 
 =- -Wn -12 
 
 (73E) 
 
316 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 and as a final check the sum of all W a loads must be by Eq. (73s) 
 
 CHAP. XV 
 
 Xi A X ' j x 
 
 w a =-Hi0 +wi +w 2 ] + -x- 
 o o 
 
 4 +4w 5 +2w 6 
 
 Ax" 
 +4w 7 +w 8 ] + [w s +4w g +2w w +4wn 
 
 o 
 
 /dx C Ax 
 
 w a . these are now used 
 El cos Jo 
 
 to determine the center of gravity and the angle /? which the x axis makes with the 
 horizontal. The y axis was taken vertically and is thus fixed as soon as the origin is 
 located. 
 
 +v 
 
 FIG. 73s. 
 
 The center of gravity is located with respect to any assumed rectangular axes 
 (z, v), Fig. 73B, with origin at the axial point A the same as shown in Fig. 73A. It is 
 most convenient to assume the z axis vertical and the v axis horizontal. The coordinates 
 (z, v) are found for each of the several axial points m previously used for the computation 
 of the W a loads, and all are tabulated together. Finally the moments zW a and vW a are 
 found and from these the coordinates l\ and z, 7 of the origin are obtained from 
 
 / v rr a , . *-*a 
 
 Z I=^TI^; and -^ 
 
 (73F) 
 
 This also fixes the y axis which is parallel to the vertical z axis through the center of 
 gravity 0. The x axis, while passing through 0, makes some angle /? with the horizontal 
 such that I,xyW a =Q, according to the last of Eqs. (72L). 
 
 The (x, y) coordinates may be derived from the (z, v) coordinates when the angle 
 ,5 is determined. Taking /? positive when measured to the right of the origin and above 
 the horizontal, as shown in Fig. 73s, then 
 
 x =li v and y =z z ' +x tan /? 
 
 (73o) 
 
ART. 74 
 
 FIXED MASONRY ARCHES 
 
 317 
 
 The angle /? is found by substituting the value of y from Eq. (73c) into the condition 
 equation %xyW a =Q, giving, as in Eq. (52j), 
 
 (73H) 
 
 The abscissae x being known from Eqs. (73c) the values HxzW a and 2x 2 W a are 
 readily found and tan /? is then obtained from Eq. (73n) and the new axes and (x, y) 
 coordinates are thus determined. 
 
 For symmetric arches the above calculations are greatly simplified and the coordinate 
 axes are readily located as follows: The y axis, being the axis of symmetry, is practically 
 given when the shape of the arch is given, and the angle 0=0 because for symmetric 
 forces the centrifugal moment ( xyW a =0 for any pair of rectangular gravity axes. 
 
 Hence the y axis and the direction of the x axis are given and the origin is located 
 by computing the ordinate z ' from any assumed horizontal v axis, giving for symmetric 
 
 arches, 
 
 SzW 
 ' ; 0=0; and y=z-z r ....... (73j) 
 
 In any case the W a , W b , and W c loads are now readily figured and from these the 
 equations for the influence lines for the redundant conditions X a , Xb, and X c are evaulated. 
 
 Solid web circular arches which are only very slightly unsymmetric may be approx- 
 imately analyzed by extending the short half to a point of symmetry and treating the 
 structure as a symmetric arch. However, when the difference in the elevations of the 
 springing points is appreciable, such approximations should not be made. 
 
 The center of gravity may be found graphically by combining the W a loads into 
 two equilibrium polygons, for the loads acting first vertically and then horizontally 
 through the centers of gravity of the arch sections. The two resultants thus obtained 
 will intersect in the required point 0. The moments of inertia 5ix 2 W a and Hiy 2 W a may 
 be obtained from the same polygons by Professor Mohr's inethod. 
 
 ART. 74. INFLUENCE LINES FOR X a , X 6 , X c , AND M m 
 
 Having located the coordinate axes (x, y) to fulfill the requirements d a b^ae ^bc = 
 the Eqs. (72E) and (72H) now become applicable to the present problem, giving 
 
 X a = 
 
 ~"2W ~JT~ riar 
 
 j vv & J-^a 
 
 1<\ <N 
 
 U mo (-* mo 
 
 ~ ~^rW, = ~tf7 = ^ 6 " ; 
 
 cos 
 
 . f\ 
 
 -Jj '/cm 
 
 (74A) 
 
 These Eqs. (74A) which are written for a load P m = l, thus represent the equations 
 for the influence lines of the redundants X a , X b , and X c . 
 
318 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XV 
 
 For X a = l, the first of Eqs. (74A) gives l-^W a = l'd ma = ^M om W a . Therefore, 
 the special moment M om equals unity and this equals the moment X a = 1 acting on the 
 principal system, which would follow from Maxwell's law. Hence, an equilibrium 
 polygon drawn for the loads W a with pole unity, would give the deflection polygon 
 for d ma due to a moment X a = I applied to the principal system. Also, the ordinates 
 of a similar equilibrium polygon drawn with a pole H a = ^W a would represent the 
 ordinates rj aw , of the X a influence line, measured to the scale of lengths of the drawing. 
 
 The influence lines for Xj, and X c are similarly found from equilibrium polygons 
 drawn for the elastic loads Wb and W& with pole distances respectively equal to 2xW b =H b 
 and cos pEyW c =H c . When axial thrust is to be considered then the pole distance 
 H c =cos p^yW c + S cos 2 <j>du/EF =cos ^yW c + 2 cos <j>Ax/EF from Eq. (72K) . 
 
 The M m influen.ce line for moments, about any point m- of the axial line and a 
 moving load P m =l, is now derived from Eq. (72D). Accordingly the ordinates jj m , 
 of the M m influen.ce line, must represent the algebraic summation of the influences due 
 to Mom, X a , Xb, and X c expressed by the equation 
 
 y m =M m =M om -X a -x m X b -y m cos t 8X c ...... (74s) 
 
 The influence line for M om is the ordinary moment influence line drawn for the 
 point m and for a simple beam on two determinate supports at a'\ and b'i, Fig. 72A. 
 It is a different line for each axial point m. 
 
 The influence lines for X n , X b , and X c remain the same for the same structure, but 
 the ordinates of the Xb and X c lines, according to Eq. (74s), require multiplication by 
 the variable coordinates x and y cos {3, which differ for each axial point m. 
 
 Hence the M m influence line is best constructed by computing the ordinates 
 (X a +x m X b +y m cos 3X C ) for the- particular values of x m and y m cos/? and plotting these 
 ordinates negatively from the M om influence line. 
 
 The M m line, so found, will serve to determine the bending moment for the axial 
 point m due to any position of a system of concentrated loads. 
 
 The normal thrust N m , and stresses on any normal arch section through m, are 
 then computed as described in Art. 49 and further discussed below by involving the 
 moments about the kernel points e and i. 
 
 ART. 75. TEMPERATURE STRESSES 
 
 The effect ,of a uniform rise in temperature is to expand the principal system 
 uniformly in all directions. But since a uniform vertical expansion is not resisted by 
 the fixed abutments, it may be assumed that the whole temperature stress is produced 
 by horizontal expansion. This manifests itself by a bending moment extending over 
 the entire arch ring, and produced by resisting abutments. 
 
 Thus for a rise in temperature, the extreme elements of the extrados at the crown 
 are in tension while at the haunches they are in compression, when the arch is not 
 otherwise loaded. The simultaneous condition of the intrados elements is exactly the 
 reverse. 
 
ART. 75 FIXED MASONRY ARCHES 319 
 
 In Eqs. (72E) the temperature term was neglected. It is now separately considered 
 without regard to the effect of the loads P. 
 
 According to Eqs. (44c), which apply to the present problem, the redundant 
 temperature stresses, for rigid abutments, become: 
 
 X at =^=0; A" M =A'=0; and X ct = d ^ ....... (75A) 
 
 Oaa Obb O cc 
 
 The first two are placed equal to zero b.ecause uniform temperature changes cannot 
 produce rotation nor -vertical deflection of the principal system, which latter is here 
 regarded as a simple beam on two supports. 
 
 The distance over which the change d ct is cumulative must be the projection of the 
 span / on the x axis, hence d ct =etl/cos^ and from Eqs. (72F) d cc = 2?/ cos 2 (3W C , giving 
 for the last of Eqs. (75A) : 
 
 ' where w --* w '- (75B) 
 
 The moment about any axial point m due to temperature alone, may now be found 
 from Eq. (72o), wherein M ow =0, X at ^0, X bt =Q, and X^ is given from Eq. (75s). 
 Then 
 
 -M mt =y m cos ?X et =^ 2 *. . . . , .. . (75c) 
 
 According t'o Eq. (75c) the only variable affecting the moment M mt is the ordinate 
 y m . Hence for y m =0, M mt =0 and for y m maximum, M mt becomes maximum. The 
 points where the x axis intersects the axial line are thus points of zero moment for 
 temperature effects, while the crown and haunch points receive maximum effects. 
 
 The temperature effects must finally be combined with the dead and live load 
 stresses to obtain absolute maxima and minima. 
 
 It is a well-established fact that when masonry arches open up cracks this generally 
 occurs during the coldest winter weather. This is because masonry or concrete is 
 scarcely ever placed during freezing temperatures, but generally in the warmest summer 
 months. Hence, the variations in temperature from that existing at the time of closing 
 the arch ring are nearly always greater in the negative direction. Thus, if the ring was 
 closed at 70 F. the completed structure might at some future period attain a temperature 
 of perhaps 90, but it is quite certain to fall to about 10 if the climate is severe. 
 This then would subject the arch to temperature stresses resulting from +20 and 80 
 from the closing temperature. 
 
 When it is considered that a lowering in temperature will induce tensile stresses 
 in the spandrels and thus cause rupture at the points of least strength, it is clear why 
 this phenomenon is of such frequent occurrence. Rising temperature would tend to 
 increase the compressive stresses, but probably not to any alarming extent. 
 
320 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XV 
 
 ART. 76. STRESSES ON ANY NORMAL ARCH SECTION 
 
 This subject is fully discussed in Art. 49 and illustrated by Fig. 49s and will not 
 be repeated here except in so far as to show the application of Eqs. (49s) to the fixed 
 masonry arch. 
 
 For any rectangular arch section the kernel points e and i are located on opposite 
 sides of the axial line at equal distances k=D/Q from the axis. The kernel point i for 
 the intrados is situated above the gravity axis, while the kernel point c for the extrados 
 is below this axis as shown in Fig. 49s. 
 
 The moments M e and M t - are found from Eq. (74s) in which the coordinates of the 
 axial point m are replaced by the coordinates (x e , y e ) for an extrados kernel point and by 
 the coordinates (xi, yj) for an intrados kernel point. This is permissible because this 
 equation is perfectly general and is true for any point in the plane of the structure. 
 
 The moment equations for the kernel points e and i, of any normal section, are 
 thus 
 
 M e =M oe -[X a +x e X b +y e c.oa t 3X e ] j ( 
 
 Mi = M oi -[X a + XiX b + yi cos pX e ] I 
 
 Designating all influence line ordinates by y and using proper subscripts as a 
 distinctive feature, then the ordinates of the kernel point moment influence lines may 
 be expressed as 
 
 wherein )j , x, and y are special for each normal arch section it' while the ordinates jj a , 
 jjfe, and rj c are the same for all influence lines but differ for each point of the span. 
 
 When the kernel moments are known for any rectangular section and due to any 
 position of a moving train of loads, the stresses on the extreme fibers of the arch section 
 and the normal thrust N and its distance v from the axial line, may be determined 
 from Eqs. (49s) thus: 
 
 J)2 ' 
 
 (76c) 
 
 i=5 ; F^l-D; and /=~ 
 
 where D is the depth and F is the area of a normal arch section of unit thickness. 
 
ART. 76 
 
 FIXED MASONRY ARCHES 
 
 321 
 
 In Eqs. (76c) M and Mi have opposite signs when TV acts between the two kernel 
 points e and i. The stresses f e and /; take their signs from M e and Mi respectively, 
 and compression is regarded as a negative stress. The offset v is positive when measured 
 from the gravity axis toward the extrados and when so applied determines a point on 
 the resultant polygon for that section. 
 
 It is not necessary to draw the M, n influence lines because the M- e and Mi Tines 
 cannot be derived from the M m line, nor is it possible to determine N and v when M m 
 alone is known except by constructing a resultant polygon. 
 
 Hence the stresses on any arch section are best found from the kernel point moments 
 M e and Mi of that section. If the redundants X a , X^, and X c are evaluated from their 
 respective influence lines for a certain position of a train of loads, then the kernel 
 moments, for any section, may be computed from Eqs. (76A). However, it is preferable 
 to conduct the entire solution by means of influence lines, as will be illustrated 
 later. 
 
 Since different portions of the arch ring are situated in the four quadrants of the 
 coordinate axes, it may be well to show how to pass from the coordinates of any axial 
 point m to those of the kernel points of the normal section through ra. 
 
 For a rectangular arch section k e =ki D/Q, so that the following relations may 
 be written out, from Fig. 76A, in terms of the lettered variable dimensions D and <f> 
 for each point, thus: 
 
 For point A, x e = x-s' 
 
 For point m, 
 
 For point n j 
 
 
 x e = x -~- sin 
 
 Xi = x+-^-sm 
 6 
 
 x e =0 
 
 and 
 and 
 and 
 and 
 and 
 and 
 
 D 
 
 sr COS 
 
 For point m', x e = x -f-^- sin and 
 
 x -^ sin 
 6 
 
 and 
 
 For point B, -x e = -x + sin (j> and 
 
 yi = y ~i~ ~TT cos <^> 
 D 
 
 7/ e = y COS 
 
 COS (f> 
 
 D 
 
 D , 
 = y cos (p 
 
 ;= 2/ + " COS^> 
 
 D 
 
 x ~-si 
 
 and iji = y + cos 96 
 
 . (76D) 
 
322 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XV 
 
 The general equations for the moment influence lines of the two kernel points may 
 then be expressed in terms of the ordinates rj and coordinates (x, y) of the axial point. 
 Thus 
 
 D , A J D 
 
 r / 
 
 -Tioe |^a ^ 
 
 COS 
 
 
 01 
 
 , (76E) 
 
 where the coordinates with their proper signs are to be taken from Eqs. (76D). 
 
 ART. 77. MAXIMUM STRESSES AND CRITICAL SECTIONS 
 
 The maximum stress for any given section is determined from the moment influence 
 line for that section by placing the maximum train of live loads over the positive portion 
 of the influence line. Similarly the minimum stress in found by loading the negative 
 portion of the moment influence line. 
 
 Any point for which the TJ ordinate of a moment influence line- is zero, must be a 
 load divide, and there is no satisfactory way of locating the load divides other than by 
 drawing the influence lines. This will be illustrated in connection with the problem 
 which follows. 
 
 By the method of influence lines we are thus enabled to find the maximum and 
 minimum stresses for any particular section. 
 
 However, this does not afford sufficient knowledge regarding the safety of a structure 
 unless those sections are examined which receive the greatest maximum stresses for the 
 most unfavorable positions, of the live load considering the structure as a whole. These 
 sections of greatest maxima are called critical sections, and for any given structure all of 
 
ART. 77 FIXED MASONRY ARCHES 323 
 
 the critical sections must be investigated while all other sections are of minor importance 
 and usually receive no consideration. 
 
 Unfortunately the theory of elasticity does not afford a direct solution for finding 
 the critical sections. The critical sections might be located after a number of sections 
 all along the arch ring had been examined, and the points of greatest maxima would 
 correspond to the critical sections. However, this would involve much labor and from 
 the nature of the case is not warranted, since the stresses in an arch ring always change 
 gradually and not abruptly. An approximate knowledge of the location of the critical 
 sections is, therefore, all that is required and this may be had from a discussion of the 
 moment equation. 
 
 Thus the general equation for moments about any point m by Eq. (74s) is 
 
 M m =Mom -[X a +x m X b +y m cos pX c ] 
 
 wherein the parenthetic quantity may be positive or negative, depending on the location 
 of the point m. 
 
 Since the stresses are direct functions of M e and M{ for a given section, then these 
 moments and stresses attain maximum values simultaneously: 
 
 1. For X a negative and Xj, and X c both positive, when y m =0 or when x m =0. 
 This would locate three critical sections, one at the crown for x m =0, and two on the x 
 axis for y m =Q. 
 
 2. For X a , X b , and X c all positive, M m may become maximum when y m has its; 
 greatest negative value or when both x m and y m ha-ve their greatest negative values. 
 This locates two other critical points, one in each abutment. 
 
 Hence, there are five critical sections in every arch which must be carefully examined 
 for maximum and minimum stresses, and the positions of the moving loads for these 
 limiting stresses are given from the influences lines drawn for the five critical sections. 
 In the case of symmetric arches, only three critical sections require investigation. 
 
 The two critical sections for ?/=0 fall very close to the quarter points of the arch, 
 a fact which is important in making preliminary designs by the method of resultant 
 polygons. The load divide may then be approximately located in the manner given 
 for three-hinged arches by treating the crown as a hinged point. 
 
 The maximum stresses are not appreciably altered by taking the quarter point 
 critical sections a little to one side or the other of the real theoretical point, while the 
 stresses are greatly affected by the shape of the axial line of the arch ring. 
 
 Thus for maximum economy, the axial line should be a line of zero moments for the 
 case of average loading. That is M m should be zero for every point of the axis of the ring 
 when the arch is carrying a dead load and half the maximum live load over the whole svan. 
 
324 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. XV 
 
 ART. 78. RESULTANT POLYGONS 
 
 The resultant polygon for any case of applied loads P may be located by finding 
 the corresponding values of X a , X^, and X c from the three influence lines for these 
 redundants, having previously located the (x, y) axes. 
 
 The redundants X c and X b , Fig. 78A, were taken as the components of H' , respectively 
 coincident with the x and y axes. The angle /?, which the x axis makes with the hori- 
 zontal, is given by Eq. (73n) and the y axis was taken vertically. Hence the horizontal 
 component H , of X c , is 
 
 H=X c cos{3 and, from Fig. 72A, 
 
 tan a= 
 
 H' = 
 
 H 
 
 H 
 
 X c cos /? 
 cos a 
 
 cos a 
 
 X a _X a 
 H' cos a ~ H 
 
 H 
 
 Co =^! ^ (tan a -tan /?) 
 ti 
 
 X a =X c z cos fi=X b z - 
 c 
 
 (78A) 
 
 These dimensions fix the closing line a\b\ on the two end verticals of the abutments, 
 also the haunch thrust H' , all in terms of X a , Xb, and X c . 
 
 The reactions R\ and #2 may be found from the vertical reactions A and B and 
 the thrust H'. The vertical reactions are those due to loads P acting on a simple beam 
 
 = ^(le) P/l and B =^Pe/l. The 
 
 and may be computed from Eqs. (72A) as 
 reactions R\ and R% must then intersect on the line of the resultant R of all the applied 
 loads P. 
 
 A force polygon, Fig. 78A, is drawn by laying off all the loads P in proper succession, 
 dividing this load line into the parts A and B and at the dividing point drawing a line 
 parallel to aib\ of length equal to H'. This determines the pole of the force polygon 
 from which the reactions RI and R 2 and the resultant polygon through i and bi are 
 easily drawn. See also the force polygon in Fig. 52A, showing how the pole is located 
 when H', A , and B are given. 
 
 The resultant polygon thus found must intersect the arch center line at least three 
 times for unsymmetric loading on an unsymmetric arch, and four times for a symmetric 
 arch, provided the arch center line was so chosen as to be coincident with the resultant 
 polygon drawn for the case of average loading. This average loading consists of the 
 total dead load, plus half the live load uniformly distributed over the entire span. It 
 
ART. 78 
 
 FIXED MASONRY ARCHES 
 
 325 
 
 will be so understood whenever average loading is referred to in connection with arch 
 designs. 
 
 The most favorable resultant polygon is the one which coincides with the axial line 
 of the arch ring. Hence the most economic shape for an arch is the one for which the 
 axial line is chosen to be the resultant polygon drawn for the case of average loading 
 as just defined. The live load, in its critical positions, will then produce minimum 
 values for the moment M m =N m v by making the offsets v, between the axial line and 
 the resultant polygon, all minimum. The absolute minimum for M m =N m v can occur 
 
 FIG. 78A. 
 
 omy for v=Q at all arch sections, and this condition can be produced only for one case 
 of loading, which is taken as the average loading, when maximum economy is to be 
 achieved in the design. 
 
 For all cases of live loads other than the average, the moments M m --=N m v could 
 never become zero for all points of the axial line, though these moments would be 
 minimum for an arch properly designed for the average loading. 
 
326 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. XV 
 
 ART. 79. EXAMPLE 150 FT. CONCRETE ARCH 
 
 (a) Given data and preliminary design. The data. Required to design a single 
 track railway arch of 150 ft. clear span and 50 ft. clear rise with profile of surface and 
 foundations as shown in Fig. 79fi. The loading to be Cooper's EGO, allowing full 
 impact according to formula (1), Eqs. (64c), 300 -r-[l +300], where I is the portion of 
 span covered by loads, for any case of loading. 
 
 The material shall be concrete, mixed in the proportion of one part American 
 Portland cement to two parts sand to four parts crushed granite or limestone. Cubes 
 of 12 inches on each side, at the age of 6 months, shall sustain a compressive stress 
 of not less than 3600 Ibs. per sq. in. and briquettes 90 days old, mixed one part cement 
 to two parts sand, shall not break below 300 Ibs. per sq. in. in tension. 
 
