ENGINEERING LIBRARY INTERNATIONAL CHEMICAL SERIES H. P. TALBOT, PH.D., Sc.D., CONSULTING EDITOR TECHNICAL GAS AND FUEL ANALYSIS Engineering Library COPYRIGHT, 1913, 1920, BY THE MCGRAW-HILL BOOK COMPANY, INC. PRINTED IN THE UNITED STATES OF AMERICA f THE MAPUE PRES9 YORK PA PREFACE TO SECOND EDITION The years intervening since the appearance of the first edition have seen a distinct increase in our knowledge of the subjects covered in this book. Standard Methods for Sampling and Analysis of Coal and Coke and also for Examination of Liquid Fuels have been studied and approved by several technical societies. Methods for analysis of gases have been critically studied and new technique developed by the staffs of various organizations, notably the Bureau of Mines and the Bureau of Standards, and by a number of individual workers. The litera- ture has been carefully reviewed for this edition and the inclusion of new material has increased the size of the book about twenty per cent. No attempt has been made to describe all of the new methods nor to illustrate all of the various forms of apparatus now on the market. On the contrary the effort has been to illustrate types and to indicate essential features. Criticism has been solicited from teachers who have used the book as a text and the needs of students have been kept in mind, but the intention has also been to make the book one which would supply the practicing engineer and chemist with the most necessary information. ANN ARBOR, MICHIGAN, May, 1920. PREFACE TO FIRST EDITION With the increased demand for the economic utilization of fuels has come an increased necessity for accuracy in testing both the raw fuel and the manner of its utilization. Our knowl- edge has recently been greatly extended by the investigations conducted by the Committee on Calorimetry of the American Gas Institute, the International Photometric Commission, the Joint Committee on Coal Analysis of the American Chemical Society and the Society for Testing Materials, the Bureau of Mines and the Bureau of Standards. The author has aimed to present the conclusions of these committees and also to indicate where there has been marked dissent from them. He desires to express his especial appre- ciation to Professor 0. L. Kowalke of the University of Wis- consin for his courtesy in revising the chapter on Determination of Heating Value of Gas; to Professor S. W. Parr of the Uni- versity of Illinois for suggestions on Calorimetry and Chemical Analysis of Coal; and to Dr. H. C. Dickinson of the Bureau of Standards for his suggestions on Calorimetry. ANN ARBOR, MICHIGAN, August, 1913. vii CONTENTS PAGE PREFACE v CHAPTER I SAMPLING AND STORAGE OP GASES . . . % Difficulties involved The problem of a fair sample Materials for sampling tubes Types of sampling tubes and their use Aspirators Solubility of gases in water Saturating water in sampling tanks Collecting an average sample representative of a definite period Collecting a representative instantaneous sample Storing gas samples. CHAPTER II GENERAL METHODS OP TECHNICAL GAS ANALYSIS 14 introduction General method The gas burette Care of stop- cock Saturating the water of the burette Drawing the sample of gas into the burette Measuring a gas volume Calibration of a gas burette Gas pipettes Connecting the burette and pi- pette Details of a simple gas analysis Accuracy of the analysis. CHAPTER III ABSORPTION METHODS FOR CARBON DIOXIDE, UNSATURATED HYDRO- CARBONS, OXYGEN, CARBON MONOXIDE AND HYDROGEN .... 28 Carbon dioxide Unsaturated hydrocarbons Oxygen by phos- phorus Oxygen by alkaline pyrogallate or hydrosulphite Oxygen by ammoniacal copper solution Carbon monoxide Absorption of hydrogen General scheme of analysis. CHAPTER IV EXPLOSION AND COMBUSTION METHODS FOR HYDROGEN, METHANE, ETHANE AND CARBON MONOXIDE 43 Available methods Apparatus for explosion analysis Manipula- tion in explosion analysis Oxidation of nitrogen as a source of error Accuracy of explosion methods Hydrogen by explosion ix X CONTENTS PAGE Hydrogen and methane by explosion Carbon monoxide, hydrogen and methane by explosion Quiet combustion of a mixture of oxygen and combustible gas Fractional combustion with palla- dinised asbestos Fractional combustion with copper oxide Detection of minute quantities of combustible gas Oxygen by ex- plosion or combustion Nitrogen Form of record of gas analysis. CHAPTER V VARIOUS TYPES OP APPARATUS FOR TECHNICAL GAS ANALYSIS. . . 4 Introduction Schlosing and Holland's apparatus Orsat's appa- ratus Bunte's burette Chollar tubes Instruments for recording carbon dioxide in flue cases Methods of gas analysis depending on thermal conductivity Gas analysis by optical methods. CHAPTER VI EXACT GAS ANALYSIS 74 Historical General methods Corrections for temperature and pressure Description of gas burettes The bulbed gas burette for exact analj'sis Manipulation of gas burette for exact analysis Calibration of burette Absorption methods in exact gas analysis Carbon dioxide Unsaturated hydrocarbons Oxygen Carbon monoxide Hydrogen Methane Errors in calculation of re- sults of explosion and combustion Nitrogen CHAPTER VII HEATING VALUE OP GAS , . . . 92 Introduction Continuous flow calorimeters Wet gas meters Corrections for temperature and pressure Control of the water Measurement of temperature Measurement of mass of water Description of calorimeter Preliminaries of a test Description of a test Calculation of observed heating value Total heating value Net heating value Calculation of total heating value Ac- curacy of method Determination of humidity of air Non- continuous water heating calorimeters Automatic and recording gas calorimeters Calculation of heating value from chemical composition. CHAPTER VIII CANDLE-POWER OF ILLUMINATING GAS 119 Introduction Method of rating candle-power The bar photo- meter Standard light Photometric units Standard candles CONTENTS xi PAGE The Hefner lamp The Pentane lamp Secondary standards of light Standard gas burners The Bunsen and Leeson photometric screens The Lummer Brodhun photometric screen The flicker photometer The gas meter The photometer bench and its equip- ment Details of a test Illustration of calculation The photo- meter room Jet photometers Accuracy of photometric work. CHAPTER IX ESTIMATION OF SUSPENDED PARTICLES IN GAS 138 Introduction The distribution of particles in the cross-section of a straight main Mean velocity in the cross-section of a gas main Influence of bends in a main Velocity of gas in a sampling tube The filtering medium Estimation of suspended tar and water Electrical precipitation of suspended particles CHAPTER X CHIMNEY GASES 144 Introduction Sampling Formation of carbon dioxide Effect of hydrogen of coal on composition of chimney gases Carbon mono- xide and products of incomplete combustion Volume of air and of chimney gases Loss of heat in chimney gases Interpretation of analysis of chimney gases. CHAPTER XI PRODUCER GAS 156 Formation of producer gas Sampling producer gas Analysis of producer gas Interpretation of analysis Heating value of producer gas Volume of producer gas Efficiency of a gas producer. CHAPTER XII ILLUMINATING GAS AND NATURAL GAS 163 Introduction Sampling General scheme of analysis Chemical composition of illuminating gas Benzene Benzene and light oils by differential pressure method Hydrogen sulphide Total sul- phur compounds Naphthalene Ammonia Cyanogen Specific gravity Natural gas Gasoline in natural gas. CHAPTER Xlll LIQUID FUELS . . 187 Introduction Sampling Heating value Specific gravity Mois- xii CONTENTS PAGE ture Proximate analysis Suspended solids Flash point Gaso- line Specifications for motor gasoline Kerosene Fuel oil. CHAPTER XIV SAMPLING COAL 204 General consideration Difference in composition of lump and fine coal A scoopful as a sample Influence of lumps of slate-^- Taking a sample Mine sampling Preparation of sample Preser- vation of sample Usual accuracy of sampling Standard methods of sampling Sampling coke Reliability of samples. CHAPTER XV THE CHEMICAL ANALYSIS OF COAL 222 Introduction Proximate analysis Preliminary examination of sample Air-drying Grinding and preserving the sample for analysis Moisture Volatile matter Ash Fixed carbon Sul- phur Ultimate analysis Carbon and hydrogen Nitrogen Phosphorus Oxygen Methods of reporting analyses Accuracy of results Slate and pyrites Standard methods for the laboratory sampling and analysis of coal Standard methods for the labora- tory sampling and analysis of coke. CHAPTER XVI HEATING VALUE OF COAL BY THE BOMB CALORIMETER 258 General methods of determining heating value The calorimetric bomb Details of the calorimetric bomb Thermometers Prepa- ration of sample Manipulation of bomb calorimeter Thermom- eter corrections Radiation corrections Corrections for oxida- tion of nitrogen Corrections due to oxidation of sulphur Correction due to combustion of iron wire Reduction to constant pressure Water value of calorimeter Adiabatic calorimeter Precision calorimetry Accuracy of results Total and net heating values. CHAPTER XVII HEATING VALUE OF COAL BY THE PARR CALORIMETER AND OTHER METHODS 282 Introduction Combustion in a stream of oxygen The Thompson calorimeter The Parr calorimeter Preparation of Parr calori- meter Care of sodium peroxide Operation of Parr calorimeter CONTENTS xiii PAGE Corrections to be applied with Parr calorimeter Accuracy of Parr calorimeter Calculation of heating value from chemical analysis. APPENDIX 290 Saturation pressure of water vapor Reduction of gas volumes Factors for reduction of gas volumes Relative humidity Correc- tions to be applied to observed heating values of illuminating gas and natural gas Emergent stem corrections for calorimeter ther- mometers Corrections for difference between inlet water tem- perature and room temperature when determining heating value of gases Table for constants of certain gases and vapors Mean specific heats of gases Volume of water vapors taken up by one cubic foot of air Baume* scale for liquids lighter than water. INDEX . 305 TECHNICAL GAS AND FUEL ANALYSIS CHAPTER I SAMPLING AND STORAGE OF GASES 1. Difficulties Involved. The problem of obtaining a repre- sentative sample of a gas for analysis presents in many cases more difficulties than the analysis itself. If the gas flowing through a main were of perfectly uniform composition throughout its cross-section and also throughout its length the problem would simplify itself to the introduction of a tube through which a por- tion of gas might be removed for analysis. The proposition becomes at once more complicated when it is necessary to sample gases which have not passed through any adequate mixing chamber and which usually travel through the main or flue in pulsations of widely varying composition. Ocular evidence of this condition is afforded by a glance at the average smoke stack with its pulsing billows of smoke. If it is desired to determine not only the com- position of the fixed gases but also the amounts of suspended solids, tar particles, water globules, etc., the problem becomes one of great complexity, capable frequently of only partial solution. This question is discussed separately in Chapter IX. 2. The Problem of a Fair Sample. Gases should be sampled as close as possible to the point of the reactions which are to be studied, thus minimizing errors due to leakage of air or gas, deposition of solids or secondary reactions. Gases travel through straight pipes in an irregular succession of waves of rather a spiral form, the velocity being greatest at the center of the pipe and least next to the walls. The shape of the wave is altered by every bend, branch or other change in the pipe and the point of maximum velocity is also shifted. The temperature of gases from hot furnaces also varies throughout the cross-section of the 1 2 GA3 AND FUEL ANALYSIS conducting pipe, being usually hottest where the velocity is greatest and coldest next to the walls and in dead bends. It is in general advisable to sample from a point of approxi- mately average velocity and temperature, but it is not possible to find a single point from which a truly representative instantaneous sample can be drawn. It is necessary to extend the sampling period over a time which will on the theory of probabilities allow so great a series of gas waves to pass the sampling tube that the resulting sample will be truly representative. A small volume of gas can therefore only be considered a representative sample when it has been drawn from a practically homogeneous mass of gas a condition which is rather closely fulfilled in illuminating gas which has been purified and further made uniform through mixing and diffusion in a large gas holder. The condition is not fulfilled in producer or chimney gases. No fixed time can be set as the most desirable for a single gas sample. If the sample is being taken from chimney gases which are supposed to be the same from one hour to another, the sam- pling process may very well extend over one hour or six hours. The longer the period the more nearly will the sample be an average one. The same thing holds true for illuminating gas coming from a holder, but it would not hold true for producer gas from a single producer nor for illuminating gas from a single re- tort, for these gases vary continuously in composition, each fresh charge of coal commencing a new cycle. When dealing with gases of this sort it is necessary to start and stop the sampling with refer- ence to some definite point in the cycle. A truly proportional sample of a constantly varying gas cannot be obtained without rather elaborate precautions. 3. Materials for Sampling Tubes. Sampling tubes must be of a material which will not react with the gases and change their chemical composition. Iron, and in general metals, cannot be used at high temperatures. Iron reacts with carbon dioxide quite rapidly at 200 C., producing carbon monoxide and oxide of iron. It also reacts readily with oxygen. An oxidized pipe will react with hydrogen causing that gas to disappear as water at equally low temperatures. Pipes with rough surface will induce a rapid catalytic decomposition of such gases as ammonia and some of the less stable hydrocarbons at a rather low red heat. SAMPLING AND STORAGE OF GASES 3 The best materials for sampling tubes are glass, quartz or porcelain. These tubes should be protected by a wrapping of asbestos in the form of paper or twine and should be slipped into an iron jacket which takes the strain off the tube and prevents sudden temperature changes. Glass begins to soften about 600 C. or 1100 F. and on long exposure to a red heat devitrifies and ultimately cracks spontaneously on cooling. Porcelain will stand about 1000 C., but at higher temperatures the glaze com- mences to soften. Unglazed porcelain will stand a higher tem- perature and special refractory mixtures may be obtained which will not soften at 1700 C., but such tubes are apt to be porous and should only be used if proved to be free from leaks. Fused quartz is in some respects admirable as a material for sampling tubes, since its coefficient of expansion is so small that it never cracks because of temperature changes. It will stand 1100 C. without softening but on long heating tends to become crystalline and brittle. The ordinary opaque electroquartz tubes are often porous and should be carefully tested for leaks before they are used. 4. Types of Sampling Tubes and Their Use. The simplest form of sampling apparatus consists of a single tube through whose open end the gases are aspirated. When sampling from a large flue the sampling tube is sometimes closed at the end and perforated at various points along its length with the expectation that gas will be drawn uniformly through the holes at intervals across the flue. A device of this sort usually fails of its purpose and is often less efficient than a single tube. If the perforations are all of the same diameter they will allow the same amount of gas to pass through only in case the suction is the same on each. This condition can be attained in practice only by making the perforations small and the suction high so that it will be practically as great at the far as at the near end of the sampling tube. But such small holes stop quickly with dust and are not practicable. A better device to secure a more uniform sample from the gases of a flue is shown in Fig. 1. It consists of a bundle of glass tubes of the same diameter but of varying lengths which are wired to- gether and slipped into an iron pipe not quite so long as the short- est glass tube. The glass tubes are cemented into this with neat 4 GAS AND FUEL ANALYSIS Portland cement in such a manner that their ends next the threaded joint are all even and about 2 in. from the end of the iron tube. This may readily be accomplished in the following manner. Plug the threaded end of the iron pipe with a cork and stand it vertically on the corked end. Plug one end of each of the glass tubes with putty and stand them in the iron tube with their lower ends resting on the cork. Fill the iron tube for 3 in. with a paste of Portland cement and water and allow to stand for twelve hours to harden. The sampler is finished after removing the cork and putty by screwing onto the iron pipe a cap which is provided with a 1/4-in. nipple to which convenient connection may be made. The suction required to draw a slow stream of gas through a smooth tube 2 ft. long is only slightly more than that required for a 1-ft. tube. The empty space at the end of the pipe acts as a mixing chamber and allows a fairly satis- factory sample to be drawn through the nipple. The sampling tube is to be inserted in the flue in such a way FIG. 1. Multiple gas sampling tube. that there will be no leakage around it. An iron gas main may be drilled and tapped to receive a threaded nipple into which the sampling tube is luted when in use and which is closed by a cap when not in use. When a hole must be cut through a brick wall a threaded nipple may be cemented permanently into the wall and closed, when not in use, with a cap. If only a single sampling tube is used it should be inserted to about the point of mean velocity of the gases, a subject which is discussed in Chapter IX. Where a multiple sampling tube of the above type is used it should be inserted so that the longest tube reaches at least to the center of the flue. It is frequent but bad practice to connect the sampling tube and the aspirator by a rubber tube. Rubber is softened and burned by hot gases and yields gaseous products which contami- nate the sample. It also dissolves tar and hydrocarbon vapors and even permanent gases like ethylene in sufficient amount to materially change the character of the gas. Furthermore, even SAMPLING AND STORAGE OF GASES 5 the best rubber is never gas tight but always allows some leakage. It is usually necessary to use a small amount of rubber tubing in making connections, but its use should be restricted to lengths of only an inch or two where it serves as a connector between glass or metal tubes. When thus used a copper wire should be twisted tightly around the rubber where it slips over the glass tube to ensure a tight joint. A tightly stretched rubber band may be used instead of the wire. Small lead or seamless copper tubes are flexible and more satisfactory than glass where the tempera- ture is not so high that the gases react with the metal. 5. Aspirators. The commonest form of aspirator consists of a bottle or tank initially filled with water which flows slowly out and is replaced by the gas. With this form of aspirator the suction constantly decreases as the head of water drops and there- fore the flow of gas slackens, which is a disadvantage. This may be overcome by making the entering gas pass down through a central tube and then bubble up through the water on the principle of the Marriotte bottle, but unless the water of the aspirator has been previously thoroughly saturated with the gas this process is certain to change the composition of the gas very materially. The extent of the error introduced by failure to observe this precaution is discussed later. A film of heavy par- affine oil is sometimes poured on the water of the aspirator bottle to prevent interaction between the water and the gas. Such an oil-film does not, however, prevent the interaction and does not even make it much slower. If oil is used it should be at least as heavy as a lubricating oil so that it will not give off any appreciable volume of vapors which even in small amount will cause difficulty in the estimation of oxygen by phosphorus. 011 should not be used when the gases sampled contain hydro- carbons, since these are quite soluble in heavy mineral oils. Where frequent samples are to be drawn a pair of aspirator tanks as shown in Fig. 2 are to be recommended. Tanks to hold approximately a cubic foot should have the cylindrical portion 12 in. in diameter and 12 in. high with the cones each 6 in. high. They may be made of galvanized iron and will be strong enough with merely soldered seams if they are not knocked about too much. The ends of the cones terminate in short 1/2-in. nipples soldered in. The lower end of the tank carries screwed to this nipple a 1/2-in. cock F, anc! beyond this another 1/2-in. nipple 6 GAS AND FUEL ANALYSIS with a hose union. The upper nipple carries a 1/2-in. tee on whose side-arm is an ordinary 1/4-in. gas cock E, and on whose upper end is a cap tapped for a 1/4-in. pipe. This pipe, cut the length of the tank, is threaded into the cap from the inside be- fore the latter is in place so that when screwed together the 1/4-in. pipe runs the whole length of the tank and projects through the cap enough to receive the gas cock D. With a pair of bottles of this sort the water flows from one to the other and is not exposed to the air so that it remains saturated with the gas. If a water or steam line is at hand an injector may be employed as an aspirator. It is desirable in most cases to have the suction a slight and steady one, a condition not readily attained with injectors running only to a small per cent, of their capacity. It is well to have the injector open more widely than is necessary to furnish the suction required and to have on the line an automatic water seal relief valve as indicated at E in Fig. 3 which will suck in air at the inlet of the aspirator if the suction rises too high. 6. Solubility of Gases in Water. The solubility of gases in water varies with the temperature and the pressure. For our treatment it will be sufficient to assume that the gas is always under atmospheric pressure and at ordinary room temperatures. Under these circumstances the solubility of the common gases is as follows: SOLUBILITY OF GASES IN WATER Expressed as the volume gas dissolved by unit volume water at the tem- perature of the experiment when in contact with the pure gas at atmos- pheric pressure. 1 770 ^ Carbon dioxide 1.070 0.826 Oxvcen 0.036 0.031 Nitrogen 0.019 0.016 Carbon monoxide 0.027 0.023 Methane 0.039 0.033 Ethylene 0.147 0.119 When mixed gases are in contact with water the amount of each gas dissolved will be in direct proportion to its volume and its solubility. Thus the composition of pure water saturated SAMPLING AND STORAGE OF GASES Composition of air, per cent. Volumes of gases in 100 vols. water Percentage com- position of gas dissolved by water CO 2 0.04 O 2 21.0 N 2 79.0 0.04X1.07 =0.043 vols. CO 2 21.0 XO. 036 =0.756 vols. O 2 79.0 X 0.019 = 1.500 vols. N 2 1.9 32.9 65.2 Total dissolved gas =2.299 vols. 100.0 The composition of the gases in water saturated with chimney gases may be calculated in the same manner. Composition of gas, per cent. Volumes of gases in 100 vols. water Percentage com- position of gas dissolved by water C0 2 O 2 10 10 10X1.07 =10.70 10X0.036= 36 85.0 2.9 N 2 so 80X0 019= 1 52 12.1 12.58 100.0 The change in the carbon dioxide from 10 per cent, of the gas sampled to 85 per cent, of the gases dissolved in water shows how serious are the errors which can arise in sampling. If the sam- pling tank originally filled with pure water should be only half emptied so that it would be half filled with gas of the above composition and half filled with water, and then it should be allowed to stand till equilibrium were attained, 7 per cent, of the gas would be dissolved by the water with the following result. Original composition of gas, per cent. Composition gas after standing over water, per cent. Composition of gases dissolved by water, per cent. CO 2 10.0 5.2 73.4 O 2 . . . 10 10 4 5.0 N,. ..80.0 84.4 21.6 The above illustration shows that in chimney gases it is not at all impossible to have an error of 50 per cent, hi the C(>2 through carelessness in sampling. Errors of similar nature, al- though hardly of such large proportions, will arise with other gases. With illuminating gas it is the important class of illuminants of 8 GAS AND FUEL ANALYSIS the ethylene series which is most seriously affected. If it is worth while taking a sample at all it is worth while to saturate the water with which the gas is to come in contact. 7. Saturating Water in Sampling Tanks. When the gas to be sampled is under pressure it may be bubbled through the water contained in a bottle, best only about two-thirds full of water and loosely stoppered. The air of the bottle will soon be replaced by the gas which will thus act on the constantly changing surface of the liquid as well as on the surface exposed to the bubbles. An occasional vigorous shake will greatly accelerate absorption. Under these conditions the water will become practically saturated in fifteen minutes. It is not wise to attempt to saturate water by bubbling the gas through it while in an open beaker, since the constant presence of the air above the liquid defeats the very object aimed at. The water in tanks of the form shown in Fig. 2 may be readily saturated if suction is available by connecting tank 1 to the gas main as shown and connecting the suction pipe to cock E of the same tank. The gases will then bubble through the water and pass out E. Where suction is not available the procedure is as follows. Start with tank 1 full of water and tank 2 empty. Fill tank 1 with gas and connect the cocks E on the two tanks with rubber tubing. Raise tank 2. Water will flow from 2 into 1 and gas from 1 to 2. When each is approximately half -full close the valves and shake. Finish passing the water from 2 into 1 and disconnect the rubber tube from E of tank 2. Draw a fresh tankful of gas into 1, allowing that in 2 to escape into the air. Connect the cocks E with rubber tubes as before and repeat the process of dividing the gas and water between the two tanks and shaking. After three such operations the water will be suffi- ciently saturated. 8. Collecting an Average Sample Representative of a Definite Period. -It is frequently desirable to determine the average composition of gases flowing through a flue for a period of time which may be 15 minutes or 24 hours. If the period is not longer than an hour the use of two cu. ft. tanks as shown in Fig. 2 is satisfactory, it being assumed that only the true gases are of importance and that suspended solids and condensable vapors are either absent or unimportant. A represents the multiple sampling tube projecting into a flue. B is a tube loosely packed SAMPLING AND STORAGE OF GASES 9 with cotton or asbestos to filter out dust. C is a bubbling tube to indicate the rate of flow of the gas. The first step in a test is to saturate the water of the tanks as previously directed. At the close of this operation tank 1 should be full of water and tank 2 of gas and the sampling tube and the filter should be full of gas. To commence the sampling operation D is opened fully and F is opened slightly until gas bubbles through C at the rate desired. It is a mistake to open widely the lower stopcock on the sampling tank and control the flow of gas by partially closing the upper stopcock, since with this procedure the gas in the tank is under FIG. 2. Apparatus for aspirating a sample of gas from a flue. an unnecessarily reduced pressure and there is an unnecessary risk of leakage. The gas in tank 2 escapes through the open cock at the top. The pressure gage at G will indicate if the sampling tube or the filter becomes choked. The rate of gas flow should be so adjusted that tank 1 will be almost filled with gas at the end of the period. There should still remain enough water to act as a stirrer for the gas when the tank is shaken vigorously to mix the contents. This mixing is a simple precau- L 1 1 1 I J -I . i 1 f, , ,1 i 1 10 GAS AND FUEL ANALYSIS mix by diffusion but the rate is a slow one and it is never safe to rely on diffusion to give a fair sample. After the gas in tank 1 is mixed a portion may be transferred through cock E to a labora- tory gas holder for analysis. To get ready for the next sample the E cocks on both tanks should again be connected by the rubber tube and the gas transferred to tank 2 so that the water in tank 2 will remain saturated. If the sampling period is to extend for much more than an hour the flow of gas through the sampling tube becomes so slow that a multiple sampling tube is not to be relied on. The use of the additional apparatus shown in Fig. 3 remedies the difficulty where a continuous aspirator is available to draw a rapid stream of gas through the sampling tube. The gas filter, bubbling tube and sampling tanks of Fig. 2 are to be attached to cock A of Fig. 3. The sampling tanks may then be set to take as slowly TuM FIG. 3. Continuous gas sampling apparatus. as may be desired a portion of the representative rapidly flowing gas stream. In Fig. 3, B is a pressure gage to indicate any obstruction of the sampling tubes, C is a bubbling bottle to give a visual control of the rate of the stream, D is a gas meter, E is a pressure control and F a pressure gage on the line from the aspirator. The suc- tion on the sampling tube as shown by gage B should be only a few tenths of an inch of water. It is difficult, however, to get an aspirator to work properly when nearly shut off so that the aspirator is opened enough to work efficiently and the suction on the line is kept down by the regulator E. The suction on B may be regulated by the depth to which the tube in E is immersed. An orifice meter may be substituted for the dry meter shown in the installation at D. Such an apparatus may be made of metal or may be made quite simply from glass tubing. 1 1 Benton, Jour. Ind. and Eng. Chem., xi, 623 (1919). SAMPLING AND STORAGE OF GASES 11 The title of this section calls for the collection of a sample representative of a definite period. The foregoing procedure only accomplishes it approximately. In order that the sample should be truly representative a given proportion, say one-tenth of 1 per cent, of the gas, should be constantly passing through the sampling tube. If the flow of gas in the main dropped to one- third of its former rate, the flow of gas through the sampling tube and into the sampling tank should also decrease. The ideal to be striven for is to have in the small sampling tank gas of the identical composition that there would be in a large gas holder where all of the main gas stream had been gathered and mixed. This can only be accomplished by a sampler which takes repre- sentative quantities as well as qualities of gas. The sampling apparatus described is supposed to work at a constant and not a proportionate rate, but it does not even do this accurately. It is not, however, possible to take a truly proportional sample without great elaboration of equipment and the method described is usually sufficient. 9. Collecting a Representative Instantaneous Sample. The problem of a representative instantaneous sample is in some ways simpler than that involved in collecting a sample to represent a longer period. If the apparatus shown in Fig. 3 is attached to a multiple sampling tube a sample drawn at A should be fairly representative. If desired an Orsat burette or similar apparatus may be attached permanently at A. 10. Storing Gas Samples. Samples of gas obtained as directed in the preceding section are too large for ready transportation to the laboratory for analysis. It is usually convenient to trans- fer a portion to a small glass gas holder which is advantageously of the type proposed many years ago by Hempel and shown in Fig. 4. This consists of a bulb terminated above by the gas inlet of capillary tubing bent in the form of a U to allow a water seal and at the bottom by a larger tube for the water supply. The transfer of gas is accomplished in the following stages. Attach at cock E of Figs. 2 and 4 a glass tee carrying at A a rubber connection and a small funnel and at B a rubber con- nection. Fill the gas holder with water taken from the large sampling tank 2 of Fig. 2, which is saturated with the gas, and connect the gas holder at B as shown in Fig. 4, with the exception 12 GAS AND FUEL ANALYSIS of the screw clamps at A and B. Open valve E and blow gas through the tee and out A to expel air. Close E and force water from the small gas holder into the funnel. Tighten clamp A. Fill the glass gas holder nearly full of gas and close E. Open A cautiously and allow water to flow from the funnel and fill the capillary seal of the gas holder. Close clamp B and disconnect the gas holder from the tee. Elevate the levelling bottle so as to put the gas in the holder under slight pressure, and close the clamp C at the bottom of the holder. The gas is now stored in a FIG. 4. Glass gas holder for storing samples. glass vessel closed at the top by a capillary tube filled with water, which in turn is closed by a clamped rubber tube at the top of the capillary. At the bottom the holder is similarly closed by a column of water. The gas is thus in contact with nothing but glass and water and if the latter has been previously saturated, SAMPLING AND STORAGE OF GASES 13 may be preserved without change for months. The above form of gas holder is always reliable. The galvanized iron sampling tanks shown in Fig. 2 may also be made of glass on a smaller scale and used for storage of gases. They are more convenient where samples are to be shipped since they do not have the pro- jecting bent capillary which is liable to be broken in shipment. Gas holders of the floating type are not to be relied upon for there is always diffusion of gases through the water, and after several hours evident changes may be found. There are nu- merous forms of gas holders where stopcocks are relied upon to prevent leakage. Stopcocks may be tight or they may not and anyone who has had experience with the irritating doubts attend- ant upon their use will prefer not to trust them more than is necessary. The same thing holds true of rubber stoppers and connections. Gases may, however, be transported quite safely in glass bottles with rubber stoppers provided a little water is left in the bottle and the stopper is wired tightly in and covered thickly with paraffine. The bottle is then to be kept upside down. Under these conditions the small amount of water in the bottle forms a seal on the inside and the paraffine a seal on the outside, re-enforcing the rubber. When samples of this sort are to be shipped by express they should be packed in crates without a top so that care will always be taken to keep the proper end up. Where gases have to be stored a long time and especially where they must be shipped, the truly safe way is to collect them in glass tubes drawn out at each end and fuse the ends in a blowpipe flame. CHAPTER II GENERAL METHODS OF TECHNICAL GAS ANALYSIS 1. Introduction. Gas analysis is an extremely useful method for controlling the operation and checking the efficiency of many industrial operations. All of the manifold industries using fuel as a source of heat and almost all industries engaged in producing fuel or utilizing it in any way find in gas analysis a valuable assistant. There are very many special industries which require the analysis of gases which are peculiar to themselves and not connected with fuels, but though the general methods of gas analysis here given will often apply to these cases, no attempt will be made to develop those applications which might better be taken up in connection with a study of the particular indus- tries. This chapter will consider only the gases arising from the utilization of fuels. The purpose for which the analysis is desired will influence the methods to be employed and the kind of apparatus to be used, for time consumed and liability to errors increase rapidly with the number of constituents to be determined. The analysis is to be made as simply as possible, small percentages of gases unimportant for the purpose of the analysis being neglected and groups of related gases being frequently determined together. In boiler firing and in the operation of all furnaces heated by combustion of fuel, the variations in the percentages of carbon dioxide, oxygen and carbon monoxide show at once the changes in efficiency of the furnace. In the operation of gas producers these three constituents with the added determination of hydro- carbons and of hydrogen suffice for most purposes. When gas is to be sold to consumers, as is the case with a city gas supply, a more complete analysis is often necessary and not infrequently some uncommon and minute constituents must be sought as an explanation of trouble. This adaptation of the means to the end is a characteristic of technical analysis, which seeks only the particular information 14 GENERAL METHODS OF TECHNICAL GAS ANALYSIS 15 of value for the purpose in hand. On this account gases from different sources will be considered separately, although there may be much in common between them. There are various operations common to almost all processes of gas analysis which may well be considered before passing to special cases, and such general considerations form the subject of this chapter. 2. General Method. The method preferably followed in technical gas analysis requires a measurement of the initial volume of the mixed gas, the absorption of a single constituent by an appropriate reagent, and the measurement of the new volume, the constituent absorbed being determined by difference. Where no suitable absorbent is known for a given constituent it is desirable to transform the gas into some more suitable compound. It is necessary in order that these changes in volume due to absorption be correctly noted, that the temperature and pressure of the gas be known at each step, and it would be ideal if both the temperatures and pressures could be kept absolutely constant throughout the process. This can never be accom- plished, although by special apparatus described in Chapter VI on Exact Gas Analysis errors due to change in external tem- perature and pressure during an analysis may be automatically eliminated. This procedure is, however, more complicated than is necessary for technical work where it is usually suf- ficiently accurate to make the assumption that the temperature of a water-jacketed burette in a laboratory does not vary during the hour that may be required for an analysis, and that the barometric pressure does not change in the same period. Since this simpler procedure is sufficiently accurate for most technical purposes, it will be described first. The historical development of the present apparatus and methods is discussed in Chapter VI on Exact Gas Analysis. 3. The Gas Burette. The gas burette has its zero point at the top and is usually of 100 c.c. capacity. It is closed at the top by a stopcock or sometimes simply by a rubber tube and a pinchcock and should always be enclosed in a water jacket. Almost all forms of gas burettes will answer the above descrip- tion, but there are many differences of detail, the most impor- tant being the style of the stopcock which closes the burette at the top. The form of apparatus which has been evolved in 16 GAS AND FUEL ANALYSIS the laboratory of the University of Michigan 1 and which has given good service during the past ten years is shown in Fig. 5, which illustrates the apparatus as it appears in service. Details are given in Fig. 6. The burette-stand may be im- provised from an ordinary iron stand, one of whose rings of ex- ternal diameter slightly greater than the water jacket of the burette has been provided with a brass collar, thus making a Rubber Stopper 100 c.c. divided c.c. ^ Rubber Stopper FIG. 5. Gas burette and pipette. FIG. 6. Details of gas burette. cup in which the rubber stopper of the water jacket rests without binding. Another ring large enough to slip loosely over the water jacket serves to keep the burette vertical. A segment is sawed out of the front of this ring to allow an uninterrupted view of the graduations, and it is wrapped with chamois skin until it fits as snugly as desired. A spring clamp made from sheet brass makes a more elegant upper support. By this simple arrangement the burette may be raised, lowered, or swung to 1 White and Campbell, /. Am. Chem. Soc,, 27, 734 (1905). GENERAL METHODS OF TECHNICAL GAS ANALYSIS 17 one side at the convenience of the operator, and may be tipped in any position while carrying, without danger of breakage. By reference to A of Fig. 6 it will be seen that the body of the burette is a perfectly straight tube. It is closed at the bottom by a one-hole rubber stopper, which need not even be wired in, unless the burette is to be filled with mercury. To clean the burette, all that is necessary is to take out the rubber stopper and lift the burette and its jacket out of the rings when it may be turned upside down, swabbed out as an ordinary burette, and then swabbed with a clean dry muslin which is more efficient than a wet cloth in removing the film of grease which causes the drops to hang to the glass. Caustic soda may be used or chromic acid, but they are not usually necessary. 4. Care of Stopcock. The stopcock of a gas burette is very carefully ground and polished so as to be as nearly gas tight as possible. Particles of grit getting into the capillary opening are apt to cut a groove around the stopcock so the opening is bored diagonally in order that the grooves worn in this manner may be parallel and non-connecting. This device is of assistance but the stopcock must still be handled carefully. To keep in good condition remove the stopcock from the burette, wipe it off with a dry cloth and see that the capillary openings are clean. Do the same with the seat into which it fits in the burette and the capillaries with which it connects. Rub a thin coat of some lubricant over the stopcock. Vaseline is inferior. A material made by melting one part of best black rubber at as low a temperature as possible and stirring into it one part of paraffine and one part of vaseline answers well when it is carefully made. The author has found anhydrous lanolin the best material. The lubricated stopcock should be pressed gently into the dried seat and turned a couple of times, when the space between the stopcock and the seat should appear perfectly translucent without air bubbles or any discontinuity. If too much lubricant is used it will work itself into the capillaries and clog them. If either the stopcock or the seat is wet the lubricant will not adhere well. These stopcocks are expensive and it is advisable to fasten them to the burette with a fine copper wire attached to the stopcock so loosely that it can turn readily. It is not advisable to use a rubber band for this 18 GAS AND FUEL ANALYSIS purpose since it sticks to the handle of the stopcock and exerts a torsion sufficient sometimes to turn the cock at an inopportune time. The stopcock should always be loosened in its seat when the burette is to be put away and it is admirable policy to always clean it and make it ready for use again before setting it aside. 5. Saturating the Water of the Burette. The liquid filling the burette is almost always water which allows much more rapid and convenient and in some ways more accurate manipula- tion than mercury does. It should be saturated with gas similar to that which is to be analyzed. If the gas is air this precaution may often be omitted, since distilled water is usually saturated with air. The precaution should not be omitted with other gases, especially those like flue gases rich in the more soluble carbon dioxide. The errors due to solubility of gases are discussed more fully in Section 6 of Chapter I. Place 200 c.c. of water in a flask or bottle of about 400 c.c. capacity and lead in a stream of gas through a glass tube passing through a loosely fitting cork, shaking occasionally. If the supply of gas is limited a smaller flask may be used and the loose cork may be pressed down tight after the air has been displaced. Shaking the flask facilitates absorption. It is not necessary that the water be absolutely saturated and five minutes is usually ample time for the operation. 6. Drawing the Sample of Gas into the Burette. The gas to be analyzed is assumed to be contained in a gas holder of a type similar to that shown in Fig. 4 of Chapter I, but the directions for use of this type of gas holder may readily be adapted to other types. The tubing to connect the burette and gas holder should be capillary so that the air contained in it may be completely swept out without wasting much gas. A tube of 1 mm. internal diameter answers well. It should not be rubber because rubber absorbs heavy hydrocarbons from gases rich in these bodies and gives them back again to gases containing them in but small amount. It is necessary to make one rubber connection at each end of the capillary tube but the surface of rubber exposed to the action of the gases should be reduced to a minimum by bringing the ends of the glass capillary tubes into direct GENERAL METHODS OF TECHNICAL GAS ANALYSIS 19 contact with each other, or, if it is necessary to leave a pinch- cock on the rubber, as close to each other as possible. The glass tubes should be cut off square and the sharp edges softened in a flame sufficiently to prevent the cutting of the rubber but not enough to constrict the capillary opening of the glass tube. The rubber tube should fit tightly to the glass and be of the best quality black gum rubber, free from any internal ridge where the seam has been joined. The inside of the rubber tube may be moistened with water or preferably with glycerine to make it slip more readily over the glass. The rubber connec- tions to the capillary should be bound with wire as a further safeguard against leakage. Soft copper wire of about 22 B. & S. gage is suitable. Finer wire is too apt to cut the rubber. Heavier wire is apt not to pull up snugly. Copper wire of 24 B. & S. gage insulated with cotton and paraffined like that used for annunciators, is the best material as it is strong enough and does not cut the rubber. It is a mistake to wrap the wire several times around the tube. A piece of wire about 2 in. long should be wrapped once around the tube. The crossed ends are then to be grasped close to the rubber with a pair of pliers and tightened with a single half turn of the wrist. No rubber joint can be relied on to be gas-tight for a long period of time. A joint prepared as directed here should, how- ever, not allow any perceptible leakage in the course of a gas analysis although it is wiser not to subject it to any higher gas pressure or suction than necessary. The method of connecting the gas holder and the burette is shown in Fig. 7 where A is the gas holder, D, a bent capillary tee tube and F the gas burette. Before making the connections the gas burette is to be filled with saturated water and the cock closed by a turn of 180 which completely seals both the capil- laries below and above the cock. The capillary tee tube is to be inserted into the open rubber tube above the clamp B, con- nected at E as shown and the joints are to be wired. The gas in the holder is to be put under pressure by raising the levelling bottle and clamp B opened. Cock C is then opened slightly until the water in the capillary of the gas holder rises to displace all the air in the funnel arm. Cock C is then closed and cock F turned 90 degrees to the position shown in Fig. 7 and at B in 20 GAS AND FUEL ANALYSIS Fig. 6. The rest of the water in the capillary A moves through D and out F driving all air ahead of it. Some gas may also blow out if the manipulator is not skilful, but this does no harm. Cock F is then turned 90 degrees opening the passage to the burette as shown at A in Fig. 6 and the gas passes into the burette. When enough gas is in the burette another 90- FIG. 7. Gas burette and gas holder. degree turn of the cock in the same direction to the position C of Fig. 6 stops the gas flow. A few cubic centimeters of water are placed in the funnel above C (Fig. 7), the pressure in the gas holder is changed to suction, cock C is opened and water flows into the capillary of the gas holder re-establishing the seal. The clamp B may then be closed and the capillary disconnected. This procedure allows transfer of the gas without GENERAL METHODS OF TECHNICAL GAS ANALYSIS 21 possibility of change in composition and restores the seal of the gas holder at the close of the operation. Instead of the tee with cock and funnel blown as one piece the simpler apparatus shown in Fig. 4 of Chapter I may be used. 7. Measuring a Gas Volume. The volume of gas is to be measured at the temperature of the water jacket and at baro- metric pressure. The temperature of the gas as drawn into the burette does not ordinarily differ very many degrees from that of the jacket water but it should be allowed to stand a few moments to allow it to attain that temperature. During these minutes the gas if not already saturated with moisture rapidly absorbs it from the moist burette walls and becomes saturated as it should be before its volume is measured. Delay in reading the volume is also necessary to allow the film of water which adheres to the burette walls as the sample is rapidly drawn in, to run down. If the walls of the burette are clean the water will have run down in three minutes so completely that the volume has become approximately constant. If this precaution is neglected the volume read may easily be in error by 0.2 c.c., even with a clean burette. Some operators object to this three-minutes wait as too great hindrance to rapid work and it is true that if the readings are made by a skillful operator immediately after the introduction of the gas, the errors of successive readings are almost constant and disappear in the subtractions. If, however, the practice is followed of detaching the used pipette and connecting the new one before reading the volume of gas, sufficient time will have elapsed without the operator's having been idle. To obtain a sample of exactly 100 c.c. a sample of slightly more than 100 c.c. is initially taken and compressed by raising the levelling bottle until the bottom of the meniscus is exactly opposite the 100 c.c. mark. The stopcock at the bottom of the burette is then closed so that the meniscus cannot change its position and the stopcock at the top of the burette opened momen- tarily to the air, to allow the pressure in the burette to equalize itself with the outside air. The volume of gas should now be 100 c.c. at atmospheric pressure but the correctness of the volume should be checked by opening the stopcock connecting the levelling bottle with the burette and raising the levelling 22 GAS AND FUEL ANALYSIS bottle until its water surface is at the same height as that of the water in the burette. The connecting stopcock may now be closed and the volume read as indicated by the bottom of the meniscus. If the operation has been properly carried out the volume should be exactly 100 c.c. This volume is subject to correction for error in the burette and if an exact 100 c.c. sample is desired the meniscus may have to be set on some other figure than the 100 mark. 8. Calibration of a Gas Burette. A gas burette may be calibrated like any other form of burette by wiring a one-hole rubber stopper carrying a stopcock into the bottom of the burette and weighing the water delivered. Burettes of good quality are usually calibrated accurately enough throughout their cylindrical portion to make this form of calibration unneces- sary for technical work. The burette should, however, be cali- brated in its upper portion, especially if its previous history is not definitely known, since its original stopcock may have been broken and the volume of the neck changed when a new stopcock was fused on, and since it may have been calibrated by the maker to be used when filled with mercury instead of with water. The error in this latter case arises from the custom of reading the mercury meniscus at the top of its convex surface and the water meniscus at the bottom of its concave surface. It will be readily seen that if the mercury meniscus stands at 10 c.c. there will be more gas in the burette than if the same burette is filled with water with the bottom of its meniscus at 10 c.c. This error may amount to 0.2 c.c. in an ordinary burette. Both of these errors are constant ones throughout the cylindrical portion of the burette and independent of the volume of gas in the burette. It will therefore be sufficient to determine them once. The most convenient method is to compare one burette with another. Draw into each burette about 10 c.c. of air, the exact amount being entirely immaterial. It is only necessary that the volume be large enough so that the reading is in the cylindrical portion of the burette. It is not desirable that the volume be large since the error caused by water adhering to the walls of the burette increases with the size of the sample. Connect the burettes by a bent capillary tube making the rubber connections while the stopcock of one of the burettes is GENERAL METHODS OF TECHNICAL GAS ANALYSIS 23 open to the air so that the air enclosed in the capillary will be under atmospheric pressure. Read the volume of the air in each burette at atmospheric pressure as usual. We will assume that Burette A which has an unknown constant error "x" is the one being tested and that Burette B is the one assumed to be correct throughout its cylindrical portion, although it may itself have a similar unknown constant error " y. " Transfer all of the air from A to B stopping the water in A just at the stopcock and read the new volume in B. The notes will then read somewhat as follows: CALIBRATION OF BURETTE A AGAINT BURETTE B Burette A B Initial vol. air 10.6 c.c.+x 9.5 c.c.-f y Second reading 20.3 c.c.+y Subtracting 10.6 c.c.+x = 10.8 c.c. x = +0.2. c.c. This result translated into words means that there must be added to each reading of Burette A 0.2 c.c. Since the probable error of observation is 0.1 c.c. several successive determinations should be made and the mean taken. It is always wiser to record this error in the notebook after each reading, 92.5+0.2, and not simply make the correction mentally and record 92.7, since there may always come a time when the analyst will be in doubt as to whether he has made the mental correction or forgotten to make it before recording. This error automatically disappears whenever one volume is subtracted from another and it is therefore only necessary to apply it when the absolute volume needs to be known as is the case with the initial volume taken for analysis. This makes the error vastly less important than it would be otherwise, for if in an analysis of flue gas an apparent 100 c.c. sample is taken which becomes when corrected 100.2 c.c., and 10 c.c. is found to be CO 2 , the percentage of CO 2 , neglecting the burette calibra- tion, is found to be 10.00 per cent, and allowing for the calibra- tion 9.98 per cent., which when rounded off becomes 10.0 per cent, as before. The ca,se is different, however, when a sample of 10.0 c.c. is taken as is the case in explosion analysis. If the 24 GAS AND FUEL ANALYSIS analysis showed 8.0 c.c. of hydrogen, the percentage neglecting the calibration would be figured as 80.0 per cent, but allowing for the calibration would be 78.4 per cent. It is preferable to record the calibration correction even though it may be later neglected in the calculation. 9. Gas Pipettes. The various gases are determined so far as possible by absorption in suitable reagents contained in pipettes of the form shown in A, B, and C of Fig. 8. The use of a separate pipette for each reagent was first brought into general use by Hempel. These pipettes differ from his in the elimination of the deep U bend in the capillary, which is retained only in the explosion pipette. This deep bend is a distinct disadvantage A. C. FIG. 8. Details of gas pipettes. since drops of reagent collect in it and are later carried into the burette. It is no longer needed when used with the burette just described. These pipettes are mounted on wooden stands, as shown in Fig. 5, which should be paraffined and not shel- lacked so that they may not be affected by reagents accidentally spilled. 10. Connecting the Burette and Pipette. The greatest manipulative error with most forms of gas analysis apparatus comes from the frequent changes of pipettes necessary. Unless the operator is skilful there is danger of loss of gas or inclusion of air. The form of burette just described prevents this error. The general arrangement of the burette and pipette is shown in Fig. 5. The stopcock A is turned to the position shown at B in GENERAL METHODS OF TECHNICAL GAS ANALYSIS 25 Fig. 6, the bent capillary B whose dimensions are immaterial, is connected to the burette and pipette and the rubber joints are wired. The operator blows through the rubber tube (D of Fig. 5) on the last bulb of the pipette, forces the liquid from the pipette up the capillary and over to the stopcock, driving all the air ahead of it, and closes the stopcock by turning it 90 to the position D of Fig. 6. The capillary tube is now entirely full of liquid and the operator has only to compress the gas in the burette by raising the levelling bottle, and turn the stopcock a half turn to the proper position (A of Fig. 6) when the gas will pass into the pipette. By this method of manipulation, it is easy to transfer the gas from burette to pipette without loss or inclusion of air. Furthermore, since the volume of the capillary is entirely immaterial, it may be chosen of larger diameter than usual, permitting more rapid work and lowering the pressure necessary to force the gas through it rapidly. It has been found advantageous to have the stopcock left-handed, as indi- cated, so that the hand manipulating the stopcock may not interfere with a clear view of the meniscus of the liquid advancing along the capillary tube. 11. Details of a Simple Gas Analysis. Accuracy in gas analysis is dependent on the exercise of very great care in ma- nipulation. When the analysis is completed there is no way of going back over the ground again as may so frequently be done in ordinary chemical analysis, and it is not always possible to make duplicate analyses. The analyst must be able to state with confidence that every precaution was taken to ensure an accurate result. Detailed directions will be given for the analysis of a gas which may be assumed to be air. This is a very convenient material for practice since it may be obtained in unlimited quantities and of practically constant composition. The burette is to be cleaned ( 3), the stopcock lubricated (4), and the water saturated (5). Draw the sample of gas into the gas burette (6), measure it ( 7), and connect it to a pipette containing NaOH ( 10), for determination of CO 2 . The connections having been properly made and the air driven out of the capillary tube as indicated, raise the levelling bottle and pass the gas into the absorbent, letting the water of the burette follow until it reaches the bottom of the capillary of V 26 GAS AND FUEL ANALYSIS the pipette. The gas is now all in the pipette. Let it remain for about three minutes gently shaking the pipette occasionally to agitate the gases and cause more rapid absorption. When it is believed that absorption is complete pass 1 c.c. of the burette water into the pipette to rinse from the capillary any reagents which might have been splashed into it, and then draw back the gas into the burette, pulling the liquid of the pipette as far as the stopcock on the burette. By a quarter turn of the stopcock to position B of Fig. 6 the connection between the pipette and the burette is closed while that from the pipette to the air is open. This causes the reagent to siphon back into the pipette. After waiting for three minutes to allow the liquid to drain from the walls of the burette read the volume as before and report the decrease in volume as the volume of the C(>2 absorbed. There is no certainty that all the gas has been absorbed. The only way for the operator to be sure is to pass the gas back again into the pipette and repeat the operation until the volume remains constant. Disconnect the pipette from the capillary tube, and rinse out the capillary tube with the wash bottle shown at E of Fig. 5, thus completing the first step of the analysis. If the gas being analyzed is air, the volume after treatment with NaOH should be the same as before since the volume of CO 2 in the air, 0.04 per cent., is too small to be measured with a burette of this type. In the practice analysis of air the next determination would be that for oxygen which would be absorbed by phosphorus, alkaline pyrogallate or ammoniacal copper solution as directed in 3, 4 and 5 of Chapter III. Air contains 20.9 per cent, of oxygen by volume. The residue is assumed to be nitrogen. Successive analyses of similar gases may be made without changing the water in the burette provided that none of the reagents have been carelessly drawn into the burette. The greatest danger is from the caustic solution which will absorb some of the C0 2 from a newly introduced sample before its volume has been measured. Phenolphthalein in the water will act as an indicator to show when it has become alkaline and should be changed. There is no objection to having the burette water faintly acid and there is the advantage that small amounts of alkali are neutralized and rendered harmless. GENERAL METHODS OF TECHNICAL GAS ANALYSIS 27 12. Accuracy of the Analysis. Account should now be taken of the magnitude of various possible errors, some of which have not been mentioned. The smallest division on the burette is usually 0.2 c.c. and the volume may not with certainty be estimated by interpolation closer than 0.1 c.c. This probable error limits the accuracy of the process to 0.1 per cent, with a sample of 100 c.c. and an accuracy of 1.0 per cent, with a sample of 10 c.c. Change of temperature of the burette water during an analysis causes a change of 0.36 per cent, in the volume of the gas for each degree centigrade. Change of barometric pressure causes a change in volume of 0.13 per cent, for each millimeter of mercury change in pressure. There are other minor sources of error which will be mentioned in Chapter VI under " Exact Gas Analysis." It will be evident that it is perfectly useless to expect an accuracy of greater than 0.1 per cent, with apparatus of this type, and that an error of 0.2 per cent, is not improbable. An analyst who reports an analysis carried to hundredths of a per cent, only shows his own ignorance. CHAPTER III ABSORPTION METHODS FOR CARBON DIOXIDE, UN- SATURATED HYDROCARBONS, OXYGEN, CARBON MONOXIDE AND HYDROGEN 1. Carbon Dioxide. This gas is determined by absorption in a strong solution of either caustic soda or potash. Very concentrated solutions dry the gas and make it necessary to let it stand in the burette before reading the volume until it has again become saturated with moisture. Dilute solutions work too slowly. A solution of 50 grm. NaOH in 150 c.c. of water is recommended to be kept in the form of pipette shown in A of Fig. 8. The absorption is rapid, three minutes being always ample. The reaction may be accelerated by gently shaking the pipette or by passing the gas back and forth. Glass rods or perferably glass tubes are sometimes introduced into the pipettes to accelerate the absorption by offering a large wetted surface with which the gas may come in contact. If this device is used care must be taken that no gas bubbles are trapped in the pieces of glass tubing as may frequently be the case if they are less than 5 mm. internal diameter or if they do not stand vertically. If it is desired to place glass tubes in the pipette use the form shown in B of Fig. 8. It will be necessary to wire the rubber stopper firmly into place to prevent leakage of the caustic. This reagent absorbs not only carbon dioxide but also sulphur dioxide, hydrogen sulphide and any other acid vapors which may be present. It may be used until almost all of the caustic has been changed to carbonate. One pipette will absorb about four liters of C02. There will be slow carbonate formation through exposure to the air but the reagent may be used with- out fear for several months provided it is used infrequently. 2. Unsaturated Hydrocarbons. These gases are determined by absorption in a liquid which by addition forms saturated compounds from the unsaturated ones. In gases from coal the 28 ABSORPTION METHODS 29 predominating constituent is ethylene, C 2 H 4 , but smaller per- centages of the other olefines are present and sometimes small amounts of acetylene, C 2 H 2 . A solution of bromine water made by diluting one volume of saturated bromine water with two volumes of water is the reagent preferred by the author. It is placed in the first bulb of the double pipette shown in C of Fig. 8, and the third bulb is filled with water to lessen the diffu- sion of bromine into the air of the laboratory. If the solution becomes bleached through formation of hydrobromic acid in the sunlight, it is sufficient to add a few drops of liquid bromine which again brings it up to its normal strength. It is unnecessary to have it so strong that the gas drawn back into the pipette shows pronounced yellow bromine fumes. When gases with more than a small per cent, of unsaturated hydrocarbons are brought into contact with bromine it is possible to observe the formation of the bromide as an oily film on the surface of the liquid. As this film retards reaction between the bromine and the gas, it is advisable to shake the pipette gently during absorption, when little drops of the heavier ethylene bromide may be seen falling to the bottom of the pipette. The gas drawn back into the burette will contain so much bromine vapor that its volume may even have increased in the process. It must be passed into a caustic solution and shaken for about one minute to remove the bromine fumes and then brought back into the burette and measured. The diminution in volume is to be reported as unsaturated hydrocarbons. A delicate test for the complete removal of these constituents is afforded when phosphorus is subsequently used as a reagent for oxygen. A fraction of a tenth of a per cent, of ethylene will completely prevent the reaction between phosphorus and oxygen. Three minutes shaking with the bromine is sufficient to remove the unsaturated hydrocarbons from most gases, but gases of high candle power, like Pintsch gas, sometimes require ten minutes. Treatment with bromine water should be repeated until phosphorus smokes when the gas is brought in contact with it. Fuming sulphuric acid acts in the same way as bromine water, but it is difficult to handle, attacks rubber tubing badly, and must be protected from the moisture of the air to avoid loss of efficiency. 