UC-NRLF ON THE THEORY AND CALCULATION CP BY MANSFIELD MERRIMAN, C.E., Ph.D., SHEFFIELD INSTRUCTOR IN SCIEN THE NEW YORK: D. VAN NOSTRAND, PUBLISHER, 23 MURRAY AND 27 WARREN STREET, 1876. \ If o PKEFACE, The following pages are divided into three chapters. The first presents by way of intro duction some of the elementary principles of continuous girders, and the fundamental ideas relating to the calculation of strains. The sec ond gives the theory of flexure as applied to the continuous trjiss^ of constant cross section, and exhibits it in formulae (I) to (VI), ready for application to any particular case ; and the third gives an example of the computation of strains in a continuous truss of five unequal spans, with some useful hints concerning the practical building of such bridges. The theory of flexure indicates that, by the use of continuous instead of single span bridges, a saving in material of from twenty to forty per cent, may be effected. f It is easy in deed to say that this advantage will be entirely swallowed up by the effect of changes of tem perature, increased labor of erection, or addi tional cost of workmanship, but by no amount of reasoning can such disadvantages be esti mated. Theory indicates a large saving, whether or not it can be realized, may only be determined by trial. Other nations have built and are building continuous bridges, and their experience has not yet shown that the system is inferior to that of single spans. The inter est now prevailing among American engineers in the subject, and the fact that at some recent bridge let tings plans have been offered for a continuous structure, seem to indicate that the system will also be tried here. This little book may then perhaps be of val ue to bridge engineers, as well as to students in general. M. M. New Haven, Conn. , July 10, 1876. THEORY AND CALCULATION OP CONTINUOUS BRIDGES. WHEN a straight bridge consists of several spans, each entirely independent of the others, it is said to be composed of simple girders. If, on the other hand, it consists of a single truss extending from one abutment to the other without any disconnection of parts over the piers it is called a continuous girder. A load placed upon any span of a continuous beam influences, to some extent, each of the other spans, and hence its complete theory is much more complex than that of the simple one. This very complexity however has rendered the subject an at tractive one to mathematicians, who, pursuing science for science s sake, have 6 investigated the laws of equilibrium which govern it. These laws with the many beautiful consequences attending them form one of the most interesting chapters of mathematical analysis, and as such have interest and value inde pendent of their application in engineer ing art. It is the object of the present paper to present in as simple a form as possible some of the main principles and laws most needed by the engineer, and to il lustrate their application as fully as space will permit to the practical de signing of continuous bridges. CHAPTER 1. The first point to be observed in con sidering either a simple or continuous girder is that all the exterior forces which act upon it are in equilibrium. The exterior forces embrace the weight of the girder and the loads upon it which act downward, and the pressures or re actions of the supports which act up- ward. In order that these may be in equilibrium, it is necessary that the sum of the reactions of all the supports must be equal to the total weight of the girder and its load. Thus, if a simple girder of uniform section and weight rest at its ends upon two supports, the reaction of each sup port will be one-half the weight. Exactly in the center between the two supports or abutments, let us suppose a pier to be placed just touching, but not pressing against the beam, which, at that point, has a deflection below a straight line joining the two abutments. Then the condition of things is in no way altered, for the weight being W, each abutment reacts with a force \ W, while the pier bears no load. Raise now the pier so as to lift the girder above the line of de flection and it receives a part of the weight W, while Jie reactions of the abutments become less than \ W. If the pier be raised higher and higher, it will at length lift the girder entirely from the abutments and bear itself the 8 whole load W. In every position, how ever, the sum of the reactions of the three supports is equal to the total load. For example, when the three are on the same level it may be shown that the re action of each abutment is -fs W, and that of the pier ii W. This illustration shows also that small differences of level in the supports occur ring after the erection of a bridge cause large variations in the reactions of its supports and in the strains in its several parts. A simple girder having a deflec tion of one inch, would, if raised one and three-fifth inches at the center, be entirely lifted from the abutments. In the first case the upper fiber would be in compression, the lower in tension; in the second case, the upper would be in ten sion, the lower in compression. If the center were raised only one inch, the re versal would be only partial, the upper fiber becoming subject to tension for a short distance on each side of the mid dle. This fact often used as an argu ment against continuous bridges, is really 9 an objection only when the piers are liable to settle after erection. Difference 8 of level, previously existing, do not act prejudicial when the bridge is built upon the piers, and with a profile correspond ing to them. The mathematical theory of the con tinuous girder enables its reactions and internal strains to be found for any as sumed levels of the supports, provided only that the differences of level are very small compared w T ith the length of the spans. However interesting such inves tigations may be in themselves, they are of little importance in practice, since it has been shown that when all the points of support are on the same level, the greatest economy of material results.* In all that follows, then, we shall regard the girder as resting on level supports, or, what is the same thing, that it was built with a profile corresponding to that of the piers. The loads upon a bridge and the reac- * Weyrauch ; Theorie der continuir lichen Trdger, p. 129. Winkler ; Lehre der Elasticitdt, p. 155. 10 tions of the supports are external forces. The equilibrium between them is main tained by means of internal forces, which, in a framed truss, are transmitted longi tudinally along the pieces as strains of tension and compression. When all the external forces are known, these internal forces or strains can be readily found. This very important point we shall now proceed to illustrate. Fig. 1 represents a portion of a con- Fig, i, A R l R tinuous girder; the first span on the left is called Z l5 which also represents its length, the second J 2 , the third / 3 , etc. ; in like manner,- the supports beginning on the left are designated by the indices 1, 2, 3, etc., and their reactions by R,, R 2 , R 3 , etc. Let the load per linear unit be iv, supposed uniformly distributed, then the weight of the first span will be wl of the second wl of the third w l 11 etc. If there are four spans the total weight will be and from the fundamental idea of equili brium, we must have the equation Each of the reactions is then a fraction al part of the total load, and by methods hereafter to be explained, their values may be readily computed, whatever be the number of spans. Granting for the present that they may be found, let us inquire how we may obtain the internal forces or strains in any part of the gir der. In the span l z let a vertical plane be passed, cutting the beam at a point whose distance from the support 3 is x. All the internal forces acting in this sec tion may be considered as resolved into two components, one vertical and the other horizontal. The sum of all the vertical components is a force which pre vents the two parts of the beam from 12 shearing asunder, and is called the shear for that section ; the horizontal compo nents acting in parallel planes are the resisting strains of tension and compres sion in the horizontal fibers, and the sum of their moments with reference to any point in the section is called the moment of resistance, or simply the moment for that section. The internal strains in any section are thus completely represented by the shear and moment. For example, if the girder in Fig. 1 be a framed truss of which Fig. 2 represents the span 1 3 Fig. 2, A enlarged, and the section be passed cut ting the three pieces E F, F e, and ef, the vertical components of the chord strains will be zero, and that of the di agonal strain e F will be the shear. Hence, if the shear be known, and the angle included between the vertical and a diagonal be 6, we have only to multi- 13 ply the shear by sec. 6 to find the strain in the diagonal. Again, let the section be moved to the left so as to pass through the point e, and let that point be taken as the center of moments. Then the mo ment of resistance will be the moment of the chord strain E F. Hence, if that moment be known, we have only to di vide it by the depth of the truss to get the strain in E F. The internal shear and moment at any section are easily found from the fundamental conditions of statical equili brium. The shear being an internal vertical force is the resultant of the exterior vertical forces on either side of the section. The exterior forces on the left of the section, for instance, have for a resultant their algebraic sum ; considering the upward forces as posi tive, and the downward ones as negative we have from Fig. 1, their sum wx as the expression for the shear in the section x. To get the internal moment 14 for the same section, we have only to consider in like manner that it is equal to the sum of the moments of all the ex terior forces on either side of the section, for if otherwise, there would be a ten dency to rotation. The moment of the force Ri with reference to x is R : (I, + Z a + a) , of R 2 is R 2 (I, + x) , of the load t0i l9 iatoi, (i^ + ^+aJ), etc. Thus from a mere inspection of Fig. 1 we write the value of the moment M, regarding those moments as positive which cause a ten sile strain in the upper fiber at x, and those as negative which cause a compres- sive one. The expression is + x ) + w I* (i k + x ) 1^3 x + wx \ x Now in these expressions for the in ternal shear S and the moment M at any point x, the lengths , 2 , x are given by the conditions of the case in hand, and the same is true of the load per linear unit w. Hence the shear and moment, and consequently the internal strains are easily obtained as soon as the reactions 15 of the supports are known. We shall hereafter give methods by which the re actions may be readily determined. By the same reasoning if we pass a section in the span Z 2 , (Fig. 1) at a point whose distance from the support 2 is x, the shear S and the moment M for that section will be W X* ^^ w 2 w x Mc= -R x (Z, + x) + w 1 1 (\ 1, + x) In a simple girder whose length is each of whose reactions is R or ^ w I the shear and moment for any section x will be I x) WX* W 2 When the number of spans is large, the expressions for the shear and moment as above deduced become long and in volve much arithmetical computation. We are, however, fortunately able to 16 place them under a much simpler form. First they may be written thus wx M= [-K, ( + wl^ + R 3 ) is the shear in the span 1 3 at a point infinitely near to the support 3 ; let this be called S 3 . Also the quantity enclosed in [ ] in the second equation is the moment of the exterior forces with reference to the point 3 ; let this be called M 3 . Then the equations become Therefore the internal shear and mo ment at any section can immediately be found) without the necessity of determin ing the reactions, provided ice know the shear and the moment for the preceding support. This method, due to Clapeyron, of using the moment at the supports in stead of the reactions greatly simplifies 17 the numerical computations of a continu ous truss. We designate the moment at 3 by M 3 , the reaction being R 3 , or the sum of the shear S 3 , in the span 1 3 and of the shear S 2 in the span 1 2 both infi nitely near to the support 3. In like manner the moments at the supports 2 and 4, will be designated by M 2 and M , the shears just to the right of those points by S 2 and S 4 , and those to the left by S j, and S 3 . In general for any sup port whose index is n, w r e have (Fig. 3) Fig. 3, < 7 * 7 *- 7. _ _ -u-i n I ^ n+i n-\ ^ n-t-i ^^^^^"^^^ji -X - on the left, the span n _i, on the right the span / n ; the shear infinitely near to n on the left is S / n _i, on the