 Accordingly the weight of one cu. ft. of concrete is assumed at 140 Ibs., and for 
 stresses between 100 and 600 Ibs. per sq. in. "=3,670,000 Ibs. per sq. in., and 
 
 13' 
 
 FIG. 79A Pier of Roadway 3'X 13'. 
 
 e =0.0000054 per 1 F. The allowable unit compressive stress will be 500 Ibs. per sq. 
 in. or 515 cu. ft. of concrete per sq. ft. of surface. All pressures will be expressed in 
 cu. ft. per sq. ft., a unit which is very convenient and nearly the same as Ibs. per sq. in. 
 Thus if 7--=144 Ibs., then Ibs. per sq. in. =cu. ft. per sq. ft. 
 
 Design of roadway It is .proposed to place the track on a floor built of concrete 
 and reinforced by rolled I-beams of sufficient strength to carry the whole wheel loads 
 over spans 10 ft. in length. This floor system is supported by concrete piers 3 ft. thick 
 and 13 ft. wide and these in turn transmit the loads to the arch ring. This floor is shown 
 in section Fig. 79A. 
 
 Each I-beam must carry two 30,000 Ib. wheel loads spaced five feet apart and 
 producing a maximum bending moment of 
 
 M =30,000X60-30,000X30=900,000 in. Ibs. 
 
 Allowing 97 per cent impact for a 10 ft. span, this moment becomes 1,733,000 in. 
 Ibs., and for an allowable unit stress of 16,000 Ibs., the required section modulus is 
 
ART. 79 FIXED MASONRY ARCHES 327 
 
 A// 16,000 = 111 in inches. One 20-inch I, 65 Ibs., has a section modulus of 117, which 
 is ample. 
 
 Weight of floor. 2 ft. of ballast at 120 Ibs. per cu. ft. . .3850 Ibs. 
 
 Concrete '. .5000 " 
 
 2 I-beams and cor. bars 150 " 
 
 2 rails 200 " 
 
 Total per foot, of roadway 9200 Ibs. 
 
 Total per sq. ft. of arch ring q=7lO Ibs. =5.07 cu. ft. 
 
 Live loads. E60 loading for 150 ft. span, all locomotives, gives 8400 Ibs. per ft. of 
 one track, and this, distributed over a 13 ft. wide arch, gives 640 Ibs. per sq. ft. Adding 
 67 per cent, impact for 150 ft. span, then the total live load p becomes 1.67x640 = 1070 
 Ibs. per sq. ft. -=7.64 cu. ft. concrete per sq. ft. 
 
 Preliminary design by resultant polygons. The methods outlined in Art. 71 are 
 applied here and the design is shown in Fig. 79s. 
 
 Taking dimensions above given and assuming that the stress for dead load plus half 
 uniform live load will be about 0.4 of the allowable, then Eq. (71 A) gives 
 
 . 
 
 ho 
 
 and Eqs. (71s) give D =6.00 ft. which value is accepted for the crown thickness of the 
 arch. 
 
 Eq. (71c) gives 81.25 ft. for the radius of the intrados at the crown, and the quarter 
 point for the parabola passing through the same three points at crown and springing, 
 is from Eq. (7 ID) 
 
 y J* = 12.5 feet and x -? =37.5 feet. 
 4 4 
 
 The point on the parabola is plotted at s, Fig. 79s, and the point s' is on the circle 
 drawn with radius 81.25 ft. The point s", midway between s and s', is chosen and a 
 three-center curve is drawn through s" and the given crown and springing points. 
 
 The thickness of the arch at the point s" is found to be 6.9 ft., using Eq. (7lE). 
 
 The remainder of the ring is drawn according to good judgment and the thickness 
 at the springing was taken at 10 ft. 
 
 The arch ring is now divided up into an even number of sections as described in 
 Art. 73, so that the same subdivisions may be used throughout the final analysis. The 
 dead loads are then computed and tabulated together with the live loads as required 
 for the construction of the resultant polygons. All this is given in Table 79*. 
 
328 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XV 
 
 m 
 
 Oi 
 
 t> 
 
 d 
 
 h-l 
 fe 
 
ART. 79 
 
 FIXED MASONRY ARCHES 
 
 329 
 
 TABLE 79A 
 DEAD LOADS AND EQUIVALENT UNIFORM LIVE LOADS 
 
 Ordinates 
 V 
 from 
 .4 
 
 ft. 
 
 Length 
 on 
 Axis, 
 
 Jw 
 
 ft. 
 
 D 
 
 ft. 
 
 DEAD LOADS IN CUBIC FEET. 
 
 Live 
 Load, 
 P 
 
 cu.ft. 
 
 tf 
 
 cu.ft. 
 
 Q+P 
 
 cu.ft. 
 
 Arch 
 Sections, 
 sq.ft. 
 
 Piers. 
 
 sq.ft. 
 
 Road- 
 way. 
 
 sq.ft. 
 
 Total Q. 
 cu.ft. 
 
 0- 0.0 
 1-5.9 
 2-12 A 
 3-18.9 
 4-25.4 
 5-31.9 
 6-38.4 
 7^4.9 
 8-51.4 
 9-57.9 
 10-64.4 
 11-71.9 
 12 79 4 
 
 
 10.00 
 9.30 
 8.60 
 8.00 
 7.60 
 7.20 
 6.90 
 6.64 
 6.46 
 6.30 
 6.20 
 6.10 
 6.00 
 
 193.44 
 145.40 
 115.47 
 95.82 
 87.33 
 91.62 
 
 108.00 
 64.80 
 42.00 
 21.00 
 14.80 
 6.00 
 
 65.91 
 65.91 
 65.91 
 65.91 
 65.91 
 76.05 
 Totals 
 
 <3 1 = 367.35 
 #3 = 276.11 
 Q 5 = 223.38 
 Q 7 = 182.73 
 #, = 168.04 
 Q u = 173.67 
 
 99.32 
 99.32 
 99.32 
 99.32 
 99.32 
 114.60 
 
 ^ = 417.01 
 P 3 = 325.77 
 P 5 = 273.04 
 P 7 =232.39 
 P 9 = 217.70 
 P u = 230.97 
 
 P' 23 = 466.67 
 P' n = 375.43 
 P' 1B = 322.70 
 P' 17 = 282.05 
 P'is = 267.36 
 P' 13 = 288.27 
 
 20.80 
 
 18.10 
 
 16.00 
 
 14.40 
 
 13.84 
 
 15.02 
 
 
 
 1391.28 
 
 611.20 
 
 1696.88 
 
 2002.48 
 
 Above loads are all for an arch ring one foot thick. 
 
 The resultant polygon for the symmetric loading Q+^P. given in Table 79A as loads PI 
 to PI i, may now be drawn so as to pass through the center of the arch ring at the joint 
 a and at the crown n. This polygon is shown in Fig. 79B, as a fine line almost coincident 
 with the center line, and the shape of the arch ring is thus found to be acceptable. Had 
 this coincidence not been so close then the shape of the arch would require modification. 
 
 For the above case of symmetric shape and loading, the resultant polygon for the 
 half span only is required, which is best constructed by computing the horizontal thrust 
 instead of finding it graphically. This is done by computing the moments of the forces 
 PI to PII about n and dividing the sum of these moments by # = 2P to obtain the lever 
 arm x of R. The vertical component of the reaction at A is equal to R and hence H is 
 found by taking moments about n of all the external forces to the left of n. This gives 
 
 H=Q 
 
 h 
 
 1696.88(79.4-45.83) 
 50.4 
 
 = 1130.2 cu.ft. 
 
 The force polygon is thus easily constructed and the resultant polygon is then drawn 
 through the two section centers and the crown. 
 
 The resultant polygon for maximum one-sided loading, is now constructed. To do this 
 the load divide for the quarter point m is found as indicated in Fig. 79s by drawing lines 
 am and bn which intersect in the required point i. 
 
 The full load p =7.64 cu.ft. per foot is allowed to cover the span from the right up to 
 the point i and the left portion is acted on by dead loads only. The loads Q\ to QQ and 
 P'n to P'zz, from Table 79A, are now united into a force polygon with assumed pole 
 
330 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. XV 
 
 distance E'\ and an equilibrium polygon a'cb' is drawn for the purpose of finding the 
 reactions R\ and R'-z and the correct pole 3 . 
 
 It is then required to pass a resultant polygon through the three points a, n and 
 6, where a, and b are middle third points and n is D/8 above the axial line. This is done 
 as follows: project the points a, n and b down vertically on the equilibrium polygon, 
 giving the points a', c and b' respectively. Draw a'c and b'c. Then in the force polygon 
 make 0> 2 c' \\ a'c and O^cf' \\cb' thus establishing the points c' and c". Now draw the lines 
 c'0 3 || an, and c"0 s \\bn giving the intersection 3 as the required pole and the line U^T [| ~ah 
 will be the true pole distance. An equilibrium polygon drawn through the point a, using 
 the force polygon OsMN, will then pass through the other two points n and b. The loads are 
 divided at T into the reactions .4 = If rand B =NT&s per Eqs. (72A) &ndO s M=R'i and 
 3 N*R>3. 
 
 By the construction given in Fig. 79s, the correct pole H' =1220 cu.ft., and the 
 required resultant polygon anb are found. The polygon intersects the arch center line in 
 three points and remains just inside the middle third of the arch ring at all critical sections. 
 
 It is seen by inspection that a more favorable resultant polygon could not be drawn, 
 hence the solution is considered completed provided the stresses resulting from this load- 
 ing do not exceed the allowable limits. 
 
 The stresses are now found from Eq. (69A) which, for a rectangular section of depth 
 D and width unity, becomes 
 
 B 
 
 D 
 
 (79B) 
 
 where v is the lever arm of the normal N about the axial point of the section. 
 
 TABLE 79B 
 STRESSES FOUND FROM PRELIMINARY DESIGN 
 
 Section. 
 
 V 
 ft. 
 
 D 
 
 ft. 
 
 UNSYMMETKIC LOAD Q + P. 
 
 LOADS Q + P OVER WHOLE SPAN. 
 
 N 
 
 cu.ft. 
 
 V 
 
 ft. 
 
 fe 
 cu.ft. 
 
 ft 
 
 cu.ft. 
 
 N 
 cu.ft. 
 
 V 
 
 ft. 
 
 fe 
 cu.ft. 
 
 ft 
 
 cu..t. 
 
 a 
 
 0.0 
 
 10.00 
 
 1980 
 
 1.67 
 
 396.0 
 
 0.0 
 
 2372 
 
 
 
 237.2 
 
 237.2 
 
 m 
 
 41.3 
 
 6.77 
 
 1390 
 
 1.13 
 
 0.0 
 
 410.7 
 
 
 
 
 n 
 
 79.4 
 
 6.00 
 
 1220 
 
 0.75 
 
 406.6 
 
 0.0 
 
 1266 
 
 
 
 422. 
 
 422.0 
 
 - 
 
 94.4 
 
 6.20 
 
 1234 
 
 1.03 
 
 398.0 
 
 0.0 
 
 
 
 
 
 b 
 
 158.8 
 
 10.00 
 
 2290 
 
 1.67 
 
 0.0 
 
 458. 
 
 2372 
 
 
 
 237.2 
 
 237.2 
 
 It is thus seen that the stresses are all well on the safe side and the preliminary 
 design is, therefore, acceptable. Had it been impossible to construct a resultant polygon 
 for the unsymmetric loading, which would remain within the middle third of the arch 
 ring, then the thickness of the ring would have to be altered accordingly. Also, if the 
 unit stresses had exceeded the allowable limits the thickness of the arch would have to 
 be increased. 
 
ART. 79 FIXED MASONRY ARCHES 331 
 
 The structure thus designed and found adequate to carry the required loads safely 
 is now subjected to a rigid analysis by applying the foregoing formulae, derived from the 
 theory of elasticity. 
 
 (b) Analysis of the preliminary design by the theory of elasticity treating the 
 structure as symmetric. The computations are all given in tabular form, thus 
 illustrating more clearly the method of conducting such analyses and furnishing a record 
 of the numerical work which is well suited for reference and checking. 
 
 The computations for the elastic loads W, and the location of the (x, y) axes are carried 
 out in Table 79c, following closely the method described in Art. 73. 
 
 The original subdivisions of the arch ring are still retained and each space is again 
 divided into two, making 24 sections in the whole ring between a and b. 
 
 The programme carried out in the table is as follows: Compute the loads W a 
 from Eq. (73A) then determine z ' and transform the original (v, z) axes to the (x, y) axes 
 according to Eqs. (73j). Finally compute the Wb and W c loads by Eqs. (73A), and the 
 pole distances H a , H b and H c from Eqs. (74A). This then furnishes all the data for con- 
 structing the X a , X b and X c influence lines. 
 
 The y axis is known to be the axis of symmetry and must, therefore, pass through 
 the crown joint n or 12. The location of the x axis is found by computing the ordinate 
 2,,' of the origin from some assumed v axis which was taken 26.4 ft. below the springing 
 points and 24, Fig. 79u. 
 
 In Table 79c, the dimensions D, z and sec <j> are tabulated from the drawing and from 
 these the values I/I, w a and zw a are derived. The abscissa; x and increments Ax are also 
 taken from the drawing, and the other values result from performing the operations 
 indicated in the headings. It should be noted that the constant modulus E was not 
 considered in computing the elastic loads W so that all these loads are E times actual. 
 
 The secants are computed from 
 
 R 115.8 , 
 
 sec = 7^-^ = ^r^~ f or P om ts between a and ra, 
 
 2+OO.O 2 +00.;) 
 
 sec & 7= '^-^ for points between m and n. 
 2+0.9 
 
 The sums 'w a = W a , 2zw a =zTF , etc., are computed by Simpson's rule for distance 
 increments Ax, according to Eqs. (73 E). 
 
 The ordinate z ' , of the center of gravity or origin of coordinates, is then found to 
 be 22TF a /2PF a =64.72 ft. Also, the sum 2zzTF a =0 when extended over the whole 
 arch, hence tan /3=0, making y=zz '. 
 
 The functions xw a , x 2 w a , yw a and y 2 w a are now figured and from these the sums 
 2.ru? =Wb, H>x 2 w n =xWb, Hyw n = W c and ^y 2 w a =yW c are again derived by applying 
 Simpsons's rule for the distance increments Ax. It should be noted that for the whole 
 arch ring 2 JF& =0 and 2JF =0, serving as a check on the computations. If these sums d6 
 not become zero, then errors have occurred which must be rectified. Small discrepancies 
 will always be found owing to insufficient accuracy in the data and not carrying sufficient 
 decimals. These may be adjusted by proportionate distribution. However, this must 
 
332 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. XV 
 
 GO 
 
 w 
 o 
 
 c 
 
 PH 
 C 
 
 o 
 
 b 
 
 05 *< 
 
 *~ w o 
 
 s tf 
 
 3 H 
 
 ^ H 
 N| 
 
 ^ 02 
 
 w 
 i 
 
 EH 
 
 fe 
 O 
 
 K 
 
 o 
 
 I I 
 
 I 
 
 ? 
 
 o 
 
 s 
 
 H 
 
 10 
 
 CH 
 
 
 
 O 
 
 , s > 
 
 "si >i 
 
 6- 
 
 C5COOOr(H>-H 
 
 IOOC5OOO1 I i-l i l 
 
 + 
 
 ooooooooooooo 
 
 I I I I I I + 
 
 1 + 
 
 ?O(MOC5t-HCDGOTtl'Ttl' IT-HT^T- 1 
 
 i I O5 iO CC CO LO Cl i i CO 
 
 (N(N(N(MIM(MGOGOOOOOOOXC 
 
 U5 t>- O i i 
 
 Oi-Hi-HlMIMfNINIMCOCOCOeOIN 
 OOOOOOOOOOOOO 
 
 co co co co co co 
 
 pppooooooopop 
 ooooooooooooo 
 
 -- 
 
 ooppppppppppp 
 odo'do'o'ooooo'oo 
 
 I * 
 
 N O 
 
 II -S 
 
 55 -I 
 
 M 
 
 was omil 
 
 o 
 H 
 
ART. 79 
 
 FIXED MASONRY ARCHES 
 
 333 
 
 be confined to small errors not exceeding ^ per cent. Mistakes must be corrected and not 
 distributed. 
 
 Lisa's 
 
 FIG. 79c. X Influence Lines for "Symmetric Arch. 
 
 The influence lines for X a , X 6 , and X c were drawn in Fig. 79c, using the results from 
 Table 79c, and neglecting the effect Scos (j>Jx/EF due to the axial thrust N on the pole 
 distance H c of the X c influence line. If this effect is to be considered then the value 
 * cos (f>4x/EF would be determined in the manner later indicated in Table 79M, omitting 
 
334 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. XV 
 
 the constant E, since it was also omitted in computing the values w a . This is permissible 
 because E being involved as a reciprocal factor in all elastic loads, as per Eqs. (72o), 
 may be considered as canceled in Eqs. (72ii). It must be remembered, however, that the 
 elastic loads are then actually E times too large, a point which must be considered in 
 computing temperature effects by Eqs. (7oB). 
 
 The X influence lines are the equilibrium polygons for the W loads when certain 
 pole distances are employed in the respective force polygons. It is generally convenient 
 to draw these diagrams so that the influence line ordinates may appear a certain number 
 of times actual when scaled to the scale of lengths of the drawing. 
 
 The X a influence line is the equilibrium polygon for the W a loads when the 
 pole distance is made equal to H a = HW a =6.564, as given by Table 79c. In the 
 present case the pole was made ^TF a so that the ordinates >j a of the X a influence 
 
 FOR INTRADOS KERNEL POINTS. 
 37/0 m n 
 
 B 
 
 FIG. 79o. Mo Influence Lines for Symmetric Arch. 
 
 line become twice actual to the scale of lengths. The scale of the force polygon is 
 chosen so as to give a reasonably large figure and to permit of accurate construction 
 of the equilibrium polygon, otherwise the scale of loads need not bear any specified 
 relation to the scale of ordinates. As a check on the drawing, the outer rays of this 
 influence line must intersect on the y axis, in fact this is the graphic method of locating 
 this axis. 
 
 The Xb influence line is the equilibrium polygon for the Wb loads when the pole dis- 
 tance is made equal to H b = 2xW b =2X5482.9 = 10965.8, from Table 79c. As this pole 
 is very large in comparison with the 2^ = 107.4, it is advisable to take Hb=Q.QlHxWb 
 = 109.66, making the ordinates r jb one hundred times actual to the scale of lengths. The 
 scale of the Wb loads was again chosen as a matter of convenience. The closing line 
 AB divides the influence line into a positive and a negative area. 
 
 The X c influence line is the equilibrium polygon for the W c loads when the pole 
 is made equal to # c =cos 8^yW c =2X560.9 = 1121.8. Here also, the pole is too large 
 
ART. 79 
 
 FIXED MASONRY ARCHES 
 
 335 
 
 to furnish a well shaped force polygon and hence the pole was made # C =0.02?/TF C , 
 giving influence line ordinates >j c fifty times actual to the scale of lengths. The scale 
 of W c loads is also chosen as a matter of convenience. 
 
 The M influence lines are ordinary moment influence lines drawn for the simple span 
 AB and the several kernel points of. the critical sections. Fig. 79o. The ordinates may 
 be made actual to the scale of lengths or, as in the case of the M ot line, the ordinates were 
 made five times actual. 
 
 The M; and M e influence lines for the kernel points of the critical sections are drawn 
 by plotting the coordinates computed according to Eqs. (76E). See Fig. 79E and 
 Tables 79E and 79F. 
 
 The coordinates of the kernel points i and e of the three critical sections at t, m and n 
 of the arch ring, are computed from the coordinates (x, y) of the axial points, using Eqs. 
 (76o) as given in Table 79D. 
 
 TABLE 79D 
 DATA RELATING TO THE CRITICAL SECTIONS 
 
 
 
 
 AXIAL POIXT. 
 
 
 
 KERNEL, POINTS. 
 
 
 
 
 
 D . 
 
 n 
 
 
 Point. 
 
 D 
 
 * 
 
 
 
 sin c 
 
 COS < 
 
 
 
 
 
 
 
 
 X 
 
 V 
 
 
 
 x i 
 
 Vi 
 
 x e 
 
 Ve 
 
 
 ft. 
 
 
 ft. 
 
 ft. 
 
 ft 
 
 ft. 
 
 ft, 
 
 ft, 
 
 ft. 
 
 ft. 
 
 t 
 
 9.75 
 
 56 30' 
 
 78.05 
 
 -36.19 
 
 1.36 
 
 0.89 
 
 79.4 
 
 -35.3 
 
 76.7 
 
 -37.1 
 
 m 
 
 6.90 
 
 30 30' 
 
 41.8 
 
 0.0 
 
 0.62 
 
 0.99 
 
 42.4 
 
 + 0.99 
 
 41.2 
 
 - 0.99 
 
 n 
 
 6.00 
 
 00' 
 
 0.0 
 
 + 12.08 
 
 0.00 
 
 1.00 
 
 0.0 
 
 + 13.08 
 
 0.0 
 
 + 11.08 
 
 From the coordinates of the kernel points Table 79o, and the ordinates in Tables 
 79E and 79r, taken from the X a , X b , X c and M influence lines, t) e and TH are computed 
 for each section of the span and for the critical sections t, m and n. 
 
 Thus, for the section at m, Eqs. (76B) may be written 
 
 (79c) 
 
 The computations are all indicated in the headings of the tables and no further 
 comment is necessary here. 
 
 With the use of M f and M e influence lines for the three critical sections of a symmetric 
 arch, the stresses in these sections may be found for any possible case of loading. 
 
 The resultant polygon for any particular case of simultaneous loading, is located 
 when the redundants A" a , X b and X c are known. Eqs. (78A) then give the necessary 
 dimensions from which the resultant polygon may be drawn. 
 
 Table 79o gives the computations for the resultant polygon for the case of average 
 loading Q +%P, over the entire span. This is the case for which the resultant polygon 
 should coincide with the axis and the results are comparable with those on Fig. 79s. 
 
336 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. XV 
 
 The lettered dimensions in the table are all shown in Fig. 7SA, but the resultant polygon 
 is not drawn for this case. 
 
