^^^^^^^^S^M^M*im^. ' The original of tiiis bool< is in tine Cornell University Library. There are no known copyright restrictions in the United States on the use of the text. http://www.archive.org/details/cu31924004083345 Cornell University Library TJ 755.C633 V.1 The gas, petrol, and oil engine, 3 1924 004 083 345 THE GAS, PETROL, AND OIL ENGINE THE GAS PETROL, AND OIL ENGINE VOL. I. THERMODYNAMICS OF THE GAS, PETROL, AND OIL ENGINE, TOGETHER WITH HISTORICAL SKETCH DUGALD CLERK, F.R.S., M.Inst.C.E. MEMBER OF THE INSTITUTION OF MECHANICAL ENGINEERS FELLOW OF THE CHEMICAL SOCIETY PRESIDENT 1908 OF THE ENGINEERING SECTION OF THE BRITISH ASSOCIATION PAST PRESIDENT OF THE JUNIOR INSTITUTION OF ENGINEERS PAST PRESIDENT OF THE SOCIETY OF BRITISH GAS INDUSTRIES PRESIDENT OF THE INCORPORATED INSTITUTION OF AUTOMOBILE ENGINEERS MEMBER OF THE ROYAL INSTITUTION NEW AND REVISED EDITION WITH ILLUSTRATIONS JOHN WILEY AND SONS 43 AND 45 EAST NINETEENTH STREET, NEW YORK 1909 All rights reserved PREFACE This work was first published in 1886 under the title of ' The Gas Engine.' Considerable additions were made to it in 1896 and the title was altered to ' The Gas and Oil Engine.' Both the science and prac- tice of Gas, Petrolj and Oil Engines have developed so largely since 1896 that it has become necessary to rewrite practically the whole book. To treat science and practice adequately necessitated consider- able additional space ; and, accordingly, the book has been divided into two volumes, each, however, complete in itself. The first volume has been entitled ' Thermodynamics of the Gas, Petrol, and Oil Engine,' and the second volume will be entitled ' The Gas, Petrol, and Oil Engine in Practice.' The present volume consists of an enlarged historical sketch, broadly dealing with the important developments down to 1908, and the science of the subject is treated in nine chapters. In this treatment the original book is followed closely, so far as arrange- ment is concerned. Chapters L and II. deal with the gas engine method and classification of gas engines. Chapter III. on ' Thermo- dynamics ' has been greatly enlarged to deal more fully with cycles of operation of practical importance to-day. Chapter IV., on ' The Causes of Loss in Gas Engines,' has also been enlarged. Chapter V., on ' Combustion and Explosion,' has been added to. Chapter VI. has been greatly enlarged to include the consideration of Cooling as well as Explosion in a Closed Vessel and the more recent experiments of the Massachusetts Institute of Technology, Grover, Clerk, the Royal College of Science, Petavel, and Hopkinson have been fully dealt with. Information upon the Laws of Explosion and Cooling has greatly increased even within the last six years. The work of the various investigators has been described in a manner which it is hoped will be useful to the engineer. Chapter VII. deals with the Discussion of Data obtainable from the work of various experimenters as to the Laws of vi THE GAS, PETROL, AND OIL ENGINE Explosion and Cooling of Gaseous mixtures in large and small vessels and at initial pressures of atmosphere and above. Chapter VIII. deals for the first time in any work on the subject with Explosion and Cooling in a Cylinder behind a Moving Piston. The recent work by the author enables approximate values to be arrived at for cooling within the Internal-Combustion Engine Cylinder, apart from tem- perature fall due to work done. The last chapter, Chapter IX., discusses the Thermal and Mechanical Efficiency of all the different types of gas engine in use. On this part of the subject more accurate knowledge exists than at any previous time. Important work has been done by English, American, and Continental investigators, and the Research Committees of the Institutions of Civil and Mechanical Engineers have made experiments of great value. All this work has been fully discussed in this chapter. In the present volume the author has attempted to systematise the knowledge existing as to the properties of the working fluid of the Internal-Combustion Engine, whether using gas, petrol, or heavy oil, so as to enable the engineer and inventor to consider not only mechanical modifications of engine construction, but more profound alterations possible by varying the actions going on in the working fluid. Such variations are, in the author's opinion, necessary to enable light and powerful Internal-Combustion Engines to be developed for marine work. For this purpose it is necessary that the engineer should be thoroughly familiar with the properties of the working fluid with which he is dealing. Appendices have been added to make clear many of the properties of gaseous explosions, and the valuable Report of the British Association Committee on Gaseous Explosions has been published in full in Appendix IV. This Report, published at Section G, Dublin, in 1908, contains the latest information available on all the pro- perties of gaseous explosions. The author has freely availed himself of the various important Papers and Reports published by the ^yal Society, the Institution of Civil Engineers, and the Institution of Mechanical Engineers. He is greatly indebted to the Institution of Civil Engineers for permission to use many blocks, and is much indebted also to the Institution of Mechanical Engineers for permission to reproduce figs. 108 to 118 inclusive, PREFACE vii The author has much pleasure in thanking his assistant, Mr. W. Grylls Adams, M.A., for his effective aid in preparing and checking the numerous calculations and curves which are given, as well as reading the proofs. He is also pleased to thank his assistant, Mr. Aubrey T. Evans, for the preparation of the Index. Engineering Laboratory, 6 Featherstone Buildings, High Holborn, London. June igog. CONTENTS CHAPTER PAGE HISTORICAL SKETCH OF THE GAS, PETROL AND OIL ENGINE TO 1908 . . .... I I. The Gas Engine Method . . . .51 II. Gas Engines Classified . . .56 III. Thermodynamics of the Internal-Combustion Engine con- sidered AS an Air Engine . .... 62 IV. The Causes of Loss in Gas Engines . . . 103 V. Combustion and Explosion .... . 109 VI. Explosion and Cooling in a Closed Vessel — Experimental Investigations 126 VII. Explosion and Cooling in a Closed Vessel — Discussion of Data Deducible 200 VIII. Explosion and Cooling in a Cylinder behind a Moving Piston ... 214 IX. Thermal and Mechanical Efficiency of the Different Types of Gas Engine in Use . .... 236 APPENDICES I. Determination of Charge Weight in Internal-Combustion Engines 331 II. Calculation of Adiabatic Lines with Varying Specific Heat of Working Fluid 335 III. Calculation of Efficiency with Varying Specific Heat of Working Fluid , 336 IV. Report of the British Association Committee on Gaseous Explosions, Dublin, 1908 337 V, Adiabatic and Isothermal Compression of Dry Air. . . 368 INDEX .... . . . . 369 LIST OF PLATES PAGE. CocKERiLi. Engines at Differdingen To face 48 Engine-Testing-room at the National Gas Engine Co.'s Works, Ashton-under-Lyne, with Engines ' L,' ' R,' and ' X ' in position „ 258 ' X ' Engine with a Clerk Optical Indicator used for Clerk Diagrams „ 259 Diagrams taken from 40 HP Crossley Engine by_^Hopkinson's Optical Indicator „ 276- Institution of Mechanical Engineers' Tests. Photograph OF Engine experimented upon „ 300- THERMODYNAMICS OK THE GAS, PETROL, AND OIL ENGINE HISTORICAL SKETCH OF THE GAS, PETROL, AND OIL ENGINE The origin of the gas engine is but imperfectly known ; by some it is dated as far back as 1680, when Huyghens proposed to use gun- powder for obtaining motive power. Papin, in 1690, continued Huyghens' experiments, but without success. The method used was a fairly practicable one. The explosion was used indirectly ; a small quantity of gunpowder exploded in a large cylindrical vessel filled with air expelled the air through check valves, thus leaving, after cooling, a partial vacuum. The pressure of the atmosphere then drove a piston down to the bottom of the vessel, lifting a weight or doing other work. In a paper, published at Leipsic in 1688, Papin stated that ' until now all experiments have been unsuccessful ; and after the combustion of the exploded powder, there always remains in the cylinder about one-fifth of its volume of air.' The Abbe Hau'efeuille made similar proposals, but does not seem to have made actual experiments. These early engines cannot be classed as gas engines. The explosion of gunpowder is so different in its nature from that of a gaseous mixture that comparison is unten- able. The first real gas engine described in this country is in Robert Street's patent, No. 1983, 1794. It contains a motor cylinder in which works a piston connected to a lever, from which lever a pump is driven. The bottom of the motor cylinder is heated by a fire ; a few drops of spirits of turpentine being introduced and evaporated by the heat, the motor piston is drawn up, and air entering mixes with the inflammable vapour, the application of a flame to a touch- hole causing explosion ; and the piston being driven up forces the pump piston down, so performing work in raising water. The details VOL. I. 8 2 THE GAS, PETROL, AND OIL ENGINE as described are crude, but the main idea is correct and was not im- proved upon in practice till very lately. Lebon, in France, describes a gas engine in his French patent, No. 356 of September 28, 1799. In it gas and air are supplied from separate compressing pumps to a combustion chamber where the gases are detonated. A motor cylinder is supplied from this chamber with the hot gases under pressure by distributing valves contained in a valve box. Both motor and pump cylinders are double-acting. The engine resembles what became known later as a constant -pressure engine, but the inventor's notions were vague, and he does not distinguish very clearly between explosion and con- stant pressure. He expects, however, to greatly improve upon the steam engine, as he states : ' But experiments on this force teach that the height should be prodigiously superior to that which measures the force of our fire engines.' The engine shown in the patent drawings is very crude ; it could not have been in practical operation. Poor Lebon had but little time to develop his ideas, as he was assassinated in 1804. In the year 1820 the Rev. W. Cecil, M.A., of Cambridge, read a paper at the Cambridge Philosophical Society with the following title : ' On the Application of Hydrogen Gas to Produce a Moving Power in Machinery, with a description of an Engine which is moved by the Pressure of the Atmosphere upon a Vacuum caused by Explo- sions of Hydrogen Gas and Atmospheric Air.' In this paper he de- scribed an engine which he had constructed to operate according to the explosion vacuum method ; and he stated that at sixty revolutions per minute the explosions take place with perfect regularity. His engine consumed, he stated, I7'6 cubic feet of hydrogen gas per hour. His hydrogen explosion appears to have been accompanied by con- siderable noise, because he states wi'h regard to a proposed larger engine, ' ... to remedy the noise which is occasioned by the explosion, the lower end of the cylinder a, b, c, d may be buried in a well, or it may be enclosed in a large air-tight vessel.' In this paper he also mentions an engine operated by non-compression explosion and also one operated by gunpowder. This paper gives an account of the first gas engine which appears to havdibeen worked in Britain, and, it is believed, in the world. Cecil appears to have been the first to attempt to measure the pressures produced by gaseous explosions ; for this purpose he used a tin cylinder, ten inches long by two inches diameter, ' made of thin tin, seamed up one side, and soft-soldered, the ends being well secured. This vessel, he states, will easily sustain without bursting the whole force of an exploding mixture of hydrogen and air : this force he deter- HISTORICAL SKETCH 3 mines as 180 lb. per square inch absolute, as found in the following manner. ' The greatest expansive force was ascertained by filling with mixed gas the cylinder just described, one end being entirely solid, the other being closed with a cork bung accurately fitted, and confined by several strings, parallel to the axis of the cylinder, and so arranged that the tension might be equally distributed. It was observed how many strings the explosion was able to break by pressing on a surface of three square inches. The same strings were then transferred to a common steelyard, and it was observed how much weight they would sustain. The result of several trials, differing but little from each other, indicated a pressure of 500 lbs. upon the three square inches. If to this be added 45 lbs. for the atmospheric pressure on the same surface, the whole being divided by 3 gives 180 lbs. nearly for the pressure upon every square inch. Cecil's result was much too high, but his method of procedure was interesting and ingenious ; his paper is important and shows very considerable knowledge of the problem to be solved. The engine actuated by ' the exploding force of the mixed gas ' alluded to in the paper is stated to have been exhibited in operation ' about three years ago ' — that is, in 1817 — at the philosophical lectures of Professor Parish. Samuel Brown's inventions come next. His patents are dated 1823 and 1826, Nos. 4874 and 5350. The principle used is ingenious and easily carried out in practice, but it is not economical, and it gives a very cumbrous machine for the amount of power produced. A partial vacuum is produced by filling a vessel with flame and expelling the air it contains ; a jet of water is thrown in and condenses the flame, giving vacuum. The atmospheric pressure thus made available for power is utilised in any engine of ordinary construction. Brown's apparatus consists essentially of a large upright cylindrical vessel fitted on the top with a movable valve cover, of the whole diameter of the cylinder. The cover is raised and lowered from and to its seat by a lever and suitable gear at proper times. The gas supply pipe enters the cylinder at the bottom ; the cylinder being filled with air, and the valve raised, the gas cock is opened and the issuing gas lighted by a small flame as it enters the cylinder. The flame produced fills the whole vessel, expelling the air it contains ; the valve being now lowered and the gas supply shut off, the water-jet is thrown in and causes condensation. To keep up a constant supply of power several of these cylinders are required, so that one at least may be always vacuous while the others are in the process of obtaining the vacuum. In the specification three are shown and three engines. The engines are all connected to the same crank- shaft. Notwithstanding this provision the motion must have been 4 THE GAS, PETROL, AND OIL ENGINE irregular. The idea was evidently suggested by the condensing steam engine ; instead of using steam to obtain a vacuum flame is employed. Brown's engine, although uninteresting theoretically, is important as being the first gas engine undoubtedly commercially at work. Brown was able and persevering, and appears to have been a man of business as well as an inventor : he succeeded in forming commercial companies for working his engines as applied to three purposes, pump- ing for water works, road locomotion and boat propulsion. According to the ' Mechanics' Magazine,' published in London in August 1824, a model had been already made which raised 300 gallons IJ'^Sigiiiiiii cawir itRHPiMiT,.] Fig. I. — Brown's Gas-vacuum Engine, 1826 of water 15 feet high on one cubic foot of gas. In 1832 it appears four engines were in use for pumping : (i) One at Croydon on the canal, raising water from a lower to a higher level. * {2) One at Soham in Cambridgeshire, for draining part of the middle Fen district. (3) One at Eagle Lodge, Old Brompton. (4) One at Eagle Lodge, Old Brompton, but of the beam type. It is stated that the cylinder of the Croydon engine was 3 ft. 6 in. diameter by 22 in. high. Engine No. 3 above was inspected by the Editor of the ' Mechanics' Magazine ' at work : its cylinder was 3 ft. 8| in. HISTORICAL SKETCH 5 diameter by 22 in. high ; and it discharged 750 gallons per stroke, four strokes per minute, 12 feet high. Brown claimed, in a circular published in 1832, that the coke and tar obtained in making coal gas for the Croydon engine was sold for such sums as produced a profit in addition to giving motive power for nothing. He states that the whole annual expense of the Croydon gas vacuum engine, including coal, wages, repairs, depreciation, and rent, amounted to 666^. 14s., while the receipts from the sale of coke and tar were 769^. 12s., so that the annual profit was 102/. i8s., without counting the value of the work done, which previously cost the canal company 275/. per annum to effect by steam engine. This state of affairs could not have lasted, as after some years' work the engines were dispensed with. In 1825 Mr. J. A. Whitfield, of Bedhngton Ironworks, describes Brown's engine as applied to a carriage, with sections of the cylinders, and a drawing of the carriage to scale. The wheels were 5 feet in diameter, wheel base was 6 ft. 3 in., and track 4 ft. 6 in. ; the weight, with gas and water, was 20 cwts. The cylinders were 12 in. diameter by 24 in. stroke. It was later stated that this carriage successfully ascended the steepest part of Shooter's Hill, where the gradient was 13-^ in. in 12 feet; the ascent was made with considerable case. The date of this test was the last week of May 1826. In January 1827 a boat was propelled by Brown's engine on two days— the ist and the 31st. In the first test the boat leaked (as the result of a collision the day before) and the river was rough, so that the trial was unsatisfactory, although the boat made some headway. On January 31 everything went well, and, starting from Blackfriars Bridge, the boat travelled on the Thames at the rate of seven to eight miles per hour, it is said, ' with all the regularity of a steamer, and the paddles worked quite smoothly, and seemed capable of continuing to go as long as gas was supplied.' The boat was 36 feet long, and the weight of the engine and frame- work came to 600 lbs. Mr. Brown stated that this test was made in the presence of the Lords of the Admiralty and a number of scientific men. Notwithstanding this partial success, the company formed to apply the engine to vessels dissolved in February 1827. Samuel Brown deserves the greatest credit for his able and perse- vering attempts to introduce gas power for the purpose of locomotion on land and water, and he appears to be the first to make even an experimental application in a practicable form on a considerable scale. W. L. Wright, 1833, No. 6525. — In this specification the drawings are very complete and the details are carefully worked out. The explosion of a mixture of inflammable gas and air acts directly upon 6 THE GAS, PETROL, AND OIL ENGINE the piston, which acts through a connecting rod upon a crank-shaft. The engine is double-acting, the piston receiving two impulses for every revolution of the crank-shaft. In appearance it resembles a high-pressure steam engine of the kind known as the table pattern. The gas and air are supplied to the motor cylinder from separate pumps through two reservoirs at a pressure a few pounds above atmosphere ; the gases (gas and air) enter spherical spaces at the ends of the motor cylinder, partly displacing the previous contents, and are Fig. 2. — Wright's Gas-exploding Engine, 1833 ignited while the piston is crossing the dead centre. The explosion pushes the piston up or down through *s whole stroke ; at the end of the stroke the exhaust valve opens and the products of combustion are discharged during the return, excepting the portion remaining in the spaces not entered by the piston. The ignition is managed by an external flame and touch-hole. The author has been unable to find whether the engine was ever made, but the knowledge of the detail essential to a working gas engine shown by the drawings indicates that it or some similar machine had been worked by the inventor. Both HISTORICAL SKETCH 7 cylinder and piston are water- jacketed, as would have been necessary in a double-acting gas engine to preserve the working parts from damage from the intense heat of the explosion. This is the earliest drawing in which this detail is properly shown. William Barnett, 1838, No. 7615. — Barnett's inventions as described in his specification are so important that they require more complete description than has been here accorded to earlier inventors. Barnett is the inventor of a very good form of igniting arrange- ment. The flame method most widely used up to about 1892 was originated by him. Barnett is also the inventor of the compression system now so largely used in gas engines. The Frenchman, Lebon, it is true, described an engine using compression in the year 1799, but his cycle is not in any way similar to that proposed by Barnett, or used in the modern gas engine. Barnett describes three engines. The first is single-acting, the second and third are double-acting ; all compress the explosive mixture before igniting it. In the first and second engines the inflammable gas and air is compressed by pumps into receivers separate from the motor cylinder, but communicating with it by a short port which is controlled by a piston valve. The piston valve also serves to open communication between the cylinder and the air when the motor piston discharges the exhaust gases. In the third engine the explosive mixture is introduced into the motor cylinder by pumps, displacing as it enters the exhaust gases resulting from the previous explosion ; the motor piston by its ascent or descent compresses the mixture. Part of the compression is accom- plished by the charging pumps, but it is always completed in the motor cylinder itself. In all three engines the ignition takes place when the crank is crossing the dead centre, so that the piston gets the impulse during the whole forward stroke. Fig. 3 is a sectional elevation of the first engine, showing the principal working parts, but omitting all detail not required for explaining the action. There are three cylinders containing pistons : A is the motor piston, B is the air-pump piston. The gas-pump piston cannot be seen in the section, but works in the same crosshead as B. The motor piston is suitably connected to the crank-shaft, and the other two are also connected by levers in such manner that all three move simultaneously up or down. The pump pistons, moving up, take respectively air and inflammable gas into their cylinders ; upon the down stroke the gases are forced through an automatic lift valve into the receiver d, and there mix. When the down stroke is com- plete and the receiver is fully charged with the explosive mixture, 8 THE GAS, PETROL, AND OIL ENGINE the pressure has risen to about 25 lbs. per square inch above atmo- sphere. At the same time as the pumps are compressing, the motor piston is moving down and discharging the exhaust gases from the power cylinder ; it reaches the bottom of its stroke just when com- pression is complete. The piston valve E then opens communication between the receiver and the motor, at the same time closing to atmosphere. The motor cylinder being in free communication wilh the receiver, the explosion of the mixture is accomplished by the igniting cock or valve F ; the pressure resulting actuates the motor Fig. 3. — Barnett Gas Engine piston during its whole upward stroke, me hot gases flowing through the port G precisely as steam would do. The volume of the receiver being constant, the pressure in the motor cylinder slowly falls by ex- pansion, due to Ihe movement of the piston, upon which work is per- formed, and by cooling, the pressure still existing in the cylinder when the stroke is complete depending on the ratio between the volume swept by the motor piston and the volume of the receiver. The down stroke again expels the products of combustion, the HISTORICAL SKETCH valve opening to atmosphere, while the compression again takes place. This cycle gives a single-acting engine. It is obvious that as the piston A does not enter the receiver it cannot displace the exhaust gases there. If means are not taken to expel these gases they must mix with the fresh explosive charge pumped in. It is very desirable that these gases should be as completely as possible discharged. An exhausting pump is described for doing this, but in small engines it adds an additional complication ; and so Barnett states that in some cases it may be omitted. The exhaust gases do not so injuriously affect the action of small gas engines. Fig. 4. — Bametl's Igniting Cock The igniting valve is very ingenious. It is shown at Fig. 4, on a larger scale. A hollow conical plug A is accurately ground inio the shell B, and is kept in position by the gland c ; the shell has two long slits, d and e ; the plug has one port so cut that as the plug moves it shuts to the slit d before opening to e. In the bottom of the shell there is screwed a cover carrying a gas burner /, which may be lit while the port in the plug is open to the air through d. The external constant flame h lights it. So long as the plug remains in this position the internal flame continues to burn quietly. If the plug be now turned io shut to the outer air, it opens to the slit e, and as that lO THE GAS, PETROL, AND OIL ENGINE contains explosive mixture it at once ignites. The explosion extinguishes the internal flame, but it is again lighted at the proper time when the plug is moved round. The valve acts well and is almost identical in principle with he flame-igniting arrangements of Hugon, Otto Langen, and Otto. Barnett's second engine is identical with his first except that it is double-acting, and therefore requires a greater number of parts. Fig. 5. — Barnett Engine Barnett's third engine is worthy of careful description. Fig. 5 is a vertical section of the principal jflfrts. It is double-acting. It has three cylinders, motor, air-pump and gas-pump ; the air and gas pumps are single-acting, the motor piston is double-acting. The pumps are driven from a separate shaft, which is actuated from the main crank-shaft by toothed wheels ; the wheel upon the pump-shaft is half the diameter of that on the motor