30 GAS AND FUEL ANALYSIS 3. Oxygen by Phosphorus. Yellow phosphorus is an extremely reliable reagent for the absorption of oxygen when used under proper conditions. It combines with oxygen producing solid oxides of phosphorus and reacts with almost no other gases which might be present. The disadvantage attending its use is the danger of the reaction not taking place at all because of the presence of poisons, because of too low temperature or because of too high a concentration of oxygen in the gas. There is, however, easy ocular evidence of the reaction so that there need be no uncertainty as to whether the reaction has taken place. When once started it goes to completion. The pipette is of the form shown in B of Fig. 8. To prepare it for use the pipette is inverted and filled with water and into it are dropped sticks of phosphorus about 5 mm. in diameter which have been cut to proper length under water. The sticks are to stand vertically and fill the pipette practically full. The only advantage of the small sticks lies in their greater surface and in the convenience with which the pipette may be filled. It is possible to prepare them in the laboratory by melting phos- phorus under water at a temperature a little above 44 C. and sucking it into a slightly conical glass tube. With expert manipulation this tube may be lifted from the warm water and plunged into cold water where the phosphorus will soldify and contract so that the stick may be pushed from the tube. It is simpler though not so elegant to mould the sticks in a tin dish about 5 in. long made with a corrugated bottom and provided with a rim about an inch high and a handle. This mould is submerged in a dish of warm water and enough phosphorus melted in it to fill the corrugations. It is then lifted out and placed in cold water till the sticks become solid. Molten phos- phorus catches fire instantly in the air and produces dangerous burns on the skin. Particles of solid phosphorus dropped in cracks of a wet table or floor have been known to smoulder twenty-four hours and finally burst into flame. Great care must always be exercised in handling phosphorus and it is usually preferable to buy it already cast in small sticks. The pipette should be kept in the dark when not in use to avoid the deterioration of the phosphorus. Pipettes made from brown glass ABSORPTION METHODS 31 retard the action of the sunlight. With proper care a phos- phorus pipette should last for years without refilling. When a gas containing oxygen is introduced into a phosphorus pipette there normally appears at once a dense white cloud of oxides of phosphorus which are evident even when the amount of oxygen is less than 0.1 per cent, of the total volume. All tech- nical gases will contain this much oxygen, for even if they did not contain it originally they will have absorbed it from the water of the sampling apparatus or of the burette, so that the presence of these clouds is a sure indication that the reaction is progressing properly. Conversely the absence of any smoke means not that there is no oxygen present, but that there is something preventing the reaction. The reaction between oxygen and phosphorus is also attended by a glow, visible only in the dark, which disap- pears with the completion of the reaction. The white cloud con- sists of solid particles of phosphoric oxides which slowly settle and dissolve in the water. It is not necessary to wait for this, how- ever, before drawing the gas back into the burette and measuring the oxygen absorbed for the particles possess very slight vapor tension and do not occupy an appreciable volume. The reaction should with certainty be completed in three minutes if the white smoke appears promptly and if the surface of phosphorus exposed is large. It will not accelerate the reaction to shake the pipette since the reagent is a solid and the pipette is so closely filled with sticks that diffusion will soon bring the oxygen to the sur- face of the phosphorus. The sticks of phosphorus at first yellow and waxy become covered with a reddish crust on long exposure to light and are in- active. Such sticks may be melted under water, skimmed, and cast into new sticks. The inactive crust may also be removed by placing in the pipette a 10 per cent, solution of K 2 Cr 2 O7 made faintly acid with H 2 SO 4 . The chromate becomes reduced to the green chromic salt with oxidation of the surface of the phosphorus. It is preferable, however, to avoid deterioration by keeping the pipette in the dark when it is not in use. If the white vapors do not appear promptly when the gas is passed into the pipette, the explanation may be sought in four directions. a. The temperature may be, too low. It should be above 15 C. 32 GAS AND FUEL ANALYSIS b. The concentration of the oxygen may be too high. Moist phosphorus does not react at all with perfectly pure oxygen at ordinary room temperature. If the partial pressure of the oxygen is diminished either mechanically by an air pump or by dilution with an inert gas the reaction commences. With 50 per cent, oxygen it is violent, flame being produced and the phosphorus melting. There is danger of breaking the pipette. When the concentration is less than 30 per cent, the reaction proceeds quietly and the phosphorus does not melt. c. A poison may be present. The most commonly occurring poison is ethylene, a few hundredths of a per cent, of which com- pletely prevents the reaction between phosphorus and oxygen. Acetylene, benzine, ether, hydrogen sulphide and many other substances possess similar though usually weaker powers. For- tunately they are practically all Removed either mechanically or chemically by bromine water followed by caustic potash and the gas should always be thus treated if the white fumes fail to appear promptly. d. The phosphorus may have been rendered inactive by a heavy dose of poison. It may be restored by replacing the water in the pipette with fresh water and passing several successive samples of air into the pipette until the phosphorus again smokes freely. Commercial compressed oxygen should be diluted with nitro- gen before analysis. The nitrogen may be conveniently pre- pared from air by passing a buretteful of air into the phosphorus pipette, disconnecting the pipette and closing it with a pinchcock. Twenty-five to thirty cubic centimeters of the oxygen are then to be drawn into the burette (note that a calibration error of 0.3 c.c. means 1 per cent, here) and the phosphorus pipette recon- nected. There will be sufficient nitrogen in it to flush out the capillary connecting tube and fill the burette as well. The oxygen may now be absorbed by phosphorus. It is desirable to introduce the oxygen into the burette before the nitrogen which thus fills the upper part of the burette and comes in contact first with the phosphorus. If the reverse procedure were fol- lowed and the oxygen introduced in the burette last, it might be brought in concentrated form in contact with the phosphorus ABSORPTION METHODS 33 and cause violent combustion, a trouble which the above scheme of procedure avoids. It has been frequently noted that beginners in gas analysis obtain rather consistently low results when determining the oxy- gen in air. This is usually due to the presence of sodium carbon- ate in the water of the burette as the result of faulty manipula- tion of the caustic pipette. The water in the phosphorus pipette contains phosphoric acid and if, in the manipulation, some of this acid is drawn back into the burette containing a little alkali, carbon dioxide is evolved in sufficient amount to cause the ap- parent percentage of oxygen in the air to be a few tenths of a per cent. low. 4. Oxygen by Alkaline Pyrogallate or Hydrosulphite. Alka- line pyrogallate is very frequently used, especially in apparatus of the Bunte or Orsat type. When used with proper precautions it gives accurate results. Pyrogallol is a trihydric phenol C 6 H 3 (OH) 3 which in alkaline solution is a strong reducing agent, becoming itself oxidized to other products. Unfortunately, un- less the solution is freshly prepared and strongly alkaline there is danger of carbon monoxide being evolved as oxygen is ab- sorbed. This not only causes the oxygen to be reported low, but also makes the carbon monoxide high. Berthelot has shown that a reagent made by taking a solution of 1 part of pyrogallol in three of water and mixing it with its own volume of a solution of 1 part of KOH in two of water will, when freshly prepared, absorb ten times its volume of oxygen without giving back more than a trace of CO, but that solutions which have been used too long or do not contain enough alkali may give up as much as 5 per cent, of the volume of oxygen absorbed as carbon monoxide. The reagent should be kept in a double pipette (C of Fig. 8), the third bulb being filled with water to prevent action of the air on the reagent. It may also be kept in a single pipette (A of Fig. 8) provided a rubber balloon is fastened to the second bulb for the same purpose. The reagent works very slowly at temperatures below 15 C. Carbon dioxide and other gases which would be absorbed by alkali must be removed before testing for oxygen with pyrogallate. Mr. H. A. Small in the author's laboratory tested the behavior of a solution of pyrogallate made up according to Berthelot's directions and kept in an ordinary four bulb pipette without 34 GAS AND FUEL ANALYSIS any special precautions to prevent diffusion of oxygen into the pipette. Duplicate samples of air were analyzed almost daily for more than a month. During the first four weeks the reagent absorbed the proper percentage of oxygen although it was fre- quently necessary to shake more than three minutes to accom- plish this. The 150 c.c. of reagent in the pipette had up to this time absorbed 1140 c.c. of oxygen. After this the absorption of oxygen became incomplete, the line of demarcation being quite sharp and the apparent oxygen of the air dropping from 20.7 per cent, to 20.4 per cent. Tests showed that the oxygen was still being quantitatively absorbed but that about 0.3 per cent, of carbon monoxide was being evolved. These results con- firm Berthelot's statements. According to our results 1 c.c. of the reagent absorbs 8.0 c.c. of oxygen. In case of doubt concern- ing the reagent it should be tested on air and should be rejected unless it absorbs 20.6 to 20.8 per cent, of oxygen. Anderson 1 recommends a much more strongly alkaline solution prepared by dissolving 15 g. pyrogallol in 100 c.c. of a solution of KOH of sp. gr. 1.55. An alkaline solution of this concentration is obtained by 1.5 to 2.0 parts of KOH in 1.0 parts of water. NaOH cannot be substituted for KOH in this concentrated solu- tion. A disagreeable feature of this concentrated reagent is the formation of a precipitate which chokes the glass tubes and ren- ders it difficult to determine the exact level of the liquid when drawing the gas back into the pipette. Anderson has described a special pipette to minimize this difficulty. Each cubic centi- meter of reagent will absorb satisfactorily 22 c.c. of oxygen from air when used in an ordinary Orsat pipette. The absorption is much more rapid than with the ordinary reagent. Anderson recommends the simple Orsat pipette over the more complicated forms, for use with this reagent. Sodium hydrosulfite has been recommended by Franzen 2 as a cheaper and better reagent than pyrogallol. The solution is prepared by dissolving 50 grm. Na 2 S 2 O 4 in 250 c.c. H 2 and mixing this with 40 c.c. of a caustic solution made by dissolving 500 grm. NaOH in 700 c.c. H 2 O. Each cubic centimeter of this reagent absorbs 10.7 c.c. oxygen. The equation for the reaction is as follows : Na 2 S 2 4 +H 2 + = l Jour. Ind. and Eng. Chem., 8, 131, 133 (1916). 2 Chem.-Berichte, 39, 2069 (1906). ABSORPTION METHODS 35 The solution is placed in a pipette containing rolls of iron gauze to increase the absorbing surface. Franzen states that the ab- sorption is complete in five minutes without shaking, that it pro- ceeds almost as rapidly at 4 C. as at room temperature, and that there is no danger of the formation of CO in the process. 5. Oxygen by Ammoniacal Copper Solution. Orsat 1 advo- cated a cold saturated solution of ammonia and ammonium chlo- ride in contact with metallic copper as a reagent for the absorption of oxygen and called attention to the limitation in the use of this reagent caused by its absorption of carbon monoxide. Badger 2 has recently made a careful study of various modifications of this solution and recommends that the reagent be prepared by saturating with ammonium chloride a mixture of one part con- centrated ammonia and one part water. This solution will ab- sorb from fifty to sixty times its volume of oxygen and then fails, not by refusing to absorb quantitatively but by the formation of so heavy a precipitate that it becomes unmanageable. The presence of metallic copper in the gas space is necessary for the proper operation of this reagent. The first bulb of a pipette of the type of B in Fig. 8 is filled with copper wires placed vertically and long enough so that the upper ends of those centrally placed reach even into the outlet of the capillary tube. This precaution is necessary when analyzing nearly pure commercial oxygen as otherwise the liquor rising as the oxygen is absorbed may sub- merge all of the wires before the oxygen is all absorbed. A rubber balloon may be fastened on the second bulb of the pipette to prevent rapid absorption of oxygen from the air. This reagent is cleaner to use and has a longer life than pyrogallate and its action is not interfered with by the poisons which inhibit the action of phosphorus. It is active at almost any temperature. Ammonia is given off from the fresh reagent in amount sufficient to make it necessary to pass the gas into a pipette containing dilute sulphuric acid before reading the volume to be recorded as that of the gas from which the oxygen has been removed. An old reagent gives off very little ammonia. This solution absorbs carbon monoxide and also acetylene and cannot be used to de- termine oxygen when these gases are present. It is recommended for the analysis of commercial oxygen. 1 Annales des Mines, Series 7, t. 8, 485 (1875). 2 Jour. Ind. and Eng. Chem., 12, 161 (1920). 36 GAS AND FUEL ANALYSIS 6. Carbon Monoxide. The methods for absorption of CO are less satisfactory than for any of the other commonly occurring gases. The usual reagent is cuprous chloride Cu 2 Cl 2 which on account of its slight solubility in water must be used either in acid or ammoniacal solution. The acid solution consists of a practically saturated solution of cuprous chloride in hydrochloric acid of specific gravity of approximately 1.12. Since the acid is only a solvent, its exact concentration is immaterial. Com- mercial acid may be used. About 150 grm. of the cuprous chloride will dissolve in a liter of acid of this concentration. This solution of cuprous chloride when pure is perfectly colorless but it darkens through slight oxidation as on exposure to the air, so that the usual solutions are black. It is possible to keep it colorless by placing in the reagent bottle or in the pipette copper turnings or copper wire, but it does not increase the efficiency of the reagent. If the oxidation proceeds so far that the solution becomes green, due to complete oxidation to the cupric state, the solution is worthless until it is again reduced to the cuprous state. The reagent is kept in a double pipette whose third bulb is filled with HC1 of sp. gr. 1.12 instead of water so that in case the cuprous chloride spills into it, it will not be precipitated. The gas which must have been previously freed from unsaturated hydrocarbons and oxygen is passed into the pipette and shaken for three minutes, then drawn back to the burette and passed into a second pipette containing fresh cuprous chloride where it is again shaken for three minutes, drawn back and measured. The HC1 vapors in the gas may be neglected. There is no method of knowing whether the absorption of the CO has been complete. The only safe way is to repeat the absorption using a fresh solution until the volume becomes constant. Usually two absorptions of three minutes each are sufficient. The reaction between carbon monoxide and cuprous chloride has not been definitely worked out. Jones 1 has shown that under certain circumstances a crystalline compound of definite composition results Cu 2 Cl 2 , 2CO. 4H 2 O. In the dilute solu- tions present in gas analysis, however, the reagent behaves l Am. Chem. Jour., 22, 287. ABSORPTION METHODS 37 exactly as if it dissolved the carbon monoxide. When a perfectly new solution is used the CO will be practically completely removed. As the CO in solution increases there comes appar- ently an equilibrium between that in the gas and that in the solution, with the result that the absorption is incomplete. If a gas with only a small amount of CO is brought in contact with a solution which has absorbed much, the gas will increase in volume due to CO given up by the solution. Each cuprous chloride pipette should bear a label on which should be recorded the number of cubic centimeters of carbon monoxide which has been absorbed. When it has absorbed more than 10 c.c. it is not safe to rely on the results. In practice the analyst should have two pipettes for cuprous chloride, one, which has absorbed considerable carbon monoxide, to be used first, and another, which should be kept almost entirely fresh to follow the other. When this second pipette has absorbed 10 c.c. of CO it should be used as the first pipette and the former first pipette should be emptied and refilled with fresh reagent. The solution may be regenerated if desired by boiling it for half an hour in a flask containing some metallic copper and provided with a reflux condenser to prevent much loss of acid. The feebly held CO is driven off by the boiling and any oxidized solution is reduced to the cuprous state again. Krauskopf and Purdy 1 suggest the use of stannous chloride as a reducing agent to be added to a solution of cupric chloride and form cuprous chloride in solution. Their solution contained 127.2 grams metallic copper and 616 c.c. of concentrated hydro- chloric acid per liter. The cupric chloride dissolved in the hydro- chloric acid was reduced by the stannous chloride. The presence of stannic and stannous chlorides even in relatively large amounts does not impair the efficiency of the solution for the absorption of carbon monoxide. The authors report that 200 c.c. of this solution will absorb over 300 c.c. of carbon monoxide quantita- tively as shown by the following extract from their Table I where Pipette I contains an old solution and Pipette II a fresh solution, used after Pipette I gave no further absorption. 1 /, Ind, Eng, bhem., 12, 158 (1920), 38 GAS AND FUEL ANALYSIS Amount of CO already absorbed 3(55. 5 e.c. Volume of gas taken 50.2c.c. Pipette I Pipette II Time, Amount CO lime, Amount CO min. absorbed min. absorbed 1 39.0 1 0.4 2 0.6 2 0.2 3 0.2 3 0.0 The authors also confirm the reliability of the process of regen- erating the solution by heating to 60-70 C. for several hours under a reflex condenser. The ammoniacal solution is made by suspending about 150 grm. of cuprous chloride in a liter of distilled water into which ammonia gas is passed until the liquid becomes a pale blue color. The ammoniacal solution slowly regenerates itself on standing, the CO becoming oxidized to (NH 4 ) 2 CO 3 and a mirror of metallic copper depositing. The NH 3 gas must be removed by an acid pipette before the correct amount of CO absorbed may be read. Carbon monoxide may be estimated by explosion or combus- tion as described in the next chapter. Minute amounts of it may be estimated by the IzOs method given in Chapter VI on "Exact Gas Analysis." 7. Absorption of Hydrogen. Hydrogen may be absorbed by palladium sponge superficially oxidized to palladous oxide, according to the method oi Hempel. It is necessary to re- generate the palladium after each experiment and the reaction is prevented by small amounts of carbon monoxide, hydrochloric acid and other constituents so that it is not a method which has found much favor. A solution of palladous chloride as prepared by Campbell and Hart 1 is a better reagent. It is used as an almost neutral 1 per cent, solution prepared by dissolving 5 grm. palladium wire in 30 c.c of HC1 to which is added 1 or 2 c.c. HN0 3 . The solution thus prepared is evaporated just to dry ness on the water bath, redissolved in 5 c.c. of HC1 (sp. gr. 1.20) and 25 or 30 c.c. of water, warmed till solution is complete and diluted to 750 c.c. It is placed in a simple Hempel pipette made to be readily 1 Am. Chem. Jour., 18, 294 (1896). ABSORPTION METHODS 39 detachable from its frame so that the bulbs may be placed in a water bath. The first bulb of the pipette should have a capacity of at least 150 c.c. to allow for expansion of the gas and water vapor when the solution is warmed. The gas freed by the usual methods from CO 2 , C n H 2n , 2 and CO, and containing H 2 , CH 4 and N 2 is passed into the pipette and the water from the burette passed over to seal the capillary of the pipette. A screw clamp is placed on the rubber connecting tube and the pipette disconnected from the burette and placed in a water bath at 50 C. for an hour and a half. A higher temperature does no harm provided it does not expand the gas so much as to cause it to bubble out of the first bulb or to leave only a small amount of reagent in contact with the gas in the first bulb. The hydrogen reacts with the palladium chloride forming metallic palladium and HC1. The total decrease in volume is reported as hydrogen. The reagent may be counted on to absorb one-third of its volume of hydrogen completely in an hour and a half. Larger quanti- ties will be absorbed more slowly. It is readily regenerated by rinsing from the pipette, evaporating just to dry ness on the water bath, dissolving in 5 or 6 c.c. HC1 and 4 or 5 drops of HNOs and again evaporating. The dry palladous chloride is dissolved by adding 2 c.c. cone. HC1 and a small amount of water and is then diluted to its original volume. The method is accurate and satisfactory. Carbon monoxide and other reducing gases behave like hydrogen and must be removed pre- vious to the test but hydrocarbons of the methane series are not affected. The chief objection to the method lies in the time consumed which makes it frequently necessary to correct the gas volumes for change in room temperature and barometric pressure. A memorandum should be made of the temperature of the burette jacket and of the barometric pressure before and after the test and corrections made if necessary. Paal and Hartmann 1 recommend a solution of colloidal palladium made by dissolving 2.44 grm. collodial palladium manufactured according to Paal's process by Kalle ( = 1.5 grm. palladium) and 2.74 grm. of sodium picrate in enough water to bring the volume to 130 c.c. Gaseous hydrogen dissolves in the aqueous palladium solution and reduces the picric acid. Hempel 2 1 Chem. Berichte, 43, 243 (1910). 2 Zeit. Angewanat. Chem., 25, 1843 (1912). 40 GAS AND FUEL ANALYSIS has made a critical study of this method and reports that a solution prepared in this manner will absorb in 15 minutes when freshly prepared 21.2 c.c. H 2 per 1 c.c. reagent, after 79 days 16.2 c.c. H 2 per 1 c.c. reagent, after 1 year 1.6 c.c. H 2 per 1 c.c. reagent. He states that a fresh solution may be safely trusted to absorb completely 7.2 c.c. H 2 per centimeter reagent in three minutes if heated to the temperature of the blood. He advocates the preparation of the solution in small quantities and its use in a pipette filled mainly with mercury. A disadvantage attending the use of the reagent is the persistent foam which results after shaking and which must be allowed to subside before the volume of the gas is read. A few drops of alcohol at once dissipate the foam but spoil the reagent for further use. When alcohol is used the pipette must be carefully cleaned before another experiment. In using this method the gas must first be freed from oxygen which is caused to unite with the hydrogen by the palladium, from unsaturated hydrocarbons which form addition products with the hydrogen, and from carbon monoxide which retards the absorption of the hydrogen. Bromine is recommended as the absorbent for the unsaturated hydrocarbons as it removes compounds of arsenic and phosphorus which might retard the catalytic action. Alkaline pyrogallate is advised for absorption of oxygen as phosphorus fumes affect the palladium. An ammoniacal solution of copper chloride is preferred to the acid solution. 8. General Scheme of Analysis. Fig. 9 shows the apparatus ready for the analysis. The details concerning the individual steps of the process were given in Chapter II but it is well to recapitulate the important points in connection with the general scheme of analysis. The water of the burette is saturated with gas similar to that which is to be analyzed and a sample of ap- proximately 100 c.c. is then drawn in and the volume read at atmospheric pressure after three minutes have been allowed to elapse so that the surplus water will have drained from the bur- ABSORPTION METHODS 41 ette walls. Any needed correction for burette error is to be applied to this reading. The caustic soda pipette is to be con- nected to the burette, and the gas passed into it and allowed to remain with gentle shaking for three minutes. C(>2 is absorbed, as well as H 2 S, SO 2 etc., the whole being usually reported as CO 2 . The gas is drawn back into the burette and while waiting for the excess water on the burette walls to run down, the capillary connecting tube is flushed out and the bromine water pipette . FIG. 9. Assembled apparatus for gas analysis. is connected. The volume in the gas burette is then read, and the gas passed into the bromine water where it is shaken for three minutes. It is drawn back to the burette, and at once passed into the caustic pipette where it is shaken one minute to re- move bromine fumes. It is then drawn back into the burette and the decrease in volume reported as unsaturated hydrocar- 42 GAS AND FUEL ANALYSIS bons. The gas is next passed into the phosphorus pipette. White smoke at once appears. If it does not, it is a sign that some poison is present, usually removable by another treatment with bromine water. Following the estimation of oxygen comes that of carbon monoxide with cuprous chloride either acid or ammoniacal, preferably acid unless hydrogen is to be absorbed by palladium. Two cuprous chloride pipettes must be used in series, the second one containing almost fresh reagent. The resi- due from this absorption consists of hydrogen, hydrocarbons of the paraffine series and nitrogen. The hydrogen may be ab- sorbed by palladium as outlined in this chapter but it is more common practice to estimate the hydrogen and hydrocarbons by combustion. The methods are discussed in the next chapter. CHAPTER IV EXPLOSION AND COMBUSTION METHODS FOR HYDROGEN, METHANE, ETHANE AND CARBON MONOXIDE 1. Available Methods. Hydrogen and carbon monoxide may be estimated by absorption as indicated in the preceding chapter. They may also be estimated after oxidation to water or carbon dioxide. There are no satisfactory absorbents for methane and ethane and so these gases are always estimated indirectly after oxidation. The oxidizing agent may be gaseous oxygen and the reaction may be violent as in explosion methods or it may be quiet combustion. There may even be combustion of hydrogen Extrathtck Wall Platinum Wire, Fused in-- Stop Cock, 2mm. Bore *3 mm. Infernal Diameter FIG. 10. Detail of explosion pipette. and carbon monoxide with the aid of a catalyzer in the presence of methane which remains unchanged. The oxidizing agent may also be an oxide, especially copper oxide, and here again there may be fractional combustion. The most rapid method and the one most frequently used is that of explosion. 2. Apparatus for Explosion Analysis. The analysis by ex- plosion is carried out in a stout glass vessel provided with elec- tric connections across whose terminals a spark may be passed to cause the explosion. The shape and dimensions of the appa- ratus may vary. The form which has given good satisfaction 43 44 GAS AND FUEL ANALYSIS in the gas laboratory at the University of Michigan for a number of years 1 is shown in Fig. 10. It is a modification of the Hempel pipette and differs from it principally in the arrangement of the electric terminals and in the incorporation of an explosion guard in the stand. In the older forms of pipette the explosion was induced by a spark made to jump a gap between two platinum wires sealed through the glass of the narrowed upper part of the bulb. When the interior of the bulb became wet as frequently happened the electric current would* sometimes travel around the wet wall instead of sparking across the gap, and it was not possible to obtain an explosion. Gill 2 modified the bulb by introducing one of the wires through a ground glass joint at the bottom of the pipette. This was a valuable modification because it made the creeping distance so long that the spark was compelled to jump the gap, but the ground glass was difficult to keep tight. The form here described introduces the wire from the bottom but adopts a simple method of sealing which is very satisfactory. The lower wire which should be stiff (and may be of nickel about 1 mm. in diameter) is pushed through the open lower end of the pipette and sealed by sucking molten sealing wax into the pipette almost to the level of the tee. As the sealing wax hardens the wire may be moved to adjust the spark gap to the desired dimensions. No difficulty has been experienced in making this joint tight. The explosion pipette and its stand are shown in Fig. 11. The bulb is enclosed in a box open at the top and with a plate glass window in front, so that the operator can observe the explosion in perfect safety. The bot- tom of this box has an irregular opening sawed in it so that the pipette as shown in Fig. 10 may be lowered into place. The bulb sits in a cup shaped hollow of the shelf which may be padded with wet asbestos paper if necessary to make it fit well. The weight of the mercury renders other fastening for the bulb unnecessary but the capillary is fastened to the rib behind it by a loop of fine copper wire passing through holes drilled in the rib. The strain of the rubber tube filled with mercury is taken off the glass tee by a ring below the shelf into which the rubber tube is 1 White and Campbell, J. Am. Chem. Soc., 27, 734 (1905). 2 J. Am. Chem. Soc., 17, 771 (1895). EXPLOSION AND COMBUSTION METHODS 45 wedged firmly by a split cork. The rack for the levelling bottle is higher than the tee of the pipette so that when the bottle is in the rack the mercury in the rubber tube is under pressure whereas if the mercury bottle were sitting on the base of the stand there would be a partial vacuum within the rubber tube. This may seem immaterial, but it must be remembered that all rubber is porous and that bubbles of air sucked into the rubber tube are certain to make their way up into the pipette and be measured FIG. 11. Explosion pipette and stand with protecting screen. as part of the gas in it. Fine copper wires soldered to the elec- trode terminals of the pipette pass to the binding posts on the stand. The fine platinum electrode is liable to be cut if it is bent back and forth where it is sealed through the glass and to protect it the copper wire from the upper electrode is brought smoothly up to the capillary and tied there firmly with thread. The design of the stand is such that if a bulb breaks a new bulb may be inserted without trouble if it has even approximately the dimensions gf the old one. The method of connecting this pipette to the burette and of transferring the gas is the same as for other 46 GAS AND FUEL ANALYSIS pipettes. This pipette is filled with mercury, since under the high pressure developed the solubility of the gases in water becomes large enough to cause appreciable error. An induction coil capable of giving a half inch spark together with its bat- tery is necessary. A coil giving a large spark such as is given by the automobile sparkers is much better than a coil giving a thin high voltage spark. 3. Manipulation in Explosion Analysis. In the explosion proc- ess a sample of gas previously freed from carbon dioxide, oxygen, unsaturated hydrocarbons and usually carbon monoxide is drawn into the burette and measured. Its volume may vary from 8.0 c.c. with pure methane to 50 c.c.with gases containing high per- centages of nitrogen. Air sufficient to fill the burette is then drawn in. The burette is connected to the pipette as usual, care being taken to have the rubber connections in good condition and firmly wired, and the mixture is passed into the explosion pipette, the water of the burette being run over through the cap- illary of the pipette until the capillary is full. This water in the capillary acts as a cushion, preventing the force of the explosion from blowing up the rubber connections. The gas in the explosion pipette is brought to atmospheric pressure by means of the levelling bottle, the stopcock is closed, and the levelling bottle replaced in its rack. It is advisable to shake the pipette to ensure thorough mixing of the gases, for diffusion proceeds some- what slowly. The gas is exploded by a spark from the induction coil. If the gas consists mainly of hydrogen there is usually no visible flame although a slight tremor of the mercury may be observed. If the gas contains much hydrocarbon a flash of flame may usually be seen. It is not advisable to spark the mixture more than a second as some nitrogen will unite with the oxygen at the temperature of the spark forming oxides of nitrogen with decrease of volume and erroneous results. The explosion com- pleted, the gas is again brought to atmospheric pressure by means of the levelling bottle, and then brought back into the burette and measured. If there has been marked contraction, the next step is to pass the gas into caustic solution and determine if there has been formation of carbon dioxide. If the decrease in volume after explosion was less than 12 c.c. it is almost certain that the explosion was incomplete. If EXPLOSION AND COMBUSTION METHODS 47 there was no decrease in volume it is not safe to assume that no combustible gas was present, for it may have been present in such a small proportion that the mixture was not explosive. The proper procedure in either case is to add about 10 c.c. of pure hydrogen made by the action of caustic soda on metallic alumi- nium and explode a second time. The addition of this amount of hydrogen ensures complete explosion. After allowance for the contraction due to the added hydrogen, the composition of the original gas may be calculated as explained later. It is advisable to determine roughly the amount of oxygen remaining after the explosion so that there may be no doubt that an excess was present. 4. Oxidation of Nitrogen as a Source of Error. Almost all technical gases contain nitrogen as do also commercial forms of oxygen. Bunsen first noted that nitrogen and oxygen react at the temperature of the electric spark or of an explosion flame to form small amounts of various oxides of nitrogen whose volume is less than that of the reacting gases and which combine with caustic. The formation of oxides of nitrogen leads, therefore, to an erroneously high contraction after explosion and to an errone- ously high figure for CO 2 due to explosion. The error is discussed more fully in Chapter VI on Exact Gas Analysis but it cannot be altogether neglected in technical work. The error increases with higher flame temperatures and the simplest way to keep it within reasonable limits is to dilute the reacting gases with some inert gas such as nitrogen or excess of oxygen. The volume of the gases participating in the explosion (combustible gas + theoret- ical volume oxygen) should be from one-third to one-fifth that of the non-exploding gases (nitrogen + excess oxygen). It will be seen that 12 c.c. H 2 + 6 c.c. O 2 require to be diluted with from 54 to 90 c.c. nitrogen or oxygen and that 8 c.c. CH 4 -f- 16 c.c. O 2 require at least 72 c. c. of diluting gases. The explosion of large samples of gas mixed with commercial oxygen, a method proposed by Hinman and endorsed by Gill, 1 involves much greater danger of blowing up the pipette and be- cause of the higher temperature of explosion, tends to cause a larger formation of oxides of nitrogen from the nitrogen necessarily present. 1 J. Am. Chem. Soc., 17, 987 (1895). 48 GAS AND FUEL ANALYSIS 5. Accuracy of Explosion Methods. The necessity of diluting the exploding gases to avoid oxidation of nitrogen restricts the size of a sample of a rich gas like illuminating gas to about 10 c.c. With this small sample each 0.1 c.c. error in reading means 1.0 per cent. A greater accuracy can therefore not be expected except by averaging a number of analyses. It may be considered safe to rely upon a single analysis for the various gases deter- mined by absorption, but explosion analyses should always be made in duplicate. The main portion of the gas after the absorption should be stored in a gas holder to be drawn upon for subsequent check analyses. 6. Hydrogen by Explosion. If hydrogen is the only com- bustible gas taking part in the explosion its volume may be calculated from the contraction after explosion according to the following equation: 2H 2 +O 2 = 2H 2 O 2+1 =2orO Expressed in volumes this means that two volumes of hydrogen combine with one volume of oxygen to form two volumes of water vapor. Since, however, the gas after explosion cools again to the temperature of the burette water and is the same as before explosion, and since it was saturated with water before explosion, the additional water formed must all condense. For our purposes, therefore, two volumes of H 2 react with one volume of 2 with complete disappearance of the reacting gases. Under these circumstances when hydrogen is the only exploding gas, two-thirds of the resulting contraction will be the volume of the hydrogen exploded. 7. Hydrogen and Methane by Explosion. Methane combines with two volumes of oxygen to form carbon dioxide and water according to the following equation: CH 4 +2O 2 = CO 2 +2H 2 1+2 =1+0 The volumetric relations are expressed by the figures of the equation, one volume of methane uniting with two of oxygen to form one volume of carbon dioxide, ami two of water vapor EXPLOSION AND COMBUSTION METHODS 49 which condense and disappear as explained in the preceding paragraph. The result of the explosion, therefore, is that there is a contraction of two volumes for every one volume of methane and the formation of a volume of carbon dioxide equal to the methane. It is possible to determine the proportion of hydrogen and methane present in a gas mixture by explosion with air. The volume of carbon dioxide resulting from the explosion equals the volume of the methane. The contraction due to the methane is twice the volume of the methane and the difference between this contraction in volume and the total contraction is the contraction due to the explosion of hydrogen. In accordance with the preceding section two-thirds of this contraction is hydrogen. The following example will serve as an illustration of the method of calculation. Sample of illuminating gas 99 . 5 c.c. Volume after absorption of CO 2 , C n H 2n , O 2 , CO 85 . 2 Sample for explosion 10 . 3 Air to 97.6 After explosion, volume 80 . 1 Contraction 17.5 After KOH, volume 74 . 9 Vol. CO 2 formed 5.2 After phosphorus, volume 69 . 4 Vol. excess oxygen 5.5 Calculation 5.2 c.c. CO 2 = 5.2 c. c. CH 4 Contraction due to 5.2 c.c. CH 4 = 2X5.2 = 10.4 Contraction due to hydrogen =17.5 10. 4 = 7.1 Hydrogen = 2/3X7.1=4.7 Vol. CH 4 = 5.2 Vol. H 2 = 4.7 Vol. N,bydiff. 0.4 10.3 Per cent. Percent. H 2 _4.7X -39.0 Percent, N 2 =0.4X^^^=3.3 50 GAS AND FUEL ANALYSIS The ratio of exploding to non-exploding gases in the above illustration may be calculated as follows: Exploding gases 5.2 c.c. CH 4 -fl0.4 c.c. O = 15.6 4.7c.c. H 2 +2.35 c.c. O 2 = 7.05 Exploding gases 22 . 65 Non-exploding gases 97 . 6 22 . 65 = 74 . 95 exploding gases 22.65 1 non-exploding gases 74.95 ~~ 3.3 The excess of oxygen is calculated from the volume of an taken for explosion, =97.6 10.3 = 87.3 c.c. air with 20.9 pel cent. O 2 = 18.24 c.c. O 2 available. Used for combustion) as above,. 10.4+2.35 = 12.75 c.c. Excess oxygen = 18.24- 12.75 = 5.49 c.c., which checks with the 5.5 c.c. found by direct experiment. 8. Carbon Monoxide, Hydrogen and Methane by Explosion. The composition of a gas mixture containing CO, H 2 , and CH< may be determined by a single explosion ii in addition to the contraction and C02 the oxygen used in the explosion is also determined. There are various methods of calculation, that given by Noyes and Shepherd 1 being as follows: 1. Gas taken =CH 4 +CO + H 2 +N 2 2. Contraction =2CH 4 +iCO+fH 2 3. Oxygen consumed =2CH 4 +JCO + fH 2 4. CO 2 iormed =CH 4 +CO Hence H 2 = Contraction O 2 consumed. CO = | (2CO 2 +|H 2 -0 2 consumed) CH 4 = CO 2 -CO N 2 = Total gas-(H 2 +CO+CH 4 ). The oxygen consumed is calculated by determining the residual oxygen and deducting this from the volume of oxygen introduced as air whose percentage of oxygen is assumed to be 20.9. This method is more rapid than the usual one in which the CO is absorbed by Cu 2 Cl 2 , and it has no systematic errors, provided the dilution is great enough to avoid oxidation oi nitrogen. It will, except in expert hands, be found less reliable 1 J. Am. Chem. Soc. 20, 345 (1898). EXPLOSION AND COMBUSTION METHODS 51 than the usual method of absorption of CO and explosion of H 2 and CH 4 because each value calculated is dependent on the accuracy of three successive operations instead of two. 9. Quiet Combustion of a Mixture of Oxygen and Combust- ible Gas. Various attempts have been made to do away with the explosion pipette by causing the gas to burn gradually. Coquillion 1 in 1876 proposed to estimate small amounts of hydrocarbons in the air from mines by placing within a pipette a spiral of platinum or palladium wire. The mine air was in- troduced into the pipette, the spiral was to be heated to redness and, the amount of combustible gas being below the explosive ratio, the hydrocarbons were to be gradually burned. He recommended that for technical gases where there was danger of explosion the platinum spiral be placed in a small bulb blown FIG. 12. Quartz combustion tube with platinum spiral. in the capillary tube between the burette and pipette. The mixture of gas and air was measured in the burette, and then passed through the capillary over the glowing spiral. The capillary tube was supposed to be adequate to prevent the explosion from flashing back into the burette. Hempel 2 has improved this latter apparatus by placing the platinum spiral in a quartz tube between two glass capillary tubes. His arrange- ment as modified by the author is shown in Fig. 12 where AB represents a tube of transparent quartz about 4 mm. internal diameter and 125 mm. long, at each end of which are glass cap- illary tees connected to it by rubber tubing. Through each tee runs a stout nickel wire connected by a spiral of fine platinum wire. The nickel wires are sealed into the glass capillaries by sealing wax drawn into the enlarged ends of the capillaries. The apparatus may therefore be readily repaired if the platinum 1 Comptes rendus, 83, 394; 84, 458 and 1503. 2 Zeit. angewandt. Chem., 25, 1841 (1912). 52 GAS AND FUEL ANALYSIS wire becomes burned out. The nickel wires should be so large that they almost fill the capillary tube which should be of about 1 mm. internal diameter. They will then not be heated per- ceptibly by the passage of an electric current sufficient to heat the platinum wire to redness and will by their cooling action help to prevent the explosion from flashing back into the gas burette. If the mixture of gas and air is passed slowly over the platinum spiral the temperature will not rise above a fair red heat and there will be little danger of formation of oxides of nitrogen, hence there is no need of diluting the gases with so much air as is necessary in the explosion process and therefore a larger sample of gas may be used. It is not safe, however, to take a large sample of gas and dilute it with pure oxygen, for the capillary tube cannot be relied upon to prevent an explosion flashing back into the burette when a very explosive mixture is used. The inaccuracy of the usual explosion methods led Dennis and Hopkins 1 to devise a process of combustion whereby a large sample of gas might be quietly burned in pure oxygen. The combustion pipette consists of a pipette such as is used for phosphorus with its second bulb cut off and a levelling bottle for mercury connected. The ignition wire in the form of a platinum coil or grid is placed within the pipette immediately under the gas inlet and connected to two heavy wires which, insulated from each other, pass through the rubber stopper at the bottom of the pipette and are fastened to binding posts. The diameter and length of the platinum ignition wire must be chosen with reference to the electric circuit so that it will be easily heated to redness and its temperature controlled without the need of cumbrous rheostats. The conducting wires within the pipette may be of platinum or one of the non-rusting nickel- chromium alloys and should be at least 1 mm. in diameter. The manipulation is as follows. The full volume of gas remaining after absorption of oxygen, consisting of CO, H 2 , CH 4 and N 2 is transferred to the combustion pipette and a clamp is screwed onto the rubber connecting tube at the tip of the burette so that the pipette may be disconnected from the burette. The burette is filled with oxygen free from CO 2 and 1 /. Am. Chem. Soc., 21, 398 (1899). EXPLOSION AND COMBUSTION METHODS 53 of known purity and reconnected to the pipette, but the stop- cock of the burette is kept closed. The levelling bottle of the pipette is placed at such a height that the gas in the pipette is under slightly diminished pressure and the electric ignition wire brought to incandescence. The stopcock on the burette is now opened and a slow stream of oxygen passed into the pipette. A slight flash is usually noticeable as ignition takes place and the platinum wire glows more brightly so that it may be necessary to interpose more resistance in the heating circuit. The volume of gas in the pipette may either increase or decrease and the height of the levelling bottle must be varied accordingly. It is usually necessary to periodically increase the external resistance to prevent the platinum wire from burning out as the hydrogen originally present gives way to water vapor. After the oxygen is all passed into the pipette, the current is interrupted, the gases allowed to cool and the CO, H 2 and CH 4 determined as in 8. The great advantage of this process lies in the large sample and the consequent diminution of the error of observation. It requires a special pipette, which is, however, easily constructed, a source of electric current and a controlling rheostat. The manipulation is somewhat complicated and it has been the author's experience that novices usually wish that nature had provided them with an extra pair of hands. The error due to oxidation of nitrogen has been found by the author 1 to be fully as large in this process as in the explosion process. The subject is discussed more fully in Chapter VI. Hempel 2 also reports unfavorably on this process on account of the formation of oxides of nitrogen when combustion is continued long enough to ensure oxidation of all the methane. 10. Fractional Combustion with Palladinised Asbestos. The well known power of palladium to bring about the union of hy- drogen and oxygen at low temperature has long been made use of as a means of separating hydrogen from methane. The use of palladinised asbestos is due to Winkler. The asbestos is prepared by soaking a small amount of selected long fibered asbestos in a concentrated solution of palladous chloride prepared according to 7 of Chapter III. The fibers are to be kept as 1 J. Am. Chem. Soc, 23, 477 (1901). *Zeit. Angewandt. Chem., 25, 1841 (1912). 54 GAS AND FUEL ANALYSIS nearly parallel as possible and after saturation are to be dried and ignited at a very dull red heat when the chloride will de- compose leaving the fibers coated with metallic palladium and possibly palladous oxide. A bundle of two or three of these single fibers about an inch long is introduced into the end of a straight capillary glass tube about 1 mm. internal diameter and eight inches long, and brought to the middle of the capillary by suction on the opposite end of the tube. A drop of water on the asbestos makes it move more freely. The capillary is then to be dried and bent to the usual form for connecting the burette and pipette. In manipulation 20 or 30 c.c. of gas freed from C0 2 , C n H 2n and usually CO is mixed with an excess of air and passed through the capillary tube containing the palladinised asbestos into a pipette containing water. If the asbestos is very active, com- bustion may begin without external heat but to make certain the tube is heated with a small gas flame or alcohol lamp. It is not necessary to heat the tube to redness. A spark frequently appears at the end of the asbestos filament when the combustible gas first strikes it. This is a sign that the gas is passing too rapidly and the speed must be decreased until the spark disap- pears. Otherwise some methane will be burned. The gas is passed back and forth through the capillary twice and then drawn back into the burette and the volume measured. If hydrogen alone has been burned two-thirds of the contraction will be the volume of the hydrogen as explained in 6. Carbon monoxide will burn as well as hydrogen in this process and where both were present, it will be necessary to determine the carbon dioxide formed in addition to the contraction. The calculations follow from the equations: 2CO+0 2 = 2C0 2 2+1 =2 Contraction = i CO or 2H 2 +O 2 = 2H 2 O 2+1 =0 Contraction = |H 2 ThereforeC0 2 = CO Total contraction = JCO+|H 2 H 2 = f (contraction JCO) EXPLOSION AND COMBUSTION METHODS 55 This method is accurate provided the palladinised asbestos is dry and active and the proper temperature is maintained. It requires care to prevent any drops of water from getting into the capillary. If this happens when the capillary is cold the thread of asbestos becomes wet and must be dried thoroughly before it is active. If a drop of water gets into the capillary while it is hot the glass tube cracks. Very little attention has been paid to the possibility of small amounts of foreign gases rendering the palladium catalyzer inactive, but from the elaborate precautions which are necessary to keep the platinum contact substance active in the sulphuric acid manufacture it is evident that this possibility should not be ignored. The capillary tube should never be heated to redness on account of danger of burn- ing methane. The combustible gases are diluted largely with air to avoid too intense combustion and also to avoid an ex- plosion of the main body of the gas which might be propagated through the capillary if the gas mixture were too rich. Dis- astrous explosions have been known to result from an attempt to burn mixtures of hydrogen and oxygen in this manner. The combined volumes of hydrogen and carbon monoxide in the sample taken for analysis should not be over 20 c.c. and the volume after dilution with air should be almost 100 c.c. This method has been investigated by Nesmjelow 1 who emphasizes the danger of burning methane if the gases are passed through the capillary at a rate faster than one liter per hour. Hempel 2 has recently reported the results of a study of this process and finds that to obtain accurate results the temperature of the capillary must not rise over 400 C. and that the gas must be passed at a speed of not over 100 c.c. in eight minutes. He rec- ommends, as a method of temperature control, that the portion of the capillary to be heated rest in a brass trough which in the middle is thickened sufficiently to contain a hole deep enough for a thermometer bulb. In default of a thermometer a glass tube sealed at the bottom and containing a little mercury may be inserted in the hole. The boiling of the mercury (358 C.) indicates when a sufficiently high temperature has been reached. 1 Zeit. Anal. Chem., 48, 232 (1909). *Zeit. Angewandt. Chem., 25, 1841 (1912). 56 GAS AND FUEL ANALYSIS 11. Fractional Combustion with Copper Oxide. The com- bustion of carbon compounds of all sorts through contact with hot copper oxide has been a method long employed by organic chemists. Campbell 1 first utilized the principle of fractional combustion in gas analysis and determined accurately the mini- mum combustion temperature for various gases both with copper oxide alone and with palladinised copper oxide. His values are as follows: Gas Initial combustion point Pure CuO Pd.-CuO H 2 175-180 C. 100-105 C. 315-325 C. 270-280 C. 320-330 C. No combustion at 455 C. 80-85 C. 100-105 C. 240-250 C. 220-230 C. 270-280 C. CO C 2 H 4 C 3 H 6 C 4 H 8 (Iso) CH 4 Jaeger 2 first proposed a convenient scheme for utilizing this principle in ordinary gas analysis and the method usually bears his name. He takes advantage of the wide difference in the ignition point of CO and H 2 as compared with CH 4 to separate the two gases by fractional combustion. His method with some modifications which the author has found desirable is as follows : The combustion tube shown at A in Fig. 13 is of hard Jena glass or preferably transparent quartz and has an internal diameter of about 10 mm. and a length of 200 mm. It is filled throughout its middle 100 mm. with granulated copper oxide kept in place by wads of asbestos fiber. The open ends of the tube are closed by elbows of glass capillary tubing which slip within each end of the combustion tube as far as the asbestos wads and are held in place by rubber tubing fitting tightly over the end of the combustion tube and also over the glass capillary. The asbestos shield shown in section at B and in elevation at C sits like a saddle over the middle portion of the tube and keepa the heat from the rubber connections during combustion. The combustion gases pass out the perforations shown in the top of the shield. A thermometer standing in the tube of the 1 Am. Chem. Jour., 17, 688 (1895). *Jour.Gasbeleucht, 41,764 (1898). EXPLOSION. AND COMBUSTION METHODS . 57 shield with its bulb touching the combustion tube indicates the temperature at which hydrogen is being burned. The whole volume of the gas from which C02, C n H 2 n and O 2 have been removed is used for the analysis. The copper oxide tube is connected to the burette on one side and on the other to a phosphorus pipette which has been previously filled with air and now contains nitrogen. This nitrogen is allowed to flow through the combustion tube and out into the air through the burette stopcock flushing out the air in the tube and rendering Jl. FIG. 13. Quartz combustion tube filled with copper oxide. unnecessary the troublesome correction involved in Jaeger's original method. The nitrogen is all driven out of the phos- phorus pipette, and the water in it blown to a mark arbitrarily fixed on the capillary stem of the pipette and the burette stop- cock turned so that connection with the outside air is shut off, the burette also remaining closed. The gas burner under the combustion tube is lighted and adjusted so that the thermometer inserted in the jacket and resting on the combustion tube shows about 250 C. The expanding nitrogen in the combustion tube is free to pass into the phosphorus pipette. When the 58 GAS AND FUEL ANALYSIS combustion tube is hot, the burette stopcock is opened and the gas passed slowly into the phosphorus pipette and back again so that it has all been exposed twice to the action of the copper oxide. A few cubic centimeters of the gas are again passed into the phosphorus pipette, the burette stopcock is closed and the flame removed. If the combustion tube is of glass it must be slowly cooled to room temperature but if it is of quartz it may be sprayed with water or wrapped with a wet cloth until it again reaches room temperature. As the tube cools there is sucked back from the pipette some of the gas purposely placed there and when it is thought that the tube has reached room temperature the liquid of the pipette is again brought to the mark in the capillary which was used at the com- mencement of the test. If after adjustment has been made the water of the pipette continues to rise in the capillary it is proof that the combustion tube has not yet reached room temperature. The volume of the gas in the burette is now measured and a caustic potash pipette substituted for the phosphorus pipette. This requires somewhat careful manipulation for the combustion tube is still filled with gas which must not be allowed to diffuse into the air. To accomplish the substitution the stopcock of the burette is opened and the gas drawn out of the capillary of the phosphorus pipette into the burette until the liquid has mounted as high as the rubber connecting tube. The glass capillaries of the pipette and the combustion tube are separated enough to allow a clamp to be screwed on the rubber tube and the phosphorus pipette is disconnected and replaced by a caustic pipette whose liquid before making the connection is blown practically to the top of the capillary by the help of a rubber tube attached to the second bulb. With care this substitution of one pipette for the other may be made with an error of only a few tenths of a cubic centimeter. The carbon dioxide formed from the CO is then determined. Since it is not feasible to drive all the gas from the combustion tube into the caustic the gas should be passed back and forth several times. The method of calculation of the H 2 and CO follows from the equations: H 2 +CuO = H 2 0+Cu CO+CuO = C0 2 +Cu EXPLOSION AND COMBUSTION METHODS 59 The metallic copper has practically the same volume as the copper oxide. The CO 2 has the same volume as the CO. The H2 completely disappears. Therefore the contraction in volume after heating to 250 is equal to the H 2 , and the C0 2 is equal to the CO. Methane is estimated by heating the combustion tube to redness and slowly passing the gas back and forth into the caustic pipette. Methane burns somewhat slowly and it is wise to pass it back and forth at least four times. The decrease in volume is read after the combustion tube has been cooled as before. The equation for the reaction is: CH 4 +4CuO = 4Cu+C0 2 +2H 2 0. If the gas had been passed back and forth into a pipette filled with water during the combustion there would have been no change in volume but since the gas was passed into the caustic pipette during the combustion process and the C0 2 was absorbed the contraction equals the methane. It is assumed in this calculation that CH4 is the only one of the paraffine series present. This is usually the case but natural gas, Pintsch gas, carburetted water gas and gas from coal dis- tilled below a red heat may contain small proportions of ethane and possibly higher homologues. Pentane vapors are present in many samples of natural gas. Any two constituents such as methane and ethane may be determined by this method if during the combustion at a red heat the gases are passed back and forth into the phosphorus pipette or other pipette filled simply with water and the contraction after combustion measured and then the COa determined. The calculation follows from the equations: CH 4 +4CuO = 4Cu+C0 2 +2H 2 0. C 2 H 6 + 7CuO = 7Cu+2CO 2 +3H,O. In the case of CH 4 , the volume is the same after combustion as before. In the case of C 2 H 6 the volume has increased by a volume of CO 2 equal to the C 2 H 6 . Any increase in volume after combustion is reported as C 2 H 6 and the volume of the C0 2 less twice the C 2 H is reported as CH 4 . In case pentane is present the increase of volume after com- 60 GAS AND FUEL ANALYSIS bustion is four volumes for each volume of pentane according to the equation C 5 H 12 +16CuO = 5CO 2 +6H 2 0. It is not usually feasible to distinguish by analysis between the various higher hydrocarbons. The copper oxide has been partially reduced to metallic copper in the combustion and must be re-oxidized by drawing air through the red hot tube. This may be done very conveniently by means of an aspirator since no attention on the part of the analyst is required. This method is perhaps the most accurate of the technical methods for the estimation of CO, H 2 and CH 4 and is to be com- mended because it does not involve any special equipment which cannot be made by the analyst himself. It is somewhat slower than the explosion methods but if a quartz tube is available it is not a tedious process. A quartz tube is highly desirable since glass tubes always break after a time and in breaking usually spoil the analysis. There is no danger of oxidation of nitrogen as in the other methods and a large sample of gas may be taken thus reducing the errors of observation to a minimum. The greatest liability to error comes from incomplete combustion of the hydrocarbons. Ethane is especially difficult to burn and it is desirable to repeat the combustion on the gas residue after the C0 2 has been absorbed to make sure that there is no further formation of C0 2 . 12. Detection of Minute Quantities of Combustible Gas. The methods described in the preceding paragraphs have been applied in specialized and portable apparatus for the detection of small amounts of combustible gases, as for instance in the air of mines, and also for the removal of small amounts of poisonous gases, such as carbon monoxide, by gas masks. In one type of detector for mine gases, the methane is oxidized by a glowing spiral and the heat of combustion is measured by the resultant change in the temperature or resistance of the platinum spiral. In another type the moisture formed by combustion is used as the indicator. Especially sensitive forms of copper oxide and other oxides were developed during the war by the Chemical Warfare Service and EXPLOSION AND COMBUSTION METHODS 61 were used in special types of gas masks. The oxides were so sensitive that they were active and protected the wearer against carbon monoxide even in winter weather. The same oxides have been used in portable apparatus for the detection of carbon mon- oxide through the heat developed by combustion. A still more sensitive detector for carbon monoxide in air is found in a mixture of iodic anhydride with fuming sulphuric acid and pumice. The detection is based on a green color of a transitory nature pro- duced in this mixture by minute quantities of carbon monoxide. By this means as little as one hundredth of a per cent, of carbon monoxide can be detected, and. by intensity of color an approxi- mate estimate of concentration up to 1 per cent, can be made. 13. Oxygen by Explosion or Combustion. The volume of oxy- gen in a gas may be determined by explosion or combustion with an excess of hydrogen. The gas must be free from hydrogen and carbon compounds. If an excess of pure hydrogen is added and the mixture exploded or burned with a platinum spiral ac- cording to the methods given in this chapter, the oxygen may be calculated from the decrease in volume. 14. Nitrogen. There is no desirable method for the direct determination of nitrogen, which is always taken by difference. This is very unsatisfactory since, although some of the errors in analysis may compensate each other, there is a tendency in a long analysis for them to pile up on the nitrogen. The Jaeger method of combustion with copper oxide just described allows all of the gases other than nitrogen to be re- moved in a single process and affords a valuable check on the accuracy of the longer analysis. A sample of 100 c.c. of the gas to be analyzed is passed through the combustion tube at red heat and into caustic. The CO 2 , CO, H 2 , and C n H m will all disappear in the process as will also the oxygen if it is present in only small amount. The residue will be nitrogen and possibly oxygen which may be removed by phosphorus. 15. Form of Record of Gas Analysis. There may of course be great variation in methods of keeping records of gas analyses. The record should in every case however be full enough to show every step of the operation. The following record is given as a sample. 62 GAS AND FUEL ANALYSIS ANALYSIS OF ILLUMINATING GAS at Chemical Laboratory, University of Michigan, Sept. 9, 1917 Sample 99.4-0.3 = 99.1 c.c. After KOH 97.4 CO 2 =2.0 c.c. =2.0% After Br 2 92 . 6 C 2 H 4 , etc., 4 . 8 c.c. 4 . 8 After P 92.2 O 2 0.4 c.c. 0.4 After Cu 2 Cl 2 84.8-0.3=84.5 CO 7. 4 c.c. 7.5 First explosion: Sample 9.4-0.3= 9.1 Air to 97.9-0.3 = 97.6 After explosion 83.0 Contraction =14.9 After KOH 79.2 CO 2 = 3.8 After P 71.0 Excess O 2 =8.2 Calculations: Factor to give percentage Q 1X99T = 9 ' 37 CH 4 =3.8X9.37 =35.6% H 2 =2/3(14. 9-2X3. 8)9. 37 =45.6% N 2 = [9.1-(3. 8+4. 87)]9.37 = 4.0% Exploding gases: 3.8c.c. CH 4 + .6c.c. O 2 =11. 4 c.c. 4.9c.c. H 2 +2.4c.c.O 2 = 7.3 c.c. 18.7 Non-exploding gases: 97.6-18.7 78.9 _, . non-exploding 78.9 Ratl exploding = 18-7 = 4 - 2 Second explosion: Sample 9.2-0.3= 8. 9 c.c. Air to 95.0-0.3 = 94.7 After explosion 80 . 4 Contraction 14 . 6 After KOH 76.7 CO, 3.7 After P 70.0 Excess O 2 6.7 Calculations : Factor to give percentage g Qx99 1 =9 ' 58 CH 4 =3.7X9.58 =35.4% H 2 = 2/3(14.6-2X3.7)9.58) =46.0 N 2 =[8.9-(3.7+4.8)]9.58 = 3.8 EXPLOSION AND COMBUSTION METHODS 63 Exploding gases: 3.7c.c. CH 4 +7.4c.c. O 2 = ll.lc.c. 4.8c.c. H 2 +2.4c.c. O 2 = 7.2 18.3 Non-exploding gases: 94.7-18.3 =76.4 . non-exploding 76 . 