 The influence line ordinates y a , TJ& and TJ C are taken from Fig. 79c and the sums of the 
 products of the loads and ordinates give the respective redundants X from which z ,c 
 and c 0} also A and B are readily found by Eqs. (78A). 
 
 TABLE 79s 
 MOMENT INFLUENCE LINES FOR EXTRADOS KERNEL POINTS. SYMMETRIC ARCH 
 
 Pt. 
 
 r]a 
 ft. 
 
 r/b 
 ft. 
 
 f)c 
 
 ft. 
 
 SECTION t 
 J2=>?o [i}a+76.7i}6 37.l7j c ] 
 
 SECTION m. 
 
 T)m = ijo - [>?a+ 4 1 .2-fjb - 0.99r; c ] 
 
 SECTION n. 
 
 T)n=T)o [ija+ll.l'jc] 
 
 90 
 
 76.7 T) b 
 
 37.lTjc 
 
 fit 
 
 ? 
 
 41.27JC 
 
 0.99i? c 
 
 Tim 
 
 Vo 
 
 H.lTJc 
 
 9 
 
 
 
 0.0 
 
 0.0 
 
 0.0 
 
 
 
 
 
 0.0 
 
 0.0 
 
 0.0 
 
 0.0 
 
 0.0 
 
 0.0 
 
 0.0 
 
 1 
 
 2.9 
 
 0.031 
 
 0.014 
 
 2.61 
 
 2.38 
 
 - 0.52 
 
 -2.15 
 
 4.5 
 
 1.28 
 
 -0.01 
 
 0.33 
 
 2.9 
 
 0.14 
 
 -0.14 
 
 2 
 
 6.0 
 
 0.062 
 
 0.056 
 
 
 
 
 
 9.4 
 
 2.55 
 
 -0.05 
 
 0.90 
 
 6.3 
 
 0.62 
 
 -0.32 
 
 3 
 
 8.8 
 
 0.086 
 
 0.120 
 
 2.39 
 
 6.59 
 
 - 4.45 
 
 -8.55 
 
 14.2 
 
 3.54 
 
 -0.12 
 
 1.98 
 
 9.5 
 
 1.33 
 
 -0.63 
 
 4 
 
 11.4 
 
 0.102 
 
 0.200 
 
 
 
 
 
 19.1 
 
 4.20 
 
 -0.20 
 
 3.70 
 
 12.8 
 
 2.22 
 
 -0.82 
 
 5 
 
 13.9 
 
 0.110 
 
 0.296 
 
 2 17 
 
 8.44 
 
 -10.98 
 
 -9.19 
 
 24.0 
 
 4.53 
 
 -0.29 
 
 5.86 
 
 16.0 
 
 3.29 
 
 -1.19 
 
 6 
 
 16.0 
 
 0.111 
 
 0.396 
 
 
 
 
 
 29.0 
 
 4.57 
 
 -0.39 
 
 8.82 
 
 19.3 
 
 4.40 
 
 -1.10 
 
 7 
 
 17.8 
 
 0.105 
 
 0.496 
 
 1.95 
 
 8.05 
 
 -18.40 
 
 -5.50 
 
 27.5 
 
 4.33 
 
 -0.49 
 
 5.86 
 
 22.6 
 
 5.51 
 
 -0.71 
 
 8 
 
 19.4 
 
 0.095 
 
 0.588 
 
 
 
 
 
 25.9 
 
 3.91 
 
 -0.58 
 
 3.17 
 
 25.9 
 
 6.53 
 
 -0.03 
 
 9 
 
 20.7 
 
 0.079 
 
 0.672 
 
 1.73 
 
 6.06 
 
 -24.93 
 
 -0.10 
 
 24.2 
 
 3.25 
 
 -0.67 
 
 0.92 
 
 29.1 
 
 7.46 
 
 0.94 
 
 10 
 
 21.7 
 
 0.058 
 
 0.734 
 
 
 
 
 
 22.7 
 
 2.39 
 
 -0.73 
 
 -0.66 
 
 32.3 
 
 8.15 
 
 2.45 
 
 11 
 
 22.4 
 
 0.030 
 
 0.782 
 
 1.48 
 
 2.30 
 
 -29.01 
 
 + 5.69 
 
 21.0 
 
 1.24 
 
 -0.77 
 
 -1.87 
 
 36.0 
 
 8.68 
 
 4.92 
 
 12 
 
 22.6 
 
 0.000 
 
 0.800 
 
 
 
 
 
 19.2 
 
 0.00 
 
 -0.79 
 
 -2.70 
 
 40.0 
 
 8.88 
 
 8.52 
 
 13 
 
 22.4 
 
 -0.030 
 
 782 
 
 1 23 
 
 -2.30 
 
 -29.01 
 
 10.14 
 
 17.4 
 
 -1.24 
 
 -0.77 
 
 -2.99 
 
 36.0 
 
 8.68 
 
 4.92 
 
 14 
 
 21.7 
 
 -0.058 
 
 734 
 
 
 
 
 
 15.6 
 
 -2.39 
 
 -0.73 
 
 -2.98 
 
 32.3 
 
 8.15 
 
 2.45 
 
 15 
 
 20.7 
 
 -0.079 
 
 0.672 
 
 0.98 
 
 -6.06 
 
 -24.93 
 
 11.27 
 
 14.0 
 
 -3.25 
 
 -0.67 
 
 -2.78 
 
 29.1 
 
 7.46 
 
 0.94 
 
 16 
 
 19.4 
 
 -0.095 
 
 0.588 
 
 
 
 
 
 12.4 
 
 -3.91 
 
 -0.58 
 
 -2.51 
 
 25.9 
 
 6.53 
 
 -0.03 
 
 17 
 
 17.8 
 
 -0.105 
 
 0.496 
 
 0.77 
 
 -8.05 
 
 -18.40 
 
 9.42 
 
 10.8 
 
 -4.33 
 
 -0.49 
 
 -2.18 
 
 22.6 
 
 5.51 
 
 -0.71 
 
 18 
 
 16.0 
 
 -0.111 
 
 0.396 
 
 
 
 
 
 9.3 
 
 -4.57 
 
 -0.39 
 
 -1.74 
 
 19.3 
 
 4.40 
 
 -1.10 
 
 19 
 
 13.9 
 
 -0.110 
 
 0.296 
 
 0.55 
 
 -8.44 
 
 -10.98 
 
 6.07 
 
 7.7 
 
 -4.53 
 
 -0.29 
 
 -1.38 
 
 16.0 
 
 3.29 
 
 -1.19 
 
 20 
 
 11.4 
 
 -0.102 
 
 0.200 
 
 
 
 
 
 6.1 
 
 -4.20 
 
 -0.20 
 
 -0.90 
 
 12.8 
 
 2.22 
 
 -0.82 
 
 21 
 
 8.8 
 
 -0.086 
 
 0.120 
 
 0.32 
 
 -6.59 
 
 - 4.45 
 
 2.56 
 
 4.6 
 
 -3.54 
 
 -0.12 
 
 -0.54 
 
 9.5 
 
 1.33 
 
 -0.63 
 
 22 
 
 6.0 
 
 -0.062 
 
 0.056 
 
 
 
 
 
 3.1 
 
 -2.55 
 
 -0.05 
 
 -0.30 
 
 6.3 
 
 0.62 
 
 -0.32 
 
 23 
 
 2.9 
 
 -0.031 
 
 0.014 
 
 0.10 
 
 -2.38 
 
 - 0.52 
 
 0.10 
 
 1.5 
 
 -1.28 
 
 -0.01 
 
 -0.11 
 
 2.9 
 
 0.14 
 
 -0.14 
 
 24 
 
 0.0 
 
 0.0 
 
 0.0 
 
 
 
 
 
 0.0 
 
 0.0 
 
 0.0 
 
 0.0 
 
 0.0 
 
 0.0 
 
 0.0 
 
 All ordinates ij in above table are expressed in feet. 
 
 Table 79n presents the case of unsymmetric loading shown in Fig. 79s. The results 
 ere plotted in Fig. 79r, and the resultant polygon is finally drawn in accordance with the 
 method given in Art. 78. 
 
 The arch being symmetric /3=0 and Zi =2 hence with the values of X a , Xb and X c 
 from Table 79n, the resulting data in that table were found. 
 
 In Fig. 79r the arch ring and the (x, y] axes are given to locate the points a and b 
 on the verticals through the end supports and then to draw the resultant polygon. 
 
ART. 79 
 
 FIXED MASONRY ARCHES 
 
 337 
 
 /\ 
 
 ' I 2 
 
 \ 
 
 \ 
 
 \ 
 
 \ 
 
 AT SECTION t. 
 1-158'.8 
 
 8 \> 10 II U l 14 IS 16 17 18 >? 30 21 22 
 
 B 
 
 LENGTHS. 
 
 Ijllll.M.I I I __l___L_ 1| 
 
 10 IO 20 30 40 SOFT. 
 OROI NATES. 
 
 I// 
 
 itifrados kernel points. 
 extrados kernel points. 
 
 FIG. 79E. M Influence Lines for the Kernel Points. Symmetric Arch. 
 
338 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES. CHAP. XV 
 
 TABLE 79F 
 MOMENT INFLUENCE LINES FOR INTRADOS KERNEL POINTS. SYMMETRIC ARCH 
 
 Pt. 
 
 Tja 
 ft. 
 
 T,b 
 ft, 
 
 1)c 
 
 ft. 
 
 SECTION t. 
 rj = ^,0 [rja + 79.4,6 35. 3, c ] 
 
 SECTION m. 
 ,m = ,o- [,3+42.4,6+0.99,0] 
 
 1 
 SECTION n 
 
 T)n=T)o [,<z + 13. Irjc] 
 
 Vo 
 
 79.4,6 
 
 35.3,6 
 
 yt 
 
 rjo 
 
 42.4,6 
 
 0.99, c 
 
 ,m 
 
 9 
 
 13.1,0 
 
 9 
 
 
 
 0.0 
 
 0.0 
 
 0.0 
 
 
 
 
 
 0.0 
 
 0.0 
 
 0.0 
 
 0.0 
 
 0.0 
 
 0.0 
 
 0.0 
 
 1 
 
 2.9 
 
 0.031 
 
 0.014 
 
 0.0 
 
 2.46 
 
 - 0.49 
 
 - 4.87 
 
 4.5 
 
 1.31 
 
 0.01 
 
 0.28 
 
 2.9 
 
 0.18 
 
 -0.18 
 
 2 
 
 6.0 
 
 0.062 
 
 0.056 
 
 
 
 
 
 9.4 
 
 2.63 
 
 0.05 
 
 0.72 
 
 6.3 
 
 0.73 
 
 -0.43 
 
 3 
 
 8.8 
 
 0.086 
 
 0.120 
 
 0.0 
 
 6.83 
 
 - 4.24 
 
 -11.39 
 
 14.3 
 
 3.65 
 
 0.12 
 
 1.73 
 
 9.5 
 
 1.57 
 
 -0.87 
 
 4 
 
 11.4 
 
 0.102 
 
 0.200 
 
 
 
 
 
 19.3 
 
 4.32 
 
 0.20 
 
 3.38 
 
 12.8 
 
 2.62 
 
 -1.22 
 
 5 
 
 13.9 
 
 0.110 
 
 296 
 
 0.0 
 
 8.73 
 
 -10.45 
 
 -12.18 
 
 24.2 
 
 4.66 
 
 0.29 
 
 O . oC 
 
 16.0 
 
 3.88 
 
 -1.78 
 
 6 
 
 16.0 
 
 0.111 
 
 0.396 
 
 
 
 
 
 28.1 
 
 4.71 
 
 0.39 
 
 7.00 
 
 19.3 
 
 5.19 
 
 -1.89 
 
 7 
 
 17.8 
 
 0.105 
 
 0.496 
 
 o.c 
 
 8.34 
 
 -17.51 
 
 - 8.63 
 
 26.6 
 
 4.45 
 
 0.49 
 
 3.86 
 
 22.6 
 
 6.50 
 
 -1.70 
 
 8 
 
 19.4 
 
 0.095 
 
 0.588 
 
 
 
 
 
 25.0 
 
 4.03 
 
 0.58 
 
 0.99 
 
 25.9 
 
 7.70 
 
 -1.20 
 
 9 
 
 20.7 
 
 0.079 
 
 0.672 
 
 0.0 
 
 6.27 
 
 -23.72 
 
 - 3.25 
 
 23.5 
 
 3.35 
 
 0.67 
 
 -1.22 
 
 29.1 
 
 8.80 
 
 -0.21 
 
 10 
 
 21.7 
 
 0.058 
 
 734 
 
 
 
 
 
 22.0 
 
 2.46 
 
 0.73 
 
 -2.89 
 
 32.3 
 
 9.61 
 
 0.99 
 
 11 
 
 22.4 
 
 0.030 
 
 0.782 
 
 0.0 
 
 2.38 
 
 -27.60 
 
 + 2.82 
 
 20.3 
 
 1.27 
 
 0.77 
 
 -4.14 
 
 36.0 
 
 10.24 
 
 3.36 
 
 12 
 
 22.6 
 
 0.000 
 
 0.800 
 
 
 
 
 
 18.6 
 
 0.00 
 
 0.79 
 
 -4.79 
 
 40.0 
 
 10.48 
 
 6.92 
 
 13 
 
 22.4 
 
 -0.030 
 
 0.782 
 
 0.0 
 
 -2.38 
 
 -27.60 
 
 7.58 
 
 16.8 
 
 -1.27 
 
 0.77 
 
 -5.10 
 
 36.0 
 
 10.24' 
 
 3.36 
 
 14 
 
 21.7 
 
 -0.058 
 
 0.734 
 
 
 
 
 
 15.1 
 
 -2.46 
 
 0.73 
 
 -4.87 
 
 32.3 
 
 9.61 
 
 0.99 
 
 15 
 
 20.7 
 
 -0.079 
 
 0.672 
 
 0.0 
 
 -6.27 
 
 -23.72 
 
 9.29 
 
 13.6 
 
 -3.35 
 
 0.67 
 
 -4.42 
 
 29.1 
 
 8.80 
 
 -0.21 
 
 16 
 
 19.4 
 
 -0.095 
 
 0.588 
 
 
 
 
 
 12.0 
 
 -4.03 
 
 0.58 
 
 -3.95 
 
 25.9 
 
 7.70 
 
 -1.20 
 
 17 
 
 17.8 
 
 -0.105 
 
 0.496 
 
 0.0 
 
 -8.34 
 
 -17.51 
 
 8.05 
 
 10.6 
 
 -4.45 
 
 0.49 
 
 -3.24 
 
 22.6 
 
 6.50 
 
 -1.70 
 
 18 
 
 16.0 
 
 -0.111 
 
 0.396 
 
 
 
 
 
 9.1 
 
 -4.71 
 
 0.39 
 
 -2.58 
 
 19.3 
 
 5.19 
 
 -1.89 
 
 19 
 
 13.9 
 
 -0.110 
 
 0.296 
 
 0.0 
 
 -8.73 
 
 -10.45 
 
 5.28 
 
 7.5 
 
 -4.66 
 
 0.29 
 
 -2.03 
 
 16.0 
 
 3.88 
 
 -1.78 
 
 20 
 
 11.4 
 
 -0.102 
 
 0.200 
 
 
 
 
 
 6.0 
 
 -4.32 
 
 0.20 
 
 -1.28 
 
 12.8 
 
 2.62 
 
 -1.22 
 
 21 
 
 8.8 
 
 -0.086 
 
 0.120 
 
 0.0 
 
 -6.83 
 
 - 4.24 
 
 2.27 
 
 4.4 
 
 -3.65 
 
 0.12 
 
 -0.87 
 
 9.5 
 
 1.57 
 
 -0.87 
 
 22 
 
 6.0 
 
 -0.062 
 
 056 
 
 
 
 
 
 3.0 
 
 -2.63 
 
 0.05 
 
 -0.42 
 
 6.3 
 
 0.73 
 
 -0.43 
 
 23 
 
 2.9 
 
 -0.031 
 
 0.014 
 
 0.0 
 
 -2.46 
 
 - 0.49 
 
 0.05 
 
 1.4 
 
 -1.31 
 
 0.01 
 
 -0.20 
 
 2.9 
 
 0.18 
 
 -0.18 
 
 24 
 
 0.0 
 
 0.0 
 
 0.0 
 
 
 
 
 
 0.0 
 
 0.0 
 
 0.0 
 
 0.0 
 
 0.0 
 
 0.0 
 
 0.0 
 
 All ordinates in above table are expressed in feet. 
 
 The construction is made by laying off z down from the origin on the y axis to 
 obtain the point s. Then lay off c up from s to determine s' on the y axis. Through 
 s' draw a line parallel to the x axis and where this line intersects the vertical at A gives 
 the point a on the resultant R'i. The line as prolonged to intersect the vertical at B 
 gives the point 6 on the resultant R' 2 . Knowing A , B and H' , the resultants R'\ and 
 R*2 can now be drawn, and their intersection must fall on the resultant R, distant r from 
 the 2 axis. 
 
 The force polygon is constructed by laying off the load line MN and_then finding T, 
 which is the point of division between A and B . Through T draw 0\T\\ob and lay 
 off the force H' , thus locating the pole Oi, and giving the reactions R'i and R' 2 . The 
 equilibrium polygon drawn through a and with pole 0\ will then pass through b as a final 
 check. The resultant polygon for unsymmetric loading intersects the axial line three 
 times, as it should. 
 
ART. 79 
 
 FIXED MASONRY ARCHES 
 
 339 
 
 o 
 
 fe 
 
 
 
 OQ 
 
 w 
 
 o 
 
 OH 
 
 HM 
 O" 
 
 CO 
 C3 Q 
 
 p ^ 
 o 
 
 H >-} ! 
 
 rl 
 
 tf 
 
 H g 
 
 O 
 
 O 
 
 OQ 
 
 3 
 3 
 
 fe 
 O 
 
 HH 
 
 H 
 
 I 
 
 RESULTING DATA. 
 
 d 
 
 II 
 
 
 H 
 
 SB 
 
 +3 
 
 II 
 ^2. 
 
 W 
 
 O 4* 
 
 o -> 
 
 . 3 
 
 ^ 
 
 *- O5 
 
 ^ ic 
 
 2 4 
 II ^ c 
 I- II - 
 
 1 ^ % 
 
 ~ " 
 Jfcj **? 
 
 w II 
 06 133 * 
 2 a 
 II ^ 
 - H 
 u 
 O 
 h 
 
 o 
 
 II 
 
 05 
 -1&5 -S 
 
 11 P 
 
 <? *; 
 
 o3 **"! 
 
 1 
 
 - 1 ."' 3 
 
 s a i 1 
 
 1 -U 1 1 
 
 5 o S II 
 5 II ci OSIN 
 
 .0 t^ || 
 
 ^-N-! ^ 
 
 II II II 
 ? K ^ 
 
 1C 
 
 i 
 
 (M 
 
 11 
 
 a 
 ^ 
 + 
 
 1 
 
 II 
 ftf 
 
 
 
 * O i-H 
 
 (X) rH 00 
 
 1^- OS Tt< Tfi O5 O 
 (N (N CO CO (M (M 
 
 -1 O Tt* 
 
 XI ^H 00 
 
 05 
 
 N" 
 
 lO O5 O 
 CO 00 
 
 1C CO O O CO ^C 
 ' 1 ^f CO 00 TJI i I 
 
 3 C5 >C 
 
 * co 
 
 1C M 
 
 2 II 
 
 
 
 
 
 
 14 
 
 ft.| 
 
 CO <M CO 
 O5 O O 
 
 O O CO CO O O ( 
 * <M O5 O (M * < 
 
 O (N CO 
 
 n o O5 
 
 1C 1C ^ 
 
 + H 
 
 ^2^ 
 
 <N 00 O 
 i-H (N CO 
 
 * r^ co CD i^ Tti c 
 
 (N rH r-l <N t 
 
 1 III 
 
 D GO (N 
 O (M rH 
 
 1 1 
 
 rH rH >/? 
 + 1 * 
 
 
 co o r^ 
 
 t~- TjH Tfl Ttl * t^ 1 
 
 ^ O CO 
 
 O 
 
 ^ 
 
 O5 t- Tfl 
 O CD OS 
 <M 00 t-~ 
 ^H (N CO 
 
 co co * -*i co co - 
 
 CO O h- I> O CO . 
 i 1 1C i 1 rH 1C r- 1 1 
 
 rfi TJI tc >C -<^ * C 
 
 * r-- o 
 
 S CO O 
 ^ GO (N 
 
 O (N rH 
 
 f:^ e 
 
 ^ 
 
 2g 
 
 3 ""On IN*; 
 
 O 00 O 
 
 * O O O l> * < 
 
 D 00 O 
 
 X 
 
 ^ I * 
 
 H| ^ 3 
 
 I> i CO 
 1-1 (N t^ 
 TJH CO C^ 
 
 (N t^ I-H T-H t^ (N C 
 CO 1 CO CO -4 CO t 
 (M (N (N (N (N (N C 
 
 O >C t^ 
 
 >. C^ rH 
 
 ^ CO -* 
 
 1 
 
 co 
 
 >^o 
 
 1C iO 1C 
 
 CO cO CO 
 
 1C 1C O O >C 1C ' 
 
 co co co co co o c 
 
 C 1C 1C 
 
 O CO CD 
 
 
 a 1 |CI 
 
 O C5 O5 
 ^* ^^ ^ 
 
 C5 C5 t^ t^ O O5 C 
 TjH TfH >C 1C Tt< Tjl 
 
 35 C5 
 * Tj( TfH 
 
 
 73 -S" * i ' 
 
 TjH 1 Tt< 
 
 t> O t^ 1^ O t^ - 
 
 * rH T(H 
 
 
 go> "3 
 0,3 3 
 
 i^ co eo 
 
 CO t^ <M 
 
 (M 00 CO CO 00 (M C 
 
 oo co i-~ t^ cc oo c 
 
 CC t^ 
 ^ t>- CO 
 
 
 
 
 
 
 
 R- 
 
 * O CO 
 
 r-H C^ 05 
 
 O ^H (M 
 
 CO (M (N (M C^J CO C 
 
 o> t^ oo oo t^ O5 c 
 
 rfn CO t^ t-- CO Th C 
 
 o o ** 
 
 R (M rH 
 
 ^ rH O 
 
 
 
 o o o 
 
 o o o o o o c 
 
 D 
 
 
 -f^> 
 
 rH CO O 
 
 O O i i 
 
 >C C5 O O C5 1C C 
 O t^ CO CO t^ O i- 
 T-I O O O O I-H ' 
 
 D CO rH 
 
 H x co 
 
 H O O 
 
 
 
 o o o 
 
 o o o o o o c: 
 
 1 1 I 
 
 D 
 
 1 1 1 
 
 
 
 O 00 O5 
 
 00 t- rH TfH t>- 00 C 
 
 s oo as 
 
 
 C- 
 
 (M QO CO 
 
 t^ O (M iM O b- C 
 
 r-4 <N (N C^l <N i-H r 
 
 5 00 (M 
 
 H 
 
 
 C 
 
 5 
 
 i i co ic 
 
 t>. os i i co >c t>- c 
 
 i 1 I 1 rH T 1 T- 
 
 5 rH CO 
 H IM N 
 
 
340 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. XV 
 
 B 
 
 C5 
 
 - 
 
 II 
 
 
 14 
 
 gO 1 
 
 O ^J 
 
 ti 
 
 -^ co 
 CO 
 
 2 8 
 
 co cs 
 
 00 ^H 
 
 + 
 
 o 
 
 n . <H 
 
 oo * 
 
 CO I-H 
 (N 
 
 co o 
 
 a i 
 
 R 
 
 ft! 
 