shaft, so that it makes two revolutions for one of the other. The pumps therefore make one up- and-down stroke for each up or down stroke of the motor piston ; HISTORICAL SKETCH ii the angles of the cranks are so set that they (pumps) discharge their contents into one or other side of the motor cylinder at every stroke ; the exhaust gases are partly displaced by the fresh explosive mixture, and the motor piston completes the compression in the motor cylinder itself. When full up or down the igniting cock acts and the explo- sion drives the piston to the middle of its stroke ; it here runs over a port in the middle of the cylinder, and the pressure at once falls to atmosphere. A is the motor piston ; b is the air-pump piston ; the gas-pump piston is behind the air-pump, and is therefore not seen in the section ; D is the main crank-shaft ; e the pump-shaft driven from the main shaft by the wheels f and G. The engine is exceedingly interesting as the first in which the compression is accomplished in the motor cylinder, but it is not so good a machine as the first because of the difficulty of obtaining a sufficient amount of expansion. From 1838 to 1854 inclusive eleven British patents were applied for ; some were not completed, but only reached the provisional stage. Of these patents by far the most important is Barnett's ; the others are interesting as showing the gradual increase of attention the subject attracted. The other names are Ador, 1838 ; Johnson, 1841 ; Robinson, 1843 ; Reynolds, 1844 ; Brown, 1846 ; Roger, 1853 ; also Bolton & Webb, making three patents for the year ; for 1854 two patents, Edington and Barsanti & Matteucci. None of the proposals in these patents are really valuable or novel, being anticipated by either Street, Wright, or Samuel Brown. Robinson's is the best, being similar to Lenoir's in some of its details, and showing distinctly a better understanding of gas-engine detail. A. V. Newton, 1855, No. 562. — This specification is interesting, and describes for the first time a form of igniting arrangement which came into use about 1885 ; it seems to be identical with the invention of the American Drake, although not described as a communication from him. It is a double-acting engine, and takes into the cylinder a charge of gas and air mixed, during a portion of the stroke, at atmo- spheric pressure. The igniting arrangement is a thimble-shaped piece of hard cast iron which projects into a recess formed in the side of the cylinder : it is hollow, and is kept at all times red-hot by a blowpipe flame projected into it by a small pump. When the piston uncovers the recess the explosive gases coming in contact with it ignite, and the pressure produced drives it forward. This is the first instance of ignition by contact with red-hot metal ; the proposal has often been made since then in varying forms. Barsanti &' Matteucci, 1857, No. 1655. — This is the first free piston engine ever proposed ; instead of allowing the explosion to act directly upon the motive power shaft through a connecting rod, at 12 THE GAS, PETROL, AND OIL ENGINE the moment of explosion the piston is perfectly free. The cylinder is very long, and is placed vertically. When the explosion occurs it expends its power in giving the piston velocity ; the expansion there- fore takes place with considerable rapidity, and the piston, gaining speed until the pressure upon it falls to atmosphere, moves on till the energy of motion is absorbed, doing work on the external air, lifting the piston and in friction. When the energy is all absorbed in this manner it stops ; it has reached the top of its stroke. A partial vacuum has been formed in the cylinder and the weight has been raised through the stroke. It now returns under the pressure of the atmosphere and its own weight ; in returning a rack attached to the piston engages the motive shaft and drives it. The cooling of the gases as the piston descends continues and helps to keep up the vacuum. The method although indirect is economical. Three advantages are gained by it — rapid expansion, considerable expansion (an ex- pansion of six times is common in these engines), and also some of the advantages of a condenser. Fig. 6 shows a vertical section of their best modification. The motor piston A working in Ihe tall vertical cylinder B-is attached to the rack c, which works into the toothed wheel d. The motor shaft E revolves in the direction of the arrow, and it is provided with a ratchet ; a pall upon the wheel D engages the ratchet on the down stroke of the piston only ; on the up stroke it slips freely past the ratchet. The piston A is therefore quite free to move without the shaft on the up stroke, but it engages on the down stroke. The cams f and G are arranged to strike projections upon the rack, and so raise or lower the piston. It is raised when the charge is to be taken in, and lowered when it has com- pleted its working stroke and the exhaust gases have to be discharged. When raised the valve h is in the position shown. Air first enters the cylinder through the port i, which also serves to discharge the exhaust. After the piston has uncovered the port k the valve h shuts on I, opening at the same time on K ; the gas supply then enters and mixes more or less perfectly with the air previously introduced. A small further movement of the piston now closes the valve and the explosion is caused by the passage of the electric spark in the position indicated upon the drawing. The piston shoots up freely to the top of its stroke, to give out the^ork stored up usefully upon its return. As the next engine to be described marks the beginning of the practicable stage of gas-engine development, it is advisable to sum- marise before proceeding. Previous to i860 the gas engine was entirely in the experimental stage. Many attempts were made, but none of the inventors suf- ficiently overcame the practical difficulties to make any of their engines HISTORICAL SKETCH 13 commercially successful. This was mostly due to the very serious nature of the difficulties themselves, but it was also due to the too great ambition of the inventors : they wished not only to compete with the steam engine for small powers, but for large powers. They thought, in fact, more to displace the steam engine than to compete with it. Fig. 6. — Barsanti & Matteucci Engine, 1857 This is clearly shown in many of their descriptions of the applica- tions of their inventions. The greatest credit is due to Wright and Barnett. Wright very closely proposed the modern non-compression system, Barnett the modern compression system. Barnett is also the originator of one of the modern flame systems for ignition. Barsanti & Matteucci follow in order of merit as the inventors of the free-piston gas engine. 14 THE GAS, PETROL, AND OIL ENGINE Lenoir occupies the honourable position of the inventor of the first gas engine ever actually introduced regularly to public use. The engine was not strikingly novel ; nothing was done in it which had not been proposed before, but its details were thoroughly and carefully worked out. It was, in fact, the first to emerge from the purely experimental stage. Lenoir's real credit consists in over- coming the practical difficulties sufficiently to make previous proposals fairly workable. The principle is exceedingly simple and evident. The piston moves forward for a portion of its stroke, by the energy stored in the fly-wheel, and takes into the cyUnder a charge of gas and air at the ordinary atmospheric pressure. The valves cut off communi- cation, and the explosion is occasioned by the electric spark : this propels the piston to the end of the stroke. Exhausting is done precisely as in the steam engine. The engine is simply an ordinary high-pressure steam engine with valves arranged to admit gas and air and discharge the products of combustion. Fig. 7 is an external elevation of a three-horse engine. It was first constructed in Paris in i860 by M. Hippolyte Marinoni. In Moigno's ' Cosmos ' of that year it is stated that two engines were . in course of manufacture — one of six horse-power, the other of twenty. The early statements of its economy were ludicrously inaccurate. A one-horse-power engine consumed, it was said, but three cubic metres (106 cubic feet nearly) of coal gas in twelve hours' work, and therefore cost for fuel not more than one-half of what a steam engine would have done. The actual consumption was speedUy shown to be much nearer three cubic metres per effective horse-power per hour. Notwithstanding the high consumption, the engine had many good points : its action was exceedingly smooth ; no shock whatever was heard from the explosion. Indeed it is quite impossible when watching the engine in motion to realise that regular explosions are occurring. The motion is as smooth and silent as in the best steam engine. In the ' Practical Mechanics' Journal ' of August 1865 there is an article describing the progress made by the engine since the date of its introduction, from which it appears that in France from 300 to 400 engines w ere then at work, the power ranging from half-horse to three-horse. The Reading Ironworks Company, Limited, at Reading, un- dertook the manufacture for this country. One hundred engines were made and delivered by them ; several of them lmve~cbntinued at work till now. Notably one engine inspected by the author at HISTORICAL SKETCH 15 Petworth House, Petworth, worked for twenty years pumping water, and is even yet in good condition. The work performed by the engines was multifarious in its character : printing, pumping water, driving lathes, cutting chaff, sawing stone, i6 THE GAS, PETROL, AND OIL ENGINE polishing marble — in fact, wherever from one-half to three horse-power was sufficient. Lenoir built an experimental road carriage propelled by one of his engines which was proved to have repeatedly circled round the works where it had been constructed in the Rue de la Roquette, in Paris, at the point of turning to Vincennes. It also made a trip from Paris to Joinville-le-Pont and returned within three hours. A two-horse-power Lenoir engine was also placed in a boat, which is said to have run between Paris and Charenton several times a week for two years. The power obtained was found to be too small, and great difficulty was experienced with cooling water on the road carriage. The fuel used was a light volatile hydrocarbon, vaporised by a surface evaporating device. The fuel was similar to petrol, though not known by that modern name. Lenoir's patent in this country was obtained by J. H. Johnson, i860. No. 335. It describes very closely the engine as manufactured both in France and England. The subsequent patent 1861, No. 107, does not seem to have been carried into effect. These specifications contain many erroneous ideas, showing the notions then prevalent among inventors of the nature of gaseous explosions. Lenoir erroneously supposed that the economy of his engine would be improved if he could obtain a slower explosion. He evidently thought that the power imparted to the piston by explosion was similar in nature to a sudden blow — a rapid rise of pressure, and a fall nearly as rapid. He therefore attempted to avoid explosion by such expedients as stratification and injection of steam or water spray. The stratification idea he very clearly expressed in his second specifi- cation, stating that ' the object of preventing the admixture of air and gas is to avoid explosion.' It is somewhat extraordinary to find notions so erroneous common at a time when Bunsen's work had clearly proved the continuous nature of the combustion in gaseous explosions, and when Hirn had made experiments which showed that the heat evolved by explosion in a gas engine was only a small part of the total heat of the combustion, the heat which did not appear during explosion being produced during expansion. Other speculations on the cause of the uneconomical working of the engine were frequent, but the #ue reason was fully explained by Gustav Schmidt in a paper read before ' The Society of German Engineers ' in 1861. He states : ' The results would be far more favourable if compression pumps, worked from the engine, compressed the cold air and cold gas to three atmospheres before entrance into the cylinder ; by this a great expansion and transformation of heat is possible.' This opinion became common at this time. Compression engines HISTORICAL SKETCH 17 were proposed with great clearness and a full understanding of the advantages to be gained. Million, 1861, No. 1840. — This Frenchman had exceedingly clear ideas of the advantages of compression ; he evidently con- siders himself as the first to propose its use in a gas engine, apparently unaware of the existence of Barnett's engine already described. He claims the exclusive right to use compression in the most emphatic language. The first engine described is exactly what Schmidt asks for. Sepa- rate pumps compress the air and gas into a reservoir, from which the movement of the motor piston, during a portion of the stroke, withdraws its charge under compression. Ignition is accomplished by the electric spark, and the piston moves forward under the high pressure produced. He states : ' In ordinary air engines the operation of the motive cylinders is analogous to that of the pumps, the result being that there are two cylinders, which act in directions contrary to each other, and that the pump, which is an organ of resistance, even works at a greater pressure than that of the motive cylinder, which is an organ of power. Thus these engines are very large in proportion to their power. On the contrary, by employing gases under the conditions above explained, these engines will exert great power in proportion to their dimensions. The sudden ignition of the gases in the motive cylinder causes the latter to work at an operative pressure much greater than that of the pumps.' The advantage of compression in a gas engine could not be more fully and clearly stated. But he goes even a step further ; he sees that the portion of the motor piston stroke spent in taking in the charge under compression is a disadvantage, and he proposes to make the whole stroke available for power by providing a space at the end of the cylinder in which the gases are compressed. ' Instead of introducing the cold gases into the cylinders, during a portion of the stroke and igniting them afterwards, when the induc- tion ceases . . . another arrangement might be adopted. The motive cylinder might be made longer than necessary in order that the piston should always leave between it and the end of the cylinder a greater or less space, according to the pleasure of the con- structor, such as one-fourth or one-third, more or less, of the volume generated by the motive piston. This space is called by the inventor a cartridge. On opening the slide valve the gases could be allowed to enter suddenly from the pressure reservoir into this cartridge towards the dead point, and this induction having ceased an electric spark would ignite the gases in the cartridge by which the driving piston would be set in motion.' VCL. I. C i8 THE GAS, PETROL, AND OIL ENGINE Such an engine would resemble in its action the best modern compression engines. The diSculties of ignition, however, are too considerable to be overcome without further detaU. The compression idea at this date was evidently widely spread, because it again crops up in a remarkably clever pamphlet by M. Alph. Beau de Rochas, published in Paris in 1862. He advances a step further than Million, and investigates the conditions of greatest eco- nomy in gas engines using compression, with reference to volume of hot gases and surfaces exposed. He states that to obtain economy with an explosion engine four conditions are requisite : 1. The greatest pDSsible cylinder volume with the least possible cooling surface; 2. The greatest possible rapidity of expansion ; 3. The greatest possible expansion ; and 4. The greatest possible pressure at the commencement of the expansion. In using boiler tubes, he states, the eliciency of the heat trans- mitted increases with reduction in the diameter of the tubes. In the case of engine cylinders, therefore, the loss of heat of explosion would be in inverse ratio to the diameter of the cylinders. Therefore, he reasons, an arrangement which, for a given con- sumption of gas gives cylinders of the greatest diameters, will give the best economy, or least loss of heat to the cylinder. One cylinder only must be employed in such an engine. But loss of heat depends also upon time ; cooling, therefore, will be proportionately greater as the working speed is slower. The sole arrangement capable of combining these conditions, he states, consists in using the largest possible cylinder, and reducing the resistance of the gases to a minimum. This leads, he states, to the following series of operations. 1. Suction during an entire outstroke of the piston. 2. Compression during the following instroke. 3. Ignition at the dead point and expansion during the third stroke. 4. Forcing out of the burned gases from the cylinder on the fourth and last return stroke. The ignition he proposes to acconl|ilish by the increase of tempera- ture due to compression. This he expects to do by compressing to one-fourth of the original volume. In our own country the late Sir C. W. Siemens proposed com- pression in 1862. The idea was exceedingly widely spread, as is evident from those numerous and independent inventions. The practical experience to enable it to be successfully effected had yet to be created, however, and this took many years of patient work. HISTORICAL SKETCH 19 The igniting arrangement was the first weak point requiring improvement. The electrical method of Lenoir was exceedingly delicate and troublesome. Hugon's engine, produced in 1865, was similar to Lenoir's ; but the igniting was accomplished by flame, a modification of Barnett's, 1838, using a slide valve instead of a lighting cock. The flame ignition was certain and easily kept. in order. In other points the engine was a great improvement upon its predecessor. The lubrication 20 THE GAS, PETROL, AND OIL ENGINE was improved by injecting water into the cylinder and the cooling water jacket was better arranged. As a result the consumption 6f gas was reduced. Fig. 8 is an external elevation of the Hugon engine. Mt. Otto n ow appears upon the scene. Before him much had been done m inventing and studying engines, but it remained for him by sheer perseverance and determination of character, to overcome all difficulties and reduce to successful practice the theories of his predecessors. In 1867 Messrs. Otto & Langen exhibited at the Paris exhibition of that year, their free piston engine, exterior elevation shown at fig. 19. It was absolutely identical in principle with the previous invention of Barsanti and Matteucci, but the details were completely and successfully carried out. The Germans succeeded commercially and scientifically where the Italians completely failed. Flame ignition was used and great economy was obtained, a half-horse engine, according to Professor Tresca, giving over half- horse power effective, on a gas consumption at the rate of 44 cubic feet per effective horse-power per hour. This is less than half the con- sumption of Lenoir or Hugon ; accordingly the prejudice excited by the strange appearance and noisy action of the engine did not prevent its sale in large numbers. It completely crushed Lenoir and Hugon, and held almost sole command of the market for ten years, several thousands being constructed in that period. The Brayton gas engine appeared in America inG873^but although more mechanical than any free piston engine, its economy was insuffi- cient to enable it to compete. It was better than Lenoir or Hugon, but not nearly so good as Otto & Langen. In this engine there are two cylinders, compressing pump and motor. The charge of gas and air is drawn into the pump on the out- stroke and compressed on the return into a receiver ; the pressure usual in the receiver varies from 60 to 80 lbs. per square inch above atmosphere. The motor cylinder takes its supply from the receiver, but the mixture is ignited as it enters, a grating arrangement preventing the flame from passing back ; the mixture, in fact, does not enter the motor cylinder at all ; what enters it, is a continuous flame. At a certain point the supply of flame is §ut off and the piston, moving on to the end of its stroke, expands the volume of hot gases to nearly atmospheric pressure before discharge. Fig. 9 is an external view of the engine. Figs. 10 and 11 are sections of the motor and pump cylinders. The action is as follows : The engine is single-acting, receiving one impulse for every revolution ; like all gas engines it depends upon the energy stored up in the fly- wheel to carry it through those parts of its cycle where the work is HISTORICAL SKETCH 21 negative. The two cylinders are inverted, and are attached to a beam rocking beneath them by connecting rods. The beam is prolonged and connected to the crank above it by a rod ; both cylinders are single-acting and the pistons are of the trunk kind. Both pump and motor cylinders are of the, same diameter, but the pump is only half the stroke of the motor. The valves are actuated from a shaft running at the same rate as the main shaft and driven from it by bevel wheels. There are four valves, all of the conical seated kind — two upon the Fig. 9. — Brayton Petroleum Engine motor, admission and discharge; two upon the pump cylinder, admission and discharge. The admission and discharge valves upon the motor are actuated from the auxiliary shaft by levers and cams, so is the pump inlet. The pump discharge valve is automatic, rising at the proper time by the pressure of compression. During the do wn-stroke the pmnp takes in the charge of gas and air, forcing tfonthe up-stroke into th e receiver. From the receiver it is led to thT^ower cynnder, passing by the mlet valve through a pair of " 22 THE GAS, PETROL, AND OIL ENGINE p erforated brass plates with wire gauze placed between them . Through this diaphragm a small stream of mixture is constantly passing into the motor cylinder ; before the engine is started, a plug is withdrawn and the current lighted ; a constant flame is therefore burning under the diaphragm. The mixture enters the cylinder through this flame, lighting as it enters ; at all times during the exhaust part of the stroke, as well as the admission, the stream of entering mixture from the Fig. 10. — Braylon Engine Section of Motor Cylinder Fig. II. — Brayton Engine Section of Pump Cylinder receiver keeps up a small constant flame which is augmented at the beginning of the stroke, so as to fill the cylinder entirely, when the admission valve is opened. When the admission valve is closed, the by-pass keeps the flame fed with sufficient mixture to keep it alight. The pressure in the cylinder thus never exceeds that in the reservoir, and the mixture burns quietly without spreading back. Figs. 9, 10, II, and 13. — a is the motor cylinder ; b the pump ; the beam and connections require no lettering ; c is the purhp inlet HISTORICAL SKETCH 23 valve (the pump discharge, which is an ordinary lift valve, is not seen in fig. II, but is lettered D in fig. 9) ; E the motor inlet ; F the igniting plug, which is withdrawn when the flame is to be lit before starting the engine (see fig. 13) ; g is the grating in section (see fig 10) ; h the exhaust valve ; the levers and cams are sufficiently indicated on the drawing ; the small pipe and stop-cock (fig. 9) communicates at all times with the reservoir and supplies the constant flame with mix- ture. The engine worked well and smoothly ; the action of the flame in the cylinder could not be distinguished from that of steam, it was as much within control and produced diagrams quite similar to steam. rIG. 12. Brayton Petroleum Pump Fig. II Plan of Grating -Brayton Grating and Valve The flame grating was the weak point ; it stood exceedingly well for a time, but if by any accident the gauze was pierced in cleaning, the flame went back into the reservoir and exploded all the mixture — the engine, of course, pulled up as the constant flame having no supply was extinguished. This accident became so troublesome that Mr. Brayton discontinued the use of g a g and r.nnverted his engine into a petro leufn engm e! Ihe light petroleum was pumped upon the gratmg into a groove filled with felt, the compressing pump then 24 THE GAS, PETROL, AND OIL ENGINE charged the reservoir with air alone. The air in passing through the grating carried with it the petroleum, part in vapour, part in spray ; the constant flame was fed by a small stream of air. The arrangements were, in fact, precisely similar to the gas engine, except in the addition of the small pump and the slight alteration in the valve arrangements. The difficulty of explosion into the reservoir was thus overcome, but a new difficulty arose — the cylinder accumulates soot with great rapidity and the piston requires far too frequent removal for cleaning. The petroleum pump is an exceedingly clever little contrivance ; fig. 12 shows its details. The amount of petroleum to be injected at each stroke is so small that an ordinary force-pump with clack valves would be uncertain. Bray ton gets over this diffi- culty by substituting a slide valve driven from the eccentric. The plunger of the pump is no larger than a black-lead pencil, yet it discharges any quantity, from a single drop per stroke up to a full throw, with unerring certainty. The plunger also is driven from an eccentric. Both eccentrics are in one piece and rotate on the end of the auxiliary shaft, driven by a pawl when the engine is in motion ; to allow of starting, the pump can be moved by a hand-crank indepen- dently. To start, the air reservoir is filled, if not already full, by turning the engine round by hand ; the plug F is then withdrawn and a little petroleum thrown upon the diaphragm by a few turns of the pump. The cock on the small pipe is then opened and a stream of air flowing from the reservoir vaporises the petroleum ; it is lit at G, and the flame having enough air for combustion retreats to the grating and remains burning within the cylinder. The plug is then inserted, the starting cock opened, and the engine starts. The flame remains alight during the whole time the petroleum continues to be supplied. The valves act well, and the motor cylinder does not suffer from the action of the flame so long as it is kept reasonably clean. If the soot, however, is allowed to accumulate, it speedily cuts up. The late Prof. Thurston, of the Stevens Institute of Technology, tested a Brayton gas engine in New York in the year 1873. The following extracts are from his report : ' The operation of the engine is precisely similar, in the action of the engine proper and in the distribution of pressure in its cylinder, to that of the steam engine. The aftion of the impelling fluid is not explosive as it is in every other form of gas engine of which I have knowledge. ' Upon the opening of the induction valve, the mixed gases enter, steadily burning as they flow into the cylinder, and the pressure from the commencement of the stroke to the point of cut-off, as is shown by the indicator diagrams, is as uniform as that observed in any steam engine cylinder. The maximum pressure exerted during my experi- HISTORICAL SKETCH 25 mental trial, and while the engine was driving somewhat more than its full rated power, was about 75 lbs. per square inch at the beginning of the stroke, gradually diminishing to 66 lbs. per square inch at the point of cut-off, where the speed of the piston was nearly at a maximum, and then declining in accordance with the law governing the expansion of gases. ' Complete combustion is insured by thorough mixture. This is accomplished by taking the illuminating gas and air, in proper proportion, into the compressing pump together, and the mixture here made becomes more intimate in the reservoir, and in its progress towards the point at which it does its work. The constantly burning jet already described insures prompt ignition on entering the cylinder. ' . . . the engine rated at 5 HP developed, as a maximum, rather more than its rated power. Its mean power during the test, as deter- mined by the dynamometer, was 3" 986 HP, the indicator showing at Max. press. 