4 Ratio r p - = <0 =4.2 exploding 18.3 Summary of analysis: I II Average CO 2 2.0 2.0% C 2 H 4 , etc., 4.8 4.8 O 2 0.4 0.4 CO 7.5 7.5 CH 4 35.6 35.4 35.5 H 2 45.6 46.0 45.8 N 4.0 3.8 3.9 99.9% CHAPTER V VARIOUS TYPES OF APPARATUS FOR TECHNICAL GAS ANALYSIS 1. Introduction. Chapter II describes the apparatus which the author believes best adapted to technical gas analysis and gives detailed directions for its manipulation. The present chap- ter will describe various other forms of technical apparatus especially those which first embodied valuable principles. The number of modifications is legion and no attempt will be made to even enumerate them. The order of description will in general be historical. 2. Schlosing and Rolland's Apparatus. Perhaps the earliest FIG. 14. Schlosing and Holland's apparatus. successful attempt to devise an apparatus for the rapid analysis of industrial gas was that of Schlosing and Holland 1 who de- vised a simple apparatus which foreshadowed closely the modern type. Their apparatus apparently attracted little attention 1 Annales de. Chim., Series 4, t.14, 55 (1868). 64 APPARATUS FOR TECHNICAL GAS ANALYSIS 65 partly because its description was embodied in a long article on the ammonia-soda process whose title did not contain any ref- erence to gas analysis. The original cut of their apparatus is reproduced as Fig. 14 as it still may serve as a model for a chemist who has to improvise his own apparatus. The following descrip- tion of the apparatus is taken from the original work. In the upper left hand corner of the cut are seen four lead pipes of small diameter coming from various pieces of apparatus in the plant. A rubber tube d connects any one of these with the copper tube t to which is attached an aspirator. The burette a terminates at the top in a tee of almost capillary tubing, one arm of which con- nects to the gas supply through the cock r and the other to the pipette b. No mention is made of a clamp on the rubber tube between a and b but necessarily such must have been used. To draw a sample of gas the cock r is opened and the levelling bottle c is raised until the water fills the burette and reaches r. The gas formerly in the burette is now in the pipe t out of which it is swept by the stream of gas which is constantly flowing. The bottle is then lowered until the gas has passed below the 100 mark. The aspirator is stopped, the rubber tube d disconnected and the bottle raised, r being again opened until the level of the water in the burette is at 100 and is at the same time coincident with the level of the water in the levelling bottle. The burette will then contain 100 volumes of gas at atmospheric pressure. The gas is then passed back and forth into the absorption pipette b filled with caustic potash and containing glass tubes to increase the absorptive surface. The volume in the burette is then read as before and the decrease in volume reported as C0 2 . 3. Orsaf s Apparatus. The original form of the Orsat 1 appa- ratus is practically the same as that frequently used today, as will be seen by Fig. 15 which is a reproduction of the original cut. It consists of a water-jacketed gas burette terminated at its upper end by a branched glass capillary tube. The pipettes, in order from right to left, contain caustic potash, alkaline pyrogallate and cuprous chloride. The cock I on the branched capillary serves for the connection of a platinum capillary in which hydrocarbons mixed with air, and added hydrogen if necessary, may be burned. The sample of gas is brought to the burette by the water-aspira- 1 Annales des Mines, Series 7, t.8, 485 (1875). 5 66 GAS AND FUEL ANALYSIS tor KLM which sucks a rapid stream of gas through the cock R and the dust filter P. The operation of the apparatus will be evident to anyone who has read the three preceding chapters. Many modifications of this burette have been proposed since it was first described, but the principle has not been altered. One group of workers has increased the complexity of the apparatus in an attempt to increase speed of manipulation. The most note- worthy change of this sort is probably the introduction of the FIG. 15. Orsat apparatus. Original form. bubbling pipette in which, by a three-way cock on the top of each pipette, the gas is made to pass down a central tube and bubble up through the liquid of the pipette to be later drawn from the top of the pipette when the three-way cock is thrown to its second posi- tion. There are decided objections to complication in any form of apparatus which may receive rough treatment in transportation and which is frequently handled carelessly by its operators. The usual modifications of the Orsat apparatus possess at least four glass stopcocks on the various outlets of the branched tee. APPARATUS FOR TECHNICAL GAS ANALYSIS 67 Unless the apparatus is always manipulated by a skilled operator it is almost inevitable that some of the alkaline reagent from the pipettes will be drawn into these stopcocks. It is apparently equally inevitable that the cocks will as a consequence stick and become broken. The branched tee is itself a source of trouble since it is fragile and difficult to clean when stopped. In Fig. 16 is shown a modification of the Orsat apparatus due to Allen and Moyer which commends itself for its simplicity and durability. The capillary glass tube is re- placed by one of hard rubber and the glass stopcocks are replaced by pinchcocks which are practi- cally as satisfactory. The pip- ettes themselves are of the test tube type and are closed at the top with a soft rubber stopper which is pressed against the upper shelf by the screw which supports the cup in which each pipette rests. This gives a firm and- yet elastic support for the pipettes which prevents breakage in shipment. The method of operating all Orsat burettes is the same. The / water of the burette should be saturated with gas similar to that ' which is to be analyzed. The gas is taken into the burette through the end of the capillary projecting out of the left hand side of the case. If the sample is drawn directly from the smoke flue the precautions given in Chapter I on Sampling must be observed. In any case the burette is filled with gas which is then wasted through the fourth side arm into the outside air, thus getting rid of the air which was in the capillary tube of the instrument. The sample of gas is now drawn into the burette and measured with the precautions given in Chapter I. addition to the gas which is in the burette there is a volume about-J. c.c. in the capillary tube which is entirely neglected. The gas is passed into the caustic pipette and CO 2 determined FIG. 16. Orsat apparatus. Allen- Moyer modification. * tie In, of I ed. 1 68 GAS AND FUEL ANALYSIS as described in 1 of Chapter III, then into the second pipette filled with pyrogallate for oxygen ( 4 of Chapter III), then into the third pipette filled with cuprous chloride for CO ( 6 of Chapter III). This apparatus is much used for analysis of smoke gases and it is sufficiently accurate for this purpose. The errors which arise from the failure of the gas in the capillary to come into contact with the reagent will hardly be more than 0.1 c.c. for each of the gases. The tendency will be for the oxygen to be slightly high due to some CC>2 which did not get into the caustic pipette and for the CO to be high due to oxygen which did not get into the pyrogallate pipette. The Orsat apparatus is not usually employed for gases like 1 illuminating gas where many constituents have to be deter- mined, although other absorption pipettes and even explosion pipettes have sometimes been made a part of the instrument. The errors mentioned above due to the gas remaining in the capillary increase with the length of the capillary and it becomes preferable to use the burette with detachable pipettes. 4. Bunte's Burette. Dr. H. Bunte 1 in 1877 described his burette which although in the main obsolete still has some uses and is illustrated in Fig. 17. It consists of a burette closed at the top by a three way cock a carrying a funnel tube, and at the bottom by a cock b. The zero of the burette is somewhat above the lower cock. A levelling bottle such as is used with the ordinary Hempel type of gas burette is to be connected to the lower cock. The sample of gas is drawn into the burette as usual and its volume may be read as usual. The method of reading the volume prescribed by Bunte is unusual. A sample of gas slightly larger than 100 c.c. is to be taken into the burette and the volume com- pressed until it reads exactly 100. Water is now poured into the funnel tube to the mark m etched upon it and cock a is opened to communicate with the funnel tube which is unstoppered. Some gas bubbles through the cock and the liquid above it and escapes into the air. The bore of the cock is so small that no water flows down and after the bubbling has ceased the volume of gas in the burette is still 100 c.c. measured under the pressure 1 Jour, fur GasbeL, 1877, 447. APPARATUS FOR TECHNICAL GAS ANALYSIS 69 of the atmosphere plus the column of water in the funnel tube. The volume of the gas is always to be read under these condi- tions. In order to introduce an absorbent such as NaOH into the burette a partial vacuum is produced by opening the lower cock and lowering the levelling bottle. The cock 6 is then closed, the levelling bottle disconnected and the reagent in a small dish is placed below the cock 6 so that the tip of the cock is immersed. On opening the cock some of the rea- gent will be sucked into the burette. The burette is then shaken to facilitate absorption, care being taken to hold it only by the tips so that the heat of the hands will not change the temperature of the gas. To read the volume after absorption the funnel tube on the top of the burette is again filled with water, and the upper cock opened. Water flows into the burette and more water is added to the funnel until with the upper cock still open the water remains stationary on the mark m. Conditions are now as they were at the first reading and the volume is again read. The diminution in volume if NaOH was the reagent, is as usual reported as CO2. Oxygen is determined by alkaline pyrogallate (4 of Chapter III). To avoid diluting the reagent the dilute caustic in the burette is sucked out as far as possible and 20 c.c. of the pyrogallate introduced. The burette is shaken at intervals for five minutes and then rinsed with fresh water introduced through the funnel tube, the pyrogallate being allowed to flow out of the lower stopcock. As soon as the walls of the burette are rinsed clean the bottom stopcock is closed and the volume read as before with the top stopcock open and the funnel tube filled with water. Carbon monoxide is absorbed by acid cuprous chloride as usual ( 6 of Chapter III), but a very considerable amount of preliminary manipulation is necessary. The alkaline pyrogallate must be thoroughly washed out by water flowing through the burette and then replaced by HC1 sp. gr. 1.12. When this has been done the Cu 2 Cl 2 reagent may be added and after the absorption the volume measured with the funnel tube filled with dilute HC1. FIG. 17. Bunte gas burette. 70 GAS AND FUEL ANALYSIS The only advantage which a burette of this type can claim is its portability. Its manipulation is cumbrous, the introduction of the large volumes of wash water changes the composition of the gas, and the temperature of the burette, which is usually not jacketed, is almost certain to change unless very unusual precautions are observed to keep it from contact with the hands of the operator and to have the temperature of the reagent and wash water the same as that of the room where the operation is being carried on. The Orsat apparatus is preferable for almost all purposes. 6. Chollar Tubes. Mr. B. E. Chollar 1 in 1888 modified A. 1 t L ; : |: r =f v\ \i 7 F FIG. 18. Chollar tubes for gas analysis. Cooper's eudiometer and produced a very practical and simple combination burette and pipette. In Fig. 18 at A are seen three of the Chollar tubes in a rack. The zero point is the top of the bulb and the graduations start at the bottom of the bulb and ex- tend down to the bend in the tube. The bulbed portion may 1 Proc. Western Gas Association, 1893, 219. APPARATUS FOR TECHNICAL GAS ANALYSIS 71 occupy varying proportions of the total volume and since it is not graduated a tube must be chosen proportioned properly for the analysis to be made. Of the three burettes shown at A, that on the right is called a 10 per cent, burette because the graduations cover only 10 per cent, of the total volume. The middle burette is a 25 per cent, burette, and the burette on the left is a 50 per cent, burette, provided also with an upper stop- cock which is convenient, but not necessary for all the forms. It is assumed that a plentiful supply of gas for analysis is avail- able and that it is under pressure. A rubber tube is slipped into the burette round the bend and up to the top through which gas is blown until the air is displaced. It is safer to fill the tube with water and displace this with the gas. The rubber tube slips into the burette more readily if it is wet. When the burette is completely filled with gas it is immersed in the cylinder of water B far enough to seal the outlet and the rubber tube is withdrawn. The burette is now pressed completely under the water and kept there by the weighted cover C for a few minutes until the gas has attained the temperature of the water which should be at room temperature. The top of the burette is then grasped by the tip of the fingers to avoid warming the gas and the burette is raised until the meniscus inside of the burette coincides with the surface of the water in the glass cylinder when the gas vol- ume is read at atmospheric pressure. In case the volume of the gas has increased through expansion so that the meniscus is below the graduations a portion of the gas must be removed by closing the lower end of the burette with the thumb while it is still under water and then by raising and tilting the burette causing a few bubbles to pass into the short arm from which they escape into the air when the thumb is removed, To introduce reagents, a portion of the water is sucked from the short arm of the burette by a pipette as shown at D. Suf- ficient water must of course be left to seal the burette. Suf- ficient reagent such as caustic sodajs then introduced to com- pletely fill the short arm which is tightly closed by a rubber stopper or the thumb. The burette is then inverted and shaken until absorption is believed to be complete. The gas which may have gotten into the short arm is now worked back into the body of the burette by turning the burette almost horizontal and the 72 GAS AND FUEL ANALYSIS burette is again immersed in the large cylinder. The rubber stopper is removed after the outlet is sealed with water, water enters to replace the gas absorbed and the volume of the gas is read as at first. The usual reagents for carbon dioxide, unsaturated hydro- carbons, oxygen and carbon monoxide may be used. A solution of arsenious oxide is recommended for hydrogen sulphide. To wash out one reagent before adding another the burette is placed on the stand shown at E in Fig. 18 with its open arm pointing down in a beaker of water. The reagent being heavier than water tends to flow out. The washing may be accelerated by passing water into the burette through a rubber tube. The instrument is very readily portable and after a little experience results of considerable accuracy may be rapidly obtained. 6. Instruments for Recording Carbon Dioxide in Flue Gases. There are a number of commercial instruments on the market which aim to make a continuous record of the percentage of carbon dioxide in stack gases. These instruments usually derive their motive power from an aspirator which sucks a sample of flue gas through a dust filter and then through the analyzing mechanism. Most of these mechanisms imitate the action of the analyst when he sucks in a buretteful of gas, passes it into caustic solution and then measures the decrease in volume. In the machine the filtered gas is made to exactly fill one measuring chamber, then passed into caustic and then into a second chamber where its volume is measured by the height to which it raises a small gas holder or by some other mechanical means. A pen attached to the gas holder indicates on a chart the height of the gas holder and, if the chart and instrument are correctly adjusted, it records directly the percentage of carbon dioxide. Machines of the above type give intermittent readings. The Uehling carbon dioxide machine operates continuously and with- out measuring chambers. It measures the percentage of carbon dioxide by recording the difference in suction caused by the dimi- nution in volume of the gas after removal of the carbon dioxide by caustic. Two standard diaphragms are inserted in the line through which the gas flows and the dry caustic absorption cylinder and the recording manometer are placed between them. With air flowing through the apparatus a certain suction will be registered. APPARATUS FOR TECHNICAL GAS ANALYSIS 73 With gases containing carbon dioxide a higher suction will be registered because of the smaller volume of gas passing the second diaphragm. A detailed description of several of these instru- ments together with results of tests on them has been made by the Bureau of Mines. 1 Some of them work very well but it must be remembered that none of them are fool proof and that ail of them require rather frequent and expert adjustment. 7. Methods of Gas Analysis Depending upon Thermal Con- ductivity. This method depends on the difference in thermal conductivity of different gases. A carefully calibrated platinum wire is heated with a constant electric current. The gas to be tested is then made to flow through the tube containing the wire and the difference of the resistance of the hot wire is measured. If the gases have not reacted with each other or been subjected to chemical change under the influence of the hot wire, this meas- urement allows the calculation of the heat conductivity of the gas, and in case only two constituents are present it permits the proportions of each to be calculated. By the use of a series of conductivity tubes with proper absorbents or oxidizing appa- ratus between them a continuous test may be made of the com- position of gases with a number of different constituents. The apparatus may be made recording. It is largely one of the devel- opments of the Bureau of Standards for war needs 2 and is hardly yet a technical instrument for general purposes. 8. Gas Analysis by Optical Methods. The refraction of a beam of light passing through a gas is in general proportional to the density of the gas. While the differences in the index of refraction are small they may, with rather complicated appa- ratus, be accurately measured. The method is most readily ap- plicable where there are only two constituents in a gas mixture. An instrument called the interferometer or refractometer has been devised for this work but it is to be regarded more as a scientific instrument than one for technical use, although Burrell and Seibert 3 state that the instrument may be successfully used for technical analysis of the following: mine air for carbon dioxide and methane; manufactured hydrogen, chlorine and other gases for purity; illuminating gas for benzol, and flue gases for carbon dioxide. Edwards 4 has published a critical review of the appli- cation of the interferometer to gas analysis. 1 J. F. Barkley and S. B. Flagg. Instruments for Recording Carbon Dioxide in Flue Gases. Bulletin 91. Bureau of Mines, 1916. 2 Weaver and others, Jour. Ind. and Eng. Chem. 1920. T> _J CHAPTER VI EXACT GAS ANALYSIS 1. Historical. Lavoisier in his Traite* Elementaire de Chemie published in 1789 devoted Chapter II of Book III to Gasometry or "The Measurement of Weight and Volume of Gaseous Sub- stances." He described eudiometers, a gasometer of the bell jar type, and methods of separation of certain gases by absorp- tion and explosion as well as the mathematical method of making correction for temperature and pressure. Bunsen and Playfair 1 in a paper "On the Gases Evolved from Iron Furnaces" presented in 1845 what was perhaps the first serious attempt to develop methods of gas analysis for technical investigation. The methods published in this paper formed the basis of Bunsen's classic book " Gasometrische Methoden" published in 1857. Bunsen used a graduated cylindrical eudi- ometer inverted over a trough of mercury both for measuring the volume of the gas and for carrying out analysis by absorp- tion and explosion. It was necessary to determine the tempera- ture and pressure of the gas when each reading of volume was made and to make arithmetical corrections to bring the volume to standard conditions. It was possible to work accurately with his apparatus but it has been replaced by simpler forms which allow more rapid work. Regnault and Reiset 2 in a paper on the. respiration of animals developed a eudiometer which was capable of accurate work but was very cumbrous. Doyere in J848 3 exhibited before the French Academy of Sciences apparatus for gas exact analysis and two years later 4 presented details of the remarkably complete and ingenious apparatus which he had devised. He used a separate pipette 1 British Ass. for Advancement of Science, 1845, 142. 2 Ann. de Chim. et de Phys. (3), 26, 299 (1849). 3 Comptes rendus de I' Academic de Science, Feb., 1848. 4 Annales de Chimie, 3 Series, 28, 5, (1850). 74 EXACT GAS ANALYSIS 75 for each reagent, measured his gases saturated with moisture instead of dry, and by means of a mechanical compensator avoided all corrections for change in gas volume due to tempera- ture and pressure. He absorbed carbon dioxide by caustic potash and estimated oxygen by explosion or by absorption with ammoniacal cuprous chloride, two pipettes being used in series followed by a third containing dilute sulphuric acid. He also studied the estimation of hydrogen by explosion. W. Hempel in the second edition of his Gas Analysis (1889) described a modification of Doyere's method in which he measured the gas volume in a bulb, varying the pressure of the gas at each reading so that its volume always filled the bulb to a definite mark. The pressure under which the gas stood was then meas- ured and correction mathematically made to find the volume under standard conditions. 2. General Methods. These earlier processes with their complications were necessitated by the imperfections of apparatus, especially stopcocks which could not be relied on to be gas- tight. With the development of reliable stopcocks came the development of apparatus so that now there is little need to use any of these older processes. The methods of exact gas analysis are now in general the same as those employed in technical analysis but with greater attention paid to the elimination of minor errors. The gas burette is graduated more accurately and an attempt is made to read to hundredths of a cubic centi- meter instead of tenths. Correction is made either arithmetically or mechanically for variations in temperature and pressure of the gas during analysis. Mercury is used instead of water as the confining liquid in the burette and errors due to diffusion in the pipette are prevented. Special methods are sometimes introduced for the estimation of minute constituents of the gas. 3. Corrections for Temperature and Pressure. According to the law of Boyle the volume of any gas varies inversely as the pressure and according to the law of Gay Lussac all gases expand under constant pressure by the same amount about 1/273 of their volume when heated from C. to 1 C., and the same for each succeeding degree. The volume of a gas is almost univer- sally expressed in scientific work as that which it would occupy 76 GAS AND FUEL ANALYSIS if completely dry and at a temperature of C., and a pressure of 760 mm. of mercury. For technical purposes the gas is fre- quently calculated to the volume which it would have at 60 F. and 30 in. of mercury pressure when saturated with water. The formulae for this calculation are given in 4 of Chapter VII on the Heating Value of Gas. The formula for the reduction of the volume of a dry gas to and 760 mm. dry follows from the laws stated. v __ V __ p^ Vp " M 7 r V -(l-.00367t)760 ~ 7 If the gas as measured was saturated with moisture at say 60 F. and it should actually be chilled to C. most but not all of the water would be condensed. It is usual in calculations to assume either that the gas becomes completely dried which is the usual assumption or that the water all remains in the vapor state which is less frequently assumed. The formula given above will be the one used in the latter case where the water vapor is assumed to obey the gas laws without condensing, just as nitrogen or hydrogen would. If the gas is to be reduced to standard conditions, dry, the volume of the water vapor must be deducted. Since the assumption is that the gas is saturated with moisture at the temperature "t" the proportion of moisture will be a constant which may be expressed either in terms of volume or more conveniently in terms of barometric pressure. Table I in the Appendix gives the vapor pressure of water, "a," for each 1 C. within the limits usually required. The formula for reduction of the volume of a gas read when saturated with moisture, to its volume when dry at C. and 760 mm. is therefore as follows: (1 -.003671)760 If the readings are in the Fahrenheit scale and in inches of barometric pressure the formula becomes V(p-a) EXACT GAS ANALYSIS 77 A. Rubb Con, Calibt ati'on MaH ection 4. Description of Gas Burettes.^ The burette for technical analysis does not allow an accuracy greater than 0.1 c.c. and the probable error is more nearly 0.2 c.c. The errors are in part due to the coarse graduations of the cylindrical burette tube and in part to liability to change in the temperature and pressure under which the gas volume is read. A device for automatically correcting for change in tempera- ture and barometric pressure was devised by Petterson 1 in an ap- paratus for the analysis of air. This was quickly adapted by Hempel to his gas burette and as used by him consisted of a closed tube connected to one arm of a manometer whose other end connected with one tip of a three- way cock at the top of the bur- ette. The volume of the gas instead of being read at a change- able atmospheric pressure was to be read at the unchanging pres- sure in the closed compensating tube. The modification of this type of apparatus developed by the author 2 is shown in Fig. 19. The burette has a specially bored stopcock as shown in the sketch which allows communica- tion to be established through the stopcock between the burette and the compensating tube. The latter consists of a U tube manometer shown in A of Fig. 19 connected to the burette by a single rubber connection at a, placed so that gas never comes in contact with the rubber and there can be no possibility of leakage a fault of the Hempel type of apparatus. The other end of the U tube bends down and terminates in a tube about the same diameter as the burette and l Zeit. anal Chem., 25,467 (1886). 2 Jour. Am. Chem. Soc., 22, 343 (1900). ' c. TT FIG. 19. Details of gas burette with automatic compensator for temperature and pressure. 78 GAS AND FUEL ANALYSIS sealed at the bottom. At the top of the U is a capillary tip which in practice is fused shut as explained later. In mounting the burette which is assumed to be clean and entirely taken apart the manometer is connected to the burette at a by a piece of good black rubber tubing wired in place, and the large rubber stopper at the lower end of the burette is pushed up until it rests snugly against the bottom of the comparison tube and presses the glass parts at a firmly together within the rubber tube. The glass water jacket tube is then slipped over the top of the burette and onto the lower rubber stopper and the upper split cork fitted into place. The burette is then placed in a stand such as is shown in Fig. 21. The levelling bottle is also connected to the burette as in technical analysis with the added precaution of wiring the rubber stopper into the burette so that it may not be blown out through the weight of the mercury column. The process of wiring in a rubber stopper is simple but since begin- ners are sometimes at a loss to accomplish it the following descrip- tion is given. A piece of rubber tubing is first slipped over the bottom of the burette to give a soft surface for the wire to grip. A piece of copper wire about 18 gage is annealed by passing several times through the flame of a Bunsen burner and a piece about 4 in. long is twisted in the middle of one about 8 in. long to form a T. The longer piece is then bent around the burette and twisted till it fits snugly. Its two ends are now brought over the rubber stopper and twisted with the free end of the 4-in. piece of wire until the cork is firmly pressed in. To prepare the comparison tube for use, a few drops of water are to be brought into its lower portion to saturate the air it contains, the manometer is to be filled with mercury and the capillary tip is to be fused shut. The two first operations can be conveniently accomplished in one operation. The burette is filled with mercury on whose surface are a few drops of water. By turning the stopcock to the position shown at D of Fig. 19, the water with the mercury following it may be passed through the stopcock and the U tube and over into the comparison tube. The progress of the mercury is to be arrested when the water has trickled down the comparison tube and the mercury is to be drawn back until there is just enough left in the U tube to fill it approximately to the calibration mark when there is atmospheric EXACT GAS ANALYSIS 79 pressure on both limbs of the U as shown in Fig. 19 at A. This may be readily accomplished after one or two trials. The cap- illary tip of the comparison tube is now to be sealed with a blow pipe and the burette is ready for use. This burette has the advantage over the usual technical type that its readings are unaffected by variations in external tempera- ture and pressure, but the volumes themselves cannot be read more accurately than with the technical type unless a reading telescope is employed. Even with this aid to the eye the results are not always certain for the presence of a few drops of water on the mercury alters the shape and boundaries of the meniscus. 6. The Bulbed Gas Burette for Exact Analysis. The idea of converting the burette into a string of bulbs connected to a side arm burette apparently originated with Bleier. 1 The author has utilized this suggestion in the design of the burette shown schematically in Fig. 20, and in perspective in Fig. 21 ; it consists of a burette with stopcock and manometer as already described. The main body of the burette contains twelve bulbs, each of a capacity approximating 12 c.c. A line is etched on each constriction and the capacity of the bulb between these marks is determined. Starting from the capillary above the top bulb a side arm springs, terminating in a small burette with total capacity of 15 c.c. and graduated in 0.1 c.c. Both these burette tubes connect at the bottom by means of heavy rubber tubes and a Y with a stopcock on each arm, to a common levelling bottle. A screw clamp on each rubber tube serves for the exact adjustment of the mercury. To measure a gas, the stopcock is placed in position shown at C of Fig. 19 and the mercury in the bulbed tube brought to the mark in one of the constricted portions by opening the proper stopcock on the Y and raising or lowering the levelling bottle. When adjusted, the mercury is held in its proper position by closing the stopcock on the Y. The stopcock leading to the small burette tube is then opened and the gas brought to approximately atmospheric pressure by proper change in the mercury level. The three-way stopcock at the top of the burette tube being now turned to position shown at D in Fig. 19, the burette is brought into connection with the manometer, which is properly set by further . d. Chem. Ges., 31, 1, 238. 80 GAS AND FUEL ANALYSIS changing the level of the mercury in the small burette. The final adjustment in both burettes is made by the screw clamps on the rubber tubes. When the gas burette is used in the manner indicated the gas volume is read under conditions which are constant but not necessarily known. It is not usually necessary in gas analysis Rubber Connection FIG. 20. Details of bulbed gas burette for exact gas analysis. FIG. 21. Burette for exact gas analysis. to know the absolute value of a gas but it may be obtained if the temperature and barometric pressure are noted at the time the tip of the manometer tube is sealed. The volume of the gas is to be read when the level of the mercury in the two arms of the man- ometer is the same. The gas has then the same volume which it would have had at the temperature and pressure prevailing when the manometer was sealed. Correction to standard con- EXACT GAS ANALYSIS 81 ditions may be made mathematically as indicated in a preced- ing paragraph. This procedure for reading the gas volume is not as accurate as the one recommended for gas analysis where the mercury is brought tangent to the rim of a metal sleeve but it is amply accurate for most purposes. 6. Manipulation of Gas Burette for Exact Analysis. It is necessary to discharge from the burette all of the residual gas in- cluding that in the manometer tube. This may be accomplished by discharging most of the gas from the burette into the air and then by turning the stopcock and lowering the levelling bottle, drawing into the burette the gas from the manometer until the mercury of the U tube is just at the stopcock. The gas in the burette may then be discharged into the air. If there is danger that alkali has been introduced into the burette in a previous operation it is advisable to draw into the burette a few cubic centimeters of acidulated water and with it rinse down the walls of the entire burette, subsequently expelling it. This also makes certain that the walls of the burette are wet so that the gas to be subsequently introduced will become saturated with water. The gas is then introduced into the burette as in technical analysis. If a volume of approximately one hundred cubic centimeters are wanted, nine bulbs are filled with gas at atmos- pheric pressure. The stopcock at the top of the burette is then closed and the gas compressed into eight bulbs. Up to this time the side arm of the burette has remained filled with mercury. The stopcock on the Y is now opened and part of the gas trans- ferred to the side arm until the whole is again under atmospheric pressure as shown by the agreement of the level of the mercury in the levelling bottle held in the hand of the operator, with the level of the mercury in the side arm. The stopcock on the Y is now to be closed and that at the top of the burette is to be turned to make communication with the manometer, the mercury in which drops at once to approximately its proper position. By using the clamps on the rubber tubes as fine adjustments the meniscus in the bulbed tube is to be brought tangent to the mark between two bulbs and also the meniscus in the manometer is to be made tangent to the metal sleeve. The volume of the gas will be read as x c.c. in the bulbs + y 6 82 GAS AND FUEL ANALYSIS c.c. in side burette + z c.c. in manometer. As there are these three readings to be made it is necessary that each be very accurate. Let us see how accurately this may be done. First, the mercury in the bulbed tube is to be brought to a specified mark in a tube of about 5 mm. internal diameter. By means of the screw clamp this may be done with such accuracy that the error is negligible. Second, the volume of gas in the side tube must be read. Each 0.1 c.c. in this tube occupies a space of a little over 2.5 mm. and it is possible to interpolate 0.01 c.c. with the eye with an error of less than 0.02 c.c. Third, the mercury in the manometer must be brought to a definite mark with such exactness that the barometric pressure, under which the gas volume is read, shall be almost identical each time. A difference of 1 mm. of mercury pressure changes the gas volume 0.13 per cent., which on a volume of 100 c.c. equals 0.13 c.c., an error far too large. It was found impracticable to attain the required accuracy when it was attempted to bring the mercury to a mark etched on the glass. The best device was found to be a band of thin, blackened copper, wrapped around the tube and cemented to the glass. It is possible to bring the mercury tangent to the lower surface of this with great exactness. In working with this burette the author is accustomed to make all readings in duplicate, readjusting at all points each time, and to repeat if the two differ from each other by more than 0.01 c.c. Dupli- cates usually agree within this limit. The greatest difficulty found in manipulation is to draw the liquid from the pipette over exactly to the burette stopcock and stop it there. If it gets into the burette, a bubble lodging in one of the capillary tubes fre- quently damps the sensitiveness of the manometer. If this happens the bubble may be shot out of its lodging place by com- pressing the rubber tube above the screw clamp with the fingers. Such a bubble may also be carried into the manometer, where it will obscure the surface of the meniscus. To remedy this it is well to keep 2 or 3 mm. of water on the surface of the mercury in the manometer. This allows a perfectly sharp reading of the mercury meniscus below the water-level. The manometer should respond to a very slight movement of the screw clamp. The advantages of this burette may be summarized as follows : It is a compact burette which, without reading-telescope or other EXACT GAS ANALYSIS 83 accessories, allows the volume to be read with an error of less than 0.02 c.c., compensates automatically for changes of tempera- ture and pressure, and avoids completely all errors due to in- clusion of air or loss of gas in making connections with the absorp- tion pipettes. The disadvantages so far developed are chiefly those inherent in all forms of apparatus which possess a stopcock and rubber connections. Both may leak; but on the other hand both may be kept so tight for limited periods of time as to in- troduce no measurable error. 7. Calibration of Burette. The burette must be carefully calibrated throughout its entire length. This can best be done by weighing the mercury discharged. One cubic centimeter of mer- cury at C. weighs 13.59 grm. Since all that is desired is a relative calibration the mercury need not be strictly pure nor need correction be made for its temperature. Ten milligrams of mercury corresponds to less than one-thousandth of a cubic centimeter so greater accuracy than this in weighing is a waste of time. Wire a stopper carrying a stopcock into the bulbed tube and fasten by a stiff rubber tube a stopcock to the side arm. It is especially important that the rubber tubing should not bulge under the mercury pressure and to prevent this it should be firmly wound with wire. Connect the tips of these stopcocks to the mercury levelling bottle and through them fill the burette with mercurj' completely to the stopcock. The portion first calibrated is the capillary tube from the bottom of the stopcock to the zero of the bulbed tube and the zero of the side arm. After that each bulb and each cubic centimeter of the side arm is separately calibrated. The accuracy of the calibration may be checked by the procedure of a regular analysis where a volume of gas is chosen such that it for instance fills 9 bulbs and a small portion of the side arm. The method of reading may then be changed to eight bulbs plus a considerable volume in the side arm but if the calibration is correct the same volume should, of course, be shown. The vol- ume of the small portion of the manometer tube above the mer- cury may best be determined by filling the burette completely with mercury and then drawing the air out of the manometer tube into the side arm of the burette where it may be measured under atmospheric pressure with an error of a few tenths of a per 84 GAS AND FUEL ANALYSIS cent. Since the total volume is only a small fraction of a cubic centimeter the method is amply accurate. 8. Absorption Methods in Exact Gas Analysis The same reagents may in general be used in exact as in technical analysis. Care should however be taken to see that the reagent has been recently saturated with gas of a sort similar to that which is to be analyzed. In case the reagent is one which does not attack mer- cury the pipette is to be filled with mercury which carries on its surface only a few cubic centimeters of the reagent. If pipettes of the ordinary type are filled with mercury the mercury rising in the second bulb places the gas under considerable pressure and greatly increases the danger of leakage at the rubber connections. An explosion pipette may with advantage be used as an absorption pipette under these circumstances since its stopcock and levelling bottle allow a regulation of the pressure within the pipette. Where the greatest accuracy is not required it is sufficient to lessen the errors due to diffusion by introducing a few cubic centimeters of mercury to form a seal in the ordinary form of pipette. In the case of solutions like cuprous chloride where mercury cannot be used reliance must be placed on the complete saturation of the reagent. Where gases are readily soluble errors due to diffusion mount up rapidly. For instance : A sample of 74 c.c. of acetylene gas when passed into an ordinary NaOH pipette as used in tech- nical gas analysis decreased to 63.9 after quietly standing for three minutes, to 58.0 after a second contact, and to 29.0 c.c. after three minutes shaking with the same reagent. After two more similar periods the residue left in the burette was only 5.2 c.c. A second sample of gas behaved similarly, and by connecting a burette with the second bulb of the pipette the acetylene diffusing through was recovered quantitatively. 9. Carbon Dioxide. Carbon dioxide is usually estimated by absorption in caustic soda as in technical analysis. If other acid gases such as H 2 S, S0 2 , or HC1 are present they may be re- moved by first shaking the gas in a pipette containing KMn0 4 very faintly acidified with H2S04. An increase in volume after this operation would indicate that oxygen had been evolved during the process. If direct evidence of the presence of CO 2 is desired a clear solution of Ba(OH) 2 should be used instead of NaOH in the pipette for CO 2 absorption. The formation of a EXACT GAS ANALYSIS 85 white precipitate completely soluble in HC1 will show the presence of carbonates. 10. Unsaturated Hydrocarbons. Unsaturated hydrocarbons as a class are estimated as in technical gas analysis by absorption with fuming sulphuric acid or bromine water. It is difficult to prevent the errors due to diffusion mentioned in 8 since both reagents attack mercury. Where it is desirable to eliminate the error as far as possible a stopcock may be placed in the line be- tween the two bulbs of the pipette which can be closed after the gas has been passed into the pipette while absorption is taking place. A separation of the constituent Unsaturated hydrocarbons is rarely carried out. It is best accomplished by bubbling a known volume of the gas through bromine and fractionating the resulting bromides. The results are at best unsatisfactory. Ernshaw 1 has shown that it is possible to calculate the average composition of the illuminants from data obtained by exploding a sample of the gas which still contains the olefines and deducting from the observed contraction and carbon dioxide the amounts due to hydrogen, carbon monoxide and the paraffines. The method de- mands very accurate work. Acetylene may be absorbed in a faintly ammoniacal solution of silver or copper salts. The precipitated acetylides are explosive and care should be taken in handling them. The ammonia vapors are to be removed from the gas by shaking with dilute acid before the volume is measured. Water dissolves more than its own volume of acetylene, so great care must be exercised to saturate the water of containing vessels before the gas to be analyzed is brought into them. Also gas from which large per- centages of acetylene have been removed must not be returned to vessels containing much water with which it had previously been in equilibrium since the water will give back to the gas material amounts of acetylene. Phosphorus forms a delicate reagent for qualitative detection of traces of Unsaturated hydrocarbons. If when the gas in question is brought in contact with phosphorus under the con- ditions prescribed for the estimation of oxygen, white fumes form, it is certain evidence of the absence of more than minute traces of Unsaturated hydrocarbons. If on the other hand the 1 Jour. Franklin Inst., 146, 161 (1898). 86 GAS AND FUEL ANALYSIS fumes fail to appear even after the addition of air it is not abso- lutely certain that the poison is an unsaturated hydrocarbon for ether, chloroform and a number of other substances behave similarly. 11. Oxygen. The estimation of oxygen by phosphorus as in technical analysis admits of little improvement in simplicity or accuracy. Alkaline pyrogallate may be used where it is not possible to remove the poisons which prevent the use of phos- phorus, and the ammoniacal copper reagent may be used, subject to the limitations given in Chapter III. 12. Carbon Monoxide. It was stated in Chapter III that the methods for estimation of carbon monoxide were unsatisfactory. They are more unsatisfactory for exact than for technical anal- ysis. The absorption by cuprous chloride given in technical methods is hardly to be considered as an accurate method. The change in the absorption spectrum of blood after treatment with carbon monoxide may be made an accurate qualitative test for carbon monoxide but does not lend itself readily to quantitative purposes. The method usually employed is to oxidize the carbon monox- ide, after having made certain that all other compounds which would be affected by the oxidizing agent have been removed. Iodine pentoxide is the most commonly employed oxidizing agent, the reaction being: The details here given are essentially those of Kinnicutt and Sanford, 1 who used substantially the method of Nicloux. The gas is first purified by being bubbled slowly through concen- trated sulphuric acid and then passed through a tube contain- ing lumps of caustic soda. This treatment removes unsatu- rated hydrocarbons, hydrogen sulphide, sulphur dioxide and similar reducing gases. The purified gas is then passed through iodine pentoxide contained in a U tube immersed in an oil bath at 150 C. Following the U tube comes another absorption tube containing about 0.5 g. potassium iodide dissolved in 10 c.c. of water. If a liter of gas is used as small an amount as 0.025 c.c. of carbon monoxide may be detected. The method is ordi- l Jour. Am. Chem. Soc., 22, 15 (1900). EXACT GAS ANALYSIS 87 narily only used for the detection of minute amounts of carbon monoxide and it is not suitable for large amounts. The absence of any liberated iodine may be considered a positive proof of the absence of carbon monoxide. The liberation of iodine is how- ever only a proof that some reducing substance was present, which can only with certainty be claimed as carbon monoxide after careful blank tests have shown that the purifying train is adequate to remove all other reducing substances and that the iodine pentoxide does not yield iodine except in the presence of such a reducing substance. Gill and Bartlett 1 tested this method on illuminating gas and report results about 9 per cent. high. They report accurate results when mixtures of carbon monoxide and air are used. Morgan and McWhorter 2 traced some of the difficulties of this method to the reduction of the iodine pentoxide by the hydro- carbon vapors given off by the lubricant and in the stopcocks of the U tube containing it. They recommend that the tips of the U tube be sealed with a flame. They further recommend the estimation of the carbon dioxide formed in the reaction as a more accurate index of the carbon monoxide than is the iodine. They insert an absorption tube containing standard barium hy- droxide in the train after the tube containing potassium iodide and after the close of the test titrate the residual barium hydrox- ide with oxalic acid. 13. Hydrogen. Hydrogen may be directly absorbed in metallic palladium but is almost always oxidized. The gas should first be freed from unsaturated hydrocarbons and other reducing gases as well as oxygen, so that it contains only hydro- gen, saturated hydrocarbons and nitrogen. There is then a choice of methods one class oxidizing the hydrogen without affecting the hydrocarbons, and the other simultaneously oxidizing both hydrogen and hydrocarbons. There are several methods for fractional oxidation of hydrogen which are reliable. .The errors which attend the method of simultaneous oxidation of hydrogen and methane have been briefly discussed in Chapter IV. The important systematic error in explosions and flame combustions is due to the oxidation 1 J. Am. Chem. Soc., 29, 1589 (1907). 2 Jour. Ind. and Eng. Chem., 2, 9 (1910). 88 GAS AND FUEL ANALYSIS of nitrogen. The formation of oxides of nitrogen increases with higher temperatures. The most favorable mixture for their formation is one in which there are equal volumes of oxygen and nitrogen. The errors are therefore minimised by keeping the temperature of combustion low and by making the diluting gas as nearly pure oxygen as possible. It is not possible to give an absolute statement of the magnitude of the error introduced for it will vary with each difference in the form of the explosion vessel and in the violence of the explosion. The following series of measurements by the author 1 will indicate the magnitude of the error and also the accuracy of the burette for exact gas analysis described in this chapter. EXPLOSION OF PURE HYDROGEN Hydrogen, c.c. Air, c.c. Contraction after explo- sion, c.c. Calculated per cent, hydrogen Contraction over potas- sium hydrox- ide, c.c. Explo- sive ratio 11.35 84.80 16.90 99.26 0.00 4.64 12.11 85.57 18.08 99.53 0.00 4.37 14.19 84.27 21.27 99.92 0.00 3.62 16.77 85.77 25.13 99.90 0.01 3.06 16.54 82.64 24.76 99.80 0.01 3.00 18.19 83.22 27.29 100.01 0.01 2.71 21.10 83.28 31.74 100.28 0.00 2.29 27.04 83.86 40.80 100.59 0.11 1.73 The hydrogen was prepared by the action of caustic potash on aluminium so as to be free from hydrocarbons. It will be noted that in this series air was used to supply the oxygen and as the diluent. A variation of 1.3 per cent, in the apparent purity of hydrogen due solely to errors inherent in the explosion process makes it evident that the method can hardly be called an accurate one. The errors attendant upon the flame combustion method of Dennis and Hopkins described in 9 of Chapter IV are illustrated in the following series. COMPARISON OP EXPLOSION AND COMBUSTION METHODS ON HYDROGEN Explosions with Air Sample hydrogen c.c. Air c.c. Contraction after explo- sion, c.c. Contraction over potassium hydroxide, c.c. Hydrogen per cent. Explo- sive ratio 15.32 18.15 85.34 82.39 22.71 26.93 0.04 0.06 98.82 98.91 3.43 2.73 A *1 f* f 1 C\(\ 1 \ EXACT GAS ANALYSIS Explosions with Oxygen 89 Sample hydrogen c.c. Air c.c. Contraction after explo- sion, c.c. Contraction over potassium hydroxide, c.c. Hydrogen per cent. Explo- sive ratio Oxygen I 14.82 93.51 22.04 0.02 99.14 3.91 16.48 82.18 24.51 0.02 99.15 3.02 20.58 80.09 30.60 0.03 99.12 2.29 Combustions by the Dennis and Hopkins Method Hydrogen c.c. Oxygen c.c. Air c.c. Contraction after com- bustion, c.c. Contraction over potassium hydroxide, c.c. Hydro- gen per cent. Oxygen in excess 91.29 58.48 89.31 51.65 53.39 40.77 54.55 50.14 50.21 136.72 87.43 133.57 0.04 0.10 0.07 99.84 99.66 99.70 13.72 40.89 3.73 The values obi ained by explosion with oxygen are remarkably concordant and are probably as accurate as we can hope to attain. The values obtained by explosion with air are higher and more irregular. The values of the Dennis and Hopkins method are also high and involve an error of about 0.6 per cent. 14. Methane. There are no absorption methods for methane which are acceptable. It is estimated by oxidation to CO 2 and H 2 O with measurement of the change in volume after oxidation and after absorption of the CO2. The general methods are given in 7 to 11 of Chapter IV. There is greater liability of error in the explosion process through formation of oxides of nitrogen than is the case with hydrogen, as is illustrated by the following series of tests of methane made from methyl iodide and the zinc copper couple. EXPLOSION OF METHANE Sample methane c.c. Air c.c. Contrac- tion after explosion Carbon dioxide c.c. Methane per cent. Hydro- gen per cent. Explo- sive ratio Ratio contrac- tion sample 7.05 8.93 9.07 10.20 92.07 104.17 98.35 98.22 13.09 16.66 17.10 19.27 6.53 8.31 8.54 9.63 92.62 93.28 94.15 94.41 0.28 0.14 0.14 0.06 4.05 3.53 3.19 2.61 1.85 1.86 1.88 1.89 90 GAS AND FUEL ANALYSIS In these experiments the explosive ratios all lie within the limits set by Bunsen, but there is a variation of 1.6 per cent, in the apparent percentage of methane as calculated by the usual methods and there is a corresponding variation in the amount of hydrogen. Similar experiments made with commercial oxygen (96.5 per cent, pure) as the diluting agent instead of air showed errors rather greater than when using air in similar amount. The greater error in analyzing methane as compared with hydro- gen results almost certainly from the higher temperatures at- tained in the explosion of methane. It is not possible to dilute the gas sufficiently to avoid this danger without running the risk of incomplete combustion of the methane. The methods of oxidation of methane which involve flame as in explosion or the Dennis and Hopkins method are always liable to material error due to formation of oxides of nitrogen. The errois may be mini- mized by keeping the gases at a low temperature while reacting. Jaeger's method of analysis by combustion wilh copper oxide in a combustion tube as described in 11 of Chapter IV cannot involve the formation of material amounts of oxides of nitrogen. This method is not so rapid or convenient as the others but, if care is taken to carefully cool the gas, including that remaining in the combustion tube, to its initial temperature before noting the change in volume, it is believed to be the most accurate method. 15. Errors in Calculation of Results of Explosion and Com- bustion. Some gases, notably carbon dioxide and the complex hydrocarbons, do not conform completely to the gas laws. The deviation is slight at low partial pressure and is usually negligible, but should be discussed. Burrill and Seibert 1 cite the following instances. If the volume relations in the explosion of methane and oxygen are accurately expressed they will be written as follows : 0.999 CH 4 + 2.000 2 = 0.995 C0 2 +(2H 2 which condenses) According to the above equation if perfectly pure methane were exploded with pure oxygen the composition as calculated from the contraction would, if calculated by the usual rules, be 100.2 per cent., while if calculated from the carbon dioxide its percent- age would be 99.4 per cent. The following table shows the 1 Bui. 42, Bureau of Mines. EXACT GAS ANALYSIS 91 molecular volumes for carbon dioxide at 20 C. and different partial pressures: Mm. of mercury Molecular vol. 100 0.9993 200 0.9986 300 0.9980 400 0.9972 500 0.9965 600 0.9958 700 0.9951 760 . 9950 If methane should be exploded with the theoretical volume of air instead of pure oxygen the error would decrease so that the apparent purity as calculated from the contraction would be 100.14 per cent, and as calculated from the carbon dioxide would be 99.86. These errors are in general within the limits of the other errors arising in analysis by explosion or combustion. Where ethane is present the error is greater 0.990 C 2 H 6 +3.500 O 2 = 1.988 C0 2 +(3H 2 which condenses) The molecular volumes of ethane corresponding to different partial pressures are shown in the following table: Mm. of mercury Molecular vol. 1.000 100 0.999 200 0.997 300 0.996 400 0.995 500 0.993 600 0.992 700 0.991 760 0.990 Burrell and Seibert cite an analysis of Pittsburgh natural gas which, calculated from the theoretical equations, showed 15.1 per cent. C 2 H 6 and 84.1 per cent. CH 4 , while by calculation from the corrected equations the results became 15.7 per cent. C2H 6 and 83.1 per cent. CH 4 . 16. Nitrogen. The method of removal of all gases except nitrogen by combustion with copper oxide as given in 12 of Chapter IV is fairly exact. The gases of the argon group are left with the nitrogen but the separation of these is not included in this work. CHAPTER VII HEATING VALUE OF GAS 1. Introduction. The heating value of gases may be deter- mined by combustion in a bomb calorimeter but the difficulty of transferring a definite quantity of gas to the bomb prevents the general adoption of such a method. The standard type of calorimeter is one in which the gas is burned with atmospheric air and the total heat of the products of combustion is transferred as completely as possible to the water of the calorimeter. A continuous flow calorimeter is usually employed wherever a suf- ficient amount of gas is available, but the intermittent type is also used. In some other types of calorimeter the heat of com- bustion is used to raise the temperature of a piece of metal, or a thermocouple placed in the flame. Calorimeters of these latter types give merely relative figures which can only be converted into heat values after very careful calibration. They are in no sense standard instruments and are not discussed in this book. It is also possible to calculate the heating value of the gas from its chemical composition with approximate accuracy. 2. Continuous Flow Calorimeters. The continuous flow calorimeter bears much resemblance to the type of instantaneous water heaters found in bath rooms. The gas, after passing through a meter, is burned in a Bunsen burner. The products of combustion give up their heat to a stream of cold water flowing in a direction opposite to that of the gases so that the gases emerge from the instrument cold and the water emerges hot. The process to be accurate requires the fulfillment of the follow- ing conditions: a. The gases must be accurately measured, be burned at a constant rate, and combustion must be complete. b. The water must enter at a constant rate and at constant temperature, and be at approximately room temperature. c. The heat of combustion must all be transmitted to the 92 HEATING VALUE OF GAS 93 water whose mass and rise in temperature must be accurately determined. The pioneer instrument of this type was that devised by F. W. Hartley in 1884. l The standard by which gas was judged was, at that time, candle power and therefore the instrument did not attract much attention. The first calorimeter to be used widely was that designed in 1893 by Hugo Junkers. 2 3. Wet Gas Meters. The gas is usually measured in a wet meter which consists of a horizontal cylindrical casing of sheet metal enclosing a horizontal drum and filled about half full of water. The drum is divided with slant partitions into three or four compartments, each with its own entrance for gas at the back and its exit at the front. Each of these compartments con- stitutes a distorted screw thread enlarged to a chamber in the center. As the inlet of one of these compartments comes out of the water the pressure of the gas entering causes the drum to revolve and the compartment fills with gas. When the compart- ment is full the inlet dips below the water and seals, the outlet unseals and the water entering the compartment drives out the gas before it. If there were only one or two compartments the flow of gas could not be continuous, but with three or four com- partments in the drum, one is always discharging and the rate of flow of gas is about constant. It will readily be seen that the capacity of each compartment and the time of sealing and unsealing the inlet and outlet will be affected by the height of water in the meter so that the exact setting of the water level becomes of great importance. Each meter is provided with a gage glass and a pointer which can be adjusted to indicate the proper water level. The Committee on Calorimetry of the American Gas Institute in its 1912 report states that even with careful work an error of 0.3 per cent, may be introduced by inaccurate adjustment of the water level and that on passing 100 cubic feet of gas through the meter an error of about equal magnitude may be introduced through evaporation of the water. The meter must be calibrated frequently for accu- rate work. The gas meter must be calibrated by passing a known volume 1 London Jour. Gas Lighting, 43, 1142 (1884). 2 Jour, fur Gasbel, 36, 81 (1898). 94 GAS AND FUEL ANALYSIS of gas through it. This known volume may most accurately be obtained by the use of what is known as a cubic foot bottle or a smaller container of the same type. Containers of this type should be rigidly made of glass or metal and tapered at the top and bottom to glass tubes of small diameter upon which zero marks are etched. The capacity of one of these containers is determined by the weight of water which it contains. They may be bought in elaborate forms and with the certificate of the Bureau of Standards. Where a standardized bottle is not avail- able a sufficiently accurate substitute may be improvised from a gas holder of the type shown in Fig. 23. This consists of a cy- linder of galvanized iron coned and terminated with a cock at both ends. If this is filled with water at room temperature and weighed, and then drained and weighed, the volume of the tank may be calculated from the following table. The volume thus obtained may be considered constant within ordinary ranges of temperature since the coefficient of expansion of mild steel is only 0.0000067 per degree Fahrenheit. The tank shown in the illustration holds one-third of a cubic foot and is weighed upon the balance shown in front of it. If larger tanks are used they must be made of heavy material so as not to change in volume when filled with water. WEIGHT OF ONE CUBIC FOOT OF WATER 50 F , 62.41 Ib. 60 F 62.371b. 70 F 62.31 Ib. 80 F 62.231b. 90 F 62,131b. In making the test the tank is to be filled absolutely full of water of almost room temperature and left standing by the side of the meter in a room of fairly constant temperature for several hours to make sure that all the parts of the system are at the same temperature. The gas supply is to be bubbled through water in order to be certain that it is saturated with water vapor. If the depth of water in the saturator is so adjusted that no gas bubbles through unless a very slight suction is applied, it will obviate mathematical corrections, because if the static pressure HEATING VALUE OF GAS 95 of the water in the saturator is adjusted to counterbalance the pressure of the incoming gas, and the outlet of the discharge from the tank is set at the level of the lower cock, a direct com- parison of the volume of gas registered by the meter and the capacity of the tank gives the calibration factor for the meter without any calculation for difference in pressure. So long as the whole system is at the same temperature it is immaterial what the temperature is. The upper cock of the tank may now be con- nected directly to the outlet of the meter. The lower cock of the tank should have a rubber tube attached which drops down to allow the formation of a water seal and rises again to discharge the water at the level of the cock. If now the lower cock on the tank is opened water will flow out through the rubber, tube and an equal volume of gas will flow through the meter. When the water is all out of the tank the gas will be automatically stopped by the water seal in the rubber outlet tube and the whole system will be under atmospheric pressure, which is a condition neces- sary for accurate work. 4. Corrections for Temperature and Pressure. The volume of the gas used in a test must be corrected for temperature and pressure. The standard conditions for the United States and Great Britian are a temperature of 60 F. and a barometric pressure of 30 in. of mercury. The gas as it comes from the wet meter is assumed to be saturated with water and correction is made for the excess of water vapor over that normal for 60. Tables giving the correction factors in convenient form have been issued by the Gas Referees of London. They are derived accord- ing to the formula. = 17.64(h-a) 460+t where n = factor sought h = height of barometric column in inches mercury a = vapor tension in inches of mercury at temp, t t = observed temp, of meter in degrees F. Although this formula apparently involves the full correction for vapor tension which would reduce the volume of gas to a dry basis at 60 F., there is a further correction contained in the figure 96 GAS AND FUEL ANALYSIS 17.64 of the formula which corrects the gas to a basis of 60 F. when saturated with water. The full formula is as follows; W 30 \ / \30-.517 The first member of the formula corrects the volume to 32 F., the second to 60 F., the third to dry gas under 30 in. barometric pressure, and the fourth to gas saturated with moisture at 60 F. the 0.51 being the vapor pressure of water at 60 F. Table III in the Appendix gives the factors for each degree Fahrenheit and each 0.1 in. pressure within wide limits. 5. Control of the Water. The water supply must always be sufficient to keep the overflow on the small elevated constant- level tank in operation. If the supply of water to the tank slackens much the rate of flow to the calorimeter will also change. The temperature of the water should also be constant and approximately that of the room. In .order to insure a water supply of even temperature and pressure it is advisable to install in the calorimeter room an elevated tank of at least 20 gallons capacity connected to the water mains and provided with an over- flow from which the supply for the calorimeter is drawn. 6. Measurement of Temperature. The temperature of the inflowing and outgoing water is determined by thermometers which should read to tenths of a degree. They must be carefully calibrated throughout their length by comparison with a stand- ard. It is immaterial whether they are Centigrade or Fahrenheit thermometers but the Fahrenheit are more convenient, when as is customary, the results >are to be expressed in British thermal units per cubic foot of gas. The thermometers are to be carefully placed in the calorimeter so that their bulbs are completely sur- rounded by water and are nowhere in contact with the metal of the calorimeter. 