 03 
 
 05 -3 
 
 CO"3<N(NOOCOCOOOcOiCTt< OS 
 
 OOCOCON.OOCOOO3OOOOO CM 
 
 COi-HCMOtNCNiOr i (iCi lie 
 I-H C^ I-H O1 I s - t* CO OS I-H OS 1C CO 
 
 00 OS CO 1C CD C^ CO C^l -H C^ 
 <M <N (M CO * >C> !> O 
 CO 
 
 Oi O^ Oi 0i O5 O^ O5 O^ Ol O5 O^ 03 
 iOGO^-*Tfl^-^-tCDOCOcDO5C<l 
 
 ^* CO CO C'l ^^ *-O ^O CO t^* C^ ^O CO 
 
 rH rH r-H CO O^ ^^ ^^ CO 00 *O O *O 
 
 COcOOi-HC s 1C^I^-COO5^ (M 
 
 O CO 
 
 o^ ^^ t^ oo r^ ^o ^o c*j ^H co oo t^ oo co 
 
 O I-H 
 
 (M i-H iH I (N (M CO CO 
 
 ' ' I I I I + I 
 
 iCC5iC(Nt-t--t--COC5iOCOCO 
 CD<NOiCr>.iCiCCOi IOOO1C 
 
 i-H C^ CO CO CO CO CD 1C 1C "^ CO i-H 1C 
 
 CD CO C^l 00 OO OO I s * C3 C^l 1C CD 00 ^^ 
 
 cD(^-CMOOCOOOOOCDOO<Mt^CO O 
 
 CO<N<Ni-Hi-H<N(M(M(MCOCOT}< 
 
 OOOOOCOCOCOCOCOCOCO 
 CO CO CO CO CO Tt^ ^^ OS OS OS OS OS 
 
 T"H I-H os os os os os 
 
 i-H Jt 
 
 t^ co co' (N oo co co oo' CM' co' CD f^ 
 
 CD t^- C^ OO CO t^- l> CO 00 CN t^ CO 
 
 CO<M<Ni ii ii Ii I i I r- i<N<MCO 
 
 TfOCOCDtNCMC^CJCDCDOTti 
 i-H(NOsOsr^OOOOt^O3OS(Mi-H 
 
 Oi itNi^cor tco^cMi-HO 
 O O O O O O O O O O O O 
 
 i (CDOiCOSOOOSiCOcOi-H 
 
 coooi-HOt^cocot^-Oi-Hooco 
 
 OOl-Hl-HOOOOl-Hr-HOO 
 
 oooooocdooooo 
 
 I I I I I I 
 
 OS OO OS OO t^- ^ "^ l>- 00 OS OO OS 
 <No6cOt^O<N<NOf~COOO<N 
 
 1-Hl-H(N<N(N(Ml-Hl-H 
 rl CO 1C I> OS i I CO 1C f>- OS T-l CO 
 
ART. 79 
 
 FIXED MASONRY ARCHES 
 
 341 
 
342 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAP. XV 
 
 The moments M e and M, and the resulting stresses on any critical section, and for any 
 case of loading, may be found from the moment influence lines Fig. 79E, drawn for the 
 kernel points of the critical sections. 
 
 The stresses for the two cases of loading, Q+^P and the one-sided load Q+P, are 
 computed in Tables 79i and 79J, using the influence line ordinates from Tables 79E and 
 79r, or from the diagrams Fig. 79E. At the bottom of the tables the values R, N, v, i g 
 and fi are computed for each section, using Eqs. (76c) . 
 
 The sections t and t', at the abutments, are so located that the intrados kernel points 
 i fall on the verticals through the end axial points, thus avoiding negative TJ O values in 
 Eqs. (79c), and other difficulties affecting the X influence lines. 
 
 It is seen from Table 79i, that the actual resultant polygon for the- case of average 
 loading really coincides very closely with the arch center line as designed. The offsets 
 v being + and show that the polygon intersects the center line four times, as it should. 
 
 MOMENTS M e AND Mi FOR THE CRITICAL SECTIONS, LOADS Q+- 
 
 SYMMETRIC ARCH 
 
 
 Loads 
 
 SECTION I. 
 
 SECTION m. 
 
 SECTION" n. 
 
 Pt. 
 
 >S 
 
 1e 
 
 ,. 
 
 M e 
 
 Mi 
 
 9 
 
 ti 
 
 Me 
 
 Mi 
 
 1* 
 
 T) 
 
 Me 
 
 Mi 
 
 
 cu.ft. 
 
 ft. 
 
 ft. 
 
 
 
 ft. 
 
 ft. 
 
 
 
 ft. 
 
 ft. 
 
 
 
 1 
 
 417.0 
 
 -2.15 
 
 - 4.87 
 
 - 896.6 
 
 -2030.8 
 
 0.33 
 
 0.28 
 
 137.6 
 
 116.8 
 
 -0.14 
 
 '-0.18 
 
 - 58.4 
 
 75.1 
 
 3 
 
 325.8 
 
 -8.55 
 
 -11.39 
 
 -2785.6 
 
 .-3710.9 
 
 1.98 
 
 1.73 
 
 645.1 
 
 563.6 
 
 -0.63 
 
 -0.87 
 
 - 205.3 
 
 - 283.4 
 
 5 
 
 273.0 
 
 -9.19 
 
 -12.18 
 
 -2508.9 
 
 -3325.1 
 
 5.86 
 
 5.35 
 
 1599.8 
 
 1460.6 
 
 -1.19 
 
 -1.78 
 
 - 324.9 
 
 - 485.9 
 
 7 
 
 232.4 
 
 -5.50 
 
 - .63 
 
 -1278.2 
 
 -2005.6 
 
 5.86 
 
 3.86 
 
 1366.9 
 
 897.1 
 
 -0.71 
 
 -1.70 
 
 - 165.0 
 
 - 395.1 
 
 9 
 
 217.7 
 
 -0.10 
 
 - 3.25 
 
 21.8 
 
 - 707.5 
 
 0.92 
 
 -1.22 
 
 "200.3 
 
 - 265.6 
 
 4-0.94 
 
 -0.21 
 
 4- 204.6 
 
 - 45.7 
 
 11 
 
 231.0 
 
 4-5.69 
 
 + 2.82 
 
 4-1314.4 
 
 4- 651.4 
 
 -1.87 
 
 -4.14 
 
 - 432.0 
 
 - 956.3 
 
 4.92 
 
 4-3.36 
 
 1136.5 
 
 + 776.2 
 
 13 
 
 231.0 
 
 10.14 
 
 7.58 
 
 2342.3 
 
 1751.0 
 
 -2.99 
 
 -5.10 
 
 -690.7 
 
 -1178.1 
 
 4.92 
 
 4-3.36 
 
 1136.5 
 
 4- 776.2 
 
 15 
 
 217.7 
 
 11.27 
 
 9.29 
 
 2453.5 
 
 2022.4 
 
 -2.78 
 
 -4.42 
 
 - 605.2 
 
 - 962.2 
 
 4-0.94 
 
 ^0.21 
 
 + 204.6 
 
 45.7 
 
 17 
 
 232.4 
 
 9.42 
 
 8.05 
 
 2189.2 
 
 1870.8 
 
 -2.18 
 
 -3.24 
 
 - 506.6 
 
 - 753.0 
 
 -0.71 
 
 -1.70 
 
 - 165.0 
 
 - 395.1 
 
 19 
 
 273.0 
 
 6.07 
 
 5.28 
 
 1657 . 1 
 
 1441.4 
 
 -1.38 
 
 -2.03 
 
 - 376.7 
 
 - 554.2 
 
 -1.19 
 
 -1.78 
 
 - 324.9 
 
 - 485.9 
 
 21 
 
 325.8 
 
 2.56 
 
 2.27 
 
 834.0 
 
 739.6 
 
 -0.54 
 
 -0.87 
 
 - 175.9 
 
 - 283.4 
 
 -0.63 
 
 -0.87 
 
 - 205.3 
 
 - 283.4 
 
 23 
 
 417.0 
 
 0.10 
 
 0.05 
 
 41.7 
 
 20.9 
 
 -0.11 
 
 -0.20 
 
 - 45.9 
 
 83.4 
 
 -0.14 
 
 -0.18 
 
 - 58.4 
 
 75.1 
 
 R = 
 
 3393.5 
 
 D= 
 
 9'. 75 
 
 4-3341.1 
 
 -3282.4 
 
 D= 
 
 6'. 90 
 
 4-1116.7 
 
 -1998.1 
 
 Z> = 
 
 6'. 00 
 
 4-1175.0 
 
 -1018.0 
 
 
 
 
 N= 
 
 2037.9 
 
 
 
 N 
 
 1354.3 
 
 
 
 JV = 
 
 1096.5 
 
 
 
 
 
 v= 
 
 4-0'. 014 
 
 
 
 v= 
 
 -0'.325 
 
 
 
 v = 
 
 4-0'. 071 
 
 
 
 
 
 /= 
 
 - 210.9 
 
 - 207.2 
 
 
 f= 
 
 - 140.8 
 
 - 251.8 
 
 
 J~ 
 
 - 195.9 
 
 - 169.7 
 
 Eqs. (76c), 
 
 f O'M e 
 
 /e= -^ = 
 
 , 
 * 
 
 v 1S positive when measured from the axis 
 
 2N D* ' * W ' 
 
 toward the extrados. 
 
 All loads in above table are expressed in cubic feet of masonry and the stresses are cubic feet per square foot. Cubic 
 feet per square foot X 0.972 = pounds per square inch. 
 
 The stresses in Table 79j are those due to the case of loading shown in Fig. 79r, and 
 hence the offsets v are those which the resultant polygon in that figure actually present. 
 The smallness of the scale, however, prevents a very close agreement. The resultant 
 polygon for the unsymmetric loading is seen to intersect the center line three times as 
 it should, since two values v are negative and three positive. 
 
ART. 79 
 
 FIXED MASONRY ARCHES 
 
 343 
 
 TABLE 79j 
 
 MOMENTS Me AND M FOR CRITICAL SECTIONS. TOTAL DEAD AND ONE-SIDED LIVE 
 
 SYMMETRIC ARCH 
 
 
 
 SECTION I. 
 
 SECTION m. 
 
 SECTION n. 
 
 Ft. 
 
 Loads 
 
 
 
 
 
 Q and P', 
 
 
 
 
 
 
 
 
 
 
 
 i 
 
 
 
 rje 
 
 fji 
 
 M e 
 
 Mi 
 
 Tje 
 
 T)i 
 
 M e 
 
 M{ 
 
 fje 
 
 rji 
 
 M e \ Mi 
 
 
 cu.ft. 
 
 ft. 
 
 ft. 
 
 
 
 ft. 
 
 ft. 
 
 
 
 ft. 
 
 ft. 
 
 
 
 1 
 
 Qi = 367.4 
 
 - 2.15 
 
 - 4.87 
 
 - 789.9 
 
 -1789.2 
 
 0.33 
 
 0.28 
 
 121.2 
 
 102.9 
 
 -0.14 
 
 -0.18 
 
 51.4 
 
 - 66.1 
 
 3 
 
 Q 3 = 276.1 
 
 - 8.55 
 
 -11.39 
 
 -2360.7 
 
 -3144.8 
 
 1.98 
 
 1.73 
 
 546.7 
 
 477.6 
 
 -0.63 
 
 -0.87 
 
 - 173.9 
 
 - 240 . 2 
 
 5 
 
 Ob = 223.4 
 
 - 9.19 
 
 -12.18 
 
 -2053.0 
 
 -2721.0 
 
 5.86 
 
 5.35 
 
 1309 . 1 
 
 1195.2 
 
 -1.19 
 
 -1.78 
 
 - 265.8 
 
 -397.6 
 
 7 
 
 Qi= 182.7 
 
 - 5.50 
 
 - 8.63 
 
 -1004.9 
 
 -1576.7 
 
 5.86 
 
 3.86 
 
 1070.6 
 
 705.2 
 
 -0.71 
 
 -1.70 
 
 - 129.7 
 
 -310.6 
 
 9 
 
 Qo= 168.0 
 
 - 0.10 
 
 - 3.25 
 
 - 16.8 
 
 - 546.0 
 
 0.92 
 
 -1.22 
 
 154.6 
 
 - 205.0 
 
 + 0.94 
 
 -0.21 
 
 + 157.9 
 
 - 35.3 
 
 11 
 
 Pn' = 288.3 
 
 + 5.69 
 
 + 2.82 
 
 + 1640.4 
 
 + 813.0 
 
 -1.87 
 
 -4.14 
 
 -539.1 
 
 -1193.6 
 
 4.92 
 
 + 3.36 
 
 1418.4 
 
 + 968.7 
 
 13 
 
 Pis' = 288.3 
 
 10.14 
 
 7.58 
 
 2923.4 
 
 2185.3 
 
 -2.99 
 
 -5.10 
 
 -862.0 
 
 -1470.3 
 
 4.92 
 
 + 3.36 
 
 1418.4 
 
 + 968.7 
 
 15 
 
 Pis' = 267.3 
 
 11.27 
 
 9.29 
 
 3012.5 
 
 2483 . 2 
 
 -2.78 
 
 -4.42 
 
 - 743 . 1 
 
 -1181.5 
 
 + 0.94 
 
 -0.21 
 
 + 252.3 
 
 - 56.1 
 
 17 
 
 PIT' = 282.0 
 
 9.42 
 
 8.05 
 
 2656.4 
 
 2270.1 
 
 -2.18 
 
 3.24 
 
 -614.8 
 
 - 913.7 
 
 -0.71 
 
 -1.70 
 
 - 200.2 
 
 -479.4 
 
 19 
 
 P 19 ' = 322.7 
 
 6.07 
 
 5.28 
 
 1958.8 
 
 1703.8 
 
 -1.38 
 
 -2.03 
 
 -445.3 
 
 - 653.0 
 
 -1.19 
 
 -1.78 
 
 - 384.0 
 
 -574.4 
 
 21 
 
 P 2 i'= 375.4 
 
 2.56 
 
 2.27 
 
 961.0 
 
 852.2 
 
 -0.54 
 
 -0.87 
 
 -202.7 
 
 - 326.6 
 
 -0.63 
 
 -0.87 
 
 - 236.5 
 
 -326.6 
 
 23 
 
 Pi/ = 466.7 
 
 0.10 
 
 0.05 
 
 46.0 
 
 23.3 
 
 -0.11 
 
 -0.20 
 
 - 51.3 
 
 - 93.3 
 
 -0.14 
 
 -0.18 
 
 - 65.3 
 
 - 84.0 
 
 *- 
 
 3508.3 
 
 D= 
 
 9'.75 
 
 + 6973.9 
 
 + 453.2 
 
 D= 
 
 6'.90 
 
 -256.1 
 
 -3556.1 
 
 D = 
 
 e'.oo 
 
 + 1740.2 
 
 -632.9 
 
 
 
 
 N= 
 
 2006.4 
 
 
 
 N= 
 
 1434.8 
 
 
 
 N= 
 
 1186.6 
 
 
 
 
 
 v= 
 
 + 1'.851 
 
 
 
 v = 
 
 -1'.328 
 
 
 
 v = 
 
 +0'.467 
 
 
 
 
 
 /= 
 
 - 440.2 
 
 t28.5 
 
 
 /= 
 
 + 32.3 
 
 - 448.2 
 
 
 != 
 
 - 290.0 
 
 -105.5 
 
 Ft. 
 
 Loads 
 Q and P', 
 
 cu.ft. 
 
 SECTION m'. 
 
 SECTION V. 
 
 ft. 
 
 ft. 
 
 M e 
 
 Mi 
 
 ft. 
 
 ft. 
 
 M e 
 
 Mi 
 
 1 
 
 Qi = 367.4 
 
 -0.11 
 
 -0.20 
 
 40.4 
 
 - 73.5 
 
 0.10 
 
 0.05 
 
 36.7 
 
 18.3 
 
 3 
 
 Os = 276.1 
 
 -0.54 
 
 -0.87 
 
 - 149.1 
 
 - 240.2 
 
 2.56 
 
 2.27 
 
 706.8 
 
 626.7 
 
 5 
 
 #6 = 223.4 
 
 -1.38 
 
 -2.03 
 
 - 308.3 
 
 - 453.5 
 
 6.07 
 
 5.28 
 
 1356.0 
 
 1179.6 
 
 7 
 
 Q 7 = 182.7 
 
 -2.18 
 
 -3.24 
 
 - 398.3 
 
 - 591.9 
 
 9.42 
 
 8.05 
 
 1721.0 
 
 1470.7 
 
 9 
 
 Q 9 = 168.0 
 
 -2.78 
 
 -4.42 
 
 - 467.0 
 
 - 742.6 
 
 11.27 
 
 9.29 
 
 1893.4 
 
 1560.7 
 
 11 
 
 Pii' = 288.3 
 
 -2.99 
 
 -5.10 
 
 - 862.0 
 
 -1470.3 
 
 10.14 
 
 7.58 
 
 2923.4 
 
 2185.3 
 
 13 
 
 Pis' -288.3 
 
 -1.87 
 
 -4.14 
 
 - 539.1 
 
 -1193.6 
 
 5.69 
 
 2.82 
 
 1640.4 
 
 813.0 
 
 15 
 
 Pis' = 267. 3 
 
 +0.92 
 
 -1.22 
 
 + 245.9 
 
 - 326.0 
 
 -0.10 
 
 - 3.25 
 
 26.7 
 
 - 868.7 
 
 17 
 
 Pn' = 282.0 
 
 5.86 
 
 + 3.86 
 
 1652.5 
 
 + 1088.5 
 
 -5.50 
 
 - 8.63 
 
 -1551.0 
 
 -2433.7 
 
 19 
 
 Pi 9 ' = 322.7 
 
 5.86 
 
 5.35 
 
 1891.0 
 
 1726.4 
 
 -9.19 
 
 -12.18 
 
 -2965.6 
 
 -3930.5 
 
 21 
 
 P 2 i' = 375.4 
 
 1.98 
 
 1.73 
 
 743.3 
 
 649.4 
 
 -8.55 
 
 -11.39 
 
 -3210.5 
 
 -4275.8 
 
 23 
 
 P 24 ' = 466.7 
 
 0.33 
 
 0.28 
 
 154.0 
 
 130.7 
 
 -2.15 
 
 - 4.87 
 
 -1003.4 
 
 -2272.8 
 
 7- 
 
 3508.3 
 
 D- 
 
 6'. 90 
 
 + 1922.5 
 
 -1496.6 
 
 Z>= 
 
 9'.75 
 
 + 1520.5 
 
 -5927.2 
 
 
 
 
 if- 
 
 1485 . 3 
 
 
 
 N= 
 
 2291.4 
 
 
 
 
 
 v = 
 
 + OM44 
 
 
 
 v= 
 
 -0'.962 
 
 
 
 
 
 /= 
 
 - 242.3 
 
 - 188.7 
 
 
 /= 
 
 - 96.0 
 
 - 374.1 
 
 The influence line ordinates ij for points TO' and tf are symmetric with those of points m and t. See note, foot of 
 Table 79i. 
 
344 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XV 
 
 By referring to Fig. 79E, it is seen that this position of loading gives maximum stresses 
 for the section I and almost maximum for section m, but not for the other sections. In 
 the present instance the effects at the five critical sections were found for one position 
 of the moving load, giving a simultaneous condition of stress, but not maximum stresses 
 at all sections. 
 
 The maximum stresses for each of the critical sections must be computed for special 
 positions of the moving train as illustrated in Table 79K, for the crown section. 
 
 In the general investigation of stresses it is best to obtain the dead load moments 
 separately for each section and combine these with the live load moments which are found 
 without regard to impact and then increased for impact prior to adding the dead load 
 moments, as shown in Table 79K. 
 
 TABLE 79K 
 
 INVESTIGATION FOR MAXIMUM STRESSES AT THE CROWN SECTION "n." 
 