68 lbs. per sq. ill. Fig. 14 Fig. 15. — Diagrams from Brayton's Gas Engine that time 8'62 HP developed in the cylinder. The amount of gas consumed averaged 32'o6 cubic feet per IHP per hour. ' The excess of indicated over dynamometric HP is to be attributed to the work of driving the compressing pump and to the friction of the machine. ' The greater portion of this appears both on debit and credit side of the account, since, although expended in the compressing pump, it is restored again in the driving cylinder.' The consumption of 32' 06 cubic feet per IHP per hour is incorrect ; it is obviously unfair to include the pump diagram in the gross power. The author has tested an engine of similar construction and dimen- sions ; he finds the friction of the mechanism to be about i-horse ; adding this number to the dynamometric power of • Prof. Thurston, the legitimate indicated power may be taken as 5 HP, the consumption therefore ^^"^ 32;o6 ^^^.^. and the gas per brake HP per hour 5'0 IS 26 THE GAS, PETROL, AND OIL ENGINE 8-62 X 32'o6 IS 3-986 = 69'3. These numbers, although showing improve- ment upon the Lenoir and Hugon, prove that the engine was much inferior in economy to the Otto & Langen engines. Mr. H. McMutrie, Consulting Engineer at Boston, took diagrams from an engine of similar dimensions which confirm these results. Fig. 16.— Simon Engine Fig. 14 is the diagram taken with full load, fig. 15 the diagram from the motor with no load on, the power being just sufficient to over- come friction and pump losses. % Full Load Diagram Area of piston Speed of piston . . Mean pressure Pressure in reservoir Initial pressure in cylinder Gross power developed . . 50'26 sq. ins. 1 80 ft. per min. 33 lbs. per sq. in. 7 5 '4 lbs. per sq. in. 68 lbs. per sq. in. 9 HP. HISTORICAL SKETCH 27 No Load Diagram Speed o{ piston . ■ S-rW resistance. . ■ Net available power • • 180 ft. per min. i8 1bs. persq. m. . . 4-87 HP. inometer. , „ -Ream Oveibead Engine v\a 17.— Bray ton Beam u , . the Bray ton engine 28 THE GAS, PETROL, AND OIL ENGINE HISTORICAL SKETCH 29 increased economy, by causing the waste heat passing into the water jacket, and the heat of the exhaust gases, to be utilised in raising steam. Fig. 16 is an external view of the engine as exhibited at the Paris Exhibition of 1878. A is the motor, B the pump, and c the added boiler ; the steam was raised in it and the water jacket. The engine, although instructive, did not successfully overcome the difficulties which caused the abandonment of the Brayton as a gas engine. Brayton constructed this engine in America also in the beam overhead style and horizontal and inverted vertical. He applied the inverted vertical engine to a tram-car, but did not succeed in running it commercially. The horizontal engine he applied to a boat. Two of the boats were in use upon the Hudson for some years. Fig. 17 is an external view of a beam overhead engine, and fig. 18 is a similar view of a horizontal engine as applied to one of the boats. Brayton was enthusiastic and indefatigable, and spent most of his life in his many experiments ; ultimately he abandoned his American attempts and crossed over to England, and died in Leeds whUe engaged with experiments on a new oil engine at a large works there. His perseverance deserved a better reward. No one, however, has yet succeeded in carrying his type of engine further than he did. In 1876 Dr. Otto superseded his former invention by the production of the ' Otto Silent ' engine, now known all over the globe. It is a compression engine, using the precise cycle described in 1862 by Beau de Rochas, but carried out in a most perfect manner and using a good form of flame ignition — a modified Otto & Langen valve in fact. The economy is greater than that of any previous engine, one indicated horse being obtained upon 20 cubic feet of gas, or one effec- tive horse upon 24 to 30 cubic feet per hour. This engine has established gas engines upon a firm commercial basis. Strangely enough, although Dr. Otto was the greatest and most successful gas engine inventor who has yet appeared, he adhered to Lenoir's erroneous ideas, and in his specification 2081 of 1876 he attributed the economy of his machine to a slow explosion caused by arrangement of gases within the cylinder. The compression, which is the real cause of the economy and efficiency of the machine, he seemed to consider as an accidental and unessential feature of his invention. Dr. Otto deserved his success ; he had fought hard and long for it. He began his work in 1854, attained his first success in 1866, and his epoch-making advance in 1876. He was born at Holzhausen, in Nassau, in 1832, and died in 1891. 30 THE GAS, PETROL, AND OIL ENGINE He devoted his whole life to the study and development of the gas engine, beginning at the age of twenty-two years and continuing steadily on until his death— thirty-seven years of persistent work. During that long working period he experimented with nearly every form and cycle of engine. His early work dealt mainly with the atmospheric type in varied forms, but after 1876 he devoted himself to the four-cycle compression engine, and he patented many modiiications, including a compound gas engine, and another engine in which the whole of the exhaust products were expelled by an auxiliary piston. His death at the comparatively early age of fifty-nine was a great loss to the gas engine industry, but he had accomplished the heavy pioneering work for the world, and other hands took up the development of the four-cycle engine in many directions. Daimler originated small high- speed engines consuming the light hydrocarbons which are now known as petrol; his first small engine was produced in 1883. It used a surface carburetter and dispensed with the slide ignition, substituting an open tube. These Daimler engines ultimately de- veloped into the modern petrol engine which performs such an important part in the life of all . nations at the present time. These little engines, although in- ternal combustion engines of the old type, have developed so many interesting problems of carburet- t#, ignition, piston speed, and so forth, that a new science has sprung up in the course of their develop- ment which has produced motors of marvellous lightness for a given power, and made not only mechanical road locomotion a startlingly rapid success, but has brought mechanical flight just within the bounds of practicability. Daimler was born at Schoendorf, in Wiirtemberg, in 1834, and he died at Canstatt in 1900, at sixty-six years of age. Fig. 19.— Otto & Langen Free Piston Engine HISTORICAL SKETCH 31 He served his time with leading German engineering firms, then came to England and entered the works of Sir J. Whitworth. He joined Dr. Otto at Deutz on the Rhine in 1870, and was with him when the ' Gas Motoren Fabrili ' of Cologne was formed, at which great factory he was managing director from 1872 to 1882, when he retired from the Otto works for the purpose of devoting himself to the small high-speed light oil engine. In 1886 he tried his first motor bicycle, and on March 4, 1887, he ran for the first time a motor propelled car. About 1887, however, he had succeeded in propelling launches and other vessels on contipental canals, and his engines were used for that purpose to a considerable extent. In 1889, Messrs. Panhard & Levassor made arrangements with him to manufacture motor carriages in France. The Daimler Canstatt works became famous, and Daimler's work had undoubtedly a great share in the progress towards the modern use of petrol engines both on land and water. Another line of development was found in the adaptation of the Otto cycle to the use of the heavier oils — that is, oils such as are used in lamps with a high flashing point. Priestman of Hull, in England, made the first success with an engine which had a spray vaporiser and electric ignition. Priestman's work began in about 1885, and resulted in the introduction of many heavy oil engines applied to all purposes on land and water. Messrs. Priestman first exhibited a 4 HP petroleum engine at the Nottingham meeting of the Royal Agricultural Society in use with ordinary lamp oil. They exhibited a 6 HP portable oil engine at the Windsor Meeting in 1888, and obtained a prize at the Plymouth meeting in 1890. Later, Messrs. Hornsby introduced an engine known as the Hornsby- Ackroyd . This engine was the invention of Mr. Stuart Ackroyd, and it applied successfully for the first time the idea of igniting and vapor- ising the oil by means of the hot walls of the combustion chamber. This Hornsby-Ackroyd engine has proved to be the most successful of its type, and it is used in very large numbers both for land and marine work. Most leading makers of gas engines in England, America, and on the Continent now manufacture both heavy and light oil engines as part of their ordinary business. Messrs. Hornsby exhibited the engine at the Royal Agricultural Society Cambridge meeting in 1894, and obtained a prize. A most important development of the heavy oil engine is found in the Diesel e ngine, in which air alone is compressed in the engine cylinder to such a pressure as to heat it above the ignition point of heavy oil. When the air is at this high pressure and temperature the heavy oil is injected into it by air compressed to a still higher 32 THE GAS, PETROL, AND OIL ENGINE pressure. The oil spray ignites as it enters and so power is produced. Diesel began his work about 1892, and by determined perseverance he has produced a most interesting engine which is now used in con- siderable numbers and of high powers. Diesel's first engine was pro- duced about the year 1895. In 1878 Mr. J. Emerson Dowson designed a complete pressure gas plant in which he intended to make a fuel gas strong enough and clean enough for factory and domestic work, and in 1881 he applied his apparatus to the supply of a gas engine. This he did at the York meeting of the British Association in 1881. The plant consisted of a closed producer containing anthracite, a boiler to raise steam, a cooler and scrubber, and a gas holder. The fuel in the producer was first raised to incandescence by means of a blower. The steam from the boiler passed through a jet blower, by which a mixture of air and steam was passed through a grate upwards through the ignited anthracite. Carbonic oxide and hydrogen were formed and passed with the nitrogen of the air through the cooler and scrubber to the gas-holder, from which the engine was fed. The engine was of 3 HP, the first engine ever operated with producer gas. At that time (1881) the largest gas engine which had been built was about 20 brake horse-power, it was called by the engineering journals of the time ' a king of gas engines.' Mr. Dowson's invention had a profound effect on the development of the gas engine. It speedily led to the construction of larger and larger engines because his method supplied fuel gas at a much lower cost per heat unit than the towns' coal gas of the time. Dowson's pressure producers were gradually increased in dimensions as years passed on till now any desired power may be obtained. From the pressure producer sprang the suction producer, first placed on the market in practical form in 1894 by M. Benier of Paris, but then presenting many difficulties which were not removed till about nine years later, when Mr. Dowson and others placed effective suction plants in use in considerable numbers. Among the most successful of these more recent suction plants are those of the National Gas Engine Company, Ltd., and Messrs. Crossley Bros., Ltd. Nearly all makers of gas engines now manufacture suction plants. Broadly, suction plant differs from pressure plant in dispensing with the steam boiler and the gas holder, and causing the engiae itself to draw through the producer the air and steam required for its operations. Suction plants have further cheapened the gas supply for gas engines to such an extent that the fuel cost is now reduced in many cases to one-tenth of a penny per brake horse-power per hour. Suction gas plants are now made in units of several hundreds of horse-power. Pressure gas plants are now made of very large dimensions in accordance with Dr. Mond's inventions, and in smaller sizes by Messrs. HISTORICAL SKETCH 33 Crossley^ which use bituminous fuel; but so far no bituminous fuel suction plant has yet appeared, although many workers are busy on the problem. Suction producers are so far practically confined to anthracite and coke. The gas producer is advancing rapidly in public favour, and undoubtedly it has greatly enlarged the field of the gas engine. Town's coal gas, however, remains the principal and the best fuel for the gas engine, when it can be obtained at a reasonable price ; and the gas engineers of Britain are now fully alive to the importance of gas power, as about one-fifth of the whole town's gas-supply is used in engines. They have made great strides in recent years in the supply of cheap and good coal gas, so that several towns now supply coal gas at is. per thousand cubic feet for engines, and many do so at IS. 6d. In 1895 the late Mr. B. H. Thwaite demonstrated that the so-called waste gases from blast furnaces could be used in gas engines, and the demonstration has undoubtedly led to the design and construction of the very large gas engines now becoming common on the Continent, in America, and in this country. It appears from his experiments that the surplus gas from the blast furnaces of Britain is capable of sup- plying at least three-quarters of a million horse-power continuously day and night, and it is calculated that in America nearly three million horse power is available from this source. Thwaite's system was put into operation in 1895 at the Glasgow Iron Works, and it was also successfully applied near Barrow-in-Furness. For many reasons the system did not take immediate root in England, but in 1898 the Societe Cockerill, of Seraing, near Liege, applied an engine designed by M. Delamere-Deboutteville to utilise blast-furnace gas of no British thermal units per cubic foot heat value. The engine had a single cylinder of 31 -5 ins. diameter and 3 feet 3 J ins. stroke; it indicated 213 HP at 105 revolutions per minute. Before dealing shortly with the development of these large gas engines it is desirable to go back again to 1876 to give a brief account of the principal mechanical changes made in the four-cycle gas engine from that time to 1908, to consider the leading points in the progress of the two-cycle engines, and to discuss the effect of the true theory of the gas engine on its practical development. Comparing the 1877 Crossley Otto engine with one built by the same firm in 1892, it wiU be found that the two to one bevel gear driving the valve shaft has been replaced by the skew gear, that the slide valve ignition has been replaced by an incandescent tube igniter, and that all engine valves are of the conical seated lift type. In the later engine of 1908 the main mechanical difference is in the use of the low-tension electrical igniter instead of the incandescent tube. These changes have been accompanied by many obvious improvements VOL. I. D 34 THE GAS, PETROL, AND OIL ENGINE in design, by better shaped combustion spaces and much higher compression pressures. The changes do not appear great, but they led to a notable improvement in thermal efficiency and in power obtained for given cylinder dimensions. The thermal efficiency is more than doubled; in 1876 it was possible to obtain 16 per cent., and in 1908 35 per cent, has been certainly attained. These changes were accompanied by a gradual increase of power. In 1878, 3 HP was considered quite a large gas engine, it was the largest ever built of the Otto and Langen type ; in 1881 a 20 horse engine was a ' king of gas engines ' ; and in 1898 the largest engine built indicated 220 HP. Writing in that year the author stated : ' It is evident that the large gas engines of ten years hence will differ as much from the large gas engine of to-day as does the latter from the engine of 1886 or 1887. There can be little doubt that in ten years, gas engines of 1000 horse-power wiU be as common as engines of 100 horse-power are now.' _^^hiLeJJig_Ot to cycle eng ine was developing, inventors werejiard aj— wnTJfjvn^tVip *".™^"-iy^^Jl^£JI!^ In Britain this work fell mostly" upon Clerk, Robson, an3 Atkinson, while on the Continent the most persevering and determined worker was Koerting. Clerk began work on the gas engine at the end of 1876. His first patent was taken out at the beginning of 1 877 and dealt with an engine of the air-pressure vacuum type in which the explosion compressed air into a reservoir and caused a partial vacuum in the explosion chamber and a vessel connected with it. Clerk's next patent was taken out in 1878 (No. 3045 of 1878), and the engine then described was exhibited at the Royal Agricultural Show at Kilburn, London, in 1879. In this engine a pump compressed a mixture of gas and air into a reservoir at the fuU pressure required for compression ; the mixture under compression was admitted to the motor cylinder during the first part of its stroke, cut off, and then ignited by a platinum igniter, the piston driven forward by the explosion and expansion and exhausting performed on the return stroke. The engine gave 3 brake horse-power, and it was the first compres- sion explosion engine which was ever run giving one impulse for every revolution of the engine. This engine did not reach the market. The engine best known as the Orerk cyc le engine was patented in 1881 (No. 1089 of 1881), and exhibited "at the Paris Electrical Exhibition of that year. In it the pump was used as a displacer only, and the mixture was transferred to the motor cylinder at some 4 lbs. per square inch above atmosphere, and the entering charge displaced the exhaust gases by way of ports overrun by the piston. This was the first engine giving an impulse for every revolution where the exhaust discharge was timed and controlled by the motor piston HISTORICAL SKETCH 35 only. This type of engine came largely into use as built by Sterne & Co., the Campbell Gas Engine Company, and many other makers, a modification of it was also largely made and sold by the Stockport Company. It feU out of use about 1890, when the Otto patent of 1876 lapsed and nearly all engineers adopted the Otto cycle. It is much in use, however, now as the Koerting engine for large gas engines. Other engines of Clerk are described later in this work. Robson began work on gas engines in 1877. His first patent is dated No. 2334 of 1877, and he produced an engine under patents 1879 and 1880 in which the front of the cylinder is enclosed and used as a pump to force gas and air into a reservoir at about 6 lbs. per square inch above atmosphere ; the piston overran ports in the cylinder, but the exhaust was not timed by it ; a separate exhaust valve was used which opened and closed at the proper time. This engine was buUt by Messrs. Tangye, and exhibited by them at the end of 1880. Atkinson began work on the gas engine in 1878 ; his first patent is dated 1879, No. 3213. This engine was of the type exhibited by Clerk at KUburn. Mr. Atkinson was indefatigable in the production of two-cycle engines, and he built a most ingenious engine called by him the ' Differential engine ' which he exhibited at the Inventions Exhibition, London, in 1885. Later he produced another engine, which he called the ' Cycle engine ' ; this engine was proved to be the most economical of all those tested at the Society of Arts trials of motors for electric lighting in 1888-89. -A- fuller account of these two-cycle engines is given in a special chapter, so they need not be further dealt with here. Broadly, however, the position at the present date is this : that Otto cycle — that is, four-cycle — engines with few exceptions monopolise the field of the smaller gas engines; while for large engines, the two-cycle or Clerk cycle engines compete on much more equal terms. The general principles developed in this work explaining the causes of the economy of the modern gas engine were first enunciated by the author in a paper read before the Institution of Civil Engineers in April 1882.' He then classified gas engines in three great groups : Type I. — Explosion, acting on piston connected to crank. (No compression.) Type 2. — Compression, with increase of volume after ignition, but at constant pressure. Type 3. — Compression, with increase in pressure after ignition, but at constant volume. ' ' The Theory of the Gas Engine,' by Dugald Clerk : Minutes, Institution Civil Engineers, London. Paper No. 1855. April 1882. 