7. Measurement of Mass of Water. The mass of water heated during an experiment may be determined either by meas- urement or by weight. A 2000-c.c. graduated cylinder is used sometimes. This is not an ideal device, for not only are the graduations coarse but the varying temperatures at which the HEATING VALUE OF GAS 97 water is measured necessitate the use of different correction factors to reduce the volumes to the equivalent weights. It is much more satisfactory to weigh the water. If the thermometers are in the Fahrenheit scale it is more convenient to have the balance weigh in pounds and hundredths of a pound. If the ther- mometers read in the Centigrade scale it is more convenient to preserve the metric unit throughout. The water should be weighed in a metal bucket or glass vessel for which a counter- poise is provided. It is convenient to counterpoise the bucket with the interior wet as it is just after the water has been poured out after a test. Its weight will be so nearly constant that it is not necessary to counterpoise after each test. The balance should be capable of carrying ten pounds on each pan with a sensitiveness of one-hundredth of a pound. 8. Description of Calorimeter. The illustrations of Fig. 22 show the Junkers calorimeter which has served as the model from which most of the American instruments have been devel- oped. Some of the American instruments embody distinct im- provements such as ease of disassembly for repairs or cleaning, and automatic tipping device for diverting the stream of heated water from the measuring bucket when the meter hand has made a complete revolution. The gas coming from the meter is burned in a Bunsen burner placed in the cylindrical combustion chamber of the instrument. The hot gases rise in the central chamber, pass down through a series of small tubes in the annular water jacket surrounding the combustion chamber, unite again in a single flue at the bottom of the instrument and pass out through an orifice whose size may be regulated by a damper. Water enters through the rubber tube w to the small elevated tank a which serves to supply water to the calorimeter under a constant head. The amount of water flowing into the^nstrument is con- trolled by the valve b and the waste water from a is discharged at c. The water traversing the instrument, passes out at d and during the actual test is measured in the graduated cylinder shown, or collected in a vessel and subsequently weighed. The operation of the instrument is shown in detail in Fig. 22. The sectional illustration shows the burner (2) properly placed in the combustion chamber (1) and the path of the combustion gases and the water. The products of combustion rise in the 98 GAS AND FUEL ANALYSIS central chamber, turn at the top, descend the annular cooling chamber (3) to the drum (4), pass the thermometer (5) and es- cape at (6). The water rises to the small overflow cup (7) tra- versing a filter of wire gauze. The excess of water overflows and passes out of the instrument to the waste pipe through a rubber FIG. 22a. Gas calorimeter. FIG. 22b. Section of gas calorimeter. tube attached at (8). The small vessel (7) must always be kept overflowing to insure a constant pressure and hence a constant supply of water through the valve (9). The water passes the thermometer (10) which registers the inlet temperature, descends in a small pipe within the outer casing to the bottom of the instru- HEATING VALUE OF GAS 99 ment and rises in the annular chamber surrounding the gas pas- sages (3), thus passing in a direction opposite to that of the com- bustion gases. The water from the annular chamber passes through the drum (11) provided with baffle plates to mix it and to insure that all parts of the stream are of uniform temperature, passes the thermometer (12) which registers the outlet tempera- ture, to the overflow vessel (13) and out at (14). The water formed in the combustion of the hydrogen and hydrocarbons of the gas is condensed as it descends the annular condenser (3) and passes out of the drip shown at e of Fig. 22a and is caught in a graduated cylinder. The working parts of the calorimeter thus described are insulated from the room by an air jacket. The outer casing of the calorimeter is nickel-plated and to be kept polished to lessen the loss of heat by radiation. It is well to drain the calorimeter when not in use by opening the cock (16). There is some danger that air brought in by the in- coming water may collect around the inlet thermometer and disturb the accuracy of its readings. The tube (h) shown at the top of the calorimeter in Fig. 22a is to allow such bubbles to escape. Any water entrained by the bubbles passes out of the small over- flow. 9. Preliminaries of a Test. The calorimeter is to be set up on a table where the light is good, where there are no draughts, and where there is a water supply and a waste pipe. Fig. 23 shows the calorimeter set in a hood which has been modified by lowering a portion of the floor to bring the thermometer more nearly on a level with the eye of the observer. On the top of the hood is a zinc-lined wooden tank connected to the water supply and provided with an overflow pipe, from which the water supply for the calorimeter is drawn. This large overflow tank is necessary to compensate for variations in the tempera- ture and pressure of the city water supply. It should preferably hold enough water for the day's tests, so that the water standing over night may come to room temperature. The first step is to turn on the water and regulate the supply so that it flows through the instrument at the rate of approxi- mately 1.5 to 2.0 liters per minute in case illuminating gas of ordinary quality is to be tested. The water must continuously overflow from both cups (13) and (7) of Fig. 226. No water 100 GAS AND FUEL ANALYSIS should drip from spout e nor from any other part of the instru- ment. The gas meter is to be levelled and water added if necessary to bring the bottom of the meniscus of the water on a level with the pointer. This is to be done when there is no gas pressure on the meter, and to ensure this state of affairs the burner cock on the outlet of the meter should be open and an opening should be made to the air on the inlet side of the meter. This may usually be most conveniently effected by disconnecting the rub- FIG. 23. Gas calorimeter and accessories. ber tube, but may also be accomplished by unscrewing the small plug at the bottom of the well below the gas inlet. This also serves the purpose of removing any water which may have spattered into the gas inlet. The probable error in adjusting the water level is 0.3 per cent, so that when a greater accuracy is required the meter should be recalibrated after adjustment. The gas supply is to be connected to the meter and then to the calorimeter. A pressure regulator is unnecessary on a city gas supply if the meter works smoothly. Directions are frequently HEATING VALUE Off G'AS ' . : 101 given to interpose a regulator of the floating ;bfcli -jar type .be- tween the meter and the calorimeter. If this is done it must be watched closely because any variation in the heights of the drum of the regulator at the beginning and end of the test causes an error in the measurement of the gas. Long connections of rubber tubing are to be avoided since rubber is porous and also exerts a selective solvent action on the hydrocarbons of the gas. The connections should be of glass with short lengths of rubber on each end. The system is to be tested for leaks by closing the burner cock and opening the inlet cock on the gas main. The large hand of the meter should not show any perceptible change in position after two minutes. When the system has been found to be tight the burner cock may be opened and the gas allowed to escape into the air of the room until the large hand of the meter has made one complete revolution. It may then be lighted without danger of explosion unless the gas is acetylene, uncarburetted water gas or similar gases having wide explosive limits. Be sure that no unburned gas gets into the calorimeter where it might cause ex- plosion later. The flame from the burner is to be regulated so that it is a clear blue color with just a tinge of yellow at the tip. The volume will depend on the quality of the gas. Special tips for gases of very high or low heating value are provided with the instrument. With ordinary illuminating gas the rate of flow should be between 5 and 7 ft. per hour. The gas should be burned with a small excess of air. The draft may be controlled by the butterfly valve (6) of Fig. 22. The valve should be set to give maximum readings of the thermometer on the water outlet. That setting which gives the highest result is the most accurate. The Burtau of Standards 1 recommends as a " nor- mal rate," the rate of 70 per cent, of the maximum attainable without incomplete combustion. The flame must burn per- fectly steadily. If it flickers the most probable cause is the presence of water in some of the connections. The rubber tub- ing should be disconnected and drained, the plug in the well below the inlet of the meter removed, and the pressure regulator examined. It may be that the flickering is due to friction within 1 Technologic Paper No. 36. 102 QA3 AND FUEL ANALYSIS the meter which causes the drum to stick. This can usually be observed by closely watching the large hand of the meter and observing if it lags at particular points of its revolution. Trouble from this source cannot usually be remedied without sending the meter back to the factory. Flickering of the flame of the test burner may also be due to fluctuations in the pressure of the gas in the mains or services. A U-gage attached to a gas outlet will show whether this is the case. If it is necessary to test gas under these conditions the flickering may be lessened by interposing between the gas outlet and the meter a large empty bottle. Care must be taken to fill this bottle completely with water and then displace this with gas to avoid danger of explosion. If the meter has been recently filled with fresh water the gas should be allowed to burn for an hour before the test in order that the water of the meter may become saturated and not cause any change in the composition of the gas passing through it. 10. Description of a Test. When the gas and water supply and the burner have been thus adjusted, the lighted burner may be placed in the calorimeter, care being taken to properly center it and to see that it projects far enough into the instrument so that the top of the burner will be above the lower level of the belt of cooling water. This operation is facilitated by a small mirror placed below the instrument. The thermometer in the outlet water will at once commence to rise, and in a few minutes will be constant to within a few tenths of a degree. The increase in temperature of the outflowing over the incoming water should not be more than 20 F. and it may be necessary to readjust the water and gas supply to obtain this condition. If it is necessary to change the adjustment of the gas the burner should be removed from the calorimeter. The butterfly valve must not be closed enough to prevent the flame from burning strongly and freely. The escaping gases should not have any odor of unburned gas. Their temperature should be only a few degrees above that of the water entering. The water formed in the combustion of the hydrogen and hydrocarbons will condense in the calorimeter and commence to drip from the small spout at the bottom of the instrument. The test should not be commenced until the thermometer on the water outlet has remained constant for several minutes and the HEATING VALUE OF GAS 103 condensed water has also commenced to drip from its spout showing that equilibrium has been reached within the calori- meter. The counterpoised bucket for the water is placed in a conve- nient position and as the large hand of the meter passes a zero mark the water is turned into the bucket. Both thermometers are now to be read at each quarter position of the meter hands or more frequently. The outlet thermometer sometimes fluctuates rapidly and it is well under such circumstances to read it as many times as possible so as to obtain a fairer average. A consumption of 0.2 of a cubic foot, or two revolutions of the large meter hand is sufficient with illuminating gas. The last thermometer readings will be made when the meter hand is in the three-quarter position. The operator then watches the meter while holding one hand on the tube through which the outlet water is flowing and as the meter hand again reaches the zero he swings the tube out of the bucket, thus marking the completion of the test. After the water has been weighed a duplicate test may at once be started. The procedure is the same in case the water is to be measured in a graduated cylinder instead of being weighed. The volume readings of the cylinder must be converted into grams by means of the table in 6 of Chapter XVI on Manipulation of the Bomb Calorimeter. It is not possible to read the volume very accu- rately in a wide graduated cylinder. The water of condensation dripping from the instrument is to be measured to allow the calculation of the net heating value. It is accurate enough to catch the water formed from 1 cu. ft. in a 50 c.c. graduated cylinder. In case the temperature readings have not stayed constant within a few tenths of a degree during the test, the results should be discarded and an attempt be made to better the conditions. If for any reason the burner goes out during a test and un- burned gas gets into the instrument, it must be expelled before again lighting the flame. This may readily be accomplished by blowing vigorously through the escape pipe of the combustion gases. 11. Calculation of Observed Heating Value. The heating value of gas is usually expressed, in English speaking countries, in British thermal units per cubic foot of gas measured when 104 GAS AND FUEL ANALYSIS saturated with moisture at 60 F. and under a barometric pres- sure of 30 in. of mercury. The method for correcting the ob- served volume of gas to these standard conditions has been described in 4. The factors for converting cubic centimeters of water at various temperatures to grams are given in 6 of Chapter XVI. A Calory is defined accurately enough for this work as the amount of heat required to heat 1 kg. of water 1 C., and a British thermal unit to be the amount required to heat 1 Ib. of water 1 F. irrespective of the temperature of the water. The formula for calculation of the heating value is where m = mass of water heated t' = mean temperature of water flowing out t = mean temperature of water flowing in v = corrected volume of gas burned If m is expressed in pounds, t in degrees Fahrenheit and v in cubic feet, the result is at once British thermal units per cubic foot. If m is expressed in kilograms, t in degrees Centigrade and v in cubic feet, the result is in the hybrid unit Calories per cubic foot which may be corrected to British thermal units per cubic foot by multiplying by 3.968. If m is expressed in grams, t in degrees Centigrade and v in liters the result will be in the metric unit, small calories per liter which is the same as Calories per cubic meter. This is the method of reporting heat values in Germany and other countries which use the metric system. To convert Calories per meter to British thermal units per cubic foot of gas measured under the same condition multiply by .1124. In scientific work the heating value of a gas is usually reported as Calories per cubic meter of dry gas at C. and 760 millimeters pressure. In tech- nical work in Germany the gas volumes are usually corrected to 15 C. and 760 mm. pressure which makes the conditions prac- tically the same as those which prevail in this country. The following data are from a test made under rather unfa- vorable conditions and where no attempt was made to secure HEATING VALUE OF GAS 105 an accuracy closer than 1 per cent. The meter used passed 0.1 cu. ft. of gas per revolution, the thermometers were in the Fahrenheit scale and the water was weighed in pounds. Meter reading at start 21 . 200 Meter reading at close 21 . 400 Meter temperature 71 F. Barometer 29 . 8 Correction factor for temperature and pressure . 965 Correction factor for meter . 996 Temperature of water In Out 53.8 70.9 53.9 70.8 53.9 70.9 53.8 70.9 53.9 71.0 53.9 70.9 54.0 71.0 Average ..... 53 . 88 F. ^ 70. 91 F. Calibration correction for inlet thermometer ............. 0.1 Calibration correction for outlet thermometer ............ +0. 1 Corrected inlet temperature ..................... 53 . 78 F. Corrected outlet temperature .................... 71 . 01 F. Rise in temperature of water .................... 17. 23 F. Room temperature ............................. 72 F. Weight of water heated ......................... 6 . 76 pounds Gas burned, corrected 0.2 X 0.965 X 0.996 = 0.192 cu. ft. Uncorrected heating value = 606 B.t.u. Correction for difference in temperature be- tween inlet water and room temperature =- 0.7 B.t.u. per degree F. (see Table VIII of Appendix) Temperature inlet water ............ 54 F. - Room temperature ................. 72 F. Difference ........................ 18 F. Correction to be subtracted 0.7 X 18 = 12 B.t.u. Observed heating value 594 B.t.u. 106 GAS AND FUEL ANALYSIS 12. Total Heating Value. The total heating value of a gas has been denned by the Bureau of Standards 1 in the following terms. "The total heating value of a gas, expressed in the Eng- lish system of units, is the number of British thermal units pro- duced by the combustion, at constant pressure, of the amount of the gas which would occupy a volume of 1 cubic foot at a temperature of 60 F., if saturated with water vapor and under a pressure equivalent to that of 30 inches of mercury at 32 F. and under standard gravity, with air of the same temperature and pressure as the gas, when the products of combustion are cooled to the initial temperature of gas and air and the water formed by combustion is condensed to the liquid state." It is believed preferable that this term replace the term "gross heating value" which has been very loosely used. The total heating value differs from the observed heating value in the application of certain corrections for errors which while minor may cause an error of as much as 2 per cent, in the final result. These are discussed in following paragraphs. 13. Net Heating Value. In most industrial operations the combustion gases are not cooled to room temperature before escaping from the apparatus and therefore some of the heat of the gas is wasted. This is due to a lack of efficiency of the appara- tus and varies with individual conditions. There are so many industrial operations, however, where the water formed in com- bustion escapes as steam and the latent heat of steam formation is so large when compared with the amount of heat required to raise the temperature of fixed gases or of water or steam through moderate temperature intervals that the value of the latent heat is frequently deducted from the gross heating value. The re- sulting lower value is called the net heating value. The net heating value is defined by the Bureau of Standards as follows:- "The net heating value of a gas, expressed in the English system of units, is the number of British thermal units produced by the combustion, at constant pressure, of the amount of the gas which would occupy a volume of 1 cubic foot at a temperature of 60 F., if saturated with water vapor and under a pressure equivalent to that of 30 inches of mercury at 32 F. 1 Waidner and Mueller, Technologic Paper No. 36. Industrial Gas Calorimetry. HEATING VALUE OF GAS 107 and under standard gravity, with air of the same temperature and pressure as the gas, when the products of combustion are cooled to the initial temperature of gas and air and the water formed in combustion remains in the state of vapor. According to the above definitions the net heating value is less than the total heating value by an amount of heat equal to the latent heat of vaporization, at the initial temperature of the gas and air, of the water formed by the combustion of the gas." The heat absorbed when one gram of water at 100 C. changes to steam at the same temperature is 0.536 Calories. The heat absorbed in the change of liquid water at 10 C. to vapor at 10 C. is very closely 0.600 Calories. This latter figure is usu- ally employed in calculating the net heating value which is obtained by multiplying the number of cubic centimeters of condensed water dripping from the calorimeter during the com- bustion of a cubic foot of gas by 0.6, and 3.968 to convert into British thermal units and then subtracting this value from the observed heating value in British thermal units per cubic foot. The method of calculating the net heating value in the test re- ported in the preceding section is as follows: Meter reading at start 21.200 Meter reading at close 22.100 Gas burned 0.900 cu. ft. uncorrected Gas burned corrected, 0.9X0.965X0.996=0.86 cu. ft. Condensed water collected, 21.8 c.c. 21 8 Condensed water formed per cu. ft. g as ?T =25.4 c.c. Latent heat of condensed water 25.4X0.6X3.968=60 B.t.u. Uncorrected heating value of gas 606 B.t.u. Correction for difference in temperature between inlet water and room temperature =0.4 B.t.u. per degree F. (see Table VIII of Appendix). Corrections to be subtracted 0.4X18= 7 Latent heat of condensed water 60 67 Net heating value of gas 539 B.t.u. 14. Calculation of Total Heating Value. The calculation of Observed Heating Value given in Section 11 must be modified 108 GAS AND FUEL ANALYSIS by taking account of the following errors in order to calculate Total Heating Value. Temperature correction for barometer. Correction for pressure of gas at meter. Correction for emergent stem of thermometer. Correction for variation in temperature of inlet water. Correction for variation in temperature of escaping gases. \; Correction for humidity of air. - Correction for humidity of gas. An illustration of a complete record and calculation with cor- rection for all errors as taken from Circular 48 of the Bureau of Standards forms is given in Fig. 24. 15. Accuracy of Method. The accuracy of gas calorimeters has been thoroughly investigated by the Committee on Calori- metry 1 of the American Gas Institute and subsequently by the Bureau of Standards. 2 Both investigations agree that continu- ous flow calorimeters give substantially accurate results when properly constructed and operated. The sources of error are as follows: Errors in Registration of Gas Volume. If the meter is in good condition and is calibrated after the water level has been ad- justed the error of the meter need not exceed one-tenth of 1 per cent. If these precautions are neglected the error may be large. The error of observation will be about one small division of the large scale corresponding to 1 part in 200 or 0.5 per cent, error in a test as ordinarily run. Errors in Measurement of Temperature. If the thermometers are accurately calibrated, the water supply of constant tempera- ture and the consumption of gas and its heat value constant dur- ing a test, there should be practically no instrumental error. If the constancy of any of these conditions is upset there will be errors due partly to lag in the thermometers which do not in- stantly respond to the changed condition. An error which is negligible except in the most accurate work is that caused by the different temperature of the bulb and the stem of the thermom- eter. There will be no correction for the emergent stem of the l Proc. Am. Gas Inst., 1908, 285; 1909, 148; 1912, 65. 2 Technologic Paper No. 36. Industr.il 1 Gas Calorimetry, 1914. HEATING VALUE OF GAS 109 HEATING VALUE TEST RECORD Place /^LLf/TJJIJJs Jg&4 . Date._(fM,J?^A2v.. Time __/8 " Preliminary &77# 9t>S7-j3 S"o t>7-fO S2 Meter therm, reading Certif. corr'n (*~8.i |65oft lj.q m ^6 () J Bupplementary 6"7?J 36 t>7-?2 if,/ bJ'fO 2 Time of 1 meter rev. Equlv. rat(eu.ft.per hr.) SzL' Average \y~/'/*3 (>793 ?6.^--5 (aj.90 fftii-S 6>-9 CertifioaU corr'n ._ _ -.2.8 -2S ] CONDENSED WATEK MeUr reading: start > nd COLLECT :D Differential corr'n ._ . -02 \-IS /? ~/g -/? 18.1 Emergent stem corr'n i-08 ) /?./ 2 O.<4- Certificated temp. &77S 8^.2-7 6775 ff6.2.6 Condensate(oc) /.6 2-1.2. Temp, rise T /g.S~2. IS.5 1 ISSLf percu.ft,(00801n.) 12.-3 2-I.J Water heated W 6./- 6.7 4- (0.12. Average A _ _ _ 3-2..I No. of rev. of meter Meter eallb. 1 rev. = 2. NET HEATING VALUE O . look Observed heating value average _. O *f- U~ Gas volume V 0.2.0/2. . -h 1 Obnrved beating rake WX T f &l^ C^3 Keductlon te netfA X 2.9) JT / Net heating value . $? V- Corr'n for heat loss .._ / / Certified as correct. 3&. Corr'n for atmoe.bnmid. / ^ Total heating value. & 4- Appendix when gas of approximately 600 B.t.u. is being tested and according to Table VI when natural gas of about 1000 B.t.u. is being heated. Under these conditions the error will be about 1 B.t.u. Before using this table the humidity of the air must be determined as directed in the following Section. The corrections are expressed in B.t.u.'s and are to be added where the sign is + and subtracted where it is . Loss of Heat by Radiation. The calorimeter is protected against interchange of heat with the outside air by an insulating air chamber enclosed in a polished metal jacket. The efficiency of this protection is given in the report of the Committee on Calorimetry of the American Gas Institute for 1908 as 99.5 per cent. The metal jacket should be kept bright in order to maintain this efficiency. Accuracy of the Process as a Whole. None of the various preceding sources of error need amount to over 0.5 per cent. They will partially offset one another. When great care is taken the total error may not be over 1 per cent. Under ordinary con- ditions it is not safe to assume that the error will be less than 2 per cent. 16. Determination of Humidity of Air. The fol- lowing directions for the measurement of atmos- pheric moisture are given by the U. S. Weather Bureau. 1 The most reliable instrument for this purpose is the sling, or whirled psychrometer. In p IG 25. special cases rotary fans, or other means, may be Sling psy- employed to move the air rapidly over the thermo- chrometer. meter bulbs. In any case satisfactory results can- 1 U. S. Dept. Agriculture, W. B. No. 235, Psychrometric Tables. I al 114 GAS AND FUEL ANALYSIS not be obtained from observations in relatively stagnant air. A strong ventilation is absolutely necessary to accuracy. The sling psychrometer consists of a pair of thermometers, provided with a handle as shown in Fig. 25, which permits the thermometers to be whirled rapidly, the bulbs being thereby strongly affected by the temperature of and moisture in the air. The bulb of the lower of the two thermometers is covered with thin muslin, which is wet at the time an observation is made. It is important that the muslin covering for the wet bulb be kept in good condition. The evaporation of the water from the muslin always leaves in its meshes a small quantity of solid material, which sooner or later somewhat stiffens the muslin so that it does not readily take up water. This will be the case if the muslin does not readily become wet after being dipped in water. On this account it is desirable to use as pure water as possible, and also to renew the muslin from time to time New muslin should always be washed to remove sizing, etc., before being used. A small rectangular piece wide enough to go about one and one-third times around the bulb, and long enough to cover the bulb and that part of the stem below the metal back, is cut out, thoroughly wetted in clean water, and neatly fitted around the thermometer. It is tied first around the bulb at the top, using a moderately strong thread. A loop of thread to form a knot is next placed around the bottom of the bulb, just where it begins to round off. As this knot is drawn tighter and tighter the thread slips off the rounded end of the bulb and neatly stretches the muslin covering with it, at the same time securing the latter at the bottom. To make an observation, the so-called wet bulb is thoroughly saturated with water by dipping it into a small cup or wide- mouthed bottle. The thermometers are then whirled rapidly for fifteen or twenty seconds; stopped and quickly read, the wet bulb first. This reading is kept in mind, the psychrometer im- mediately whirled again and a second reading taken. This is re- peated three or four times, or more, if necessary, until at least two successive readings of the wet bulb are found to agree very closely, thereby showing that it has reached its lowest tempera- ture. A minute or more is generally required to secure the cor- rect temperature. In whirling and stopping the psychrometer HEATING VALUE OF GAS 115 the arm is held with the forearm about horizontal, and the hand well in front. A peculiar swing starts the thermometers whirling, and afterward the motion is kept up by only a slight but very regular action of the wrist, in harmony with the whirling ther- mometers. The rate should be a natural one, so as to be easily and regularly maintained. If too fast, or irregular, the ther- mometers may be jerked about in a violent and dangerous man- ner. The stopping of the psychrometer, even at the very highest rates, can be perfectly accomplished in a single revolution, when one has learned the knack. This is only acquired by practice, and consists of a quick swing of the forearm by which the hand also describes a circular path, and, as it were, follows after the thermometers in a peculiar manner that wholly overcomes their circular motion without the slightest shock or jerk. The ther- mometers may, without very great danger, be allowed simply to stop themselves; the final motion in such a case will generally be quite jerky, but, unless the instrument is allowed to fall on the arm, or strikes some object, no injury should result. The tables from which humidity may be calculated form Table IV of the Appendix and give the data for a barometric pressure of 29.0 inches of mercury. Their use is illustrated by the following example. Air temperature t =75.0 F. Wet bulb reading t' = 66.0 F. t-t' = 9.0 F. In table opposite 75 in column 9.0 is found 63. Relative humidity = 63 per cent. If the barometric pressure had not been 29 in. a slight error would have been introduced whose magnitude may be judged from the following examples of the same problem at different barometric pressures. Barometric pressure 30 Relative humidity 63 Barometric pressure 27 Relative humidity 63 Barometric pressure 25 Relative humidity 64 116 GAS AND FUEL ANALYSIS The Weather Bureau report 235 referred to above gives fuller tables and may be obtained from the Bureau for 10 cents. 17. Non-continuous Water Heating Calorimeters. In the third edition of his Gas Analysis, Hempel described a calorimeter where a volume of about a. liter of gas was measured in a glass cylinder, passed through a small burner and burned in a stream of oxygen within a calorimeter containing a known mass of water. The rise in temperature of the water gave the data for the calcu- lation of heat value, after the instrument had been calibrated by the combustion of hydrogen. The Graefe calorimeter is a commercial instrument of the same general type but somewhat larger. The instrument is rather crudely constructed and allows the exhaust gases to escape at an unduly high temperature so that it is necessary to calibrate it against some standard calorimeter. An inherent defect in calorimeters of this type comes from the increasing temperature of the exhaust gases as the test proceeds and the water of the calorimeter becomes warmer. The Parr 1 calorimeter aims to compensate for this error and errors due to moisture in the exhaust gases by providing two du- plicate calorimeters, one of which runs on pure hydrogen while the other is testing the unknown gas. The variation from the correct result shown by the hydrogen calorimeter is taken as the correction to be applied to the other result. The Committee on Calorimetry of the American Gas Institute in its 1912 report states that this calorimeter if properly operated gives correct results but that it is rather complicated in construction, and requires more skill for its proper operation than the other types. The instrument gives total heating values only. The Doherty calorimeter is a compact instrument which meas- ures the gas in an annular cylinder surrounding the combustion chamber and its water jacket. The gas is displaced by the warmed water which has flowed through this water jacket or heat absorption chamber, and thermometer readings are taken as the water level passes fixed points on the gage glass. No meter is required and the water is neither weighed nor measured. The Committee on Calorimetry of the American Gas Institute in 1 J. Ind. and Eng. Chem., 2, 337 (1910). HEATING VALUE OF GAS 117 its 1912 report states that this calorimeter when operated prop- erly gives the same efficiency as the Junkers calorimeter. 18. Automatic and Recording Gas Calorimeters. The form- ula for the calculation of the heating value of a gas as given in 11 of this chapter is E.V. -. It is evident that if the ratio can be kept a constant and t can also be kept con- stant that the heating value can be readily determined from a single reading of t' or can be continuously determined by a re- cording thermometer showing the temperatures of the outlet water. In the Junkers continuous calorimeter the ratio is kept constant by passing both gas and inlet water through meters whose drums are geared together by a chain forcing them to rotate always proportionately. There are various other types of automatic calorimeters. In every case they should be checked occasionally by a direct determination with a standard instru- ment. 19. Calculation of Heating Value from Chemical Composi- tion. If the heating value and the proportion of each constituent in a mixed gas is accurately known, it is possible to calculate the heating value of the mixture. Table IX gives the heating value as well as other properties of a number of gases. If each constituent in the gas were known, the heating value calculated from this data would probably give accurate results. However, when it is noted that among the defines, propylene has a heating value approximately 50 per cent, greater than ethy- lene, and butylene a heating value almost double that of ethylene it will be seen that there is dangerous latitude for arbitrary assumptions as to the constituents of the olefines. When the " illuminants " as reported include not only the olefines but ben- zene the error involved in an arbitrary assumption of the mean heating value becomes still greater. The varying members of the methane series also possess widely differing heating values. Earnshaw 1 gives an analytical method for determining the mean composition of the olefines and for differentiating between me- 1 Jour. Franklin Inst., 146, 161 (1898). 118 GAS AND FUEL ANALYSIS thane and ethane. The method is, however, difficult analytically, and the results when obtained are not entitled to the degree of confidence which pertains to those obtained directly in a calori- meter. The probable error involved in this method of calculating the heating value of gas (unless Earnshaw's complex analysis is followed out) is about 5 per cent. In the case of carburetted water gas it is still higher. In the case of producer gas where the total percentage of hydrocarbon is low and where all suspended tar particles have been removed the results are more accurate,. CHAPTER VIII CANDLE POWER OF ILLUMINATING GAS 1. Introduction. The use of candle power as a standard test of illuminating gas was practically universal before 1900. The most efficient of the early types of gas burner was the Argand and quite naturally it was used as the test burner. When carburetted water gas of high candle power came into use a bats-wing burner was found to be more efficient and was in some, cases allowed. The Welsbach mantle was developed later and its efficiency was found to be more nearly in proportion to the heating value of the gas than to its candle power. The proportion of gas burned in luminous flames is now so small in proportion to that burned for development of heat that tests of candle power have become of minor importance and are in a fair way to become obsolete as a criterion of quality of gas. Photometry deals with the measure- ment of the intensity of light. The term light as used here in- cludes only those rays which excite vision in the human eye, which thus necessarily becomes the final arbiter in photometric tests. The eye cannot estimate absolutely the amount of light which stimulates it. It can compare roughly the intensity of illumination from two sources and it can determine with more precision when the intensities from two sources are the same. In the sense in which it is here used, photometry consists in the comparison of two lights, one of which is .a standard. The photometer is a device which assists the eye in determining when the two lights are of the same intensity. The intensity of light entering the photometer is changed by varying the distance between the photometer and the light, the intensity of light from a given source varying inversely as the square of the distance. When the adjustments have been made so that the intensity of light impinging on the photometer from the two sources is the same, as shown by the equal illumination of the two photometer faces, the ratio of the unknown light to the standard light be- comes mathematically calculable from the relative distances of the lights from the point of equal illumination. The value of 119 120 GAS AND FUEL ANALYSIS luminous intensity is in English-speaking countries and in France expressed in candle-power. 2. Method of Rating Candle-power. The light emitted from a single incandescent particle would illuminate uniformly every point of an enveloping sphere and the intensity of illumination might be measured equally well at any point on the sphere. When the light to be measured comes from a surface of finite size as is always the case in practice, there is interference with the free path of the light waves from a single particle in one or more direc- tions so that the illumination of the enveloping sphere is no longer uniform. It is possible by the use of reflecting mirrors to deter- mine the illumination at various points on the circumference of a polar circle and to plot from this data a curve showing the dis- tribution of light at various angles. Methods of this sort are often resorted to in a study of illumination where it is desired to determine the value of a light source for a particular purpose. This method is, however, rarely followed where it is simply a question of testing the quality of the gas. The simpler custom of. taking the horizontal candle-power given by a conventionalized burner under conventional conditions as indicating the value of the gas, has become well established. 3. The Bar Photometer. The bar photometer consists of a graduated bar which carries at one end a standard light and at the other end the test light. Upon the bar slides a carriage with an apparatus for comparing the illumination from the two sources. The carriage is to be moved back and forth until the point is found where the illumination from the two lights is equal, and its position on the graduated bar recorded. If now the distance of the comparison box from the standard light be called "a" and that from the unknown light "b," then the illumination of the unknown as compared with the standard light is expressed by the proportion, unknown _b 2 standard a 2 There are many modifications of this type of photometer but all involve the four essentials: a standard light, the unknown light, a photometric screen and a means of measuring the distance of each light from the comparison box. It is customary to have the two lights fixed at opposite ends of the bar, in which case the CANDLE POWER OF ILLUMINATING GAS 121 sum of a+b in the preceding formula is a constant. Some- ' times however the standard lamp is placed on a sliding carriage connected by a rigid link with the photometric screen so that the distance "a" of the formula becomes a constant. A modification of the bar photometer in use in England is the table photometer where the two lights and the comparison box are all rigidly fast- ened at the points of a triangle. Comparison is effected by vary- ing the rate of combustion of the gas being tested until equality of illumination is reached. Its candle-power is then mathematic- ally determined. This method of determining candle-power is not in use in Germany or America. The various essential parts of a bar photometer will be con- sidered separately and the details of its operation will then be described. 4. Standard Light. The early photometrists used as their standards candles of varying size. In 1860, the English parlia- ment adopted as standard the sperm candle 7/8 in. in diameter and burning at the rate of 120 grains per hour. In 1884, Hefner v. Alteneck brought out the amylacetate lamp which became a widely used standard for testing the candle-power of gas. The Harcourt 10 candle pentane lamp was proposed in 1898 and has been adopted as the official source of light by the Gas Referees of London. It has many advantages. All of these flame standards vary materially in candle-power with change in atmospheric con- ditions and are, for scientific work, to be corrected to standard conditions of temperature, pressure, humidity and percentage of carbon dioxide in the air. In ordinary work when used in meas- uring candle-power of gas flames they are however not thus corrected but the assumption is made that the standard light and the gas light are equally affected by atmospheric conditions. The only satisfactory standard not affected by atmospheric conditions is the incandescent electric lamp. Incandescent lamps properly aged may be bought with the certificate of the Bureau of Standards and furnish the most reliable photometric stand- ards when used under proper conditions. These conditions, how- ever, require that the lamp shall be supplied with current at perfectly definite voltage from a large storage battery equipped with suitable rheostats and electrical measuring instruments so that the installation is an expensive one, and is used only in re- 122 GAS AND FUEL ANALYSIS search laboratories. When the incandescent electric standard is used in the photometry of gas flames correction must be made for the effect of atmospheric conditions on the flame. This correc- tion is not infrequently as much as ten per cent., and has been accurately determined for only a few of the standard lights. 6. Photometric Units. The international candle is the com- mon unit of intensity in England, France and America, having been officially adopted by agreement of the government standard- ising laboratories of the three countries in 1909. Prior to that date the official unit in this country had nominally been the British Parliamentary candle but there had not been definite agreement as to its value. In Germany the photometric unit is the Hefner which equals 0.90 International Candles. Conversely 1 International Candle equals 1.11 Hefners. The history of the adoption of the International Candle may be found in the reports of the Bureau of Standards and in the Proceedings of the Ameri- can Gas Institute. 1 Although there is thus an international unit of light, the international candle, it does not follow that this unit is best obtained by burning any actual candle. In fact various other standard 'lights are preferable. 6. Standard Candles. In 1860 the Gas Referees of the City of London adopted as their unit the light emitted by a sperm candle of 1/6 Ib. weight when burned at the rate of 120 grains per hour. From time to time they issued specifications for the manu- facture of candles to fulfill this requirement but were never suc- cessful in ensuring uniform quality and in 1897 entirely discon- tinued the use of candles. The Dutch Photometric Commission reported in 1894, after an exhaustive study, that the average light from a good English Parliamentary candle might exceed or fall below that of the average candle by nine per cent. The use of candles as standards is deservedly decreasing. When candles are to be used they are burned on a candle balance placed on the photometer bench. A simple form of balance is illustrated in Fig. 26. A long candle is cut in two and both halves used simultaneously. They are to be allowed to burn until the cups have formed normally and the wicks have bent over till the tips are glowing in the outer flame. The l Proc. Am. Gas. Inst. 2, 454, 528 (1907); 3, 403 (1908); 4, 78 (1909). CANDLE POWER OF ILLUMINATING GAS 123 candles are then to be turned so that the glowing end of one wick points towards the photometer and that of the other points in a direction at right angles to that of the first. They should project 1 to 1 1/2 in. above the holder and should burn clearly and without guttering. When all is in readiness for a test, the counterpoises are adjusted so that the candles are slightly too heavy for a perfect balance. As they burn away the pointer on the scale falls and as it passes the zero mark the stop watch is started. A 20-grain weight is then placed on the pan below the candles and photometric readings are made each half-minute. At the expiration of four and a half minutes the observer returns to the candle balance and stops the watch when the pointer again is at the zero mark, indicating that the 20 grains of sperm have been burned. If the candles are burning at exactly the proper rate the watch should show that exactly five minutes have elapsed. If the variation in the amount of sperm burned per hour is not over 5 per cent, from the standard amount it is permissible to make a mathe- matical correction, the as- sumption being that the light evolved is in direct proportion to the weight of candles burned . If the observed weight of sperm burned by the two candles is 250 grains per hour instead of 240 the value 250 of the light is said to be 2 X ^ = 2.08 candles. If the devia- tion is greater than 5 per cent, the test must be rejected and a different candle used. Improper ventilation and too high a temperature in the photometer room will affect the burning of the candles. This subject is discussed in 18. 7. The Hefner Lamp. The dimensions of the Hefner lamp have been rigidly specified by the German Reichsanstalt 1 which 1 Jour, fur GasbeL, 36, 341 (1893). FIG. 26. Candle balance. 124 GAS AND FUEL ANALYSIS will certify a lamp to be correct if it is properly made mechanic- ally and gives a light which does not differ more than 2 per cent. from the official lamp of the Reichsanstalt. The construction of the lamp is shown in Fig. 27. It consists of a brass bowl into which a head screws carrying the German silver wick tube and the mechanism for controlling the height of the flame. The flame height is determined by a gage which clamps to the head-piece. The older form engage shown at A consists of two sights, one on each side of the flame. The newer Krtiss optical gage shown at C consists of a ground glass screen and a magnify- FIG. 27. Hefner lamp. ing lens which allows more delicate adjustment of the flame tip to the horizontal line across the gage. Each lamp is provided with a control gage shown at B which fits over the wick tube and sits squarely on the head-piece. With the control gage in this position and the lamp level an observer looking toward the light should see through the openings D a very fine ray of light less than 0. 1 mm. wide between the top of the wick tube and the control gage, and looking through the optical gage should see the cross-hair in exact coincidence with the broad top of the control gage. The wick tube is screwed into the head-piece and if it becomes necessary to change its height the control gage, CANDLE POWER OF ILLUMINATING GAS 125 1 inverted, is to be pushed down the wick tube and used as a handle. The exact material of which the wick is made is not of importance but it must fill the tube snugly but not tightly. It is best to use only that furnished by the manufacturers. The amyl acetate must be of good quality and certified to be fit for photometrical purposes. In using the Hefner lamp the bowl is to be filled about two- thirds full of amyl acetate and after the wick has been moistened by capillary action it is to be screwed somewhat above the wick tube and cut squarely off. The lamp is then to be lighted and allowed to burn at least ten minutes with occasional regulation of the flame height before a test is commenced. The temperature of the photometer room should be between 60 and 70 F. The lamp is to sit on a horizontal support in a room free from drafts and adequately ventilated. The flame height of the Hefner lamp is to be carefully adjusted since a deviation of 1 mm. from the correct flame height of 40, introduces an error of about 3 per cent. It is the luminous tip of the flame which is to be 40 mm. high. With the Kriiss optical gage the frosted glass cuts out the almost colorless outer flame so that there is no possibility of confusion. With the older Hefner gage the luminous tip should appear tangent to the lower edge of the sighting plane. If the lamp is used only infrequently it should be emptied after use and both lamp and wick should be washed with alcohol. It is wise to throw away the old amyl acetate and clean the lamp in this manner at intervals even when it is in frequent use since the amyl acetate decomposes somewhat on standing. The Hefner lamp gives a light of. 0.9 international candles when burned in pure air under 760 mm. barometric pressure and containing 8.8 liters of water vapor per cubic meter. Although atmospheric conditions must be controlled and correction made in exact scientific work corrections are usually omitted in taking candle-power of gas on the assumption that atmospheric condi- tions affect the Hefner lamp and the gas burner to a similar de- gree. The errors involved in this assumption are discussed briefly in 18. The Hefner lamp is a widely used standard. It is portable, cheap, and relatively accurate. Its disadvantage is its 126 GAS AND FUEL ANALYSIS low candlepower, and the tendency of the flame to flicker, es- pecially at summer temperature. 8. The Pentane Lamp. The 10 candle pentane lamp or Harcourt lamp was adopted as stand- ard by the London Gas Referees in 1898. l Fig. 28 shows this lamp with some improvements in details recommended by the Bureau of Standards and added by the American manufacturers. In this lamp air entering at A passes over pentane and becomes saturated with pentane vapor. The air-gas so formed descends by gravity to an Argand burner B enclosed in a metal hood. The flame is drawn into a definite form and the top of it is hidden from view by a long brass chimney C. The chimney is surrounded by a larger brass tube D in which air, warmed by the chimney, rises and descends through the tube E, which is also the main standard of the lamp, to the cen- ter of the Argand burner where it aids in the combustion of the gas. The lamp may be obtained with the cer- tificate of the Bureau of Standards. Before using the lamp the satu- rator is to be filled about two-thirds full of pentane and both cocks on the saturator are to be closed. As pen- tane is very volatile and mixtures of its vapor and air within certain proportions are explosive care must be taken that no flames are burning in the room while the lamp is being filled. The inner chimney above the burner must be centered by the adjusting screws, turned so that the mica window is away from the photometer box and set at the proper height by plac- ing on the burner the 47 mm. block which accompanies the l Jour. of Gas Lighting, 71, 1252 (1898). FIG. 28. 10 candle power pentane lamp. CANDLE POWER OF ILLUMINATING GAS 127 lamp, and lowering the chimney till it rests lightly on the block. To prepare the lamp foT lighting, open the outlet cock on the saturator and the drip cock. This will fill the feed pipe with pen- tane vapor and air. Open the inlet cock on the saturator, close the drip cock, open the regulating cock at the burner and light the gas at once. It requires about fifteen minutes for the flame to become constant and during this period the top of the flame should be kept approximately on the cross bar of the mica window. The lamp should be set for maximum luminosity which condition is attained when the flame is just high enough so that the non- luminous upper portion is cut off from the photometric screen by the chimney. In case of doubt the proper setting may be deter- mined by lighting the gas flame at the other end of the bench and determining with the photometer the setting of the lamp which gives maximum illumination. In leaving the lamp after a test both the inlet and the outlet cocks of the saturator should be closed. After about a gallon of pentane has been burned the liquid remaining in the saturator should be emptied out and thrown away. 9. Secondary Standards of Light. The Hefner lamp, the 10 Candle Pentane lamp and, to a lesser degree, standard candles are primary standards since they are readily, if not entirely ac- curately, reproducible. There are various secondary standards which are convenient to use when frequent candle-power deter- minations are to be made but which must be standardized at intervals by direct comparison with a primary standard. Most of these standards are based on the fact that the brightest por- tion of^a lamp flame is of almost constant luminosity. The"fedgerton Standard burner consists of a Sugg "D" burner provided with a glass chimney 1 3/4 in. in diameter and 7 in. high. Outside of this glass chimney is a brass sleeve with a horizontal slot 13/32 of an inch high through which the light passes to the photometer. This is nominally a five candle-power standard but will actually vary from four to seven candles. After the value with a given gas has been fixed it will not vary much if the candle-power of the gas feeding it does not vary over two candles. The chimney must be cleaned frequently and the lamp restandardized each time a new chimney is put into service. The Elliot lamp is a student lamp of special design with a 128 GAS AND FUEL ANALYSIS flat wick and a rather large chimney and a screen which cuts off all but the desired portion of the flame. The lamp uses kerosene as its fuel and is nominally a ten candle-power lamp. Its illu- minating value with a single lot of good kerosene is of very satis- factory constancy. 10. Standard Gas Burners. At the time when gas testing commenced to be standardized the Argand burner was the form in common use. This type was therefore naturally adopted as the standard. It was also recognized that it was only right to FIG. 29. D Argand burner. FIG. 30. Metropolitan No. 2 Argand burner. test the gas in a burner which was adapted to it and therefore various standards came into vogue in England, such as the Sugg D Argand illustrated in Fig. 29, which is intended for gases of less than 16 candle-power. The Sugg F burner is intended for gases of 16-20 candle-power. In 1905, in connection with a readjustment of the price and candle-power of the gas supplied in London the Gas Referees were directed by Parliament to use a burner adapted to obtain from the gas the greatest amount of light when burned at the rate of five cubic feet per hour. In accordance with these instruc- CANDLE POWER OF ILLUMINATING GAS 129 tions the Gas Referees adopted the Metropolitan No. 2 Burner shown in Fig. 30. This burner differs from the older types mainly in having an adjustable air supply to the center of the burner. The burner is designed for all qualities of gas up to 20 candle-power. When lighting the burner the air regulating disc A is to be screwed down so that the full supply of air passes to the burner, and the burner is to be adjusted to approximately the five cubic foot rate. After allowing it to burn for fifteen minutes to become thoroughly warm the gas is to be adjusted carefully to the rate of 5 cu. ft. an hour, after which the air regulator is to be screwed upwards until the flame rises in the chimney as high as possible without smoking. The Metropolitan No. 2 Argand gives results materially higher than the ordinary Argand on gases of low candle-power. Bray's No. 7 Slit Union burner is frequently used with car- buretted water gas of more than 20 candle-power. The rate of gas consumption is as usual adjusted to 5 cu. ft. an hour. 11. The Bunsen and Leeson Photometric Screens. The Bunsen photometric disc dates from 1841 and in its simplest form consists merely of a sheet of paper with a grease spot in the center. This is mounted so that it may be moved back and forth between the two lights. When looking toward the stronger light the translucent grease spot appears bright. As the carriage is slowly moved away from the stronger light the constrast between the spot and the surrounding paper diminishes and almost dis- appears when equality of illumination is reached. If the carriage is moved still further in the same direction the grease spot stands out dark against the white background. The paper screen is usually mounted in a box as shown in Fig. 34 where by an arrangement of mirrors 'the observer standing in front of the instrument may see both sides of the screen at once. This form of apparatus is still frequently used. The Leeson star disc is a modification of the Bunsen screen. It consists of a piece of opaque paper from whose center is cut a star and which is pressed between two sheets of translucent paper. 12. The Lummer-Brodhun Photometric Screen. This is a very accurate form of photometer which is shown diagrammatic- ally in Fig. 31 and in perspective in Fig. 32. It consists 130 GAS AND FUEL ANALYSIS of a series of reflecting surfaces and prisms which direct light rays from the two sources into a telescope tube. Light entering from the two opposite sources R and L is diffusely reflected by the b-" FIG. 31. Diagram of Lummer-Brodhun photometric screen. opaque plaster of paris disc P onto the mirrors MI and M 2 and by them to the prisms AB. The prism A has most of its hypothe- Fia. 32. Lummer-Brodhun photometric screen. nusal face ground away, o.nly a small circular plane being left in the center. The two prisms are clamped closely together and become optically homogeneous over this small circular area shown CANDLE POWER OF ILUMINATING GAS 131 FIG. 33. Field of Lummer -B r o d h u n contrast photomet- ric screen. at ab. Of the light coming from L only that reaches the telescope which passes through this circular spot, the path of the rays being shown from L 2 . Of the light from R, that which strikes the spot ab passes on undeflected and is absorbed by the black walls of the box. The path of these rays is shown from R 2 . The other rays suffer total reflection into the telescope as shown in the rays from RI. The image in the telescope appears, therefore, as a circular spot illuminated from L in a circular field illuminated from R. In this equality photometer when the illumination from the two sources is identical the spot and the field are not to be distinguished from one another. In the Lummer-Brodhun contrast photom- eter advantage is taken of the physiological fact that the eye is able to perceive a smaller degree of difference in contrast than difference in brightness. By suitably cutting the prisms the image in the telescope is divided into four portions as shown in Fig. 33. In this figure the shaded trapezoidal space A' is illuminated from the same source as the shaded semi-circular area A. Simi- larly B and B' are illuminated from the same source. However, although the areas A and A' are illuminated from the same source, they are not equally illuminated, for through the interposition of a plate of glass before A' it receives about four per cent, less light than A. B' is in the same way and to the same degree less brilliantly illuminated than B. If , now, the light from the two sources is exactly the same both in intensity and color the semi-circular fields A and B will be identically illuminated and will not be distinguishable from one another. The trapezoidal figures A' and B' will also be identi- cally illuminated and will stand out with the same relief from their respective backgrounds. This can only happen when A and B are equally illuminated. It affords a more sensitive ocular test of the equality of A and B than can be obtained by comparing them directly. The lights at the two ends of the bench are never of absolutely the same, and are sometimes of a widely differing, color. When a Welsbach light is tested against the Hefner lamp the field illuminated from the mantle burner is a clear blue color while the other is a yellow. The eye cannot determine with much 132 GAS AND FUEL ANALYSIS accuracy when the yellow field and the blue one are illuminated to the same extent, but it can determine with greater accuracy when the yellow trapezoid A' stands out from its blue background with the same distinctness that the blue trapezoid B' stands out from its yellow background. The eye judges slight contrasts more accurately than large ones and therefore it is most sensitive when the photometer is almost at the neutral point. It is well to make an approximate setting for equality of A and B and then focus the attention on the contrast between the trapezoids and their respective backgrounds and complete the adjustment. 13. The Flicker Photometer. In the various forms of flicker photometer the light from each source is presented alternately and rapidly to the eye by means of revolving discs or prisms in the photometer box. When the intensity of light from the two sources is the same the flicker vanishes. No difficulty is experi- enced with lights of varying colors, but the photometer is fa- tiguing to the eye and its proper adjustment requires consid- erable skill. 14. The Gas Meter. The meters used in photometric work are of the same general type of wet meter as those described in 3 of Chapter VII for calorimetric work. They must be cali- brated with the same care and used with the same precautions. It is more convenient, however, to use a smaller meter which passes only 1/12 cu. ft. per revolution. When the gas is being burned at the rate of 5 cu. ft. an hour this meter will make exactly one revolution a minute. The dial of the meter is graduated into five parts with finer subdivisions. An observation of one minute will therefore give directly the uncorrected gas consumption in cubic feet per hour. Great care must be taken to see that the water of the meter is saturated with gas of the sort that is to be tested, for the " illuminants " of the gas are relatively soluble in water and a slight change in their percentage makes an appreci- able difference in the candle-power of the gas. 16. The Photometer Bench and Its Equipment. The preceding sections have discussed the various types of standard lights, burners and photometers which may be used. It is evident that wide latitude may be exercised in the choice of units and the method of assembling them to form a photometer bench. The details regarding the length of bar, type of standard light, CANDLE POWER OF ILLUMINATING GAS 133 form of test burner, kind of comparison box, and the directions for testing are in some cases controlled by legal enactment and are in some cases matters of arbitrary choice. It is under all cir- cumstances necessary that the meter, standard light, and gas burner be thoroughly reliable. The photometric screen may be of cheaper type and the bench itself may be of simple wooden construction with the scale made of yard sticks joined together. The bar most commonly used in America is 60 inches long. This is sufficiently accurate where ordinary gas flames are being tested. For lamps of high candle-power longer bars are desirable. FIG. 34. Photometer bench. A photometer bench frequently used is shown in Fig. 34. At the right hand end is shown the meter and next to it the candle balances in position, with the Edgerton standard burner on a swinging arm so that it may be used instead of the candles. At the left hand end is the Argand burner for the gas and between the two the bar itself with its screens and photometer. The piping for the gas is entirely below the table top but the pressure regulators and gages are shown above the table at the back. Precision photometers usually follow the German, or Reichs- 134 GAS AND FUEL ANALYSIS anstalt, pattern in which the bar is built up of a rigid steel track on which the carriage of the photometric screen travels. The standard light and the test light are usually also mounted on a travelling carriage with provision for clamping them rigidly at any desired point of the bench. The length of the bench may thus be varied at will. 16. Details of a Test. The details of a test will of course vary with the equipment of the photometer bench and especially with the type of standard light employed. Directions for the use of these lights have been given in preceding sections. In the follow- ing paragraphs will be found general directions which are appli- cable to most forms of apparatus. The meter is to be examined to see that the water-level is correct and if necessary more water is to be added. In case the test is unusually important the meter should be calibrated against a tank of known volume. The tightness of connections between the meter and the burner is to be assured by turning the gas into the meter while keeping the stopcock on the burner closed. The meter hand should not show any perceptible motion in one minute. If there is a small leak allowance may be made for it in the calculations, but it is vastly better to have the whole apparatus tight. A clean chimney is to be placed on the Argand burner and the burner lighted. The pressure gage between the diaphragm governor and the burner should indicate about 1 in. of water pres- sure depending on the exact type of burner used. The consump- tion of gas is to be set so that the meter hand makes a revolution in approximately a minute and the light is then allowed to burn for at least fifteen minutes. If the meter has had much fresh water added to it, or if it was last used for a gas of a different quality than the one soon to be tested, or if the gas is being drawn from long lengths of pipes where the gas lies dead, a longer time than fifteen minutes must be allowed to elapse before commencing the test which must not be started until it is certain that the gas burning is of representative quality and that it has not been changed by contact with the water of the meter. The final adjustment of the gas is made after taking into con- sideration the meter temperature and the barometric pressure. The desired rate of consumption being 5 cu. ft. measured under CANDLE POWER OF ILLUMINATING GAS 135 standard conditions, the correct apparent rate may be mentally calculated by adding to the 5 ft. 0.01 cu. ft. for each degree Fahrenheit shown by the meter thermometer above 60, and add- ing 0.03 ft. for each 0.1 in. of mercury pressure below 30. For example, if the meter temperature is 80 F., and the barometric reading is 29.5 the uncorrected consumption of gas per hour should be 5.0+.20+. 