 SYMMETRIC ARCH 
 
 
 Dead 
 
 Total Dead Load Moments. 
 
 No. 
 
 l-I'L 1 
 
 
 Max. + Live Load Moments. 
 
 Point. 
 
 Loads 
 
 
 
 
 
 vv neel 
 
 Live 
 
 
 
 
 
 
 
 
 
 
 
 from 
 
 Loads, 
 
 
 
 
 
 
 Q 
 
 ije 
 
 1,1 
 
 M e 
 
 Mi 
 
 Head of 
 Train. 
 
 
 f)e 
 
 'Ji 
 
 M e 
 
 Mi 
 
 
 cu.ft. 
 
 ft. 
 
 ft. 
 
 
 
 
 cu.ft. 
 
 ft. 
 
 ft. 
 
 
 
 1 
 
 367.4 
 
 -0.14 
 
 -0.18 
 
 - 51.4 
 
 - 66.1 
 
 1 
 
 16.5 
 
 0.00 
 
 -1.19 
 
 0.0 
 
 - 19.6 
 
 3 
 
 276.1 
 
 -0.63 
 
 -6.87 
 
 -173.9 
 
 -240.2 
 
 2 
 
 33.0 
 
 1.30 
 
 + 0.10 
 
 42.9 
 
 + 3.3 
 
 5 
 
 223.4 
 
 -1.19 
 
 -1.78 
 
 -265.8 
 
 -397.6 
 
 3 
 
 33.0 
 
 2.55 
 
 1.10 
 
 84.2 
 
 36.3 
 
 7 
 
 182.7 
 
 -0.71 
 
 -1.70 
 
 -129.7 
 
 -310.6 
 
 4 
 
 33.0 
 
 4.10 
 
 2.55 
 
 135.3 
 
 84.2 
 
 9 
 
 168.0 
 
 + 0.94 
 
 -0.21 
 
 + 157.9 
 
 - .35.3 
 
 5 
 
 33.0 
 
 6.10 
 
 4.45 
 
 201.3 
 
 146.9 
 
 11 
 
 173.7 
 
 4.92 
 
 + 3.36 
 
 854.6 
 
 + 583.6 
 
 6 
 
 21.4 
 
 6.40 
 
 4.80 
 
 137.0 
 
 102.7 
 
 13 
 
 173.7 
 
 4.92 
 
 + 3.36 
 
 854.6 
 
 + 583.6 
 
 7 
 
 21.4 
 
 4.40 
 
 2.80 
 
 94.2 
 
 59.9 
 
 15 
 
 168.0 
 
 0.94 
 
 -0.21 
 
 157.9 
 
 - 35.3 
 
 8 
 
 21.4 
 
 2.55 
 
 1.00 
 
 54.6 
 
 21.4 
 
 17 
 
 182.7 
 
 -0.71 
 
 -1.70 
 
 -129.7 
 
 -310.6 
 
 9 
 
 21.4 
 
 1.30 
 
 0.00 
 
 27.8 
 
 0.0 
 
 19 
 
 223.4 
 
 -1.19 
 
 -1.78 
 
 -265.8 
 
 -397.6 
 
 10 
 
 0.0 
 
 0.00 
 
 -1.19 
 
 0.0 
 
 0.0 
 
 21 
 
 27fi 1 
 
 fi^ 
 
 087 
 
 mo 
 
 74O 9 
 
 
 
 
 
 
 
 ft A 
 
 23 
 
 fff \f A 
 
 367.4 
 
 \J . UO 
 
 -0.14 
 
 . 01 
 
 -0.18 
 
 . U 
 
 - 51.4 
 
 HI . , 
 
 - 66.1 
 
 Max.+ L.L.= 
 
 + 777.3 
 
 + 435.1 
 
 
 
 
 
 
 
 84% 
 
 
 652.9 
 
 + 365.5 
 
 
 
 
 R = 
 
 2782.6 
 
 Z> = 
 
 6'. 00 
 
 + 783.4 
 
 -932.4 
 
 
 
 
 
 
 
 N = 
 
 857.9 
 
 
 TotaFD.L.= 
 
 783.4 
 
 -932.4 
 
 
 
 
 v = 
 
 -0'.087 
 
 
 Total max. + A/ = 
 
 2213.6 
 
 -131.8 
 
 
 
 
 f= 
 
 -130.6 
 
 -155.4 
 
 max.A' r = 
 
 1190.8 
 
 =-<y.88 
 
 
 
 
 
 
 
 max./= 
 
 -369.0 
 
 - 22.9 
 
 
 
 
 
 
 1 
 
 
 
 All stresses / are in cubic feet per square fook Multiply by 0.972 to reduce to pounds per square 
 
 300 
 
 inch. Impact for 56 feet is =0.84. 
 oUO+56 
 
 The live load moments are found by placing the train of wheel loads for Cooper's 
 E 60 loading over the positive portions of the M e and M., influence lines for the section 
 in question to obtain the simultaneous maximum positive moments for the kernel points. 
 Similarly the negative portions of these influence lines will yield the maximum negative 
 moments for the same section. 
 
ART. 79 FIXED MASONRY ARCHES 345 
 
 Since the loads are distributed over a roadbed 13 feet in width the train of wheel 
 loads, for the present purpose, is obtained by dividing the standard axle loads by 13, and 
 also converting the loads into cubic feet of masonry, all as shown in Fig. 79o. The load 
 diagram must of course be drawn to the same horizontal scale as the influence lines, which 
 is not the case in the figures as here reproduced. 
 
 The wheel loads, in kips, are converted into cubic feet of masonry by multiplying 
 kips by 1000/140-7.14. 
 
 The maximum stresses at the crown section n are computed in Table 79ic. The dead 
 load moments are first found and these are combined with the live load moments which 
 latter are increased for impact. The maximum stresses are found from the total maximum 
 moments. 
 
 The live load diagram Fig. 79c, was placed over the influence line for section n, Fig. 
 70E, so that the first load came at the point 8, which is the zero point for positive ordinates. 
 One locomotive then covered the distance from point 8 to 16, and the remainder of the 
 span was considered empty. The same position was used for both M e and Mi moments, 
 because Eqs. (76c) apply only to simultaneous kernel moments. 
 
 UNIFORM J.OAO OP 0.461 kiM p.tT. 
 
 V $' 5' S' 
 
 FIG. 79c. Cooper's E 60 Loading for an Arch Ring One Foot Thick. 
 
 Comparing the result for max. f e = 369 cu.ft., obtained here with the value given 
 in Table 79j as 290 cu.ft., the relative effect of the two cases of loading is clearly seen. 
 
 Each of the critical sections should be examined in the manner shown in Table 
 79K, and the positive as well as negative influence line areas should be considered as sepa- 
 rate cases. 
 
 The temperature stresses are easily determined according to the method given in 
 Art. 75. 
 
 Referring back to the given data, E =3,670,000 Ibs. per sq.in. =3,755,720 cu.ft. 
 per sq.ft.; =0.000,005,4 per 1 F; =-80; Z = 158.8 ft; and cos /? = !, then etf = 
 -0.0686 ft., and by Eq. (75c) 
 
 dlE 0.0686 X3,775,720 
 
 where the value cos 2 /?t/TF c is taken from Table 79c for W c loads which are E times 
 actual. 
 
 The stresses on any section are then found by evaluating the kernel moments for 
 temperature and combining these with the kernel moments previously found for the case 
 of maximum loading. 
 
346 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XV 
 
 Thus, for the haunch section t, Table 79o gives D =9'. 75, y e = -37'. 1 and t/ t -= -35'.3 
 and maximum M e and Mi are taken from Table 79j as follows: 
 
 M et = -37.1 X 230.9= -8566.4 
 M e from Table 79 j- +6973.9 
 
 Total M e =-1592.5 
 
 .... , 6X1592.5 , inn ,. 
 
 giving f e = + , = + lOO.o cu.ft. 
 
 9.7o~ 
 
 Also, 
 
 Mil= -35.3X230.9= -8150.8 
 Mi from Table 79 j = + 453.2 
 
 Total M t =-7697. 6 
 
 . . , 6X7697.6 ,. e ,, 
 
 giving/, = ., = 458.8 cu.ft. 
 
 9.75" 
 
 Similarly for the crown section n where Z>=6'.0, y e = ll'.OS and i/; = 13'.08, then for 
 the maximum moments from Table 79K 
 
 M et = 10.08 X 230.9 = + 2558.4 
 M e from Table 79K - + 2213.6 
 
 Total M e = +4772.0 or/ e = -795.3 cu.ft., 
 
 and M it = 13.08 x 230.9 = + 3020.2 
 
 Mi from Table 79K = - 131.8 
 
 Total M;= +2888.4 or /= +481.4 cu.ft. 
 
 It is readily seen that these stresses are excessive and would undoubtedly cause 
 cracks on the tension elements occurring in the intrados at the crown and in the extrados 
 at the haunches. 
 
 (c) Analysis as under (b), treating the structure as unsymmetric. It is presumed 
 that for good and sufficient reasons the B abutment had to be founded on a lower stratum 
 of rock, a condition which developed after the original design for the symmetric arch 
 was accepted. The question to be answered here is whether such a change in the location 
 of a foundation would seriously alter the resulting stresses and necessitate a change in 
 the design. 
 
 The location of the (x, y) axes and computation of the elastic loads is given in Tables 
 79L and 79.M, following the method outlined in Art. 73, and using the same data as in the 
 previous case of the symmetric arch with the addition of two sections at the right-hand 
 end, producing the unsymmetric condition illustrated in Fig. 79u. . 
 
 The (z, v) axes are located through the axial points of the haunches, which is the 
 same position previously chosen for these axes in dealing with the symmetric condition. 
 
 The dimensions D, v and z are tabulated in Table 79L, and from these the values 
 1/7, sec (j>, w a , vw a and zw a are computed, using the formulae for sec < above given, p. 331. 
 
ART. 79 
 
 FIXED MASONRY ARCHES 
 
 347 
 
 O O O C5 
 
 co TP co 
 
 K 
 
 
 
 + I I I I I I I I I I I I I 
 
 B 
 tf 
 
 H 
 
 i + r I I I I J M I I I I 
 
 J5 N 'T* * .p "5 
 
 h-*. W I! *-H iO 00 C^ CM ^O 00 C^ CM CO t^ ^2 *"i CO I^ CO CM CO 00 *O C^ ^D GO ^O 01 i~H Cw CO **"* PH 
 
 *" f-H T i I-H I-H C^4 CM C^ C^ CO CO CO CM CM C^J C^l i~~i *~H I-H i^ CM M 
 
 t>. .^ 
 
 !>. CO OXOOCOCOCOQOOOOO' i^QOO' iLOOOOOCOT-HOOr^OOO * "5 
 
 CC - C^^CCOCN' iiOiOC v J^ ( O l OQO'^t | OOC^lT-HCM <^ fOOC^QO'^O'~'Ot^- CO 
 
 O5 Ti* S 
 
 r 1 P 
 
 K fe^ OOOOOOOOOOOOOOOOOOOOOOOOOOO CD i "- 
 
 a 
 
 ^ ^ PN 
 
 fa ^ 
 
 o 
 
 ^ ~ T ^ ^ ^ ^ ^ ~ ~ il, " ~ ~ ^-, ^. ^- .- ^ ^ ~ ^ N a 
 
 ?) OOi-Hi-H'-t(N(MCOCOCOCO^Tt < TtiCOCOCOCOCMCM'-li-H' lOOOO 
 
 H 
 
 O g Or-HCOCOGO(M_COO^OOCOOOTt<COOOp^HCMr-(pl> p-HOOt--^H M C5 
 
 H 2. __^ SS S 
 
 S PI 1 c^^i^^^^ooot-Cicoooococnt-ooo^^LCic^^as-H T T ^ 
 
 -j i"^ OOOOOOOOOOOOOOOOOOOOOOOOOOO 
 
 do'dddooodoooooooooooooooooo 
 
 fa 
 
 CO I - 
 
 "- coooooOOTc^t^^^^c^doocM^^^t^cM'osboQGOcocD'-o 
 
 o 
 
 H^ 
 
 H 
 
 f^ II *j p p p p p c5 p p p p p p p p p p p p p p p p p p p ~ 
 
 ^ r-i i >~H o' o' o o o' o d d o d o o' o' o o' o o' d d o o o' d o o o o 
 
 o 
 p 
 
 2 CMCO^f^O^COCOcOt-t^-t^-t^-t t t- t^-t 5>cOCD'O l O^COCM> I 
 
 p ^ 
 
 t-H 
 
 1 1 
 
 ^~* OOOOOOO-^cO. OOOOOOOcOrPOOpOOOOOO 
 
 ___^ co^f 
 
 O I-H (N CO -^ 1C CO t^'OO 
 PH 
 
348 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 TABLE 
 COMPUTATION OF W FORCES AND POLE 
 
 Pt. 
 
 W a 
 
 XZW a 
 
 .V. 
 
 = xz W a 
 
 ^ 
 
 Eq. (73c), 
 
 ?/ 2 2 ' ~\~ X T ' ill 3 
 
 XW a 
 
 
 x tan 3 
 
 2 2 ' 
 
 y 
 
 ft. 
 
 
 
 0.0232 
 
 49.8 
 
 153.2 
 
 155.1 
 
 446.6 
 
 -4.22 
 
 -37.17 
 
 -41.39 
 
 1.885 
 
 1 
 
 0.0251 
 
 66.6 
 
 142.6 
 
 414.5 
 
 885.0 
 
 -3.91 
 
 -28.37 
 
 -32.28 
 
 1.892 
 
 2 
 
 0.0285 
 
 85.0 
 
 135.2 
 
 551.2 
 
 877.8 
 
 -3.57 
 
 -20.27 
 
 -23.84 
 
 1.963 
 
 3 
 
 0.0325 
 
 101.4 
 
 126.4 
 
 656.0 
 
 817.5 
 
 -3.24 
 
 -13.57 
 
 -16.81 
 
 2.027 
 
 4 
 
 0.0354 
 
 110.4 
 
 110.5 
 
 716.8 
 
 718.8 
 
 -2.90 
 
 - 7.77 
 
 -10.67 
 
 1.978 
 
 5 
 
 0.0394 
 
 118.3 
 
 96.0 
 
 765.2 
 
 622.3 
 
 -2.56 
 
 - 2.77 
 
 - 5.33 
 
 1.945 
 
 6 
 
 0.0429 
 
 119.7 
 
 78.8 
 
 774.2 
 
 511.5 
 
 -2.22 
 
 + 1.53 
 
 - 0.69 
 
 1.839 
 
 7 
 
 0.0458 
 
 114.4 
 
 60.6 
 
 739.5 
 
 393 . 9 
 
 -1.89 
 
 5.13 
 
 + 3.24 
 
 1.666 
 
 8 
 
 0.0477 
 
 102.0 
 
 42.6 
 
 661.0 
 
 278.3 
 
 -1.55 
 
 8.03 
 
 6.48 
 
 1.425 
 
 9 
 
 0.0499 
 
 86.1 
 
 27.3 
 
 556.8 
 
 178.8 
 
 -1.21 
 
 10.23 
 
 9.02 
 
 1.166 
 
 10 
 
 0.0513 
 
 65.2 
 
 14.6 
 
 450.6 
 
 102.8 
 
 -0.88 
 
 11.73 
 
 10.85 
 
 0.865 
 
 11 
 
 0.0530 
 
 38.0 
 
 4.6 
 
 283.2 
 
 38.0 
 
 -0.49 
 
 12.93 
 
 12.44 
 
 0.497 
 
 12 
 
 0.0555 
 
 8.0 
 
 0.2 
 
 51.2 
 
 5.2 
 
 -0.10 
 
 13.23 
 
 13.13 
 
 0.104 
 
 13 
 
 0.0530 
 
 - 22.8 
 
 1.7 
 
 -168.6 
 
 16.3 
 
 + 0.29 
 
 12.93 
 
 13 . 22 
 
 -0.298 
 
 14 
 
 0.0513 
 
 - 50.7 
 
 8.8 
 
 -349.1 
 
 62.8 
 
 0.68 
 
 11.73 
 
 12.41 
 
 -0.674 
 
 15 
 
 0.0499 
 
 - 72.3 
 
 19.2 
 
 -467.3 
 
 126.4 
 
 1.02 
 
 10.23 
 
 11.25 
 
 -0.980 
 
 16 
 
 0.0477 
 
 - 89.2 
 
 32.6 
 
 -577.8 
 
 213.4 
 
 1.36 
 
 8.03 
 
 9.39 
 
 - 1 . 246 
 
 17 
 
 0.0458 
 
 -102.7 
 
 48.8 
 
 -663.6 
 
 317.5 
 
 1.69 
 
 5.13 
 
 6.82 
 
 -1.494 
 
 18 
 
 0.0429 
 
 -109.3 
 
 65.7 
 
 -706.6 
 
 426.6 
 
 2.03 
 
 1.53 
 
 3.56 
 
 - 1 . 679 
 
 19 
 
 0.0394 
 
 -109.3 
 
 82.0 
 
 -706.8 
 
 531.7 
 
 2.37 
 
 - 2.77 
 
 - 0.40 
 
 - 1 . 798 
 
 20 
 
 0.0354 
 
 -103.0 
 
 96.2 
 
 -668.5 
 
 625.8 
 
 2.71 
 
 - 7.77 
 
 - 5.06 
 
 -1.845 
 
 21 
 
 0.0325 
 
 - 95.3 
 
 111.7 
 
 -615.4 
 
 722.4 
 
 3.04 
 
 -13.57 
 
 -10.53 
 
 -1.906 
 
 22 
 
 0.0285 
 
 - 80.4 
 
 120.9 
 
 -521.2 
 
 784.9 
 
 3.38 
 
 -20.27 
 
 -16.89 
 
 - 1 . 856 
 
 23 
 
 0.0251 
 
 - 63.3 
 
 128.8 
 
 -388.0 
 
 782.1 
 
 3.72 
 
 -28.37 
 
 -24.65 
 
 - 1 . 798 
 
 24 
 
 0.0174 
 
 - 35.6 
 
 104.6 
 
 -214.3 
 
 621.6 
 
 4.02 
 
 -37.17 
 
 -33.15 
 
 -1.349 
 
 25 
 
 0.0129 
 
 - 16.7 
 
 89.8 
 
 - 99.6 
 
 544.1 
 
 4.33 
 
 -48.07 
 
 -44.74 
 
 -1.076 
 
 26 
 
 0.0131 
 
 0.0 
 
 104.5 
 
 - 8.2 
 
 301.0 
 
 4.64 
 
 -63.57 
 
 -58.93 
 
 -1.170 
 
 
 
 
 
 + 620.3 
 
 11953.1 
 
 
 
 
 
 All elastic loads W 
 
 The sums ^w a = W a , ^vw a =vW a and Hzw a =zW a are now computed by Simpson's 
 rule according to Eqs. (73E), using the lengths Jv for the horizontal distances between 
 the arch sections. 
 
 Eqs. (73r) then give the coordinates l\, z ' of the new origin with respect to the 
 (v, z) axes. Thus 
 
 2vW a 545.7264 
 
 6.7149 
 
 =81.272 feet, 
 =63.574 feet, 
 
 and x=li v. 
 
ART. 79 
 
 FIXED MASONRY ARCHES 
 
 349 
 
 79M 
 
 DISTANCES. UNSYMMETRTC ARCH 
 
 ft. 
 
 z^-r, 
 
 yw a 
 
 2 yWa We 
 
 y 2 
 
 * 
 
 
 
 cos <f> 
 
 2 cos <j>Jx 
 
 Point. 
 