36 THE GAS, PETROL, AND OIL ENGINE It was proved that under comparable conditions the relative theoretic eiiftciencies of the three types were Type I = 0-21 Type 2 = 0-36 Type 3 = 0-45 It was also shown that in the actual engines the real efficiency could not be so high as the theoretic, mainly because of the large proportion of heat lost through the sides of the cylinder, by the exposure of the flame which filled the cylinder to the comparatively cold enclosing walls. A balance-sheet was given showing the disposal of 100 heat units by a compression engine. Of the 100 heat units, i7"83 were con- verted into indicated work, 2g'28 were discharged with the exhaust gases, and 52-89 units passed through the sides of the cylinder into the water jacket. The economy of the Otto engine over its predecessors, the Lenoir and Hugon engines^ was clearly proved to be due to the fact of its using compression previous to explosion. These conclusions were very generally accepted by scientific and practical men who had studied the subject, and in February 1884 the late Prof. Fleeming Jenkin, then Professor of Engineering at the University of Edinburgh, delivered a lecture at the Institu- tion of Civil Engineers in London, on ' Gas and Caloric Engines.' ^ He had recalculated the efficiencies due to compression, with the result of corroborating the present writer's conclusions. He states : ' If I were to compress gas to 40 lbs., a pressure which is used not unfrequently, the theoretical efficiency would be 45 per cent. We actually get something like 24 or 23 per cent. ; we know that one- half of the heat is taken away by external cooling. Thus we find a very close coincidence between the calculated efficiency of those engines and that which we actually obtain, only we throw away about one-half of the heat in keeping the cylinder cool enough to permit lubrication. If we compress to 80 lbs. we have a theoretical efficiency of 53 per cent. If we do not compress at all, as Mr. Clerk has told you, we have a theoretical efficiency of only 21 per cent., so that we have it in our power to increase the theoretical efficiency very greatly by increasing the pressure of the gas%nd air before ignition. I have no doubt that the great gain of efficiency in the Clerk and Otto engines is really due to the fact of the compression ; this being done in a workmanlike way and carried to a very considerable point.' The advantages of compression could not be stated with more clearness and truth. ' ' Heat in its Mechanical Applications ' -. Institution Civil Engineers' Lectures, Session 1883-84. HISTORICAL SKETCH 37 In the same year there was pubHshed in Paris an able work entitled ' Etudes sur les Moteurs a Gaz Tonnant/ by Professor Dr. Aime Witz, of Lille, in which the theoretic efficiencies of the different types of cycle are calculated for a maximum temperature of explosion of 1600° C, and temperature before explosion of 15" C. He adopts the same classification as the present writer did in 1882, and finds the efficiencies : Type I = 0-28 Type 2 = 0-38 Type 3 = 0-44 which are almost identical with the author's figures. He also arrives at the conclusion that compression is the great- source of economy in the modern gas engine. At p. 53 he says : ' I find myself again in agreement with Mr. Dugald Clerk when he affirms that the success of Otto is due to compression alone, and not to the extreme dilution of the explosive mixture in the pro- ducts of the combustion of a precedent explosion.' He then proceeds to quote from the present writer's paper, and adheres to the statement that — ' Without compression previous to ignition an engine cannot be produced giving power economically and with small bulk.' Compression previous to ignition gives two great advantages : (i) A thermodynamic advantage (unproved theory of the cycle) ; (2) Higher available pressures and smaller cooling surfaces — the joint result being an economy in practice nearly fourfold that of the old non-compression engines. Mr. Otto's Theory. — Previous to 1882 the nature of the improve- ment obtained by compression was imperfectly understood, and this notwithstanding the very clear, though qualitative, statements of Schmidt, Million, and Beau de Rochas. An erroneous theory of the cause of the economy of the Otto engine was widely circulated and gained considerable support. It was enunciated in Mr. Otto's specification of 1876, No. 2081, and it weis supported by men so distinguished as Dr. Slaby of Berlin, Professor Dewar of the Royal Institution, the late Sir Frederick Bramwell, and the late Mr. John Imray. According to Mr. Otto, all gas engines, previous to his patent of 1876, obtained their power from the explosion of a homogeneous charge of gas and air. By the explosion excessive heat was evolved, and the pressures produced rapidly fell away : the excessive heat was rapidly absorbed by the enclosing cold walls. This caused great loss and gave very wasteful engines. Two methods were open to obtain better economy : 38 THE GAS, PETROL, AND OIL ENGINE ist, by using a very rapid expansion, so that the heat had but little time to be dissipated ; 2nd, by using slow combustion ; that is, by causing the inflammable mixture to evolve its heat slowly, so that the production of excessive temperatures and pressures was avoided. By the first method aU the heat was supposed to be evolved at once, and a high temperature was produced : by the second method the heat was evolved gradually so as to give a low temperature and pressure which was sustained throughout the stroke, and which was advantageously utilised by the piston while moving at a moderate speed. Mr. Otto states that this gradual evolution of heat may be produced by stratifying the charge of gas and air. Instead of using the homogeneous charge of Lenoir and Hugon, Mr. Otto uses a charge which he states is not homogeneous but heterogeneous. He affirms that his invention lies in the method or process of forming this stratified charge in a gas-engine cylinder, and that in addition to the explosive mixture, there must be present in the cylinder a mass of inert gas which does not burn but which serves to absorb the heat of the explosion and prevent the loss which would otherwise occur by the cooling effect of the cylinder walls. The ' inert ' gas may be either air alone which is capable of support- ing combustion, or the products of combustion which are incapable of supporting combustion, or a mixture of both. It is not sufficient that a mere film of this inert gas be present ; there must be what is termed a ' notable ' quantity. Mr. Otto proposes to form this heterogeneous or stratified charge by first drawing into the cylinder a charge of air alone ; and, second, a charge of explosive mixture, or by leaving in the cylinder a sufficient quantity of the products of a previous combustion to form a ' notable ' quantity of inert diluent. The compression space in the Otto engine is supposed to con- tain a sufficient volume of burned gases to form the inert diluent, so that the whole stroke of the piston is available for taking in the explosive charge. Suppose the piston to begin its charging stroke : the coal-gas and air mixture flows into the cylii^er through the inlet port and mixes to some extent with the inert gas aheady in the space ; but the mixing is incomplete, and at the piston itself the charge is sup- posed to consist entirely of exhaust gases. So that, while the charge at the igniting port is readily explosive, that at the piston is not explosive at aU, and between the igniting port and the piston the com- position of the charge varies from point to point. This ■ arrangement of the gases ' is supposed to be retained during compression, and exist at the moment of explosion. The compression HISTORICAL SKETCH 39 space contains a ' packed charge,' which consists of an explosive mixture at the one end, and between the explosive mixture and the piston a cushion of inert fluid, which is uninflammable and serves the double purpose of relieving the piston from the shock of explosion and absorbing heat which would otherwise be lost by conduction. By this device heat is gradually evolved. The flame originated in the port burns at first with great energy and spreads from one combustible particle to another, more and more slowly as it approaches the piston, where the particles are dispersed more and more in the inert gas. The mixture is so arranged that this burning lasts through- out the whole stroke, and is complete very shortly before the exhaust valve opens. The entire cylinder is never completely fiUed with flame, but the charge at one end has burned out before the flame arrives at the other end. Dr. Slaby comes forward in support of this hypothesis in an interesting report published as an Appendix to Prof. Fleeming Jenkin's lecture already referred to. Dr. Slaby states : ' The essence of Otto's invention consists in a definite arrangement of the explosive gaseous mixture, in con- junction with inert gas, so as to suppress explosion (and nevertheless insure ignition). ' At the touch hole, where the igniting flame is applied, lies a strong combustible mixture which ignites with certainty. The flame of this strong charge enters the cylinder like a shot, and during the advance of the piston it effects the combustion of the further layers of dispersed gaseous mixture, whilst the shock is deadened by the cushion of inert gases interposed between the combustible charge and the piston. ' The complete action takes place in a cycle of four piston strokes. The first serves for drawing in the gases in their proper arrangement and mixture ; the second compresses the charge ; during the third the gases are ignited and expand ; and finally, by the fourth the pro- ducts of combustion are expelled. The essential part of the working is performed by the first of these strokes, by which the charge is drawn in and arranged, first air, then dilute combustible mixture, and finally strong combustible mixture. This arrangement is obtained by the working of the admission slide. Moreover, after discharge of the products of combustion, a portion remains in the clearance space of the cylinder, and this constitutes the inert layer next the piston. By this peculiar arrangement of the gases, the ignition and combustion above described are rendered possible, whilst the products of previous combustion form a cushion, saving the piston from the shock of the 40 THE GAS, PETROL, AND OIL ENGINE explosion of the strongly combustible mixture at the further end of the cylinder.' Having stated the essence of Otto's invention, Dr. Slaby pro- ceeds to compare the Otto and Lenoir indicator diagrams, to show that the Otto diagrams prove that the above actions occur in the engine. He finds that the Otto expansion line is somewhat above the adiabatic line, and that the Lenoir expansion line is below it. That is, the Otto diagram gives evidence of heat being added or com- bustion proceeding in the cylinder during the whole expansion stroke, and the Lenoir diagram gives evidence of loss of heat, not gain, during a similar period. If a mass of expanding gas traces on the diagram the adiabatic line, then it appears as if no loss of heat occurred ; but as the temperature of the flame filling the cylinder is known to exceed 1200° C, it must be losing heat to the water jacket. To make the expansion line keep up to the adiabatic a great flow of heat into the gas must be taking place, and as the only source of heat is combustion, it follows that the gas is burning during the expansion period. Dr. Slaby calculates the proportion of heat evolved by the explosion in the Otto engine as 55 per cent., leaving 45 per cent, to be evolved during expansion. This he states is due to the portion of the charge which continues to burn after the explosion. The curve differs from Lenoir's in this, that while in Lenoir's engine all the heat is evolved at the moment of explosion, leaving none to be evolved during expansion, in Otto's only a part is evolved at first, and the reserved portion keeps up the temperature during expansion. He concludes from his experiments that the action of the Otto engine is truly as Mr. Otto states in his specification — explosion is suppressed and a slow evolution of heat is obtained, and this slow evolution of heat is the result of the invention and the cause of the economy of the engine. In addition to this indirect proof, experiments have been made at Deutz and elsewhere to show directly that stratification has a real existence in the Otto engine. An Otto engine was constructed, specially fitted with two igniting valves ; one valve was placed on the^ide of the cylinder at the end of the explosion space next the piston, so that it could ignite the gases at the piston ; the other valve was the usual one at the end of the cyUnder, igniting the gases in the admission port. Experiments were made to discover if the side valve would fire the mixture at the piston ; it was found that it did so. Consecutive ignitions were obtained there. Diagrams were taken for comparison, with the end and the side valves in alternate action, care being taken to keep the charge in HISTORICAL SKETCH 41 the same proportions during the trials. It was found that although the side valve ignited as regularly as the end valve, yet the diagrams were different. Instead of the usual rapid ascending explosion line, the explosion took place more slowly, and the maximum pressure was not attained till late in the stroke. The ignitions were slower from the side valve than from the end valve. If an uninflammable cushion, such as Dr. Slaby so clearly describes, existed at the piston, one would expect that the side valve would fail entirely, but it ignited quite regularly although more slowly than the end valve. This experiment is considered to prove stratification. To make stratification visible to the eye, a small glass model was constructed. It consisted of a glass cylinder of about i| in. internal diameter, containing a tightly packed piston connected to a crank ; the stroke was about 6 ins. ; when full back, the piston left a considerable space to represent the explosion space. A brass cover was fitted to the end of the tube, and in it was bored a hole of about I in. diameter, representing the admission port ; in this hole was screwed a pet cock to which a cigarette was affixed. On lighting the cigarette and then moving the piston forward by the crank, it was seen that the smoke of the cigarette which passed in did not completely fill the cylinder ; the smoke slowly oozed in and left a large clear space between it and the piston. The smoke was supposed to represent the charge of gas and air rushing in, and the clear air behind the piston the cushion which was said to exist in the Otto engine. It was supposed that in the glass cylinder was repeated on a small scale the action of the gases occurring on a larger scale in the Otto engine. In a paper in a German engineering journal, Dr. Slaby recounts this experiment, and lays great weight upon it. He considers that it undoubtedly proves the truth of the Otto theory. In discussion Mr. John Imray concisely states the Otto position as follows : ' The change which Mr. Otto had introduced, and which rendered the engine a success was this : that instead of burning in the cylinder an explosive mixture of gas and air, he burned it in company with and arranged in a certain way in respect of, a large volume of incom- bustible gas which was heated by it, and which diminished the speed of combustion.' And Mr. Bousfield states it in similar terms : ' In the Otto gas engine the charge varied from a charge which was an explosive mixture at the point of ignition to a charge which was merely an inert fluid near the piston. When ignition took place, there was an explosion close to the point of ignition that was gradually 42 THE GAS, PETROL, AND OIL ENGINE communicated throughout the mass of the cylinder. As the ignition got further away from the primary point of ignition, the rate of trans- mission became slower, and if the engine were not worked too fast the ignition should gradually catch up the piston during its travel, all the combustible gas being thus consumed. When the engine was worked properly the rate of ignition and the speed of the engine ought to be so timed that the whole of the gaseous contents of the cylinder should have been burned out and have done their work some little time before the exhaust took place, so that their full effect could be seen in the working of the engine. This was the theory of the Otto engine.' From these quotations it will be seen that Mr. Otto's supporters agree that Mr. Otto has invented a means of suppressing explosion and substituting for explosion a regulated combustion, and that this process is the cause of the economy of the engine. They are agreed that he has succeeded in preventing explosion, and that he does this by arranging or stratifying the charge which is to be used. They consider that engines previous to Mr. Otto's were wasteful because they used a homogeneous and therefore explosive charge, and that Mr. Otto's engine is economical because it uses a heterogeneous or stratified charge, which is consequently non-explosive. Discussion of Mr. Otto's Theory. — The primary fallacy of Mr. Otto's theory lies in the assumption that previous engines were more explosive than his, and that in previous engines all the heat was evolved at once : as a plain matter of fact this is incorrect. In the Lenoir and Hugon engines, as in all explosive engines, little more than one-half of the total heat is evolved by the explosion,' and the portion reserved is evolved during the stroke of the engine. The following test of a Lenoir engine, made by the author in London, very clearly shows the suppression of heat at first : Lenoir engine rated at i horse-power. Cylinder y\ ins. diameter ; stroke iif ins. Average revolutions during test, 85 per minute. Gas consumed in one hour, 86 cubic feet. With full load, indicated horse-power, x'xy (average of 9 diagrams). Gas consumed per indicated h#:se-power per hour, 73^5 cubic feet. Maximum temperatures of explosion, 1100° to 1200° C. Mixture in engine i vol. coal gas, I2'5 vols, of air and other gases. Heat evolved by explosion, 60 per cent, of total heat. ' This assumes the working fluid to be air of constant specific heat at high as at low temperatures. See discussion on the working fluid at a later part of this book. HISTORICAL SKETCH 43 The proportion of the mixture was calculated from the points of cut-off on the diagram and after making allowance for the volume of burned gases in the clearances of the engine. It will be observed that only 60 per cent, of the gas is burned at first, leaving 40 per cent, to be burned during the stroke, and also that the temperature of the explosion never exceeds 1200° C. Now in the Otto engine, according to Thurston, 60 per cent, of the heat is evolved at explosion, and 40 afterwards, and the usual maximum temperature is about 1600° C. So that, so far as the slowness of the explosion is concerned, there is no difference, and in the intensity of the temperature produced the Otto exceeds the Lenoir. It is difficult to understand how Dr. Slaby could fall into so obvious an error as he did, and suppose that more heat was kept back in the case of the Otto explosion. At the time he wrote his report, accounts of Hirn's, Bunsen's, and Mallard's experiments on explosion were in existence, aU of them agreeing on the fact of a large suppression of heat at the maximum temperature of the explosion, although differing in the explanation of the fact. Him even stated that in the Lenoir engine the pressures fell far short of what should be, if all the heat were evolved at once. Yet Dr. Slaby, in the presence of all this definite and carefully ascertained knowledge, is astonished when he finds only 55 per cent, of the total heat evolved by the explosion in the Otto engine, and the only explana- tion which occurs to him is that of stratification. If stratification exists at all in the engine, then it produces no measurable change in the explosion ; it neither retards the evolution of heat nor does it moderate the temperature. The explosion and expansion curves are precisely what they would have been with a homogeneous charge. The mere fact that heat is suppressed in the Otto explosion proves nothing, because a precisely equivalent amount of heat is suppressed in aU gaseous explosions, and Dr. Slaby's contention, based upon the supposed peculiarity of the Otto, falls to the ground. Dr. Slaby has been led into error by the fact that the expansion line of the Lenoir diagram falls below the adiabatic, while the expan- sion line of the Otto diagram remains slightly above it or upon it. He assumes that in the Lenoir no heat is being added during expan- sion, whereas just as much heat is being added, or just as much com- bustion is proceeding during the Lenoir stroke, only the cooling of the cylinder walls is greater, and the heat is abstracted so rapidly that the line falls below the adiabatic. This is due to two causes : (i) the greater proportional cooling-surface exposed by the Lenoir engine, and (2) a longer time of exposure. The absence of compression and the slow piston speed makes the loss greater. 44 THE GAS, PETROL, AND OIL ENGINE Although quite as much heat is evolved during the stroke, it is overpowered by the greater cooUng, and the line falls under the adiabatic. This fall is evidence of greater cooling, not of less evolution of heat. In a paper,^ ' Die Verbrennung in der Gasmaschine,' Professor Schottler makes this explanation of the difference between the lines, and states that ' Whether stratification exists or does not exist in the Otto engine it is unnecessary, and is not the cause of the slow falling of the expansion line.' In all crucial points the Otto theory breaks down, as proved by diagrams taken from his engine. The explosion is not suppressed ; the maximum temperatures produced are not lower than those previously used ; the mixture used is not more diluted than in the previous engines, and the in- tensity of the pressures, as well as the rate of their application, is greater. The mixture in the engine from Slaby's figures is i vol. coal gas to io"5 vols, of other gases, and from Thurston's figures i vol. coal gas to 9"i vols, of other gases, while Lenoir often used i vol. coal gas to 12 of air. The engine, instead of using a less explosive power than the Lenoir engine, uses one more intensely explosive. The effect of the reduction of cooling surface and increase of piston velocity is to diminish the loss of heat to the cylinder waUs, and the slowly descending line is not the cause of the economy, but is the effect and evidence of it. Stratification. — The inquiry into the existence or non-existence of stratification in the cylinder has no practical bearing on the question of economy, as the explosion curves act precisely as they would with homogeneous mixtures. Scientifically, however, the question is interesting and will be shortly considered. The evidence which it is considered proves its existence in the Otto engine is in the author's opinion most unsatisfactory. Dr. Slaby distinctly asserts the existence of an inert stratum next the piston, ' interposed between the combustible charge and the piston,' and Mr. Imray speaks of the ' arrangement of the charge in respect of a large volume of incombustible^ases,' and Mr. Bousfield of ' a charge which was merely an inert *uid next the piston.' Yet all the evidence in support of these positive assertions is given by one experiment made with an Otto engine, and one with a small glass model. The evidence given by the experiment on the engine itself, in the author's opinion, disproves stratification in the Otto sense altogether. If the inert stratum next to the piston had any real ' Zeitschrift. des Vereines deutscher Ingenieure. Band xxx. , Seite 209. HISTORICAL SKETCH 45 existence, then the side igniting valve, in the experiment made by Mr. Otto, should not have ignited the mixture at all. The fact that it did ignite regularly and consecutively proved most distinctly that the gas next the piston was not inert but was explosive, and being explosive in itself it could not act as a cushion to absorb heat or shock. That experiment alone settles the question, and proves at once the visionary nature of the cushion of inert gas next the piston. The fact that the ignitions were slower than those from the end slide does not get rid of the fact that ignition did take place, and to those who understand the sensitive nature of any igniting valve, it will not be difficult to comprehend how small a difference in adjust- ment will cause late and slow ignitions. At the very utmost the experiment points to a small difference in the dilution of the explosive mixture at the piston and that at the end port. Experiments made by the author also prove that the mixture in the Otto cylinder is present in explosive proportions close up to the piston. The piston of a 3 J HP Otto engine was bored and fitted with a screw plug, which carried a small spiral of platinum wire in electrical connection with a battery ; the platinum spiral proj acted from the inner surface of the piston by a quarter of an inch. When the engine was running in the usual way, the wire was made incan- descent by the battery and the external light was put out. It was proved that by a little care in getting the platinum to a certain tem- perature the engine worked as usual, igniting regularly and consecu- tively. The spiral was made just hot enough to ignite when com- pression was complete, but not hot enough to ignite before compressing. If an incombustible stratum had existed even so close to the piston as i in. then the wire should never have been able to ignite the charge at all. If the wire was made too hot, then ignition often took place while the charge was still entering, proving that no stratification existed even while the charge was incomplete. A little consideration of the arrangement of the Otto engine will show that stratification cannot have any existence in it. The end of the combustion space is usually flat, and sometimes the admission port projects slightly into it ; the area of the admission port is about ^'^th of the piston area ; accordingly the entering gases flow into the cylinder at a velocity thirty times the piston velocity, or at the Otto piston speed, about 120 miles an hour. Great commotion inevitably occurs ; the entering jet projects itself through the gases right up against the piston, and then returns eddying and whirling till it mixes thoroughly with whatever may be in the cylinder. The mixture becomes practically homogeneous even before compression commences. 46 THE GAS, PETROL, AND OIL ENGINE Experiments made by the late Dr. John Hopkinson and the author on full-size glass models of the Otto cylinder show this mixing action very beautifully. A 3| HP Otto cylinder was copied in every propor- tion in glass, and the valve was so arranged that it passed a charge of smoke at the proper time. The piston was placed at the end of its stroke, leaving the compression space filled with air. When pulled forward the valve opened to a chamber filled with smoke, and the smoke rushed through the port, projected right through the air in the space, struck the piston, and filled the cylinder uniformly, much faster than the eye could follow it. It mixed instantaneously with the air in the cylinder without evincing the slightest tendency to arrange itself in the manner imagined by Mr. Otto. Mr. Otto's experiment with a cigarette and glass cylinder does not, in the most remote degree, imitate the conditions occurring in his engine ; the proportions are quite wrong. The model is much too small, and the glass cylinder is too long in proportion t-o its diameter ; then the gases are so badly throttled by passing through the cigarette, that when the piston is moved forward it leaves a partial vacuum behind it, and only a little smoke enters, not nearly enough to foUow up the piston, but only sufficient to ooze into the back of the cylinder while the piston moves forward and expands the air which is already in the cylinder. It was easy for Mr. Otto to have copied his cylinder and valve full size and imitated precisely the conditions existing in his engines. Had he done this he would have proved complete mixing instead of stratification. Why did he refrain from doing this ? The question at issue is not, Can stratification be obtained by a specially devised form of apparatus ? — no one doubts that it can — but. Does stratification exist in the Otto engine ? If it does not exist in the Otto engine then it is perfectly plain that it cannot be the cause of the economy of the motor, and it is quite certain that it cannot exist in the Otto engine. Professor Schottler, in the paper already referred to, also arrives at the conclusion that stratification has no existence in the Otto engine, and that Mr. Otto's small glass model does not truly represent the actions occurring in the engine. In aU gas engines, when the charge enters the cylinder through a port the residual gases in the port%re swept into the cylinder, and while the port itself is filled with gas and air mixture, free from admixture with residual gases, the cylinder contains the gas and air mixture diluted with whatever residual gases exist in the engine which have not been expelled by the piston. The mixture in the port is accordingly stronger and more inflammable than the mixture in the cylinder. In the Lenoir and Hugon engines this occurred to a marked HISTORICAL SKETCH 47 extent ; in the Hugon engine as much as 30 per cent, of the whole charge consisted of residual gases, and the charge in the cylinder was considerably more dilute than that in the admission port. In the Otto engine this also occurs, but it is not stratification, and it is not a new invention ; the cylinder is filled with explosive mixture more dilute than that in the ignition port, but still explosive throughout. Some space has been devoted here to the discussion of the erroneous theory of the gas engine, as it was very necessary to arrive at correct ideas in order to see the line of true advance. Had the stratification theory been true, advance could not have been made by increasing compression pressures by diminishing the volumes of the compression spaces. This course, however, has been followed, and while in 1876 the compression pressure was only about 30 lbs. per square inch above atmospheric, in 1908 the latest Otto cycle engines utilise pressures as high as 170 lbs. per square inch. The effect of correct theory upon gas engine advance has been most marked, and facts have proved how much it was necessary to arrive at true conceptions. Recent experiments have thrown a flood of light upon many of the properties of the working fluid, which will aid materially in further advance, but these matters are fully dealt with later. Large Gas Engines The Societe Cockerill, of Seraing, continued their work on the large gas engine, and at the end of 1899 they had in operation there a single-cylinder Otto cycle engine, cylinder 51-2 ins. diameter and stroke 55'i3 ins. driving a blowing cylinder for a blast furnace. The following particulars will give an idea of its huge dimensions. Diameter of crank shaft, i8-ii ins. Piston rod diameter, 9-6 ins. Height of engine above ground, I3'i feet ; length, 36'i feet ; width, 197 feet. Total weight of engine and fi}^wheel without blower, 124-9 tons. The engine runs at ninety revolutions per minute, gives 600 brake horse-power, and converts 28 per cent, of the heat given to it into indicated work in the cylinder. A number of these engines are shown in their external aspect at fig. 20 ; their dimensions will be appreciated by observing the workmen shown standing by them. The engine was fully tested by Professor Hubert, of Liege University, early in 1900, and reported as working weU and economically. It was inspected by the author in 1901, and it was then performing its work with great smoothness and regularity. 