15 = 5.35. The gas is to be set to this desired flow with an error of less than 1/10 cu. ft.. The stop- watch is started as the hand crosses the zero and stopped after one complete revolution. It should read between 59 and 61 seconds. After the gas has been satisfactorily adjusted and the standard lamp given a final adjustment the test may be commenced. The observer starts the stopwatch as the large hand of the meter passes the zero and steps quietly to the photometer avoid- ing sudden movements which would create drafts, and makes and records the first observation. Four more readings are made at intervals of about twenty seconds and then the photometric screen is reversed and five similar readings taken. If the lights flicker during an adjustment the observer must wait until the drafts have subsided before completing the observation. The series of ten observations usually requires about five minutes. At their conclusion the operator steps back to the meter and stops the watch as the large hand of the meter again passes the zero. If a stopwatch is not available it is better to make the test during an even number of minutes rather than during the con- sumption of an even number of cubic feet of gas. If the observer holds the watch close to the meter and keeps his eyes on the watch till the second hand reaches the zero and then reads the position of the large meter hand, and follows the same procedure at the close of the test, the error will be well within the other necessary errors of the process. In case a stopwatch is available, it is more accurate to start the watch as the meter hand comes to its zero and to conclude the observation when the meter hand passes its zero, after the photometric observations have been completed. 17. Illustration of Calculation. The calculation which follows is for a test made on a 2500 mm. bench with a Hefner light as the standard. 136* GAS AND FUEL ANALYSIS Date, Feb. 26 Source of Gas. Proportional Tank. Test 42. Experimental Gas Plant. Gas burned in London Argand. Standard Light Hefner. Time Meter Reading Commencement of Test 10 : 16 : 00 A. M 47 . 8 End of test 10:21:00 50+24.6=74.6 Duration 5 : 00 Difference in meter readings 26 . 8 Cubic feet gas per hour uncorrected 5 . 36 Meter Temperature 80 F. Error in meter less than . 1 per cent. Barometer 29 . 5 inches. Correction factor . 930 Cubic feet gas per hour uncorrected 5.36. Corrected 4.98 Bar Readings 454, 456, 455, 462, 460 460, 458, 460, 463, 464 Average 462. ,11.- r(2500-462) 2 XI 5.00 Calculation [ (462) 2 79J X 4T98 ==17 ' 6 candle -P wer - 18. The Photometer Room. The photometer bench must be placed in a room of reasonably constant temperature which is free from drafts and yet well ventilated. A room ventilated so that the carbon dioxide does not rise above ten parts in 10,000 during a test is as much as can be expected in ordinary work. The carbon dioxide may rise to twenty parts without the air being more polluted than in an ordinary crowded street car in winter. The water vapor normally present in the air and that given off by the flames and the respiration of persons in the photometer room exerts an even greater influence on flames than the carbon dioxide but since its accurate determination is difficult, the carbon dioxide is usually taken as the measure of contamina- tion of the air. The committee on Photometry of the American Gas Institute 1 has published some curves showing the variation of certain flames with increased carbon dioxide when compared with an incandescent electric lamp. The humidity of the air varied so much from day to day that a comparison of one day's work with another could not be made and the following figures from their curves must be taken as merely illustrative of the large errors that may arise. l Proc. Am. Gas. Inst., 11, 480 (1907). CANDLE POWER OF ILLUMINATING GAS 137 COMPARISON OF PENTANE LAMP WITH GAS FLAME Parts CO 2 in 10,000 10.0 20.0 Per cent, loss of candle-power of gas flame 4.0 20.0 Per cent, loss of candle-power of pentane flame 7.5 31. COMPARISON OF PENTANE LAMP WITH CANDLES Parts CO 2 in 10,000 10.0 20.0 Per cent, loss of candle power of candles 19.0 27. Per cent, loss of candle power of pentane flame 13 . 16.5 It is therefore evident that the usual assumption that the standard light and the test light are equally affected by atmos- pheric conditions, is erroneous and that care should be taken to have the test made under as favorable atmospheric conditions as possible. 19. Jet Photometers. There are two main types of jet pho- tometers. In one type, gas passes through a pressure regu- lator and issues at constant pressure through a small round orifice where it burns in a jet whose height as read on the glass chimney is assumed to give candle power directly. In the other type of jet photometer the gas flame is kept at a constant height and the pressure required to force the gas through the burner opening is measured as an indication of candle power. No type of jet photometer can be relied on to do more than give approxi- mate determinations. They should be calibrated frequently against a bar photometer. 20. Accuracy of Photometric Work. When it is recalled that a Hefner lamp is considered as correct if it is within 2 per cent, of the standard, and that the absolute value of the pentane lamp may vary as much as 25 per cent. 1 in the course of a year on account of changing atmospheric conditions and further that the human eye is a very inaccurate scientific instrument, a greater accuracy than half a candle can hardly be expected with illumi- nating gas tested under ordinary conditions. Much larger errors may creep in unless care is taken. 1 J. B. Klumpp, Proc. Am. Gas Light Assoc., 1905. Appendix. CHAPTER IX ESTIMATION OF SUSPENDED PARTICLES IN GAS 1. Introduction. The estimation of particles held in suspen- sion in gases is daily becoming of greater importance on account of legal restrictions on pollution of the air and on account of insistence on closer control of industrial operations by manu- facturers. The problem is one of great difficulty and is usually susceptible of only approximate solution. Not only is it difficult to obtain the suspended solids present in a flue at a given point and time, but it is difficult to determine whether the solids thus determined were normal in amount or whether they were, for instance, low because of the deposition of an unusually large pro- portion prior to the point of sampling on account of slower veloc- ity of gas in the main, or because of lower temperature or for some other reason. 2. The Distribution of Particles in the Cross-section of a Straight Main. If the main is horizontal it is evident that there will tend to be a stratification of the particles, the large and heavy particles separating faster than the fine and light. This tendency to settle is, however, resisted by the whirling motion which gases traversing flues frequently possess and which is frequently caused by the inequalities in pressure produced by bends in the pipe. The velocity of gas in a straight main at ordinary working speeds is greatest at the center and least at the walls. The shape of the wave front varies with the speed of the gas, high velocities accentuating the difference. Solid particles are pushed gradually out of the zone of high velocity into one of lower velocity in the same way that a piece of wood in a river is grad- ually pushed to the still waters along the bank. This action takes place in a vertical as well as a horizontal main. The solids contained in a gas at the point of greatest velocity will therefore be the least in amount, the smallest in size, and the lowest in specific gravity. The quantity of particles, their size and specific gravity will all increase in the regions where velocity 138 ESTIMATION OF SUSPENDED PARTICLES IN GAS 139 is least. In a normal round main this point of greatest velocity is the center where will be found the fewest and lightest solid particles. Their quantity and magnitude increase in successive rings to the circumference. If the velocity of the gas is decreased until the main is also acting as a settling chamber there will be little difference in the velocity throughout the cross-section and the region near the top of the main will contain the fewest solid particles. This uneven distribution of suspended particles in a gas stream may take place very rapidly as was shown by the author 1 some years ago in an attempt to determine the amount of sus- pended tar in unpurified illuminating gas. A 14-in. main con- taining unpurified illuminating gas was tapped on its horizontal axis at a point a few feet beyond the exhauster and two sampling tubes inserted, one extending to the middle of the main and the other projecting through the wall only about an inch. Four tests were made and in each case the suspended tar caught in the sampling tube near the edge of the main was more than twice as great as that found in the tube projecting to the center. 3. Mean Velocity in the Cross-section of a Gas Main. Threl- fall 2 has shown that it is necessary to investigate the distribution of velocity for each individual case as it arises, but that in general the radius of the circle of mean velocity is about 0.775 of the ra- dius of the pipe. In one case it was as high as 0.9 of the radius but in no case did it sink to 0.69 which is the figure quoted for water flowing through a long and smooth pipe. The radius of mean velocity did not change with varying speed of gas flowing through the pipe within the ranges of 600 ft. and 3600 ft. per minute, which marked the limit of the experiments. Threlfall's experi- ments were on pipes varying from 6 to 36 in. in diameter. 4. Influence of Bends in a Main. If gas flowing through a straight main comes to a bend there will be a change in the rela- tive velocities of the particles of the gas throughout the cross- section. The kinetic energy of a body is represented by the expression l/2mv 2 where m represents the mass of a body and v its velocity. It is evident therefore that the particles with the greatest mass and the greatest velocity will be projected beyond 1 Proc. Mich. Gas. Ass., 1906. 2 Proc. Inst. Mech. Eng., 1904, 1, 245. 140 GAS AND FUEL ANALYSIS their companions. The point of maximum velocity will shift from the center to a point nearer the opposing wall and will then slowly return to its normal position with a spiral movement. Solid particles on account of their greater mass may strike the opposing wall and if they or the wall are sticky may adhere there and build up deposits. It will be evident, from what has preceded, that it will not be possible to find a single point in a gas main from which it is pos- sible to draw a fair sample of gas for the determination of sus- pended solids. If a sample can only be taken at a single point, the termination of the sampling tube should be at about the point of mean velocity, as explained in the preceding section. In im- portant tests it is advisable to draw a number of samples from various points in the cross-section of the main. A tube with nu- merous perforations along its length is useless for this work. Sep- arate tubes should be used each with its own filter and aspi- rator as explained in Chapter I. 5. Velocity of Gas in a Sampling Tube. The rate of flow through the sampling tube has a ma- terial effect on the accuracy of sam- * pling as has also the inclination of the sampling tube to the gas stream. It __> N. B is evident that if -a sampling tube is * inserted at A of Fig. 35 at right angles to the flow of gas, even assuming that the solid particles are uniformly dis- tributed, the result will be incorrect FlG 35> _ Diagram show . for the heavy particles will tend to be i ng method of inserting sam- carried past the open end of the tube pling tube in gas main, and not drawn into it. The suspended solids will be reported low even if the speed of gas within the sam- pling tube is as high or even higher than that in the main. If, on the other hand, the opening of the sampling tube faces the approaching stream of solid particles as at B, on the same assump- tion of uniform distribution of particles, the result may be cor- rect, or it may be high or low. If the speed of the gas in the sam- pling tube is the same as that in the main the result should theo- retically be correct. The whole column of gas opposite the open- ing of the tube should enter without distortion. If, however, the ESTIMATION OF SUSPENDED PARTICLES IN GAS 141 velocity in the sampling tube is lower than that in the main the column of gas approaching the opening will be disturbed and part of it will be forced aside. 'The solid particles will on account of their momentum not be pushed aside so readily and will there- fore enter the tube in unduly great amount, giving a high result. If the velocity of the gas in the sampling tube is greater than that in the main there will again be a disturbance in the approaching column of gas. A column of gas larger than the opening of the tube will be sucked in, but the solid particles in the outer shell of gas thus sucked in will not be diverted from their course and will pass by the opening of the tube, giving a low result. Brady 1 states that in sampling blast-furnace gas an error of more than 44 per cent, was caused when the sampling speed was dropped to half that in the main. It is thus evident that the speed with which gas enters the sampling tube must be carefully controlled. The velocity of the gas must however be reduced before it passes through the filtering medium or the finely divided particles will not be taken out. The usual sampling tube has therefore a relatively small aperture. Care must be taken that the aperture is not so small that it will become clogged, which readily happens when tarry matters are present. Each case must be studied independently. 6. The Filtering Medium. Where conditions permit, filter paper discs or shells make satisfactory filtering media. The Brady gas filter for dust in blast-furnace gas is described in the article referred to in the preceding section. A filter using a disc of filter paper as developed by Mr. W. S. Blauvelt 2 of the Semet- Solvay Company has been used by the author with good results. Various commercial filters of this type are now on the market. When large amounts of tar are present a weighed tube filled with a fibrous material may with advantage be inserted ahead of the filter paper. The filtering materials to be inserted in the sam- pling tube will vary with conditions. If the temperature is high, sand or ignited asbestos is suitable. Ignited asbestos is usually to be preferred since on account of its fibrous nature it makes a more efficient and a lighter filter. This last consideration is of impor- tance since the suspended solids collected frequently weigh only a few milligrams and it conduces to accuracy to have the increase 1 Jour. Ind and Eng. Chem., 3, 662 (1911). z Proc. Am. Gas. Inst., 4, 795 (1909). 142 GAS AND FUEL ANALYSIS in weight of the filter relative to its initial weight as large as may be. The tubes may be of glass, porcelain or quartz protected if desirable by an iron j acket. The tubes after filling and before use should be placed in an air bath heated to the temperature to which they are to be exposed later and dry air should be drawn through them until they are constant in weight. They should then be cooled in dry air, weighed, carefully stoppered and if possible kept in a dessicator until used. The asbestos for this purpose should not be soft enough to pack readily and choke the tube. The fine washed asbestos used for analytical work is not so good for this purpose as a cruder sort. Sometimes when much tar is present it is advantageous to procure the crude asbestos rock and merely crush it coarsely in an iron mortar. 7. Estimation of Suspended Tar and Water. The amount of suspended matters caught by a filter paper may be estimated either by color or, if sufficient in amount, may be determined by weight. A second weight after drying at 105 C. for an hour will give by difference the moisture and other volatile matter, while the weight after ignition will give the mineral matter, cor- rection being made if necessary for the change in composition due to ignition. The Steere Engineering Company manufactures a convenient filtering device which they name a tar camera and which is illustrated in Fig. 36. They furnish with it a colori- metric chart from which the amount of sus- FIG. 36 Tar camera pen( jed tar may be determined directly by for colorimetxic tar de- h J termination. Comparison of colors. Where asbestos filters have been used a similiar procedure may be followed provided the asbestos has been ignited before use. In drying the tube however it is not sufficient to heat it externally. Dry air must be drawn through until it comes to constant weight. The water will be driven off and also approximately 25 per cent, of the weight of the tar. The non-volatile tar remaining may afterward be extracted with ESTIMATION OF SUSPENDED PARTICLES IN GAS 143 chloroform or carbon bisulphide, and this figure increased by one- third will give a rough estimate of the amount of tar present. It is not possible to determine accurately the amounts of water and suspended tar since it is not feasible to determine how much of the material volatilized is tar and how much is water. 8. Electrical Precipitation of Suspended Particles. Where the expense warrants the installation of the process, the method of electrical precipitation as developed on the large scale so success- fully by Cottrell 1 may be applied. The equipment consists of a small step-up auto transformer capable of giving 15,000-30,000 volts, a rectifier for this high-tension alternating current and a precipitating vessel which may be made of an iron pipe with an insulated electrode in the center. An exhauster for aspirating the gas and a meter for measuring it must also be provided. This apparatus will quantitatively precipitate all suspended solid and liquid bodies including tar and operates on such large volumes that the precipitated materials can readily be examined. 1 Jour. hid. and Eng. Chem., 3, 542, (1911). CHAPTER X CHIMNEY GASES 1. Introduction. A knowledge of the chemical composition of the gases escaping from a chimney aids much in controlling the efficiency of the furnace. It makes very little difference whether the fuel is burned to raise steam or to melt steel and it is of equally small importance whether the fuel burned be solid or liquid. The only assumption is that it is desirable to burn the fuel as completely as possible without the introduction of any unnecessary excess of air. When gaseous fuels are burned the same general principles apply but there is somewhat greater complication in calculation. This chapter therefore limits itself to a study of the gases arising from complete combustion of solid or liquid fuels. Let us see how much light a knowledge of the composition of the gas can throw on the operation of the furnace. 2. Sampling. Samples should be drawn from a point as near to the fire as possible, while still allowing time for complete combustion. Irregular streams of unburned gases may arise from a bituminous coal fire with adjacent streams of almost unchanged air. Kreisinger, Augustine, and Ovitz 1 have shown that if samples are taken a short distance beyond the firebox of a fur- nace such as is used in boiler plants the percentage of carbon dioxide may vary from 0.8 to 15.6 per cent, in two samples taken simultaneously and only sixteen inches apart. When composite samples obtained over a period of eight minutes were drawn from points only eleven inches apart and four feet from the combustion chamber there was still a difference of 1.0 per cent, in the carbon dioxide of the two samples. If, however, the investigator goes a long distance from the point of combustion in an effort to obtain a fair sample, he runs the danger of finding his gases diluted by air pulled through leaks in the setting. The very greatest care must therefore be exercised in sampling chimney gases. 1 Bui. 135 Bureau of Mines. Combustion of Coal and Design of Furnaces. 144 CHIMNEY GASES 145 3. Formation of Carbon Dioxide. Air is composed of prac- tically 21 volumes of oxygen and 79 volumes of nitrogen and other inert gases. When oxygen unites with carbon there is formed carbon dioxide which is stable unless it comes into intimate con- tact with carbon or other reducing agent at a high temperature. Chemically the result is expressed as follows: C + 2 = C0 2 . The expression means not only that carbon dioxide is formed by the union of carbon and oxygen, but also indicates that one volume of carbon dioxide is formed from one volume of oxygen and that the volume of the smoke gases after cooling is the same as that of the air which was used. This follows from the law of Gay Lussac which states that a molecule of one gas occupies the same volume as that of any other gas under like conditions. The simplicity of this volume relation makes it extremely desirable to work with volumes instead of weights in problems where gases are involved. Since one volume of oxygen forms one volume of carbon dioxide it follows that the theoretically best composition of the chimney gases from the combustion of carbon would be 21 per cent. CO 2 and 79 per cent. N 2 . This is unattainable in practice because the strong reducing action of the glowing carbon on the carbon dioxide will cause formation of carbon monoxide (CO) which will not be again oxidized unless it is brought in contact with free oxygen while still at a high temperature. An excess of oxygen is in practice necessary to ensure this. It follows from the fact that the volume of the carbon dioxide is the same as that of the oxygen which formed it, that all chimney gases resulting from the com- bustion of pure carbon to carbon dioxide will contain 21 per cent, of CO 2 +O 2 and 79 per cent, of N 2 . 4. Effect of Hydrogen of Coal on Composition of Chimney Gases. -The simple relation stated in the preceding section only holds where carbon is the only fuel burned, a condition which is quite closely fulfilled with a coke fire and approximately ful- filled when anthracite coal is the fuel. When fuels contain not- able percentages of hydrogen, as does bituminous coal, and to a greater extent petroleum and most gaseous fuels, part of the oxygen of the air burns to water which escapes from the furnace as steam. 146 GAS AND FUEL ANALYSIS The changes which air undergoes on combustion with coal may be illustrated as follows : 100 cu. ft. dry entering air is distributed: 79 cu. ft. N 2 10 cu. ft. O2 to combine with C 9 cu. ft. O 2 as excess 2 cu. ft. O 2 to combine with H of coal Products of combustion from 100 cu. ft. dry air after cooling tc initial temperature. 79 cu. ft. N 2 10 cu. ft. CO 2 9 cu. ft. O 2 4 cu. ft. H 2 O 102 cu. ft. total volume The water in this illustration will have a volume of 4.0 cu. ft. if it remains in the state of vapor after cooling to the initial tem- perature of the air before combustion. If this combustion had taken place in the bomb calorimeter where, instead of dry air, oxygen saturated with moisture is used, the 4.0 cu. ft. of water would have condensed to liquid when the products had cooled to their initial temperature. If combustion gases taken from the bomb are analyzed in the usual way in a burette filled with water so that the gases always remain saturated with water vapor, no account whatever would be taken of the water vapors and the analysis reported would be the same as if the gas volume had been : 79 cu. ft. N 2 80 . 6 per cent. 10 cu. ft. CO 2 10 . 2 per cent. 9cu. ft. O 2 9. 2 per cent. 98 cu. ft. 100 . per cent. Let us now work backwards from this gas analysis which may be assumed to represent the composition reported for a stack gas. When the gas sample was being drawn part of the steam formed may have condensed. If the gas sample was stored over water it certainly became fully saturated with water vapor so that its volume became entirely independent of the amount of steam which it contained in the chimney. The resulting gas CHIMNEY GASES 147 composition would then be the same as if the combustion had taken place in the bomb calorimeter. The calculation would be as follows: 100 cu. ft. of air contained 79 cu. ft. of nitrogen which is now 80.6 per cent, of the chimney gas, therefore the 79 volume of the gas is gQ~^ = 0.98, or 98 per cent, of the initial volume of entering air measured under the same conditions of temperature and pressure. It follows that 2.0 of the 21 volumes of oxygen have combined with hydrogen to form water. The volume of the water formed, so long as it remains in the vapor form, will be twice that of the oxygen from which it was formed as shown by the equation. 2H 2 + 2 = 2H 2 0. The hydrogen in this case is contained in the coal and is con- sidered as a solid just as the carbon is. In the case of gaseous fuels the problem is a little more complicated and is treated under Pro'ducer Gas. 5. Carbon Monoxide and Products of Incomplete Combus- tion. The presence of carbon monoxide, hydrogen or hydro- carbons is a sign of incomplete combustion and represents loss of heat which would have been liberated in the furnace had combustion been complete. Heating value 1 Ib. C to CO 2 14,600 B.t.u. 1 Ib. C to CO 4,450 B.t.u. Since carbon burning to CO only evolves 30 per cent, of the heat obtainable by complete combustion it is evidently uneconomical to allow more than small amounts of this gas to appear in chimney gases. It is frequently stated that carbon monoxide is formed when carbon burns with an insufficient supply of air. This is only a partial truth for with a bed of coals at a dull red heat it is difficult to form carbon monoxide no matter how much the air supply is limited. If the free oxygen in the chimney gases is below 3 per cent, it will be entirely normal to find products of incomplete combustion. The presence of carbon monoxide and other in- completely burned gases is abnormal when associated with much more than 3 per cent, free oxygen. It indicates either a faulty design of the furnace or carelessness on the part of the fireman. 148 GAS AND FUEL ANALYSIS Furnaces intended for coal high in volatile matter must have roomy combustion chambers so that the streams of gas given off by the coal may have time to mix with air and burn before they become chilled by contact with cold surfaces. Furnaces designed for anthracite coal do not have such large combustion chambers and hence do not give good results with bituminous coal. As mentioned in Chapter III, the estimation of carbon monox- ide presents some difficulties and the careless analyst may readily report a fraction of a per cent, of carbon monoxide when none is there. On the other hand, the natural tendency is to fail to find hydrogen when it is present in only small amounts. The presence of soot in chimney gases is not necessarily an in- dication that measureable amounts of unburned gases are present for the particles of tar and carbon formed by the destructive distillation of the coal burn much more slowly than do the gases and also have higher ignition temperatures and so are more likely to escape combustion. 6. Volume of Air and of Chimney Gases. The volume of the air used in combustion per pound of carbon and the volume of the chimney gases may be calculated from the gas analysis. The method is based on the assumption that the nitrogen of the air passes through the furnace unchanged in volume, and that all of the nitrogen of the chimney gases is derived from the air. This assumption is practically correct, the small amount of nitrogen derived from the coal introducing only a negligible error. It is necessary also to have some factor to connect the weight of carbon burned with the volume of the chimney gases. One pound of carbon burning to CO 2 requires 32.1 cu. ft. of oxygen measured wet at 60 F., and 30 in. barometric pressure and yields 32.1 cu. ft. carbon dioxide. If the air were perfectly dry only 31.4 cu. ft. would be needed per pound of carbon and the volume of carbon dioxide would be 31.4 cu. ft. Let us assume the following gas analysis : CO 2 8.5 per cent. O 2 10.8 per cent. N - 80 . 7 per cent. It was shown in 3 that the increase in the percentage of the nitrogen over 79 was due to the condensation of steam formed by CHIMNEY GASES 149 the union of hydrogen of the coal with oxygen of the air. The volume of these gases referred to 100 of air may be obtained by 79 multiplying them by the factor 5^ = 0.979. oU. i 8.5X0.979= 8.3CO 2 10.8X0.979= 10.6 O 2 80.7X0.979= 79.0 N 2 97.9 O 2 which has disappeared as steam 2 . 1 forming 4 . 2 steam. 100.0 The volume of air used per pound of carbon may now be ob- tained. To burn 1 Ib. carbon =32. 1 cu. ft. moist O 2 forming 32. 1 cu. ft. CO 2 32.1X10.6 Oxygen in excess 5-0 =41 . 1 o. o , A 32.1X2.1 Oxygen forming steam ^5 = 8.1 o . o Total oxygen per pound carbon, 81 . 3 cu. ft. 79 Accompanied by ^ X 81 . 3 = 305 . 5 cu. f t. N t Corresponding to 386 . 8 cu. ft. air. The excess of air may be determined from the ratio Oxygen used ^32. 1+41. 1+8. 1 = 81.3 Oxygen required 32.1+8.1 40.2 ' The volume of the chimney gases is obtained directly from the above, it being remembered that the volume of the C0 2 is the same as that of the O 2 forming it and that the volume of the steam (assumed to be cooled to standard temperature without condensa- tion) is twice the volume of the oxygen forming it. Volume of chimney gases from 1 Ib. carbon in the above example : CO 2 32. 1 cu. ft. H 2 O vapor 2X8.1 16.2 cu. ft. O 2 41 . 1 cu. ft. N 2 t 305.5 cu. ft. Total chimney gases 394. 9 cu. ft. 150 GAS AND FUEL ANALYSIS 7. Loss of Heat in Chimney Gases. The bomb calorimeter gives the heating value of coal with 100 per cent, efficiency. In this instrument the gases are cooled to practically the same temperature as before ignition and all water formed in combus- tion is condensed to liquid water. The calorimeter makes no distinction between water present as moisture in the coal or as combined water in the coal, and the water formed by combustion of the available hydrogen or hydrocarbons. In calculating the loss of heat in chimney gases the clearest procedure is to deter- mine what would have been the state of the products of combus- tion if combustion had taken place in a bomb calorimeter, and then calculate the losses caused by the higher temperature of the gases and the presence of water in the state of vapor rather than as liquid. The heat carried away by these gases may be determined by multiplying their volume by the rise in temperature and by their specific heat. It was first shown in 1883 by Mallard and Le- Chatelier that the specific heats of gases are not constant but in- crease with rising temperature. Engineers have been slow to adopt these variable specific heats but there can be no question as to their general correctness. The mean specific heats ex- pressed in British thermal units per cubic foot and per pound at constant pressure have been calculated by the author from the data of Holborn and Henning 1 and are given in Tables X and XI of the Appendix. It will be noted that the specific heats of oxygen, nitrogen and all permanent gases are the same per cubic foot, an agreement which holds true only for specific heats by volume and not for those for which the unit basis is weight. The loss of heat per pound of carbon in the particular case given above will be calculated as follows, a temperature of 600 F. for the escaping gases being assumed : Temperature through which gases are heated 600 60 = 540. Use mean specific heats from Table X corresponding to 600 F. Heat lost in CO 2 , 32 . 1 X . 0253 X 540 = 439 B.t.u. Heat lost in steam, 16 . 2 X . 0221 X 540 = 193 Heat lost in oxygen, 41 1 j 346<6x o. 0177X540 = 3310 B.t.u. Heat lost in nitrogen, 305 . 5 J 3942 1 Annalen der Physik, 23, 809 (1907). CHIMNEY GASES 151 It is necessary to know the percentage of carbon in the coal before this loss of heat per pound carbon can be calculated to the desired basis of loss per pound of coal. The loss of heat per pound of dry coal = Loss per pound carbon X per cent, carbon in dry coal 100 Moisture present in the coal when placed on the fire will be vaporized, and, in case combustion is complete, will escape from the stack as steam. It is immaterial whether or not it underwent decomposition in the fire, only the initial and final states being important. The amount of water thus vaporized calculated from the analysis of the coal is reported in pounds and is most conveniently kept in that form throughout the calculation. The mean specific heats by weight at constant pressure are given in Table XI of the Appendix. Moisture present in the air intro- duced into the firebox will be heated from room temperature to that of the escaping gases. Its amount may be determined from observations with a wet and dry bulb thermometer from which the percentage humidity may be calculated as described in 16 of Chapter VII. The volume of water vapor per cubic foot of air for various temperatures is given in Table XII of the Appendix. There is also steam in the stack gases which is derived from the union of the hydrogen and oxygen of the coal with each other. It is sufficiently accurate to assume that all of the oxy- gen of the coal unites with the hydrogen of the coal to form water, that the excess or available hydrogen unites with the oxygen of the air td form water and that all of the carbon of the coal unites with the oxygen of the air to form carbon dioxide. The volume of steam due to this union of the hydrogen and oxygen of the coal with each other can only be accurately calculated from an ultimate analysis. Fortunately its amount is small and fairly constant for a given type of coal. The weight of water so formed, sometimes called " combined water," may be taken as: 2.5 per cent, for anthracite coals. 6.0 per cent, for Eastern bituminous coals. 10.0 per cent, for bituminous coals of the Western or Illinois type. These figures may for this purpose be added to the percentage of moisture in the coal. 152 GAS AND FUEL ANALYSIS There are also changes in the ash of the coal which may in- volve absorption of oxygen or liberation of SO2 or CO2 but they are negligible in a calculation of this sort. PROBLEM ILLUSTRATING CALCULATION OF LOSS OF HEAT IN CHIMNEY GASES ~ . Average composition of Data Coal as charged chimney gases Moisture 9.3 per cent. Volatile matter. . 31.7 CO 2 9. 6 per cent. Fixed carbon 53.7 O 2 9.8 Ash 5.3 N, 80.6 100.0 100.0 B.t.u. per Ib 12,456 Temp, escaping gases. . 720 F. Per cent, total carbon. . 71 . 6 Temp, inlet air 70 F. Relative humidity 75 per cent. Distribution of air entering furnace. Factor to correct for change of volume caused by formation of water in 79 combustion o7r~g = . 98 9 . 6 X . 98 = 9 . 4 O 2 f or burning carbon 9.8X0.98= 9.6 O 2 in excess 80.6X0.98= 79.0 N 2 98.0 2 . O 2 for burning hydrogen 100.0 Volume gases from 1 Ib. carbon. 1 Ib. carbon produces 31 .4 cu. ft. dry CO 2 . Factor ^^=3. 34 9.4X3.34= 31.4cu. ffc. CO 2 4.0X3.34= 13.4cu. ft. H 2 O vapor 9.6X3.34= 32.7cu. ft. O 2 79. 0X3. 34 = 263. 8 cu. ft. N 2 Volume dry air required for combustion 100X3.34 = 334 cu. ft. Moisture in air for combustion assumed as 75 per cent, of saturation at 70 F. 0.75X0.026=0.019 cu. ft. per cu. ft. air 334 . 00 X . 019 = 6 . 35 cu. ft. for 1 Ib. carbon CHIMNEY GASES 153 Losses due to sensible heat of gases. Gases heated from 70 - 720 F. Loss in CO 2 = 3 1.4X650X0. 0257- 525 B.t.u. Loss in H 2 O (sensible heat only) vapor from entering air .................... 6.35 cu. ft. formed from available H of coal ............. 13.4 cu. ft. moisture (9.3) and combined water (6.0) of coal = . 153 Ib. per Ib. coal carbon with . 0476 cu. ft. per Ib. (Table IX, Appendix) ............... 4.5 cu. ft. Total water vapor ......................... 24. 25 cu. ft. Heat lost = 24 . 25 X 650 X . 0221 = 347 B.t.u. Loss in O 2 and N 2 32.7+263.8 = 296.5X650X0.0177= 3410 B.t.u. Total B.t.u. lost in sensible heat per Ib. carbon ................ 4282 Total B.t.u. lost in sensible heat per Ib. coal 4285X0.716= 3065 Losses due to latent heat of water. Moisture in coal 9.3 per cent. =0.093 Ib. per Ib. coal Combined water in coal 6.0 per cent. =0.060 Ib. per Ib. coal Water formed by combustion of available H of coal forming 13 . 4 cu. ft. H 2 O vapor with . 0476 Ib. per cu. f t. = 13 . 4 XO . 0476 = . 638 Ib. per Ib. carbon = 0.638X0.716 =0.4571b. per Ib. coal Total water which would have condensed had com- bustion taken place in bomb calorimeter 0. 610 Ib. per Ib. coal Latent heat of vaporization . 610 X 1067 = 651 B.t.u. Total heat losses per Ib. coal burned. Sensible heat of gases = 3065 B.t.u. Latent heat of water = 651 B.t.u. Total heat lost =3716 B.t.u. O*7 "I f* Per cent, heat lost v>456 = 2 ^ 8 P er cent * 8. Interpretation of Analysis of Chimney Gases. An analysis is of no value unless the sample is representative. Some of the difficulties in sampling are mentioned in Section 2. If the sam- pling and analysis have been properly performed the conclusions 154 GAS AND FUEL ANALYSIS to be drawn from the preceding paragraphs may be summarized as follows: Carbon Dioxide. The higher the percentage of CO 2 in chimney gases without the presence of CO or hydrocarbons, the more efficient is the furnace. When the fuel is coke or anthracite coal the sum of the percentages of carbon dioxide and oxygen should be between 20.5 and 20.8. If the fuel is bituminous coal the sum of the carbon dioxide and oxygen will drop to 19 per cent, and if the fuel is oil or gas the figure will be still smaller. In ordinary practice the percentage of carbon dioxide should be as large as the oxygen, and with well-equipped and operated plants the pro- portion of CO 2 to O 2 should be as high as 2 to 1. With liquid or gaseous fuels the proportion of CO 2 will be still higher. The CO 2 as reported includes a small amount of S0 2 from the sulphur of the coal, which does not usually amount to more than a few hundredths of a per cent. Oxygen. When solid fuel is burned on an ordinary grate it is necessary to have an excess of air to insure complete combustion. This excess should be kept as small as possible. Carbon Monoxide and Products of Incomplete Combustion. These products should be entirely absent from chimney gases. Their presence indicates waste of fuel. Unless the analytical work is carefully done as much as 0.2 per cent. CO may readily be reported through error. Nitrogen. Nitrogen is present in air to the extent of 79 per cent, by volume and will be present in at least that percentage in chimney gases. With bituminous coal the percentage will rise to 81 per cent, and with oil or gaseous fuel the percentage will be higher. The percentage of nitrogen can only fall below 79 per cent, through the introduction of some gas which makes the total volume of the chimney gases greater than that of the air from which they were derived. The formation of carbon mon- oxide will affect this result since it occupies twice as much space as the oxygen from which it was derived. The amount of CO in chimney gases is, however, too small to exert any appreciable influence of this sort. Loss of Heat in Chimney Gases. The loss of heat will depend on the temperature and volume of the gases. The volume of the gas is in general indicated by the relative percentage of carbon CHIMNEY GASES 155 dioxide and oxygen. The higher the per cent, of oxygen and the lower the per cent, of carbon dioxide the greater is the loss of heat. The loss will vary in steam-boiler practice between 15 and 45 per cent. With smelting furnaces where the gases escape at high temperatures the loss may be much higher. CHAPTER XI PRODUCER GAS 1. Formation of Producer Gas. Producer gas is formed whenever air is brought into contact with fuel under such con- ditions that carbon monoxide is an important constituent of the products resulting from their combination . The formation of pro- ducer gas is frequently said to be due to incomplete combustion, but the statement is only a half truth, for a limited quantity of air supplied to a fire will not necessarily produce carbon monox- ide. The primary product formed when carbon burns in air is carbon dioxide, the equation being written C+0 2 = C0 2 If this carbon dioxide comes into intimate contact with glowing carbon, it unites with more carbon and carbon monoxide is formed according to the equation C0 2 +C<=*2CO. These two reactions are frequently combined into one and the typical reaction of the gas producer is usually written 2C+0 2 = 2CO. This equation shows that one volume of oxygen is converted into two of carbon monoxide. The composition of the resulting gas may be shown as follows : f 210 2 =42CO = 34.7 per cent. CO. = I 79N 2 = 79N 2 =65.3 per cent. N 2 . When steam is introduced in the bottom of the producer the reaction desired is: C+H 2 O = CO+H 2 . 156 PRODUCER GAS 157 If the temperature of the producer is low, this reaction may proceed in part as follows: C+2H 2 = C0 2 +2H 2 . If bituminous coal is placed in the producer there will also be products of destructive distillation including hydrocarbons both saturated and unsaturated, hydrogen and carbon monoxide. The largest single constituent of producer gas is nitrogen, which will not often fall below 50 per cent. Carbon monoxide and hydrogen rank next in percentage. The hydrocarbons are practically always under 5 and are usually less than 3 per cent. Carbon dioxide should be low. It is, however, frequently as high as 10 per cent. The following are some analyses of producer gas. 1 TYPICAL ANALYSES OF UP-DRAFT PRESSURE-PRODUCER GAS (Per cent, by volume) From Bituminous Coal From Lignite From Peat Carbon dioxide (CO 2 ) 9.84 10.55 12.40 Oxvcen (Q<>) .04 0.16 0.00 Ethylene (C 2 H 4 ) .18 0.17 0.04 Carbon Monoxide (CO) 18 28 18 72 21.00 Hydrogen (Ha) 12.90 13.74 18.50 Methane (CH 4 ) 3.12 3.44 2.20 Nitrogen (N 2 ) 55.64 53.22 45.50 TYPICAL ANALYSES OF DOWN-DRAFT PRODUCER GAS (Per cent, by volume) From Bituminous Coal From Lignite From Peat Carbon dioxide (CO 2 ) 6.22 11.87 10.94 Oxygen (O 2 ) 0.13 0.01 0.41 Ethylene (C 2 H 4 ) 0.01 0.00 0.06 Carbon monoxide (CO) Hvdrocen (H) 21.05 12 01 16.01 14.76 16.91 10.19 Methane (CH 4 ) Nitrogen (N 2 ) 0.49 60.09 0.98 56.37 0.66 60.83 1 From Bulletin 13, U. S. Bureau of Mines. "Re'sume" of Producer-Gas Investigations by R. H. Fernald and C. D. Smith. 158 GAS AND FUEL ANALYSIS 2. Sampling Producer Gas. The quality of gas yielded by a given producer may change quickly. Soon after a charge of bituminous coal has been added, the amount of volatile tarry vapors -and of gaseous hydrocarbons in the gas increases. Within a half hour the larger part of the volatile matters may have dis- tilled off leaving the gas almost free from hydrocarbons and from tar vapors. Rapid changes will also be noted after poking the fire. An average sample of producer gas may be obtained only by extending the sampling over a long period, as directed in Chapter I. There will be especial difficulty in determining the quantity and heat value of the suspended tarry particles. Yet these values must be ascertained if the heat value of the crude gas is to be determined accurately. The method of collecting the tar particles on filter papers, given in Chapter IX, may be followed and the weight of tar per cubic foot of gas thus obtained. The papers and tar may then be burned in a bomb calorimeter and after deduction of the heat due to the known amount of filter paper, the heating value of the tar may be obtained. This determination is not often made, as it is difficult to carry it out accurately. However, it is not possible to find the true heat balance on a furnace fired with crude producer gas, especially if from a bituminous producer, unless such a determination is made. Ordinarily the determination of tar and suspended particles is neglected and the sampling then is to be conducted as described in Chapter I. 3. Analysis of Producer Gas. The constituents to be determined in producer gas are carbon dioxide, unsaturated hydrocarbons, oxygen, carbon monoxide, hydrogen and methane. The methods are given in Chapters II, III and IV. No difficulty will be experienced except with hydrogen and methane. The percentage of these gases, except in water gas, is usually so small that a sample after removal of the absorbable constituents is no longer explosive when mixed with air. It should be emphasized that failure to obtain an explosion does not mean the absence of hydrogen and methane but merely that they are present in less than explosive amounts. Explosion may be brought about by addition of a known volume of pure hydrogen to form an explosive mixture but it is usually simpler to use a method which PRODUCER GAS 159 does not involve explosion. Combustion with a hot platinum spiral as in the Dennis and Hopkins method ( 9 of Chapter IV) or with copper oxide as in the Jaeger method ( 11 of Chapter IV) affords a satisfactory method for determination of these constituents. 4. Interpretation of Analysis. The important constituents are carbon dioxide and carbon monoxide. Oxygen should be entirely absent, as it should all have been brought into combina- tion in its passage through the producer. Its presence in a pro- ducer gas may be an indication of leakage in sampling or of leakage into the flue prior to sampling. Rarely, if operating conditions in the producer are bad, and the fire is thin, there may be such a channel formed in the producer that air will rush through the producer without its oxygen becoming combined. Such a condition will be indicated by extremely high carbon dioxide, with low percentages of combustible gases. When a producer is running under normal conditions its operation may be quite closely checked by the percentage of carbon dioxide alone. High carbon dioxide is in practically all cases an un- favorable symptom. It may be due to a cold fuel bed in the producer caused either by an excess of steam or by slow running, it may be due to a thin fuel bed which does not allow sufficient time and contact for the reduction of the carbon dioxide to monoxide, and it may be due to channels or chimneys in a deep fire which allow uncombined air to get through the fuel bed and burn above the coals. A cold fuel bed in a producer burning bituminous coal will tend to increase the percentage of unsaturated hydrocarbons, but in no case will they amount to more than a few tenths of a per cent. A hot and thin fuel bed and especially a channeled fuel bed will cause the unsaturated hydrocarbons to practically disappear, since they are decomposed at the high temperature and of all the gases show the greatest avidity for oxygen. The carbon dioxide is almost a direct measure of the thermal efficiency of the producer, the only exception being its appear- ance as the result of the interaction of carbon and steam at a relatively low temperature as in the Mond producer where it is accompanied by a high percentage of hydrogen. Under other circumstances high carbon dioxide means low thermal efficiency 160 GAS AND FUEL ANALYSIS for the 70 per cent, of the energy of the carbon which should have been converted into the potential energy of the carbon monoxide is all changed to the sensible heat of the carbon dioxide and accompanying gases. 6, Heating Value of Producer Gas. The heating value of producer gas may be determined in a calorimeter as described in Chapter VII for illuminating gas. A special tip must be used on the burner and care be taken to see that the flame burns clear. The heating value of producer gas may be as low as 100 British thermal units per cubic foot and it frequently happens that it does not burn readily in a Bunsen burner. The gas must be carefully cooled and cleaned before testing. This operation separates tar whose amount and heat value must be determined as directed in 2 of this chapter. The heating value of the purified gas may also be calculated from the analysis as indicated in 17 of Chapter VII. The low per- centage of unsaturated hydrocarbons in producer gas makes the errors of calculation less than is the case with illuminating gas. On account of the difficulty in cleaning the gas and in keeping a steady flame in the calorimeter, the heating value is usually obtained by calculation. It is worth while to emphasize again that the heating value of the cleaned gas from bituminous coal is lower than that of the hot gas which still contains tar vapors and that allowance must be made for the tar vapors in calculating the heat value of the gas which is used while hot. 6. Volume of Producer Gas. It would be very desirable to be able to calculate the volume of producer gas per pound of coal, as is done in Chapter X for chimney gases. There are, how- ever, so many possible reactions in the gas producer and the changes in volume are so complicated, especially in a bituminous producer, that it is only possible to make such calculations for simple cases. Assume a producer burning pure carbon in dry air. It is manifest that the only products of combustion will be C02,CO and N2 Assume the following composition of the gas COj 5.5 per cent. CO 25 . 6 per cent. N 2 68.9 per cent, PRODUCER GAS 161 The air entering the producer was composed of 79 volumes of nitrogen for every 21 volumes of oxygen. The change in percentage of the nitrogen in the producer gas is due to changes resulting from the union of oxygen with carbon. The first step is to trace the changes taking place when 100 volumes of air pass through the producer and find the relative volumes of C0 2 and CO for 79 volumes of N 2 . 7Q CO 2 5.5X-= 6.3= 6.3vols. O 2 79 CO 25.6Xgg^=29.4 = 14.7 vols. O 2 79 . N 2 100.0 vols. air. One pound of carbon burning to carbon dioxide requires 32.1 cu. ft. of oxygen (at 60 F. and 30 in. of mercury pressure) and yields 32. 1 cu. ft. of carbon dioxide. One pound of carbon burning to carbon monoxide requires 16.05 cu. ft. of oxygen and yields 32.1 cu. ft. of carbon monoxide. It follows that the weights of carbon burning to CO and CO 2 are proportional to the volumes of the two gases. In the present instance CO 2 6. 3=^7X100 = 17. 7 per cent. oq 4 CO 29. 4=~*X100 = 82.3 per cent. One pound of carbon yields 0.177X32.1= 5.7cu. ft. CO 2 0.823X32.1= 26.4cu. ft. CO 3.76 X32. 1 = 120.7 cu. ft. N 2 152 . 8 cu. ft. producer gas. The sensible heat will be calculated as in Chapter X. 5. 7 X. 0268X1000= 153 B. t. u. 26.4 120^7 147.1X.0180X1000 = 2647 B. t. u. 2800 B. t. u. 11 162 GAS AND FUEL ANALYSIS l Energy in 26.4 cu. ft. of CO 8540 B. t. u. Sensible heat in gases 2800 B. t. u. Total energy in gas at 1000 F. 11340 B. t. u. Efficiency of producer when gas is ueed at 1000 F. = X 100 = 77.6 per cent ' 7. Efficiency of a Gas Producer. In the simple instance cited above it is easy to calculate the energy contained in the gas. The only potential energy in the gas from one pound of carbon is contained in the 0.823 lb., which is now in the form of 26.4 cu. ft. carbon monoxide. The heating value of this is: 26.4X323.5= 8540 B.t.u. The total energy of the coal if burned to carbon dioxide would be 14600 British thermal units. If the gas is cooled before being burned so that its only energy is the potential energy of the carbon monoxide the efficiency of the producer is 8540^ 14600 ^X 100 = 58 .5 percent. If the gas is burned while still hot, say at 1000 F., there should be credited to the producer also the sensible heat of the gas, as calculated in the preceding section where the efficiency was shown to be 77.6 per cent. The above simple relations do not hold when steam is being injected into the producer with the air nor when bituminous coal is being used as a fuel. On account of the varied possibilities of chemical reaction in these cases the volume of the gases cannot be calculated from their chemical composition. If the volume of the gases is measured by a Venturi meter, or otherwise, then it is possible to calculate the sensible and potential energy of the gases as indicated in this chapter and the one preceding. _ CHAPTER XII ILLUMINATING GAS AND NATURAL GAS 1. Introduction. Chemical analysis plays a minor role in \ testing illuminating gas and natural gas. The determination ',. of heating value is described in Chapter VII and of candle-power in Chapter VIII. The ordinary chemical analysis as described in Chapters III and IV usually includes the determination of carbon dioxide, unsaturated hydrocarbons, oxygen, carbon mon- oxide, hydrogen and methane, nitrogen being taken by difference. A separate determination of benzene is sometimes desired in illuminating gas and of gasoline vapors in natural gas. Sulphur may be called for in both gases. Naphthalene and ammonia are / frequently determined in coal gas. 2. Sampling. The sampling of natural gas and of purified illuminating gas usually offers little difficulty, since the gases are thoroughly mixed and contain such small amounts of sus- pended particles that they are usually negligible. The chief point to be observed in sampling from service pipes in cities is to see that the gas is allowed to run long enough to flush out the pipe and bring to the sampling cock gas which is representative of that flowing in the mains. Illuminating gas of high candle- power must not become chilled in the sampling process, for there is danger of condensing the benzene or other hydrocarbon vapors which it contains. Rubber connections are also to be mini- mized in sampling because of the solubility of the hydrocarbons in rubber. The water used in the sampling vessels must be carefully saturated before use, because unsaturated hydrocarbons are relatively soluble in water. In case unpurified illuminating gas is to be sampled, additional precautions must be taken on account of the presence of material amounts of ammonia, hydrogen sulphide, carbon dioxide and other very soluble gases, as well as suspended tar particles. In case it is sufficient to determine what the approximate composi- tion of the gas would be after purification it is sufficiently accu- 163 164 GAS AND FUEL ANALYSIS rate to sample in the usual way and trust the water of the samp- ling tank to remove the ammonia, hydrogen sulphide and part of the carbon dioxide. In case the actual composition of the unpurified gas is desired, these soluble constituents must be separately determined as indicated in succeeding sections. 3. General Scheme of Analysis. Carbon dioxide, unsaturated hydrocarbons, oxygen, carbon monoxide, hydrogen and methane are usually determined according to the methods of Chapters III and IV. In the case of unpurified illuminating gas there may be appreciable amounts of hydrogen sulphide absorbed with the carbon dioxide. If it is desired to separate the two the hydrogen sulphide may be estimated according to the method of 7 of this chapter. The estimation of carbon dioxide, oxygen and carbon monoxide does not present any peculiarities, although emphasis should be laid on the necessity of complete removal of the unsaturated hydrocarbons before the estimation of oxygen by phosphorus. In the case of Pintsch gas it sometimes requires five minutes shaking with bromine water to affect such a com- plete removal of the hydrocarbons that the phosphorus will smoke when the gas is subsequently passed over it. The deter- mination of hydrogen and methane in illuminating gas offers no marked peculiarity. In natural gas, higher hydrocarbons are present and complicate the calculation. Ethane may be present in natural gas and also in water gas and Pintsch gas. The methods for its determination have been discussed in Chapter IV. Nitrogen is taken by difference and as the analysis is a rather long and complicated one, the errors piling up on the nitrogen are apt to be material. A direct combustion of the gas with copper oxide, as described in 12 of Chapter IV, gives a more accurate determination of the residual nitrogen. 4. Chemical Composition of Illuminating Gas. The so-called " coal gas" is made by the destructive distillation of coal in closed retorts. ' The composition of the gas is dependent on the compo- sition of the coal, the temperature of the retort and to some extent its shape and size. There is nothing, however, which distinctly characterizes gas from the large retort of the by-product coke oven from that of the small horizontal retort of the gas works. The oxygen of the coal appears in the gas partly as carbon dioxide, partly as carbon monoxide, and partly as water vapor. ILLUMINATING GAS AND NATURAL GAS 165 None of it will be evolved as gaseous oxygen. The carbon mon-J oxide will be greater at a high temperature than at a low one, but will usually stay between the limits of 5 and 9 per cent. A high retort temperature will cause cracking of the hydrocarbons with decrease of their percentage and increase of hydrogen. A frac- tion of 1 per cent, of free oxygen is normally present in illumin- ating gas, partly because of air entering during the operation of charging and drawing the retort, partly because of leaks in the long condensing system and partly because of liberation of the oxygen dissolved in the water used in the scrubbers. In so far as this oxygen comes from the air it must be accompanied by four volumes of nitrogen. Nitrogen must always be present in this amount. Less than four volumes of nitrogen for one of oxygen TYPICAL ANALYSES OF ILLUMINATING GAS 1 2 3 4 5 C 6 H 6 * 1.2 2 * 1.3 CO 2 1 5 1 4 1 3 2 7 3 5 CnH 2 n 4.6 4.0 2.0 6.5 11.6 O 2 3 0.9 0.5 1.1 0.7 CO 7.1 4.6 4.8 12.4 31.6 H 2 46.4 49.6 50.7 38.4 35.7 CH 4 36.3 35.0 38.1 28.4 9.0 N 2 . 3.7 3.3 2.4 10.5 3.7 C 2 H 6 2.9 * Not separately reported. 1. Coal gas of 17 c.p. and 650 B.t.u. 2. Coke oven gas enriched by benzol to 16.4 c.p. and 626 B.t.u. (Proc. Am. Gas Inst., 6, 519 (1911).) 3. Coke Oven Gas. Fuel Gas (Proc. Am. Ga* Inst., 6, 519 (1911).) 4. Mixed Coal and Water Gas 15 c.p. and 615 B.t.u. 5. Carbureted Water Gas of 24 c.p. and 649 B.t.u. (Proc. Am. Gas Inst., 7, 739 (1912).) indicates a faulty analysis. The unavoidable nitrogen in the gas arising from the destructive distillation of the nitrogenous com- pounds of the coal will be between 1 and 1.5 per cent. High percentages of nitrogen unaccompanied by a corresponding amount of oxygen indicate that suction has been maintained on the porous retorts, so that air has been sucked in. The oxygen thus brought in contact with the hot gas will at once burn with formation of carbon dioxide or water while the nitrogen will 166 GAS AND FUEL ANALYSIS remain and appear as such in the purified gas. In the manufac- ture of water gas a high percentage of nitrogen will result if the gasmaker turns the gas into the holder before all the gas pro- duced while blowing air is flushed from the machine by the water gas. 5. Benzene. Benzene is a normal constituent of coal gas and also probably of water gas, but its amount in unenriched coal gas is always less than 1 per cent., and it is not usually deter- mined. Its solubility in water and in caustic soda is slight so that in the ordinary analysis the benzene is not absorbed by the caustic but passes on to the bromine pipette where it dissolves in the ethylene bromide formed in the reaction between ethylene and bromine and is therefore estimated with the ethylene as "illuminants." As benzene has a very high illuminating power this grouping is logical and sufficiently satisfactory for most purposes. Hempel recommends that 1 c.c. of absolute alcohol be placed in a pipette otherwise filled with mercury. An explosion pipette answers well for this purpose. A sample of gas is to be passed into the pipette and shaken with the alcohol to saturate it with ethylene, and the sample to be analyzed is later passed into this same pipette and shaken three minutes. The gas is then drawn back into the burette and passed into a second mercury pipette containing 1 c.c. of distilled water which removes the alcohol vapors. The decrease in volume from the initial reading is recorded as benzene. The method is to be considered only an approximate one. Morton 1 recommends that, after removal of carbon dioxide by caustic soda as usual, the gas be passed into an ordinary simple absorption pipette containing concentrated sulphuric acid (sp. gr. 1.84) and shaken vigorously for one minute. The decrease in volume after drawing back into the burette represents benzene. Dennis and McCarthy 2 dispute the accuracy of this method and propose ammoniacal nickel cyanide as a reagent. The gas after carbon dioxide has been removed by caustic is passed into the pipette containing the ammoniacal nickel cyanide solution and drawn back and forth between the burette and 1 Jour. Am. Chem. Soc., 28, 1728 (1906). 2 Jour. Am. Chem. Soc., 30, 233 (1908). ILLUMINATING GAS AND NATURAL GAS 167 pipette for about two minutes. It is then passed into a five per cent, solution of sulphuric acid and shaken until the ammonia is absorbed, which requires about two minutes. According to the authors, the absorption is quantitative and the result unaffected by ethylene. The process of Haber and (Echelhauser 1 is based on Bunte's observation of the solubility of benzene in ethylene bromide. The authors treat the gas with a fresh solution of bromine water to which is added, immediately after the reaction an excess of strong potassium iodide. The liberated iodine is titrated with thiosulphate and the difference between the figures of this titration and those obtained by a blank test on an equal volume of the original solution represents the bromine which has combined with the ethylene. One cubic centimeter deci- normal thiosulphate corresponds to 1.12 c.c. ethylene at C. and 760 mm. or to 1.22 c.c. at 60 F. and 29.5 in. mercury pres- sure. The diminution in volume of the gas after the usual treat- ment with bromine water followed by caustic gives the sum of the benzene and ethylene. The difference between this volume and that indicated by the titration for ethylene gives the benzene. If a more exact method of estimation of benzene is desired and a large sample of gas can be obtained the method of Harbeck and Lunge 2 may be used. It consists in aspirating 10 liters of the gas through a mixture of equal weights of fuming nitric acid and concentrated sulphuric acid. Part of the resulting dinitrobenzene crystallizes out after the acids are diluted, cooled and neutralized, and may be filtered and weighed. The dinitrobenzene remaining in solution is extracted from an ali- quot portion of the filtrate with ether and also weighed. 6. Benzene and Light Oils by Differential Pressure Method. Davis and Davis 3 have described a differential pressure method for determining benzene and light oils which requires relatively small samples of gas. The apparatus as illustrated in Fig. 37 consists of a pair of flasks connected by a differential manometer. The saturation vapor pressure of a liquid is a function of the temperature and is independent of the amount of liquid (provided 1 Jour, fiir Gasbel, 43, 347 (1900). *Zeit. fiir anorg. Chem., 16, 41 (1898). 3 J. Ind. & Eng. Chem., 10, 709-718 (1918). 168 GAS AND FUEL ANALYSIS there is an excess of liquid after saturation is reached) and of the pressure on the inert gas into which it evaporates. If each of the flasks A' and B in Fig. 37 is filled with air and closed, the mano- meter will show no pressure difference between them. If now the small bulb a containing benzene is broken, the benzene will evaporate into A and the manometer will after equilib- rium is reached, show an in- creased pressure in the vessel corresponding to the satura- tion pressure for benzene; since the flask B contains air with no benzene. If flask B is filled with coal gas at atmospheric pressure and the flasks are closed the manom- eter will indicate no pressure difference. If the bulbs a and 6 containing benzene are now broken the vapor pressure in A will rise as before but the vapor pressure in B will not increase so much since the gas in B had already con- tained some benzene vapors. The manometer connecting the two flasks will therefore register a pressure equal to the partial pressure of the benzene originally in B, from FIG. 37. Davis differential pressure apparatus for benzene in gas. which the amount of benzene in the original gas can be calcu- lated. Coal gas carries not only benzene but also toluene and small amounts of other hydrocarbon vapors. The liquid in the bulbs should therefore be light oil instead of benzene and under such circumstances the reading will give percentage of light oil vapors. If benzene alone is to be determined the flasks must be cooled to almost the freezing temperature of benzene, at which temperature the vapor pressure of the other constituents is negligible. ILLUMINATING GAS AND NATURAL GAS 169 7. Hydrogen Sulphide. Hydrogen sulphide should be present only in minute traces in purified illuminating gas. The usual test is an approximate one based on the reaction of hydrogen sulphide and lead acetate to form black lead sulphide. A strip of white filter paper moistened with colorless lead acetate is exposed to the gas and the depth of the resulting black stain noted. The details of the test differ widely. The New York State Commission prescribes that gas shall show no hydrogen sulphide when tested by exposing the paper moistened with lead acetate to a current of gas burning at the rate of 5 cu. ft. per hour. The paper must not become discolored after thirty sec- onds of such exposure. Ramsburg 1 and McBride, 2 Weaver and Edwards have given a full discussion of the various methods of testing for hydrogen sulphide in gas. Hydrogen sulphide is always present in unpurified illuminating gas. In the ordinary gas analysis it is absorbed by the caustic soda simultaneously with the carbon dioxide and reported with it. Its quantitative estimation may be carried out as follows. One or more liters of the gas is bubbled through a solution of ammoniacal cadmium chloride and the resultant cadmium sul- phide is filtered. If the precipitate contains much tar it may be washed on the filter with benzol. Place filter and precipitate in cold dilute hydrochloric acid till dissolved and titrate with standard iodine. If the iodine solution is made by dissolving 1.0526 grm. iodine to the liter, 1 c.c. will be equivalent to 1/10 c.c. of hydrogen sulphide measured damp at 60 F. and under 30 in. of mercury pressure. A solution of ten times the above strength is more convenient if much hydrogen sulphide is present. A rapid approximate estimation of hydrogen sulphide may be made in the Bunte gas burette described in 4 of Chapter V. The burette is first filled with water containing a little thin starch paste to act as indicator, and the sample of gas drawn in and measured rapidly to prevent error due to the solubility of the hydrogen sulphide in the water of the burette. Standard iodine solution is now admitted from the reservoir at the top of the burette, about a cubic centimeter at a time, until the blue color formed by the reaction between the iodine and starch 1 Proceedings Am. Gas. Inst., 4, 453, 1909. 