 
 D 
 
 D 
 
 
 5.56 
 
 -0.960 
 
 - 2.76 
 
 1713.1 
 
 39.74 
 
 110.5 
 
 052 
 
 0.158 
 
 
 
 5.9 
 
 11.77 
 
 -0.810 
 
 - 5.04 
 
 1042.0 
 
 26.15 
 
 163 . 4 
 
 0.064 
 
 0.394 
 
 1 
 
 6.5 
 
 12.77 
 
 -0.679 
 
 - 4.43 
 
 568.3 
 
 16.20 
 
 106 8 
 
 0.077 
 
 0.500 
 
 2 
 
 6.5 
 
 13.13 
 
 -0.546 
 
 - 3.54 
 
 282 6 
 
 9.18 
 
 60.6 
 
 090 
 
 0.583 
 
 3 
 
 6.5 
 
 12.88 
 
 -0.378 
 
 - 2.47 
 
 113.8 
 
 4.03 
 
 27.4 
 
 0.101 
 
 0.657 
 
 4 
 
 6.5 
 
 12.61 
 
 -0.210 
 
 - 1.36 
 
 28.4 
 
 1.12 
 
 8.3 
 
 0.113 
 
 0.732 
 
 5 
 
 6.5 
 
 11.92 
 
 -0.030 
 
 - 0.13 
 
 0.5 
 
 0.02 
 
 1.0 
 
 0.123 
 
 0.799 
 
 6 
 
 6.5 
 
 10.80 
 
 + 0.148 
 
 + 0.95 
 
 10.5 
 
 0.48 
 
 3.7 
 
 0.135 
 
 0.872 
 
 7 
 
 6.5 
 
 9.26 
 
 0.309 
 
 1.99 
 
 42.0 
 
 2 00 
 
 13.3 
 
 0.144 
 
 0.930 
 
 8 
 
 6.5 
 
 7.56 
 
 0.450 
 
 2.90 
 
 81.4 
 
 4.06 
 
 26.3 
 
 0.153 
 
 0.987 
 
 9 
 
 6.5 
 
 5.99 
 
 0.557 
 
 3.89 
 
 117.7 
 
 6.04 
 
 42.6 
 
 0.158 
 
 1.110 
 
 10 
 
 7.5 
 
 3.71 
 
 0.659 
 
 4.91 
 
 154.8 
 
 8.20 
 
 61.0 
 
 0.163 
 
 1.225 
 
 11 
 
 7 .5 
 
 
 
 
 
 
 
 
 
 
 
 0.73 
 
 0.729 
 
 5.39 
 
 172.4 
 
 9.57 
 
 70.7 
 
 0.167 
 
 1.250 
 
 12 
 
 7.5 
 
 - 2.22 
 
 0.701 
 
 5.22 
 
 174.8 
 
 9.26 
 
 68.8 
 
 0.163 
 
 1.225 
 
 13 
 
 7.5 
 
 4.65 
 
 0.637 
 
 4.45 
 
 154.0 
 
 7.90 
 
 55.3 
 
 0.158 
 
 1.110 
 
 14 
 
 6.5 
 
 6.35 
 
 0.561 
 
 3.62 
 
 126.6 
 
 6.32 
 
 40.8 
 
 0.153 
 
 0.987 
 
 15 
 
 6.5 
 
 - 8.08 
 
 0.448 
 
 2.89 
 
 88.2 
 
 4.21 
 
 27.4 
 
 0.144 
 
 0.930 
 
 16 
 
 6.5 
 
 - 9.66 
 
 0.312 
 
 2.01 
 
 46.5 
 
 2.13 
 
 - 14.1 
 
 0.135 
 
 0.872 
 
 17 
 
 6.5 
 
 - 10.86 
 
 0.153 
 
 0.99 
 
 12.7 
 
 0.54 
 
 4.1 
 
 0.123 
 
 0.799 
 
 18 
 
 6.5 
 
 - 11.63 
 
 -0.016 
 
 - 0.08 
 
 0.2 
 
 0.01 
 
 0.8 
 
 0.113 
 
 0.732 
 
 19 
 
 6.5 
 
 - 11.98 
 
 -0.179 
 
 - 1.16 
 
 25.6 
 
 0.91 
 
 6.9 
 
 0.101 
 
 0.657 
 
 20 
 
 6.5 
 
 - 12.31 
 
 -0.342 
 
 - 2.21 
 
 110.9 
 
 3.60 
 
 24.4 
 
 0.090 
 
 0.583 
 
 21 
 
 6.5 
 
 - 12.05 
 
 -0.481 
 
 - 3.14 
 
 285.3 
 
 8.13 
 
 54.2 
 
 0.077 
 
 0.500 
 
 22 
 
 6.5 
 
 - 10.94 
 
 -0.619 
 
 - 3.75 
 
 607.6 
 
 15.25 
 
 92.6 
 
 0.064 
 
 0.393 
 
 23 
 
 5.9 
 
 
 
 
 
 
 
 
 
 
 
 8.03 
 
 -0.577 
 
 - 3.44 
 
 1098.9 
 
 19.12 
 
 114.2 
 
 0.047 
 
 0.275 
 
 24 
 
 5.9 
 
 6.53 
 
 -0.577 
 
 -3.52 
 
 2001 . 7 
 
 25.82 
 
 158.7 
 
 0.032 
 
 0.187 
 
 25 
 
 5.9 
 
 3.40 
 
 -0.772 
 
 - 2.18 
 
 3472.7 
 
 45.49 
 
 124.5 
 
 0.020 
 
 0.064 
 
 26 
 
 
 + 118.69 
 
 
 + 39.21 
 
 
 
 1482.4 
 
 
 19.511 
 
 
 
 -118.69 
 
 
 -39.21 
 
 
 .#:,*. 
 
 = 1480.6 
 
 
 
 
 are E times actual. 
 
 When the ordinates x are found then the quantities xzw a and x 2 w a and their sums 
 xzW a and xWi, are determined in Table 79M, using Simpson's rule in summing for the 
 horizontal increments Ax. 
 
 Eq. (73n) then gives 
 
 620.3 
 
 tan 8 = 
 
 11953.1 
 
 = -0.0519. 
 
 Eq. (73o) now furnishes the ordinates y and from these the quantities yw a and 
 y 2 w a are computed. As a check STF&=0 and SW C =0, and any small discrepancies 
 must be adjusted. 
 
350 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XV 
 
ART. 79 
 
 FIXED MASONRY ARCHES 
 
 351 
 
 The W loads were again derived without including E so that these loads are all E 
 times actual, a fact which should not be overlooked if temperature stresses are to be 
 found from Eq. (75c). 
 
 In Table 79M the quantities cos (f>dx/D are given, showing that the pole distance 
 H c =cos 52?/TF c + 2cos (f>Ax/D from Eq. (72ic), would be increased by 1.3 per cent if axial 
 thrust is included in the determination of X c . This would tend to diminish X c , and the 
 ordinates of the X c influence line. 
 
 The influence lines for X a , X b and X c are shown in Fig. 79i, noting that the length of 
 span /, is now taken as the horizontal distance between the intrados kernel points of the 
 haunch sections. This was done to obviate the negative M moments otherwise encoun- 
 tered here and leading to complications which are rather far reaching. In the previous 
 case of symmetric arch, the real haunch sections could not be examined becuase the span 
 was chosen as the distance between the verticals through the axial points and the sec- 
 tion t was accordingly taken as the nearest approach to the haunch section. It is thus 
 interesting to note the two methods of treating this matter. The length does not 
 enter into the computations, but must be considered in drawing all the influence 
 lines. 
 
 The detailed description of the method of drawing these influence lines need riot 
 be repeated here, and the reader is referred back to the treatment of the symmetric arch 
 in the first part of this article. 
 
 Axial thrust was not included in Fig. 79i, thus making all the results of both investiga- 
 tions comparable. However, the effect is so small that it may be regarded as a negligible 
 quantity in the present case. 
 
 Having located the coordinate axes, this determines the critical sections, and the data 
 required for the computation of the ordinates for the M e and Mi influence lines are given 
 in Table 79N. 
 
 TABLE 79N. DATA RELATING TO THE CRITICAL SECTIONS UNSYMMETRIC ARCH 
 
 
 
 
 Coordinates, 
 
 
 
 Coordinates Extrados Kernel 
 
 Coordinates Intrados Kernel 
 
 
 
 
 Axial Point. 
 
 
 
 Points. 
 
 Points. 
 
 T> ' 
 
 
 
 
 D . 
 
 D , 
 
 
 
 
 
 
 X 
 
 y 
 
 g sm <j> 
 
 g COS $ 
 
 x e 
 
 Ve 
 
 cos @y e 
 
 *i 
 
 Vi 
 
 COS &/j 
 
 
 ft. 
 
 
 ft. 
 
 ft. 
 
 ft. 
 
 ft. 
 
 ft. 
 
 ft. 
 
 ft. 
 
 ft. 
 
 ft. 
 
 ft. 
 
 A 
 
 10.0 
 
 58 51' 
 
 + 81.27 
 
 -41.39 
 
 1.43 
 
 0.86 
 
 + 79 . 85 
 
 -42.25 
 
 -42.20 
 
 + 82.70 
 
 -40.53 
 
 -40.48 
 
 m 
 
 6.9 
 
 31 04 
 
 + 41.20 
 
 0.00 
 
 0.59 
 
 1.00 
 
 + 40.61 
 
 - 1.00 
 
 - 1.00 
 
 + 41.79 
 
 + 1.00 
 
 + 1.00 
 
 n 
 
 6.0 
 
 00 
 
 0.00 
 
 + 13.25 
 
 0.00 
 
 1.00 
 
 0.00 
 
 + 12.25 
 
 + 12.23 
 
 0.00 
 
 + 14.25 
 
 + 14.23 
 
 m" 
 
 7.2 
 
 34 58 
 
 -43.90 
 
 0.00 
 
 0.69 
 
 0.98 
 
 -43.21 
 
 - 0.98 
 
 - 0.98 
 
 -44.59 
 
 + 0.98 
 
 + 0.98 
 
 B 
 
 14.7 
 
 73 11 
 
 -89.33 
 
 -58.93 
 
 2.35 
 
 0.71 
 
 -86.98 
 
 -59.64 
 
 -59.56 
 
 -91.68 
 
 -58.22 
 
 -58.14 
 
 /?=2 58' and cos = 0.9987 from Table 79L 
 
 The M influence lines for the kernel points of the several critical sections are shown 
 in Fig. 79J, the data for the construction of these being taken from Table 79x. It should 
 be noted that for the haunch sections A and B, M ot -=0. 
 
352 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XV 
 
 1=174138 
 
 FIG. 79i. X Influence Lines for Unsymmetric Arch 
 
ART. 79 
 
 FIXED MASONRY ARCHES 
 
 353 
 
 From the coordinates of the kernel points in Table 79N, and the ordinates in Tables 
 79o and 79p, taken from the X a , X b , X c and M influence lines, Figs. 79i and 79J, the T) e 
 and j)i are computed for the odd points and for the five critical sections according to Eqs. 
 (76B). The operations are all indicated in the headings of Tables 79o and 79p, thus 
 requiring no further description here. 
 
 These ordinates, plotted in Fig. 79K, furnish the M e and M{ influence lines from 
 which the moments and stresses for any position of the live loads may be found for each 
 of the five critical sections. In practice it is well to compute all the ordinates TJ 
 instead of merely the odd ones as given here. 
 
 FOR INTRADOS KERNEL POINTS. 
 
 FORCXTRAOOS KERNEL POINTS, 
 
 FIG. 79j. M Influence Lines for Unsymmetric Arch. 
 
 Table 79o, gives the computations for the resultant polygon for the same case of 
 loading previously used in Table 79n, adding one additional load P^ 5 at a point on the 
 right abutment. The results of this computation are shown in Fig. 79n, where the resultant 
 polygon is drawn through the points a and b which are located on the verticals through 
 the intrados kernel points of the A and B sections. 
 
 It should be noted that c and c are both negative, indicating that they must be meas- 
 ured up from the line ah when found. Hence z is measured down on the y axis to 
 locate the point s and c =ss' is then laid off upward from s. The line as'||z 
 axis, locates the point a at the left abutment and the line as prolonged fixes the point 6 
 at the right abutment. The force polygon is therl easily constructed by laying off 
 the reactions A and B and finding the pole 0\ by drawing the line 0\T \\ ab and equal in 
 length to H', 
 
354 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XV 
 
 TABLE 79o. MOMENT INFLUENCE LINES FOR 
 
 Point. 
 
 90 
 
 ft. 
 
 fib 
 ft. 
 
 rye 
 ft. 
 
 SECTION A. 
 
 r> A = r> - [ij a + 79.857/6 - 42.27j c ] 
 
 SECTION TO. 
 rjm = ijo [i)a + 40.6j?& l.Oqc] 
 
 r lo 
 
 79.85 t)b 
 
 -42.2r,c 
 
 IA 
 
 T>0 
 
 + 40.67,6 
 
 T)m 
 
 1 
 
 3.5 
 
 0.036 
 
 0.012 
 
 2.74 
 
 2.87 
 
 - 0.51 
 
 - 3.12 
 
 5.2 
 
 1.46 
 
 0.2t> 
 
 3 
 
 9.8 
 
 0.082 
 
 0.100 
 
 2.52 
 
 6.55 
 
 - 4.22 
 
 - 9.61 
 
 15.2 
 
 3.33 
 
 2.17 
 
 5 
 
 15.2 
 
 0.102 
 
 0.252 
 
 2.31 
 
 8.14 
 
 -10.63 
 
 -10.40 
 
 25.1 
 
 4.14 
 
 6.01 
 
 7 
 
 19.7 
 
 0.092 
 
 0.424 
 
 2.10 
 
 7.35 
 
 -17.89 
 
 - 7.06 
 
 31.0 
 
 3.74 
 
 7.98 
 
 9 
 
 22.9 
 
 0.064 
 
 0.580 
 
 1.88 
 
 5.11 
 
 -24.48 
 
 - 1.65 
 
 28.0 
 
 2.60 
 
 3.08 
 
 11 
 
 25.0 
 
 0.014 
 
 0.692 
 
 1.64 
 
 1.12 
 
 -29.20 
 
 + 4.72 
 
 24.5 
 
 0.57 
 
 -0.38 
 
 13 
 
 25.5 
 
 -0.044 
 
 0.720 
 
 1.40 
 
 - 3.51 
 
 -30.38 
 
 9.79 
 
 21.0 
 
 -1.79 
 
 -1.99 
 
 15 
 
 24.5 
 
 -0.100 
 
 0.652 
 
 1.18 
 
 - 7.99 
 
 -27.51 
 
 12.18 
 
 17.4 
 
 -4.06 
 
 -2.39 
 
 17 
 
 22.1 
 
 -0.132 
 
 0.520 
 
 0.97 
 
 -10.54 
 
 -21.94 
 
 11.35 
 
 14.2 
 
 -5.36 
 
 -2.02 
 
 19 
 
 18.7 
 
 -0.148 
 
 0.360 
 
 0.75 
 
 -11.62 
 
 -15.19 
 
 8.86 
 
 11.1 
 
 -6.01 
 
 -1.23 
 
 21 
 
 14.3 
 
 -0.136 
 
 0.196 
 
 0.53 
 
 -10.86 
 
 - 8.27 
 
 5.36 
 
 8.0 
 
 -5.52 
 
 -0.58 
 
 23 
 
 9.1 
 
 -0.100 
 
 0.070 
 
 0.32 
 
 - 7.99 
 
 - 2.95 
 
 2.16 
 
 4.8 
 
 -4.06 
 
 -0.17 
 
 25 
 
 3.9 
 
 -0.046 
 
 0.016 
 
 0.13 
 
 - 3.67 
 
 - 0.68 
 
 0.58 
 
 2.0 
 
 -1.87 
 
 -0.01 
 
 All ordinates in above 
 
 TABLE 79p. MOMENT INFLUENCE LINES FOR 
 
 Point. 
 
 T/a 
 ft. 
 
 >?& 
 ft. 
 
 T)c 
 ft. 
 
 SECTION A. 
 ^ = 0-[7j a +82.77j6-40.5rjc] 
 
 SECTION m. 
 
 fjm = rjo [f)a + 41 .87J6 + 1.0)jf] 
 
 S2.7rj b 
 
 -40.57y c 
 
 r >A 
 
 1)0 
 
 41.87/6 
 
 l.OlJc 
 
 TJm 
 
 1 
 
 3.5 
 
 0.036 
 
 0.012 
 
 2.98 
 
 - 0.49 
 
 - 5.99 
 
 5.4 
 
 1.50 
 
 0.01 
 
 0.39 
 
 3 
 
 9.8 
 
 0.082 
 
 0.100 
 
 6.78 
 
 - 4.05 
 
 -12.53 
 
 15.4 
 
 3.43 
 
 0.10 
 
 2.07 
 
 5 
 
 15.2 
 
 0.102 
 
 0.252 
 
 8.44 
 
 -10.21 
 
 -13.43 
 
 25.2 
 
 4.26 
 
 0.25 
 
 5.49 
 
 7 
 
 19.7 
 
 0.092 
 
 0.424 
 
 7.61 
 
 -17.17 
 
 -10.14 
 
 30.0 
 
 3.85 
 
 0.42 
 
 6.03 
 
 9 
 
 22.9 
 
 0.064 
 
 0.580 
 
 5.29 
 
 -23.49 
 
 - 4.70 
 
 27.0 
 
 2.68 
 
 0.58 
 
 0.84 
 
 11 
 
 25.0 
 
 0.014 
 
 0.692 
 
 1.16 
 
 -28.03 
 
 + 1.87 
 
 23.8 
 
 0.59 
 
 0.69 
 
 -2.48 
 
 13 
 
 25.5 
 
 -0.044 
 
 0.720 
 
 - 3.64 
 
 -29.16 
 
 7.30 
 
 20.2 
 
 -1.84 
 
 0.72 
 
 -4.18 
 
 15 
 
 24.5 
 
 -0.100 
 
 0.652 
 
 - 8.27 
 
 -26.41 
 
 10.18 
 
 17.0 
 
 -4.18 
 
 0.65 
 
 -3.97 
 
 17 
 
 22.1 
 
 -0.132 
 
 0.520 
 
 -10.92 
 
 -21.06 
 
 9.88 
 
 14.0 
 
 -5.52 
 
 0.52 
 
 -3.10 
 
 19 
 
 18.7 
 
 -0.148 
 
 0.360 
 
 -12.24 
 
 -14.58 
 
 8.12 
 
 11.0 
 
 -6.18 
 
 0.36 
 
 -1.88 
 
 21 
 
 14.3 
 
 -0.136 
 
 0.196 
 
 -11.25 
 
 - 7.94 
 
 4.89 
 
 7.8 
 
 -5.68 
 
 0.20 
 
 -1.02 
 
 23 
 
 9.1 
 
 -0.100 
 
 0.070 
 
 - 8.27 
 
 - 2.84 
 
 2.01 
 
 4.8 
 
 -4.18 
 
 0.07 
 
 -0.19 
 
 25 
 
 3.9 
 
 -0.046 
 
 0.016 
 
 - 3.80 
 
 - 0.65 
 
 0.55 
 
 2.0 
 
 -1.92 
 
 0.02 
 
 0.00 
 
 All ordinates in above 
 
 It is seen that the resultant polygon intersects the axial line four times, showing con- 
 sistency with Professor Winkler's theorem. 
 
 As a final check, the normal thrusts TV, offsets v and stresses /are computed in Table 
 79n for the same case of loading; used in Table 79Q, and the close agreement of the results 
 shows both solutions to be satisfactory. The only stresses, however, which are maximum, 
 are those on the section A, the others are simply for the simultaneous case of loading 
 and have no special interest. 
 
 The stresses on the haunch section A are thus found to be/ e = 549 and/ t -= +152.9 
 
ART. 79 
 
 FIXED MASONRY ARCHES 
 
 355 
 
 EXTRADOS KERNEL POINTS UNSYMMETRIC ARCH 
 
 SECTION n. 
 
 , = ,- [, a +12.23,c] 
 
 SECTION m'. 
 
 SECTION B. 
 
 Point. 
 
 r/o 
 
 12.23 7j c 
 
 r/n 
 
 VO 
 
 -43.2,6 
 
 -.098, c 
 
 T)m 
 
 fjo 
 
 -87.0,6 
 
 -59.56, c 
 
 r ,B 
 
 3.7 
 
 0.14 
 
 + 0.05 
 
 1.9 
 
 -1.56 
 
 -0.01 
 
 -0.03 
 
 0.18 
 
 - 3.13 
 
 - 0.71 
 
 0.52 
 
 1 
 
 10.6 
 
 1.22 
 
 -0.42 
 
 5.5 
 
 -3.54 
 
 -0.10 
 
 -0.66 
 
 0.54 
 
 - 7.13 
 
 - 5.96 
 
 3.83 
 
 3 
 
 17.2 
 
 3.08 
 
 -1.08 
 
 9.1 
 
 -4.41 
 
 -0.25 
 
 -1.44 
 
 0.89 
 
 - 8.88 
 
 -15.01 
 
 9.58 
 
 5 
 
 24.2 
 
 5.19 
 
 -0.69 
 
 13.0 
 
 -3.97 
 
 -0.41 
 
 -2.32 
 
 1.24 
 
 - 8.00 
 
 -25.25 
 
 14.79 
 
 7 
 
 31.1 
 
 7.10 
 
 + 1.10 
 
 16.4 
 
 -2.76 
 
 -0.57 
 
 -3.17 
 
 1.59 
 
 - 5.57 
 
 -34.54 
 
 18.80 
 
 9 
 
 38.5 
 
 8.46 
 
 5.04 
 
 20.3 
 
 -0.60 
 
 -0.68 
 
 -3.42 
 
 1.99 
 
 - 1.22 
 
 -41.22 
 
 19.43 
 
 11 
 
 40.8 
 
 8.80 
 
 6.50 
 
 24.6 
 
 + 1.90 
 
 -0.71 
 
 -2.09 
 
 2.38 
 
 + 3.83 
 
 -42.88 
 
 15.93 
 
 13 
 
 34.2 
 
 7.97 
 
 + 1.73 
 
 28.4 
 
 4.32 
 
 -0.64 
 
 + 0.22 
 
 2.76 
 
 8.70 
 
 -38.83 
 
 8.39 
 
 15 
 
 28.1 
 
 6.36 
 
 -0.36 
 
 32.0 
 
 5.70 
 
 -0.51 
 
 4.71 
 
 3.10 
 
 11.48 
 
 -30.97 
 
 0.49 
 
 17 
 
 22.0 
 
 4.40 
 
 -1.10 
 
 33.4 
 
 6.39 
 
 -0.35 
 
 8.66 
 
 3.47 
 
 12.88 
 
 -21.44 
 
 - 6.67 
 
 19 
 
 15.6 
 
 2.40 
 
 -1.10 
 
 23.7 
 
 5.88 
 
 -0.20 
 
 3.72 
 
 3.82 
 
 11.83 
 
 -11.67 
 
 -10.64 
 
 21 
 
 9.5 
 
 0.85 
 
 -0.45 
 
 14.4 
 
 4.32 
 
 -0.07 
 
 + 1.05 
 
 4.17 
 
 8.70 
 
 - 4.17 
 
 - 9.46 
 
 23 
 
 4.0 
 
 0.20 
 
 -0.10 
 
 5.8 
 
 1.99 
 
 -0.02 
 
 -0.07 
 
 4.50 
 
 4.00 
 
 - 0.95 
 
 - 2.45 
 
 25 
 
 table are expressed in feet. 
 