right S n ; the sum of S n _i, and S n makes the reaction R n ; and the moment over the support is M G . If the load be uniform and equal to w per linear unit, the internal 18 shear S and moment M for any section x are given by S = S n wx M=M n S n x + 1 w x* To find then the internal strains in the diagonals of a continuous truss due to dead load only, we have to pass a section cutting each diagonal and find the value of S, this is the shear which the strain in the diagonal must resist o and multiplied by sec. 8 (8 being the in clination of the diagonal to the vertical,) it gives the required strain. To find the s trains in the upper chord we have to take the lower chord apices as centers of moments and compute the values of M ; these divided by the depth of the truss give the strains, which are tensile if M is positive, compressive if M is neg ative. To find the lower chord strains, we choose the upper apices from which to measure the values of a?, and divide the resulting values of M by the depth of the truss ; if M is positive these give compressive strains ; if negative, tensile ones. Everything is thus known, except the shears and moments at the supports, and for these formulae and methods will be presented in Chapter II, by which they may be found for all cases. We give here, however, two tables from which they may be found for the common case when all the spans are equal, and which, by a simple law, may be extend ed to include any number of such spans. As before, let w be the unifornily dis tributed load per linear unit; let I be the length of each span, then will w I be the weight of one span. The shear at any support is a fractional part of w I, or s A w I A being a fraction given in the follow ing triangle : 20 21 Each of the squares composing this triangle represents one of the supports, and its left hand division gives the left hand shear S n _i, and the right hand one the shear S n (Fig. 3). Thus, in the gir der of three spans, the triangle shows that the first support, beginning at the left, has on the left no shear, and on the right YQ 10 /, that the second support has on the left a shear of ^ w /, and on the right one of -f w I. The sum of the two shears for any supports is of course its reaction. For example, a girder of six equal spans has at its middle support a reaction of I8t w I. The moment at any support will be a fractional part of ivl^or 7. Moment = B w l^ B being a fraction given in the following triangle ; in which like the preceding one the spaces indicate the supports of the girder. Thus, the fourth horizontal line refers to a girder of four spans, the moments at the first and last supports being 0, at the second and fourth - 2 3 8 - w I* and at the middle one w I 2 . 22 23 The triangles can be extended to any required length by the application of the following law which obtains in &\\ oblique columns. Any fraction belonging to an even number of spans, may be obtained by multiplying both numerator and de nominator of the preceding fraction by two and adding the numerator and de nominator of the fraction preceding that. 33 Thus for eight spans, the fraction 2X11 + 11 is equal to - - or to 2X142 + 104 <388 ,. 2X142 + 104 accordm S as we use one oblique column or the other. For an odd number of spans, any fraction is found by adding the two preceding fractions, numerator to numerator and denomina tor to denominator. Thus for seven spans, 12 8+4 12 9 + 3 or 142 104 + 38 142 104 + 38 These tables* furnish the data for solv- * These triangles were first given by the author in the Journal of the Franklin Institute for March, 1875. A de monstration of the laws governing them may be seen in the same Journal for April, 1875. 24 ing all questions concerning continuous girders whose supports are on the same level, whose spans are all equal, and which are loaded uniformly throughout their entire length. The reader should first acquire facility in the use of the tables. We give, therefore, a few ex amples for practice : 1. In a girder of six spans, what is the reaction at the second support ? 118 Ans. K. = - - w I. 104 2. In one of eight spans, what is the reaction at the middle support? 386 Ans.K=-wl 3. In one of ten spans, what is the mo ment over the fourth support from the left? 4. In one of seven spans, what are the shears S 2 and S ? (see Fig. 3.) 25 5. In one of six spans, what is the mo ment M and the shear S 6 ? Ans. M= Having thus found from the triangles, .the moment M n and the shear S n for the n th support, the shear S and the moment M for any section in the n ih span are readily found from the formulae O - On W X Tf C I 1 nn rf -LT_Ln c?n th ~r 2^ w **j which we have demonstrated above, and Fig. 3. n-i n n+i in which x is the distance from the sup port n to the assumed section. If in these x be made equal to I, they will, of course, give the shear S n at the left of the w + l th support, and the moment M n+ i over that support. We will illus trate their use by a few examples : 26 6. In a continuous girder of three spans, what is the shear and the moment at the center of the middle span ? 5 We have from the table S a =--- wl and 1 10 M,=-r. Hence S== w I iv x 1 5 M= ivF-w l and placing x equal to /, we have S=0 M=-irP 40 7. In a girder of six spans, find the shear and moment at the center of the second span ? Ans. S= A wl M = -- - w ? 104 208 8. In one of four spans, what is the shear and moment in the third span for x-=.\ I and x=%l? Ans. S= wl and S= -- wl 14 14 27 In computing a framed truss we need to find the value of S for a section cut ting every diagonal and that of M for one passing through each panel apex, from these we readily derive the strains in the webbing and chords by the rules explained above. A single example will render the whole process clear. (We here treat of the dead load only ; com putations involving the live or rolling load will be presented hereafter.) Let the truss represented in Fig. 2 consist of seven continuous spans, each sixty feet in length. Let the uniformly distributed load per linear foot be two hundred pounds, one half of which rests upon the upper chord and the other half upon the lower. The lower chord is di vided into six bays, each of ten feet, and is connected with the upper one by a Warren system of diagonals. The depth of the truss is seven feet. Let it be re quired to compute the strains in all the pieces of the third span, due to this dead load. We first take from the triangles for a 28 girder of seven spans. S q = wl and M 3 =- wl*. We have for our case, w=200 Ibs. and 1=60 feet. Hence S 3 5916 Ibs. andM 3 =55775 Ibs. ft. Inserting these and the value of w in the above general formulae, we have S = 5916 wx M = 55775 5916 ^ + 100 x* as the value of the shear and moment for anv section x. v Now, since this is a framed truss, and the several pieces are to be subjected only to longitudinal strains, the load should not be strictly uniformly distrib uted but concentrated on the upper chord at the panel points B, C, etc. and on the lower chord at a, b, c, etc. Allow ing that each of these points receives an 29 equal weight and that a and g, count as but one point, we have at a 500 Ibs., at g 500 Ibs. and at each of the others 1000 Ibs. In finding the shear for the diag onal a B, we pass the section anywhere between a and B and takewx as 500, for B b 10 x is 1500, for b C 2500 and so on ; these subtracted from S 3 give the required shears. This is in fact nothing but taking the algebraic sum of all the exterior forces between the left hand of the truss and the diagonal under consid eration, for S 3 is the sum of those forces from the left end to the beginning of the span. For the diagonal Fe we have, for example, S = 5916 - - 8500 = - - 2584 Ibs. Thus by successive subtraction we find the shears for all diagonals. Multi plying them by the secant of the angle between a diagonal and the vertical or 30 and we have the required strains. To determine their character we have simply to consider that a positive shear causes ( tensile ) a -{ I strain in a diagonal ( compressive j which slopes \ u P ward \ toward the ( downward ) left hand support, while a negative shear produces the reverse. In the following table, the results thus determined are recapitulated. The first column shows the name of the diagonal corresponding to Fig. 2, the second gives the shears, the third gives the slopes, + indicating an upward inclination toward the left, and - a downward one, and the last column gives the strains, -f indicating tension, and compression. In forming the last column from the two preceding ones, it will be noticed that the rule of signs is observed : 31 DIAGONALS. (See Fig. 2.) Piece. Shear. Slope. Strain. Cb J -5416 -3416 + -66561bs. +5427 -4198 Cc DC h2416 -1416 -U2969 -1740 Ed + 416 - 584 -1584 - + 511 + 718 -1947 Ye -2584 -3584 ^ +3176 -4405 S{ -4584 +5634 -6863 We will now pass to the computation of the chord strains. In the above ex pression for M the quantity ^ 10 cc 2 or 100 x* is the moment of the load between the point 3 and the assumed section and its value is the same whether the load be considered as uniformly distributed or concentrated at the apices as above. Hence to find the moments for the upper chord we have in the expression M 55775 -- 5916^+lOOar 2 simply to give to x the successive values 32 0, 10, 20, etc., since for the bays A B, B C, C D, etc., the opposite vertices #, b, c etc., must be taken as centers of moments. Thus if the bay C D be cut rotation will at once begin around the point c ; we take then x =20 and find for the mo ment of the strain in C D, M= - - 22545 Ibs. ft. and dividing this by its lever arm or seven feet we have 3221 Ibs for the strain. The character of the strain is found by recollecting that a positive mo- ( tensile } ment causes a \ . > strain in ( compress^ve \ the ** [ chord, while a negative ( lower } moment produces the reverse. If we designate then tension by -f and com pression by , the signs of the strains in the upper chord will be the same as those of the moments. In this way it is easy to compute the following results : 33 UPPER CHORD. (See Fig. 2). Bay. Moment. Strain. AB BC CD DE EF FG GH +55775 Ibs. ft. + 6615 -22545 -31705 -20865 + 9985 +60845 +7968 Ibs. + 945 -3221 -4529 -2981 +1426 +8692 For the lower chord the calculation is very similar. The centres of moments are taken at the points B, C, etc., the successive values of x are 5, 15, 25, etc., the strains are numerically one-seventh of the moments, and their signs are op posite to that of the moments. Thus for the bay e/,a?=45, M= - 7945 Ibs. ft. and the strain in ef is 1135 Ibs. tension. The results are in the following table : 34 LOWER CHORD. (See Fig. 2.) Bay. Moment. Strain. ab +28695 Ibs. ft. -4099 Ibs. be -10465 -1495 cd -29625 -4218 de -28785 _ -4112 f - 7945 -1135 fg +32895 -4699 and the strain sheet for the span is now complete. In the same way the strains for each of the other spans may be readily found. From the symmetry of the truss it is evident that the fifth span will be exact ly the same as the third, the sixth the same as the second, and the seventh the same as the first. For the fourth span the value of S and M for any section x are w x M=60848 6000& + 100 x 3 and the strains will be the same on each side of its center. For the first span Sj is the same as the reaction R 1 the mo ment M, is zero and we have 35 8 = 4732 wx M= 4732 x+100 It will be seen then that the compu tation of the strains in a continuous girder is exactly the same as in a simple one, except only in the preliminary de termination of the shears and moments at the supports. In a simple girder the end shears are the same as the reactions which are known from the law of the lever, and the moments at the supports are zero. In a continuous one these quantities must be determined by formu lae, or, for the case of equal spans uni formly loaded,* taken from the triangles which we have given above. They may also be found by a graphical process. If the truss above discussed were built with seven simple girders, the strains in each would be the same. It may prove interesting then to compare the results above found with those for a simple gir- * Other convenient tables for concentrated loads and for uniform loads over one span only are given in the ar ticle above referred to in the Journal of the Franklin In. stitute. 36 der. The mode of computation is essen tially the same ; the end shears are each one half the total load, the pieces A B and G PI in Fig. 2 disappear, and for any section x we have 8=6000 wx Considering as before that the load is concentrated at the panel points we have 500 Ibs. at a and y and 1000 Ibs. at B, C, D, , c, d, etc., respectively. We then find the shears and moments and from them deduce the strains as above de scribed. The results are given below compared with those for the third span of the continuous truss. (See Table on following pages.