2 Technologic Papers 20 and 41, Bureau of Standards. 170 GAS AND FUEL ANALYSIS persists after repeated shaking, showing that the hydrogen sul- phide has all been oxidized. If iodine solution of the concentra- tion given above has been used, the volume of hydrogen sulphide may be read directly from the volume of iodine used, which must be obtained by measuring the iodine solution still remaining in the reservoir of the burette. Tutwiler has modified the burette by making the reservoir longer and graduating it so that the iodine used may be read directly. 8. Total Sulphur Compounds. Illu- minating gas and almost all other gases used for fuel contain, in addition to hydrogen sulphide, compounds of sulphur and carbon such as carbon bisulphide and more complex compounds like the mercaptans. These compounds are usually estimated after complete com- bustion, in which process all the sulphur, whatever its previous combination, is converted into sulphur dioxide and sulphur trioxide. These gases are ab- sorbed, oxidized to sulphuric acid and weighed as barium sulphate. Care must be taken to see that the air used for combustion, which is usually ten times the volume of the gas, is itself free from sulphur compounds, and that combustion is complete. The form of burner and absorption apparatus is immaterial, except for convenience. The older design due to Drehschmidt has been modified by Harding, 1 Jenkins, 2 and the Bureau of Standards. 3 -Any of these modifications may be made from ordinary laboratory apparatus. The apparatus described and illustrated in Fig. 38 is that of the Bureau of Standards. The entire apparatus consists of pressure regulator, U water gage, meter, sulphur apparatus, wash bottles, 1 Jour. Am. Chem. Soc., 28, 537 (1906). 2 Jour. Am. Chem. Soc., 28, 542 (1906). 3 Circular No. 48, Bureau of Standards. FIG. 38. Bureau of standards form of sulphur apparatus. ILLUMINATING GAS AND NATURAL GAS 171 and jet pump, connected in the order named; soda-lime tower for the purification of the air; and battery and spark coil, connected ( to the burner. The burner is a porcelain tube of 3-4 mm. internal diameter. The gas is ignited by an electric spark between plati- num terminals which are soldered to nickel leads, in order that only a short length of platinum wire will be needed. One of the leads is placed within the burner tube, its lower end being brought j out through a small side tube which is sealed to the glass tube just below the rubber stopper; the wire can be held in place and the opening closed by sealing wax. The terminal outside is wired to the porcelain burner tube. The platinum wire becomes heated by the flame and thus reduces the likelihood that the flame will be extinguished by fluctuations in gas pressure. This igniter therefore eliminates the principal difficulty both of lighting and of regulating the burner which is experienced with apparatus of this type. The stopper which closes the lower end of the combustion chamber also serves as a connector, the porcelain burner tube and the glass T piece being firmly fastened into it by means of Kho- tinsky or sealing wax. The small tip through which the gas enters just above the primary air inlet is also held in with Kho- tinsky cement. The tip can be easily removed for cleaning, or tips of various sizes adapted to the gas to be burned can be inserted. The air necessary for complete combustion after being purified by passage through the large soda-lime tower is supplied to the flame in two portions. The primary air is drawn in by the gas as it passes through the small tip; the secondary air enters through the two inlets at the side of the combustion chamber. The combustion chamber, made of Pyrex glass tubing, is about 360 mm. long and about 25 mm. in internal diameter. The nar- row tube at the top may be used for introducing water when it is desired to rinse out the apparatus; when in operation this tube is closed with a small cork. Satisfactory drainage is pro- vided by the sloping layer of paraffin or sealing wax covering the stopper at the bottom. When the burner is lighted the second- ary inlet air keeps the base of the apparatus cool, but the rest of the combustion chamber up to the side tube is heated so that no condensation takes place on the walls. For that reason it is usually unnecessary to rinse out the combustion chamber. By 172 GAS AND FUEL ANALYSIS means of a cork connector the first of a series of wash bottles is attached to the apparatus. Rubber tubing must not be used at this point on account of the danger of introducing sulphur from it; but the wash bottles may be connected to each other and to the suction pump by rubber tubing, which may also be used to connect the air inlets to the soda lime-tower Only one wash bottle is shown in the illustration, but three are usually required for satisfactory operation. In order that the suction may pull the gas steadily through the wash bottles it is necessary that the end of the inlet tube of the first bottle be perforated with a number of small holes. With a single, large opening the opera- tion of the burner is not steady. The wash bottles may be of any of the ordinary forms. Each bottle should contain enough absorbent so that the products of combustion will bubble through a depth of 1 to lj^ inches of liquid. The air is drawn in and the products of combustion are drawn through the apparatus by the suction of a small water jet pump, or its equivalent. The spark for igniting gas is produced by a single dry cell and an induc- tion coil of the size rated as giving a quarter-inch spark. Before beginning a determination the apparatus should be adjusted to burn gas at the required rate, not more than 2.5 cu. ft. per hour, and to use the proper amount of primary and secondary air. The amount of gas burned should be adjusted by removing the small glass inlet tip from the burner and reduc- ing or enlarging the opening as required. The opening may be reduced by heating carefully in a flame; it may be enlarged by filing back the tip until the required internal diameter is reached. To adjust the air supply the wash bottles are filled with water to the depth of 1 to lj^ inches above the lower end of inlet tubes. The jet pump is then turned on to draw air through the appa- ratus at a rapid rate, the battery circuit is closed to produce a continuous spark, and the gas is turned on last. This order should be followed every time the burner is lighted. If the gas is turned on before both the air flow and the spark are started, an explosion may result. As soon as the gas has ignited the battery circuit may be opened. The amount of air entering the burner tube must be regulated rather carefully, so that the flame is entirely nonluminous with a clearly defined inner cone. The amount of secondary air flowing ILLUMINATING GAS AND NATURAL GAS 173 through the apparatus must, of course, be sufficient to give com- plete combustion. This is assured when the outer cone of the flame is steady and sharply defined. If the outline of the flame appears " ragged" or indistinct, some of the sulphur is certain to escape oxidation. There is little danger of having too much secondary air, but the amount is limited by the capacity of the wash bottles. To insure complete absorption and prevent mechan- ical loss of the sulphate solution, it is desirable to keep this rate of air flow reasonably low, but it is better to use too much air than too little. The primary air is regulated by the pinch cock on the inlet tube, but the adjustment of secondary air should be made by regulating the jet pump rather than by closing the air inlet. When these adjustments have been completed a test for leaks should be made, the gas-supply line purged, and the meter ad- justed. While the line is being purged the soda lime tower is filled and sufficient 5 per cent, solution of sodium carbonate (Na 2 CO a ), with a few drops of hydrogen peroxide or of bromine water, is introduced into each wash bottle to bring the liquid 1 to lj^ inches above the bottom of the inlet tube. The burner is now connected to the meter and the meter read- ing recorded. The suction, spark, and gas are then turned on in order, and pressure and temperature readings are made and re- corded. The burner should usually be adjusted to consume about 1 cubic foot of gas per hour. When enough gas has been burned the gas is turned off first; then the valve controlling the jet pump is closed carefully to prevent tap water being sucked back into the wash bottles. The meter, barometer, thermometer, and mano- meter readings are again recorded. The contents of the wash bottles are transferred to a beaker and the bottles rinsed twice with a little water. It is ordinarily unnecessary to wash out the burner chamber since it is dry at all times and the sulphur is mostly present in the form of sulphur dioxide, which passes on quanti- tatively. The hot walls prevent the condensation of any sulphur trioxide which may be present. Sulphate is determined in the solution by any of the usual methods. 1 grm. BaSO 4 = 0.1373 grm. S. or 2. 1 19 grains S. The result is usually reported as grains sulphur per 100 cu. ft. of gas measured under standard conditions. 9. Naphthalene. The amount of the hydrocarbon, naphtha- lene CioH 8 , usually present in gas is less than one-tenth of 1 per 174 GAS AND FUEL ANALYSIS cent. Its small amount would make it unworthy of considera- tion were it not for its disagreeable property of causing stoppages in gas mains. Its estimation is therefore sometimes demanded as one of the steps in controlling the manufacturing process. The usual method for purified gas is that devised by Coleman and Smith, 1 who based their method on Kiister's 2 method for separating naphthalene from other hydrocarbons. The method depends upon the property which naphthalene possesses of com- bining molecule for molecule with picric acid to form an insoluble compound. In the author's laboratory it has been customary to make the picric acid about 1/20 normal, which is an almost saturated solution, and to use as alkali Ba(OH) 2 1/5 normal, with phenolphthalein, lacmoid or methyl red as indicator. The color change is not difficult to observe, but the same conditions must always be observed in the analyses that are maintained in the standardization. When the gas to be tested has been puri- fied and there is no danger of naphthalene deposition at room temperature the gas may be passed through a wet experimental meter, then through a wash bottle containing dilute tartaric or other non-volatile acid to remove all traces of ammonia, then through a bottle containing water to stop acid which might have spattered over and then through a bulbed gas washing tube with about ten bulbs containing 50 c.c. of the standard picric acid. A yellow precipitate betrays the presence of naphthalene. After five or more cubic feet of gas have been bubbled through the apparatus, the bulbed tube is disconnected and washed into an Erlenmeyer flask of about 150 c.c. capacity, the bulbed tube being rinsed with 50 c.c. of picric acid from a pipette. The flask is then closed with a rubber stopper carrying a glass tube through which most of the air is sucked from the flask by a filter pump in order that it may not blow up when heated later. The evacuated flask is then placed in a water bath which is maintained at the boiling temperature for an hour. Rut ten 3 says that it is sufficient to heat to 40 C. for half an hour. The flask is then removed and allowed to cool when the naphthalene picrate will have recrystal- lized as a definite compound CioH 8 . CeHaOHfNC^s. This is l Jour. of Gas Lighting, 75, 798 (1900); 80, 1277 (1902). 2 Berichte, 27, 1101. 3 Jour, fur Gasbel, 52, 694 (1909). ILLUMINATING GAS AND NATURAL GAS 175 | filtered off and an aliquot part of the filtrate is titrated with alkali. The difference between the amount of alkali required \ for this titration and that which would have been required on a , blank titration gives, when calculated for the whole volume of ; picric acid present, the naphthalene absorbed. One cubic centi- meter of N/5 Ba(OH) 2 is equivalent to 0.0256 grm. naphthalene. If great accuracy is not required the heating of the naphthalene picrate in its solution may be dispensed with. When gas freed from tar particles but otherwise unpurified is to be tested for naphthalene, it must be freed from ammonia and hydrogen sulphide, which affect the titration. The gas must not be allowed to cool below the temperature of the main during this purification process or naphthalene may deposit. The purifying train may be placed in an oven such as that shown in Fig. 40, placed so that it is in close proximity to the main and heated to the desired temperature. The first washer may contain lead acetate, the second a dilute acid and the third water. The tube with picric acid must not be placed in the oven since naphthalene will not be absorbed by a warm solution, but must be immediately outside. If naphthalene is deposited in the glass connecting tube it may be vaporized and driven forward by heat. Where much naphthalene is present two bulbed tubes should be used in series and if much precipitate appears in the second tube the liquid in the first tube should be renewed since a dilute solution of picric acid does not remove naphthalene com- pletely. The precipitated naphthalene is estimated in the same manner as in the case of purified gas. Where naphthalene must be estimated in crude gas containing suspended tar and it is desired to separate the naphthalene pres- ent as vapor in the gas from that which is carried in the dissolved tar particles, the method becomes still more complicated. The method devised by the author 1 to separately determine the naph- thalene present as vapor in the gas from that which is carried in the dissolved tar particles involves precipitation of naphthalene as the picrate with subsequent recovery of the naphthalene. Considerable care is necessary in sampling gases which contain tar, if it is desired to distinguish between the naphthalene actu- ally present as vapor in the gas and that existing dissolved in the l Proc. Mich. Gas. Ass., 1904; 1905, 83. J. Gas Lighting 88, 262 and 92, 388. 176 GAS AND FUEL ANALYSIS fine, mist-like particles of tar suspended in the gas and which will be removed later by mechanical scrubbing. It is out of the ques- tion to collect a tank of gas, say from the foul main, and transport it to to the laboratory for examination. The condensation and separation of tarry products in the gas holder would nullify the value of the figures obtained. It is necessary to separate the suspended tar and remove the naphthalene before change of temperature has had time to alter the conditions prevailing at the point of sampling. This requires that the tar vapors shall be mechanically filtered from the gas without any change Glass Sleeve ' Covered with Rubber FIG. 39. Details of naphthalene train. in temperature. The result is attained by inserting horizontally into the main a glass tube about J^ in. in diameter filled with glass wool or asbestos fiber. This tube should project into the main at least 6 in. and at as short a distance outside the main as possible should connect with the picric acid absorbing train, and then with a gas holder of about 1 cu. ft. capacity and of known volume as shown in Fig. 2 of Chapter I. The strong solvent power of tar for naphthalene renders it absolutely essential that the gas shall not be scrubbed by passing through cold tar, as would be the case if the tube for separation of the tar were placed outside the main and, for example, connected to a pet cock. After the sample has been drawn, the picric acid contains the naphthalene and tar which were still present in the gas as vapor. The solution and precipitate is washed into a 200 c.c. Erlenmeyer flask and treated with alkali to neu- ILLUMINATING GAS AND NATURAL GAS 177 tralize the picric acid. It is our custom to add here an excess of solid alkali, making an almost saturated solution when hot, merely to prevent so much moisture being carried into the dry- ing tube. As this addition of solid alkali makes the solution hot, it should not be added until just before the apparatus is con- nected up so as to avoid loss of naphthalene. If the glass con- necting tube contains naphthalene deposited from the gas by condensation the tube should be broken into fragments and dropped into the same flask. FIG. 40. Oven for naphthalene determinations. The flask is then corked with a stopper carrying two glass tubes, one long enough to reach nearly to the bottom of the flask and the other terminating just below the cork and extending above the cork to make connections to the tube containing lime and phosphorus pentoxide as shown at A of Fig. 39. At the other end of the drying tube B close connection is made to a small weighed U tube C which can be immersed in ice water. The tube containing the asbestos and tar is connected directly to a similar drying tube and U tube. The arrangement of the train is shown in Fig. 39 and the whole set in the oven is shown in Fig. 40. The oven is heated to 70-80 C. and air slowly drawn through the system, volatilizing the naphthalene and moisture in the tar. The moisture is taken up by the dryer lime and phosphorus pentoxide while the naphthalene passes on to be frozen out in the U tube placed directly outside of the oven in a trough filled full of cracked ice. The analysis is complete when the 12 178 GAS AND FUEL ANALYSIS weight of the naphthalene U tube becomes constant or nearly so at consecutive weighings after a two- or three-hon interval. The volatilization of the naphthalene from the gas ii usually complete in six hours. The time required for an analysii of tar thus deposited varies with the amount of tar in the sampl< and usually takes thirty or forty hours for samples drawn from th< standpipe when the weight of tar amounts to 8 or 10 grm When the analysis is complete, the tar volatilizing tube is agaii weighed, the loss giving the weight of moisture and naphthalene given off. Having the weight of naphthalene in the U tube, w( have the weight of moisture also, which, however, is at best onlj approximate, because there is always more or less light oil, sue! as benzene, given off from the tar with the moisture and naph thalene, which cannot easily be estimated. Finally the volatiliz- ing tube is set in a Soxhlet extractor and the remaining content! extracted with chloroform until free of all soluble material After drying, the 'tube is weighed, this giving the weight of fre< carbon in the tar. The oven used is of galvanized iron and is 20 in. high by 16 in wide by 14 in. deep. It is shown in Fig. 40 and is arranged s< that eight samples may be worked at a time. The drying trail consists of a heavy glass tube about J^ in. internal diamete: and 12 in. long. It contains broken lime for about two-thirds o its length and phosphorus pentoxide thoroughly incorporated ii glass wool for the other one-third. This introduction of the glasi wool with the phosphorus pentoxide prevents the gas fron forming channels in the latter and thus aids in rendering the ex traction of moisture complete before reaching the naphthalene U tube. The lime used must be extremely rapid in its reactioi with water in order to avoid too great expense for phosphorus pentoxide. It may be readily made by igniting crushed lime stone to a dull red heat for two hours in a muffle. If the lumps o: lime are too small, the expansion attendant upon their slacking will crack the tube. If the lumps are too large, the gas will not b< dried sufficiently. A satisfactory mixture is obtained by taking everything that will pass a four-mesh sieve and will not pass f twelve-mesh. Connection with the naphthalene U tube is mad( as shown in Fig. 39 at C, through a glass sleeve made air tight b} a piece of rubber tubing placed over the whole. This prevents ILLUMINATING GAS AND NATURAL GAS 179 the naphthalene from coming in contact with the rubber. It is well known that rubber absorbs naphthalene. Nevertheless, it has been found safe to use rubber stoppers in the volatilizing oven. Rubber stoppers in frequent use absorb all they can take up and after a few runs cause no further trouble. This method allows the estimation of naphthalene as vapor in the gas, and of water, non-volatile tar, free carbon and naphtha- lene in the suspended tar. The separation of the water and non- volatile tar is not very accurate, the light oils which are vaporized by the air drawn through being reported as water. The estima- tion of naphthalene is, however, quite accurate. The air passing through will leave the system saturated with naphthalene at the temperature of ice water, but this loss need not amount to over a milligram for each ten hours run. It is possible that the naph- thalene may be contaminated by other hydrocarbons and the naphthalene deposit is sometimes slightly oily and has a low melting point. 10. Ammonia. Crude coal gas before scrubbing may contain as much as three-quarters of a per cent, of NH 3 by volume. After the gas has traversed the scrubbers the amount of ammonia should be reduced to a trace. The gas to be tested is bubbled through standard acid, suc- tion being produced by an aspirator holding a cubic foot, which also acts as a measuring device. The excess of acid is then titrated back with standard alkali, cochineal or sodium alizarine sulphonate being used as an indicator. If the gas contains, much suspended tar the end of the titration cannot be observed sharply and it is necessary to make the solution alkaline and redistill the ammonia into standard acid before titrating. One cubic centimeter of N/10 acid equals 0.0017 grm. NH 3 . 11. Cyanogen. Cyanogen compounds exist in small amounts in unpurified gas. In the purification process they are partly removed in the ammonia scrubbers and largely in the iron-oxide purifiers or in special scrubbers containing compounds of iron. The gas is estimated by bringing it into contact with an alkaline solution carrying suspended ferrous hydroxide and titrating the resultant ferrocyanide. The method according to Mueller 1 is as follows: l Proc. Am. Gas Inst., 5, 249 (1910). 180 GAS AND FUEL ANALYSIS 11 To determine the amount of cyanogen in gas, the cyanogen is converted into potassium ferrocyanide by passing the gas through a caustic potash solution containing freshly precipitated ferrous hydrate in suspension. After filtering, the potassium ferrocyanide is determined in the clear solution by acidifying and titrating with a standard solution of zinc sulphate until all ferrocyanide has been precipitated as zinc ferrocyanide. The end reaction is determined as follows: A drop of a 1 per cent, solution of ferric chloride is put on a piece of white filter- paper absolutely free from iron. A drop of the liquid being tested is then put on the paper near the drop of ferric chloride so that the liquor as it spreads out on the paper will come in contact with the ferric chloride. Care must be taken that the precipitate of zinc ferrocyanide does not come in contact with the iron solution. As long as there is any ferrocyanide left in solution, a blue color will appear where the two drops come in contact due to the formation of prussian blue. When all ferrocyanide has been precipitated this color will no longer appear, which indicates the end point of the titration. The zinc sul- phate solution is made by dissolving approximately 5 grm. of pure zinc sulphate in 1 liter of water with the addition of 10 c.c. of sulphuric acid. This solution is standardized with a solution of 10 grm. of potas- sium ferrocyanide (K 4 Fe(CN') 6 .3H 2 0) dissolved in water and diluted to 1 liter. Twenty-five cubic centimeters of the potassium ferrocyanide solution are put into a beaker and titrated with the zinc sulphate solution, the end reaction being determined as above. One cubic centimeter of the ferrocyanide solution is equivalent to 0.0570 grains of cyanogen, from which the value of the zinc solution is calculated. "To test for cyanogen in gas put 15 c.c. of a 10 per cent, ferrous sulphate (FeS0 4 .7H 2 0) solution into each of three wash-bottles. Add 15 c.c. of 20 per cent, caustic soda solution to each bottle and pass about 3 cu. ft. of gas through these bottles at the rate of about 1 cu. ft. per hour. Rinse the contents of the bottles into a beaker, add 20 c.c. more of the caustic soda solution and heat to boiling. Filter and wash with hot water until a few drops of the filtrate no longer show a blue color when acidified and tested with a drop of 1 per cent, ferric chloride solution. Transfer the filtrate to a 500 c.c. graduated flask, dilute to the mark and shake well. Take 100 c.c. of this solution and transfer to a beaker by means of a pipette. Slowly add dilute sul- phuric acid (sp. gr. about 1.5) stirring constantly until the solution becomes slightly acid toward litmus. Then run in the zinc sulphate solution a few drops at a time until the drop test as explained above shows that the ferrocyanide has all been precipitated. From the amount of zinc sulphate solution used the amount of cyanogen in the gas is calculated." ILLUMINATING GAS AND NATURAL GAS 181 12. Specific Gravity. The simplest method of determining the specific gravity of gases makes use of the law that different gases streaming through a given orifice at the same temperature and pressure flow through the orifice at a rate inversely propor- tional to the square root of their specific gravity. Since the time of flow is inversely proportional to the rate, the specific gravity becomes proportional to the square of the time of flow. Bunsen devised an inge- nious instrument to measure specific grav- ities in this way and Schilling later gave it the form which is shown in Fig. 41. It consists of a glass cylinder open at the bottom and fastened to a metal base which keeps it vertical within the larger cylinder of water. It is closed at the top by a metal cap carrying two cocks. A is for the in- troduction of the gas to be tested. B is a three-way cock which in one position dis- charges the gas through a side arm to flush the aparatus. When this cock is in the vertical position the gas passes through a small opening in a platinum plate at C. The apparatus should be standardized against air each time it is used. The calibration is made by opening the air cock and raising the inner cylinder until it is nearly out of water. It will then be filled with air. After closing both cocks it is to be again lowered into the cylinder of water. The observer opens the cock B so that the air streams out through the capillary platinum opening, starts a stop watch as the meniscus passes the lower mark on the glass tube and stops the watch as it passes the upper mark. The instrument is now thoroughly flushed with gas and the time required for the volume of gas between the two calibration marks to stream out of the opening is de- termined in the same way. The calculation then follows from the formula sp.gr. gas^t 2 gas sp. gr, air t 2 air FIG. 41. Schilling's specific gravity appa- ratus. 182 GAS AND FUEL ANALYSIS Edwards 1 has made a careful study of the accuracy of this effusion method of determining gas density and finds that al- though the apparatus may serve well for control work, errors of ten per cent, in the absolute measurement are not unusual. If calibrated orifices are employed, better results may be obtained. The modification of apparatus shown in Fig. 42 is recommended for approximate work and more elaborate models are illustrated where mercury is to be used as the displacing agent. A more accurate determination of the specific gravity of gas is afforded by the specific gravity balance. Edwards 2 has re- viewed the literature of the subject and has devised a portable apparatus capable of giving the specific gravity with an error of less than one part in a thousand. According to Boyle's law the density of a gas is propor- tional to its pressure; and the buoyant force exerted upon a body suspended in a gas is proportional to the density of the gas and, therefore, to its pressure. Hence, if the buoyant force exerted upon a body is made the same when suspended successively in two gases, then the densities of the two gases must be the same at these pressures; or the densities of the two gases at normal pressure are in inverse ratio to the pressures when of equal buoyant force. The Edwards gas density balance is illustrated in Fig. 43 and consists of a balance beam B carrying a sealed cylinder on one end and a counterweight on the other. The balance beam with its sup- port is mounted in a gas-tight chamber to which is attached a mercury manometer. In operation, the balance case and manometer connections are filled with dry air through the inlet / and the pressure adjusted by removing the excess gas through the needle valve E until the beam just balances, as determined by observation (through the adjustable lens L) of the cross line on the end of the beam. After determining this pressure, the 1 Bureau of Standards, Technologic Paper No. 94. 2 Bureau of Standards, Technologic Paper No. 89. FIG. 42. Bureau of Standards form of simple specific gravity apparatus. ILLUMINATING GAS AND NATURAL GAS 183 balance is evacuated through E and filled with the gas, the pressure is then adjusted until the beam is again in equilibrium. The specific gravity of the gas is then the ratio of the total pressure (manometer reading plus atmospheric pressure) required to balance the beam in air to the total pressure required to balance it in the gas. FIG. 43. Edwards specific gravity balance for gases. 13. Natural Gas. Natural gas is ordinarily distributed and used without any attempt at purification. There is, therefore, much less call for analysis of this product. The most important determination is that of heating value, which is carried out in a gas calorimeter as described for illuminating gas in Chapter VII. If the burner of the calorimeter is adjusted for coal gas it will have to be readjusted for the natural gas and a different tip may have to be inserted. The volume of the natural gas should be controlled to give about the same rise in temperature in the calor- imeter as is desirable for coal gas. When a knowledge of the total sulphur is desired it is estimated by the method of 8 of this chapter. A few analyses of natural gas selected from the reports of the Bureau of Mines are given below. It will be noted that no mention is made of unsaturated hydrocarbons, oxygen, carbon monoxide, or hydrogen, which are believed to be entirely absent from American natural gas. The small amounts which may be found in an ordinary analysis are due to air, absorption of gases from water, solubility of the gas in reagents, or error in the explosion or combustion of the hydrocarbons. The hydro- carbons in the above table are reported entirely as methane and ethane. Further discrimination is not possible by ordinary methods of analysis. 184 GAS AND FUEL ANALYSIS ANALYSES OF NATURAL GAS : Charleston, W. Va. Altoona, Pa. Piqua, O. Los Angeles, Cal. CH 4 76 8 90 78 3 59 2 C 2 H 6 22 5 9 12 6 13 9 CO 2 2 2 26 2 N 2 7 8 8 9 7 Calculated heating value at 60 F 1169 1065 1010 841 Calculated sp. gr. air = 1 ... 0.67 0.60 0.66 0.88 * Burrell and Robertson, Bureau of Mines Technical Paper 158 (1917). The combustion of the hydrocarbons of natural gas offers considerable difficulty. Not only is it difficult to burn hydro- carbons with copper oxide but Anderson 1 states that there is difficulty in burning natural gas with oxygen in a combustion pipette if considerable amounts of gasoline vapors are present. In a later article Anderson 2 gives detailed corrections to be ap- plied to the results of combustion. The effect of deviation from the theoretical volume has been discussed in Chapter VI. An- derson advocates the expression of the results of analysis of natural gas in the form of the average number of carbon atoms per molecule of paraffine hydrocarbons. If n is the average num- ber of carbon atoms 3C0 2 2 contraction CO2 The volume V of paraffine hydrocarbons will then be Corrections for deviation from theoretical formulae are worked out as curves and formulae and corrections for calculating specific gravity and heating value from analysis are also given in the paper referred to. 1 /. Ind. & Eng. Chem., 9, 142 (1917). 2 /. Ind. & Eng. Chem., 11, 299 (1919). ILLUMINATING GAS AND NATURAL GAS 185 Burrell and Seibert 1 have developed a method of analysis of gases by fractional distillation at low temperatures. A compari- son of the results of the application of this method to the analysis of a Pittsburgh natural gas as compared with the ordinary analy- sis is given below. It will be noted that there are material differences. COMPOSITION OF NATURAL GAS FROM PITTSBURGH Analysis by ordinary methods Analysis by fractional dis- tillation at low temperature CH 4 79.2 84.7 C 2 H 6 19.6 9 4 C 3 H 8 3 C4Hio (mainly) 1 3 N 2 1.2 1.6 14. Gasoline in Natural Gas. Some gases contain enough gasoline vapors to make it pay to condense them by compression and refrigeration. Burrell 2 reports that the specific gravity of the gas gives good indication of its value for this purpose. Pitts- burgh natural gas with a specific gravity of 0.64 when compared with air, does not yield commercial quantities of gasoline. Gases with specific gravity of 0.95 to 1.60 yield commercially from one to five gallons of 75 to 98 Be. gasoline per thousand cubic feet of gas. Heavy oils of various sorts may also be used as absorbents for gasoline vapors and the process may be successfully applied to gases yielding less than a pint of gasoline per 1000 cu. ft. of gas. Absorption in oil is also used as an analytical method to determine the amount of gasoline in the gas. Dykema 3 illus- trates various types of commercial testing apparatus. A meas- ured quantity of gas is bubbled through a heavy petroleum oil preferably contained in a series of washers. If only a single washer is used the percentage of saturation should be kept be- 1 J. Am. Chem. Soc., 36, 1538 (1914). 2 Bull. 88, U. S. Bureau of Mines. 3 Bureau of Mines Bull. 176. Recent developments in the absorption process for recovering gasoline from natural gas. 1919. 186 GAS AND FUEL ANALYSIS low 4 per cent. In another type of apparatus applicable espe- cially to rich gases a cubic foot tank is filled with the gas and about 850 cc. of absorption oil injected through one of the valves. The tank and contents are then violently agitated for twenty minutes to make sure that the oil has extracted all -of the gaso- line, after which the oil is removed and 800 cc. of the oil is dis- tilled. In any absorption process the amount of gasoline is finally determined by distillation of the absorption oil, using a condenser cooled with ice water. CHAPTER XIII LIQUID FUELS 1. Introduction. The liquid fuel most frequently used is petroleum in either a crude or semi-refined form. Coal tar and tar products rank next in importance. Alcohol may become important in the future. These fuels, which are to be burned directly, are usually blown into the furnace in a fine spray by means of steam or compressed air. The main points to be deter- mined are their heating value, their behavior hi the burner and the relative danger which attends their storage. Fuels which are to be vaporized before combustion, as is the case in internal combustion engines, kerosene lamps, etc., require more elaborate tests. 2. Sampling. The main difficulty in getting a representa- tive sample of liquid fuel is caused by the layer of water and sedi- ment which frequently accumulates on the bottom of a tank of oil or on the surface of one of tar. The main portion of the liquid may also be stratified if various grades have been mixed. The U. S. Bureau of Mines 1 recommends that the oil be sampled as delivered and that either a small stream be run off continuously into a drum from which, after mixing, a smaller sample shall be taken, or that at regular intervals a small dipperful shall be taken from the main stream and placed in a mixing drum. Where it is necessary to sample from a small tank, a proportional sample may be obtained by slowly lowering a glass tube vertically through the oil till the lower end rests on the bottom. If now the upper end be closed by the thumb a column of liquid may be drawn out which represents the composition of the vertical section at the point of sampling. For large tanks the glass tube is replaced by one of tin carrying throughout its length a stiff wire on whose lower end is a tapering cork. When the cork hits the bottom 1 Technical Paper 3, Bureau of Mines. Specifications for the Purchase of Fuel Oil for the Government with Directions for Sampling Oil and Natural Gas. 187 188 GAS AND FUEL ANALYSIS of the tank as the tube is lowered, it is forced up into the tube, sealing the latter so that the sample may be drawn to the surface without leakage. In default of a sampling tube a corked empty bottle with a string tied to the cork may be lashed to a stick and be used. When the bottle has been lowered to the desired depth a pull on the string removes the cork and allows the bottle to fill with oil which can be withdrawn and form part of a com- posite sample. 3. Heating Value. The heating value of liquid fuels may be determined in either the bomb or Parr calorimeter in accordance with the general directions in Chapters XVI and XVII. Some difficulty in combustion will be experienced since all the com- pounds volatilize very rapidly during combustion and there is danger that some of the vapors may break through the flame zone without being completely burned. Incomplete combustion may usually be detected on opening the calorimeter by the odor, and the presence of soot on the inside of the cover. The difficulty becomes greater with volatile liquids like gasoline or alcohol, both because of their greater volatility in the calorimeter and because of the difficulty of weighing the sample accurately. Slightly volatile liquids such as tar and heavy petroleum oils may be weighed directly into the capsule of the bomb calorimeter and burned completely, if oxygen under 25 atmospheres pressure is used. It is advisable to place on the oil a small weighed pellet of sugar or benzoic acid to start combustion. More volatile liquids must be weighed in thin- walled bulbs of about 0.5 c.c. capacity with capillary necks which the analyst may blow for himself out of fine glass tubing. These are filled by warming the weighed bulb and immersing the open neck in the liquid. Contraction of the air in the bulb draws up a little of the liquid and by repetition of the process the bulb may be filled. The capil- lary neck may then be sealed close to the bulb with a small blow- pipe flame. The increase in weight of the bulb plus the portion of the neck fused off gives the weight of oil in the sample. The sealed bulb is placed on the capsule of the calorimeter and around it is piled about 0.25 grm. of sugar or benzoic acid in contact with the fuse wire. The combustion of this material breaks the bulb and ignites the contents. Richards and Jesse 1 have shown that 1 Jour. Am. Chem. Soc., 32, 268 (1910). LIQUID FUELS 189 . even this method fails to give complete combustion with volatile liquids like benzene. They recommend the following procedure as successful. "The benzene in a very thin glass bulb was placed in the bottom of a narrow platinum crucible, 2 cm. in diameter and 2.5 cm. high. A few millimeters above the bulb was fixed a small platform of thin glass bearing a weighed quantity of powdered sugar. The passage of a cur- rent through the coil of iron wire ignited the sugar, which in its turn burst the bulb and ignited the benzene at a moment when the whole top of the narrow crucible was filled with flame from the burning sugar. Thus none of the benzene vapor could escape ignition. The trouble with the old method had been that the larger crucible was too wide. Moreover, the sugar had been beneath the benzene instead of above it, so that some of the benzene escaped unconsumed. The amount which thus escaped was greater when there was more nitrogen present than when there was less. Obviously, with non-volatile compounds like sugar the width of the crucible would make no difference." Gelatine capsules such as used by pharmacists have also been recommended as containers for volatile oils, but their moisture content changes so rapidly in the air that it is difficult to keep constant the necessary correction factor for the gelatine. The heating value of oils may also be determined in the Parr calorimeter, which is described in Chapter XVII. Non-volatile oils may be weighed directly into the calorimeter which already contains the peroxide mixture, and mixed thoroughly with the charge by means of a stiff wire. Volatile liquids may be placed in the calorimeter in a thin-walled glass bulb as directed for the bomb calorimeter, and the charge of chemicals placed upon it. The calorimeter is then closed with the cap provided and shaken violently until the bulb is broken and the oil is mixed with the peroxide. A correction in addition to those specified in Chapter XVII must be deducted for the heat liberated by reaction between the peroxide and the glass of the bulb. Professor Parr gives this correction as 0.017 C. per 0.1 grm. glass. The weight of oil taken should be about 0.3 grm. and the charge 10 grm. of Na 2 2 and 1 grm. of KC1O 3 . The use of 0.2 grm. benzoic acid is also advantageous. Care must be taken that crude petroleum and tars do not carry much emulsified water, for the water reacts with the peroxide with evolution of heat. In ex- 190 GAS AND FUEL ANALYSIS treme cases a violent explosion may take place, wrecking the calorimeter. If the oil is of such a type that it may be burned without smoke in a burner without a wick and there is at least a pint of the oil available, the most convenient method of determining heat value is in a calorimeter of the type whose use for determining the heating value of gases is described in Chapter VII. The apparatus as modified for liquids requires a suitable burner for the oil which hangs upon a balance during the determination as FIG. 44. Calorimeter for heating value of oils. shown in Fig. 44. The lamp as shown in the illustration requires 150-200 c.c. of oil. To start the lamp the cup L is filled with alcohol which is lighted to preheat the burner head, n. When the alcohol is nearly burned away, air-pressure is placed upon the liquid by a hand pump connected to m. The oil rises in the bur- ner, vaporizes and ignites in the alcohol flame. The pumping is continued until a freely-burning blue flame results when the pump is disconnected. After the water is flowing normally through the calorimeter, the lighted lamp is inserted and centered in the com- bustion space. When equilibrium has been reached, the balance 12 LIQUID FUELS 191 is brought to zero by proper adjustment of weights in the pan and the experiment started. After a definite weight of oil, usually 5 or 10 grm., has been burned, the experiment is inter- rupted and from the rise in temperature and the weight of water heated the heating value of the fuel may be calculated. The details and precautions are in general the same as given for the gas in Chapter VII. Especial care must be taken that the flame does not impinge directly against the metal of the calor- imeter since incomplete combustion will result from the sudden cooling of the gases while combustion is still in progress. 4. Specific Gravity. The specific gravity of petroleum products ,is less than 1 and is usually reported on the Baume scale f for liquids lighter than water. Tar is usually heavier than water and its specific gravity is reported directly. If a sufficient quantity of material is available and it is not too viscous, the specific gravity may be determined with approximate accuracy by a hydrometer spindle. If greater accuracy is required or if only a small sample is available a pycnometer or Westphal balance must be used. If a specific gravity on water-free material is demanded, the oil must be put into a flask without the addition of any diluent and distilled slowly till the water is off. The oil distilled is then separated from the water and returned to the residue in the still after it has cooled. A comparison of the Baume* scale for liquids lighter than water and the corresponding specific gravities is given in Table XIII of the Appendix. 5. Moisture. The various methods for the determination of water hi petroleum have been carefully examined by Allen and Jacobs. 1 They recommend a method which involves the measurement of the hydrogen evolved by the action of the water on metallic sodium and also the method of distillation, either with or without the addition of water-saturated toluene or xylene. The latter method will probably give the better results in inexperienced hands. It may be used for tar as well as petroleum products. The toluene is added to diminish the viscosity of the mass and lessen the danger of foaming and bumping. Instead of toluene, xylene or petroleum benzine with a boiling-point of 110 to 150 C. may be used. The diluent must, however, be first shaken with water and then allowed to stand until perfectly clear in order that it may not dissolve technical Paper 25, U. S. Bureau of Mines, 1912. 192 GAS AND FUEL ANALYSIS any of the water of the sample. The sample of about 100 grin, is weighed into a distilling flask holding at least 500 c.c. and to it is added a roughly measured volume of 100 c.c. of the diluent, or 200 c.c. if the sample is very viscous. The distillation is started slowly and continued until the distillate no longer comes over turbid and approximately as much oil has distilled as was added as a diluent. The distillate is caught in a graduated cylinder and the volume of water read directly after sufficient time has been given for it to settle by gravity. Allen and Jacobs state that the method may be made accurate to approximately 0.033 grm. water for each 100 c.c. of benzene and oil in the distillate. The details of a similar method for the determination of water in tar as used in the laboratories of the Barret Manufac- turing Company have been published by S. R. Church. 1 He specifies exactly the dimensions of the still, the manner of placing the thermometer and other details. The distillation is to be continued until the thermometer in the vapor has reached 205 C. He recommends a convenient form of graduated separatory funnel for the distillate and states that a clean separation of the oil and water can be obtained if 25 c.c. of benzene is introduced into the separatory funnel before the distillation is started. 6. Proximate Analysis. A proximate analysis in the sense in which it is used in coal analysis is not often made on liquid fuels because they are so largely volatile that the test has little meaning. The ash gives some measure of suspended earthy solids and in the case of tar, the fixed carbon gives an indication of the amount of "free carbon" in the tar. It is necessary to modify the standard method for volatile matter in coal by heating the crucible gently until all foaming has stopped. 7. Suspended Solids. Suspended solids which in the case of crude petroleums usually are earthy matters and in the case of tars are fine particles of coke forming the so-called "free carbon" are separated by filtration and washing. The oil or tar is first filtered through a 30- or 40-mesh sieve to remove coarse foreign bodies accidentally included. A weighed sample of 5 or 10 grm. is then diluted with pure benzene or toluene until it will filter readily. The solution is filtered through a * Jour, Ind. and Eng. Chem,. 3, 228 (1911). LIQUID FUELS 193 pair of weighed heavy filter papers or through a Gooch funnel and the filter washed with more of the warm solvent until the extraction is complete. The filter is then dried at 105 C. The increase in weight gives suspended solids. If there is much 'water in the liquid being examined it may be retained on the filter in the form of drops during the first filtration. This water may be driven off by gentle heating and the extraction of soluble material then continued. 8. Flash Point. The flash point of an oil indicates the temperature at which the oil gives off com- bustible vapors with sufficient rapidity to form an explosive mix- ture with the air above it. The flash point will depend upon the rate of heating the oil, the volume of air above it, the rapidity with which the air is replaced and many other variables. It is evi- dent that the conditions must be closely specified in order that the results may be of value. The figure is of great importance with kerosene oil and most of the states have definite and for the most part different regulations on the subject. The older forms of apparatus had open cups, but the more modern FlG 45 ._ Tag c i ose d tester for forms have closed cups. The flash point of oils. American Society for Testing Ma- terials has adopted as its standard the Tag Closed Tester illustrated in Fig. 45. The United States Fuel Administration has adopted the same instrument for testing the flash-point of kerosene. The method of determining flash point is as follows: 1 1. (a) Flash point shall be determined with the Tag Closed Tester, operated in accordance with the directions given below. 1 Proceedings Am. Soc. Test. Mat., 18, 685 (1918). 13 194 GAS AND FUEL ANALYSIS (b) For unofficial tests any suitable closed type of tester such as the Abel, the Abel-Pensky or the Elliott may be used. 2. (a) If gas is available, connect a ^-in. rubber tube to the corru- gated gas connection on the oil cup cover. If no gas is available, un- screw the test flame burner-tip from the oil chamber on the cover, and insert a wick of cotton cord in the burner-tip and replace it. Put a small quantity of cotton waste in the oil chamber, and insert a small quantity of signal, sperm or lard oil in the chamber, light the wick and adjust the flame, so that it is exactly the size of the small white bead mounted on the top of the tester. (6) The test shall be performed in a dim light so as to see the flash plainly. (c) Surround the tester on three sides with an inclosure to keep away draughts. A shield about 18 in. square and 2 ft. high, open in front, is satisfactory, but any safe precaution against all possible room draughts is acceptable. Tests made in a laboratory hood or near ventilators will give unreliable results. (d) See that the tester sets firm and level. (e) For accuracy, the flash-point thermometers which are especially designed for the instrument should be used, as the position of the bulb of the thermometer in the oil cup is essential. 3. Put the water-bath thermometers which are especially designed for the instrument in place, and place a receptacle under the overflow spout to catch the overflow. Fill the water bath with water at such a temper- ature that, when testing is started, the temperature of the water bath will be at least 10 C. below the probable flash point of the oil to be tested. 4. Put the oil cup in place in the water bath. Measure 50 c.c. of the oil to be tested in a pipette or a graduate, and place in the oil cup. The temperature of the oil shall be at least 10 C. below its probable flash point when testing is started. Destroy any bubbles on the surface of the oil. Put on the cover, with flash-point thermometer in place and gas tube attached. Light the pilot light on the cover and adjust the flame to the size of the small white bead on the cover. 5. Light and place the heating lamp, filled with alcohol, in the base of the tester and see that it is centrally located. Adjust the flame of the alcohol lamp so that the temperature of the oil in the cup rises at the rate of about 1 C. per minute, not faster than 1.1 nor slower than 0.9 per minute. 6. (a) Record the barometric pressure which, in the absence of a laboratory instrument, may be obtained from the nearest Weather Bureau Station. (b) Record the temperature of the oil sample at start. (c) When the temperature of the oil reaches about 5 C. below the LIQUID FUELS 195 probable flash point of the oil, turn the knob on the cover so as to intro- duce the test flame into the cup, and turn it promptly back again. Do not let it snap back. The time consumed in turning the knob down and back should be about one full second, or the time required to pro- nounce distinctly the words "one-thousand-and-one." (d) Record the time of making the first introduction of the test flame. (e) Record the temperature of the oil sample at the time of the first test. (/) Repeat the application of the test flame at every 0.5 C. rise in temperature of the oil until there is a flash of the oil within the cup. Do not be misled by an enlargement of the test flame or halo around it when entered into the cup, or by slight flickering of the flame; the true flash consumes the gas in the top of the cup and causes a very slight explosion. (g] Record the time at which the flash point is reached. (h) Record .the flash point. (i) If the rise in temperature of the oil, from the "time of making the first introduction of the test flame" to the "time at which the flash point is reached" was faster than 1.1 or slower than 0.9 C. per minute the test should be questioned, and the alcohol heating lamp adjusted so as to correct the- rate of heating. It will be found that the wick of this lamp can be so accurately adjusted as to give a uniform rate of rise in temperature of 1 C. per minute and remain so. 7. (a) It is not necessary to turn off the test flame with the small regulating valve on the cover; leave it adjusted to give the proper size of flame. (b) Having completed the preliminary test, remove the heating lamp, lift up the oil cup cover, and wipe off the thermometer bulb. Lift out the oil cup, and empty and carefully wipe it. Throw away all oil samples after once used in making a test. (c) Pour cold water into the water bath, allowing it to overflow into a receptacle, until the temperature of the water in the bath is lowered to 8 C. below the flash point of the oil, as shown by the previous test. With cold water of nearly constant temperature, it will be found that a uniform amount will be required to reduce the temperature of the water bath to the required point. (d) Place the oil cup back in the bath and measure into it a 50-c.c. charge of fresh oil. Destroy any bubbles on the surface of the oil, put on the cover with its thermometer, put in the heating lamp, record the temperature of the oil, and proceed to repeat the test as described above in Sections 4 to 6, inclusive. Introduce the test flame for first time at a temperature of 5 C. below the flash point obtained on the previous test. 196 GAS AND FUEL ANALYSIS 8. If two or more determinations agree within 0.5 C., the average of these results, corrected for barometric pressure, shall be considered the flash point. If two determinations do not check within 0.5 C., a third determination shall be made and if the maximum variation of the three tests is not greater than 1 C., their average, after correcting for barometric pressure, shall be considered the flash point. 9. A correction table furnished with each instrument, for converting the results of tests made at varying barometric pressures to equivalent temperatures at the standard barometric pressure of 760 mm. 9. Gasoline. Gasoline is defined in Webster's new Inter- national Dictionary as "a volatile inflammable liquid used as a solvent for oils, fats, etc., as a carburetant, and to produce heat and motive power." The gasoline of commerce has changed its composition markedly within recent years. Originally one of the volatile fractions obtained by "straight" distillation of petroleum, it has changed to be largely a product of destructive distillation with progressively lower Baum gravity and higher average boiling point. Rather heavy oils are made to flash more readily through mixture with the very volatile casing-head gasolines recovered from natural gas. The definition given above is broad enough to cover benzene, and other products derived from destructive distillation of coal, as well as alcohol and other products from destructive distillation of wood or fer- mentation of grain or molasses. It is obvious that with such a wide range of chemical composi- tion, only the broadest tests can be applied. The determination of flash point is unnecessary because all gasolines flash at ordi- nary temperatures. Heating value is of some importance with the mixed products for, as shown by Table IX of the Appendix, the heating values may vary widely. The following figures will illustrate this: HEATING VALUE OF LIQUID FUELS Name Formula Sp. gr. Heating value Per pound Per gallon Pentane CsHi 2 CeHi4 C 6 H 6 GH 3 OH C,H R OH 0.6273 0.6640 0.8846 0.8027 0.7946 21,177 20,914 18,447 10,250 13.325 110,670 115,620 135,950 68,540 88.200 Hexane Benzene Methyl alcohol Ethvl alcohol.. LIQUID FUELS 197 Since liquid fuels are usually sold by volume it is the heat units in a gallon which are important. On this basis benzene is seen to be distinctly the best and methyl alcohol the poorest. However, other considerations than heating value enter into the efficiency with which these fuels are used in an internal combus- tion engine. Benzene with its high carbon content tends to form free carbon in the engine cylinder which cuts down its efficiency. Alcohol with its higher oxygen content is free from this trouble and in admixture with hydrocarbons lessens forma- tion of free carbon. Methyl and ethyl alcohol may both be used with higher compressions than gasoline without preignition, so that a gallon of ethyl alcohol is, in a suitably designed engine, practically the equivalent of a gallon of gasoline. The best test of a motor spirit is an actual operating test in an engine similar to that in which it is to be used. The distillation test is the best laboratory guide in predicting how an oil will behave in the carburetor and cylinder. 10. Specifications for Motor Gasoline. The committee on Standardization of Petroleum Specifications appointed by the United States Fuel Administration has adopted specifications effective November 25, 1919, of which the following is a copy. Quality. Gasoline to be high grade, refined, and free from water and all impurities, and shall have a vapor tension not greater than ten pounds per square inch at 100 F. temperature, same to be determined in ac- cordance with the current "Rules and Regulations for the transporta- tion of explosives and other dangerous articles by freight," as issued by the Interstate Commerce Commission. Inspection. Before acceptance the gasoline will be inspected. Sam- ples of each lot will be taken at random. These samples immediately after drawing will be retained in a clean, absolutely tight closed vessel and a sample for test taken from the mixture in this vessel directly into the test vessel. Specifications. (a) Boiling point must not be higher than 60 C. (140 F.). (6) 20 per cent, of the sample must distill below 105 C. (221 F.). (c) 50 per cent, must distill below 140 C. (284 F.). (0 90 per cent, must distill below 190 C. (374 F.). (e) The end or dry point of distillation must not be higher than 225 C. (437 F.). 198 GAS AND FUEL ANALYSIS (/) Not less than 95 per cent, of the liquid will be recovered in the receiver from the distillation. Test. One hundred cubic centimeters will be taken as a test sample. The apparatus and method of conducting the distillation test shall be that adopted by Sub-Committee XI of Committee D-l of the American Society for Testing Materials 1 (as shown in Fig. 46), with the following modifications : FIG. 46. Apparatus for distillation of petroleum. First: The temperature shall be read against fixed percentage points, and, second: the thermometer shall be as hereinafter described: Flask. The flask used shall be the standard 100 c.c. Engler Flask, described in the various textbooks on petroleum. Dimensions are as follows: Dimensions Cm. Outside diameter of bulb 6.5 Outside diameter of neck 1.6 Length of neck 15 . Length of vapor tube 10 . Outside Diameter of vapor tube 0.6 Inches 2.56 0.63 5.91 3.94 0.24 1 American Society for Testing Materials, Year Book for 1915, pp. 568- 569; or pt. 1, Committee Reports, 1916, vol. 16, pp. 518-521. See alsc Bureau of Mines Technical Papers Nos. 166 and 214. LIQUID FUELS 199 Position of vapor tube, 9 cm. (3.55 in.) above the surface of the gaso- line when the flask contains its charge of 100 c.c. The tube is approxi- mately in the middle of the neck. The observance of the prescribed dimensions is considered essential to the attainment of uniformity of results. The flask shall be supported on a ring of asbestos having a circular opening \Y in. in diameter; this means that only this limited portion of the flask is to be heated. The use of wire gauze is forbidden. Condenser. The condenser shall consist of a thin walled tube of metal (brass or copper) K in. internal diameter and 22 in. long. It shall be set at an angle of 75 from the perpendicular and shall be sur- rounded with a cooling jacket of the trough type. The lower end of the condenser shall be cut off at an acute angle and shall be curved down for a length of 3 inches. The condenser jacket shall be 15 in. long. Thermometer. The thermometer shall be made of selected enamel- backed tubing having a diameter between 5.5 and 7 mm. The bulb shall be of Jena normal or Corning normal glass, its diameter shall be less than that of the stem and its length between 10 and 15 mm. The total length of the thermometer shall be approximately 380 mm. The range shall cover C. (32 F.) to 270 C. (518 F.) with the length of the graduated portion between the limits of 210 to 250 mm. The point marking a temperature of 35 C. (95 F.) shall not be less than 100 inm. nor more than 120 mm. from the top of the bulb. For commercial use the thermometer may be graduated in the Fahrenheit scale. The scale shall be graduated for total immersion. The accuracy must be within about 0.5 C. The space above the meniscus must be filled with an inert gas, such as nitrogen, and the stem and bulb must be thoroughly aged and annealed before being graduated. Source of Heat in Gasoline Distillation. The source of heat in dis- tilling gasoline may be a gas burner, an alcohol lamp, or an electric heater. PROCEDURE AND DETAILS OF MANIPULATION IN CONDUCTING DISTILLATIONS 1. If an electric heater is used it is started first to warm it. 2. The condenser box is filled with water containing a liberal portion of cracked ice. 3. The charge of gasoline is measured into the clean, dry Engler flask from a 100 c.c. graduate. The graduate is used as a receiver for distillates without any drying. This procedure eliminates errors due to incorrect scaling of graduates and also avoids the creation of an 200 GAS AND FUEL ANALYSIS apparent distillation loss due to the impossibility of draining the gasoline entirely from the graduate. 4. The above-mentioned graduate is placed under the lower end of the condenser tube so that the latter extends downward below the top of the graduate at least 1 in. The condenser tube should be so shaped and bent that the tip can touch the wall of the graduate on the side adjacent to the condenser box. This detail permits distillates to run down the side of the graduate and avoids disturbance of the meniscus caused by the falling of drops. The graduate is moved occasionally to permit the operator to ascertain that the speed of distillation is right, as indicated by the rate at which drops fall. The proper rate is from 4 c.c. to 5 c.c. per minute, which is approximately two drops a second. The top of the graduate is covered, preferably by several thicknesses of filter paper, the condenser tube passing through a snugly fitting opening. This minimizes evaporation losses due to circulation of air through the graduate and also excludes any water that may drip down the outside of the condenser tube on account of condensation on the ice-cooled condenser box. 5. A boiling stone (a bit of unglazed porcelain or other porous mate- rial) is dropped into the gasoline in the Engler flask. The thermometer is equipped with a well-fitted cork and its bulb covered with a thin film of absorbent cotton (preferably the long-fibered variety sold for surgical dressing). The quantity of cotton used shall be not less than 0.005 nor more than 0.010 g. (5 to 10 milligrams). The thermometer is fitted into the flask with the bulb just below the lower level of the side neck opening. The flask is connected with the condenser tube. 6. Heat must be so applied that the first drop of the gasoline falls from the end of the condenser tube in not less than five or more than ten minutes. The initial boiling point is the temperature shown by the thermometer when the first drop falls from the end of the condenser tube into the graduate. The operator should not allow himself to be deceived as sometimes (if the condenser tube is not dried from a previous run) a drop will be obtained and it will be sometime before a second one falls; in this case the first drop should be ignored. The amount of heat is then increased so that the distillation proceeds at a rate of from 4 c.c. to 5 c.c. per minute. The thermometer is read as each of the selected percentage marks is reached. The maximum boiling point or dry point is determined by continuing the heating after the flask bottom has boiled dry until the column of mercury reaches a maximum and then starts to recede consistently. 