 INTRADOS KERNEL POINTS UNSYMMETRIC ARCH 
 
 SECTION n. 
 'Jn= >?o [)Ja+ 14.23^ c ] 
 
 SECTION m'. 
 5}m /==) ?o -[Jjo 44.6i;6 + 0.987j c ] 
 
 SECTION B. 
 ?B = 0-[?-9l.7ij6-58.14ij c ] 
 
 Point. 
 
 rio 
 
 14.23ij c 
 
 Tjn 
 
 90 
 
 -44.67?6 
 
 0.98)? c 
 
 Tim 
 
 -91.7>j6 
 
 -58.14/jc 
 
 IB 
 
 3.7 
 
 0.17 
 
 + 0.03 
 
 1.8 
 
 -1.61 
 
 0.01 
 
 -0.09 
 
 -3.30 
 
 - 0.70 
 
 + 0.50 
 
 1 
 
 10.6 
 
 1.42 
 
 -0.62 
 
 5.4 
 
 -3.66 
 
 0.10 
 
 -0.84 
 
 -7.52 
 
 - 5.81 
 
 3.53 
 
 3 
 
 17.2 
 
 3.58 
 
 -1.58 
 
 8.9 
 
 -4.55 
 
 0.25 
 
 -2.00 
 
 -9.35 
 
 -14.65 
 
 8.80 
 
 5 
 
 24.2 
 
 6.03 
 
 - 1 . 53 
 
 12.4 
 
 -4.10 
 
 0.41 
 
 -3.61 
 
 -8.44 
 
 -24.65 
 
 13.39 
 
 7 
 
 31.1 
 
 8.25 
 
 -0.05 
 
 16.0 
 
 -2.85 
 
 0.57 
 
 -4.62 
 
 -5.87 
 
 -33.72 
 
 16.69 
 
 9 
 
 38.5 
 
 9.85 
 
 + 3.65 
 
 19.6 
 
 -0.62 
 
 0.68 
 
 -5.46 
 
 -1.28 
 
 -40.23 
 
 16.51 
 
 11 
 
 40.8 
 
 10.24 
 
 5.06 
 
 23.8 
 
 + 1.96 
 
 0.71 
 
 -4.37 
 
 + 4.03 
 
 -41.86 
 
 12.33 
 
 13 
 
 34.2 
 
 9.28 
 
 0.42 
 
 27.8 
 
 4.46 
 
 0.64 
 
 -1.80 
 
 9.17 
 
 -37.91 
 
 4.24 
 
 15 
 
 28.1 
 
 7.40 
 
 -1.40 
 
 31.2 
 
 5.88 
 
 0.51 
 
 + 2.71 
 
 12.10 
 
 -30.23 
 
 - 3.97 
 
 17 
 
 22.0 
 
 5.12 
 
 -1.82 
 
 33.4 
 
 6.60 
 
 0.35 
 
 7.75 
 
 13.57 
 
 -20.93 
 
 -11.34 
 
 19 
 
 15.6 
 
 2.84 
 
 -1.54 
 
 24.2 
 
 6.06 
 
 0.20 
 
 3.64 
 
 12.47 
 
 -11.40 
 
 -15.37 
 
 21 
 
 9.5 
 
 0.99 
 
 -0.59 
 
 14.6 
 
 4.46 
 
 0.07 
 
 0.97 
 
 9.17 
 
 - 4.07 
 
 -14.20 
 
 23 
 
 4.0 
 
 0.23 
 
 -0.13 
 
 6.0 
 
 2.05 
 
 0.02 
 
 0.03 
 
 4.22 
 
 - 0.93 
 
 - 7.19 
 
 25 
 
 table are expressed in feet. 
 
 cu.ft., which are quite excessive as compared to those previously obtained for section 
 t of the symmetric arch, where f e = 440 and J\ = +28.5 cu.ft. 
 
 If the investigation of stresses were extended to all the critical sections it would 
 be found that the dimensions found safe for the symmetric arch are now quite insufficient 
 when the arch is treated as an unsymmetric structure using the same axial line as before. 
 The shape of the ring might be so adjusted as to reduce the stresses within the required 
 limits by computing a resultant polygon for the average loading Q+%P, and shifting 
 the axial line by the amounts v obtained for the several critical sections. 
 
356 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES CHAP. XV 
 
 A 
 
 ^^rrr^r^ SECTION n. ^r^rrri' 
 
 rf rl'"-T T^jNa ip n la is K i5/fcJ"'T T 
 
 m 5*7 8S X ] T / T>I7 18 19 ; 
 
 V 
 
 intrados kernel points. \ 
 
 extrados ke 
 
 rnel points. \ //uml 
 \ / '~~^ 
 
 / LENGTHS. 
 
 / l...iliml I I I I l| 
 
 / 10 O |O ZO 3O 4O SO PT. 
 
 OROINATCS. 
 
 III! 
 
 SFT. 
 
 v' 
 FIG. 79K. M Influence Lines for the Kernel Points. Unsymmetric Arch. 
 
ART. 79 
 
 FIXED MASONRY ARCHES 
 
 357 
 
 TABLE 79q. 
 THE RESULTANT POLYGON FOR LOADS Q + P UNSYMMETRIC ARCH 
 
 
 
 
 
 Dead 
 
 Live 
 
 Total 
 
 r,(Q + P) for 
 
 Moments about a 
 
 Point. 
 
 r ia 
 
 16 
 
 r, c 
 
 Loads. 
 
 Loads. 
 
 Loads. 
 
 
 
 
 
 
 
 
 
 
 
 
 Q 
 
 P 
 
 
 x a 
 
 x b 
 
 X c 
 
 r 
 
 r(Q+P)' 
 
 
 ft. 
 
 ft. 
 
 ft. 
 
 cu.ft. 
 
 cu.ft. 
 
 cu.ft. 
 
 
 
 
 ft. 
 
 
 1 
 
 3.5 
 
 0.036 
 
 0.012 
 
 367.4 
 
 0.0 
 
 367.4 
 
 1285.9 
 
 13.23 
 
 4.40 
 
 5.9 
 
 2167.7 
 
 3 
 
 9.8 
 
 0.082 
 
 0.100 
 
 276.1 
 
 0.0 
 
 276.1 
 
 2705.8 
 
 22.64 
 
 27.61 
 
 18.9 
 
 5218.3 
 
 5 
 
 15.2 
 
 0.102 
 
 0.252 
 
 223 .4 
 
 0.0 
 
 223.4 
 
 3395.7 
 
 22.82 
 
 56.30 
 
 31.9 
 
 7126.5 
 
 7 
 
 19.7 
 
 0.092 
 
 0.424 
 
 182.7 
 
 0.0 
 
 182.7 
 
 3599.2 
 
 16.81 
 
 77.46 
 
 44.9 
 
 8203.2 
 
 9 
 
 22.9 
 
 0.064 
 
 0.580 
 
 168.0 
 
 0.0 
 
 168.0 
 
 3847.2 
 
 10.75 
 
 97.44 
 
 57.9 
 
 9727.2 
 
 11 
 
 25.0 
 
 0.014 
 
 0.692 
 
 173.7 
 
 114.6 
 
 288.3 
 
 7207 . 5 
 
 4.03 
 
 199 . 50 
 
 71.9 
 
 20728.8 
 
 13 
 
 25.5 
 
 -0.044 
 
 0.720 
 
 173.7 
 
 114.6 
 
 288.3 
 
 7344.8 
 
 -12.68 
 
 207.59 
 
 86.9 
 
 25053.3 
 
 15 
 
 24.5 
 
 -0.100 
 
 0.652 
 
 168.0 
 
 99.3 
 
 267.3 
 
 6548.9 
 
 -26.73 
 
 174.28 
 
 100.9 
 
 26970.6 
 
 17 
 
 22.1 
 
 -0.132 
 
 0.520 
 
 182.7 
 
 99.3 
 
 282.0 
 
 6232.2 
 
 -37.22 
 
 146.64 
 
 113.9 
 
 32119.8 
 
 19 
 
 18.7 
 
 -0.148 
 
 0.360 
 
 223.4 
 
 99.3 
 
 322.7 
 
 6034.5 
 
 -47.76 
 
 116.17 
 
 126.9 
 
 40950.6 
 
 21 
 
 14.3 
 
 -0.136 
 
 0.196 
 
 276.1 
 
 99.3 
 
 375.4 
 
 5368.2 
 
 -51.05 
 
 73.58 
 
 139.9 
 
 52518.5 
 
 23 
 
 9.1 
 
 -0.100 
 
 0.070 
 
 367.4 
 
 99.3 
 
 466.7 
 
 4247.0 
 
 -46.67 
 
 32.67 
 
 152.9 
 
 71358.4 
 
 25 
 
 3.5 
 
 -0.040 
 
 0.012 
 
 680.0 
 
 84.0 
 
 764.0 
 
 2674.0 
 
 -30.56 
 
 9.17 
 
 165.9 
 
 126747.6 
 
 
 
 
 
 
 
 4272.3 
 
 60490.9 
 
 -162.39 
 
 1222.81 
 
 
 428890.5 
 
 
 
 
 
 
 
 = # 
 
 = X a 
 
 =x b 
 
 =X C 
 
 
 = M 
 
 Z = 174'.38, r, = 82'.7, / 
 Then H = X c cos /?= 1221.22 cu.ft, 
 
 TT 
 
 H' = = 1242 cu.ft. 
 
 cos a 
 
 I 
 
 c = -X&=-23.19ft. 
 n 
 
 M 
 
 r = 7r= 100.388 ft. 
 ti 
 
 tan 13= -0.0519, cos/3 = 0.9987 
 
 V 
 
 tan a = tan 
 
 -77 
 
 1 
 
 = -0.1849 
 
 = = 49.53 ft. 
 
 -^Xi-- 11.00 ft. 
 
 H 
 
 B = 
 
 =2494.6 cu.ft. 
 
 A = R B = 1777 7 cu.ft. 
 
 The investigation for temperature is not repeated here except to evaluate the moment 
 Eq. (75c) for this case using the same data just given for the symmetric arch. 
 
 For cos,/? =0.9987 span 1 = 174.4 ft., and 2t/ W c = 1482.6 from Table 79M, then 
 
 etl = -0.0752 ft. and 
 _0.0752X3,775,720 
 
 from which the temperature stresses for any section are readily obtained by computing 
 the moments about the two kernel points of the section. 
 
 A complete investigation of the stresses in aiiy arch would then consist in finding 
 the kernel moments due to the dead loads, the kernel moments due to the train of con- 
 centrated live loads, Fig. 79o, making proper allowance for impact; and finally the kernel 
 moments due to temperature effects; all for each of the five critical sections. These 
 moments for the same section are algebraically combined into the total M e and M t - moments 
 and from them and Eqs. (76c) the maximum stresses are found. 
 
358 
 
 KINETIC THEORY OF ENGINEERING STRUCTURES 
 
 CHAT. XV 
 
 TABLE 79E 
 
 COMPUTATION OF MOMENTS AND STRESSES ON THE CRITICAL SECTIONS 
 FOR LOADS Q+P UNSYMMETRIC ARCH 
 
 
 
 SECTION A. 
 
 SECTION m. 
 
 SECTION n. 
 
 Point. 
 
 Loads 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 Q + P 
 
 fie 
 
 m 
 
 Me 
 
 Mi 
 
 i?e 
 
 ft 
 
 Me 
 
 Mi 
 
 to 
 
 w 
 
 M e 
 
 *< 
 
 
 cu.ft. 
 
 ft. 
 
 ft. 
 
 
 
 ft. 
 
 ft. 
 
 
 
 ft. 
 
 ft. 
 
 
 
 1 
 
 367.4 
 
 - 3.12 
 
 - 5.99 
 
 -1146.3 
 
 -2200.7 
 
 0.25 
 
 0.39 
 
 91.9 
 
 143.3 
 
 + 05 
 
 + 0.03 
 
 + 18.4 
 
 + 11.0 
 
 3 
 
 276.1 
 
 - 9.61 
 
 -12.53 
 
 -2653.3 
 
 -3459.5 
 
 2.17 
 
 2.07 
 
 599.1 
 
 671.5 
 
 -0.42 
 
 -0.62 
 
 - 116.0 
 
 - 171.2 
 
 5 
 
 223.4 
 
 -10.40 
 
 -13.43 
 
 -2323.4 
 
 -3000.3 
 
 6.01 
 
 5.49 
 
 1342.6 
 
 1226.5 
 
 -1.08 
 
 -1.52 
 
 - 241.3 
 
 - 339.6 
 
 7 
 
 182.7 
 
 - 7.06 
 
 -10.14 
 
 -1289.9 
 
 -1852.6 
 
 7.98 
 
 6.03 
 
 1457.9 
 
 1101.7 
 
 ^0.69 
 
 -1.53 
 
 - 125.1 
 
 - 279.5 
 
 9 
 
 168.0 
 
 - 1.65 
 
 -4.70 
 
 - 277.2 
 
 - 789.6 
 
 3.08 
 
 0.84 
 
 517.4 
 
 141.1 
 
 4-1.10 
 
 -0.05 
 
 + 184.8 
 
 - 8.4 
 
 11 
 
 288.3 
 
 + 4.72 
 
 + 1.87 
 
 + 1360.8 
 
 + 539.1 
 
 -0.38 
 
 -2.48 
 
 - 109.5 
 
 - 715.0 
 
 5.04 
 
 + 3.65 
 
 1453.0 
 
 + 1052.3 
 
 13 
 
 288.3 
 
 9.79 
 
 7.30 
 
 2822 . 4 
 
 2104.6 
 
 -1.99 
 
 -4.18 
 
 - 573.7 
 
 -1205.1 
 
 6.50 
 
 5.06 
 
 1874 . 1 
 
 1458. S 
 
 15 
 
 267.3 
 
 12.18 
 
 10.18 
 
 3255.7 
 
 2721 . 1 
 
 -2.39 
 
 -3.97 
 
 - 638.8 
 
 -1051.2 
 
 -1-1.73 
 
 -1-0.42 
 
 -1- 462.4 
 
 + 112.3 
 
 17 
 
 282.0 
 
 11.35 
 
 9.88 
 
 3200 . 7 
 
 2786.2 
 
 -2.02 
 
 3.10 
 
 - 569.6 
 
 - 874.2 
 
 -0.36 
 
 -1.40 
 
 - 101.5 
 
 394.8 
 
 19 
 
 322.7 
 
 8.86 
 
 8.12 
 
 2859 . 1 
 
 2620 . 3 
 
 -1.23 
 
 -1.88 
 
 - 396.9 
 
 - 606.7 
 
 -1.10 
 
 -1.82 
 
 355.0 
 
 587.3 
 
 21 
 
 375.4 
 
 5.36 
 
 4.89 
 
 2012.1 
 
 1835.7 
 
 -0.58 
 
 -1.02 
 
 - 217.7 
 
 - 382.9 
 
 -1.10 
 
 -1.54 
 
 - 412.9 
 
 - 578.1 
 
 23 
 
 466.7 
 
 2.16 
 
 2.01 
 
 1008 . 1 
 
 938.1 
 
 -0.17 
 
 -0.19 
 
 - 79.4 
 
 - 88.7 
 
 -0.45 
 
 -0.59 
 
 - 210.0 
 
 - 275.4 
 
 25* 
 
 764.0 
 
 0.42 
 
 0.40 
 
 320.9 
 
 305.6 
 
 -0.01 
 
 0.00 
 
 7.6 
 
 - 0.0 
 
 -0.08 
 
 -0.11 
 
 - 61.1 
 
 - 84.0 
 
 Totals 
 
 4272.3 
 
 D= 
 
 10'. 
 
 + 9149.7 
 
 + 2548.0 
 
 D= 
 
 6'.90 
 
 + 1415.7 
 
 -1739.7 
 
 D= 
 
 6'. 00 
 
 + 2369.8 
 
 - S3.S 
 
 
 
 
 N= 
 
 1980.5 
 
 
 
 N= 
 
 1372.0 
 
 
 
 A" = 
 
 1226.9 
 
 
 
 
 
 v = 
 
 + 2'.953 
 
 
 
 v = 
 
 -OM19 
 
 
 
 v = 
 
 -f 0'.932 
 
 
 
 
 
 f 
 
 - 549.0 
 
 + 152.9 
 
 
 /= 
 
 - 178.4 
 
 - 219.7 
 
 
 f 
 
 - 395.0 
 
 - 14.0 
 
 
 
 SECTION in'. 
 
 SECTION B. 
 
 Point 
 
 Loads. 
 
 
 
 
 
 
 
 
 
 
 Q + P 
 
 T ie 
 
 9 
 
 M e 
 
 Mi 
 
 1e 
 
 fc 
 
 M e 
 
 tfi 
 
 
 cu ft. 
 
 ft. 
 
 ft. 
 
 
 
 ft. 
 
 ft. 
 
 
 
 1 
 
 367.4 
 
 -0.03 
 
 -0.09 
 
 11.0 
 
 - 33.0 
 
 0.52 
 
 0.50 
 
 191.0 
 
 183.7 
 
 3 
 
 276.1 
 
 -0.66 
 
 -0.84 
 
 - 182.2 
 
 - 231.9 
 
 3.83 
 
 3.53 
 
 1057 . 5 
 
 974.6 
 
 5 
 
 223.4 
 
 -1.44 
 
 -2.00 
 
 - 321.7 
 
 - 446.8 
 
 9.58 
 
 8.80 
 
 2140.2 
 
 1965.9 
 
 7 
 
 182.7 
 
 -2.32 
 
 -3.61 
 
 - 323.9 
 
 - 659.5 
 
 14.79 
 
 13.39 
 
 2702.0 
 
 2446.4 
 
 9 
 
 168.0 
 
 -3.17 
 
 -4.62 
 
 - 532.6 
 
 - 776.2 
 
 18.80 
 
 16.69 
 
 3158.4 
 
 2803.9 
 
 11 
 
 288.3 
 
 -3.42 
 
 -5.46 
 
 - 986.0 
 
 - 1574 . 1 
 
 19.43 
 
 16.51 
 
 5601.7 
 
 4759.8 
 
 13 
 
 288.3 
 
 -2.09 
 
 -4.37 
 
 - 602.5 
 
 -1259.9 
 
 15.93 
 
 12.33 
 
 4592.6 
 
 3554.7 
 
 15 
 
 267.3 
 
 + 0.22 
 
 -1.80 
 
 + 58.7 
 
 - 481.1 
 
 8:39 
 
 4.24 
 
 2242 . 6 
 
 1133.4 
 
 17 
 
 282.0 
 
 4.71 
 
 + 2.71 
 
 1328.2 
 
 + 764.2 
 
 0.49 
 
 - 3.97 
 
 138.2 
 
 -1119.5 
 
 19 
 
 322.7 
 
 8.66 
 
 7.75 
 
 2794.6 
 
 2500.9 
 
 - 6.67 
 
 -11.34 
 
 -2152.4 
 
 -3659.4 
 
 21 
 
 375.4 
 
 3.72 
 
 3.64 
 
 1396.5 
 
 1365.4 
 
 -10.64 
 
 -15.37 
 
 -3994.3 
 
 -5769.9 
 
 23 
 
 466.7 
 
 + 1.05 
 
 0.97 
 
 + 490.1 
 
 452.7 
 
 - 9.46 
 
 -14.20 
 
 -4415.0 
 
 - 6627 . 1 
 
 25i 
 
 764.0 
 
 -0.05 
 
 0.02 
 
 - 38. 2 
 
 15.3 
 
 - 1.76 
 
 - 6.80 
 
 -1344.6 
 
 -5195.2 
 
 Totals 
 
 4272.3 
 
 D = 
 
 7'. 20 
 
 + 3070.0 
 
 - 374.0 
 
 D = 
 
 14'. 7 
 
 + 9917.9 
 
 -4548.7 
 
 
 
 
 N= 
 
 1435.0 
 
 
 
 ;V = 
 
 2952.4 
 
 
 
 
 
 V 
 
 + 940 
 
 
 
 V 
 
 + 0' . 909 
 
 
 
 
 
 /= 
 
 - 355.3 
 
 43.3 
 
 
 /- 
 
 - 274.7 
 
 - 126. C 
 
 Eqs. (76c, 
 
 3 
 
 -Jj { 
 
 
 D 2 
 
 v is positive when measured 
 
 from the axis toward the extrados. All loads in above table are expressed in cubic feet of masonry 
 and the stresses are cubic feet per square foot. Cubic feet per square foot X 0.972 = pounds per square 
 inch. 
 
BIBLIOGRAPHY 
 
 THEORY OF ELASTICITY 
 
 HOOKE. Philosophical tracts and collections, 1678. Hooke's law of proportionality between 
 
 stress and strain. 
 
 BERNOULLI, 1667-1748, was the first to give an analytic solution for the elastic curve. 
 EULER. Methodus invemendi hneas curvas, 1741-4. 
 COULOMB. Memoires de Pacademie des Sciences, 1784. 
 LAGRANGE. Mecamque analytique, Paris, 1788. 
 YOUNG. A course of lectures on natural philosophy and mechanical arts, 1807 Gives Young 3 
 
 modulus of elasticity E. 
 NAVIER. Memoire sur la flexion des verges elastiques courbes, 1827. Law of distribution of stress 
 
 on any section subjected to bending. 
 
 POISSON. Memoire de 1'Academie des Sciences, VIII, 1829. 
 CLAPEYRON. Sur 1 equilibre interieur des corps solides homogcnes. Memoires des savants etrangers, 
 
 IV, 1833. 
 
 LAME. Lemons sur la theorie mathematique de 1'elasticite, 1852. 
 DE SAINT-VENANT. Memoires des savants etrangers, XIV, 1856. 
 MENABREA. Nouveau principe sur la distribution des tensions dans les systemes elastiques 
 
 Comptes rendus, 1858. 
 
 CASTIGLIANO. Theorie de 1'equilibre des systemes elastiques, 1879. 
 FRAENKEL. Das Prinzip der kleinsten Arbeit, etc., Zeitsch. des Arch.- und Ing.-Vereins, Hannover, 
 
 1882. 
 MOHR. Ueber die Darstellung des Spannungszustandes und des Deformationszustandes eines 
 
 Koerperelementes. Civilingenieur, 1882. 
 MOHR. Ueber die Elastizitaet der Deformationsarbeit. Zeitschr. des Arch.- und Ing.-Vereins, 
 
 Hannover, 1886. 
 MOHR. Welche Umstaende bedingen die Elastizitaetsgrenze und den Bruch eines Materials. 
 