} Adding these strains regardless of sign we find the two sums to be the same. It can be easily demonstrated that for the dead load such should be DIAGONALS. (See Fig. 2.) Piece. Continuous truss. Simple truss. Ea -6656 Ibs. 6760 Ibs. Cb +5427 -4198 +5531 4301 Cc DC +2969 1740 +3072 1844 Dd Erf + 511 + 718 -1947 + 614 + 614 1844 e +3176 -4405 +3072 4301 G +5634 6863 +5531 6760 Sams. . . 44244 Ibs. 44244 Ibs. the case. As far as the diagonals are concerned, the two structures require an equal amount of material. UPPER CHORD. (See Fig. 2.) Bay. Continuous truss. Simple truss. AB BC CD DE EF FG GH +7968 Ibs, + 945 3221 4529 2981 +1426 +8692 7143 Ibs. 11429 12857 11429 7143 Sums. . . 29762 Ibs. 50001 Ibs. 38 Adding the strains in the upper chord we observe that the sum for the simple truss is about 1.7 times that of the other. If the amount of material is to be pro portional to the strain, a considerable saving will here be expected. LOWER CHORD. (See Fig. 2.) Bay. Continuous truss. Simple truss. ab -4099 Ibs. 3929 Ibs. be -1495 9643 cd -4218 12500 de -4112 12500 ef -1135 9643 fff -4699 3929 Sum.... 19758 Ibs. 52144 Ibs. The lower chord in the simple truss would then be subjected to about 2.6 times as much strain as in the continuous one. If we suppose that the same working strength may be allowed for compression as for tension, we may obtain an estimate of the saving in material by employing a 39 continuous truss instead of a simple one. The amount of material will be propor tional to the strain and to the length of the piece strained. Regarding the bays of the chord as unity, the diagonals will be represented in length by 0.86. The pro portionate amounts of iron will then be found by multiplying the above sums by unity for the chords and by 0.86 for the diagonals. Thus we have a COMPARISON. Continuous truss. Simple truss. Diagonals .... Upper Chord. . Lower Chord. . 38050 29762 19758 38050 50001 52144 Total 87570 140195 from which we see that the amounts of material in the two cases are in the ratio of the numbers 87570 and 140195 or as 1 to 1.6. For this particular span then a saving in material of thirty-seven and a half per cent, is effected by using a con tinuous truss instead of a common one. 40 It is capable of demonstration that for girders subjected only to dead load, the total amount of strain in the webbing will be the same for simple as for con tinuous trusses, and also that under the most favorable circumstances, the total strain in the chords of the first is to that in the chords of the second as /\/2^ is to 2 or nearly as 2.6 to 1. In studying the theory of girders many interesting questions arise which are of little importance in practice. One of these is the determination of the inflec tion points. At these points the curva ture of the beam changes, the strain passes from tension to compression and the moment is zero. At any point in the nth span the moment is M = M n - - S n x +-J- w x* Making in thisM equal to zero and solv ing the equation with reference to x we find ; 2 " w 41 For a girder of equal spans and uniform ly loaded we may hence write for the two inflection points, in which A and B are to be taken from the above triangles, A always being taken for the right hand side of the support under consideration, for example, in a girder of eight spans the inflection points for the fourth span are at the points 195 V388/ 388 or for 33=0.22 I and ^=0.79 L The point of maximum moment, or the point near the center of the beam, where the chord strain is the greatest is more important and readily determined from the above general value for M. Differentiating it with reference to x we have = S n + W 23=0 a x that is, the maximum moment obtains at 42 the point where the shear is zero. Its S value is found by replacing x by which gives ft 2 Max. M=M n 2 w as the greatest negative moment. The following examples will enable the reader to test his knowledge of the preceding principles : 9. In a girder of two spans uniformly loaded what are the maximum positive and negative moments ? Ans. 0.125 wT and 0.071 wl\ 10. In one of three spans what is the maximum negative moment in the mid dle span ? Ans. 0.025 wF. 11. In one of eight spans where are the inflection points in the fifth span ? Ans. x= 0.2lZandcc=( 12. A continuous girder of three spans, each equal to fifty feet, is divided into 43 five panels on the lower chord, and has bracing similar to that shown in Fig. 2. Supposing a load of five tons applied at each of the lower panel points, what are the strains in each of the pieces of the middle span ? the height of the truss being six feet. Ans. In a &, --11.7 tons; in b c, +1.7; in B&, +13 ; etc. In this chapter we have treated of the continuous girder when affected only by dead load or its own weight. In the fol lowing chapters we shall take up the ac tion of the live or variable load. 45 CHAPTER II. A continuous girder, loaded in any manner, is held in equilibrium by the up ward pressures or reactions of the sup ports, and, as we have seen, these reac tions are alone sufficient for the complete determination of the strains in every part of the girder. But if, regarding the question as one of pure statics, we con sider the beam as rigid, we find it impos sible to determine the reactions when the number of supports is greater than two. This does not arise from the fact that in an actual case the question is indeterm inate, but simply because in considering the girder as rigid we have restricted the data to the mere weight, neglecting en tirely the physical properties of the ma terial. By taking into account the elas- 46 ticity of the girder, the problem becomes determinate ; we find the reactions, or what is equivalent, the shears and mo ments at the supports, and from these the investigation of the internal forces or strains is easy. THE ELASTIC LINE. When a girder is acted upon by verti cal forces, a change of shape arises, which causes the originally parallel fibers, to be on one side lengthened, and on the other shortened. Between the lengthen ed and shortened fibers, there is a plane which undergoes no change of length ; the central line of this plane is called the neutral axis or the elastic line. Thus, in the bent beam represented in Fig. 4, m o is the neutral axis, the fibers above it be ing shortened "or compressed, and those below it lengthened or tensioned. We derive the equation of the elastic line upon three hypotheses : 1st, that all planes perpendicular to the neutral axis before the bending or flexure, preserve 4 - during the bending their perpendiculari ty and their form as planes ; 2d, that the change of length of a body subjected to a force is, within certain limits called the elastic limits, proportional to the intensi ty of the force ; and 3rf, that the change of shape is so little that the length of the neutral axis is sensibly the same as its horizontal projection. In Fig. 4 we have a longitudinal sec tion of a portion of a bent beam ; the two planes a b and d 6, originally parallel, re maining perpendicular to the neutral axis m o, and intersecting in c the center of curvature. Hence, drawing f g paral lel to ab through o, the lines /V7, ge, etc., denote the elongations and compressions of the respective fibers, and we see from the figure that od:od r \\df: d f or the change of length in the fibers is proportional to their distances from the neutral axis. This is the consequence of the first hypothesis. Designating by H and H the force 48 acting in the fibers df and d /", the sec ond hypothesis says that H:H :;<7/: d f FIG. 4. Hence, by combining these two propor tions, H : H ; I o d : o d r or, the horizontal forces are directly pro- portional to their distances from the neutral axis. Therefore, if we denote the distance of any fiber from the axis by 2 the strain upon it by H , the dis tance of the remotest fiber by e and the strain upon it by H, we have H : H ; ; z : e or H = e We have thus far considered the cross- section of the fibers as unity. If the actual section be a, the force in each is evidently - - . Each of these forces, as for instance H in the figure, tends to turn the beam around the point o with a lever arm o d or z, and its moment or the measure of that tendency to rotation is the product of the force * by the dis- TT y 8 tance z, or - . The sum of all these moments is M=- e 50 or, since 2 a is the moment of inertia of the section a &, we have 6 as the expression for the sum of the mo ments of the internal forces, H being the strain in the remotest fiber, e its distance from the neutral axis, and I the moment of inertia of the cross-section. The line df denotes the change of length of the fiber ad due to the force H. Hence if E be the coefficient of elasticity,* ad: df: ; E: H Designating the radius c o by r we have from the similar figures o df and cad (since mo=ad), a d : df. \r : e * The Coefficient of Elasticity is the ratio of the force of displacement to the amount of displacement taken upon a cube whose edge is unity ; Hence for the above J 4* case E=H-$ - . The term modulus of elasticity pro- wtt perly relates to the impact of bodies, and is a measure of elasticity in the common sense of the word, unity indi cating perfect elasticity or restitution of form. These terms are often confounded by writers. 51 Combining these proportions we find H_E e r and hence, for the value of the internal moment, we have The radius of curvature of any plane curve, whose length is u, and co-ordinates x and y is* r= d x d* y And as by our third hypothesis we may place du=dx, this becomes r _dx* /*y ^ i \At */ which, inserted in the above value of M, gives (1) d* y M dx* El as the differential equation of the elastic * See any work on the Differential Calculus. 52 curve, applicable to all bodies subject to flexure, which fulfill the condition im posed by the third hypothesis. The co efficient of elasticity and the moment of inertia may be different in every section. CONDITIONS OF EQUILIBRIUM. Let us consider the r thv span of a con tinuous girder whose length is l Ty and let a single concentrated load P r be placed on this span at a distance M r from the left hand support r. This load, the loads on the other spans, and the weight of the girder itself, are held in equili brium by the vertical reactions R r i, R Tj etc., of the several supports. (See Fig. 6.) Let us pass a section between the load P r and the support r-j-1 at a distance x from the r th support. As shown in the last chapter, all the internal forces in this section are represented by a shear S and a moment M. The shear S is equal to the algebraic sum of all the external forces upon the left hand side of the sec- 53 tion, and the moment M is equal to the sum of the moments of those forces with respect to the section as a center. Hence, regarding upward forces as positive, and a moment as positive when it tends to cause tension in the upper fiber of the section, w r e have (2) S = S, - P r in which S r is the shear at the right of the r th support, and M r the moment at that support. In like manner for a sec tion between r and P r> we have S=S r M=M r - - S r x The internal forces at any section can then be found as soon as the shear and moment at the preceding support are known. If, in the above expression, we make x equal to / r , M becomes M r _j_ i, and we de duce 54 The shear and moment at any section can then be determined as soon as M r and M r 4_ i, the moments at the preceding and following supports, are known. These conditions of equilibrium are entirely independent of variations in the dimensions or material of the beam, or in the relative heights of the supports of tlie girder. THE EQUATION OF THE ELASTIC LINE. In order to apply equation (l) to the case of continuous girders, we have to insert for M, E and I their values as func tions of x and integrate the equation twice. E, however, cannot under any ordinary circumstances be a function of x, it being dependent upon the elasticity of the material alone, which is nearly the same in one and the same beam, and we hence regard it as constant. In a beam of uniform section I is constant, and although it varies in common bridge o o trusses, we shall be obliged in order to bring the investigation within the limits 55 of this paper to consider it always as con stant, taking care to point out after wards the slight error thus introduced. Inserting then in (1) the value of M from (2), we have as the differential equation, applicable to girders of constant elasticity and uniform cross section. Integrating this, the constant is t r , the tangent of the in clination of the elastic line at the sup port r and dy 2M r x-S r a * + -- we have thus far taken no account of the relative heights