7. Distillation loss is determined as follows: The condenser tube is allowed to drain for at least five minutes after heat is shut off, and a LIQUID FUELS 201 final reading taken of the quantity of distillate collected in the receiving graduate. The distillation flask is removed from the condenser and thoroughly cooled as soon as it can be handled. The condensed residue is poured into a small graduate or graduated test tube and its volume measured. The sum of its volume and the volume collected in the receiving graduate, subtracted from 100 c.c. gives the figure for dis- tillation loss. 11. Kerosene. Kerosene is usually ranked as an illuminating oil rather than as a fuel oil and tests are framed to determine its suitability for use in lamps. The most important test for this purpose is the flash test which insures safety from explosion caused by accumulation of combustible vapors in the bowl of the lamp. Various specifications have been written for kerosene but the recommendations of the Committee on Standardization of Petroleum Specifications 1 will probably supplant the others. The tests and specifications for water-white kerosene are quoted below. Tests and specifications for long-time burning oil, 300 degree mineral seal oil and signal oil are also included in the same bulletin. WATER- WHITE KEROSENE. METHODS OF TEST Flash. To be taken on the Tag closed cup, A. S. T. M. standard; oil to be heated at the rate of 2 F. per minute; test flame to be applied every 2, commencing at 105 F. Color. To be determined on the Saybolt colorimeter or its equivalent. Sulphur. Test to be made by burning at least 2 grams of the oil in a small flask and absorbing the gases of combustion in a standard solution of Sodium Carbonate and titrating the excess of Sodium Car- bonate with the standard solution of Sulphuric Acid. Floe. Directions for making test: Take a hemispherical iron dish, and place a small layer of sand in the bottom. Take a 500 c.c. Florence or Erlenmeyer flask and into it put 300 c.c. of the oil (after filtering if it contains suspended matter). Suspend a thermometer in the oil by means of a cork slotted on the side. Place flask containing the oil in the sand bath, and heat bath so that the oil has reached a tempera- ture of 240 F. at the end of one hour. Hold oil at temperature of not less than 240 F. nor more than 250 F. for six hours. The oil may become discolored but there should be no suspended matter formed in 1 Bulletin No. 2, U. S. Fuel Administration, 1918. 202 GAS AND FUEL ANALYSIS the oil. The flask should be given a slight rotary motion and if there is a trace of "floe" it can be seen to rise from the Center of the bottom. Distillation Test. The oil shall all distill below temperature of 600 F. The test is made as described by the Bureau of Mines, Technical Papers 166, using A. S. T. M. apparatus with wet bulb and total immersion thermometer. Cloud Test. Directions for making test: Take a 4-ounce oil sample bottle and introduce therein 1)^ ounces of the oil to be tested; insert cork with cold-test thermometer so that thermometer is suspended in the oil. Place bottle in a freezing mixture and cool to F. Keep oil cooled to this temperature for 10 minutes. Bottle should be given a rotary motion occasionally so as not to supercool the sides. The oil should not be clouded from crystals of paraffin wax at the end of 10 minutes. Reaction. Two ounces of the oil should be shaken with one-half ounce of warm neutral distilled water and allowed to cool and separate. The water when separated shall react neutral to methyl-orange and phenol-phthalein. Burning Test. The oil must burn freely and steadily in a lamp fitted with a No. 1 sun hinge burner. It must give a good flame for a period of 18 hours without smoking or forming "ears" or "toad-stools" on the wick. The chimney must be only slightly clouded or stained at the end of the test. SUMMARY OF SPECIFICATIONS WATER-WHITE KEROSENE Appearance. Oil must be free from water, glue, and suspended matter. Flash. Not less than 115 F., Tag closed cup, A. S. T. M. standard. Color. To be 21 color on Saybolt colorimeter or its equivalent on a Lovibond tintometer, these being equal to color of a solution of Potassium Bichromate containing 0.0048 grams per liter. Sulphur. Not more than 0.06 per cent. Floe. Oil to be free from floe. Distillation. Oil to distill below temperature of 600 F. Cloud Test. Oil should not show cloud at F. Reaction. Must be neither acid nor alkaline. Burning Test. As stated above. 12. Fuel Oil. Tentative regulations for the storage and use of the fuel oil were adopted by the Committee on Inflammable Liquids of the National Fire Protection Association in 1919. 1 1 Chemical and Metallurgical Engineering 21, 781 (1919). LIQUID FUELS 203 This committee defined oil-burning equipments as those using only liquids having a flash point above 150 F. in a closed cup tester. Specifications for fuel oil have not been standardized. In addition to flash point a distillation test is sometimes speci- fied, since a mixed product of a heavy residuum and a light naph- tha does not behave so well in the burners as a straight product. The oil should be free from water and excessive amounts of sul- phur. The latter may be determined by fusion with sodium peroxide according to the method given for coal in Chapter XV. CHAPTER XIV SAMPLING COAL 1. General Consideration. However accurate an analysis of coal may be, the results are of little value and are often worse than useless if the sample submitted to the analyst is not a repre- sentative one. Elaborate methods have been worked out for sampling gold and silver ores but cost precludes the application of anything but the simplest methods to coal. It is manifest that it is unwise to spend ten cents a ton to determine whether the price is two cents a ton too high. If we assume a shipment of a single car of coal weighing forty tons, which must be sampled and analyzed by itself, it will be seen that a charge of eight dollars for this service will add twenty cents per ton to the price of coaL This is nearly 10 per cent, of the cost of the coal at the mouth of the mine and is an expense which is hardly justifiable. However if the test is worth making at all it must be on a sample which has fair claim to representativeness. The coal sampler stands eternally between the devil of inadequateness and the deep sea of excessive cost. In large plants where the coal is immediately crushed and removed to storage bins by conveyors a representative sample may readily be obtained. In most cases, however, the coal must be sampled as it comes from the car and the problem is more difficult. To many people coal is black and all that is black is coal. The more careful observer may detect bits of slate, and streaks or nodules of the brassy looking pyrites. The chemist knows that in addition the fine particles which have crushed because of their greater friability differ in composition from the lump coal and are usually higher in ash, though sometimes the reverse is the case, and that coal is far from being a mass of uniform composition. 204 SAMPLING COAL 205 2. Difference in Composition of Lump and Fine Coal. The following tests taken from the author's record of cooperative tests undertaken jointly by the University of Michigan Gas Ex- periment Station and the United States Bureau of Mines to determine the availability of various coals for gas manufacture show some interesting variations. The coals had mostly been shipped in small lots in sacks and were therefore considerably cru&hed in transit. They were screened in lots of about 600 Ib. on a three-quarter inch bar screen preparatory to gas tests and the screenings and lump coal were separately sampled. The sampling of the fine coal presented no difficulty since it was already in small lumps and could be crushed as fine as de- sired. The lump coal could not be finely crushed without detriment to the gas tests and so it was sampled by breaking the large lumps and then taking about two scoopf uls which were crushed and sampled as usual. Of the eleven coals tested in this manner four, one each from West Virginia, Colorado, New Mexico and Wyoming showed very little difference between the lumps and screenings. One coal showed decidedly less ash in the screenings than in the lump coal. Six coals showed notice- able and in some cases notable increases of ash and sulphur in screenings with corresponding decreases of heating value, as shown by the following analyses of the coals calculated to a dry basis. In the coal from Hellier, Kentucky for which there are three tests, the average heating value of the screenings is 1080 B.t.u. lower than that of the lump. This is entirely due to difference in ash as shown by the figures for heating value figured to coal dry and free from ash, where the difference disappears, the average heating value of the screenings being only 6 B.t.u. below that of the lump. The same thing is true of the other coals of the list the variation in heating value of lump and screenings disappears al- most completely when calculated to a moisture and ash-free basis. Sulphur is never lower in the screenings than in the lump and is in some cases nearly twice as high. The last coal in the above table differs from all the others in that the screenings are much lower in ash than the lump. It might be thought that the sample had been labelled incorrectly 206 GAS AND FUEL ANALYSIS 12 u PQ tn QQ a .s S 8 PH S rH O O O O i- rH C COC4. Solutions and Reagents. Barium Chloride. Dissolve 100 g. of BaCl 2 .2H 2 in 1000 c.c. of distilled water. Saturated Bromine Water. Add an excess of bromine to 1000 c.c. of distilled water. Eschka Mixture. Thoroughly mix 2 parts (by weight) of light cal- cined MgO and 1 part of anhydrous Na2COa. Both materials should be as free as possible from sulphur. Methyl Orange. Dissolve 0.02 g. in 100 c.c. of hot distilled water and filter. Hydrochloric Acid. Mix 500 c.c. of HC1, sp. gr. 1.20, and 500 c.c. of distilled water. Normal Hydrochloric Acid. Dilute 80 c.c. of HC1, sp. gr. 1.20, to 1 liter with distilled water. Sodium Carbonate. A saturated solution, approximately 60 g. of crystallized or 22 g. of anhydrous Na2COa in 100 c.c. of distilled water. Sodium-hydroxide Solution. Dissolve 100 g. in 1 liter of distilled water. This solution may be used in place of the Na2COs solution. Method. Preparation of Sample and Mixture. Thoroughly mix on glazed paper 1 g. of coal and 3 g. of Eschka mixture. Transfer to a No. 1 porcelain capsule, 1 in. deep and 2 in. in diameter, or a No. 1 crucible or a platinum crucible of similar size, and cover with about 1 g. of Eschka mixture. Ignition. On account of the amount of sulphur contained in artificial gas, the crucible shall be heated over an alcohol, gasoline or natural gas flame as in procedure (a) below, or in a gas or electrically heated muffle, as in procedure (b) below. The use of artificial gas for heating the coal and Eschka mixture is permissible only when the crucibles are heated in a muffle. (a) Heat the crucible, placed in a slanting position on a triangle, over a very low flame to avoid rapid expulsion of the volatile matter, which tends to prevent complete absorption of the products of com- bustion of sulphur. Heat the crucible slowly for 30 minutes, gradually increasing the temperature and stirring after all black particles have disappeared, which is an indication of the completeness of the procedure. THE CHEMICAL ANALYSIS OF COAL 249 (6) Place the crucible in a cold muffle and gradually raise the tem- perature to 870-925 C. (cherry-red heat) in about 1 hour. Maintain the maximum temperature for about \% hours and then allow the crucible to cool in the muffle. Subsequent Treatment. Remove and empty the contents into a 200- c.c. beaker and digest with 100 c.c. of hot water for H to % hour, with occasional stirring. Filter and wash the insoluble matter by decanta- tion. After several washings in this manner, transfer the insoluble matter to the filter and wash 5 times, keeping the mixture well agitated. Treat the filtrate amounting to about 250 c.c., with 10 to 20 c.c. of satu- rated bromine water, make slightly acid with HC1 and boil to expel the liberated bromine. Make just neutral to methyl orange with NaOH or Na 2 CO 3 solution, then add 1 c.c. of normal HC1. Boil again and add slowly from a pipette, with constant stirring, 10 c.c. of a 10 per cent, solution of BaCl 2 .2H 2 0. Continue boiling for 15 minutes and allow to stand for at least 2 hours, or preferably over night, at a tem- perature just below boiling. Filter through an ashless filter paper and wash with hot distilled water until a AgN0 3 solution shows no precipi- tate with a drop of the filtrate. Place the wet filter containing the precipitate of BaSC>4 in a weighed platinum, porcelain, silica or alundum crucible, allowing a free access of air by folding the paper over the pre- cipitate loosely to prevent spattering. Smoke the paper off gradually and at no time allow it to burn with a flame. After the paper is prac- tically consumed, raise the temperature to approximately 925 C. and heat to constant weight. The residue of MgO, etc., after leaching, should be dissolved in HC1 and tested with great care for sulphur. When an appreciable amount is found this should be determined quantitatively. The amount of sulphur retained is by no means a negligible quantity. Blanks and Corrections. In all cases a correction must be applied either (1) by running a blank exactly as described above, using the same amount of all reagents that were employed in the regular deter- mination, or more surely (2) by determining a known amount of sul- phate added to a solution of the reagents after these have been put through the prescribed series of operations. If this latter procedure is adopted and carried out, say, once a week or whenever a new supply of a reagent must be used, and for a series of solutions covering the range of sulphur content likely to be met with in coals, it is only necessary to add to or subtract deficiency or excess may have been found in the appropriate "check" in order to obtain a result that is more certain to be correct than if a " blank" correction as determined by the former procedure is applied. This is due to the fact that the solubility error 250 GAS AND FUEL ANALYSIS for BaSCh, for the amounts of sulphur in question and the conditions of precipitation prescribed, is probably the largest one to be considered. BaS(>4 is soluble in acids and even in pure water, and the solubility limit is reached almost immediately on contact with the solvent. Hence, in the event of using reagents of very superior quality or of exercising more than ordinary precautions, there may be no apparent "blank," because the solubility limit of the solution for BaS0 4 has not been reached or at any rate not exceeded. As shown in the preliminary report, the Atkinson and sodium- peroxide methods give results in close agreement with theEschka method. Regester has shown that if 5 per cent, of nitrogen is present in the gases contained in the bomb calorimeter the sulphur of a coal is almost completely oxidized to H 2 S04 and the washings of the calori- meter may be used for the determination of sulphur. The permissible differences in duplicate determinations are as follows: Same analyst, Different analysts, per cent. per cent. Sulphur under 2 per cent . 05 0.10 Sulphur over 2 per cent 0. 10 0. 20 DETERMINATION OF PHOSPHORUS IN ASH Method No. 1. To Cover All Cases. To the ash from 5 g. of coal in a platinum capsule is added 10 c.c. of HN0 3 and 3 to 5 c.c. of HF. The liquid is evaporated and the residue fused with 3 g. of Na 2 C03. If unburned carbon is present 0.2 g. of NaN0 3 is mixed with carbonate. The melt is leached with water and the solution filtered. The residue is ignited, fused with Na 2 C0 3 alone, the melt leached and the solution filtered. The combined filtrates, held in a flask, are just acidified with HN0 3 and concentrated to a volume of 100 c.c. To the solution, brought to a temperature of 85 C., is added 50 c.c. of molybdate solution and the flask is shaken for 10 minutes. If the precipitate does not form promptly and subside rapidly, add enough NH 4 N0 3 to cause it to do so. The precipitate is washed six times, or until free from acid, with a 2-per-cent. solution of KN0 3 , then returned to the flask and titrated with standard NaOH solution. The alkali solution may well be made equal to 0.00025 g. phosphorus per cubic centimeter, or 0.005 per cent, for a 5-g. sample of coal, and is 0.995 of one-fifth normal. Or the phosphorus in the precipitate is determined by reduc- tion and titration of the molybdenum with permanganate. Note on Method 1. The advantage of the use of HF in the initial attack of the ash lies in the resulting removal of silica. Fusion with THE CHEMICAL ANALYSIS OF COAL 251 alkali carbonate is necessary for the elimination of titanium, which if present and not removed will contaminate the phospho-molybdate and is said to sometimes retard its precipitation. Method No. 2. When titanium is so low as to offer no objection, the ash is decomposed as under method No. 1, but evaporation is carried only to a volume of about 5 c.c. The solution is diluted with water to 30 c.c., boiled and filtered. If the washings are turbid they are passed again through the filter. The residue is ignited in a platinum crucible fused with a little Na a CO s , the melt dissolved in HNOs and its solution, if clear, added to the main one. If not clear it is filtered. The sub- sequent procedure is as under method No. 1. The fusion of the residue may be dispensed with in routine work on a given coal if it is certain that it is free from phosphorus. ULTIMATE ANALYSIS Carbon and Hydrogen. The determination of carbon and of hydro- gen is made with a weighed quantity of sample in a 25-burner combus- tion furnace of the Glaser type. The products of combustion are thoroughly oxidized by being passed over red-hot CuO and PbCrO 4 , and are fixed by absorbing the water in a weighed Marchand tube filled with granular CaCl s and by absorbing the CO 2 in a Liebig bulb con- taining a 30 per cent, solution of KOH. The apparatus used consists of a purifying train, in duplicate, a combustion tube in the furnace, and an absorption train. The puri- fying train consists of the following purifying reagents arranged in order of passage of air and oxygen through them: H 2 SO 4 , KOH solu- tion, soda lime, and granular CaCl?. One of the trains is for air and one for oxygen. In the H 2 S0 4 and KOH scrubbing bottles the air and the oxygen are made to bubble through about 5 mm. of the purifying reagent. Both purifying trains are connected to the combustion tube by a Y-tube, the joint being made tight by a rubber stopper. The combustion tube is made of hard Jena glass. Its external diam- eter is about 21 mm., and its total length is 1 meter. The first 30 cm. of the tube are empty; following this empty space is an asbestos plug (acid-washed and ignited), or in its place a roll of oxidized copper gauze may be used; the next 40 cm. are filled with "wire" CuO; a second asbestos plug separates the copper oxide from 10 cm. of fused PbCrO 4 , which is held in place by another asbestos plug 20 cm. from the end of the tube. The end of the tube is drawn out for rubber-tubing connec- tion with the absorption train. The absorption train consists, first, of a Marchand tube filled with granular CaCl 2 to absorb moisture. The CaClj should be saturated 252 GAS AND FUEL ANALYSIS \vith CO 2 before using. The Marchand tube is followed by a Liebig bulb containing a 30 per cent. KOH solution, in which any possible impurities, as ferrous iron or nitrites, have been oxidized by a little KMnC>4. A guard tube containing granular CaCl 2 and soda lime, is attached to the Liebig bulb to absorb any CO 2 escaping the KOH solution and any water evaporating from that solution. The train is connected to an aspirator which draws the products of combustion through the entire train. A guard tube of CaCl 2 prevents moisture from running back into the absorption train. The suction is maintained constant by a Mariotte flask. The advantage of aspirating the gases through the train rather than forcing them through by pres- sure is that the pressure on the rubber connections is from the outside, so that gas-tight connections are more easily maintained than if the pressure is on the inside of the tube. The connections are made as tight as possible. The usual test for tightness is to start aspiartion at the rate of about three bubbles of air per second through the potash bulb, and then to close the inlet for air and oxygen at the opposite end of the train; if there is no more than one bubble per minute in the potash bulb, the apparatus is considered tight. Before starting a determination when the train has been idle some hours, or after any changes in chemicals or connections, a blank is run by aspirating about 1 liter of air through the train which is heated in the same manner as if a determination on coal were being made. If the Liebig bulb and the tube containing calcium chloride show a change in weight of less than 0.5 mg. each, the apparatus is in proper condition for use. A porcelain or platinum boat is provided with a glass weighing tube of suitable size, which is fitted with an accurately ground glass stopper. The tube and empty boat are weighed. Approximately 0.2 g. of the air-dry coal (60-mesh and finer, or better 100-mesh if much free im- purity is present) are quickly placed in the boat. The boat is at once placed in the weighing tube, which is quickly stoppered to prevent moisture change in the coal while weighing, and transferring to the furnace. The absorption tubes are connected and the boat and sample are transferred from the weighing tube to the combustion tube, which should be cool for the first 30 cm. The CuO should be red hot and the PbCrO 4 at a dull-red heat. The transfer of the boat from weighing tube to combustion tube should be made as rapidly as possible. As soon as the boat is in place near the asbestos plug at the beginning of the copper oxide the stopper connecting with the purifying train is inserted and the aspiration started with pure oxygen gas at the rate of three bubbles per second. One burner is turned on about 10 cm. back THE CHEMICAL ANALYSIS OF COAL 253 from the boat, and the aspiration is continued carefully until practi- cally all the moisture is expelled from the sample. The heat is then increased very gradually until all the volatile matter has been driven off. In driving off the volatile matter the heat must be applied gradu- ally in order to prevent a too rapid evolution of gas and tar, which may either escape complete combustion or may be driven back into the puri- fying train. The heat should be slowly increased by turning on more burners under the open part of the tube until the sample is ignited; then the temperature can be increased rapidly, but care should be taken not to melt the combustion tube. Any moisture collecting in the end of the combustion tube or in the rubber connection joining it to the CaCl 2 tube is driven over into the CaCl 2 tube by carefully warming with a piece of hot tile. The aspiration with oxygen is continued for 2 minutes after the sample ceases to glow, the heat is then turned off and about 1200 c.c. of air are aspirated. The absorption bulbs are then disconnected, wiped with a clean cloth, and allowed to cool to the balance-room temperature before weighing. 11.19 X (increase in weight of CaCl 2 tube) Percentage of hydrogen = Weight of sample 27.27 X (increase in weight of KOH bulb) Percentage of carbon = ~ 7 . , , Weight of sample The ash in the boat is weighed and carefully inspected for any un- burned carbon, which would destroy the value of the determination. Method with Electrically Heated Combustion Furnace. An electrically heated combustion furnace of the Heraeus type is used by the Bureau of mines. It consists of three independent heaters, two of which are provided with sheave wheels, and are mounted on a track so that they are movable along the tube; the third heater which surrounds the PbCr04, is stationary. The furnace as provided by the manufacturer does not include the small stationary heater. This can be made in the laboratory by winding an alundum tube 12 cm. in length with No. 20 nichrome II wire and enclosing it in a cylinder packed with magnesia asbestos. The movable heaters have very thin platinum foil, weighing about 9 g. in all, wound on a porcelain tube of 30 mm. internal diam- eter. The larger one which heats the CuO, is 350 mm. in length, and the smaller one, which heats the sample in the boat, is 200 mm. in length. The Jena glass or fused silica combustion tube, about 21 mm. external diameter and 900 mm. in length, is supported by an asbes- tos-lined nickel trough. The current through each heater is regulated independently by separate rheostats, mounted on the frame of the furnace. The two platinum-wound heaters require an average current 254 GAS AND FUEL ANALYSIS of about 4.5 amperes at a pressure of 220 volts, although for heating rapidly a larger amperage is necessary. The oxygen or air entering the combustion tube is purified by passing through a Tauber's drying apparatus, which contains the following reagents arranged in order of the passage of air or oxygen through them: H 2 S0 4 , for removing posible traces of ammonia, 30 per cent. KOH solution, granular soda lime, and granular CaCl 2 . One side of the train is connected directly to a Linde oxygen tank, which is provided with a reducing valve for regulating the oxygen pressure ; the other side of the train is used for purifying the air supply. The absorption train consists of a 5-in. U-tube, filled with granular CaCl 2 to absorb moisture. Before using, the CaCl 2 should be saturated with C0 2 to avoid possible absorption of C0 2 during a determination by any traces of CaO that may be present. This saturating is done most conveniently by placing a quantity of CaCl 2 in a large drying jar, and % filling the jar with C0 2 . After standing over night, dry air is drawn through the jar to remove the CO 2 . The treated CaCl 2 is kept in well-stoppered bottles. The CaCl 2 tube is connected to a Vanier potash bulb containing a 30 per cent. KOH solution and granular CaCl 2 . Six to eight determin- ations can be made without recharging this bulb. The potash bulb is connected to an aspirator through a guard tube containing granular CaCl 2 and soda lime, and a guard tube containing granular CaCl 2 and soda lime and a Mariotte flask. The Mariotte flask keeps the pressure constant. In general, the method of determination is the same as the one used with the gas furnace. By moving the heaters toward the end of the tube where the gases enter, and cutting in the electric current, the air can be warmed enough to thoroughly dry the tube and its contents. The current is then cut off from the small heater ,and the large heater is moved over the CuO; about 250 mm. of that part of the combustion tube between the two heaters where the boat containing the sample to be placed is kept exposed. The full current is then turned on the large heater to bring the CuO to a red heat. When this temperature is reached it is necessary to reduce the current with the rheostat to avoid melting the tube. In the meantime the absorption train is weighed and connected, and the boat containing the sample is placed in the exposed and cooler part of the tube between the two heaters. The current is then passed through the shorter heater. By manipu- lating the rheostat and by gradually pushing this heater toward the boat, the rate of evaporation of moisture and evolution of volatile matter can be readily controlled. THE CHEMICAL ANALYSIS OF COAL 255 After combustion is complete, the electric current is turned off the smaller heater and this heater moved back to allow the tube to cool for the next determination. The final aspiration of air and the weighing of the absorption train is conducted as described under the gas-furnace method. Note. In place of granulated CaCl 2 , concentrated H 2 S0 4 , may be used for collecting the water formed by combustion. In such cases the air and oxygen entering the combustion tube and the gas leaving the potash bulb must also be dried by H 2 S0 4 . Other suitable forms of absorption vessels than those indicated in the above procedure may be used. NITROGEN The Kjeldahl-Gunning method is recommended for the determination of nitrogen This method has the advantage over either the simple Kjeldahl or the Gunning method, in requiring less time for the complete oxidation of the organic matter, and in giving the most uniform results. The Kjeldahl-Gunning Method. One gram of the coal sample is boiled with 30 c.c. of concentrated H 2 S0 4 , 7 to 10 g. of K 2 S0 4 , and 0.6 to 0.8 g. of metallic mercury in a 500 c.c. Kjeldahl flask until all particles of coal are oxidized and the solution nearly colorless. The boiling should be continued at least 2 hours after the solution has reached the straw- colored stage. The total time of digestion will be from 3 to 4 hours. The addition of a few crystals of KMn0 4 after the solution has cooled enough to avoid violent reaction, tends to insure complete oxidation. After cooling, the solution is diluted to about 200 c.c. with cold water. If the dilution with water has warmed the solution, it should be cooled again and the following reagents added: 25 c.c. K 2 S solution (40 g. K 2 S per liter) to precipitate the mercury; 1 to 2 g. of granular zinc to prevent bumping; and finally enough strong NaOH solution (usually 80 to 100 c.c.) to make the solution distinctly alkaline. The danger of loss of NH 3 may be minimized by holding the flask in an inclined position while the NaOH solution is being added. The alkaline solu- tion runs down the side of the flask in an inclined position while the NaOH solution is being added. The alkaline solution runs down the side of the flask and forms a layer below the lighter acid solution. After adding the alkaline solution, the flask is at once connected to the condensing apparatus and the solution mixed by gently shaking the flask. The NH 3 is distilled over into a measured amount (10 c.c.) of stand- ard H 2 S0 4 solution, to which has been added sufficient cochineal indi- cator for titration. Care should be taken that the glass connecting tube on the end of the condenser dips under the surface of the standard 256 GAS AND FUEL ANALYSIS acid. The solution is slowly distilled until 150 to 200 c.c. of distillate has passed over. To avoid mechanically entrained alkali passing over into the condenser, the rate of distillation should not exceed 100 c.c. per hour. The distillate is titrated with standard NH 3 solution (20 c.c. NH 4 OH solution = 10 c.c. H 2 S0 4 solution = 0.05 g. nitrogen). Stand- ard NaOH or KOH solution with methyl orange or methyl red as indi- cator may be used instead of NH 3 and cochineal. A blank determination should be made in exactly the same manner as described above, except that 1 g. of pure sucrose (cane sugar) is substituted in place of the coal sample. The nitrogen found in this blank determination is deducted from the result obtained with the coal sample. The K 2 S and NaOH may be dissolved in a single stock solution. Sufficient K 2 S is dissolved in the water before adding the NaOH to make a solution in which the quantity necessary for a nitrogen deter- mination (80 to 100 c.c.) contains 1 g. of K 2 S. Twelve grams of K 2 S and 500 g. of NaOH in one liter of water are required for the above proportions. Coke and anthracite should be ground to an impalpable powder as they are very difficult to oxidize. Even if this is done the digestion may require 12 to 16 hours. OXYGEN There being no satisfactory direct method of determining oxygen, it is computed by subtracting the sum of the percentages of hydrogen, carbon, nitrogen, sulphur, water and ash from 100. The result so ob- tained is affected by all the errors incurred in the other determinations and especially by the change in weight of the ash-forming constituents on ignition; iron pyrite changes to ferric oxide, increasing the ash and causing a negative error in the oxygen equivalent to three-eighths of the pyritic sulphur. On the other hand, there is always a loss on igni- tion, of water of composition from the clayey and shaley constituents, C0 2 from carbonates, etc., which tends to compensate the absorption of oxygen. Corrected Oxygen. When a more correct oxygen value is desired, it may be obtained by making the corrections indicated in the following formula: Corrected oxygen = 100 - [(C - C') plus (H - H') plus N plus H 2 plus S' plus corrected ash.] in which C equals total carbon C' equals carbon of carbonates THE CHEMICAL ANALYSIS OF COAL 257 H equals total hydrogen less hydrogen of water H' equals hydrogen from water of composition in clay, shale, etc. N equals nitrogen HaO equals moisture as found at 105 C. S' equals sulphur not present as pyrite or sulphate. This is usually small. In many types of coal it may be disregarded. CORRECTED ASH Corrected ash equals mineral constituents originally present in the coal. For most purposes this can be determined with sufficient accur- acy by adding to the ash, as found, five-eighths of the weight of pyritic sulphur, the CO 2 of carbonates and the water of composition of clay, shale, etc. See also Determination of Ash. 19. Standard Methods for Laboratory Sampling and Analysis of Coke. 1 The American Society for Testing Materials has issued a standard method for sampling and analysis of coke which in general follows the lines laid down for coal. Coke is so very abrasive that especial care must be taken not to in- crease the ash of the sample during crushing. The samples may be coarsely crushed with a jaw or roll crusher or by hand on a chilled iron or hard-steel plate by impact of a hard bar or sledge, avoiding all rubbing action. The use of rubbing sur- faces such as a disk pulverizer or a bucking board is never per- missible for grinding coke. The final grinding to 60 mesh may be made in a porcelain jar mill. No special care is needed in determining moisture which may be done in an ordinary oven at 104-110 C. The other methods follow those for coal quite closely. The American Society for Testing Materials Standards 1918, p. 709. 17 CHAPTER XVI HEATING VALUE OF COAL BY THE BOMB CALORIMETER 1, General Methods of Determining Heating Value. The heating value of a fuel is determined either directly in a calori- meter or indirectly by calculation from its chemical composition. The direct calorimetric method involves the combustion of the fuel by oxygen supplied either as free oxygen gas or as combined oxygen of some chemical compound, the operation being carried out in a closed vessel immersed in a known mass of water under conditions which ensure that the heat evolved in the oxidation shall be with as little loss as possible transferred to and retained by the calorimetric vessel and the water. The heat evolved is calculated from the rise in temperature of the system. Modern methods of calorimetry really commenced with the invention by Berthelot of his bomb calorimeter described in 188 1. 1 This has become the standard method for the determin- ation of heating values and therefore the whole of this chapter is devoted to it. Other methods, especially that of Parr, are described in the following chapter. 2. The Calorimetric Bomb. Berfchelot showed that if com- bustion of carbon compounds took place in a closed vessel in an atmosphere of oxygen compressed to at least seven atmospheres and with a weight of .combustible such that only 30 to 40 per cent, of the oxygen initially present was consumed, combustion was rapid and complete. His bomb was lined with heavy plat- inum and was very expensive. Hempel in the second edition of his gas analysis published in 1889 described a much cheaper bomb which had no lining and which has been found to be mechanically unsatisfactory. Mahler 2 in 1892 reported a careful study of Berthelot's method as applied to coals, and described 1 Annales de Chimie, 5 Serie, 23, 160 (1881). Annales de Chimie, 6 Serie, 6, 546 (1885). 2 Bui. de la Societe d. Encouragement, 1892, 319. 258 HEATING VALUE OF COAL BY THE BOMB CALORIMETER 259 a bomb of improved construction with an enamel instead of a platinum lining. This bomb is mechanically better than HempePs, but there is still the objection that the top as it screws down, roughens the gasket., Atwater 1 in 1894 described a modi- fication of the calorimetric bomb distinctly superior mechanically to the preceding iorms. It resembled more closely the Berthelot bomb than either of the others but, whereas the Berthelot bomb was closed by a tapered plug held in place by a screwed cap, Atwater's was closed by a flat cap held in place by a collar slipped over it and screwed over threads on the outside of the bomb after the manner of a union pipe-fitting. In this way all tearing of the gasket was avoided. The Atwater bomb may be provided with a gold or platinum lining. Many modifications of the calorimetric bomb have been made by other workers, but the principle has not been changed. Anyone familiar with one instrument can readily learn to use any other. 3. Details of the Calorimetric Bomb. The calorimetric bomb which has been in use in the calorimeter laboratory of the Uni- versity of Michigan since 1908 is shown in Figs. 55 and 56. It is in general patterned after the Atwater bomb, but possesses several improvements. One of these, due to Mr. Edwin H. Cheney, is the octagonal belt on the body of the bomb which fits into a recessed plate and holds the bomb rigidly while the cover is being screwed on. Another, due to Professor S. W. Parr, is the deeply recessed groove for the gasket in the cover of the bomb into which the straight lip of the bomb fits closely so that a rubber gasket may safely be used. Improvements in various details are due to Mr. J. H. Stevenson, instrument maker of the University of Michigan. Details of the bomb are shown in Fig. 55. It consists of a cylinder of about 300 c.c. capacity on which sits a cover carrying the oxygen inlet and needle valve. The original models were made of steel and some of them have withstood continuous use by classes of beginners for ten years. They become coated with a layer of dense adherent scale on the inside and after that show very little change. One of these bombs was cut into sections after ten year's use and showed no measurable decrease in thickness of the metal due to corrosion. 1 Storrs Conn. Experiment Station Report, 1894, 135; also J. Am. Chem. Soc., 25, 659 (1903). 260 GAS AND FUEL ANALYSTS Monel metal was soon substituted for steel as the material from which the heads were made to avoid corrosion of the needle valve, and within the last few years the entire bomb, except the collar, has been made of monel metal. It stays bright both inside and out with no especial care, and the small amount of metal dissolved by the acids formed in combustion introduces only a negligible error in the calorimetric work. Compressed oxygen is admitted through a flexible metal tube soldered at A FIG. 55. Details of calorimetric bomb. to the steel tube with the coned head B. This tube AB slides freely in the threaded sleeve C. Its coned head makes a gas tight joint with the bomb when C is screwed up. The needle valve D closes the bomb when it is screwed down. The coal sits in a flat nickel or quartz capsule E supported on a ring which screws into the head piece. The insulated electrical connection FG is a rod coned where it passes through the head piece from which it is insulated by a bit of thin rubber tubing. The binding post G screwing down on the threaded end of F which projects HEATING VALUE OF COAL BY THE BOMB CALORIMETER 261 through the cover pulls the cone tightly into its seat and makes a gas tight joint. A mica disc placed between the binding post and the bomb completes the electrical insulation of the electrode from the bomb. The gasket which fits into the groove H may be of lead, hard fiber or rubber. Rubber gives a tight joint with the least pressure of the spanner and is therefore to be preferred. If it is cut to fit the groove accurately the inner lip of the bomb projecting into the recessed head will effectually protect it from the hot gases evolved in the bomb during com- bustion. Fig. 56 shows the various parts of the calorimeter. Two bombs are shown, one assembled and one taken apart and with FIG. 56. Bomb calorimeter. the head sitting on a stand in position for adjustment of the fine iron firing wires. The nickel-plated copper can, the stirrer and the insulating buckets are also shown. The insulating buckets as shown in Fig. 56 consist of two con- centric fiber pails with air in the space between them. It is in some ways preferable to have this space filled with water, whose temperature may be set at any desired point. This minimizes the effect of draughts in the room and enables the operator to use a Beckman thermometer without having to shift its zero when the room temperature fluctuates. The temperature of the water must not, however, be so far below room temperature 262 GAS AND FUEL ANALYSIS that dew will deposit on the walls of the calorimeter vessel. When a water jacket is thus used in the calorimeter, it should be provided with a stirrer and its temperature should be recorded. Where a large number of calorimetric determinations are to be made in a laboratory the multiple unit installation designed by the Bureau of Mines 1 may profitably be installed. 4. Thermometers. Thermometers for the calorimeters should be made especially for the purpose with a stem long enough to allow the bulb of the thermometer to be opposite the center of the bomb and should be carefully calibrated. It is necessary that it be possible to read the rise in temperature to at least 0.01 C. The best thermometers are those of the Beckman type with a scale length of 6 and a zero point adjustable between 12 and 25. This type of thermometer is always to be recom- mended where the calorimeter room is of relatively constant temperature so that it is not necessary to change the zero point often. Where this desirable condition is not fulfilled calori- metric thermometers with a fixed scale running from 15 to 30 C. must be used. These are usually divided only into 0.02 to avoid the excessive length of stem which would otherwise result. The thermometer should in any case have been carefully cali- brated since an error of 0.01 on the average rise of 3 means 0.3 per cent, or approximately 40 B.t.u. per pound of coal. 5. Preparation of Sample. The methods to be followed in ob- taining a representative sample from a large quantity of coal and the precautions necessary in grinding, sampling, and drying this large sample have been given in Chapters XIV and XV. It is assumed here that the sample is already ground to a fineness of at least 60-mesh and has been air-dried. The amount of mois- ture is immaterial so far as the operation of the bomb calorimeter is concerned, but an air-dried sample is less likely to change during the operation of weighing. When powdered bituminous coal is burned in compressed oxy- gen, combustion is so violent that there is danger that gas and even solid particles will be projected unburned through the flame zone. The rate of combustion may be materially lessened by reducing the surface of coal exposed to the oxygen. This is best accomplished by briquetting the coal. Most bituminous coals 1 David and Wallace, Technical Paper 91, Bureau of Mines (1918). HEATING VALUE OF COAL BY THE BOMB CALORIMETER 263 may be readily compressed into pellets in a screw press. The pressure should be slowly applied and allowed to remain for a few minutes. The resulting pellet may be trimmed to approxi- mate weight with a penknife. It is advantageous to break it into two or more pieces and discard the dust before weighing. The advantage of cutting the pellet lies in the readier ignition, for pellets which have been pressed very hard are sometimes so dense on the surface that they fail to ignite. It is possible to compress some bituminous coals so firmly on the surface that the gas evolved in the interior of the briquette by destructive distillation explodes the briquette and blows a cap of coke out of the crucible. If these dense briquettes are cut into several pieces, as directed above, the trouble will be obviated. It is not necessary to briquet anthracite coals or coke. Indeed, it is not possible to do so without the addition of a binder such as sugar or bituminous coal. 6. Manipulation of Bomb Calorimeter. The bomb is taken apart and examined to see that it is in good condition and that the gasket is not cut. A few drops of water are placed in the bottom part of the bomb which is set in its receptacle in the table-top. The top part of the bomb is placed on a ring of an ordinary ring stand as shown in Fig. 56 which allows the heavy terminals to drop through in a convenient position for adjust- ment of the fuse wire and sample. The weighed sample of coal is placed on a shallow thin quartz, nickel or platinum capsule resting on the supporting ring suspended from the head of the bomb. The capsule must be almost flat to allow free access of oxygen from the edges as the flame flares up. Otherwise com- bustion may be incomplete. It must be thin or it will chill the flame and prevent complete combustion. It is advisable with anthracite and coke to place a thin pad of ignited asbestos on the metallic capsule in order to decrease still further the cooling effect of the metal. A measured length, preferably not more than two inches, of the fine iron ignition wire 34 B. & S. gage is attached to the heavy wire terminals by winding the ends of the fine wire several times around the heavy ones, leaving the fine wire in the form of a loop between the terminals. After making connections the loop is pushed down until it rests on the fragments of coal. Care 264 GAS AND FUEL ANALYSIS is to be taken that the wire does not touch the metal capsule and form a short circuit. The cover with the sample in position is placed carefully on the bomb and the threaded collar slipped over it and screwed down, pressure finally being applied with the spanner. A novice will nearly always screw the cover down harder than necessary, thus shortening the life of the gasket. A moderate pressure will suffice if the gasket is a good one. Gaskets cut from ordinary red fiber packing are too porous to be tight unless screwed down with great pressure. They may be much im- proved by vacuum impregnation with a solution of 5 grm. of glue in 5 c.c. of glycerine and 100 c.c. of water. After impreg- nation the gaskets are to be dried in air and rubbed with paraffine or graphite to keep them from sticking to the metal. The loose joint on the end of the flexible metal tube from the oxygen tank is screwed into the head of the bomb, the needle valve of the bomb opened at least a full turn, and then the valve on the oxygen tank is opened slightly, the gas entering in a slow stream from the tank until the gage shows 20 atmospheres pressure. If the valve on the tank is opened relatively more than the one on the bomb the gage between the two may show 20 atmospheres before there is that much pressure in the bomb. The oxygen valve on the tank is to be closed first and after that the valve on the bomb. If the valve on the bomb is closed before that on the tank, the pressure on the gage will rise very quickly to the full pressure of the oxygen tank which may be 2000 Ib. and the gage may be blown up. The bomb is disconnected and placed in the water of the calorimeter or in a separate vessel of water to test for leaks. If bubbles of gas appear around the threaded ring the cover must be screwed down more tightly, and possibly the gasket may have to be replaced. If bubbles of air come from the head it is evi- dent that the needle valve is leaking. It is worse than useless to try and force it to become tight by screwing down the needle with great pressure A needle valve ground truly into its seat is tight with slight pressure. If a particle of grit comes between the metal surfaces the application of pressure causes it to scour the polished surface and the valve will leak until it has been again ground to a true surface. In case of a leaking needle valve HEATING VALUE OF COAL BY THE BOMB CALORIMETER 265 the pressure must be relieved, the bomb opened and the needle valve unscrewed entirely out of the head. The lock nut into which it was threaded is also to be removed. The coned seat into which the needle valve is ground may now be seen in a strong light. The best policy is to grind the needle valve into its seat, an operation requiring not more than ten minutes if the valve has not been abused. The needle valve is dipped into a paste of fine emery or carborundum in water and ground into its seat by rotating it back and forth with the fingers. A polished ring will soon be visible on the cone point and a corresponding ring in the. seat. When this appears, unless the metal has been badly scratched, the process may be considered complete and the grinding interrupted. The valve is to be thoroughly cleaned from grit and dried, when it is ready for use. The bomb when charged is to be carefully centered in the calorimeter vessel which is in turn centered in the outer vessels. The stirrer and thermometer are placed in position and two liters of water whose temperature is approximately 3 below room temperature is added. A glass flask which holds 1000 c.c. of water, contains the following weights of water, when balanced against brass weights in air. 1 15 C 998.05 grm. 20 C 997 . 18 grm. 25 C 996.04 grm. 30 C 994.66 grm. The liter flasks of various makers differ in the amount of water which they discharge and the flask should be calibrated by direct weight for some one temperature. It is more conven- ient for calculation purposes to calibrate the flask to deliver 2000 grm. of water at the temperature most frequently used, or sometimes to deliver such an amount of water that the sum of the water added and the water value of the calorimeter shall be 2500 grm. It is in many ways better to weigh the water directly into the counterpoised calorimeter vessel as it sits on the balance. Especial care is to be taken to see that the thermometer is centered in the space between the bomb and the edge of the 1 Bureau of Standards Bull. 4, 600 (1907-08) 266 GAS AND FUEL ANALYSIS vessel. If it touches either, or even if it is a little off center, the rise of the thermometer will not be even and the result may be in error. After the adjustments are complete the stirrer is operated for at least two minutes before the first temperature reading is made on an even minute. Readings are to be made each minute thereafter for at least five minutes, the stirrer being kept going steadily at 30-40 strokes per minute and the temperature slowly and steadily rising with each reading as heat is absorbed from the air of the room, or dropping if the calorimeter is above room temperature. This ends the preliminary period, which must show at least five readings changing by regular increments due solely to heat transfer to or from the outside air. The firing circuit is closed simultaneously with the last read- ing of the preliminary period. The iron wire becomes heated to redness, the coal ignites and the iron wire fuses almost in- stantly. It is well to have an electric lamp in the firing circuit which lights when the current is turned on and is extinguished when the wire fuses. An ammeter in the circuit answers the same purpose showing that the ignition is prompt and that an undue amount of heat is not imparted to the calorimeter by the electric current. Current for ignition may best come from a storage battery or group of dry cells giving about 12 volts. Higher voltages are apt to cause insulation troubles. Within a half minute after ignition the thermometer begins to rise so rapidly that it is not possible to make the thermometer readings accurately. They should be taken as accurately as possible, and regularly on each minute. After about three minutes the thermometer reaches its maximum, but the stirring and temperature readings must be kept up without intermission for a total of ten minutes after ignition to obtain data for the radiation corrections. After the termination of the thermometer readings the bomb is removed from the calorimeter, wiped dry, and placed in its receptacle on the table. The needle valve is opened and after the pressure is relieved the top is removed. The coal should be perfectly burned and the ash should appear as fused beads. The iron wire has burned as far as the heavy conductors and in accurate work the length of the wire unburned should be deter- HEATING VALUE OF COAL BY THE BOMB CALORIMETER 267 mined to enable the proper correction to be made for the weight of wire burned. The weight of wire burned comes to be almost a constant for each operator and may be taken as such in ordinary work. If soot appears in the bomb or on the capsule, the deter- mination should at once be rejected. With inexperienced ope- rators this trouble is frequently caused by opening the valve on the oxygen tank too fast when filling the bomb, with the result that the gage on the connecting tube jumps to the proper reading before the indicated pressure is reached in the bomb. If trouble persists it may be necessary to increase the oxygen pressure to 25 atmospheres. The bomb is to be rinsed out carefully and unless it is to be used again at once, is to be dried best in an oven at a temperature of about 35-40 C. Careful drying is especially necessary unless non-corrodible alloys are used throughout the construction of the bomb. 7. Thermometer Corrections. The calorimetric thermometer should have the certificate of the Bureau of Standards. In addi- tion to the corrections indicated on the certificate as inherent in the thermometer on account of variation in the diameter of the capillary tube, etc., minor corrections must be made in ac- curate work for variations due to the conditions under which the thermometer is used. The Bureau of Standards calibrates ther- mometers when totally immersed in a bath of the temperature indicated. In calorimetric work the bulb and part of the stem is within the calorimeter, while part of the stem projects through the cover of the calorimeter into the air of the room. A small correction must be made for this emergent stem. In the case of Beckmann thermometers an additional ''setting factor correc- tion" must be used in case the thermometer is set for a different zero from that used in the calibration. The formulae for these corrections vary with different sorts of glass and are given in full in the certificate of calibration accompanying each thermometer. The corrections rarely amount to more than a few thousandths of a degree. 8. Radiation Corrections. The combustion of the coal in a bomb calorimeter is probably a matter of only a few seconds, but it requires several minutes for the heat to be transmitted to the water and for the thermometer to register the rise in temperature. With the usual type of instrument radiation corrections must 268 GAS AND FUEL ANALYSIS be made in spite of careful jacketing of the calorimeter. Their magnitude is lessened by adjusting the temperature of the water placed in the calorimeter with reference to room temperature and to the rise in temperature expected. If the rise in tempera- ture is to be 3, the water poured into the calorimeter should be about 3 below room temperature. When equilibrium is reached at the time of ignition the temperature will be about 2.5 below that of the room and after combustion it will be about 0.5 above room temperature. This arrangement minimizes the errors. The temperature rises very rapidly after ignition to one so nearly that of the room that changes due to radiation are slight and repeated readings may be made to obtain the final temperature. If the final temperature of the calorimeter is slightly above that of the room there should be a maximum point in the thermometer readings with a slow decrease thereafter. It is common practice to consider that the actual maximum thermometer readings represent the actual maximum tem- perature of the calorimeter, but it is not a safe assumption, for if the thermometer bulb is unduly close to the bomb or if the stirring is inefficient the thermometer may rise too high and fall rapidly again to the temperature representing the true average value of the system, after which it will change slowly and regularly through radiation. It is, therefore, unsafe to use the maximum temperature in calculations. The final tempera- ture of the combustion period should be taken only after sufficient time has elapsed so that it is certain that the system has come to equilibrium. Five minutes is usually sufficient. Radiation corrections are based on the principle that the inter- change of heat between the room and the calorimeter is propor- tional to the difference in the temperature between them. The temperature of the room is assumed to be a constant during any one operation and need not even be known. The formula for the correction as used by Regnault and developed by Pfaundler 1 is somewhat complicated in appearance, but is simple in use. Regnault-Pfaundler Formula. Three sets of temperature read- ings are to be made. The initial set must not start until after the temperature of the calorimeter has commenced to change regularly due to radiation. It consists of at least five readings 1 Poggendorfs Annalen, 129, 115 (1866). HEATING VALUE OF COAL BY THE BOMB CALORIMETER 269 made one minute apart. Only the first and last readings and the time interval enter into the calculation, but it is advisable to record the intermediate readings as a check on the accuracy of the two important ones and to make sure that the change of temper- ature is uniform as it should be. At the moment of taking the final reading of the initial period the firing key is pressed and the reading just taken is recorded, both as the final reading of the initial period and the first reading to the combustion period. It is to be marked TO. During the combustion period readings are to be made and recorded regularly, not only till the thermometer reaches its maximum, but also till it is certain that the changes in tempera- ture are again due solely to radiation. This period may be five to ten minutes. There follows a final period of five minutes to fix the radiation losses for the latter portion of the test. The derivation of the Regnault-Pf aundler formula is as follows : Let t = mean temp, of initial period. t' = mean temp, of final period, v = loss per time interval in initial period, v' = loss per time interval in final period. To, TI, T 2 , T n = temperature readings in combustion period. ti, t2, . . . . t n = average temp, of each interval during com- bustion period; i.e., ti = - -, etc. A a, QJ. o r A Fio. 57. Diagram showing derivation of Regnault-Pf aundler formula. The special case assumed by Pfaundler is one where the initial temperature is only slightly different from room temperature, giving a small value for v. The final temperature is considerably above room temperature and the value of v' is larger than v. The geometrical construction for the Regnault-Pf aundler formula is shown in Fig. 57. The demonstration is as follows: 270 GAS AND FUEL ANALYSIS Lay off OA = t. Lay off OA' = t'. LayoffOai = ti. Oa 2 = t 2 .... Oa n = t n At A erect perpendicular A V = v. At A' erect perpendicular A'V = v'. Join V and V by a straight line and at ai a 2 . . . . a n erect perpendicular intersecting W. Any ordinate a r V r = AV+ p V r . A'V AV On account of similar triangles pV r = --- T-T-/ pV. A A C = the algebraic sum of all the ordinates = correction sought. n = the number of observations in the combustion period proper. ' ... t n -nt). Heat received by the calorimeter from the outside air is con- sidered as negative and therefore in the especial case assumed by Pfaundler where the initial temperature was slightly under room temperature v was negative. The correctness of the formula is independent of the relative values and signs of v and v'. The corrected rise in temperature of the calorimeter R = T n -T +C The need of an elaborate correction for radiation is naturally less when the calorimeter is provided with an adequate stirrer so that the heat interchange between the bomb and the water is quickly effected, and also less when the insulating jacket is good than when it is poor. With a well designed calorimeter the largest part of the rise in temperature occurs in the first minute and if the final temperature is only slightly above room tempera- ture radiation in succeeding minutes is almost negligible. The HEATING VALUE OF COAL BY THE BOMB CALORIMETER 271 standard method of the American Society for Testing Materials prescribes that a turbine stirrer shall be used because it stirs more efficiently than a reciprocating stirrer. SAMPLE OF RECORD Determination of Heating Value of Coal in Bomb Calorimeter. Sample No. U. 38 Date Nov. 10, ' Calorimeter No. 2 Thermometer No. 4 Water Value of Calorimeter 475 grm. Water Used 2000 c.c. = 1995 grm. Total water equivalent 2470 grm. Sample of coal (air-dried) 0.9922 grm. Thermometer readings by minutes 19.68 19.68 19.69 19.69 19.69 21.4 22.58 To 22.95 23.09 23.11 T 2 T 4 T 6 Factors v = -0.0025 v' = +0.0025 t = 19.69 t' = 23.11 T!+T 2 +T 3 +T4=90.0 To + Ts =21.4 n =5 Thermometer corrections T n =23.11-0.045 =23.065 To = 19.69-0.040 = 19.65 23.11 23 . 1 1 23. 10 23 .10 = 5X -0.0025 + ll-i (90.0+21.4-5X19.7) = +0.006 C. R =23.065 -19.65 +0.006= 3.421 C. 3.421 X2470 = 8450 calories Deduct for 0.025 grm. fuse wire 40 Deduct for 1.0 per cent, sulphur (20 X. 9922) = 20 60 OO QA 8390 calories 8455 calories per gram of air-dried coal. u.yy^z 8455X1.8 = 15,219 B.t.u. per pound of coal. 272 GAS AND FUEL ANALYSIS Proximate Analysis of Coal Moisture . 32 per cent. Volatile Matter 22.87 t t Fixed carbon 72.67 / 95 ' 5 P er Ash 4.14% 100.00 Heat evolved per pound coal dry and free from ash = ggg- = 15936 B.t.u. Radiation Correction by the Dickinson Formula. Dr. H. C. Dickinson 1 has worked out a simple method for determining radiation corrections whose use is recommended in the Standard Method of the American Society for Testing Materials, 2 from which the following quotation is made: Observe (1) the rate of rise (rj of the calorimeter temperature in degrees per minute for five minutes before firing, (2) the time (a) at which the last temperature reading is made immediately before firing, (3) the time (b) when the rise of temperature has reached six-tenths of its total amount (this point can generally be determined -by adding to the temperature observed before firing sixty per cent, of the ex- pected temperature rise, and noting the time when this point is reached), (4) the time (c) of a thermometer reading taken when the tempera- ture change has become uniform some five minutes after firing, (5) the final rate of cooling (r 2 ) in degrees per minute for five minutes. " The rate TI is to be multiplied by the time b a in minutes and tenths of a minute, and this product added (subtracted if the temperature were falling at the time a) to the thermometer reading taken at time a. The rate r 2 is to be multiplied by the time c-b and this product added (subtracted if the temperature were rising at the time c and later) to the thermometer readings taken at the time c. The difference of the two thermometer readings thus corrected, provided the cor- rections from the certificate have already been applied, gives the total rise of temperature due to the combustion. This multiplied by the water equivalent of the calorimeter gives the total amount of heat liberated. This result, corrected for the heats of formation of nitric and sulphuric acids observed and for the heat of combustion of the firing wire when that is included, is to be divided by the weight of the charge to find the heat of combustion in calories per gram. Calories per gram multiplied by 1.8 give the B.t.u. per pound. Example: 1 Jour. Ind. and Eng. Chem. 5, 525 (1913). Scientific Paper No. 230, Bureau of Standards. 2 A. S. T. M. Standard, 1918, HEATING VALUE OF COAL BY THE BOMB CALORIMETER 273 OBSERVATIONS Water equivalent 2550 grm. Weight of charge 1.0535 Approximate rise of temp. 3.2 60 per cent, of approximate rise 1.9 Time Temp. Corrected temp. 10-21 15.244 (Thermometer corrections from the certificate.) 22 15 . 250 23 15.255 24 15.261 25 15.266 (a) 26 15.272 15.276 Charge fired (b) 27-12 17. 2 01 (c)31 18.500 18.497 32 18 . 498 33 18.497 34 18.496 35 18.494 36 18.493 COMPUTATION fi =0.028 ^5 =0.0056 per minute, b a = 1.2 minutes The corrected initial temperature is 15.276 +0.0056 X 1.2 = 15.283. r 2 = 0.007 4- 5 =0.0014 per minute; c-b =3.8 minutes. The corrected final temperature is 18.497 +0.0014 X 3.8 =18.502 Total rise 18.502 - 15.283 , = 3.219 Total calories 2550X3.219 = 8209 Titration, etc 7 Calories from 1.0535 grm. coal 8202 Calories per gram 7785 or B.t.u. per Ib 14013 In practice, the time b a will be found so nearly constant for a given calorimeter with the usual amounts of fuel that b need be determined only occasionally. 9. Corrections for Oxidation of Nitrogen. When coal is burned on a grate minute amounts of oxides of nitrogen are formed by the combination of some of the nitrogen of the air and possibly also of the fuel, with the oxygen of the air. At the 1 The initial temperature is 15.27; 60 per cent, of the expected rise is 1.9. The reading to observe is then 17.2. 18 274 GAS AND FUEL ANALYSIS higher temperature of combustion in the compressed oxygen of the calorimeter more oxides of nitrogen are formed and account should be taken of the heat evolved in their formation. The heat of formation of aqueous nitric acid from nitrogen, oxygen, and water is represented, according to Thomsen, by the following equation. 2N+ 5O+ H 2 = 2HNO 3 + 29800 calories. This corresponds to 1058 calories per gram of nitrogen or 238 calories per gram of HNOs. The American Society for Testing Materials adopts the value 230 instead of 238. The nitric acid formed may be estimated in bombs with platinum or gold linings by rinsing out the bomb and titrating the washings with standard alkali. From this total acidity is deducted the sulphuric acid formed and the balance is considered nitric acid. The amount of nitrogen oxidized is roughly about one per cent, of the total nitrogen present whether introduced as free nitrogen with the oxygen or as combined nitrogen of the coal. The correction is not usually more than 8 calories and may be considered to be offset by the heat absorbed in keeping the gases in the calorimeter at constant volume. (See 12.) 