48 THE GAS, PETROL, AND OIL ENGINE About the same time the Gasmotoren Fabrik Deutz took up the subject and built an engine of looo HP with four cyhnders, each of 33-in. diameter and 39-3-in. stroke, which was set to work at the Horde Iron Works : it indicated 1200 HP at 135 revolutions per minute. This engine was coupled direct to a dynamo. The Deutsche - Gas - Kraft Gesellschaft built large Oechelhauser engines on the two-cycle principle, and two 600 HP sets of these engines were also erected at Horde ; each set had two cylinders having two pistons in each : the diameter of the cylinders was i8"9 ins., and the combined stroke of each pair of pistons was 63 ins. Messrs. Korting, of Hanover, then appeared with a two-cycle engine operating on the Clerk cycle, which speedily took an important position among large gas engines. Messrs. Crossley Brothers, in England, took up the large gas engine at an early date, and a 400 HP engine by them was at work at Messrs. Brunner, Mond & Co.'s works, Winnington, in 1900 ; it had two cylinders of 26 ins. diameter and 36 ins. stroke running at 150 revo- lutions per minute. The Premier Co., too, have long devoted attention to the large gas engine ; they also had a 500 HP engine at work at Winnington in 1900 : it had two cylinders 28^ ins. diameter by 30 ins. stroke, and ran at 125 revolutions per minute. It was supplied with a scavenging pump. They now build engines up to 2000 HP. The National Gas Engine Co., the Hornsby- Stockport Co., Tangyes Ltd., the Campbell Gas Engine Co., Messrs. Fielding & Piatt, Ltd., and other English gas-engine builders now construct compara- tively large engines, but the largest engines of the double-acting type have been built on the Continent mainly. Most English makers of long experience prefer open-cylinder engines with pistons having no watering arrangements, whereas the standard continental type for high powers is now double-acting four- or two-cycle — single motor cylinders for the two-cycle type, double tandem cylinders for the four- cycle. English makers have now taken up the manufacture of the leading continental types. Messrs. Richardson, Westgarth & Co., Ltd., the Cockerill engine ; Messrs. Mather & Piatt, Ltd., the Korting two-cycle ; Messrs. Beardmore, the OechelhausA; the Lilleshall Co., the Niirnberg engines ; but in the case of the Cockerill, the Korting, and the Oechel- hauser considerable alterations have proved necessary to suit the engines to English use. Korting engines are built giving 1000 HP per motor cylinder, but Messrs. Ehrhardt & Sehmer claim to exceed this on an Otto cycle engine with an engine giving 1,100 HP per double-acting cylinder. Such an engine with two cranks and two cylinders in tandem to each HISTORICAL SKETCH 49 crank gives a unit of 4,400 HP. Such an engine has cylinders of about 45J ins. diameter by 51J ins. stroke, and runs at 94 revolutions per minute. Large engines are undoubtedly making great progress, as will be seen from the following interesting figures by Mr. R. E. Mathot, of Brussels, giving the numbers and horse-power of large gas engines recently manufactured in Europe : Messrs. Crossley Brothers, Ltd., 57 motors, with an aggregate of 23,660 HP ; Messrs. Ehrhardt & Sehmer, 59 motors, total 69,790 HP ; the Otto Gasmotoren Fabrik, 82, total 47,400 HP ; Gebriider Korting, 198, total 165,760 HP ; Societe Alsacienne, 55, total 23,410 HP ; Soci^te John Cockerill, 148, total 102,925 HP ; Societe Suisse, Winterthur, 67, total 8,620 HP ; Vereinigten Maschinenfabrik Augs- burg and Niirnberg, 215, total 256,240 HP. The mean power of each gas engine made by Messrs. Ehrhardt & Sehmer and the Augsburg and Niirnberg companies is in each case 1,200 HP. It is stated that in one factory there are gas engines representing a total output of 35,000 HP. These European large gas engines thus give nearly 575,000 HP between them. In America, too, the large gas engine has made considerable progress. Mr. E. L. Adams estimates that 350,000 HP is now at work or in construction in the United States. The first large engines installed were of the Korting-Clerk type, built by the De La Vergne Co., of New York: sixteen blowing engines of 2000 HP each and eight electric driving engines of 1000 HP each. They were set to work in igo2. Large engines have been built later by the Westinghouse Co., Pitts- burg, of the horizontal twin tandem type, having two cranks and four double-acting cylinders on each unit, cylinder 38 ins. diameter by 54 ins. stroke ; and the Snow Steam Pump Co. have under construction similar horizontal tandem engines having cylinders of 42 ins. diameter by 54 ins. stroke. The Westinghouse Co. in England has also de- voted great attention to the large gas engine, and they have built very interesting multiple-cylinder engines of the single-acting open- trunk type, one of which was at work at the Franco-British Exhibition in the year 1908. It gives 750 HP, is vertical, and has four cranks and eight cylinders, using no watering for the pistons. Many interesting developments are now in progress; over 2,000,000 HP of smaller gas engines are at work in the world, and certainly over 1,000,000 HP of petrol motors. The application of large gas engines to marine work, the com- pounding of the gas engine, and many other matters are being strenuously pursued. It may be said, indeed, without exaggeration, that the whole world is now alive to the possibilities of the internal-combustion VOL. I. E 50 THE GAS, PETROL, AND OIL ENGINE motor, and that progress will be more and more rapid. These motors have almost fulfilled the expectations of those engineers who — like the author — have devoted a large part of their lives to their study and advancement. They are looking forward now to the completion of the work begun so many years ago, and expect, at no distant date, to find the internal-combustion motor competing with the steam engine even in its latest form, the steam turbine, on sea as vigorously as it does at present on land. CHAPTER I THE GAS ENGINE METHOD Gas ENGINES, while differing widely in theory of action and mechanical construction, possess one feature in common which distinguishes them from other heat engines : that feature is the method of heating the working fluid. The working fluid is atmospheric air, and the fuel required to heat it is inflammable gas. In aU gas engines yet produced, the air and gas are mixed intimately with each other before introduction to the motive cylinder; that is, the working fluid and the fuel to supply it with heat are mixed with each other before the combustion of the fuel. The fuel, which in the steam and in most hot-air engines is burned in a separate furnace, is, in the gas engine, introduced directly to the motive cylinder and burned there. It is, indeed, part of the working fluid. This method of heating may be called the gas-engine method, and from it arises at once the great advantages and also the great difficulties of these motors. Compare first with the steam engine. In it there exist two great causes of loss : water is converted into steam, absorbing a great amount of heat in passing from the liquid to the gaseous state ; after it has been used in the engine it is rejected into the atmosphere or the condenser, still existing as steam. The heat necessary to convert it from the liquid to the gas is consequently in most part rejected with it. Loss, occurring in this way, would be small if high tempera- tures could be used ; but this is the point where steam fails. High temperatures cannot be obtained without pressure so great as to be quite unmanageable. The attempt to obtain high temperatures by superheating has often been made, but without any substantial success. Although the difiiculty of excessive pressure is avoided, another set of troubles is introduced. AU the heat to be given to the gaseous steam must pass through the iron plates forming the boiler or superheater, which plates will only stand a comparatively low temperature, certainly not exceeding that of a low red heat, or about 600° to 700° C. Steam, being a gas, is much more difficult 52 THE GAS, PETROL, AND OIL ENGINE to heat than water ; it follows that even these temperatures cannot be attained without enormous addition to the heating surface. The difficulties of making a workable engine using high-temperature steam are so great that even so distinguished an engineer and physicist as the late Sir C. W. Siemens failed in his attempts, which extended over many years. It may be taken, then, that low temperature is the natural and unavoidable accompaniment of the steam method, arising from the necessary change of the physical state of the working fluid and the limited temperature which iron will safely bear. The originators of the science of thermodynamics have long taught that the maximum efficiency of a heat engine is obtained when there is the maximum difference between the highest and lowest temperatures of the working fluid. So long ago as 1854 Professor Rankine read a paper before the British Association, ' On the means of realising the advantages of the Air Engine,' in which he expresses his belief that such engines wiU be found to be the most economical means of develop- ing motive power by the agency of heat. In this opinion he stood by no means alone. Engineers so able as Stirling, Ericsson, and Siemens, physicists so distinguished as Dr. Joule and Sir Wm. Thomson, devoted much energy and study to their practice and theory. Not- withstcmding all their efforts, aided by a host of less able inventors, the difficulties proved too formidable ; and although more than fifty years have now passed since Rankine announced his belief, the hot-air engine proper hcis made no real advance. Similar causes to those acting in the steam engine impose a limit here. It is true the complication of changing physical state is avoided, but the limited resistance of iron to heat acts as powerfully as ever. Air is much more difficult to heat than water, and therefore requires a much larger surface per unit of heat absorbed. In the larger hot-air engines, accordingly, the furnaces and heating surfaces gave great trouble. Very low maximum temperatures were attained in practice. In a Stirling engine giving out thirty-seven brake horse-power, the maxi- mum temperature was only 343° C. ; in the engines of the ship ' Ericsson,' the maximum was only about 212° C, according to Rankine, the indi- cated power being about 300 horses. These figures show that the heating surfaces were insufficient, as in botk cases the furnaces were pushed to heat the metal to a good red. A method of internal firing was proposed, first by Sir George Cayley and afterwards carried out with some success by others ; the furnace was contained in a completely closed vessel, and the air to be heated was forced through it before passing to the motor cylinder. The plan gave better results, but the temperature of 700° C. was stUl the limit, as the strength of the iron reservoir had to be considered, and the hot gases had to pass through valves. Wenham's engine, described in a paper read before the THE GAS ENGINE METHOD 53 Institution of Mechanical Engineers in 1873, is a good example of this class. In it the highest temperature of the working fluid, as measured by a pyrometer, was 608° C. ; higher temperatures could easily have been got, but the safety of the engine did not permit it. Professor Rankine in his work on the steam engine has very fully discussed the disadvantages arising from low maximum temperatures. He cal- culates that in a perfect air engine without regenerator an average pressure of 8'3 lbs. per square inch would only be attained with a maximum of 2i6'6 lbs. per square inch, thus necessitating great strength of cylinder and working parts for a very small return in effective power. In the ' Ericsson,' the average effective pressure was less than this, being only about 2 lbs. per square inch ; it had four air cylinders each of 14 feet diameter, and only indicated 300 horse-power. Stirling's motor cylinder did not give a true idea of the bulk of the engine, as the real air-displacer weis separate. Even with Wenham's machine the bulk was excessive, an engine of 24 ins. diameter cylinder and 12 ins. stroke giving 4 horse-power. Those facts sufficiently illustrate the practical difficulties which prevented the development of the hot-air engine proper. All flow from the method of heating. Low temperature is necessary to secure durability of the iron. All hot-air engines are, therefore, very large and very heavy for the power they are capable of exerting. The friction of the parts is so great that although the theo- retical efficiency of the working fluid is higher than in the best steam engines, the practical efficiency or result per horse available for external work is not nearly so great. The best result ever claimed for Stirling's engine is 27 lbs. of coal per brake horse-power per hour, probably under the truth, but even allowing it, a first-class steam engine of to-day will do much better. According to Professor Norton, the engines of the ' Ericsson ' used xSy lb. of anthracite per indicated horse-power per hour ; but the friction must have been enormous. Compared with the steam engine, the practical disadvantages of the hot-air engine are much greater than its advantage of theory. Owing to the great inferiority of air to boiling water as a medium for the convection of heat, the efficiency of the furnace is much lower ; owing to the high maximum and low available pressure, the friction is much greater — which disadvantages in practice more than extinguish the higher theoretical efficiency. The gas engine method of heating by combustion or explosion at once disposes of those troubles ; it not only widens the limits of the temperatures at command almost indefinitely, but the causes of failure with the old method become the very causes of success with the new. 54 THE GAS, PETROL, AND OIL ENGINE The difficulty of heating even the greatest masses of air is quite abolished. The rapidly moving flash of chemical action makes it easy to heat any mass, however great, in a minute fraction of a second ; when once heated the comparatively gradual convection makes the cooling a very slow matter. The conductivity of air for heat is but slight, and both losing and receiving heat from enclosing walls are carried on by the process of convection, the larger the mass of air the smaller the cooling surface relatively. Therefore the larger the volumes of air used, the more economical the new method, the more difficult the old. The low conductivity for heat, the cause of great trouble in hot-air machines, becomes the unexpected cause of economy in gas engines. If air were a rapid carrier of heat, cold cylinder gas engines would be impossible. The loss to the sides of the enclosing cylinders would be so great that but little useful effect could be obtained. Even as it is, present loss from this cause is sufficiently heavy. In the earlier engines as much as three-fourths of the whole heat of the combustion was lost in this way ; in the best modern engines so much as one-quarter is still lost. A little consideration of what is occurring in the gas-engine cylinder at each explosion will show that this is not surprising. Platinum, the most infusible of metals, melts at about 1700° C. ; the ordinary temperature of cast iron flowing from a cupola is about 1200° C. ; a temperature very usual in a gas-engine cylinder is 1600° C, a dazzling white-heat. The whole of the gases filling the cylinder are at this high temperature. If one could see the interior it would appear to be filled with a blinding glare of light. This experiment the writer has tried by means of a smaU aperture covered with a heavy glass plate, carefully protected from the heat of the explosion by a long cold tube. On looking through this window while the engine is at work, a continuous glare of white light is observed. A look into the interior of a boiler furnace gives a good notion of the flame filling the cylinder of a gas engine. At first sight it seems strange that such temperature can be used with impunity in a working cylinder ; here the convenience of the method becomes evident. The heating being quite inde- pendent of the temperature of the KaUs of the cylinder, by the use of a water jacket they can be kept at any desired temperature. The same property of rapid convection of heat, so useful for generating steam from water, is essential in the gas engine to keep the rubbing surfaces at a reasonable working temperature. In this there is no difficulty, and notwithstanding the high temperature of the gases, the external surface of metal itself never exceeds the boiling-point of water. So good a result cannot of course be obtained without careful THE GAS ENGINE METHOD 55 proportioning of the cooling surfaces for the amount of heat to be carried away ; in all modern engines this is carefuUy attended to, with the gratifying result that the cylinders take and retain a polished surface for years of work just as in a good steam engine. The gas engine method gives the advantage of higher tempera- ture of working fluid than is attainable in any other heat engine, and at the same time the working cylinder metal may be kept as cool as in the steam engine. It also allows of any desired rate of heating the working fluid in any required volumes. In consequence of high temperatures the available pressures are high, and therefore the bulk of the engine is small for the power obtained. It realises all the thermodjmamic advantages claimed for the hot-air engine without sacrificing the high available pressures and rapid rate of the generation of power which is the characteristic of the steam engine. For rapid convection of heat existing in the steam boiler is sub- stituted the still more rapid heating by explosion or combustion, a rapidity so superior that the power is generated for each stroke separately as required, there being no necessity to coUect a great magazine of energy. The only item to the debtor side of the gas engine account is the flow of heat through the cylinder walls, which disadvantage is far more than paid for by the advantages. CHAPTER II GAS ENGINES CLASSIFIED Although the gas-engine patents now in existence number many thousands, the essential differences between the inventions are not great. In their working process they may be divided into a few well-defined types : 1. Engines igniting at constant volume, but without previous compression. 2. Engines igniting at constant pressure, with previous com- pression. 3. Engines igniting at constant volume, with previous com- pression. The first type is the simplest in idea ,- it is the most apparent method of obtaining power from an explosion. In it the engine draws into its cylinder gas and air at atmospheric pressure, for a part of its stroke, in proportions suitable for explosion ; then a valve closes the cylinder, and the mixture is ignited. The pres- sure produced pushes forward the piston for the remainder of its travel, and upon the return stroke the products of the combustion are expelled exactly as the exhaust of a steam engine. By repeating the same process on the other side of the piston, a kind of double-acting engine is obtained. It is not truly double-acting, as the motive impulse is not applied during the whole stroke, but only during that portion of it left free after performing the necessary function of charging with the explosive mixture. The working cycle of the engine consists of four operations : 1. Charging the cylinder with explosive mixture. 2. Exploding the charge. 3. Expanding after explosion. 4. Expelling the burned gases. To carry it out in a perfect manner, the mechanism must be so arranged that during the charging the pressure of the gases in the cylinder does not fall below atmosphere ; there must be no throttling of the entering gases. The cut-off and the explosion must be absolutely simultaneous and also instantaneous, so that the heat may be applied without change of volume, and thereby GAS ENGINES CLASSIFIED 57 produce the highest pressure which the mixture used is capable of giving. The expansion will be carried far enough to reduce the pressure of the explosion to atmosphere ; and the exhaust stroke will be accomplished without back pressure. The charge in entering must not be heated by the walls of the cylinder, but should remain at the temperature of the atmosphere till the very moment previous to ignition. At the same time, the cyUnder should not cool the gases after the explosion and no heat should disappear except through expansion doing work. Although all these conditions are necessary to the perfect cycle, it is evident that no actual engine is capable of combining them. Some throttling at the admission of the mixture, and a little back pressure during the exhausting are unavoidable ; some time must .elapse between the closing of the inlet valve and the explosion, in addition to the time taken by the explosion itself. Heat will be communicated to the entering gases and lost by the exploded gases to the walls of the cylinder. The actual diagram taken from an engine will therefore differ .considerably from the theoretical one. The theoretical conditions are to a great extent contradictory. The idea of the type, however, is easily comprehensible, and .evidently suggested by the common knowledge of the destructive effect of accidental coal-gas explosions which occurred soon after the introduction of gas into general use. 'The power is there, let us use it like steam in the cylinder of a steam engine,' said the early inventors. The two most successful engines of this type were Lenoir's and, later, Hugon's. The second type is not so simple in its main idea, and required much greater knowledge of detail, both mechanical and theoretical. As a hot-air engine its theory was originally proposed by Sir George ■Cayley, and, later, by Dr. Joule and Sir Wm. Thomson. As a hot- .air engine it failed for the reasons discussed in the previous chapter. In it the engine is provided with two cylinders of unequal capacity ; the smaller serves as a pump for receiving the charge and compressing it, the larger is the motor cylinder, in which the charge is expanded .during ignition and subsequent to it. The pump piston, in moving forward, takes in the charge at atmospheric pressure, in returning compresses it into an intermediate receiver, from which it passes into the motor cylinder in a compressed state. A contrivance similar to the wire gauze in a Davy lamp commands the passage between the receiver and the cylinder, and permits the mixture to be ignited on the cylinder side as it flows in -without the flame passing back into the receiver. 58 THE GAS, PETROL, AND OIL ENGINE The motor cylinder thus receives its working fluid in the state of flame, at a pressure equal to, but never greater than, the pressure of compression. At the proper time the valve between the motor and the receiver is shut, and the piston expands the ignited gases till it reaches the end of its stroke, when the exhaust valve is opened, and the return expels the burned gases. The ignition here does not increase the pressure, but increases the volume. The pump, say, puts one volume or cubic foot into the receiver ; the flame causes it to expand while entering the cylinder to two cubic feet. It does the work of two cubic feet in the motor cylinder, so that, though there is no increase of pressure, there is nevertheless an excess of power over that spent in compressing. In the first type of engine the heat is given to the working fluid at constant volume, in the second type the heat is given to the working fluid at constant pressure during change of volume. The working cycle of the engine consists of five operations : 1. Charging the pump cylinder with gas and air mixture. 2. Compressing the charge into an intermediate receiver. 3. Admitting the charge to the motor cylinder in the state of flame, at the pressure of compression. 4. Expanding after admission. 