of the supports. For the reasons mentioned in the last chapter, we shall consider them as all upon the same level. The constant for the second integration is then 0, the origin being at r, and we have 3 M x 2 -S x* + P r x 56 as the equation of the elastic curve be tween the load P r and r-fl th support (Fig 5). If there be several loads we have only to affix the sign of summation 2 to the term involving P r> and if that term be omitted we shall have the equa tion between the load and the r ih sup port, since for any section between those points M=M r S r #. If in (5) we make # / r , y becomes 0, and inserting for S r its value from (3), we find (6) 6EH=-2M r t Thus the equation of the curve is completely determined, when we know M r and M r + 1 the moments at the sup ports r and r-f-1. These may be found by the remarkable theorem of three mo ments. THE THEOREM OF THREE MOMENTS. In Fig. 5 is represented a portion of a continuous truss. Beginning at the left hand end, the lengths of the spans are 57 denoted by l^ / 2 , ..... l r , etc , and the sup ports are designated as 1, 2, ..... r, etc. Upon the spans l r \ and / r are loads P r i and P r> whose distances from the nearest left hand supports are // r i an d Jc l^ k being any fraction less than unity, and not necessarily the same in the two cases. The equation of the elastic line between P r and the support r + 1 is given by (5), and the tangent of the angle which the curve at the section x makes with the axis of abscissae is iven (4). Fio. 5. ^ ---- if-i ~" x ----- it- If in (4) we substitute for S r its value from (3), and for t r its value from (6), and make cc=/ r , -~ becomes t^\\ the ax tangent of the inclination of the curve at r-f-1, and we find 58 Now, if we consider the origin moved from the support r back to r 1, we may derive a value for t r by simply diminish ing each of the indices in the above ex pression by unity, hence (7) 6 EH=M r _i? r _i + 2 M r / r _i -P r _!Z 2 r _!(&-& ) Equating the values of 6 E I t r given by (6) and (7), we have (8) M r _ 1 / r _ which is the most general form of the theorem of Three Moments for girders of constant cross-section. By prefixing the sign 2 to the terms in the second member, it becomes applicable to any number of single loads. For uniform c_> loads ^ r i and w r per linear unit, we have only to place P r _ i = w r id(kl r i) = w r il r \dk P r = w r d (k l r )=io r l r d k And to replace the sign of summation 2 by that of integration /. If these 59 loads extend over the entire spans l r and / r 1> we take the integrals between the limits k=Q and &=1 and have r (/r + ?r - l) + M r + l 1 T W _ 1 3 _ W T I s T which is the theorem as first announced by Clapeyron. For each support of a continuous girder an equation may be therefore written involving the moment at that support, and those at the preceding and following support. In a girder of s spans there are 5+1 supports, and since the moments at the first and last sup ports are zero we have s 1 moments, whose values may be found by the so lution of the 51 equations. The mo ments give the shear at any support, and by (2) the internal forces or strains may be determined for every section of the girder. REMARKS ON THE PRECEDING THEORY. The laws of the theory of continuity above deduced can be regarded as onlyap- 60 proximate. Of the three hypotheses upon which the differential equation of the elastic line is deduced, the first, although a most reasonable assumption, has not been definitely verified by experiment, and the second is rendered somewhat doubtful by the extreme difficulty in delicate experiments of assigning the elastic limits. Nevertheless they are universally regarded by all writers as sufficiently accurate to form the basis of a working theory, and must continue to be thus used until we attain to a more thorough knowledge of matter and force. The third hypothesis, however, is a limi tation of the data, which we are at per fect liberty to make, since we know that the increase in length of the girder by deflection is too small to be practically measured. We may conclude then that the equation (1) is an extremely close approximation to the actual law govern ing straight elastic beams. From the time of Navier to the present it has been so accepted and used. The next hypothesis or limitation of 61 data, which we make, is that E, the co efficient of elasticity, is constant through out the girder. In a solid beam of ordinary homogeneous material, there can be no reason for supposing it other wise. In the Journal of the American So ciety of Civil Engineers for May, 1876, appeared an article by Charles Bender, C. E., in which the use of continuous bridges is strongly opposed. One of his main objections is that the theory upon which such bridges are computed is un reliable in consequence of the assump tion of a constant coefficient of elasticity, and he quotes the records of experi menters to show that values for the coefficient of iron and steel have been observed ranging from 17,000,000 to 40,000,000 pounds per square inch. These limiting values are, however, decidedly exceptional, but even grant ing that such variations may exist in materials and forms like soft iron wire, steel rails and wrought iron eye- bars, it cannot be supposed that they 62 will occur in one and the same structure, where the material is of one kind, of similar cross sections and which has been subjected in the same mill to the same process of manufacture. The mere statement that Morin has observed val ues of the coefficient of elasticity as low as 17,000,000 has very little weight when unaccompanied by any reference to the kind of iron experimented upon. Let us see what Morin himself actually says in recapitulating the results of experi ments upon wrought iron.* " Iron of superior quality, which comes from standard ores, and which has been manufactured exclusively with charcoal, or iron from sheet metal, many times refined, may give for the coefficient of elasticity values as high as E=28,- 400,000, or even E= 31, 200,000 Ibs. per square inch, equal to those furnished by ordinary steel. Iron of ordinary manu facture reduced with common coal, and drawn into forms like rails, T irons, *Morin ; Resistance des Materiaux, Paris, 1862. Vol. I, p. 443. 63 flanges, etc., give such values as E = 24,100,000 and E = 25,600,000. Finally the most soft and ductile iron furnishes values as low as E=21, 300,000, or E= 19,800,000, or even E=: 17,000,000. It is well, then, in calculations upon the strength of iron, to ascertain the quality of the material and the process of manu facture." Mr. Bender likewise alludes to ex periments upon wrought iron bars in which the coefficient of elasticity was found to be 40,000,000, and it seems to be implied by his language that such values are of common occurrence. The fact, however, that a standard writer on the strength of materials, like Morin, regards 34,000,000 as an exceptionally high figure for iron, may justify us in demanding that when a value like 40,000,000 is quoted, we should be fur nished with some details concerning the quality of such iron, the process of manufacture, as well as a description of the testing machine and the manner of measuring the small extensions or com- V 64 pressions, from which the coefficient is. calculated, or at least that we should be referred to the book or journal where such experiments are described. And further as it is well known that by strain ing a bar beyond the elastic limits, very low values of E can be deduced, are we not justified in asking similar informa tion concerning experiments which fur nish such values ? Undoubtedly there is some variation in the elasticity of different pieces of iron, even when great care has been taken to ensure uniformity of material and manufacture, and it is greatly to be desired that experiments should be made to determine how it varies with the cross section and length of the piece. As soon as such a law of variation is discovered (if any exist), we shall be obliged to consider E as variable in in vestigating a continuous truss. But if no law exists and we know only the fact that there are slight variations in the elasticity of different pieces in the same truss, what is to be done ? Nothing but 65 to regard E as constant, assured that the distribution of the variable pieces throughout the structure will be governed by the law of proba bility, and that hence the girder as a whole will conform closely to the theoretic form of the elastic line. The next argument which it is our duty to criticise in that article is, that the theory of continuous girders is un reliable, because the calculated deflection does not generally agree with the actually measured deflection. The ac curacy of the computed strains must depend upon the accuracy of the theorem of three moments, and this it is asserted depends upon the calculated deflection. And because the deflection as actually measured is sometimes no more than one-half of the calculated one, hence, it is said, the same differences may occur in the strains, and the whole theory is unworthy of consideration. This we can only regard as a striking instance of the jncompetency of prac tical men to draw conclusions from even 66 simple experiments. The reader who has followed onr presentation of the theory of the elastic line, will see at once that the value of the deflection given by (5) only enters the discussion as an auxiliary for finding (6), the tangent of the inclination angle at the support. The process supposes, indeed, that E is constant, but it supposes noth ing whatever concerning the value of the deflection at any point. When we pass to the next span and find again in (7) a second value of the tangent, the actual value of the deflection there is likewise not considered. And when by the combination of (6) and (7), we de duce (8) in which E does not appear, its very absence is a proof that the moments and hence the strains are entirely in dependent of its value or of the actual deflection. If two trusses of the same spans, height and form are continuous over several supports, one of steel hav ing a coefficient of elasticity of 31,000,000 Ibs. per square inch, and the other of wood having a coefficient of only one- 67 twentieth as much, the reactions, shears, moments and strains would be the same in each. The measurement of the actual deflections of these bridges under given loads is only useful for determining and comparing their stiffness or elasticity, or in connection with theory for finding the values of E and I. The theory of con tinuity rests not upon absolute deflec tions, but on relative ones on the form of the elastic curve, and this again upon the three universally accepted hypotheses included in our equation (1), with the additional assumption that the coefficient of elasticity is practically constant. One more remark and we close for to day a discussion which shall be resumed in our next chapter. Mr. Bender advises us to abandon the theory of flexure, to make no further advance in bridge building, to remain content with the simple lever, or at the utmost with the continuous (sic) patent hinged truss. But until such advice is enforced by more logical arguments than we have yet seen, we must continue our work 68 in support of a theory and system which is universally accepted as only slightly deviating from the exact ex isting conditions, which is applied in the erection of continuous bridges by every nation except our own, and which perhaps if carried out by us might lead to more perfect structures than the world has yet seen. The great majority of coefficients of elasticity quoted in his paper, made by such men as Staudinger, Baker, Morin and Wohler, were in fact found by measuring the deflections of beams, and then from the theory of flexure computing the value of E. He accepts those values, and on their evi dence condemns the theory by which they were deduced ! Is not Morin s conclusion, which we have quoted above, by far the most reasonable? MOMENTS AT THE SUPPORTS. The theorem of three moments given by (8) furnishes the means of finding the moments at the supports due to any as signed system of loading. The actual 69 solution of those equations is, however, quite tedious when the number of spans is large, and we proceed therefore to de velop a general solution, by which the values of the moments may be formulated and placed in a convenient form for nu merical computation. In the designing of continuous bridges it is only necessary to consider single loads concentrated at the panel apices or uniform loads extending over an entire span. Let us, then, consider a continu ous girder of constant cross-section and homogeneous material whose supports are on the same level. Let, as in Fig. 6, the supports beginning at the left hand end be