10. Corrections due to Oxidation of Sulphur. When sulphur or pyrites burns in the air only about 5 per cent, of the sulphur is oxidized to SO 3 , the rest of it remaining as SO2. When com- bustion takes place under high oxygen pressure in the bomb calor- imeter a much larger percentage burns to SOa and correction must be made for it. The equations are. S+ 20 = SO 2 gas + 69,100 calories S+ 3O+ H 2 (excess) = dilute H 2 SO 4 + 141,100 calories. One gram of sulphur burning to S0 2 evolves 2165 calories and to dilute H 2 S04 evolves 4410 calories. There should therefore be a deduction made of 2245 calories for each gram of sulphur thus oxidized in the bomb. The difficulty is enhanced by the fact that sulphur may be present in coal as free sulphur, as sul- phur in organic combination, as pyrites or as calcium sulphate and that the corrections will vary for each of these various forms. For free sulphur burning to H 2 SO 4 the correction will be 2245 calories per gram as given above, for sulphur as pyrites 2042 l 1 Somermeier J. Am. Chem. Soc., 26, 566 (1904). HEATING VALUE OF COAL BY THE BOMB CALORIMETER 275 calories, while for sulphur as gypsum or sulphate of iron no cor- rection is to be made since it is already in the oxidized form. The conditions governing the oxidation of sulphur in the bomb calorimeter were studied by Regester 1 who found that the oxida- tion of SO 2 to SOs in the bomb was largely a function of the quantity of oxides of nitrogen which were present. The oxides of nitrogen are derived in part from the combined nitrogen of the coal and in part from the free nitrogen in the gases contained in the bomb. Commercial oxygen may contain very little nitro- gen and therefore the air which is originally present in the bomb should not be flushed out but should be left in the bomb to ensure the presence of a sufficient quantity of nitrogen to provide for the oxidation of the sulphur. It is customary to assume that all of the sulphur in coal exists in the form of pyrites and to deduct two calories for each milligram of sulphur in the sample of coal. This procedure is not above criticism for Parr 2 has shown that the amount of sulphate in fresh coal may be as high as one per cent, and that it doubles after six months storage of the ground sample in the laboratory. A source of error which should be considered here is that due to the possible action of the dilute sulphuric acid formed upon the inner surface of an unlined bomb. A steel bomb soon be- comes coated with oxide on its inner surface so the action will be between iron oxide and sulphuric acid. According to Thorn- sen the reaction Fe 2 O 3 xH 2 O+ 3H 2 SO 4 (dilute) evolves 33,840 calories. This means 353 calories for each gram of sulphur in- volved or 3.5 calories as the maximum error involved for 1 grm. sample of a coal containing 1 per cent, of sulphur. If the acid acts upon steel rather than iron oxide the heat evolved is even less. The error from this source is totally negligible. An ex- pensive calorimeter lined with gold or platinum is unnecessary except where the greatest refinements of accuracy are sought. 11. Correction Due to Combustion of Iron Wire. -The iron fuse wire which burns to Fe 3 O 4 evolves 1600 calories per gram, or 1.6 calories per mg. of iron burned. 12. Reduction to Constant Pressure. Combustion in the bomb calorimeter takes place at constant volume whereas in 1 Jour. Ind. and Eng. Chem., 6, 812 (1914). 2 Jour. Ind. and Eng. Chem., 5, 523 (1913). 276 GAS AND FUEL ANALYSIS ordinary furnace work combustion takes place at constant pres- sure. Wherever a decrease in volume takes place on combus- tion as where oxygen unites with hydrogen to form water which condenses, the gases in the calorimeter which should normally have contracted after combustion have had work done upon them to keep them at constant volume with the disappearance of an equivalent amount of heat. The correction amounts to 541 cal- ories for each gram molecule of gas which dissappears. When gaseous oxygen combines with carbon to form CC>2 there is no change of volume and hence no correction. The oxy- gen in the organic matter of the coal, may for the purposes of this calculation be considered to unite with the hydrogen of the coal to form water. No correction is needed here since both the hydrogen and the oxygen were in the solid state before combus- tion and the water formed is a liquid. There is always present in coal an excess of hydrogen over that sufficient to combine with the oxygen and this so-called available hydrogen burns with gaseous oxygen to form water which condenses The change in volume is shown by the equa- tion 4H(solid)+ 2 = 2H 2 O (liquid). The gas which disappears is oxygen in the proportion of one gram molecule for each 4 grm. of hydrogen. The amount of available hydrogen in coals varies from 3 to 5 per cent, so that on a gram sample there would be on an average 0.04 grm. of hydrogen which would unite with 0.01 grm. molecule of oxygen causing a correction of 5 calories a negligible amount except as it may be considered as balancing other minor errors such as that due to the oxidation of nitrogen. With petroleum the correction will be about three times as great as with coal. 13. Water Value of Calorimeter. When combustion occurs in a calorimeter there follows a rise in the temperature of both the water and of the calorimeter vessel. It is necessary to find how many calories are required to heat the metal parts of the calorimeter one degree and when this has been accomplished the value is translated for convenience of calculation into grams of water and called the water value of the calorimeter. The water value is usually determined in one of four ways. The first HEATING VALUE OF COAL BY THE BOMB CALORIMETER 277 is by calculation from the weight of the metal parts and their specific heats; the second is by the combustion of a pure sub- stance, such as sugar, benzoic acid or naphthalene, whose heating value is known; and the third is by the addition to the calori- meter of a definite amount of hot water with the determination of the rise in temperature resulting; the fourth is by the input of a definite amount of electrical energy. The first method is simple but of only approximate accuracy. The second method has the advantage of tending to compensate for any errors such as the oxidation of nitrogen in combustion and even errors in the thermometer and in radiation corrections in so far as these are constant. The third method has the advantage of being an absolute one, but it is difficult to get it accurate. The fourth method is accurate but requires considerable equipment, such as described by Dickinson. 1 First Method. The first method requires simply that the weights of each of the several different materials contained in the bomb, the stirrer, the thermometer and the water-containing vessel be known. By multiplying these weights by the specific heats as given in the following table the number of calories is obtained directly. TABLE OF SPECIFIC HEATS* Sp. ht. Tool steel. 0. 1087 Gun steel 0.1114 German silver . 094 Platinum . 032 Lead 0.030 Oxygen (constant vol.) 0. 157 Brass 0.094 Mercury . 033 Glass 0.19 The inaccuracy of the method lies partly in the fact that it is not possible to determine the individual weights of each of these constituents e.g., the mercury in the thermometer, and partly in the fact that not all of the materials thus weighed are heated in actual practice to the temperature indicated by the thermom- 1 Scientific Paper No. 230, Bureau of Standards. 2 Atwater and Snell, /, Am, Chem, Soc., 25, 694, 278 GAS AND FUEL ANALYSIS eter immersed in water. A large part of the thermometer is outside of the calorimeter, a part of the stirrer is, in some types of apparatus, constantly passing in and out, and the top of the calorimeter vessel although within the calorimeter is not in con- tact with the water. On the other hand there is some transfer of heat from the calorimeter vessel to its jackets of which no account is taken. Fortunately all these errors are minor ones but the method can hardly be considered accurate within 3 per cent. Second Method. The method of determining the water value of a calorimeter by the combustion of a substance of known heat- ing value is the most commonly employed and the most reliable one. Sugar, naphthalene and benzoic acid are substances which are readily obtained in a state of purity and whose heating value has been determined by a number of independent observers. Dickinson 1 has accurately determined the heats of combustion of these substances and gives the following figures as the heat of combustion per gram of substances weighed in air: Sucrose 3940 + 2 calories (20) Naphthalene 9622 + 2 calories (20) Benzoic Acid 6329 + 2 calories (20) Benzoic acid is recommended as the most desirable for a com- bustion standard. The procedure in determining the water value is exactly the same as for the combustion of a fuel. A sufficient amount of the pure material is pressed into a pellet so that the heat evolved by its combustion will be 7000-8000 calories. This is placed in the bomb which is charged with oxygen, set in the calorimeter and fired, the temperature readings being made as usual. In the final calculations the unknown to be solved for is the mass of water equivalent to the calorimeter which has been heated. The difference between this value and the mass of water actually added gives the water value of the calorimeter. The method of conducting the combustion is the same as that for coal and should be recorded according to the form, 8. The following example gives the method of calculation. i Scientific Paper No. 230, Bureau of Standards (1914). HEATING VALUE OF COAL BY THE BOMB CALORIMETER 279 Total heat evolved from benzoic acid 1 .0856X6320 = 6870 calories from iron wire 0.022 X1600 = 35 6905 Corrected rise in temperature 2.854 C. Heat absorbed by water 2000X2.854 5708 Heat absorbed by calorimeter 1197 calories 1 107 Water value =^^ = 419 The accuracy of this process is dependent first on the purity of the materials used as a standard. Samples of pure substances with certified heat value should be obtained from the Bureau of Standards. The accuracy is also affected by errors in the ther- mometer, errors due to oxidation of nitrogen, etc., but in this very fact lies one of the valuable points of the method. For if it be assumed that with a thermometer set at a given zero there is an error of 0.02 in a rise of three degrees and that there is a correction of 8 calories to be made with oxygen from a certain tank when a sample of benzoic acid which gives a rise of 3, is burned, and both of these corrections be neglected, it is evident that the water value obtained will be in error. But if this erron- eous water value be used in the calculations of the heating value of a coal where the errors due to the thermometer and the oxi- dation of nitrogen are the same as in the combustion of sugar, and where the total rise in temperature is approximately the same, the erroneous water value will compensate for the errors on the coal test and the result of the coal test will be correct. 14. Adiabatic Calorimeters. Corrections for transfer of heat between the calorimeter and its jacket must always be made with the ordinary type of instrument. If the calorimeter bomb is placed in a Dewar vessel, the heat interchange may be lessened but not eliminated. Corrections may only be avoided by com- pletely surrounding the calorimeter by a vessel whose tempera- ture changes constantly during the test to keep pace with the changing temperatures in the calorimeter. The temperature of the jacket may be varied by addition of warm water or by elec- trical means. Such adiabatic calorimeters are somewhat com- plicated in operation but allow simple calculations. They do not necessarily give more accurate results. 280 GAS AND FUEL ANALYSIS 15. Precision Calorimetry. The requirements for precision in calorimetric work have been considered by Dickinson 1 and W. P. White. 2 It is useless to try to attain precision merely by in- creasing accuracy of thermometric readings as by substitution of an electric for a mercurial thermometer. All sources of error must be reduced to a minimum and correction must be made for them. When it is considered that the error in sampling coal is usually over one per cent., it will be seen that methods of high precision are not called for in coal calorimetry. 16. Accuracy of Results. It is the aim to determine by com- bustion in the bomb calorimeter the amount of heat which would be evolved by the combustion of a fuel in the outside air. This standard is not an absolute one, for not only will carbon be burned to carbon dioxide and hydrogen to water in an ordi- nary fire but also sulphur will be burned, in part to SO 2 and in part to SO 3 , and small amounts of nitrogen will be burned. In applying corrections to the figures obtained in the bomb calor- imeter it is customary to assume that all of the sulphur of the coal burns to SOs in the bomb, and that all of it burns to SO 2 in the air, and that no nitrogen is oxidized on combustion of coal in air. The errors introduced by these assumptions are small and are usually neglected. The accuracy of the estimation of the amount of heat evolved by combustion in the bomb will depend on the accuracy with which the water value of the system is known, the care taken in making radiation corrections and the accuracy with which the rise in temperature is measured. This last usually involves the largest error. If the error is 0.01, it will amount to about 0.3 per cent, or 40 British thermal units. When care is taken in every detail and apparatus of superior quality is used the agreement between duplicate determinations will be closer than this but it is certainly not safe to claim a closer absolute accuracy since according to Jesse 3 the highest authorities differ by 0.25 per cent, as to the absolute heating value of sugar and benzoic acid. The Committee on Coal Analysis states that in its judgment results obtained by a single analyst should not differ more than 0.3 per cent, and that 1 Scientific Paper No. 230, Bureau of Standards. 2 Jour. Franklin Inst., 186, 279 (1918). 3 Jour. Ind. and Eng. Chem., 4, 748 (1912). > HEATING VALUE OF COAL BY THE BOMB CALORIMETER 281 results obtained by different analysts should not vary by over 0.4 per cent. This high standard can only be attained when every precaution is observed. 7. Total and Net Heating Values. The heating value of the fuel computed by the method given above gives the total heat developed when the water formed condenses to a liquid within the calorimeter. This gives the total heat. In most industrial operations the water escapes as steam, and if deduction is made for its latent heat, a lower net heating value is obtained. This net figure gives a closer approximation to the heat which is ordi- narily utilized, but it does not give it accurately, because an arbitrary assumption has been made as to the temperature of the escaping gases. It is customary to report the total heating value of the coal and allow the consumer to put such a factor on it as will indicate its relative efficiency for the purpose to which he intends to put it. The net heating value is obtained by de- ducting from the total heating value the latent heat of the water present in the fuel or formed in combustion. The water is determined from the hydrogen as shown by ultimate analysis. The formula for calculation of net heating value is: Net heating value = total B.t.u. 1040 (hydrogen X9) Attempts have frequently been made to calculate the total heating value from the proximate analysis. Very little success has attended these efforts, but Fieldner and Selvig 1 have shown that the correction to be applied to the total heat of combustion to obtain the net heating value may be calculated with consider- able accuracy. By the use of curves constructed from 2000 analyses, the hydrogen content of bituminous coal, semi-bitu- minous coal and anthracite may be estimated from the volatile matter to within 0.6 per cent. With sub-bituminous and lig- nitic coals the error is somewhat greater. The original paper must be referred to for details. 1 Technical Paper 197, Bureau of Mines. CHAPTER XVII HEATING VALUE OF COAL BY THE PARR CALORIMETER AND OTHER METHODS 1. Introduction. -The preceding chapter was devoted to a discussion of the bomb calorimeter as the standard instrument for the determination of the heating value of coal. The present chapter will deal with other methods, such as combustion in a stream of oxygen, combustion with chemicals like sodium per- oxide, and calculation of the heating value from the chemical composition of the coal. 2. Combustion in a Stream of Oxygen. Calorimeters of this type have become obsolete on account of difficulties of manipu- lation and sources of error. The temperature of the oxygen flowing in must be accurately measured and also the temperature of the gases flowing out, for correction must be made for the heat which these streams of gas carry. The great source of error is the incomplete combustion of the coal. With bituminous coals smoke may frequently be seen issuing from the instrument and even with anthracite and coke, carbon monoxide may always be found in the escaping gases. Accurate results have been obtained with this type of calorimeter but only after laborious correction for the large number of errors. 3. The Thompson Calorimeter. A very crude form of calor- imeter which has also become obsolete was that of Lewis Thomp- son. He mixed powdered co'al with potassium chlorate and nitrate, placed the mixture in a calorimeter vessel and fired the charge. The method and apparatus were crude throughout but the greatest source of error lay in the heat absorbed in the de- composition of the chlorate and nitrate. When coal burns under a boiler it unites with gaseous oxygen to form CO2 and H 2 O. Essentially the same result takes place in a bomb calorimeter. When, however, the oxygen is taken from one form of chemical combination and is made to combine with the coal to form a different compound, the result is not at all the same as that 282 HEATING VALUE OF COAL BY THE PARR CALORIMETER 283 obtained in the combustion of the coal with gaseous oxygen and the corrections to be applied must be worked out with great care. Scheurer-Kestner 1 determined that if 15 per cent, was added to the heating value obtained with the Thompson calorimeter, the results never differed by more than 4 per cent, from those ob- tained by the Favre and Silverman calorimeter which burns the coal in a stream of oxygen. 4. The Parr Calorimeter. Parr 2 proposed sodium peroxide as a chemical to be used in oxidizing coal in a calorimeter, worked out the corrections to be applied, and devised a very practical calorimeter. He writes the probable reactions in the calorimeter as follows : 2Na 2 O 2 + C = Na 2 CO 3 + Na 2 O Na 2 O 2 + Na 2 O+ 4H+ O = 4NaOH. There is more heat evolved in each of these reactions than in the combustion of carbon and hydrogen with gaseous oxygen but fortunately the reduction factor is closely the same for both of them. The heat evolved by the combustion of carbon and hy- drogen in the Parr calorimeter multiplied by 0.73 gives the true heat value. Smaller corrections are to be made for the dissocia- tion of the KClOs used, for the oxidation of sulphur, the com- bustion of the fuse wire, the fusion of the ash, and the hydroxyl or combined water present in coal. Since the oxygen is introduced as a solid and the Na2COa and NaOH formed in the reaction are also solids, the bomb need not be made to resist high gas pressures but may be made of thin metal. The general arrangement of the calorimeter is shown in Figs. 58 and 59. Fig. 58 shows a section of the fibre buckets which act as heat insulators and of the can which holds the water. The fusion cup is shown in position. The stirring is accomplished very effectively by the removable wings attached to the bomb which force the water down the annular space between the fusion 1 Bull Soc. Ind., Mulhouse, 506, 1888. J Jour. Am. Chem. Soc., 22, 646 (1900). Jour. Am. Chem. Soc., 24, 167 (1902). Jour. Am. Chem. Soc., 29, 1606 (1907). The Chemical Engineer, 6, 253 (1907). /. Ind. and Eng. Chem., 1, 673 (1909). 284 GAS AND FUEL ANALYSIS cup and the centering cylinder, out of openings at the bottom and up again on the outside. Details of the fusion cup are shown in Fig. 59, where C is the brass fusion cup closed at the top by the headpiece and the rubber gasket G. The outer shell A and the removable bottom B are separated by an air space from the fusion cup which pre- vents too rapid cooling of the charge during the first stages of fusion. FIG. 58. Parr calorimeter. FIG. 59. -Details of Parr calorimeter. 5. Preparation of Parr Calorimeter. The bomb must be thoroughly dry and the gasket in good condition. It is best to dry the parts after each test in an oven and to examine them carefully before putting them together. The lower cap B is fitted into place, the fusion cup is inserted in its shell, and the head piece firmly screwed down with the wrench provided. Water leaking into the bomb always spoils the determination and may cause an explosion. HEATING VALUE OF COAL BY THE PARR CALORIMETER 285 The coal is to be ground to pass a 100-mesh sieve in order that it may react rapidly with the peroxide. Anthracite coal, coke, semi-bituminous and eastern bituminous coals in which the moisture will not exceed 3 per cent, may be used in an air-dry condition. Other bituminous coals, lignites, peats, etc., must be dried at 105 C. before use to avoid the reaction between the hygroscopic moisture and the peroxide with evolution of heat in the calorimeter. The danger of change in the coal sample during the processes of drying and grinding is treated in Chapter XV on Chemical Analysis. It is necessary to secure very intimate mixture of the coal, the Na2O 2 and the KC10 3 added as an accelerator of combustion. The charge consists of 1 grm. of the dry and finely ground KC1O 3 , 0.5 grm. of the coal prepared as directed above and approxi- mately 10 grm. of sodium peroxide which may be measured with sufficient accuracy in the scoop provided with the instrument. The contents of the bomb are next to be thoroughly mixed by shaking. If this were done after the regular top and firing wire were in place the firing wire would almost certainly become twisted and short circuited, so it is better to use the false top provided. It is well at the beginning of the shaking process to invert the cartridge and tap it sharply on the desk to dislodge any coal which may have stuck to the bottom. When the mixing is complete the bottom of the cartridge may be tapped lightly against the desk to dislodge any material sticking to the cap. The regular top to which about 3 in. of fine iron wire (32 or 34 American gage) has been attached in a loop as shown in Fig. 59 is now placed carefully in position and screwed into place. Care should be taken after this has been adjusted, not to tip the bomb since the fine iron wires are easily crossed. The spring stirring clips may now be adjusted and the apparatus set up as shown in Fig. 58. The strength of the firing current will vary between 2 and 4 amperes. It should be adjusted by trials in the open air until the wire fuses promptly on closing the switch. 6. Care of Sodium Peroxide. Sodium peroxide is hygroscopic and absorbs moisture from the air, even when preserved in ap- parently well-stoppered bottles, forming Na 2 O 2 .2H 2 O. The effect of this hydrate formation is illustrated by an experiment 286 GAS AND FUEL ANALYSIS made by Professor Parr. He exposed 10 grm. of sodium peroxide on a watch glass in the laboratory for an hour and found that it had gained in weight nearly 0.5 grm. This peroxide when used in the calorimeter gave a rise in temperature higher by 0.194 than the pure peroxide. As the total correct rise of temperature in this experiment was only 2.180 it will be seen that the error was almost 9 per cent. The sodium peroxide should not only be pure and anhydrous but should also be in grains of the proper size. If the peroxide is too coarse the powdered coal tends to sift to the bottom of the bomb and escape combustion. If peroxide is too fine the reac- tion is sometimes very violent. The manufacturers of the calor- imeter furnish reliable peroxide in small hermetically sealed cans and also furnish a clamp-top fruit jar which is said to preserve the contents of a single can during its use. Sodium peroxide reacts at ordinary temperatures with all organic substances in presence of moisture with evolution of heat often sufficient to produce flame or explosion. Mixtures of sodium peroxide and coal which may have to be disposed of are not to be thrown into waste jars. They may be cautiously and slowly poured into a vessel containing considerable water which will absorb the heat and prevent violent reactions. Sodium peroxide causes bad burns on the skin. 7. Operation of Parr Calorimeter. The fusion cup prepared as above directed is placed in the calorimeter vessel and properly centered. Two thousand grams of water are then added to the calorimeter vessel. It is preferable that the temperature of the water should be about 2.0 C. below room temperature for the radiation correction will be less under these conditions. The thermometer is adjusted and the bomb started to rotating at the rate of about 150 revolutions a minute. Temperatures are to be read at the end of each minute. Within two or three min- utes the readings of the thermometer should become almost constant except for the regular and very slight change due to radiation. Should the thermometer rise irregularly and more rapidly than 0.01 or 0.02 per minute during this preliminary period, there is probably a slight leak of water into the bomb. The operation must be at once stopped, the bomb taken out, wiped dry on the outside, opened and emptied. A leaky bomb is HEATING VALUE OF COAL BY THE PARR CALORIMETER 287 not only inaccurate but dangerous. After five readings at inter- vals of one minute each in the preliminary period have shown only this slight and regular change of temperature, ignition is effected by closing the switch on the cover of the instrument. The thermometer rises very rapidly owing to the thin walls of the fusion cup and the efficient stirring and usually reaches its maximum after two minutes. Nevertheless observations should be continued for at least eight minutes after ignition to allow corrections for radiation to be made. The bomb is then removed from the water, wiped dry and opened. There should be no trace of unburned carbon visible, nor any odor. The lower plug is removed and the inner vessel with its fused contents placed in a casserole containing about 500 c.c. of water, until all the fused peroxide has dissolved. It is then removed, rinsed out and dried. The solution in the casserole is to be tested for unburned car- bon. As an alkaline solution it contains black flakes of oxides of iron and copper. After acidification it becomes a clear yellow solution, with carbon as the only matter in suspension, the silicic acid remaining in solution in the large volume of water. If any unburned carbon is visible it should be filtered on a Monroe or Gooch crucible, dried and weighed. A correction of 8.1 calories must be made for each milligram of unburned carbon. This precaution should never be omitted by a beginner nor where accurate work is important. The corrected rise of temperature may be calculated according to the formulas given in Chapter XVI, but a simpler method will usually suffice since this calorimeter is not used where the great- est accuracy is required. It is sufficiently accurate to assume that the average change in temperature per minute during the final period represents also the change due to radiation for each minute during the combustion period. The errors involved in this assumption are small since the temperature in the combustion period rises to almost its full value in the first minute. The tem- perature at the end of the fourth minute after ignition, may be taken as the end of the combustion period and corrections as shown by the next four minutes' readings, applied to it. Cor- rections for potassium chlorate, sulphur, ash, etc., as given in 288 GAS AND FUEL ANALYSIS the following section are to be deducted from this reading, the result being the corrected final temperature. The older model of instrument furnished by the makers had a standard water equivalent of 135 grm. The corrected rise in temperature multiplied by 2135 and by the factor .73 and divided by the weight of the sample in grams gave the heat value in calories per gram. The figure may be converted into British thermal units per pound by multiplying it by 1.8. The present model is heavier and has a water value of 3100 grams when ready for a test. 8. Corrections to be Applied with Parr Calorimeter. Parr has worked out very carefully the correction to be applied and pub- lished his results in the journals cited as references at the com- mencement of this chapter. He has shown that the ratio of the heat evolved in the combustion of carbon and hydrogen with gaseous oxygen to that evolved on combustion with sodium peroxide is very closely represented by the factor 0.73. The KClOs used as an accelerator evolves an additional amount of heat. Similar corrections are required for ash, sulphur, fuse wire, and hydroxyl constituents of the coal. Since the calor- imeter is always furnished with a standard water value these corrections may be calculated in terms of temperature. The corrections for the earlier model 1 have been modified to conform to the altered water equivalent of the new apparatus and are now given as follows : Electric fuse wire equals 0.0030 C. or 0.005 F. Per cent, ash is multipied by 0.0025 C. or 0.005 F. Per cent, sulphur is multiplied by . 0050 C. or . 010 F. 1 gram accelerator equals 0. 1500 C. or 0.270 F. Hydrogen factors: For all Bituminous coals . 0400 C. or . 070 F. For black lignites 0.0560 C. or 0. 100 F. 9. Accuracy of Parr Calorimeter. The work of Professor Parr has shown definitely that accurate results may be obtained with this calorimeter, but it has also shown that this accuracy can be gained only by the observation of precautions and the use of corrections which deprive the process of much of the simplicity 1 J, Ind, and Eng. Chem., 1, 673 (1909). HEATING VALUE OF COAL BY THE PARR CALORIMETER 289 which formerly characterized it. Accurate results are dependent on the quality of the peroxide used, a point which must usually be taken on faith. The calculations involve a knowledge of the ash, sulphur, and volatile matter of the coal and the application of corrections for these constituents. These points prevent the method from being a standard one as is combustion in the bomb calorimeter, but do not prevent it from being a useful commercial instrument where the highest accuracy is not required. 10. Calculation of the Heating Value from Chemical Analysis. If an ultimate analysis is available, the heating value of a coal may be calculated with fair accuracy from the Dulong formula which is usually given as 8080C+34460 fe~) +2500S Calorific power = JLUU where C, H, and S represent the respective percentages of these various elements shown by the analysis. This formula gives results in Calories per kilogram which when multiplied by 1.8 are converted into British thermal units per pound. Tests of 57 coals made by the U. S. Geological Survey 1 show an average error of 87.5 calories in the calculated result and a maximum error of 312 calories. Inasmuch as it is much simpler to deter- mine the calorific value directly than to make an ultimate analy- sis, the value of this formula has come to lie largely in its confir- mation of the correctness of the ultimate analysis. A formula which would correlate proximate analysis and heating value would be much more useful, but on account of the variable composition of the volatile matter in different types of coal no general formula can be devised which will fit all cases. The combustible matter of coal from a given seam is, however, quite constant in composition and after its value has been experi- mentally determined this figure may be used with considerable accuracy as a basis for the calculation of the heating value of similar coals whose moisture and ash content are known. 1 U. S. G. S., Professional Paper 48 (1906); Bulletin 382 (1909) p. 24. 19 290 GAS AND FUEL ANALYSIS TABLE I. SATURATION PRESSURE OF WATER VAPOR From 0-50 C. Phys. (4), 31, 731 in millimeters of mercury. (1910). Scheel and Heuse, Ann. d. Temp.C. I mm. Hg. Temp. C. mm. Hg. 4.579 26 25.217 1 4.926 27 26.747 2 5.294 28 28.358 3 5.685 29 30.052 4 6.101 30 31.834 5 6.543 31 33 . 706 6 7.014 32 35.674 7 7.514 33 37.741 8 8.046 34 39.911 9 8.610 35 42 . 188 10 9.210 36 44.577 11 9.845 37 47.082 12 10.519 38 49.708 13 11.233 39 52.459 14 11.989 40 55.341 15 12.790 41 58.36 16 13.637 42 61.52 17 14.533 43 64.82 18 15.480 44 68.28 19 16.481 45 71.90 20 17.539 46 75.67 21 18.655 47 79.62 22 19.832 48 83.74 23 21.074 49 88.05 24 22.383 50 92.54 25 23.763 APPENDIX 291 TABLE II. REDUCTION OF GAS VOLUMES TO AND 760 MM. MERCURY PRESSURE AND DRYNESS If the gas is already dry the reduction formula is 1 + . 00367 t 760 where t is the temperature and h the barometric pressure corresponding to the volume V. If the gas is saturated with moisture there must be deducted from the oberved barometric reading the value e for the yapor pressure of water corresponding to the temperature t as given in Table I. The reduction formula then becomes V = h-e 1 + .00367 t 760 The following table gives the values for l+0.00367t for each degree from to 50 C. t l+0.00367t t l+0.00367t 1.00000 26 1.09542 1 1.00367 27 1.09909 2 1.00734 28 1.10276 3 1.01101 29 1.10643 4 1.01468 30 1.11010 5 1.01835 31 .11377 6 1.02202 32 .11744 7 1.02569 33 .12111 8 1.02936 34 . 12478 9 1.03303 35 . 12845 10 1.03670 36 . 13212 11 1.04037 37 . 13579 12 1.04404 38 . 13946 13 1.04771 39 . 14313 14 1.05138 40 . 14680 15 1.05505 41 . 15047 16 1.05872 42 .15414 17 1.06239 43 . 15781 18 1.06606 44 . 16148 19 1.06973 45 . 16515 20 1.07340 46 . 16882 21 1.07707 47 1.17249 22 1.08074 48 1.17616 23 1.08441 49 1.17983 24 1.08808 50 1 . 18350 25 1.09175 odd Tj< 00 1-H t- t 00 00 00 00 000 000 000 00 i-H -^ .-I CM (N o o o o CO CO O5 " ^ t< O3 O5 O5 o o o d d c Tf 00 00 00 o d o I-H 10 00 05 05 05 GO GO GO odd odd rH Tj< 00 O5 O5 03 odd (M (N (N O5 05 05 odd o co co ^ rt< r}< O5 OS 03 odd rjn lO u 05 05 C doc t> O Tt< 00 00 00 00 t-H It5 t^- 00 00 00 00 00 00 O3 O5 00 00 00 odd OS 05 05 odd CO 05 (N eo co Tf 05 05 05 Tf< t>- i-H 00 T* 00 i-l t^ t^ 00 00 00 00 000 O 00 00 00 00 odd o "* J-H O5 O5 O5 odd o 1-H Tjl CO OO 00 00 odd i-H ^ CO t~ l> 00 00 00 odd 00 i-H l> 00 00 CO 00 00 S8SS 00 CO 00 odd o o 00 O5 od o o O3 S S odd o ^ 00 i-H 00 00 00 odd 5 00 IN l>- t^- CO 00 00 00 000 odd O 00 05 03 00 CO ssl odd CO CO ( O5 O5 C odd l> i-H 10 10 iO 00 00 odd i-l IO 00 <* 00 00 00 CO GO d co d 00 00 00 odd OO GO 00 odd O5 05 05 O O O rH .Tt< t co co co 00 00 OO odd t- t^ l> 00 CO 00 05 05 05 00 00 00 o o o O3 03 05 o o o i-l (N C 05 O5 C CO C3 CO T}< IO O 00 00 00 odd 00 00 00 odd GO CO CO odd O5 CM t 00 o o o t> O5 00 00 dd 00 CO 00 odd t^ 00 00 00 00 00 d d d 00 00 00 00 00 d d t^ O CO 10 co co 00 00 CO lO 00 i-l 00 CO 00 00 00 00 00 00 00 (N lO C O5 O5 < 00 00 C o oo i-- 00 00 00 V odd CM 1C 00 l>- t" t OS OS O boo (> O CM OS OS O O CO fl t o o o 000 00 00 OS O3 88 CD O5 rH sss * t^ c - 00 03 03 OS odd rH 2 C C iC 00 O 1C 1C CO odd CO "5 t^ 888 O CO u C 000 000 odd ^ CO 03 00 00 00 OS 03 OS d d d O rH r-I - 1C CO CO odd o r o CO CO t^ OS os OS odd CN C 00 b I s - b OS OS OS odd 00 00 00 OS OS OS d d d CO CO ^ odd C 00 O rt< r< 1C 05 05 OS odd ill odd r~ CD CO CO OS O> OS odd (N <<* t^ t^ OS O3 odd r^ o OS (N 1C 1C CO OS OS O5 o CN 1C t^ d d d h- OS os as as odd S os os odd 00 1-1 * CO TJ4 Tj< O5 OS O3 odd CD -OS i-H -tf ( 1C O5 OS O5 il 000 000 OS OS 05 d d d CO O5 r- O5 OS C d d c OS OS OS odd OS OS 05 OS O3 d d d O CO 1C OS OS OS d d d 00 rH CO odd CO 1C N t- t>- t> O3 O3 C d d c os os a. odd OS - O cN H (N CN OS OS OS ift t^ N CN OS OS CN UJ 00 co co ro 09 0) 0> odd -I CO C Tj< T}4 T}< 05 OS OS 1C t> O 1C U5 CD OS OS O3 d o' d (N 1C co co OS OS si's o o odd O 00 I CO O b isi S M 10 UJ CO O 00 sss T}< I l^ OS IN Tf* | TX r< iO 00 -< CO H ?1 IN 000 O CO OS 3 05 odd t^ OS CN OS OS OS odd d o d CO OS i-l r-- t oo OS OS OS CO 00 - 00 00 O CC 88 T}< J> OS ? odd ^H CO CO odd ^ CO OS GO 00 OO 10 CO CO o o o O t* OS O O iO 000 CN M CO o o o 00 o o o 00 i-i CO S88 00 O Tt* O o o -H * SSS t OS (N CO CO Tf o o o CO CO o o ssl (N Tt< t* o o o fH CO III 00 O N S8.8 33 -< CO >O sss II ll CO CO O rH ^ IN o o o CN "5 t^ o o o o o O t* 88 t- OS t* t o o CO 5 00 t>- t b. o o o CO b o o OS (N * CO ^ t^ o o o s o o III 00 O CO ^t o >o o o o CO CO 00 000 rH CO 10 O .cooocot^ococoooot^t> t>-t>l > t0000000000 iOOCOI>t>t>-t>- Oi-i(NCO'?tiiOOcOt>.r^OOOO OOOOOOOOOOOOOOOOOOOOOOOO OOCOOO5'-iC s; l < ^OOl>-tOOO5OSOOO'-i' I Otl>t>.OOOOOOOOOOOOOOOOOOOOOSOiO5O5O5 O5'-iC y: lCOCO 1 ^Ttt> OOO5O5O5O5O5O5OSO5O5O5O5O5O5OSO5O5O5O5 7 s CO 10 9 58 (N IO 00 rH * CO CO O5 ^^ CO .... . . 43 3 rH IO O5 O M PH^ Ifl ^ *& ^fl 1-4 &3 G 10 ^ b- 1 X ^f 5 CO SI > oooo _ t^" oT o" od co" o* co CN oo o c'-HiOO>C^t^^HTft^ O O O p O C -i -i ^ (N O r-H -< --i O ^H ^ (N (N O"S >U'R ::: 03 ' : : g-| :| : 'f! J2 .ti !?^lsj^jijj 1 S S c 1 : "5L a iilsiii^'3|j "3 c : -sj J B s 80 S 'S 1 O g C^ CJ P3 - O 4H]iiiiniiifiKikJti 'Ms Iiiltilill|tljf &1 6 H!^0^KSfcB!MOOS2odQOfc3 lOrHt*.i6coc^cs O QCO I ) 10 O Ob-t ^ O CO T-< t. U5 CO N 00 00 00 00 CO 00 >-* CO U) t* 00 OS rH -i Tt< CO -^ i i ! I! ; i i i !.i.i i i !. i i.i i ; TTTT! = ; ....... e*t-t-t- * . ^H M CO * U? C l^ (N CO * 1C --H IN CO CO << pp -pppppppppppppppppp -ppppp MrH 1-1 W CO * 1C O N CO * "5 N CO TtO l> 00 OS O i* (N I-H 1-1 C* SS! JSSSSI APPENDIX 303 TABLE X. MEAN SPECIFIC HEATS OF GASES AT CONSTANT PRESSURE IN B.T.U. PER CUBIC FOOT AT 60 F. AND 30 IN. OF MERCURY CALCULATED FOR THE INTERVAL 60 F.-T T, deg. F. Carbon dioxide Water vapor Nitrogen, oxygen and other diatomic gases 200 0.0237 0.0220 0.0174 400 0.0246 0.0220 0.0175 600 0.0253 0.0221 0.0177 800 0.0260 0.0222 0.0178 1000 0.0268 0.0224 0.0180 1200 0.0275 0.0226 0.0181 1400 0.0282 0.0229 0.0183 1600 0.0287 0.0232 0.0184 1800 0.0292 0.0236 0.0186 2000 0.0298 0.0240 0.0187 2200 0.0302 0.0245 0.0189 2400 0.0306 0.0250 0.0190 2600 0.0309 0.0256 0.0192 2800 0.0312 0.0263 0.0194 3000 0.0314 0.0270 0.0196 3500 0.0317 0.0288 0.0201 4000 0.0319 0.0312 0.0206 *^ TABLE XL MEAN SPECIFIC HEATS OF GASES AT CONSENT PRESSURE IN B.T.U. PER POUND CALCULATED FOR THE INTERVAL 60 F.-T T I Carbon dioxide | Water vapor | Nitrogen 200 F. 0.2067 0.4653 0.2365 400 F. 0.2143 0.4657 0.2386 600 F. 0.2216 0.4673 0.2407 800 F. 0.2285 0.4698 0.2428 1000 F. 0.2348 0.4735 0.2449 1200 F. 0.2406 0.4782 0.2470 1400 F. 0.2462 0.4841 0.2491 1600 F. 0.2512 0.4910 0.2512 1800 F. 0.2559 0.4990 0.2534 2000 F. 0.2601 0.5081 0.2555 2200 F. 0.2638 0.5182 0.2576 2400 F. 0.2670 0.5294 0.2597 2600 F. 0.2698 0.5420 0.2618 2800 F. 0.2722 0.5557 0.2639 3000 F. 0.2742 0.5702 0.2660 3500 F. 0.2770 0.6093 0.2707 4000 F. 0.2790 0.6599 0.2755 304 GAS AND FUEL ANALYSIS TABLE XII. VOLUME OF WATER VAPORS TAKEN UP BY ONE CUBIC FOOT OF AIR WHEN SATURATED AT VARIOUS TEMPERATURES Temperature Cubic feet of water vapor 0F. 12 F. 22 F. 32 F. 42 F. 52 F. 62 F. 72 F. 82 F. 92 F. 102 F. 0.001 0.002 0.004 0.006 0.009 0.013 0.019 0.027 0.038 0.053 0.073 TABLE XIII. COMPARISON OF THE BAUME SCALE FOR LIQUIDS LIGHTER THAN WATER AND SPECIFIC GRAVITIES Degrees Baum6 Specific gravity Degrees Baum6 Specific gravity Degrees Baum6 Specific gravity 10 1.0000 36 0.8433 62 0.7290 i 0.9929 0.9859 0.9790 37 38 39 0.8383 0.8333 0.8284 63 64 65 0.7253 0.7216 0.7179 14 0.9722 40 0.8235 66 0.7142 15 0.9655 41 0.8187 67 0.7106 16 0.9589 42 0.8139 68 0.7070 17 0.9523 43 0.8092 69 0.7035 18 0.9459 44 0.8045 70 0.7000 19 0.9395 45 0.8000 71 0.6965 20 0.9333 46 0.7954 72 0.6931 21 0.9271 47 0.7909 73 0.6897 22 0.9210 48 0.7865 74 0.6863 23 0.9150 49 0.7821 75 0.6829 24 0.9090 50 0.7777 76 0.6796 25 0.9032 51 0.7734 77 0.6763 26 0.8974 52 0.7692 78 0.6731 27 0.8917 53 0.7650 79 0.6699 28 0.8860 54 0.7608 80 0.6667 29 0.8805 55 0.7567 81 0.6635 30 0.8750 56 0.7526 82 0.6604 31 0.8695 57 0.7486 83 0.6573 32 0.8641 58 0.7446 84 0.6542 33 0.8588 59 0.7407 85 0.6512 34 0.8536 60 0.7368 90 0.6364 35 0.8484 61 0.7329 95 0.6222 SUBJECT INDEX Absorption methods in gas analysis, 28-42 Acetylene, absorption of, 85 inhibiting reaction between phosphorus and oxygen, 32 solubility in water, 85 Adiabatic coal calorimeters, 279 Air and water vapor, table of vol umes, 304 Air, determination of relative hu- midity, 113 dissolved in water, change in composition, 7 required for combustion of one pound carbon, 150 impure affecting candle power, 136 Alcohol as fuel, see liquid fuel Alkaline pyrogallate, reagent for oxygen, 33 Ammonia, estimation in illuminating gas, 179 Ammoniacal copper solution as re- agent for oxygen, 35 Amyl acetate for photometric pur- poses, 125 Analysis, see coal, gas, etc. Argand gas burners, 128 Argon group of gases, 91 Ash in coal, composition, 230 corrected, 257 standard method of analysis, 245 Aspirators, 5 Automatic instruments for analy- zing chimney gases, 72 Bar photometer, 120, 132 Barium hydroxide as reagent for carbon dioxide, 84 20 305 Baume* scale for liquids lighter than water, table, 304 Benzene in illuminating gas, 166 and light oils in illuminating gas, 167 inhibiting reaction between phosphorus and oxygen, 32 Benzoic acid as standard in calori- metry, 278 Blast furnace gases, sampling, 144 Bomb calorimeter, 258 see also heating value of coal British thermal unit defined, 104 Bromine water reagent, 29 Bulbed gas burette for exact analy- sis, 79 Bunte's gas burette, 68 Burners, standard gas, 128 Calculation of candle power, of explosion analysis, of heat lost in chimney' 150 of heating value of coal, 273 of heating value of gas, 103, 107, 117 of volume of chimney gases, 146, 149 Calibration of gas burette, 22, 83 of gas meter, 94 Calorimeter, see heating value, for coal, 258 for gases, 97 for oils, 190 Calory, definition of, 104 Candle power, accuracy of photo- metric work, 137 atmospheric conditions influ- encing, 137 bar photometer, 120 Bray's burner, 129 Bunsen screen, 129 306 SUBJECT INDEX Candle power, calculation, 135 candle balance, 123 candles as standards, 122 decreasing significance of, 119 details of a test, 134 Edgerton standard, 127 Elliot lamp, 127 equipment of photometer bench, 132 flicker photometer, 132 gas meter, 132 Harcourt lamp, 126 Hefner lamp, 123 humidity of air, 136 impure air, 137 incandescent electric standard, 121 jet photometer, 137 Leeson screen, 129 Lummer-Brodhum screen, 129 method of rating, 120 Metropolitan No. 2 burner, 129 mtane lamp, 126 >meter room, 136 >metric units, 122 Sotometer bench, 132 saturating water of meter, 132 secondary standards, 127 setting flow of gas, 135 solubility of ilium inants, 132 standard candles, 122 standard gas burners, 128 standard lights, 121 Sugg D burners, 127 table photometer, 121 types of photometer, 119 units, 120 Capillary tube preventing explosion, 52 Carbon in coal, method of deter- mination, 251 Carbon dioxide, determination of, 28 Carbon dioxide, formation in chim- ney gases, 145 Carbon dioxide, recording instru- ments for chimney gases, 72 Carbon dioxide, specific heat, table, 303 Carbon monoxide, causing change in blood, 86 combustion with copper oxide, 56 estimation of by acid cuprous chloride, 36 estimation of by ammoniacal cuprous chloride, 38 estimation of by iodine pento- xide, 86 evolved from pyrogallate solu- tion, 33 explosion analysis, 50 formation in chimney gases, 147 fractional combustion with cop- per oxide, 56 fractional combusion with pal- lodinised, asbestos, 53 hydrogen and methane, simul- taneous explosion, 51 initial combustion temperature, 56 incomplete absorption, 37 quiet combustion with oxygen, 51 small quantities in air, 60 Carbonates in coal ash, 230, 245 Carbonic acid, see carbon dioxide Carbonic oxide, see carbon monoxide Carburetted water gas, composition, 165 Caustic soda, reagent, 28 Chemical analysis, see coal, gas, etc. Chimney gases, 144 calculation for loss of heat in, 150 calculation of volume changes, 146, 148 carbon dioxide formation, 145 carbon dioxide percentage, 154 carbon monoxide, 154 carbon monoxide formation, 147 change in composition in con- tact with water, 7 SUBJECT INDEX 307 Chimney gases, continuous record- ing instruments for analysis of, 72 effect of hydrogen in coal on composition, 145 interpretation of analysis, 153 loss of heat in, 150, 154 loss of heat due to water vapor, 151 nitrogen percentage, 154 oxygen percentage, 1M problem illustrating loss of heat in, 152 sampling, 144 solubility in water, 7 specific heats, table, 303 volume of air per pound of car- bon, 148 volume of, per pound of car- bon, 148 Chollar tubes for gas analysis, 70 Clinkering properties of ash, 232 Coal analysis, 222 accuracy of results, 238 air-drying, 223 air-drying sample, 223 ash composition, 230 ash fusibility, 232 ash, standard method for, 245 carbon, total, 251 combined water in, 151 corrected ash, 257 deterioration of samples, 227 Eschka method for sulphur, 247 fixed carbon, 232, 247 fusibility of ash, 232 grinding and preserving sample, 25 hydrogen, 251 method of reporting, 237 moisture, 227 moisture, standard method, 243 nitrogen, 237 . nitrogen, standard method, 255 oxygen, 237, 256 phosphorus, 237 phosphorus in ash, 250 Coal analysis, preliminary examina- tion of sample, 223 preparation of laboratory sam- ples, 240 preservation of sample, 226 proximate analysis, 222 slate and pyrites, 239 standard method, 240 sulphur, 232 sulphur, standard method. 247 true coal, 238 ultimate analysis, 236 ultimate analysis, standard method, 251 volatile matter, 228 volatile matter, standard method, 245 Coal, briquetting sample for bomb calorimeter, 262 changes after mining, 211 changes in air-drying, 224 chemical analysis, see coal analysis combined water, percentage, 151 difference in composition of lump and fine, 204 effect of hydrogen in, on chim- ney gases, 145 grinding, see coal sampling heating value of, see heating value washing, 239 Coal gas, see gas analysis, heating value, candle power, etc. Coal sampling, 204 accuracy, 211 a scoopful as a sample, 207 difference in composition of lump and fine coal, 204 from cars, 209, 214 from wagons, 209, 214 grinding and preserving sample, 218, 227 influence of slate, 208 mine sampling, 209 preparation of sample, 210, 215 308 SUBJECT INDEX Coal sampling, preservation of sample, 210, 218 reducing gross sample, 215 reliability of samples, 221 standard method, 214 size of sample, 207, 215 taking of sample, 209 variation in results, 211 washing, 239 Coal tar, see liquid fuels, Coke, chemical analysis, 220, 257 Coke, sampling, 219 specification of chemical com- position, 221 standard method for chemical analysis, 257 standard method of laboratory sampling, 257 Coke oven gas, composition, 165 Colorimetric tar determination, 142 Combustion, see gas analysis, chim- ney gas, producer, gas, etc. tabular data on volumes, heats, etc., 302 Constants of gases and vapors, table, 301, 302 Continuous recording instruments for analyzing chimney gases, 72 Conversion factors for heating value of gas, 104 Copper oxide method of fractional combustion, 56 Copper solution, ammoniacal as re- agent for oxygen, 35 Corrections in gas volume for tem- perature and pressure, 75, 95, 291, 292-5 Cuprous chloride, acid solution, 36 ammoniacal solution, 38 preservation of, 36 reagent for carbon monoxide, 36, 86 reagent for oxygen, 35 regeneration, 36 Cyanogen in illuminating gas, 179 Diffusion, errors due to, 84 prevention of, 84 Distillation test for gasoline, 200 Dulong formula, 289 Electrical precipitation of suspended particles, 143 Ethane, analysis by combustion with copper oxide, 59 variation from gas laws, 91 Ethylene, absorption of, 29 initial combustion temperature, 56 Exact gas analysis, see gas analysis Explosion analysis, see gas analysis Explosion, not prevented by capil- lary tube, 52, 55 Filters for solid particles in gas, 141, 142 Fixed carbon in coal, analysis, 232 standard method of analysis, 247 Flash point of oils, 193 Flue gases, see chimney gases Form of record of gas analysis, 61 heating value of gas, 105, 109 Fractional combustion, see gas analysis Fuels, liquid, see liquid fuel Fuel oil, 202 Fuming sulphuric acid for olefines, 29 Fusibility of coal ash, 232 Gas, analysis of, see gas analysis Gas, candle power of, see candle power Gas, heating value of, see heating value Gas, suspended particles, see gas analysis Gas, table of properties, 301, 302 Gas, table of specific heats, 303 Gas, see also illuminating, natural, producer, etc. SUBJECT INDEX 309 Gas analysis, absorption methods, 28-42 accuracy of technical, 27 acetylene, 85 , alkalinity of burette water, 26 apparatus Allen-Moyer form, 67 Bunte's, 68 Chollar's, 70 Orsat's, 65 White's, 16, 43, 77, 79 calibration of burette, 22, 83 carbon dioxide, 28, 84 carbon monoxide by absorp- tion, 36 carbon monoxide by explosion, 50 carbon monoxide, hydrogen, and methane by explosion, 50 carbon monoxide by iodine pen- toxide, 86 combustion methods, 43, 51 continuous recording instru- ments, 72 corrections for temperature and pressure, 75 details of simple analysis, 25 errors due to diffusion, 84 errors in explosion of hydrogen and methane, 88 exact methods, 74-91 explosion analysis, accuracy, 48 explosion analysis, calculations, 49 explosion pipette details, 43 explosion analysis manipula- tion, 46 explosion analysis, hydrogen and methane, 48 form of record, 61 fractional combustion with cop- per oxide, 56 fractional combustion with pal- ladinised asbestos, 53 gas burette, simple, 16 gas burette, bulbed, 79 Gas analysis, gas pipettes, 24 general methods, 14 general absorption methods, 28 general scheme, 40 historical, 74 humidity, table, 295, 296 hydrogen by absorption, 38 hydrogen by colloidal palla- dium, 39 hydrogen by explosion, 48, 87 illuminating gas, 163 initial combustion temperature of different gases, 56 methane by explosion, 50, 89 minute quantities of combus r tible gas, 60 natural gas, 183 nitrogen, 61 olefines, 29, 85 optical methods, 73 order of absorptions, 41 explosion analysis, oxidation of nitrogen, 47 oxygen by explosion or combus- tion, 61 oxygen, 30-35, 86 oxygen, commercially pure, 32 palladous chloride for hydrogen, 38 pipettes, White's, 24 producer gas, 158 quiet combustion with oxygen, 51 quiet combustion of hydrocar- bons, 52 recording types of apparatus, 72 reduction of volume to standard conditions, 75, 291 saturating burette water, 18 suspended particles, 138 thermal conductivity methods, 73 ' transferring gas from holder, 19 see also carbon monoxide and hydrogen and other indivi- .dual gases ,, 310 SUBJECT INDEX Gas analysis, table of constants of certain gases and vapors, 301, 302 tables for reduction of volume to and 760 mm. dry, 291 tables for reduction of volume to 60 and 30 in. wet, 292-4 tar fog, 138 unsaturated hydrocarbons, 28, 85 variation from gas laws, 90 Gas burette, calibration, 22 cleaning, 17 description of simple form, 15 drawing sample of gas, 18 measuring gas volume, 21 saturating water of, 18 for exact analysis, 79 see gas analysis Gas burners, standard for photo- metry, 128 Gas calorimeters, continuous flow, 92-116 automatic and recording, 117 various types, 116 see also heating value of gas Gas holders, for samples, 13 transferring gas for analysis, 19 Gas mains, sampling from, 140 Gas meters, calibration, 94 description of wet, 93 for candle power tests, 132 Gas pipettes, see gas analysis Gas producers, see producer gas Gas Referees, 122, 126, 128 Gas sampling, 1-13 apparatus for aspirating sample of gas, 9 aspirators, 5 collecting average sample, 8 collecting instantaneous sample, 11 continuous apparatus, 10 errors due to solubility, 7 fair sample, 1 forms of sampling tubes, 2 Gas sampling, glass gas holders, 12 illuminating gas, 163 influence of bends in mains on suspended particles, 140 materials for sampling tubes, 2 point of mean velocity in cross section of gas main, 139 producer gas, 158 saturating water in sampling tubes, 8 solubility in water, 6 storage of samples, 1 1 velocity of gas in sampling tubes, 140 Gas volume, corrections for tempera- ture and pressure, 95 table for reduction to and 760 mm. dry, 291 60 and 30 in. wet, 292-4 defined, 196 Gasoline, in natural gas, 185 specifications, 197 Heat of combustion, table for va- rious materials, 301 Heat of formation, table for various materials, 302 Heat lost in chimney gases, 150 lost in chimney gases problem, 152 Heating value of coal in bomb calori- meter accuracy of results, 280 action of acid on calorimeter bomb, 275 adiabatic calorimeters, 279 calorimeter bomb, 258 calculation of heating value, 273 correction for combustion of iron wire, 275 corrections for oxidation of ni- trogen, 273 corrections for oxidation of sulphur, 274 corrections for radiation, 268- 273 SUBJECT INDEX 311 Heating value of coal in bomb calorimeter details of calorimetric bomb, 259 form of record, 271 manipulation, 263 net heating value, 281 oxidation of nitrogen, 273 oxidation of sulphur, 274 precision calorimetry, 280 preparation of sample, 262 radiation corrections, 267-273 Regnault-Pfaundler formula, 268 Dickinson formula, 272 standard method, 272 reduction to constant pressure, 275 sample of record, 271 thermometers, 262 thermometer corrections, 267 total heating value, 281 water value of calorimeter, 276 Heating value of coal in Parr calori- meter accuracy, 289 care of sodium peroxide, 285 corrections, 288 description, 283 operation, 286 Heating value of coal in Thompson calorimeter, 282 Heating value of coal calculation from chemical analysis, 289 combustion in a stream of oxy- gen, 282 general method of determina- tion, 258 peroxide method, 283 Heating value of gas, 92-118 accuracy of method, 108 automatic calorimeters, 117 calibration of meter, 95 calculated from chemical com- position, 117 Heating value of gas, calculation of observed heating value, 103 calculation of total heating value, 107 calculation of net heating value, 106 gas calorimeters, various types see gas calorimeters conversion factors from metric to British units, 104 corrections to be applied to ob- served heating value of illuminating gas, table, 297 corrections to be applied to ob- served heating values of natural gas, table, 298 corrections for difference be- tween inlet water tempera- ture and room tempera- ture, table, 300 corrections for emergent stem of thermometer, table, 299 corrections for temperature and pressure, 95 description of calorimeter, 97 description of test, 102 errors itemized, 108 form of record, 105, 109 gross heating value, 106 Hempel calorimeter, 116 . humidity of air, influence, 113 illustrations of calculations, 105 Junkers calorimeter, 97 measurement of mass of water, 96 measurement of temperature, 96 net heating value, 106 normal rate of gas flow, 101 observed heating value, 103 preliminaries of a test, 99 recording calorimeter, 117 saturating water in meter, 102 Test Record of Bureau of stand- ards, 109 total heating value, 106 uncondensed water vapor, 1 12 312 SUBJECT INDEX Heating value of liquid fuels, 188 Humidity of air, affecting candle power, 136 Humidity of air, determinations, 113 relative tables, 295, 296 Hefner lamp, 123 Hydrogen, accuracy of estimation affected by explosive ratio, 88 analysis by absorption, 38 carbon monoxide and methane, simultaneous explosion, 51 affecting compositions of chim- ney gas, 145 method of analysis, 251 fractional combustion with pal- ladinised asbestos, 53 fractional combustion with copper oxide, 56 explosion analysis, 48, 87 initial combustion temperature, 56 quiet combustion with air, 51 Hydrogen sulphide arsenious acid as reagent, 72 in illuminating gas, 169 with carbon dioxide, 28 Hydrometers, comparison of Baum6 scale for liquids lighter than water and specific gravity, table, 304 Hydrosulphite as reagent for oxygen, 33 Illuminants, average composition, 85 solubility of, 132 Illuminating gas, 163 ammonia, 179 benzene, 166 benzene arid light oils, 167 candle power, see candle power chemical composition, 164 cyanogen, 179 general scheme of analysis, 164 hydrogen sulphide, 169 naphthalene, 173 sampling, 163 Illuminating gas, scheme of analysis, 164 specific gravity, 181 suspended tar, 179 total sulphur compounds, 170 typical analyses, 165 Incomplete combustion in chimney 'gases, 147 Initial combustion temperatures of various gases, 56 International candle, 122 Iodine pentoxide for carbon mono- xide, 61 Iron, heat of combustion, 275 Jet photometer, 137 Kerosene, flash point, 193 tests, 201 Kjehldahl method for nitrogen in coal, 255 Liquid fuels, 187 distillation test, 198 flash point, 193 fuel oil, 202 gasoline, 196 heating value, 188 kerosene, 201 moisture, 191 proximate analysis, 192 sampling, 187 specific gravity, 191 suspended solids, 192 Ix)ss of heat in chimney gases, 150 Lubricant for stop cocks, 17 Measuring gas in burette, 21 Meter for candle power determina- tion, 132 Meter, wet gas, 93 Methane, accuracy of estimation affected by explosive ratio, 89 carbon monoxide and hydrogen, simultaneous explosion, 50 SUBJECT INDEX 313 Methane, by combustion with copper oxide, 59 by explosion, 48, 89 small quantities in air, 60 variation from gas laws, 91 initial combustion temperature, 56 quiet combustion with air, 51 Mine air, examination, 60 Mine sampling of coal, 209 Moisture in air, see humidity Moisture in coal, 227 standard method of analysis, 243 Moisture in gas, correction for, 75 Moisture in liquid fuels, 191 Naphthalene, estimation in illumi- nating gas, 173 estimation in tar, 176 as standard in calorimetry, 278 ; Natural gas, analysis, 183 analysis by fractional distilla- tion at low temperatures, 185 errors in analysis due to varia- tions from gas laws, 91 typical analysis, 184 gasoline vapors in, 185 specific gravity, 182 Net heating value of gas, 106 Net heating value of coal, 281 Nitrogen in gas analysis, 61, 91 Nitrogen in coal, 237 method of analysis, 255 Nitrogen, oxidation in bomb calori- meter, 273 oxidation in explosion analysis, 47, 88 table of specific heats, 303 defines, absorption of, 29 estimation of mean composi- tion, 85 Optical methods in gas analysis, 73 Orsat apparatus, 65 Oxidation of nitrogen, error caused by in gas analysis, 47, 88 corrections for in coal calori- metry, 273 Oxygen, always present in industrial gases, 31 analysis of commercially pure, 32 determination of by ammonia- cal copper solution, 35 determination of by alkaline pyrogallate, 33 determination of by hydrosul- phite, 33 determination of by phos- phorus, 30 by explosion or combustion, 61 Oxygen and phosphorus, poisons inhibiting reactions, 29, 31 Oxygen, table of specific heats, 303 Oxygen in coal, 237 method of analysis, 256 Palladinised asbestos, 53 Palladinised copper oxide for frac- tional combustion, 56 Palladium, colloidal reagent for. hydrogen, 39 Palladous chloride, reagent for hy- drogen, 38 Parr calorimeter, see heating value of gas see heating value of coal see liquid fuels' Pentane, combustion with copper oxide, 59 Pentane lamp, 126 Permanganate as reagent for re- ducing gases, 84 Peroxide, sodium, care of, 285 Peroxide calorimeter, see Parr calori- meter Peroxide method for sulphur in coal, 233 Petroleum, see liquid fuels, gaso- line, kerosene, etc. 314 SUBJECT INDEX Phosphorus, as reagent for oxygen, 30, 185 Phosphorus as reagent showing ab- sence of unsaturated ^ hydrocarbons, 29, 85 precautions in handling, 30 sometimes inactive, 32 Phosphorus in coal, 237 standard method of analysis, 250 Photometry, see candle power Pintsch gas, 29 Pipettes for gas analysis, see gas analysis Poisons affecting reaction between oxygen and phosphorus, 32 Portable gas analysis apparatus, 64-72 Pressure of gases, correction for, 95, 291, 292, 293 Pressure of water vapor, table, 290 Producer gas, analysis, 158 efficiency of producer, 162 formation, 156 heating value of, 160 sampling, 158 interpretation of analysis, 159 typical analyses, 157 volume per pound of carbon, 161* Proximate analysis of coal, 222 Proximate analysis of liquid fuels, 192 Psychrometer, 113 Pyrogallate, as reagent for oxygen, 33, 86 Quartz combustion tube for copper oxide, 57 Quartz combustion tube with plati- num spiral, 51 Radiation corrections, see heating value of coal, and gas Record form for gas analysis, 61 heating value of coal, 271 heating value of gas, 105, 109 Reduction of gas volumes to 60 and 30 in. table, 292-4 and 760 mm. dry, table, 291 Relative humidity, table, 295, 296 Rubber connections, danger of, in gas analysis, 18 Sampling, apparatus for gas, 1-13 blast furnace gas, 141 chimney gases, 144 crude illuminating gas for naphthalene, 1 75 coal, see coal sampling coal, standard method for laboratory sampling, 240 coke, 219 standard method of laboratory sampling, 257 see gas sampling liquid fuels, 187 producer gas, 158 suspended particles in gas, 140 tubes for gases, 2 Saturating water in sampling tubes, 8 Saturating water of burette, 18 Saturation pressure of water vapor, table, 290 Saturating water in gas meter, 102, 132 Slate, as cause of error in coal sampling, 208 separation by washing, 239 Sodium hydrosulphite, as reagent for oxygen, 33 Sodium hydroxide, reagent, 28 Sodium peroxide, care of, 285 Solubility of gases in water, 6 Specific gravity compared with Baume" scale for liquids lighter than water, table, 304 Specific gravity of gas, determina- tion of, 181 Specific gravity of gases, table, 301 Specific gravity of liquid fuels, deter- mination, 191 SUBJECT INDEX 315 Specific heat of gases, determination of, 150 Specific heats of gases, tables 303 material in bomb calorimeter, 277 Stack gases, see chimney gases Standard method of coal analysis, 240 coke analysis, 257 coal sampling, 214 coke sampling, 216 Standard conditions for measure- ment of gases, 76, 95 Stopcocks, care of, 17 Storage of gases, 11 Sugar as standard in calorimetry, 278 Sulphates in coal ash, 230 Sulphur, analysis of in coal, 232 Atkinson method, 233 Parr's photometric method, 235 Peroxide method, 233 in washings from bomb calori- meter, 235 Sulphur, corrections for oxidation in bomb calorimeter, 274 total in illuminating gas, 170 Sulphur dioxide, removal with car- bon dioxide, 28, 84 Sulphuretted hydrogen, see hydro- gen sulphide Sulphuric acid, fuming as reagent, 29 Suspended particles in gas, 138 Suspended particles in gas, distri- bution in main, 138 Suspended particles in gas, mean velocity, 139 Suspended particles in gas, sampling, 140 Suspended solids in liquid fuels, 192 Suspended tar in gas, estimation, 142 Table photometer, 121 Tar as fuel, see liquid fuel Tar camera, 142 Tar particles in gas, electrical preci- pitation, 143 Tar, estimation of naphthalene in, 176 moisture in, 192 Tar suspended in gas, estimation, 142 Temperature of gases, corrections for, 95 Temperature of initial combustion of various gases, 56 Temperature measurement in calori- metry, errors in, 108, 267, 280 Thermal conductivity as method of gas analysis, 73 Thermometers, table of corrections for emergent stem, 299 Ultimate analysis of coal, 236 standard method, 251 Unsaturated hydrocarbons, deter- mination of, 28, 85 Variations from gas laws, 90 Volatile matter in coal, 228 standard method of analysis, 245 Volume of gases, formula for reduc- tion to standard con- dition, 76 Volume of chimney gas per pound of coal, 148 Volume of gases, tables for reduction to standard conditions, 291-4 Volume of producer gas per pound coal, 160 Water, combined in coal, 151 saturated with air, composition of dissolved gases, 7 Water value of calorimeter, see heating value of coal Water vapor, table of saturation pressure, 290 Water vapor, table of specific heats, 303 316 SUBJECT INDEX Water vapor, volume taken up by Water, weight of one cubic foot at one cu. ft. air, table, various temperatures, 94 304 Weathered coal, changes in, 224, 262 weight percu. ft. of saturated air, Wet gas meter, see gas meter 112 Wet and dry bulb thermometers, 113 Water, weight of one liter at various Wiring rubber connections, 19 temperature, 265 Wiring stopper into burette, 78 INDEX OF AUTHORITIES CITED Allen and Jacobs, 191 Allen and Moyer, 67 American Chemical Society, see Committee on coal analy- sis, American Gas Institute, 108, 111, 136 Anderson, 34, 184 American Society for Testing Ma- terials, 193, 198, 214, 219, 240, 257, 272 Atkinson, 234 Atwater, 259 Atwater and Snell, 277 Badger, 35, 227 Bailey, 208 Barker, 233 Barkley and Flagg, 73 Bartlett, 87 Benton, 10 Berthelot, 32, 258 Blauvelt, 141 Bleier, 79 Brady, 141 Brodhun, 129 Bunsen, 129, 181 Bunsen and Playfair, 74 Bunte, 68, 167 Bureau of Mines, 144, 157, 187, 213, 214, 224, 243 Bureau of Standards, 106, 108, 109, 170, 182, 265, 297, 298 Burrell, 185 Burrell and Robertson, 184 Burrell and Seibert, 73, 90, 185 Campbell, 16, 38, 44, 56 Cheney, 259 Chollar, 70 Church, 192 Coleman and Smith, 174 Committee on coal analysis, 228, 229, 239, 240, 280 Coquillion, 51 Cottrell, 143 Davis, see Field ner Davis and Davis, 167 Davis and Fairchild, 239 David & Wallace, 262 Dennis, 52 Dennis and Hopkins, 88 Dennis and McCarthey, 166 Dickinson, 272, 277, 280 Doherty, 116 Doyere, 74 Drehschmidt, 170 Dykema, 185 Earnshaw, 85, 117 Edwards, 73, 182 Eschka, 232, 247 Fairchild, 239 Favre and Silverman, 283 Fernald and Smith, 157 Field, 232 Fieldner and Davis, 228 Fieldner, Hall and Field, 232 Fieldner and Selvig, 281 Flagg, 73 Franz en, 34 Gas Chemists hand book, 301, 302 Gill, 44, 47 Gill and Bartlett, 87 Graefe, 116 Haber and Oechelhauser, 167 Hall, 232 Harbeck and Lunge, 167 317 318 AUTHOR INDEX Harcourt, 126 Harding, 170 Hart, 38 Hartmann, 39 Hartley, 93 Hefner, 121 Herapel, 38, 39, 51, 53, 55, 75, 116, 166, 258 Henning, 150 Hillebrand and Badger, 227 Hinman, 47 Heuse, 290 Holborn and Henning, 150 Holmes, 209 Hopkins, 52, 88 Hulett, 227 Jacobs, 191 Jaeger, 56 Jenkins, 170 Jones, 36 Jesse, 280 Junkers, 93 Kinnicut and Sanford, 86 Klumpp, 137 Krauskopf, 37 Kreisinger, Augustine and Ovits, 144 Kroeker, 236 Kiister, 174 Lavoisier, 74 Le Chatelier, 150 Leeson, 129 Lord, 239 Lummer, 129 Lunge, 167 McBride, Weaver and Edwards, 169 McCarthy, 166 McWhorter, 87 Mack and Hulett, 227 Mahler, 258 Mallard, 150 Morgan and McWhorter, 87 Morton, 166 Moyer, 67 Mueller, 179 National Fire Protection Associa- tion, 202 Nesmjelow, 55 Nicloux, 86 Noyes, 50 Oechelhauser, 167 Ovitz, 211 Orsat, 35, 65 Paal, 39 Parr, 116, 228, 230, 233, 235, 236, 238, 259, 283 Pennock and Morton, 233 Petterson, 77 Pfaundler, 268 Playfair, 181 Pope, 214 Porter, 225 Porter and Ovitz, 211 Purdy, 37 Ramsburg, 169 Regester, 235, 250 Regnault and Reiset, 74 Regnault and Pfaundler, 268 Reiset, 74 Richards and Jesse, 188 Robertson, 184 Rolland, 64 Rutten, 174 Sanford, 86 Scheel and Heuse, 290 Scheurer-Kestner, 283 Schilling, 181 Schlosing and Rolland, 64 Seibert, 73, 90, 185 Selvig, 281 Shepherd, 50 Silverman, 283 Small, 33 Smith, 157 Snell, 277 Somermeier, 274 Steere, 142 AUTHOR INDEX 319 Stevenson, 259 Sundstrom, 233 Thompson, 282 Threlfall, 139 Tutwiler, 170 U. S. Fuel Administration, 201 United Gas Improvement Co., 301, 302 U. S. Geological Survey, 289 U. S. Weather Bureau, 113 Waidner and Mueller, 106 Wallace, 262 Weather Bureau, 113 Weaver, 73, 169 White, A. H., 16, 44, 53, 77, 88, 139, 175 White, W. P., 280 H UNIVERSITY OF CALIFORNIA LIBRARY This book is DUE on the last date stamped below. Fine schedule: 25 cents on first day overdue 50 cents on fourth day overdue One dollar on seventh day overdue. '=.ti/MG LIBR. ^HY DEC 13 1947 DEC A ? J94X MOV A? 19T 1S50..L ? OCT 25 1952^ '- / DEC DEC 10 1952 DEC **** rfc 2 4 1952 ,) T f NOV 12 1S49/ ' / DEC 2 1349^ / DEC 1 9 1949^ . JAN I S I9so / OGT 6 1950/ !T 2 5 1950 LD 21-100m-12,'46(A2012s 16)4120 . THE UNIVERSITY OF CALIFORNIA LIBRARY