5. Expelling the burned gases. To carry out the process perfectly the following conditions would be required : No throttling during admission of the charge to the pump. No heating of the charge as it enters the pump from the atmo- sphere. No loss of the heat of compression to the pump and receiver walls. No throttling as the charge enters the motor cylinder from the receiver. No loss of heat by the flame to the sides of the motor cylinder and piston. And last, No back pressure during the exhaust stroke. The exhaust gases also must be completely expelled by the motor piston ; that is, the motor cylinder sj|ould have no clearance. The requirements of this type, although sufficiently numerous and exacting, are not so contradictory among themselves as in the first. Although every engine of the kind yet made fails to fulfil them, it is quite possible that a machine very closely approximating may be yet constructed. The most successful engines of this kind have been Brayton's. and Simon's^ the first an American invention, and the second an. GAS ENGINES CLASSIFIED 59 English adaptation of it. Sir C. W. Siemens proposed such an engine in 1861, but does not seem to have been successful in carrying it out. In i860 it was also proposed by F. Million, but without a sufficient understanding of the mechanical detail necessary for a working machine. Brayton's engine was made in considerable numbers in America, and was applied by him to drive a good-s'^ed launch, petroleum being used as the fuel instead of gas. It was exhibited at the Centennial Exhibition in Philadelphia ; and at the Paris Exhibition of 1878 by Simon. The third type is the best kind of compression engine yet intro- duced ; by far the largest number of gas engines in everyday use throughout the world are made in accordance with its requirements. In theory it is more easily understood as requiring two cylinders, compression and power. The leading idea, compression and ignition at constant volume, was first proposed by Barnett in 1838, then by Schmidt in more general terms, very fuUy by Beau de Rochas in i860 and also by F. Million in the same year. Otto, however, was the first success- fully to apply it, which he did in 1876. The compression cylinder may be supposed to take in the charge of gas and air at atmospheric temperature and pressure ; compress it into a receiver from which the motor cylinder is supplied ; the motor piston to take in its charge from the reservoir in a compressed state ; and then communication to be cut off and the compressed charge ignited. Here ignition is supposed to occur at constant volume, that is, the whole volume of mixture is first introduced and then fired ; the pressure therefore increases. The power is obtained by igniting while the volume remains stationary and the pressure increases. Under the pressure so produced, the piston completes its stroke, and upon the return stroke the products of the combustion are expelled. In this case the working cycle of the engine consists of six opera- tions : 1. Charging the pump cylinder with gas and air mixture. 2. Compressing the charge into an intermediate receiver. 3. Admitting the charge to the motor cylinder under com- pression. 4. Igniting the mixture after admission to the motor. 5. Expanding the hot gases after ignition. 6. Expelling the burned gases. To carry out the process perfectly, similar conditions are neces- sary to those in the second type. But the conditions are more 6o THE GAS, PETROL, AND OIL ENGINE contradictory. The gases entering the cylinder under pressure must not be heated by its walls ; no heat should be added till the ignition ; then, after ignition, the gases must not lose heat to the cylinder — conditions which it is impossible for the same cylinder to fulfil simul- taneously. In the engines constructed the receiver is dispensed with, for reasons which will be explained in discussing the practical difficulties of construction ; but this does not in any way modify the theory, which shall first be discussed. The most considerably used engines of this kind are of the Otto or the Clerk type. In neither of these types does any part of the working cycle require either the heating or the cooling of the working fluid by the relatively slow processes of convection and conduction. Heating is accomplished by the rapid method of explosion or, if the term be preferred, combustion, and for the cooling necessary in all heat engines is substituted the complete rejection of the working fluid with the heat it contains and its replacement by a fresh portion taken from the atmosphere at the atmospheric temperature, which is the lower limit of the engines. This is the reason why those cycles can be repeated with almost indefinite rapidity, and why gas engines can be run at speeds equal to steam engines, while the old hot-air engines could not be run fast, because of the very slow rate at which air could be heated and cooled by contact. There still remains one important type of gas engine not included in this classification ; in it part of the efficiency is dependent on cooling by contact, and consequently only a slow rate of working stroke can be obtained. It is the kind of engine known as the free piston or atmospheric gas engine. It may be regarded as a modification of the first type. The first part of its action is precisely similar, the latter part differs considerably. It may be called Type i A. In it the piston moves forward, taking in its charge of gas and air from the atmosphere at the atmo- spheric pressure and temperature. When cut off it is ignited instan- taneously, the volume being constant and the pressure increasing ; the piston is not connected directly Jo the motor shaft, but is free to move under the pressure of the explosion, like the ball in a cannon. It is shot forward in the cylinder (which is made purposely very long) ; the energy of the explosion gives the piston velocity ; it therefore con- tinues to move considerably after the pressure has fallen by expansion to atmosphere ; a partial vacuum forms under the piston till its whole energy of motion is absorbed in doing work upon the exterior air. It then stops, and the external pressure causes it to perform its instroke, during which a clutch arrangement yokes it to the motor GAS ENGINES CLASSIFIED 6i shaft, giving the shaft an impulse. The explosion is made to give its equivalent in work upon the external air, in forming a vacuum in fact ; the vacuum is increased by the cooling of the hot gases during the return of the piston. The piston proceeds completely to the bottom of the cylinder, expelling the products of combustion. So far as the working fluid of the engine is concerned the cycle consists of five operations : 1. Charging the cylinder with explosive mixture. 2. Exploding the charge. 3. Expanding after explosion. 4. Compressing the burned gases after some cooling. 5. Expelling the burned gases. To carry it out perfectly, in addition to the requirements of the first type, the expansion should be carried far enough to lower the temperature of the working fluid to the temperature of the atmo- sphere, and the compression to atmospheric pressure again should be conducted at that temperature ; that is, the compression line should be an isothermal. This kind of engine was proposed first by Barsanti and Matteucci in 1854, by F. H. Wenham in 1864, and then by Otto and Langen in 1866. The last-named inventors were successful in overcoming the practical difficulties, and many engines were made and sold by them. Their engine, although cumbrous and noisy, was a good and economical worker. The next best known engine of the kind was Gillies's, of which a considerable number were constructed and sold. CHAPTER III THERMODYNAMICS OF THE INTERNAL-COMBUSTION ENGINE CONSIDERED AS AN AIR ENGINE Beginning with Professor Rankine, able writers have so fully treated the thermodynamics of the air engine that but little can be added to the knowledge of the subject now in existence. The gas engine method of heating, however, introduces limits of temperature so extended and cycles of action so different from those possible in the air engine proper, that something remains to be done in applying the existing data. So far as the author is aware, this had only been previously attempted by three writers prior to 1885 — Professor R. Schottler, Dr. A. Witz, and himself. Before proceeding with the special consideration of the subject, it is advisable for the sake of completeness to state briefly the general laws. In doing so Rankine will be followed as closely as possible. Thermodynamics Defined ' It is a matter of ordinary observation that heat, by expanding bodies, is a source of mechanical energy, and conversely, that mechanical energy, being expended either in compressing bodies or in friction, is a source of heat. ' The reduction of the laws according to which such phenomena take place to a physical theory or connected system of principles constitutes what is called the science of thermodynamics.' First Law of Thermodynamics Heat and mechanical energy are mutually convertible, and heat requires for its production, and produces by its disappearance, mechanical energy in the proportion of 1,390 foot-pounds for each Centigrade heat unit, a heat unit being the amount of heat necessary to heat one pound weight of water through 1° C. This is Joule's law, having been first determined by him in 1843. It holds with equal truth for other forms of energy, and is a general statement of the great truth that in the universe energy is as incapable of creation THERMODYNAMICS 63 or destruction as matter. Energy may change its form indefinitely while passing from a higher to a lower level, but it can neither be created nor destroyed. The energy of outward and visible move- ment of matter may be arrested and caused to disappear as move- ment of the whole mass in one direction, but its equivalent reappears as internal movement or agitation of the particles or molecules com- posing the body. Energy assumes many forms, but the sum of all remains a constant quantity, incapable of change of quantity, but capable of disappearing in one form and reappearing in another. Second Law of Thermodynamics Although heat and work are mutually convertible and in definite and invariable proportions, yet no conceivable heat engine is able to convert all the heat given to it into work. Apart altogether from practical limitations, a certain portion of the heat must be passed from the hot body to the cold body in order that the remainder may assume the form of mechanical energy. To get a continuous supply of mechanical energy from heat depends upon getting a continuous supply of hot and cold substances : it is by the alternate expansion and contraction of some substance, usually steam or air, that heat is converted into mechanical energy. Perfect heat engines are ideal conceptions of machines which are practically impossible, but whose operations are so arranged that, if possible, they would convert the greatest conceivable proportion of the heat given to them into mechanical work. Efficiency. — The efficiency of a heat engine is the ratio of the heat converted into mec^ianical work to the total amount of heat which enters the engine. In this work the word Efficiency, when used without qualification, bears this meaning only. The efficiency of a perfect heat engine depends upon two things alone : these are, the temperature of the source of heat and the tempera- ture of the source of cold (allowing the expression). The greater the difference between these temperatures the greater the efficiency. That is, the greater will be the proportion of the total heat converted into mechanical energy, and the smaller the proportion of the total heat which necessarily passes by conduction from the hot to the cold body. Properties of Gases. — Gases are the most suitable bodies for use in heat engines ; they are almost perfectly elastic, and they expand largely under the influence of heat. A gas is said to be perfect when it completely obeys two laws : 1. Boyle's law. 2. Charles's law. 64 THE GAS, PETROL, AND OIL ENGINE Boyle's Law. — Suppose unit volume of gas to be contained in a cylinder fitted with a piston which is perfectly tight at unit pressure. Suppose the temperature to be kept perfectly constant. Then, according to Boyle's law, however the volume may be changed by moving the piston, the pressure is always inversely proportional to volume — that is, if the volume becomes two, the pressure becomes one-half ; volume becomes three, pressure becomes one-third. The product of pressure and volume is always constant. Denoting pressure by p, and volume by v, Boyle's law is, pv = constant. Charles's Law. — If a gas kept at constant volume is heated, the pressure increases. If a gas is kept behind a piston which moves without friction so that the pressure upon the gas is always constant, the heat applied will cause it to expand. One volume of gas at o° C, if heated through i° C. will expand -?T3-j and become i^^t-.^ volume, if the pressure is constant. If the volume is constant, then its pressure wiU increase by t^ j^, that is, its pressure will become x^^s of the original. In the same way if cooled 1° C. below o° C, it will contract or diminish in pressure by ^4.?, its volume or pressure becoming |"f| of what it is at 0° C. For every degree of heat or cold above or below 0° C. a perfect gas expands of contracts by ^}-g of its volume at 0° C. From this it is evident that a perfect gas, if cooled to 273 Centi- grade degrees below 0° C. will have neither volume nor pressure. This originally gave rise to the conception of absolute zero of temperature. The absolute temperature of a body is ordinary tem- perature in degrees Centigrade + 273, just as the absolute pressure of any gas is its pressure above atmosphere plus atmospheric pressure. The absolute temperature of a body is its temperature above Centigrade zero -I- 273. The pressure or volume of a gas is therefore directly proportional to its absolute temperature. If ^= pressure for absolute temperature f, and p^ pressure for t' temperature, also absolute, then |-^|,; or if V be the volume at absolute temperature t and u' at t\ then ^=^. The Second Law {quantitative). — If heat be supplied to a perfect heat engine at the absolute temperature t, and the absolute tem- THERMODYNAMICS 65 perature of the source of cold is t', then the efficiency of that engine is, denoting it by e, T T ' It is unity minus the lower temperature divided by the upper t' temperature. The efficiency is greater or less as the fraction - is less or greater. This fraction may be diminished either by reducing T or by increasing t. The lowest available temperature is not capable of great variation, being in our climate about 290° C. absolute. It therefore follows that efficiency could only be increased by increas ingT. Suppose t' = 290" absolute and t = 580° absolute. Then E = i-it§=i-J = 0-5. Suppose t' = 290°, and T = 1450°, a temperature common in gas engines, then E = i-rm'V = i-i- = o-8. The efficiency increases with increase of the maximum tempera- ture. The second law, in its quantitative form, is the statement of the efficiency of any perfect heat engine in terms of absolute temperatures of the source of heat and the source of cold. Thermal Lines. — If a volume of air is contained in a cylinder having a piston and fitted with an indicator, the piston, if moved to and fro, wiU alternately compress and expand the air, and the indicator pencil wiU trace a line or Hues upon the card, which lines register the changes of pressure and volume occurring in the cylinder. If the piston is perfectly free from leakage, and it be supposed that the temperature of the air is kept quite constant, then the line so traced is caUed axi Isothermal line, and the pressure at any point when multiplied by the volume is a constant according to Boyle's law, pv = a. constant. If, however, the piston is moved in very rapidly, the air will not remain at constant temperature, but the temperature v/ill increase because work has been done upon the air, and the heat has no time to escape by conduction. If no heat whatever is lost by any cause, the line will be traced over and over again by the indicator pencil, the cooling by expansion doing work precisely equalling the heating by compression. This is the line of no transmission of heat, therefore, known as Adiabatic. Fig. 21 shows these two lines for air starting from atmospheric pressure and temperature. VOL. I. F 66 THE GAS, PETROL, AND OIL ENGINE The pressures at different points of the curve are related by the equation fv' = constant. The pressure when multiplied by the volume raised to the y power is always constant. The power y is the ratio between the specific heat of the air at constant pressure and its specific heat at constant volume. Accord- ing to Rankine y = i'4o8 for air. Imperfect Heat Engines. — For a complete description of the working cycle of perfect heat engines, the reader is referred to works upon the I I so r- 32 9% •• 60 SO \ \ 1 10 100 90 80 70 60 sp 40 30 20 \ V in "c \ - \ N?^ \ s % \ % > \ k \ \ i' 50 ".5 C \ ^ > ^ i^ ■s^ ^ \ ■^ a—. 10 Atmospheric Hue 20 30 40 70 80 90 100 Fig. 21.- 50 60 Volume -Compression lines for air (dry), Adiabatic and Isothermal steam engine, which contain the fullest possible details both of reasoning and of results. The working cycles of practicabk heat engines are always imperfect, that is, the operations are such that, although perfectly carried out, the maximum efficiency possible by the second law of thermo- dynamics could not be attained by them. Each cycle has a maximum efficiency peculiar to itself, which is invariably less than •*• ~ •*• , but which does not necessarily vary with T and T'. It does not always follow that increase of the higher tempera- ture causes increase of efficiency ; conversely, it does not always THERMODYNAMICS 67 follow that diminution of the upper temperature causes diminution of efficiency. Under some circumstances, indeed, the opposite effect is produced — increase of the upper temperature diminishes efficiency, whUe its diminution increases it, of course within certain limits. All the gas-engine cycles described in the previous chapter are imperfect in this sense, but all are practicable. It follows that if any one of them gives a higher efficiency than another in theory, it will also do so in practice, provided the practical losses do not increase with improved theory. It is necessary before discussing the practical losses to see how the cycles compare with each other, if each be perfectly carried out. The results obtained can then be modified by examination of the way in which unavoidable practical losses affect each cycle. Efficiency Formula If H is the quantity of heat given to an engine, and H" the amount of heat discharged by it after performing work, then, the portion which has disappeared in performing work is H — H^, supposing no loss of heat by conduction or other cause, and the efficiency of the engine is H Type I. — A perfect indicator diagram of an engine of this kind is shown at fig. 22 : the line a b c is the atmospheric line, represent- ing volume swept by the piston, the line a d is the line of pressures. From a to b the piston moves forward, taking in its charge, at atmospheric temperature and pressure; at b communication is in- stantaneously cut off, and heat instantaneously supplied, raising the temperature to the maximum, before the movement of the piston has time to change the volume. From e, the point of maximum temperature and pressure, the gases expand without loss of heat, the temperature only falling by reason of work performed till the pressure again reaches atmosphere. The curve e c is therefore adia- batic. In all cases let t be the initial temperature of the air in absolute degrees Centi- grade. T the absolute temperature after explosion or heating. t' the absolute temperature of the gases after adiabatic expansion. p the atmospheric pressure. Po the absolute pressure of the explosion. v„ the volume at atmospheric temperature and pressure. V the volume at the termination of adiabatic expansion. In the particular case of diagram fig. 22, where the expansion is 68 THE GAS, PETROL, AND OIL ENGINE continued to the atmospheric line, the formula expressing the efficiency is very simple. Calling k^, the specific heat of air at constant volume, s o U ■«>.■«, 0. BBS, ssinsssjj TT '^. s^ ^Q- .^ ^ 5tt.«l y o, E 1^ = » » ?5 _p s ■*!"*< ^ a B 5j I I I 1 I r- ■ > -CI u o d S ■o >- uX ^ OOOQOOOO On and K^ the specific heat at constant pressure, then the heat supplied to the engine is H = K, {T-f), THERMODYNAMICS and the heat discharged from it is therefore efficiency H' = K, (Ti - t) ; E_K.(T-0-K^(T'-0 and therefore E-I-,r-^^ 69 (I) It is evident that for every value of t there is a corresponding value of t', which increases with the increase of T. If T is known in terms of t, then the calculation of efficiency is very rapid, as all that is required is a knowledge of the maximum temperature of the explosion to calculate the efficiency of an engine using that maximum temperature, and perfectly fulflUing this cycle. For any adiabatic curve, the pressure multiphed by volume which has been raised to the power y is a constant ; therefore Po t>/ = py (see diagram, fig. 22), (a) and - = / which, as * = *„ is the same as — ; ^ P P, also = r- .: in equation (a) T may be substituted for p„ t for p„, t for v^, and t' for V, giving = '(-?) (2) In most engines of this type the expansion is not great enough to reduce the pressure to atmosphere before opening the exhaust valve ; it is therefore necessary to give formulse where the best con- dition is not carried out. Fig. 23 is a diagram of a case of this kind. The pressure at the termination of the stroke has fallen to p„, and the temperature to x' . The heat supplied to the engine is the same as in the first case H = K„ (t — t). The heat discharged by it cannot be so simply expressed. Sup- pose the hot gases at the pressure p„ to be allowed to cool by contact with the sides of the cylinder at constant volume till the atmospheric pressure p is reached, then the temperature P^ 70 THE GAS, PETROL, AND OIL ENGINE or in terms of volume and t t^ =''Lt, and the heat lost is k^ (t' — f). The heat to be still abstracted before the air returns to its original condition at t, and pressure p is K, {f - t). Total heat discharged by exhaust, therefore, H^ = K, (t' - t^) + K^ ((!' - i). The efficiency consequently is E = _ K,{T:-t) - {K, (t1 - f) + K, {f - t) K.(T-<) T — t ^ ' In this case there is no fixed relationship between x the tempera- ture of the explosion, and t' the temperature of the gases at the ter- mination of adiabatic expansion. As the expansion is more or less complete, so do x^ and <' change. In no case, however, can the efficiency be so great as that in the first case. Type 2. — A perfect indicated diagram of an engine of this tjrpe is shown at fig. 24. Although the cycle requires two cylinders, producing two diagrams, they are better compared when superposed. The whole diagram may be supposed to come from the motor cylinder, the shaded portion of it representing the available work of the cycle, and the unshaded part the part done by the compressing pump. The atmospheric line is ab c. The pump volume is a b, the motor volume is a c. The pump takes in the volume a 6 at atmospheric pressure ; it compresses it into an intermediate receiver, the compression line (adiabatic) is bf, passing into receiver, line/e. From the receiver it enters the cylinder at the constant pressure of compression on the line e/g, supply of heat cut off at g. Then expansion (adiabatic) to the point c atmospheric pressure. The part b/g c is the part avail- able for work, the part b/e a repreggnting the work of the compressing pump, which is deducted from the total motor cylinder diagram a e g c. The total volume of air passed through the pump is v„ volume swept by motor piston, w. So far as the heat operations are concerned, the part of the diagram to volume v^ may be disregarded ; it repre- sents the pressing of the compressed charge into the reservoir after reaching the maximum pressure of compression (it is called v^ because it is volume of compression). The admission to the motor cylinder is THERMODYNAMICS 71 identical, so that work done in pump in that part equaJs work done upon the motor piston. In addition to the letters used in type i, Vc is volume of compression. Vj, volume at point g on diagram. pc is pressure of compression. tc is temperature of compression. The temperature, volume, and pressure letters are figured below the diagram to make matters clear. Compression is carried on from volume v„ at atmospheric pressure and temperature to volume v^ at pressure p, and temperature 4, the curve being abiabatic. After compression, heat is added without allowing the pressure to increase, but the piston moves out till the maximum temperature T is attained, and the supply of heat being completely cut off, adiabatic expansion follows till the atmospheric pressure is reached ; the exhaust valve is then opened, and the hot gases discharged. It is evident that as the pressure is constant, while heat is being given, the amount of heat given to the engine in all is H = K, (T - 4), and the heat discharged from it is also at constant pressure. Hi = K^ (tI - t). The efficiency is therefore E Kf (T - Q T^ -t , , The compression and expansion curves being adiabatic. Compression p,vl = p jd/, Expansion p^ v/ = p„v' , •• - =-^%, but ^ = />, V/ p, V %otha±'"-i='"~I (a) and -' = ^^, also ?" = 4- Vf T I) T' Substituting in equation (a) t,_ t T T T t, and — = -. 72 THE GAS, PETROL, AND OIL ENGINE As the efficiency is E = I — T^ —t T-t' it may be either ■■ — or = I - - T t. (5) That is, when expansion is carried to the same pressure as existed before compression, the efficiency depends upon the compression alone, t being the temperature before compression, and t, the tem- -^ y 'S'^ '^ ^ ^ w ■^•^■<, B B a a a, s 3ajnssaj(j ><> viS'ta ^St^ t^ V <;) " ~ rt s W -SVj'^'S B B a E> a, H THERMODYNAMICS 73 perature of compression. The efficiency being i — -, the greater the temperature 4 the ]ess is the fraction -, and the more nearly does E approach unity. In most working engines of this kind, the expansion is not con- tinued long enough to make the pressure after expanding fall to atmosphere ; so that the efficiency is never so great, as when that is done a greater portion of the heat is discharged than need be. The modification of the formulae is precisely as in type i for similar circumstances. A diagram of the kind is shown at fig. 25. The temperature t^ is found as before : = 783°. The efficiency therefore T -t, ^ 1873 - 462 E = o-5i. With compression 100 lbs. above atmosphere, ^. = 524° go THE GAS, PETROL, AND OIL ENGINE and t' is therefore t^ = 290 ( — Z^ j imS"_ ^3-5° and E = i-i-4o8M5.