designated by the indices 1, 2, 3, .... r, etc, and the lengths of the spans Z t , l# l l n etc. Call the num ber of spans s; then the last span will be / and the last support s+l. The ends of the girders rest upon abutments in the usual manner, the lengths of the spans may be all different and their number may range from one to infinity. In the span 1 T let a single load P be 70 placed at a distance k 1 T from the sup port r, (k being any fraction less than FIG. 6. unity), or let this span be loaded uni formly from r to r+ 1 with a weight W, (W being equal to w / r , if w is the load per linear unit), all the other spans be ing unloaded. By reference to (8) we notice that there are two functions of such loads which enter the equations of moments. If the single load P is alone V--- considered these functions are the first entering into the equation for the preceding support r, and the second into the one for the following support r 4- 1. If the uniform load over the whole span is alone considered, these become each equal to J W C, as we have shown above in discussing the theorem of three 71 moments. In the following investiga tion, we place therefore for abbreviation A=P 5 (2* 3 ** + * ) ) for a single B=P II (* i*) j load in span I,. (I) A \ -WZJ ) for a uniform load over B=J WZ* f whole span Z r . By introducing the letters A and B to represent these functions, our discussion will apply equally well to a single load P, or to a weight W uniformly distribut ed over the whole of a single span. Since the girder is not fastened at the abutments 1 and s-fl, the moments at those points will be zero. The moments at 2, 3, .... r, etc., we designate by M 2 , M 3 , .... M r , etc., and from (8) we may write an equation for each of those sup ports. As there is no load considered except on the span Z r , the right hand member of the equation for the support r will be A, of that for r + 1 will be B and of all the others will be zero. Thus we have the following equations : >7f 2 t The number of these equations is s 1, the same as the number of unknown mo ments. Their solution is best effected by the method of indeterminate multi pliers. Let then the first equation be multiplied by a number c 2 , the second by c a , etc., the index of the as yet inde terminate numbers corresponding with that of the M in the middle term. Then let all the equations, thus multiplied, be added, and the resulting equation be ar ranged according to the coefficients of the unknown moments M 2 , M 3 , etc. Now, if we require that such relations exist between the multipliers, that all the terms in the first member shall reduce to zero, except the last containing M 8> the value of M 8 is M= ^I/S^JL+SA (/8-i K) And the values of the multipliers will be given by the equations etc., etc After deducing the values of c from these equations, the value M 8 is at once known. Again, if we multiply the equations of moments, beginning with the last, by the indeterminate numbers d d 3 , etc., all the moments except M 2 may be eliminated, and we have and the multipliers will bo given by the equations etc., etc., etc. 74 The values of the numbers in the series c and d need only satisfy the equations as given above. Assuming then c,=l, and d, 1, we get the following : c, = <V=l 2 A+A C 3 _ ^ (II) Ci =-2c 3 3 c.= -2c,-(2c 4 + c.) J c 6 =+2c 6 -(2c B + c 4 )-^ 6 etc., etc. etc., etc. 75 which reduce to numerical form as soon as the lengths of the spans for any par ticular case are substituted. Since the 51 equations of moments are of the same form as the equations of the multipliers c and d, we must have or, if n indicate the index of any sup port, M n = c n JVI 2 when n <r -f 1 M n = ^s_ n -f2 MS, when n>r Inserting in these the values of M 2 and M 8 as found above, we have M n = (III) M n = when which give the values of the moments at all supports in terms of the quantities A and B, depending only upon the charac ter of the load and its position in the 76 span /r, and the numbers c and d depend ing only upon the lengths and number of the spans of the girder.* To find their numerical value for any given case is hence a simple arithmetical exercise. Example 1. A continuous girder has four unequal spans, ^ = 80 ft.,/ 2 =100ft., 1 3 = 50 ft., and / 4 =40 ft. (Let the reader draw the figure). On the span Z 2 is a single load P=10 tons, whose distance from the support 2 is &Z 2 =40ft. To find the moments at the supports. Since // 2 = 40, and 2 =100, we have &=0.4. Inserting then in (I), the values of &, l z and P, we find A=38400 tons ft. B- -33600 tons ft. Inserting next the lengths of the spans in (II), we have ^=0 ^=0 c,= l 4, = ! c 3.6 c?=: 3.6 * In the London Philosophical Magazine for September, 1875, where the above demonstration was first give ", the author has- extended the method to girders witn fastened or walled-in ends. 77 Since the load is in the second span r 2, also 5=4; hence r/ s _ r _f_ 2 =f/ 4 =10.3, c r = 2 =1, etc., and / 8 _ ^^^SO, etc. By- inserting these values of c, d, I, A and B in (III), we obtain M n = 82.01 c n , when n<3 M n = --24.05 / 6 -n, when n>2 For the abutment or left hand support, we have n=l r c,=0, and hence IV^ 0, for the second support, n = 2, ^ 2 =1, and M 2 =S2.01 tons ft. For the third sup port, w=3, f/ 6 _ n r= 3.6, and M 3 = 88.56 tons ft. For the next, ?i=4, r/e n : =l> and M 3 = 24.65 tons ft. Lastly, for the right hand abutment, w = 5, and M 6 0. A positive moment, it will be remember ed, causes tension in the upper chord of a truss, while a negative moment causes the reverse. 2. A girder of four equal spans has a load P at any point on the tirst span. Find the moment at each pier due to P. Ans. M 4 =if PJ (k & 3 ), etc. 78 SHEARS AND REACTIONS AT THE SUPPORTS. It is thus easy from (I), (II) and (III) to find the moments at all supports due either to single concentrated loads or to a uniform load over an entire span, and these are the only two kinds of loading which we need to consider in designing O O a continuous bridge. We next need the shear S u at the right of any support due to these same loads. In computing strains in a continuous truss we take up each span separately. The index n refers always to the particu lar span under consideration, while the index r referring to the span in which the load is for the moment considered, may be less than, equal to or greater than n. For single loads the shear S r is given directly by (3), fora uniform load P (1 k) in that expression becomes ^ wl^ while for an unloaded span those terms disappear. Thus we have -ft), for Fig. 6. (IV) S r =* for Fig. 3. 79 M n -M n , whenn>r, orn<r, &n for the shear in the span / n infinitely near to the support n. The shear in the span / n at a point in finitely near to the n+l ih support is called S n (see Fig. 3). For its values we deduce - S f = I- + P k, for Fig. 6. S r = ll + $ wfti for Fig. 3. r Q/ M n + l M n S n = - -,for n>r, or n<r. n The reaction I? n at any support n is evidently the sum of the shear S n in the span l n ; and of the shear S n _i in the span / n i O1 " f r all cases Rn = n + n ! Example. A girder of four equal spans has a single load P at the center of the third span from the left end. Find the reaction at the third support. Ans. E = 4JP. 80 SHEAR AND MOMENT AT ANY SECTION. These are given directly by the sim ple conditions of static equilibrium. For a single load we have from (2), (see Figs. 5 and 6), S = S r P, for a section between (Y) Pandr-fl S=S n> for any other section. as expressing the internal shear for any section x (see Figs. 5 and 6). Also, we have M=M r - - S r x + P (xkl r \ between (VI) Pandr-fl M=M n -- S n ce, for any other section, for the internal moment for any section x in a span either unloaded or containing the weight P. Similar expressions may be also written as in Chapter I, if the load be taken as uniformly distributed. JZxample. A girder of three equal spans has a load P at the center of the first spans. What is the shear and mo ment at the center of the middle span ? Ans. S=*P, M=ftP/. 81 MAXIMUM SHEARS AND MOMENTS. The formulae (I) to (VI) inclusive are sufficient in connection with an arithmeti cal process of tabulation to determine the maximum strains in all the pieces of a properly designed continuous truss. Having for instance to calculate tbe span /n, we may take at every panel apex throughout the bridge a single load P, and compute the shear and moment at any section due to every possible posi tion of P. These arranged in a table, afford a clear view of the distribution of loading giving the maxima ; the great est positive shear, for example, occurring when the live load covers those portions of the bridge which furnish plus values of S, while at the same tim.3, it is absent from those portions giving minus values of S. Adding then all the plus values thus found, the maximum is determined by combination with that due to the al ways existing dead load. It is therefore not absolutely neces sary that the engineer should be ac- 82 quainted with the theory of the distri bution of loading giving rise to the maxi- murn strains in the various pieces of the truss. As such a knowledge, however, is of great assistance in checking the ac curacy of the calculation, we shall here state without demonstration the cases under which such maxima and minima i arise. First the shear ; from this we obtain the strains in the webbing by the simple multiplication by a constant, a positive shear producing tension in a diagonal which slopes upward toward the left hand support. The maximum positive shear in the span l u at the section whose distance from the support n is cc, occurs under a distribution of loading such as is represented in Fig. 7, in which the shaded portions denote the live or rolling load. From this we see that the nearest span on the left and each alternate one FJG. 7. u-/ n 83 are covered with the live load ; that from the section x to the support n-f-1 the live load extends ; and that the sec ond span on the right and each alternate one are also covered with the live load ; all other portions being subjected only to the dead or actual load of the truss. The minimum positive, or, what is the same thing, the maximum negative shear obtains under exactly reverse conditions, the loaded portions in Fig. 7 being un loaded, while the empty ones receive the live load. Let the reader draw a figure for this case, and imagine the section x to move from n to n-h 1. Next the moment ; from this we ob tain the chord strains by dividing by the constant depth of the truss, a positive moment producing tension in the upper chord. Here the maximum positive mo ment in the span / n> occurs near the sup port n under a distribution of loading like that represented in the first illustra tion of Fig. 8, near the middle of the span as in the second and near the sup port tt-fl as in the last. Fig. 8 repre- 84 sents one and the same beam with the cases of loading causing maximum posi tive moments at three different sections in the span l Uf the first a section between n and a point i, the second between i and i and the third between i and w-f 1. These points i and i are called fixed in flection points, and they enjoy the pro perty that all loads on the spans to the FIG. 8. -Ln 71H 71 71*, atb. Mb. right of l Ut produce no moment at i, while all loads on the spans to the left of l n produce no moment at i . The position of these points depend only up on the lengths and number of the spans 85 of the girder, or upon the numbers c and d given by (II). If the distance from n to i be denoted by i, and that from n to i r by i 9 the following simple formulae <?n ni .,__ $s n-f-2 ^n n-j-2 < n-f 1 will give the position of the fixed^inflec- tion points in any span n . In order to render these distributions of load clear, let us imagine the sec tion x to move from the support n to Ti+1. When the section is at n the live load covers the whole span n to render the moment at a; a maximum, as x passes toward i the load recedes rapidly toward n + 1, until when x reaches i the span l n becomes empty, and the loads on the following spans shift as shown in Fig. 8. As x passes from i to i r the span l u remains empty as in the second sketch and when it reaches i the loads on the preceding spans shift. As soon as x passes i the load begins to come on 86 at n, which rapidly increases as x moves, until it covers the whole span when x coincides with n+l. The arrangements of loading for caus ing the maximum negative moment in any section depend likewise upon the position of that section with reference to the fixed inflection points, and are in all cases exactly the reverse of those for the positive moment. It will be seen, then, that the maxi mum moments between the supports and the fixed inflection points cannot be de termined by cases of loading, for such cases are different for every section. In a giider of two equal spans for example, one of these points in each span coin cides with the abutments, the others are at one-fifth the length of the span from the pier. Between those points the maximum strains are not to be found by parabolic curves of moments drawn from a few assumed arrangements of loading. Here have some late writers fallen into grave error. The above completes, what seems to 87 us a simple presentation of the theory of the continuous girder of constant cross- section. We have disconnected it en tirely from the properties of the simple girder, have avoided the use of artificial angles and couples, parabolic moment and shear curves, static and elastic reac tions and other paraphernalia which are too often introduced to complicate the subject. The formulae (I) to (VI) which may be written on a page of the engi neer s note book include indeed the whole theory, and are sufficient for the determination of the maximum strains in a continuous truss of any number or lengths of spans. 