 Zeitshcr. des Verems deutscher Ing., 1900. 
 
 GRAPHOSTATICS 
 
 PONCELET. Cours de Mecanique industrielle, 1826-9. 
 CULMANN. Graphische Statik, 1866. 
 
 CLERK MAXWELL. On reciprocal figures and diagrams of forces. Phil. Mag., 1864. 
 WINKLER. Vortrag ueber die Berechnung von Bogenbruecken, Mitth. des Arch.- und Ing.-Vereins, 
 Boehmen, 1868. Introduces influence lines. 
 
360 BIBLIOGRAPHY 
 
 MOHR. Beitrag zur Theorie der Holz und Eisenkonstruktionen. Zeitschr. des Arch.- und Ing. 
 
 Vereins, Hannover, 1868. Introduces influence lines. 
 CREMONA. Le figure reciproche nella Statica Grafica, 1872. 
 MOHR. Die graphische Statik und das graphische Rechnen. Civiling., 1875. 
 WILLIOT. Notions pratiques sur la statique graphique. Genie Civil, Oct., 1877. 
 MUELLER-BRESLAU. Die graphische Statik der Baukonstruktionen, 1881-1905. 
 MOHR. Ueber Geschwindigkeitsplane und Beschleunigungsplane. Civiling., 1887. 
 W. HITTER. Anwendungen der graphischen Statik, 1890-1906. 
 
 GENERAL TREATISES 
 
 SCHWEDLER. Theorie der Brueckenbalkensysteme. Zeitschr. fuer Bauwesen, 1851. 
 A. RITTER. Elementare Theorie und Berechnung eiserner Dach- und Brueckenkonstruktionen, 1863. 
 CLERK MAXWELL. On the calculation of the equilibrium and stiffness of frames. Phil. Mag., 1864. 
 MOHR. Beitrag zur Theorie des Fachwerks. Zeitschr. des Arch.- und Ing.-Vereins, Hannover, 
 
 1870, 1874, 1875, 1877. 
 
 WEYRAUCH. Theorie und Berechnung der Kontinuierlichen und einfachen Traeger, 1873. 
 WINKLER. Vortraege ueber Brueckenbau, 1873-1884. 
 FRAENKEL. Anwendung der Theorie des augenblicklichen Drehpunktes auf die Bestimmung der 
 
 Formaenderung von Fachwerken. Civiling., 1875. 
 FRAENKEL. Ueber die unguenstigste Einstellung eines Systemes von Einzellasten auf Fachwerk- 
 
 traeger mit Hilfe von Influenzkurven. Civiling., 1876. 
 FOEPPL. Theorie des Fachwerks, 1880. 
 
 MUELLER-BRESLAU. Graphische Statik der Baukonstruktionen. 1881-1908. 
 SWAIN. On the application of the principle of virtual velocities to the determination of the deflec- 
 tion and stresses of frames. Jour. Franklin Inst., 1883. 
 MELAN. Beitrag zur Berechung statischunbestimmter Stabsysteme. Zeitschr., des Oestr. Arch. 
 
 und Ing. Vereins, 1884. 
 LAND. Die Gegenseitigkeit elastischer Formaenderungen, etc. Wochenblatt fuer Baukunde, Jan., 
 
 1887. 
 LAND. Beitrag zur Ermittelung der Biegungslinien ebener und elastischer Stabwerke. Civiling., 
 
 1889. 
 
 L. v. TETMAJER. Angewandte Elastizitaets und Festigkeitslehre. 1888, 1902. 
 MUELLER-BRESLAU. Festigkeitslehre, 1893. 
 LANDSBERG. Beitrag zur Theorie des raemulichen Fachwerks. Zentr. der Bauverwaltung. 1898 
 
 and 1903. 
 
 MEHRTENS. Statik der Baukonstruktionen. 3 vol., 1903-5. 
 MOHR. Abhandlungen aus dem Gebiete der technischen Mechanik. 1905. 
 KECK-HOTOPP. Elastizitaetslehre, 1905. 
 Handbuch der Ingenieurwissenschaften. Der Brueckenbau. 
 
 SECONDARY AND ADDITIONAL STRESSES 
 
 ENGESSER. Ueber die Druchbiegung von Fachwerktraegern und die hierbei auftretenden zusaetz- 
 
 lichen Spannungen. Zeitschr. fuer Baukunde, 1879. 
 
 ASIMONT. Hauptspannung und Sekundaerspannung. Zeitschr. fuer Baukunde, 1880. 
 MANDERLA. Die Berechnung der Sekundaerspannungen, etc., Allgemeine Bauzeitung, 1880. 
 MUELLER-BRESLAU. Ueber Biegungsspannungen in Fachwerken. Allgemeine Bauzeitung, 1885. 
 
BIBLIOGRAPHY 361 
 
 LANDSBERG. Ebene Fachwerksysteme mit festen Knotenpunkten, etc. Centralb. der' Bauver- 
 waltung, 1885. 
 
 LANUSBERG. Beitrag zur Theorie des Fachwerks. Zeitschr. des Arch.- und Ing.-Vereins, Han- 
 nover, 1885 and 1886. 
 
 MUELLER-BRESLAU. Zur Theorie der Biegungsspannungen in Fachwerktraegern. Zeitschr. des 
 Arch.- und Ing.-Vereins, Hannover, 1886. 
 
 WINKLER. Querkonstruktionen. Zeitschr. des Arch.- und Ing.-V., Hannover, 1886. 
 
 Handbuch der Ingenieurwissenschaften, Vol. II, 1890. 
 
 MOHR. Die Berechnung der Fachwerke mit starren Knotenverbindungen. Civiling., 1891 and 1892. 
 
 ENGESSER. Die Zusatzkraefte und Nebenspannungen eiserner Fachwerke, 1892-3. 
 
 HIROI. Statically Indeterminate Bridge Stresses, 1905. 
 
 MEHRTENS. Statik der Baukonstruktionen, Vol. Ill, 1905. 
 
 MOHR. Abhandlungen aus dem Gebiete der technischen Mechanik, 1906. 
 
 GRIMM. Secondary Stresses in Bridge Trusses, 1908. 
 
 SPECIAL TREATISES ON ARCHES 
 
 LAME and CLAPEYRON. Journal des Voies de Communication, 1826. This contains the first applica- 
 tion of force and equilibrium polygons to fixed arches. 
 
 PONCELET. Memorial de 1'officier du genie, 1835. 
 
 GERSTNER. Handbuch der Mechanik, 1831. Line of thrust method. 
 
 MOSELEY. Phil. Mag., 1833. Discussion of most probable position of line of thrust. 
 
 HAGEN. Ueber Form und Staerke gewoelbter Bogen, 1844. 
 
 SCHWEDLER. Theorie der Stuetzlinie. Zeitschr. fuer Bauwesen, 1859. 
 
 CULMANN. Die graphische Statik, 1866. 
 
 FRAENKEL. Berechnung eiserner Bogenbruecken. Civiling., 1867, 1875. 
 
 MOHR. Beitrag zur Theorie der elastischen Bogentraeger. Zeitschr. des Arch.- und Ing.-Ver.. 
 Hannover, 1870, 1874, 1881. 
 
 WINKLER. Beitrag zur Theorie der elastischen Bogentraeger. Zeitschr. des Arch.- und Ing.-Ver., 
 Hannover, 1879. 
 
 WEYRAUCH. Theorie der elastischen Bogentraeger. Zeitschr. fuer Baukunde, 1878. 
 
 ENGESSER. Theorie und Berechnung der Bogenfachwerktraeger, 1880. 
 
 MUELLER-BRESLAU. Theorie und Berechnung der eiserner Bogenbruecken, 1880. 
 
 W. RITTER. Der elastische Bogen berechnet mit Hilfe der graphischen Statik, 1886. 
 
 MUELLER-BRESLAU. Graphische Statik der Baukonstruktionen, 1892, 1907. 
 
 WEYRAUCH. Die elastischen Bogentraeger, 1897. 
 
 TOLKMITT. Leitfaden fuer das Entwerfen gewoelbter Bruecken, 1895. 
 
 MEHRTENS. Statik der Baukonstruktionen, 3 vols., 1903-5. 
 
 Handbuch der Ingenieurwissenschaften, Vol. I, 1904. 
 
INDEX 
 
 Abutment displacements, fixed framed arches . 188 
 for general case of re- 
 dundancy 208 
 
 in two-hinged arches, 
 
 156, 165, 168 
 
 stresses due to 32 
 
 Additional stresses due to dynamic influences . 256 
 
 Arches, bibliography of special treatises on 361 
 
 fixed framed 178, 197 
 
 fixed solid web or masonry 298 
 
 three-hinged 72,. 78 
 
 two-hinged 153, 166, 176 
 
 without hinges 178, 298 
 
 B 
 
 Beams, deflection of 117, 118, 123, 125, 126 
 
 work of deformation for 41 
 
 Betti's law 28 
 
 Bibliography of treatises, etc 359 
 
 Cantilever bridges by influence lines 67 
 
 Castigliano's law, derivative of work equation . 30, 31 
 
 Centrifugal force due to curved track 260 
 
 Changes in the angles of a triangle due to 
 
 stress 107 
 
 Choice of the redundant conditions 202 
 
 Clapeyron's law 2, 14 
 
 Column formulae 270 
 
 Combination of stresses as a basis for design- 
 ing 268 
 
 Continuous girder over four supports 130 
 
 three supports. 129, 143, 148 
 Critical loading for max. moments, direct 
 
 loading 59 
 
 Critical loading for max. moments, indirect 
 
 loading 60 
 
 Critical loading for max. shear, direct loading. . 62 
 
 , indirect loading 62 
 
 D 
 
 PAGE 
 
 Dead-load stresses in determinate structures . . 210 
 Definitions of terms used throughout this work . vii 
 
 Deflection influence lines 49 
 
 of a solid w r eb beam 117 
 
 polygons according to Prof. Laud . . 107 
 polygons according to Prof. Mohr, 
 
 100, 104 
 polygons for determinate structures. 99 
 
 polygons for the loaded chord 104 
 
 Deformations 11 
 
 Deformation, work due to 1, 29 
 
 Derivative of the work equation; Castigliano's 
 
 law... .\ 30, 31 
 
 Designing, combination of stresses as a basis 
 
 for 268 
 
 Determinate structures, dead load stresses in .. 210 
 displacements of 
 
 points 18 
 
 displacement influence 
 
 lines for 122 
 
 influence lines for. 46, 65 
 
 stresses in 210, 218 
 
 Direct and indirect loading 47, 48, 50 
 
 Displacement influence lines, determinate 
 
 structures 122 
 
 for a cantilever. . 124 
 for a simple beam 
 ortruss.123,125,126 
 
 Displacements, horizontal 114 
 
 of points for determinate struc- 
 tures 18 
 
 Distortion of a statically determinate frame by 
 
 graphics. 
 
 87 
 
 Distortions due to changes in the lengths of 
 
 members 87 
 
 Double intersection trusses by influence lines . . 65 
 
 Dynamic impact 262 
 
 work of deformation due to. ... 43 
 
 influences ' 256 
 
 363 
 
364 
 
 INDEX 
 
 E 
 
 PAGE 
 
 Eccentric connections, stresses produced by. . . 255 
 
 Effects due to unusual loads 266 
 
 Effect of shop lengths on determinate structures 256 
 indeterminate struc- 
 tures 208 
 
 Elastic deformations 11 
 
 Elastic loads w in terms of angle changes. 109, 111 
 
 Equation of an influence line 47 
 
 Equations, solution of simultaneous 295 
 
 Externally indeterminate, definition of 6 
 
 External redundant, influence line for one . 128, 140 
 
 F 
 
 Fatigue of material 265 
 
 Features in design intended to diminish secon- 
 dary stresses 267 
 
 Fixed framed arches 178 
 
 effect of abutment dis- 
 placements 188 
 
 example 189 
 
 influence lines for the re- 
 
 dundants 184 
 
 location of the coordinate 
 
 axes 182 
 
 resultant polygon for .... 187 
 stress influence lines for.. 185 
 temperature stresses in . . 188 
 
 Fixed masonry arches 298 
 
 coordinate axes and 
 elastic loads, 314, 331, 346 
 
 critical sections 322 
 
 determination of the 
 
 redundants 309 
 
 example of symmetric. . 326 
 example of unsymmet- 
 
 ric 346 
 
 influence lines for the 
 
 redundants. 317, 333, 351 
 maximum stresses, 
 
 322, 344, 357 
 modern methods of con- 
 struction 301 
 
 preliminary design for, 
 
 304, 326 
 resultant polygons, 
 
 324, 335, 357 
 stresses on any normal 
 
 section 320, 344 
 
 temperature stresses in, 
 
 318, 345, 357 
 
 Framed structures, theorems, laws and formulae 
 for 11 
 
 G 
 
 General case of redundancy 203 
 
 Girder continuous over four supports 130 
 
 three supports. 129, 143, 148 
 fixed at one end, supported at the other . 140 
 
 Glossary of terms vii 
 
 Graphostatics, bibliography 359 
 
 H 
 Horizontal displacements 114 
 
 Impact due to dynamic effects ............... 
 
 formulae ........................... 
 
 Indeterminate, externally ................... 
 
 internally .................... 
 
 Indeterminate straight beams, work of defor- 
 mation .................... 
 
 structures by Maxwell's law. ... 
 
 by Mohr's work equa- 
 tion .............. 
 
 effect of abutment 
 displacements. 32, 
 effect of shop lengths 
 effect of temperature 
 influence lines for, 
 
 127, 
 
 influence line for one 
 redundant ....... 
 
 stress influence lines 
 for ...... 137, 139, 
 
 Influence line, equation for .................. 
 
 defined ....................... 
 
 for cantilever bridges .......... 
 
 deflections ................ 
 
 determinate structures, 
 
 46, 65, 
 direct loading .............. 
 
 displacements of determinate 
 structures ............... 
 
 double intersection trusses ... 
 indeterminate structures. 127, 
 indirect loading ............ 
 
 moments ................ 49 
 
 one external redundant. 128, 
 one internal redundant ..... 
 
 one redundant condition. . . . 
 
 shear ................... 43 
 
 262 
 
 264 
 
 6 
 
 41 
 24 
 
 19 
 
 208 
 208 
 206 
 
 140 
 127 
 
 140 
 
 47 
 46 
 67 
 49 
 
 218 
 
 48 
 
 122 
 65 
 140 
 50 
 , 50 
 140 
 128 
 127 
 , 50 
 
INDEX 
 
 365 
 
 PAGE I 
 
 Influence line for skew plate girder on curved 
 
 double track 80 
 
 stresses in truss members ... 52 
 
 reactions 48, 50 
 
 redundants in fixed masonry 
 
 arches 317, 333 
 
 three-hinged arches 72 
 
 three-hinged masonry arches 78 
 three-hinged solid web 
 
 arches 78 
 
 trusses with subdivided 
 
 panels 83 
 
 two redundant conditions. . . 130 
 Internal redundant, influence lines for one .... 128 
 
 Internally indeterminate 6, 7 
 
 Introduction 1 
 
 Isotropic bodies, theorems, laws and formulae 
 for 34 
 
 K 
 
 Kernel moment influence areas for three- 
 hinged solid web arch 79 
 
 Kernel moment influence areas for two-hinged 
 solid web arches 159 
 
 Lateral trusses, stresses due to wind, etc . . 258, 261 
 
 vibrations 260 
 
 Least work, theorem of 29 
 
 Live load stresses 218, 219 
 
 author's method 219 
 
 Load divide for a truss 51 
 
 Loading, direct and indirect . 47, 48, 50 
 
 M 
 
 Masonry arches, see under Arches. 
 
 Maxwell's theorem of reciprocal displacements . 27 
 
 Menabrea's law of least work 29, 31 
 
 Mitering lock gates 271 
 
 example 279 
 
 theory of the analysis .... 273 
 
 Mohr's rotation diagram 89 
 
 work equations 16, 17 
 
 Moment influence lines 49, 50 
 
 Moving load stresses 218, 219 
 
 Multiple redundancy 132, 137 
 
 simplification of influence 
 lines for 132, 135 
 
 N 
 Nature of secondary stresses 226 
 
 Plate girder continuous over four supports .... 130 
 over three supports, 
 
 129, 143, 148 
 
 deflection of 117 
 
 fixed at one end and supported at 
 
 the other 140 
 
 Positions of a moving load for max. moments. . . 57 
 
 for max. shears 62 
 
 Preface v 
 
 Principle of virtual velocities 2, 16 
 
 work 1, 15 
 
 R 
 
 Reaction conditions 5 
 
 influence lines 48, 50- 
 
 summation influence lines 54 
 
 Redundancy, simplification of influence lines 
 
 for 132, 135 
 
 solution of the general case of ... 203- 
 stress influence lines for struc- 
 tures involving same . 137,139,140 
 
 Redundant conditions. 5 
 
 influence lines for one ... 128 
 influence lines for two . . . 130 
 
 on the choice of 202 
 
 Ritter's method of moments 211 
 
 Rotation diagram, Mohr's 89 
 
 of a rigid frame about a fixed point. . 89 
 
 S 
 
 Secondary and additional stresses, bibliog- 
 raphy 360 
 
 Secondary stresses 226 
 
 as effected by certain de- 
 signs 267 
 
 concluding remarks on 267 
 
 due to riveted connections, 
 
 227, 235 
 
 due to various causes 251 
 
 in members due to eccentric 
 
 connections 255 
 
 in members due to fheir own 
 
 weight 251 
 
 in pin-connected structures. 253 
 in riveted cross-frames, load 
 
 effects 244 
 
 in riveted cross-frames, wind 
 
 effects 247 
 
 the nature of 226 
 
 Shear influence lines 49, 50 
 
 Shearing stress, work of deformation due to. . 38 
 
366 
 
 INDEX 
 
 Simplification of influence lines for multiple 
 
 redundancy 132, 135 
 
 Skew plate girder on curved double track by 
 
 influence lines 80 
 
 Solid web beam, deflection of 117 
 
 Solution of simultaneous equations 295 
 
 Special applications of influence lines to inde- 
 terminate structures 140 
 
 Statically determinate frame, distortion of by 
 
 graphics 87 
 
 structures, deflection 
 
 polygons for 99 
 
 indeterminate structures 140, 194 
 
 influence lines for. 140 
 methods of pre- 
 liminary designing 194 
 
 Stresses by Ritter's method of moments 211 
 
 the method of influence lines 218 
 
 stress diagrams 213 
 
 Stresses, combination of as a basis for designing 268 
 
 due to abutment displacements 32 
 
 in statically determinate structures, 
 
 210, 218 
 
 on any arch section 157, 320 
 
 Stress influence lines for indeterminate struc- 
 tures 137, 139, 140 
 
 for truss members 52 
 
 Structural redundancy, tests for 6,7 
 
 Summation influence lines for reactions . . 54 
 
 Temperature effects on determinate structures . 256 
 indeterminate struc- 
 tures 206 
 
 stresses follow Menebrea's and 
 
 Castigliano's laws 31 
 
 fixed framed arches 188 
 
 fixed masonry arches, 
 
 318, 345 
 
 girder on three supports . 146 
 truss on three supports .. 152 
 two-hinged arches, 
 
 156, 165, 167 
 
 Tests for structural redundancy 7 
 
 Theorem of least work 29 
 
 Theorems relating to work of deformation .... 29 
 
 Theory of elasticity, bibliography 359 
 
 Three-hinged framed arches by influence lines . 72 
 solid web and masonry arches by 
 influence lines. . 78 
 
 Tractive forces, effect of j 
 
 Trusses with subdivided panels ; 
 
 Truss on three supports, problem 1 
 
 Two-hinged arch with cantilever side spans. . . 170 
 
 tension member 1 
 
 framed arch, abutment displace- 
 ments 168, 1 
 
 deflection for any 
 
 point 168, 1 
 
 example 166, 1 9 
 
 stress* influence areas 
 
 for 168, 1 
 
 temperature stresses 
 
 in 167, 17" 
 
 with column supports I.'/ 
 X a influence line for, 
 
 166, 1 
 
 solid web 'arch, abutment displace- 
 ments 156, ] 
 
 example 153, 1 
 
 kernel moment in- 
 fluence areas . . . . ] 
 temperature stresses, 
 156, ] 
 
 X a influence line for, 
 154, 
 
 U 
 
 Unusual load effects. 
 
 V 
 
 Virtual velocities , 2, 
 
 work, principle of 1 , 
 
 W 
 
 Williot displacement diagrams 
 
 Williot-Mohr diagrams 
 
 Wind pressure 2 
 
 stresses in cross-frames 
 
 in main and lateral trusses 2 
 
 Work equations for any frame 14 
 
 for isotropic solids 
 
 of deformation 1 
 
 due to dynamic impact.. . 
 due to shearing stress .... 
 
 due to moments 
 
 for any indeterminat i 
 
 straight beam 
 
 theorems relating to 
 
THIS BOOK IS DUE ON THE LAST DATE 
 STAMPED BELOW 
 
 AN INITIAL FINE OF 25 CENTS 
 
 WILL BE ASSESSED FOR FAILURE TO RETURN 
 THIS BOOK ON THE DATE DUE. THE PENALTY 
 WILL INCREASE TO SO CENTS ON THE FOURTH 
 DAY AND TO S1.OO ON THE SEVENTH DAY 
 OVERDUE. 
 
 SEP. 
 
 DEC 10 1932 
 
 MAR IB 1934 
 
 MAR 23 1S23 
 
 J940 
 
 APR 28 1947 
 
 14 r 
 
 LD 21-50?re-8,'32 
 
YD 02782 
 
 UNIVERSITY OF CALIFORNIA LIBRARY