=l^o 1873-524 E = 0-55. Taking, next, 1000° C. as the highest temperature, first with the lower compression, and after with the higher compression, with 60 lbs. compression t^ is 595° absolute with 100 „ t' is 545° E = o'47 at 60 lbs., E = 0'52 at 100 lbs., with 1000° C. In this case the efficiency varies both with the maximum tem- perature of the explosion and with the compression temperature previous to explosion. A glance at the numbers placed together will show clearly the relationship. Max. temps, in °C 1600° 1600° 1000° 1000° Pressure of compression above atmo- sphere 60 lbs. 100 lbs. 60 lbs. 100 lbs. Ef&ciency 0-51 0-55 0-47 0-52 2nd Case. — Here the expansion after explosion is not carried on far enough to reduce the pressure to atmosphere. It terminates when the volume is the same as existed before compression, that is, the volume swept by the motor piston when the air is expanding doing work is identical with that swept by the pump piston in compressing up to maximum pressure. Pump and motor are equal in volume. To this case of type 3 belong all compression engines in which the motor piston compresses its charge into a space at the end of the cylinder. In this case, as in case i, type 2, the theoretical efificiency of the engine is quite independent of the maximum temperature of the explosion. So long as the volume after expansion is the same as that before compression, it does not matter in the least how much heat is added at constant volume of compression ; whether only a few degrees' rise occurs or 1000° or 2000°, it is all the same so far as the proportion of added heat converted into work is concerned. That proportion depends solely upon the amount of cogipression. For 60 lbs. adiabatic compression, temperature 462° absolute, the efificiency is 0-37 ; for 100 lbs. above atmosphere it is 0-45. Given by the formula e = i — -. (See p. 80) E depends absolutely upon the temperature of the atmosphere and the temperature of compression t and t,. If the relative volumes of space swept by piston and compression space be known, then the efficiency can be at once calculated. THERMODYNAMICS 91 2,rd Case. — Here the expansion is carried further than the original volume before compression, but not far enough to reduce the pressure to atmosphere. Efficiency is always less than in the first case with corresponding temperature of explosion and compression, but greater than in the second case. It is found by the formula : E_i_ (T^-^^ )+y(^'-0 f depends on the relationship between the volumes v^ and v the volume at atmosphere and the volume of discharge after expansion. It is always : t' is also found by the same method as in types i and 2. It is better to postpone calculating any particular case of this at present, as no engine doing this has yet got into public use, and it can be considered further on in discussing the effect of increased expan- sion in the actual engines. Type I A. — The efficiency of this type of heat cycle depends to a considerable extent upon cooling during the return stroke ; in its best form, cooling at the lowest temperature during isothermal compression, it cannot be carried out without introducing the very disadvantages with which the hot-air engine was saddled — namely, a dependence upon the slow convection of air for the discharge of the heat necessarily rejected from the cycle. The rapid performance of this operation is impossible, and accordingly it is hardly fair to com- pare this type with those preceding ; they could all of thern be greatly improved in theory by introducing greater expansions and cooling by convection at the lowest temperature, but all at the expense of rate of working. The efficiency of type i a will be found to be high ; but it is to be kept constantly in mind that the penalty of slow rate of work was fully exacted in the practical examples of the kind in public use. They are exceedingly cumbrous, and give but a trifling power in comparison with their bulk and weight. The efficiency in this type is dependent upon T and t only. (y-l) :0-48. VOL. I. H 98 THE GAS, PETROL, AND OIL ENGINE Then p„ is 500 lbs. per sq. in. absolute, and f^ is 141-8 lbs. per sq. in. absolute. Maximum volume of cycle is the initial volume before compression. The mean effective pressure is 105 lbs. per sq. in., and ratio of maximum to mean effective ^ =4-6. 105 Constant -pressure and constant volume types at the same com- pression. Fig. 34 is a diagram combining constant pressure and constant volume types employing the same compression and the same maximum temperature. In both cases — t = 290° C. absolute t. = 559° C. „ T = 1973° C. t' = 1023° C. I _ I r ~ 5 and E =0-48 In i\ie constant pressure AiSigxdim the maximum pressure is 141 '8 lbs. absolute per sq. in., the mean effective pressure throughout the stroke is 37 lbs. per sq. in. The effective maximum is I4i'8 — 15 = i26'8 lbs. Ratio of maximum to mean effective - ^ = 3'8. The 37 maximum volume of the cycle is 3-5 times the initial volume at 290° C. absolute. In the constant volume diagram it has been already pointed out that the effective maximum is 485 lbs. per sq. in., the mean effective 105 lbs., and the ratio of maximum to mean 4-6, while the maximum volume is equal to the initial volume before compression. Here we have two diagrams of equal thermal efficiency, of which the constant pressure type requires an engine of 3-5 times the cylinder capacity of the constant volume, but where the ratio of maximum to mean is more favourable, 3'8 against 4 '6, so that for equal power the constant pressure engine would require to withstand smaller stresses than the constant volume. Practical considerations alone would determine here which cycle should be adopted. Such calculations cannot be made until we study the real as well as th^deal conditions assumed. Comparison of Results The three maximum temperatures used, 1700° C, 1600° C, and 1000° C, with the lowest temperature, 17° C, give, in a perfect heat engine, efficiencies 1700° C. = 0-853 1600° C. = 0-85 nearly, 1000° C. = 0-77 „ THERMODYNAMICS To compare easily the following table will be useful : Table of Theoretic Efficiencies calculated in the Examples Given 99 — Max. temp. °C abs Compression E Temp °C. abs. Ratio r Type I Expanding to atmosphere 1873° — — 0-29 ,, ., .... 1373° — — 0-23 Expanding to twice initial volume . 1873° 1273° — — 0'22 0-20 Type 2 Expanding to atmosphere . \ 462° 524° 31 I 0-37 0-45 Expanding to twice volume existing before compression . . . ' 1873° 1873° 1273° 462° 524° 462° 0-30 0-40 o'36 *■ 1273° 524° 0-44 Type 3 1873° 462° 0-51 Expanding to atmosphere 1873° 1273° 524° 462° 0-55 0-47 1273° 524° — 0-52 Expanding to the same volume as existed before compression . ) 462° 524° 3'i I 0-37 0-45 4'4 Type I A Expanding from maximum temperature . to lowest temperature . 1873° 1273° — ^ 0-66 0-56 Symmetrical cycles of constant constant volume. temperature, constant pressure, and r E 4 0-246 -3- 0-36 i 0-43 i 0-48 T 0'55 tV o-6i 2V • • ■ ■ 070 Tk 0-85 100 THE GAS, PETROL, AND OIL ENGINE Comparing these types it is evident that types 2 and 3, which employ compression previous to ignition or heating, are far superior to types I and i A, which employ no previous compression. It is not true, however, that the two latter are both non-compression, because i A involves not only great expansions, large volumes at low pressures, but also recompression after expansion, this recom- pression being effected isothermaUy at the lower temperature limit. This is an impracticable process for long expansions and consequently ~1 !— -T /, 73 °C \bs 1 i ) 1 11 A S is' C. / T- m i" 0.3 hs. ^ \ II \ ^i^ 1 i 1 1 >> V, ^ *< T ■IC 23 'C. all s. ftTM, 0- 1 ^ n % 1 i I t- ZS 0° ^.3 )s. — - _ r- Q •3 c. !b: ^ ' Compression pressure . Maximum pressure Mean pressure Pressure at end of expansion Efficiency .... VOLUMES. Constant pressure. 141 "8 lbs. per sq. in. abp. 141*8 lbs. per sq. in. abs. y] lbs. per sq. in. 14 "7 lbs. per sq. in. abs. 0-48 Constant volume. 141*8 lbs. per sq. in. abs. 500 lbs. per sq. in. abs. 105 lbs. per sq. in. 5 1 '8 lbs. per sq. in. abs. Fig. 34. — Diagrams of Air Engines Constant Pressure and Constant Volume Types, with the same compression and the same maximum temperature. correspondingly long compressions. High theoretical efficiencies, however, are attainable by the method. Type i has neither prior nor subsequent compression, and so gives only low thermal efficiencies, because only small expansions are possible. These two types may be dismissed from further consideration. The three symmetrical cycles of constant temperature, constant pressure, and constant volume are all perfect cycles for their assumed conditions and all three are theoretically capable of yielding thermal THERMODYNAMICS loi efficiencies which depend solely on the degree of adiabatic compression employed. Of the three, however, the first is impracticable because of the very low effective mean pressure obtainable compared with the very high relative maximum pressure. In the example calculated, the ratio is 8i to I, a quite impossible ratio for a practicable engine. It must always be borne in mind in reasoning on these thermodynamic cycles that the compression ratio which may be chosen is dependent on the maximum temperature available. Thus a compression ratio of j;}^ could not be adopted unless that maximum temperature were at least 1600° C, assuming the lower temperature as 17° C, because adiabatic compression from 100 to i volume would raise the temperature from 17° C. to 1600° C. In such a case the constant temperature cycle would be the only one available, as compression has already attained maxi- mum temperature and heat at that temperature can only be added isothermally. This consideration excludes such an extreme compression range as -ylij, that is it confines practicable engines to types 2 and 3 — that is, to constant pressure and constant volume cycles. In practice the ratio now adopted varies between J and ^V ^^ t)oth cases, so that their thermal efficiencies vary between 0'43 and 0'65. Quite favourable maximum and mean effective pressure ratios are obtained between these limits. In the example calculated for constant presstire conditions, as in fig. 32, it is 4'i4 to i, with an efiBciency of 0"56. The corresponding constant volume diagram, fig. 33, gives 4'6 to i with an efficiency of 0'48 ; here the constant pressure cycle is best from both points of view. In both cases the total volumes of working fluid used are the same, and the maximum volume is the same. Com- paring the cases shown on fig. 34 of constant pressure and constant volume, efficiencies both 0'48, the maximum to mean effective pressure ratio is 3"8 for constant pressure and 4.6 for constant volume, using identical volumes of working fluid, but the maximum volume required for the constant pressure engine is 3 '5 times that of the constant volume. It is entirely a question for practice whether it is better to have a smaller engine with high pressures or a larger engine with low pressures but parts strained relatively less. So far practice favours the smaller high-pressure engine, and with engines of small power the choice is justifiable. Whether this will be so in the case of large gas engines has not been yet settled. The Otto cycle engines made under type 3 far outnumber all others in present use, but high economies are obtained by the Diesel engine operating under type 2, although on the Otto mechanical cycle also. Undoubtedly these two types, constant volume and constant pressure, are of vital importance to-day. Throughout the present chapter the working fluid has been 102 THE GAS, PETROL, AND OIL ENGINE assumed to be dry air obeying perfectly the laws of Charles and Boyle ; its specific heat has also been assumed to be constant through- out the temperature range. Its specific heat at constant volume K^ has been taken as 0'i68, and specific heat at constant pressure K^ = 0-237 ^nd K„ o-i68 It is now known that the specific heat of air is not quite constant between o° and 1400° C. Holborn and Austin's recent determinations of specific heat at constant pressure by a calorimetric method and electrical heating show a rise, which, however, is not large. The mean K^, between 100° C. and 1400° C. is about 8 per cent, higher than that between 100° and 200° C. Also the working fluid of the gas engine, as has already been pointed out, is not air but a mixture of nitrogen, carbonic acid, steam, and oxygen, so that, apart from chemical questions of combustion, the physical properties of the real working fluid are different from the idealised air which has been assumed in the foregoing formulae and examples. Nevertheless the author considers it desirable to treat the problem of ideal efficiency in its most simple form, as many valuable lessons have been learned from the simplified conditions and many valuable deductions have been made in the past from calculations on the ' air standard ' first introduced by the author in 1882. But it must be remembered that the efficiencies and mean pressures determined by these calculations for ideal air are not the efficiencies and mean pressures which would be proper to the actual working fluid. The air standard supplies a relative but not an absolute standard, as will be discussed later on when the actual working fluid has been studied. Meantime, however, it may be taken that the reasoning and con- clusions reached in this chapter are valuable when properly used. Conclusions. — The best cycles for maximum efficiency and maximum power for given stresses are those of constant pressure and constant volume in which adequate compression forms an essential feature. By such compression we are enabled to subject our heated working fluid to any predetermine effective expansion. Without adiabatic compression, effective expansion is restricted within very narrow limits ; with compression our limits may be chosen with a view to a pre-arranged thermal efficiency. Compression previous to ignition does for the gas engine what the condenser does for the steam engine. It extends the range of effective expansion. It does more than that, but we must first discuss the laws of heat loss before dealing with effects other than purely thermo- dynamic. CHAPTER IV THE CAUSES OF LOSS IN GAS ENGINFS In calculating the efficiency of the different kinds of engines, it has been assumed that the conditions peculiar to each cycle have been perfectly complied with. In actual engines this is impossible ; it is therefore necessary to discover in what manner practice fails in performing the operations required by theory. The actual engines differ from the ideal ones in several ways : 1. The working fluid loses heat to the walls enclosing it after its temperature has been raised to the highest point ; 2. The working fluid often gains heat when entering the cylinder at a time when it should remain at the lowest temperature ; 3. The supply of heat is never added instantaneously as is required in some types ; 4. The working fluid is not the pure dry air assumed in Chap- ter III. ; it is a mixture of nitrogen, carbonic acid, steam, and oxygen, with a different specific heat from air, which specific heat increases considerably with rise in temperature ; 5. Combustion is not complete when the maximum temperature of the working fluid is attained, and in some cases it is not complete when the exhaust valve opens ; 6. Combustion changes the volume of the working fluid so that the volume which is heated is different from the volume which is compressed ; 7. The admission, transfer, and expulsion of the working fluid are not accomplished without some resistance, throttling during admission, back-pressure during exhaust ; 8. Loss of heat to the cj'linder walls during compression. The first cause of loss is by far the most considerable and will be considered first. Loss OF Heat to the Cylinder and Piston Although this is the most considerable source of loss in all gas engines, the stock of information in existence upon the subject is quite 104 THE GAS, PETROL, AND OIL ENGINE insufficient to justify any attempt to state a general law. Recent experiments have been made by the author, Hopkinson, and Petavel to determine the rate at which a mass of heated gases, at from 1000° to 1600° C, loses heat to the comparatively cool metal surfaces which enclose it. These experiments will be described later, meantime it is sufficient to say that the rate of flow is rapid. Otherwise, it would be impossible to raise steam with the relatively small heating surfaces generally used in boilers. Before applying the efficiency values obtained to actual practice it is necessary to know at what rate a cubic foot of gases at about 1600° C. in contact with metal walls at from 17° C. to 100° C. will lose heat ; also to know how that rate changes with change of temperature and density. Much is known of the laws of cooling at lower temperatures, but little positive data exist for temperatures so high as those occurring in the gas engine. A hot gas loses heat to the colder walls enclosing it mainly by circulation or convection. The conduc- tivity of gases for heat is very slight, and unless in some way a large surface of the gas is exposed to the cooling surface, practically no heat would escape from the working fluid in the short time during which it is exposed in gas engines. Any arrangement which favours or hastens convection will therefore increase loss by increasing the extent of hot gaseous surface exposed to the walls. The smaller the surface to which a given volume of working fluid is exposed the less heat wiU it lose in a given time. So far as loss of heat is concerned, then, the best type of engine is that which exposes a given volume of working fluid to the smallest surface in performing its cycle. Suppose that in the three types the pistons move at the same velocity, then that which requires to move through the smallest volume, the areas of the pistons being supposed equal, will take the shortest time to perform its cycle. In the first engine the piston moves through 27 vols., with the hot air filling the cylinder ; the second, through 37 vols. ; and the third, through 2'4 vols, (see figs. 22, 24, and 26). As the volumes are proportional to the time taken to perform each cycle the third type has the best of it, the time of exposure of the hot working fluid being the least ; the second type is worse than the first. There is still another circumstance in addition t§ surface exposed and time of exposure, that is, the average temperature of the hot gas which is exposed. If the average temperature is lower in one type than in another during exposure to a given surface for a certain time, then obviously less heat wiU be lost in the one than in the other. Com- paring the average temperatures it is found that in the first the tem- perature ranges from 1600° C. to 817° C. ; in the second from 1600° C. to 901° C. ; and in the third from 1600° C. to 510° C. The third will therefore show a lower average temperature than the others. Three CAUSES OF LOSS IN GAS ENGINES 105 conditions are requisite in the engine which is to lose the minimum of heat from its working fluid : 1. In performing its cycle it should expose a given volume of its working fluid to the least possible cooling surface ; 2. It should expose it for the shortest possible time ; 3. The average temperature during the time of exposure should be as low as possible — which conditions are best fulfilled by the third type. In addition to its advantage in theoretic efficiency it possesses the further good points in practice of proportionally small cooling surfaces, short time of exposure, and rapid depression of temperature due to work done, consequently small loss of heat to the cylinder and piston. The diagrams, figs. 22, 24, and 26, have been selected from the others belonging to each type because the pressures, temperatures, and relative volumes closely correspond with those which would be best and at the same time readily practicable. The flow of heat really occurring in the gas engine cylinder will be discussed when the actual diagrams come under consideration ; mean- time, it is sufficient to have proved that the third type will in practice give results more closely approaching its theory than the others. If in each case a constant proportion of the heat supplied were lost to the cylinder and piston, the ratio of the efficiencies would remain constant, and although it would be impossible from present data to predict the actual values, yet the relative values would be known. Gain of Heat by the working Fluid when entering the Engine In all types of gas engine it is found most economical to keep the motor cylinders as hot as possible ; they are generally worked at a temperature close upon the temperature of boiling water.^ This is done to diminish the loss of heat from the explosion. It follows that if the working fluid is introduced at a lower temperature it becomes heated. In the first type, the charge should be admitted and remain at the lowest temperature until the moment of explosion, which is of course impossible if the cylinder is at 100° C. As the piston itself is hotter than that, it may be supposed that the charge is heated to that point. Taking an extreme case and Ccdculating the effect of having an absolute temperature of 390° for the lower limit, it will be found that the efficiency is diminished. In case i, type i, where the expansion is ' In large gas engines the water jackets are kept as cold as possible. io6 THE GAS, PETROL, AND OIL ENGINE carried to atmosphere with a maximum temperature of 1873° absolute = 1600'' C, the value becomes reduced to 0'23. With a maximum temperature of 1273° absolute = 1000" C. the efficiency is 0"i6. Type I. Initial temp, of working fluid Max. temp. Efficiency 17° c. 117° c. 17° c. 117° c. 1600° C. 1600° C. ICK)0° C. 1000° c. 0-29 0-23 0-23 o-i6 Here heating while introducing the charge will always cause dimi- nution in efficiency, the proportion of loss being greater with the lower maximum temperature. At 1600° C. the loss is nearly one-fifth, while at 1000° C. it is close upon one-fourth. It is very difficult to say whether it is better to work with the cylinder hot or cold. The constructor finds himself in a dilemma ; if the cylinder is kept as cold as the surrounding, air, then the hot gases cool more rapidly. If he keeps the cylinder hot to diminish this, the efficiency falls also. Experiment alone can decide the question. In engines of type 2 it is a usual proceeding to leave the com- pression cylinder entirely without water- jacketing, under the impres- sion that heat is thereby saved ; the temperature consequently rises to very nearly that of compression, and the entering charge becomes considerably heated before compression. This is especially the case if the admission area is small, and throttling occurs ; all the energy of velocity of the entering gas becomes transformed into heat. As in the previous case the charge may be considered to rise to 117° C. before compression. Where expansion is carried to atmosphere it has been shown that the efficiency is quite independent of the maximum temperature, but is determined by one circumstance only — the amount of the compres- sion. As E = I — ~ and t is the temperatur*. absolute before compressing *c >) )) }) jj alter ,, 4 and as hm-- " follows that with a constant ratio between the pressures before and after compression, the ratio of temperature before and after compressing will also remain constant ; that is, the efficiency is not in any way affected by heating the working fluid, provided the same degree of compression is used. Increase of CAUSES OF LOSS IN GAS ENGINES 107 temperature previous to compression causes a proportional increase of temperature after compressing without in any way disturbing the ratio between them. This is an important if in appearance a somewhat paradoxical fact, and it may be stated in another way : If an engine receives all its supply of heat at one pressure, and rejects all its waste heat at another pressure, after falling from the higher to the lower pressure by expansion doing work, the efficiency is constant for all maximum temperatures of the working fluid. The proportion of heat converted into work is not changed in any way by increasing the temperature before compressing, and if only one degree of heat be added after compressing, the same proportion of that one degree is converted into work, as would be done with any addition of heat however great. Where the expansion is not continued enough to reduce the pressure after heating, to atmosphere, as in the cases of this type which occur in practice, this is not quite true ; the compression still remains the most powerful element of efficiency, but heating before compression produces some change, just as increase of temperature after compression pro- duces change. The change is not great, and it is always in the direction of improvement with a limited expansion. If the lower temperature t is increased, the compression temperature 4 increases in proportion, and is accordingly nearer the maximum temperature. The volume increases less on heating, so that the effect upon efficiency is the same as if the expansion had been increased ; the terminal pressure will more closelj' approach atmosphere, and therefore come nearer to the condition of maximum efficiency. In engines of type 3 the compression and expansion are often performed in the same cylinder. For this purpose it is necessary to leave at the end of the cylinder a space into which the charge is to be compressed. As the piston does not enter this space, a considerable volume of exhaust gases remains to mix with the fresh cold charge. Partly from this and partly from the heating effect of the cylinder and piston, the charge becomes considerably heated before compression. The temperature of 200° C. is not unusual. Here the simplest case is that where the expansion is continued to the same volume as existed