89 CHAPTER III. We will now apply the above theory to the practical calculation of the strains in a continuous trus, and to show the perfect generality of our method we will take one of t ftve unequal spans. Fig. 9 shows the relative lengths of the several spans, each support and span receiving an index according to the no tation previously adopted. The first span Z, has a length of 70 feet, the FIG. 9. It 72 3 Iff- 15 y & """nr" ^ & / * 3 * 56 second l^ of 100 feet, the third l % of 80 feet, the fourth 1 4 of 120 feet and the last J t of 90 feet. This girder is 90 to be subject to a live load of 0.8 tons per linear foot per truss ; the dead load we estimate at 0.6 tons per linear foot per truss. It is divided into panels of ten feet each and its height is also ten feet, the webbing being a simple series of isosceles triangles as shown in Fig. 10, which represents the FIG. 10. <u I, = SO- , A BC D E 3 r is H tc I span 1 3 enlarged. The live load is ap plied at the panel points on the lower chord. It is required to calculate the maximum strains in all the pieces of the span 1 9 due to the above live and dead loads. We take up first the live load of 0.8 tons per linear foot, or eight tons per panel. Since every load in the span ^ affects every section in 1 3 in a similar manner, we may, instead of considering 91 the separate panel loads on / take them as uniformly distributed in the prelimi nary determination of the shear and moment at 3. (The load of eight tons at the points 1, 2, 3, etc., give reactions only at those points, and cannot affect the span Z 3 ). On the span /, there are seven panels and six apices, hence the live load in that span is W, = 6XS = 48 tons. In the same way we have on the spans J a , / 4 and Z 6 , to consider the live loads applied at the panel apices as uni formly distributed over the spans; but in the span 1 3 we must consider each panel load separately, since different arrange ments of those loads give maxima for different sections. Thus we have On 1 19 the load W 1 = 48 tons, On ^, the load W a =72 tons, On * 3 , the loads P,, P u , P 3 , etc., (see Fig. 10) each equal to 8 tons, On Z 4 , the load W 4 =88 tons, On Z 6 , the load W 5 =64 tons. We now turn to formulae (I) of the 92 preceding chapter, and determine the quantities A and B due to each of these loads. For that on /, we have, for ex ample, W t =48, /\=:4900, hence A = B =58800. For P t we have P=8, l\ 6400, k%, hence A= 10500, and B= 6300 ; in like manner, for P 2 , P 3 , etc., we place &=t, ^^f, etc., and find for each a value of A and B. Thus we have for the several loads : For W t , A=B= 58800 For W 2 , A=B= 180000 ForPj, A= 10500 B= 6300 ForP,, A=16800 B=12000 ForP 3 , A=19500 B=16500 ForP 4 , A=19200 B = 19200 ForP 6 , A 16500 B = 19500 ForP 6 , A-12000 B = 16800 ForP 7 , A= 6300 B= 10500 For W 4 , A=B = 316800 ForW 6 , A=B= 129600 We next turn to formulae (II), and sub stitute the lengths ^ = 70, /,= 100, etc., and thus obtain 93 c,=0 d, * "* 04. /7 Q K 3 ^ ? "3 *** 9 c= 44.567 < 54.8 o 5 as the values of the multipliers c and d for the case under consideration. We are now able to find from (III) the moments M s and M 4 at the supports 3 and 4 for each of the above loads. Since there are five spans, 55, d n = 54.8, c7 8 _i 16, etc. ; substituting these in (III), we have r -,whenn<r-fl *r c r -fr r 4_i M n = -^ T -n- -, when n>r 1 t U O A n being any index (in our case either 3 or 4), and r the index of that span, which, for the moment, we regard as loaded. Taking the load on the first span, we have r=l, and since n>r, we use only the second of the above formu lae, which becomes 94 A c, 4- B <r M n - -rf T _ n _-i__ - 3.452 <7 T _ n Substituting in this TI 3 and w 4, and we have M 3 = 3.452 (16) = 55.23 tons ft. M 4 = 3.452 (3.5) = 12.08 tons ft. Taking the load in the second span, we have r=2, and -,.- . M n = -^ 7 _ n -- 3 - in which, by making n=3 and n=4, we get the values of M 3 and M 4 . For the single loads on the span 1 31 we must use the first of formulae (III) to obtain M 3 , and the second to obtain M 4 . Making then r 3, we have, 3.4 3 17032 17032 _ A ^ + B C 4 3.5 M ^2 // 4 (16A-3.5B) 17032 17032 (-3.4A + 14.05 B) and inserting in these the values of A and B, as given above, we find the mo ments at 3 and 4 from each of the loads from P x to P 7 . For the load on the fourth span, we make r=4; for that on the fifth span, r=5; and the first of for mulae (III) give the moments. Thus, by very simple arithmetical work, we obtain the moments M 3 and M 4 due to each of the single loads in Z 3 , and each of the uniform loads in the exterior spans, and arrange them in the second and third columns of the following table : Load M 3 M 4 83 tons ft. tons ft. tons. w, - 55.23 + 12 08 0.841 W 2 +40"). 82 - t^.77 - 6.182 p, + 81.19 h 10.83 - 7.254 P 2 4- 45.36 - 22.85 - 6.281 - 50.85 - 83.93 - 6.211 pi r 48.00 - 41.92 - 4 078 p.. -3D. 15 - 44.65 - 2.9U PC ~ 2(5.64 - 39.92 + 1.834 P 7 - 12.81 - 25.85 + 0.837 W 4 - 158.10 -653.34 -10.143 W 8 + 25.87 -106.91 + 1.660 The last column of the table, which gives the shear S 3 in the span 1 3 at a 06 point infinitely near to the support 3, is found from the quantities M 3 and M 4 by means of formulae (IV). The load W s , for example, gives a positive moment of 405.82 tons ft. at 3, and a negative one of 88.77 tons ft. at 4. From the last formula of (IV), we have then 80 Also for the load P 3 on the span / 3 , we have P=8, &=i, and from the first for- / o / mulae of (IV) p _ 50.85-33.93 (1 ax fl 91 , tnns , 80 and in the same way the other shears in the last column are computed. All of these refer, of course, only to the live load of eight tons per panel. We are now ready to proceed with the computation of the maximum strains in the span 7 S . And first we take up the webbing. The maximum strain in any diagonal in Fig. 10, is equal to the maximum shear for that section multiplied by the secant of the angle between the diagonal and a vertical. We proceed first to find the maximum shears. The shear at any section due to the dead load is constant, increases or de creases as the live load comes upon the bridge, and becomes a maximum or minimum under certain particular distri butions of loading. To determine these o it is only necessary to tabulate the shear due to each separate load. This is easily done from the values of S 3 and the formulae (V). In the following table the vertical column headed aB/> includes the shears which may act upon the diagonals a B and B b, b C c those for b C and C c, etc. The horizontal column numbered 1 gives then the shears at every section due to the live loads ; the load W, for example producing a negative shear of 0.84 tons in every panel or S=S 3 , the load P 3 giving in the three panels on its left S S. = +5.21 tons and in the five 5 on its right S = S 3 P= + 5.21 8 2.79 tons. A mere inspection of this table rH fci) o o> ^ CO w <5 W w OQ M GO T-I t- .- t- - "i O T t T 1 r -, CO GO to 01 o TO CO Ol O CO o O CO O T- 1 + -" Ol T- TJ 1C CO > C 1 1 5 T-H -+ *> GO W GO > > t- Z -. O T-. GO r ~t sO GOO 01 - a 7 O CO O r- OJC 5 1C CO C ; + GO i-l I 00 1 s 00 T-I L^ t- 1 Ci O i- C 1 - CC "* T 5 O GO GO T- s CO s 11 o Ol CO CO O CO- O T- 1 + I O? C- "++ S 5 + O 1C +* * -f GO 1C O o co o T- 1 + ? Ci O " Ol C * CO CO -f T ^ol o c +++^ (^ + -^ CO co o 53 01 CO ^ o o ^f Ol GO 1C p "* OO 1C O" 00 ^* C- C? > C5 00 CO CiO Tf ^ CO C O C^ GO -X) r-i CO t> CO T 1 r 1 GO CO o 4- GO GO -rf< Oi GO 1C i r 1 + 1 t+] T++++^ + 00-tiS^O30ctI3o CO 1C GO co cc O O O rr o o o T- l + l C "^ O? O O T I +++++7 + O> CO +7 C5 CO -f C5 CO OA 1 -*" X> JC GO T-H X C>0 CO ^* Tr t^ -o T-I CO O i> GO 1-H CO 01 T- ( + =1 + 1 ++++++7 + Cl T-I +7 1 . sssgsssssss O GO Ol C5 GO Ol 01 " o GO T < Ol 01 1C 1C > O CO ?" TO 1C "^ Ol T ( O O i -H-H--H-++7 + CO 7C T-I + 1 *V--v=.v sV Ss CO 5 CQ T; oJ O ei o Q + 1 t-t 1 CO 99 shows the distributiorr of live load caus ing the maximum or minimum shear in c_? any section. Thus for the panel c?Ee the maximum occurs when those loads giving positive shears are present, namely, W,, P 4 , P., P., P,. and W. and when all the others are absent, and the minimum occurs when only those giving negative shears are on the bridge. If then we add all the positive quantities in 1 and likewise all the negative ones and place the results in the horizontal column 2, we have for the panel dl&e, 4-17.52 tons and -16.24 tons as the greatest and least shears which can occur in that panel due to the live load, and these need only to be combined with the shear due to the dead load to obtain the absolute maximum and minimum. If the dead load be regarded like the live load as concentrated at the panel points on the lower chord, its effect will be a fractional part of that of the live considered as uniformly distributed. Adding algebraically the quantities in 2 we have in 3 the shears produced by a 100 uniformly distributed live load of eight tons per panel, since this is the same as taking the algebraic sum of all the quan tities in 1. The live load if extending over the whole bridge will then produce in dTSe a shear of 4-1.28 tons, and since the actual dead load is three-fourths of the live, the dead load must produce in ttuit panel a shear equal to fxl.28 = 0.96 tons. Taking then three-fourths of the quantities in the horizontal col umn 3 we have in 4 the shears due to the dead load of six tons per panel. The shears in 4 always must exist, while those in 2 may exist under certain positions of the live load. The absolute maxima arc therefore found by adding algebraically the quantities in those two horizontal rows. Thus for<#Ee, -f-0.96 always obtains, and if +17.52 also oc curs their sum 4-18.48 is the positive maximum shear ; and if -16.24 occurs, 4-0.96 16.24 = 15.28 is the minimum or negative maximum. Placing these results in column 5 we have the required maximum shears in every section of the 101 span under consideration. If the dead load shear in 4 is greater than the live load shear of opposite sign in 2 only one kind of shear can prevail ; thus in bCc the greatest possible value is +12.96 + 29.01 =+41.97 tons, the least possible is +12.96 --11.73 = + 1.23 tons and the diagonals b C and Cc will be subject to only one kind of strain. In the case be fore us three panels cDr7, c/Ee and e~Ff have both a positive and negative maximum and the diagonals in those panels may be subject to either tension or compression. The maximum shears in 5 multiplied by sec. 9, or the secant of the inclination of diagonal to vertical give the maximum strains, tension of the diagonal slopes up ward toward the left hand support, com pression if it slopes downward. The depth of the truss being ten feet and the half panel length 5 feet, sec. 6 is 1.118. We have then the following table of maximum strains in the 102 DIAGONALS. (See Fisr. 10.) Ba B6 Od Cc DC T>d Ed HA KA JLk 61.7 tons +61.7 46.9 +46.9 2 or +7.2 tons 2 or 7.2 20.6 or +17.1 +20 6 or 17 1 +28.2 or 9.4 28.2 or +9.4 +40.5 40.5 +.",4.0 54.0 +688 68.3 in which + denotes tension and com pression. In the same manner we may tabulate the moments at every section due to P each load and deduce the maximum chord strains. For the upper chord panels A B, B C, C D, etc., the centers of moments are at the opposite vertices a, b, c, etc., and hence in formulae (VI) we must take 0, 10, 20, etc., as the suc cessive values of x. To find the moment for CD due to the load W, on the 103 span ,, we have only to insert # 20, and the values of M 3 and S 3 as found above for that load (see Fig. 10), giving M = 55.23 - (0.841 X 20) = -- 38.41 tons ft. the negative sign denoting that W, causes compression in the upper chord. Also to find the moment in the same panel due to the load P , we have as before z=:20 and M = 31.197.254X20 + 8X10 = 33.89 tons ft. In this way we readily compute the mo ments for every panel due to every live load and arrange them as in the follow ing table of MOMENTS FOR UPPER CHORD. (See Fig. 10.) W r H GO oc v c ^ C5i > G Ci C OC ft H^ > o GO O5 t* TH O T 1 O7GOC r- GO r + 1- ^O7CO -rCSiCC - O7 CO rf ^ CO O7 L.I <. w < C C, T- 9 9 - 1C H coc GO r t r. l - M w O Oi -: fr- r-l t GO O CO or *> GO CO GO CO GO 07 T1 GO CO CO CO 1C Ci 2 07 CO C - lO C <? rr 1C C lO co c t^_ ** 1.^^ * - % HH HH O 3 - D C5 T- K^T CO 4 CO GO O * Tf -^ JC Z> ^ O < ^ GO - T- r - r- H CO CO C 7 CO 1C - O Cf O 7C O t- d I*** 1 ^ GO C I! 5 2 1C f ( 1 r- CO O T-I C0_07 d V P 7 J> tf -CO^r- H O r H r- -i CO 1C CO T* O7 TH : co - - c: T- - -: ; - i *C X 1- r C-- O C, T r 3: Ci t- 507^^ CO > ~. a ^ ^ \ 3 H + + o co *C CO "Vji ^^ + 1 H V f- - * * 1C O O CO rH i t- - f>- CO iO rH O7 1 1 o rH O7 -t GO c: ^ 7 1C I-* T; | : * r * - tC C5 - r- t- T rn D cr t c pi O7 ^ "^ + > 4 O 2 J, - m ^- > ? If : CO "* P - c " ^"fr- ^ ^ ^^ ^ H rH Tf CO GO ^ H CO C i Ci CO CO ~f ^ 5 O :> 07 O 07 - - > - - : ic : _ ^ -f c: o . t> Tt - D O7 co r i T^ ^^ O rH j ! ?> .CO r ^ < Q * \( h J CO.TJ1 P 01 7 H 07 3 _ ? -^coicc: ?GfeO D Ci* > O CO t- 07 07 rH 5 o > r 2 GO C 07 1 r 7 : ? ; u : cc o c 5 CO O7 - 2 CO ^ ?> 07 a CO ^ H 3 ! + CO CO O7 + o* n y r }O^ I 1C CO O7 GO 00 ^ t - CO t> + + r -^ ^ CO H t< - -- I ( t> C5 GO ** C 5 C5 66 c: CC r- : - PQ O7 GO O7 ^ 3" 5 O 07 ^ 5 Xr- < d CO CO 07 i- CO CO T-t o TH O O rH JO O GO Si O O7 y C O CO ^n" 1C ^r CO O7 i zjL-H-t-H-H-- J l!t ft 1C C y^ CO O > S V---v> V -~ A > ^ l~> CC a rt O + 1 T-I 05 CO !"* 105 The horizontal column 1 of this table shows at a glance the distribution of live load giving the maximum strain in any bay : in the bay B C for instance the loads W 2 W 5 and P 4 to P 7 inclusive produce positive moments, and hence the greatest tensile strain obtains when those loads are on the bridge and all others are absent, while the least tensile strain in B C occurs when W 1? W 4 , P,, P, and P 3 are present and the others ab sent. Adding separately therefore these positive and negative moments, we have in 2, the greatest and least moments for every bay due to the live load. Adding those algebraically and we have in 3 the moments when the live load covers the entire bridge ; taking three-fourths of the quantities in 3 we have in 4 the mo ments due to the actual dead load. Lastly combining the moments in 4 which always must exist with those in 2 which may exist, we get in 5 the absolute maxima and minima. For example, in CD we have due to the live load the moments +327.0 and -281.3, the sum 106 of these or +45.7 is the moment when the live load extends over the whole girder, three-fourths of this or -f 34.2 is the value due to the actual dead load, and finally + 34.2 -f 327.0= -f- 361.2 tons ft. + 34.2 281.3:^247.1 tons ft. are the maximum positive and negative moments, C D may then be subject to two kinds of strain. Dividing these results by the depth of the truss and remembering that a posi tive moment causes tension in the upper chord, we have the maximum strains for the UPPER CHORD. (See Fig. 10.) AB -104.0 tons BG - 547 CD - 36 1 or 24.7 tons DE - 33.2 or 44 9 EP - 30.1 or 41 I FG - 4~>.0 or 43.3 GH - (>0.2 or 21.4 Hit - 860 KL -140.3 107 From this we see that the whole upper chord may under certain positions of the rolling load be subject to tension. This is due to the short length of the span compared with the adjacent ones. The calculation of the maximum strains in the lower chord is entirely similar, the centers of moments being at the opposite vertices B, C, D, etc., and the corresponding values of x being 5, 15, 25, etc. We leave therefore as an exercise for the student the formation of the tabulation, merely giving the results to which it will lead, viz. LOWER CHORD. (See Fig. 10.) ab 76.9 tons be 41.4 gry "+" ^"l cd 34.7 or - -34. 6 tons de 84.7 or - -478 ef 40.7 or - -47.1 fg 51.9 or - -34.0 gh 70.7 lik 127.9 The strain sheet for the span l z (Fig.9) is now finished, and in a similar way each of the other spans may be comput ed. The method we have presented is entirely general and applicable to any number of continuous spans, whether equal or unequal. In each case we take the load in the exterior spans as uniform and that in the span under consideration as applied at the panel apices, and find for each the quantities A and B from (I) and from the lengths of the spans we find by (II) the multipliers c and d. These enable us to deduce from (III) the moments at the supports due to each load, from which (IV) give us the shear. It is only in this preliminary computation of moments and shears that the calcula tion of continuous girders differs from that of ordinary simple trusses. In the latter the moments at the ends are known to be zero, and the shears coincide with the reactions which are found from the law of the lever. The simple truss is thus but a particular case of the continu ous one as may be readily seen by placing 5=1 in our formulae (I) to (IV). In an end span of a continuous truss the mo- 109 ment at the abutment is also zero and the shear is the same as the reaction.* GIRDERS WITH SPANS ALL EQUAL. This is one of the most common cases. Making in (II) all the Ps equal, we have c^ d^~ c,=d t = 1 C 3 " i = 4 C 4 =<= 15 c s =rf.= -56 c 6 =d 9 = 209, etc., which are the well-known Clapeyronian numbers first deduced by the discoverer of the theorem of three moments. f Each of these numbers is equal to four times the preceding one less the one next pre ceding, and their signs are alternately positive and negative. They are here seen to be a particular case of our gener al formulae (II). * For an example of the computation of a continuous truss of two spans, see Van Nostrand s Engineering Magazine for July, 1875. t See Comptes JKendus, 1857, p. 1076. 110 GIRDERS WITH SYMMETRICAL SPANS. If the two end spans of the bridge are each equal to ft I and the others each equal to I, the multipliers c and d become also equal. Their values are c b d^ 26 -- 30 /3 c G d^ 97 + 112 /?, etc., each being equal to four times the pre ceding one less the one preceding that. Having established the value of ft (usual ly taken at about 0.8) these reduce at once to numerical form. If ft be unity the spans become all equal and the mul tipliers reduce to the Clapeyronian num bers. Example. A girder of four spans has a single load P in the second^ span at a distance k I from the second support ; the two end spans being equal to 0.8/ and the central ones to I. Find the mo ment at the second support. Ans. M = P/0.52* 0.901 7 Ill CONSTANT AND VARIABLE CKOSS SECTION. Having computed the maximum strains in a continuous truss, we choose for the various pieces cross sections of an area and form sufficient to resist those strains, thus making the girder one of uniform strength. The theory by which we have computed the strains supposes however that the cross section of the girder is constant. The question now arises what error is introduced by this hypothesis. As we have been unable to present in the short limits of this paper the theory of the continuous girder with variable cross section, we cannot place before the reader a mathematical comparison of the two cases, and are hence obliged to re ly on the computations of others and on general considerations. Computations of strains in continuous girders have been made by Bresse, Mohr, Winkler, Weyrauch and others, consid ering the cross section both constant and variable. The general conclusion to be derived from their investigations is, that 112 the maximum moments over the supports are greater, when the variable cross sec tion is taken into account, but rarely more than six per cent, that the maximum moments near the centers of the spans are generally slightly less, and that the shearing forces do not sensibly differ.* For example, in a truss of two equal spans, the maximum moment at the pier is 0.125 wl* for constant cross section and 0.133 wl* for variable ; the maximum negative moment is 0.070 wl* for constant and 0.067 w I* for variable, and the reac tions of the pier are 1.250 iv I and 1.266 w I respectively. If then we compute continuous trusses as if they were of constant cross section, we are liable to slight errors in the chord strains. These strains are however com puted on the assumption of a distribution of live load which can never occur in practice, and in proportioning the sec- * Mohr. in Zeitschrift des Arch. u. Ing. Ver. zu Hanno ver, 1860, 1862, Winkler ; Die Lehre von der Elasticitdt, p. 150. Weyrauch ; Theorie der continuirlichen Trdger, p. 22., p. 143. 113 tions to those strains a factor of safety involving five or six-fold security is in troduced. Considering then that it must be almost impossible for the live load to be arranged on the bridge as Fig. 8 represents, we may be well as sured that our computations on the hy pothesis of constant moment of inertia give greater strains than can ever obtain. After having computed on both hypo theses a girder with four spans, two of sixty-five meters in length and two of fifty-two meters, Weyrauch says : " We are now able to answer the question, whether it is allowable to calculate con tinuous girders with variable cross sec tion by the formulae for constant cross section, in the affirmative. The maxi mum moments arising from the two cal culations differ but slightly. In our ex ample, where the cross sections vary be- tw T een 1 and 24, the greatest difference was 6 per cent, the next following only 2.7 per cent. The shears change scarcely at all. Only for bridges with extremely long spans, is it desirable to make a sec- 114 ond computation on account of variable cross section." LEVELS OF SUPPORTS. In Chapter I we alluded to the fact that small changes in the relative levels of the piers produce great variations in the strains of the pieces of the truss. Continuous girders should not for this reason be used when the piers are liable to settle. Whether the supporting points are exactly upon level when the bridge is built is a matter of no importance, al though, of course, no great differences can be allowed. If in finding the equation of the elas tic line we had considered the supports on different levels, a term containing those differences of level and the term El would have .entered the theorem of three moments. Now if a straight beam be laid across two points, a downward force is necessary in order that it should touch a third point at a lower level. If this downward force be furnished by the 115 weight of the beam, a certain part of this weight will be effective in producing the deflection or satisfying the term E I, while the remaining part will act exactly as if the three supports were on the same level. But if the beam, instead of being originally straight, were of such a shape that it exactly fitted the three points it is in the same condition as the horizontal one after it has undergone the deflection. Hence its action is independent of the variations in level, for no exterior force is required to compel it to correspond with the points of support. We therefore conclude, that if a bridge be built, by suitably adjusting its false works, corresponding to the profile of the piers, all the strains obtain exactly as if those piers were on one and the same level. Except the well-known rule that the lengths of the spans should be so adjust ed that the cost of the piers and super structure may be equal, there is little to 116 be said upon this subject. Although most writers give a mathematical discus sion of the most economical relations of spans, the fact that no two of them agree except in the simplest case, only indicates that the theory contains no principle which will lead to general con clusions. Viewing the matter from a practical point of view, but not neglecting the in vestigations of mathematicians, we may give the following rules. For two spans the lengths should be equal. For others the span should be symmetrical, the in terior ones being equal and. the end ones shorter by about one-fifth or one- sixth, or making the central ones each equal to /, the end ones should be about f I or I L Such an arrangement equalizes the mo ments due to the dead load and being pleasing to the eye, it is advisable to re gard it in designing continuous bridges. PRACTICAL CONSTRUCTION. In the construction of continuous bridges, the following points should be carefully observed : 117 1. The iron should be of a uniform qnality, and have undergone as near as possible the same process of manufac ture. 2. The truss should be built with par allel chords, each capable of resisting either tension or compression ; the web bing should be simple. With double and triple systems of webbing the strains cannot be accurately determined. 3. Joints in the chords should never be made over the piers. 4. The false works should be so adjust ed that the bridge may be built with its points of support on the same relative level with the actual bed plates. 5. The bearing surface of the bridge upon the piers should be as .small as pos sible consistent with considerations of strength and safety, and arrangements for longitudinal variation, due to changes in temperature, should be provided. 6. Continuous girders cannot be used if the piers are liable to variations in level after erection. 118 ADVANTAGES OF THE CONTINUOUS SYS TEM. In favor of the continuous girder versus the simple one, we may mention : 1. Greater stiffness, since the deflec tion under a rolling load is much less than that of independent simple spans. 2. Ease of erection in cases where false works are difficult and expensive ; the girder may then be built on shore and pushed out over the piers. 3. Saving in material for the piers, since a less bearing surface is required than for two ends of single span bridges. 4. Saving in iron, amounting to from twenty to forty per cent, over the ordinary construction of single spans. 5. Simplicity of construction, when an angle of skew exists in the piers: in such cases the cross girders may be placed at right angles in the continuous structure, and the difficulties of oblique connec tions entirely avoided. In a simple girder, whose length is /, and live load per unit of length w, the 119 maximum deflection due to the live load is ^r-f- In two continuous spans each 384 El equal to /, the maximum deflection, which occurs when one span is covered with 3.7 iv r the live load, is - -^TTJ or only three- 384 xL 1 fourths as much. With many continu ous spans, the deflection is much less, its greatest value being in the end spans. In the case of a girder with horizontally fastened ends, the deflection is only one- fifth of that of ordinary simple spans. The saving in iron is large, and alone sufficient to recommend the continuous system, particularly for long spans. This saving occurs wholly in the chords where material can best be spared. In the web bing the quantity of material is slightly increased. The exact percentage of sav ing depends upon the number and lengths of spans, the proportion of live to dead load, the arrangement of webbing, and will be the same in no two cases. In the ex ample of Fig. 9, the center span which we computed does not afford scarcely 120 any saving in material owing to the in fluence of the larger adjacent spans. For girders of two hundred feet in length with spans nearly equal, calcula tion indicates a saving of about thirty per cent. For the extreme case a sin gle span with horizontal fixed ends the saving is fifty per cent. DISADVANTAGES OF THE CONTINUOUS SYS TEM. In our last chapter we referred to an article by Charles Bender, C. E., which contains many ingenious arguments against the use of continuous bridges. There are, in fact, fifteen " conclusions to which he is led and which may be seen by the reader on p. 109 of the cur rent volume of The Journal of the American Society of Civil Engineers. Of these we will give a short abstract and append a running commentary. 1. The theoretical calculation of curves of moments, without consideration of proportions and details, is exceedingly 121 fallacious. Other things being equal, the calculation of strains does furnish an estimate of amounts of material. The method of curves moreover is neither so accurate nor so quick as our process of tabulation. 2. This fallacy will be greater, if the theory stands upon false premises. Un doubtedly. 3. The theory of continuity is fallaci ous and unreliable, because it supposes the coefficient of elasticity constant, whereas it has been shown that it may vary for wrought iron from 17,000,000 to 40,000,000 Ibs. per square inch. In our last chapter we have shown that the proper interpretation of the variability of E is, that different qualities of iron have different degrees of elasticity. The values of E, from which it is concluded that the laws of flexure are fallacious, were in fact found by the theory itself from the measured deflections of beams, and it is hence more than fallacious to regard them as condemning that theory. 4. With several diagonal systems the 122 strains in a continuous girder cannot be calculated, but only guessed. A good objection and it applies also to simple girders and draw spans. 5. The theory needs a correction if the chords are made variable in cross section. The amount of this correction we have indicated above. 6. The calculation of continuous trusses is excedingly tedious, and if gen erally introduced would greatly impede the business of bridge building in this country. A conclusion to which those who know how to calculate decidedly ob ject. For a construction costing half a million dollars, it matters little whether one day or one week be spent in compu tation, and if the one week saves ten per cent or more on the cost, it is certainly well employed.* * In the discussion of this subject at the late Engineers 1 Convention, one of the speakers mentioned an instance where the strain sheet for a continuous revolving draw bridge was furnished for $40. As that strain sheet was made by the author of this paper it is not improper that he should remark that its price would have been consid erably less bad the computations been made by tabula tion of apex loads, instead of by the consideration of cases of loading. 123 7. Pieces which resist two kinds of strain must be proportioned to resist the maximum tension plus the maximum compression. Further experiments are much needed in this connection ; if Woehler s conclusions are confirmed, pieces must be so proportioned and hence the percentage of saving lowered. 8. Continuous girders require very accurate workmanship. Are we to infer that ordinary trusses do not ? 9. The foundations and masonry of the piers must be of excellent quality. Does not the same hold for the simple girder? See No. 11. 10. If the girders are built on shore and pushed out over the piers, additional computations must be made and extra pieces introduced. The additional com putations are of the simplest character and in many cases no extra pieces would be needed. 11. If improperly placed on their bed plates, greater strains arise than are con templated, an inch difference in level producing great variations in strains. 124 We have shown above that unless piers are liable to settle after the erection of the bridge, such differences in level pro duce no effect. 12. If one chord be protected from the heat of the sun the strains are much disturbed. The camber of the bridge is altered, as also occurs in single spans. It should be remembered however that although certain strains cause a curva ture, a certain curvature does not neces sarily cause corresponding strains. 13. Continuous bridges have proved to be more economical in Europe than simple spans, because the latter have been improperly proportioned. Trial is necessary to prove that they would or would not be more economical in this country. 14. By designing two bridges of 200 feet each, a two span continuous truss twenty-five feet high, and a simple truss 27 feet high, with different details, Mr. Bender finds that the latter is more economical Perhaps for other propor tions this might be reversed. In a com- 125 parison of amount of material, ought not other things to be taken equal ? 15. Continuous bridges deflect as much as single spans. Not if they have the same height and span, and are subject to the same loads ; a fact which every schoolboy knows. In regard to objections 2, 6, 7, 8, 9, 10, 12 and 13 one remark further is neces sary. They are objections which may be made to every new proposed con struction in engineering art. In the days of wooden bridges, they were ad vanced against the use of iron ; they have been made against the suspension system and against the braced arch. But their value can be estimated in only one way, by trial. On the other hand theory can estimate one at least of the advantages claimed for the continuous systems, and that estimate is a saving of twenty to forty per cent, in material; how much of this must be deducted for extra care in workmanship, labor of erec tion, effects of temperature, or varia tions in the elasticity of the material can 126 only be determined by the actual erec tion of continuous bridges, by experi ments extending through a long series of years. mr Having thus stated briefly, but fairly, the arguments for and against the use of continuous bridges, we leave it to prac tical builders to decide whether or not the system is worth a trial. Other na tions have long been using it ; profiting by their experience and by our own im proved methods of manufacture and modes of erection, it may, perhaps, turn out that we shall find it better and more economical than the present system of single independent spans. HfSTORY AND LITERATURE. The literature on the theory of con tinuous girders is very extensive in the German and French languages, and very limited in English. We can only give here a few hints concerning its develop ment and history. About the year 1825, Navier founded the present theory of flexure by intro- 127 ducing the hypothesis that the exten sions and compressions of the fibers on each side of the neutral axis were pro portional to their distances from that axis. From this he deduced the equa tion of the elastic line, and applied it to the discussion of continuous girders. His method consisted in determining first the reactions of the supports, and from these the internal forces or strains, which, although the most logical, was exceedingly tedious in practice. In the following works the reader may find de tailed information concerning his meth od : Kayser : Handbuch der Statik, Carls- ruhe, 1836, chap. X. Molinos et Pronnier : Traite de la construction des ponts metaliques, I^aris, 1857. From the time of Navier to the pub lication in 1857 by Clapeyron of the method of using the moments over the piers as auxiliaries in the computation instead of the reactions, many continu ous girders were built in France, Ger- 128 many and England. Among these may be mentioned the Britannia tubular bridge, of four spans, two of 231 feet, and two of 460 feet ; the Boyne Via duct with three spans of 141, 267 and 141 feet ; and the bridge over the Weichsel at Dirschau with six continu ous spans, each 397 feet in length. By Clapeyron s happy discovery of the the orem of three moments, a great impetus seemed to be given toward the erection of such bridges, for in the twenty years following 1857, we find them extensively built in Germany and France, although in England the unfortunate example of the Britannia tube caused a tendency to other forms of construction. A mere list of such bridges would occupy pages. They are generally of shorter span than those mentioned above, rarely exceeding 300 feet, while the number of spans varies from two to seven. To give here a list of the books which has appeared since Clapeyron s time will also be impossible. We can only indi cate two or three which are at the same time valuable and easily accessible : 129 Bresse : La mecanique appliqu&e, Paris, 1865. (Vol. Ill contains a tolera bly complete mathematical discussion, with tables for facilitating the calcula tion of moments.) Winkler: Theorie der J5rM6c&6?i, Vienna, 1872. (Contains the graphical method of Culmann and Mohr, with also analyt ical investigations.) Weyrauch : Theorie der continuir- lichen Traeger^ Leipzig, 1873. (A com plete analytical discussion of the whole subject. The best work which has yet appeared.) In our own country have been issued during the past year the works of Greene, Herschel and DuEois, each of which contains valuable contributions to the literature of the subject, and which should be in the library of every student of engineering. On the other hand, works by English authors, which treat of the subject at all, do it in such an imper fect and unsatisfactory manner, that we are forced to consider them as twenty years behind the age. 130 As the above mentioned works treat only of the theory and calculation of continuous girders, we ought perhaps to say that a book giving an account of the most important continuous and simple, together with suspension and arch bridges, is the admirable descriptive work by Heinzerling, Die BruecJcen in Eisen: Leipzig, 1870, which is illustrat ed by over a thousand engravings, and presents a complete history of iron bridge construction. ERRATA. Page 15, line 13 ; for ^ read I. 11 21, "17 and 18 ; for wl 9 read wl*. 14 DAY USE RETURN TO DESK FROM WHICH BORROWED LOAN DEPT. This book is due on the last date stamped below, or on the date to which renewed. Renewed books are subject to immediate recall. 2Jun 58 OS ~~ YA ( fOut UNIVERSITY OF CALIFORNIA LIBRARY