before compression. The efficiency depends solely upon the amount of the compression ; for any given degree of compression it is constant, whether the addition of heat at constant volume after compression be great or small. The efficiency is E = i — - as in type 2 (see p. 80) ; and the two absolute temperatures vary in the same ratio, that is, if the charge is heated before compression, the temperature after com- pression will be increased in the same ratio. The two temperatures io8 THE GAS, PETROL, AND OIL ENGINE will therefore bear a constant ratio to each other, whatever the initial temperature may be, provided the compression is constant. Heating the charge before compression will consequently have no disturbing effect upon the theoretical efficiency.' Where the expansion is carried to atmosphere the case is different. The diagram (fig. 26) may be considered to be made up of two parts giving two different efficiencies, the sum of which in this case is o'5i. In expanding from the compression volume v, to the original volume v„ (compression 75 lbs. per sq. in.) the total efficiency is 0'37, and from that volume to v and atmospheric pressure, 0'i4. The latter portion still obeys the same law as in a similar case of type i ; so that if the initial temperature at volume v„ be supposed 117° C. it will lose efficiency in a similar way. The temperature 901° C. will still exist at that point of the expanding line, so that it may be taken as similar to the case calculated on p. 106, where 1000° C. is the maximum. The loss of efficiency there is from 0'23 to 0'i6 for an initial temperature of 117° C, which makes 0'i4 become nearly O'lo. The total efficiency would therefore be 0-47 instead of o'Si without previous heating. Efficiency diminishes with increased temperature of working fluid before compressing if the expansion is carried to atmosphere, but does not change where the expansion is limited to the initial volume. Other Causes of Loss The third, fourth, fifth, sixth, seventh and eighth causes of loss require for their examination a comparison of the actual diagrams, and a knowledge of the phenomena of explosion and combustion, and so cannot be discussed at this stage. ' It is here necessary to distinguish between theoretical and practical efficiency. Heating before compression diminishes efficiency in practice by increasing maximum temperature, and therefore loss of heat. CHAPTER V COMBUSTION AND EXPLOSION In the preceding chapters the gas engine has been considered simply as a heat engine using air as its working fluid ; it has been assumed tliat in the different cycles the engineer is able to give the supply of heat either instantaneously, or slowly, at will ; and also that he can command temperatures so high as 1000° C. or 1600° C. It is now necessary to study the properties of gaseous explosive mixtures in order to understand how far these assumptions are true. On True Explosive Mixtures When an inflammable gas is mixed with oxygen gas in certain pro- portions, the mixture is found to be explosive : a flame approached to even a smaU volume contained in a vessel open to the air will produce a sharp detonation. Variation of the proportions wiU cause change in the sharpness of the explosion. There is a point where the mixture is most explosive ; at that point the inflammable gas and the oxygen are present in the quantities requisite for complete combination. After explosion the vessel will contain the product or products of com- bustion only, no inflammable gas remaining unconsumed, or oxygen uncombined, both having quite disappeared in forming new chemical compounds. That mixture may be called the true explosive mixture. Definition. — When an inflammable gas is mixed with oxygen in the proportion required for the complete combination of both gases, the mixture formed is the true explosive mixture. If the chemical formula of an inflammable gas is known, the volume of oxygen necessary for the true explosive mixture can be at once calculated. Elementary substances combine chemically with each other in certain weights known as the atomic or combining weights : chemical symbols are always taken as representing those weights of the elements indicated. In dealing with inflammable gases used in the gas engine it is convenient to remember the following symbols and weights ; no THE GAS, PETROL, AND OIL ENGINE Element Symbol Combining weight Oxygen Hydrogen Nitrogen Carbon Sulphur o H N C s i6 I 14 12 32 In entering or leaving any compound the elements invariably enter or leave in weights proportional to those numbers or multiples of them. Thus hydrogen and oxygen combine with each other, forming water ; the formula of the compound is HjO, meaning that 18 parts by weight contain 16 parts of and 2 parts of H. Similarly when carbon combines with oxygen two compounds may be formed, according to the conditions, carbonic oxide or carbonic acid, formulae CO and CO2, the former containing in 28 parts by weight, 12 parts of carbon and 16 parts of oxygen ; the latter in 44 parts by weight con- taining 12 parts of carbon and 32 parts of oxygen. The formula of a compound therefore not only indicates its nature qualitatively, but it also indicates its quantitative composition. H2O not only tells the nature of water, but it represents 18 parts by weight ; CO means 28 parts by weight of carbonic oxide : CO2 means 44 parts by weight of carbonic acid. The numbers 18, 28, and 44 are known as the molecular weights of the three compounds in question. When dealing with gases it is more convenient to think in volumes than in weights. It is easier, for instance, to measure the proportions of explosive mixtures by volume and to say this mixture contains one cubic inch, one cubic foot, or one volume of inflammable gas to so many cubic inches, feet, or volumes of oxygen. Fortunately there exists a simple relationship between the volumes of elementary gases and their combining weights, and also between the volumes of compounds and their molecular weights. If equal volumes of the elementary gases are weighed, under similar conditions of temperature and pressure, it is found that their weights are proportional to the combining weights. Taking the weight of the hydrogen as i, then the weigh* of equal volumes of nitrogen and oxygen are 14 and 16 respectively. If then it is wished to make a mixture of hydrogen and oxygen gases in the proportion of 2 parts by weight of the former to 16 parts by weight of the latter, it is only necessary to take 2 vols. H and i vol. 0. The law may be stated in two ways, as follows : Taking hydrogen as unity the specific gravity of the elementary gases is the same as their combining weights ; or COMBUSTION AND EXPLOSION III The combining volumes of the elementary gases are equal. Instead of troubling to weigh out portions of the gases it is at once known that one volume of nitrogen weighs 14 parts, the same volume of hydrogen weighing i part, oxygen 16 parts, and so on through all the gaseous elements, under the same temperatures and pressures. Knowing that water is the compound formed by the combustion of hydrogen and oxygen, and that its formula is H2O, it is at once apparent that the true explosive mixture of these gases is 2 vols. H. and I vol. O. By experiment it is found that the volume of the water produced is less (of course in the gaseous state) than the volume of the mixed gases before combination. The measurement requires to be made at a temperature high enough to keep the steam formed in the gaseous state. Measure 2 vols. H and i vol. O into a strong glass vessel heated to 130° C. ; the total is 3 vols. ; fire by the electric spark over mercury. It will be found that the steam formed when it has cooled to 130° C. after the explo- sion measures 2 vols. It has been found to be true for all gaseous compounds, that however many volumes of elementary gases combine to form them the product is always two volumes. In elementary gases, one volume always contains the combining weight ; in com- pound gases, two volumes always contain the molecular weight. Compared with hydrogen, therefore, the specific gravity of a gaseous compound is always one-half of the molecular weight. As before, the law may be stated in two ways : Taking hydrogen as unity, the specific gravity of a compound gas is half its molecular weight ; or The combining volume of a compound gas is always equal to double that of an elementary gas. These laws are known as Gay-Lussac's laws, and form part of the very basis of modern chemistry. Using them, the true explosive mixtures by volume and the volumes of the products of the combination can be found for any gas or mixture of gases, whether elementary or compound. The inflammable compound gases, used in the gas engine, forming some of the constituents of coal gas are : Inflammable gas Formula Molecular weight Molecular vol. Marsh gas Ethylene Carbonic oxide .... CH^ CO 16 28 28 2 2 2 Applying Gay-Lussac's laws, the oxygen required fpr true explosive H,0 Steam. . 2 vols. C0= Carbonic acid. . 4 vols. 2 vols. . 4 vols. 4 vols. — 2 vols. . 8 vols. 8 vols. 112 THE GAS, PETROL, AND OIL ENGINE mixtures and the volumes of the products of combustion are as follows for all the inflammable gases used in the gas engine ; 2 vols, hydrogen (H) require i vol. oxygen (O) forming 2 vols, marsh gas (CHJ require 4 vols, oxygen (O) forming 2 vols, ethylene (C2H,,) require 6 vols, oxygen (O) forming . 2 vols, carbonic oxide (CO) require i vol. oxygen (O) forming 2 vols, tetrylene (C,H(,) require 12 vols, oxygen (O) forming With hydrogen and oxygen 3 volumes before combination become 2 volumes after combination. CH4 and 0, also C2H4 and O, the volumes of the products of combustion, are equal to the volumes of mixture. With carbonic oxide and oxygen 3 volumes before become 2 volumes after combination. On Inflammability Previous to 1817, Sir Humphry Davy made the admirable researches which led him to the' invention of the safety lamp. He then made experiments upon different explosive mixtures, and found that under certain conditions they lost the capability of ignition by the electric spark. True explosive mixtures, he observed, may lose inflammability in two ways : by the addition of excess of either of the gases or of any inert gas such as nitrogen, and by rarefaction. The hydrogen explosive mixture, if reduced to one-eighteenth of ordinary atmospheric pressure, cannot be inflamed by the spark. Heated to dull redness at this pressure it will recover its inflammability and the spark will cause combination. One volume of the mixture to which has been added nine volumes of oxygen is uninflammable, but if the density is increased or. the termperature raised, it recovers its inflammability. Eight volumes of hydrogen added, produces the same effect as the nine volumes of oxygen, but only one volume of marsh gas or half a volume of ethylene is required. The excess which destroys inflam- mability varies with the temperature, increasing with increase of temperature. Heating the mixture widens the range, both of dilution with excess or inert gas and reduction of pressure. The point where inflammability cftises by diluting is very abrupt and sharply defined. The author has found that a coal gas which will inflame by the spark in a mixture of i gas and 14 air wUl not inflame with 15 of air. If the experiment be repeated on a warmer day, it may inflame with 15 of air but will not with 16 air. As the proportion is fixed for any given temperature it will be convenient to call that pro- portion for any mixture the ' critical proportion.' Any mixture in the critical proportion becomes inflammable by a very small increase of COMBUSTION AND EXPLOSION 113 temperature or pressure. The exact limits of dilution, temperature, and pressure have yet to be discovered. Passing from any true explosive mixture by dilution to the mixture in the critical proportion, the inflammability slowly diminishes, the explosion becoming less and less violent, till at last no report whatever is produced, and the progress of the flame (if a glass tube be used) is easily followed by the eye. In his great work on gas analysis Professor Bunsen confirms Davy's observations in every particular, proving loss of inflammability by dilution and reduction of pressure as well as its restoration by heating, increase of pressure and slight addition of the inflammable gas. His work, however, was not published till 1857. On the Rate of Flame-Propagation The sharp explosion of a true explosive mixture is due to the very rapid rate at which a flame, initiated at one point, travels through the entire mass and thereby causes the maximum pressure to be rapidly attained. With a diluted mixture the flame travels more slowly. Dilution therefore diminishes explosiveness in two ways — by increasing the time of getting the highest pressure and also by diminishing the highest pressure which can be got. Professor Bunsen's experiments are the earliest attempts to measure the velocity of flame movement in explosive mixtures. His method is as follows : The explosive mixture is allowed to burn from a fine orifice of known diameter, and the rate of the current of the issuing gas carefully regu- lated by diminishing the pressure to the point at which the flame passes back through the orifice and inflames the explosive mixture below it. This passing back of the flame occurs when the velocity with which the gaseous mixtures issue from the orifice is inappreciably less than the velocity with which the inflammation of the upper layers of burning gas is propagated to the lower and unignited layers. Knowing then the volume of mixture passing through the orifice and its diameter, the rate of flow at the moment of back ignition is known. It is identical with the rate of flame-propagation through the mixture. Bunsen made determinations for the true explosive mixtures of hydrogen and carbonic oxide. Velocity of Flame in true Explosive Mixtures. {Bunsen) Hydrogen mixture (2 vols. H and i vol. O) . -34 metres per sec. Carbonic oxide mixture (i vol. CO and i vol. O). i metre per sec. nearly. The method is a singularly simple and beautiful one, and answered thoroughly for Professor Bunsen's purpose at the time he devised it. Several objections, however, may be brought against it. The mixture in issuing from the jet into the air as flame, becomes mixed to some extent VOL. I. I 114 THE GAS, PETROL, AND OIL ENGINE with the air and so cools down ; the metal plate also, pierced with the orifice, exercises a great cooling effect. If the hole were made small enough the flame could not pass back at all, however much the flow is reduced, because the heat would be conducted away so rapidly as to extinguish the flame. This had been shown by Davy in 1817 ; indeed it is the principle of the safety lamp. These causes probably make Bunsen's velocities too low. MM. Mallard and Le Chatelier have made velocity determinations by a method designed to obviate those sources of error. The explosive mixture is contained in a long tube of considerable diameter, closed at one end, open to the atmosphere at the other. At each end a short rubber tube terminates in a cylindrical space closed by a flexible diaphragm. A light style is fixed upon the dia- phragms. A drum revolves close to each style, both drums upon the same shaft. A tuning-fork, vibrating while the experiment is being made, traces a sinuous line upon the drum and so the rate of revolution is known. The mixture is ignited at the open end, and the flame in passing the lateral opening leading to the first diaphragm ignites the mixture there, and so moves the style and marks the drum ; the arrival of the flame is signalled at the other end in the same way. The drums revolving together, the distance between the two style-markings measured by the vibration marks of the tuning-fork gives the time taken by the flame to move between the two points. The numbers got in this way are the rates of the communication of the flame through the mixture, back into the tube, while the flame can freely expand to the air ; when both ends are closed the velocity is much greater. Then not only does the flame spread from particle to particle of the explosive mixture at the rate due to contact of the inflamed particles with the uninflamed ones, but the expansion produced by the inflam- mation projects the flame mechanically into the other part and so produces an ignition, which does not travel at a uniform rate, but at a continually accelerating one. In the same way, using the open tube but firing at the closed end, the expansion of the first portion adds to the apparent velocity of propagation, and projects the last portion of the mixture into the atmosphere. The true velocity of the propaga- tion is the rate at which the flame pmceeds from particle of inflamed mixture to uninflamed particle by simple contact ; the true velocity depends upon inflammability alone, the rate under other conditions depends also upon heat evolved, and therefore movement due to expansion, mechanical disturbance of the unignited by the projection of the ignited portion into its midst. These conditions may vary much ; the inflammability remains constant. Mallard and Le Chatelier's results for the true velocity of propaga- tions are : COMBUSTION AND EXPLOSION 115 Velocity of Flame in true Explosive Mixtures (Mallard and Le Chatelier) Per sec. 20 metres 2 "2 ,, Hydrogen mixture (2 vols. H and i vol. O) . . Carbonic oxide (2 vols. CO and i vol. O) Bunsen's rate for hydrogen mixture seems to have been too great, and for carbonic oxide mixture too little. The rate for a true and very explosive mixture such as hydrogen is liable to be inaccurately determined, as temperature variation makes a great change, and it is difficult even with Mallard and Le Chatelier's method to obtain con- cordant experiments. With less inflammable mixtures the difficulty disappears. As true explosive mixtures are never used in the gas engine, their properties concern the engineer only as a preliminary to study of diluted mixtures. The most explosive mixture which can be made with air contains a large volume of nitrogen inevitably present as diluent. The following are some of their results with diluted mixtures, which are stated to be correct within 10 per cent, error of experiment : Velocity of Flame in Diluted Mixtures. (I\/allard and Le Chatelier) Per sec. 1 7 '3 metres 10 . 18 II-9 8-1 ' vol. hydrogen mixture + \ vol. oxygen . + 1 vol. oxygen . ,, + J vol. hydrogen + 1 vol. hydrogen + 2 vols, hydrogen These rates show that the true explosive mixture of hydrogen and oxygen when diluted with its own volume of oxygen falls from 20 metres per second to 10 metres, that is, it becomes one-half as inflammable ; when its own volume of hydrogen is the diluent, the velocity only falls to ii'g metres per second. Hydrogen therefore has less effect in diminishing inflammability than oxygen. Remembering the fact that the atmosphere contains one-fifth of its volume of oxygen, the remaining four-fifths being nearly all nitrogen, it is easy to get the proportions for the strongest explosive mixture possible with air. Two volumes hydrogen require i volume oxygen, and therefore 5 volumes air. The strongest possible mixture with air is two-sevenths hydrogen, five-sevenths air. The following experiments are for hydrogen and air in different proportions : Velocity of Flame in Diluted Mixtures. Mixture, vol. H and 4 vols, air H and 3 vols, air H and 2\ vols, air H and if vols, air H and i^ vols, air H and 1 vol. air H and J vol. air (Mallard and Le Chatelier) Per sec. 2 metres 2-8 „ 3 '4 .. 4'i .. 4'4 .. 3-8 .. 2-.3 M I 2 ii6 THE GAS, PETROL, AND OIL ENGINE Very strangely the velocity is greatest when there is an excess of hydrogen present. To get just enough of oxygen for complete burning, I volume H requires 2| volumes air, which would be naturally supposed to be the most inflammable mixture, as it gives out the greatest heat, but for some reason it is not. When the hydrogen is increased beyond i volume H to i^ volume air the velocity again falls off. A determination for coal gas and air gave i volume gas, 5 volumes air a velocity of I'oi metre per second, and i volume gas, 6 volumes air o"285 metre per second. With coal gas also the maximum velocity is got with the gas slightly in excess. So far, these rates of ignition or inflammation are measures of inflammability, and are the rates for constant pressure ; the rates for constant volume are very different, and the problem is a more complex one. Inflaming at the closed end of the tube, they found that even very dilute mixtures gave a sharp explosion, and in the case of hydrogen true explosive mixture, the velocity became looo metres per second instead of 20. With hydrogen and air 300 metres per second were obtained. MM. Berthelot and Vieille have proved that under certain con- ditions even greater velocities than these are possible. The con- ditions, however, are abnormal, and the generation of M. Berthelot's explosive wave is exceedingly undesirable in a gas engine. It is generated by inflaming a considerable portion of the mixture at once, and so causing the transmission of a shock from molecule to molecule of the uninflamed mixture : this shock causes an ignition velocity nearly as rapid as the actual mean velocity of movement of the gaseous molecules at the high temperatures of combustion. The difference between this almost instantaneous detonation and the ordinary flame-propagation may be compared with similar differences in the explosion of gun-cotton discovered by Sir Frederic Abel. Gun- cotton lying loosely, and open to the air, will burn harmlessly if ignited by a flame ; indeed, a considerable portion may be laid upon the open hand and ignited by a flame without the smallest danger. The same quantity in the same position, if fired by a percussive detonator, will occasion the most violent explosion, the nature of the shock given to the gun-cotton by the detonator causing a transmission of the kind of vibration necessary to cause its almost instantaneous resolution into its component gases. The explosive wave in gases seems to originate in like conditions. Its velocity for the true explosive mixture of hydrogen and oxygen is 2841 metres per second, and for carbonic oxide mixture 1089 metres per second. The velocity is independent of pressure between half an atmosphere and one and a half atmosphere. It is independent, too, of the diameter of the tube used, within considerable limits, or of the COMBUSTION AND EXPLOSION 117 material of the tube, rubber and lead tubes giving similar results. Diluting the mixtures diminishes, and heating increases it. The experiments are very interesting and important, from a physicist's standpoint, but, fortunately for the inventor dealing with gas engines, the explosive wave is not easily generated in a gas engine cylinder ; if it were, it would be impossible to run the engines without shock and hammering. The velocity which really concerns the engineer is that due to inflammability, and expansion produced by inflaming — the velocity, in fact, with which the inflammation spreads through a closed vessel. As it cannot be discussed without considering other matters — heat evolved by combustion, and temperatures and pressures produced — it will be advisable first to give the heat evolved by combustion, and then devote a complete chapter to explosion in a closed vessel. Heat evolved by Combustion Careful experiments upon the heat evolved by the combustion of gases in oxgyen have been made by Favre and Silberman, and also by Professor Andrews. The physicists first named burned the gases at constant pressure in a specially devised calorimeter. Professor Andrews mixed the gases in a thin spherical copper vessel, closed it, and exploded by the spark : the vessel being surrounded by water gave up its heat to the water, the weight of which being known, the rise of temperature gave the heat evolved. Quantities of heat are measured by taking water as the unit. In this work, unless otherwise stated, a heat unit always means the amount of heat necessary to raise unit weight of water through 1° C. Taking an average of Favre and Silberman and Andrews's results, the inflammable gases used in gas engines evolve upon complete com- bustion the following amounts of heat : Heat units Unit weight of hydrogen completely burned to H2O evolves . . 34,170 Unit weight of carbon completely burned to CO2 evolves . . . 8,000 Unit weight of carbonic oxide completely burned to CO^ evolves . 2,400 Unit weight of marsh gas completely burned to CO2 and H2O evolves 1 3,080 Unit weight of ethylene completely burned to CO2 and H„0 evolves" 1 1,900 That is, one pound weight of hydrogen burned completely to water will evolve as much heat as would raise 34,170 lb. of water through 1° C, or the converse. One pound of carbon in burning to carbonic acid evolves as much heat as would raise 8000 lb. of water through 1° C. These numbers give the amount or quantity of heat evolved. The intensity or temperature of the combustion may be calculated on the assumption that the whole heat is evolved under such conditions that no heat is lost, or is applied to anything else but the products of ii8 THE GAS, PETROL, AND OIL ENGINE combustion. To make the calculation it is necessary to know the specific heat of the products. The amount of heat required to heat unit weight of water through one degree is i heat unit, the specific heat of any other body is the number of heat units required to heat unit weight of the body through one degree. Gases have two different specific heats depending upon whether heat is applied while the gas is kept at constant volume, or at constant pressure ; both are required in dealing with gas engine problems. The specific heat at constant volume is sometimes known as the true specific heat ; in taking the specific heat at constant pressure the gas necessarily expands, and so does work on the external air ; this specific heat is therefore greater than the former by the amount of work done. For the gases used in the gas engine the two values are as follows. The ratio between the two is also given, as it is frequently required in efficiency calculations. The experimental numbers are Regnault's, the calculated specific heat at constant volume, Clausius. Specific Heats of Gases to 200° C. RegnauU (For equal weights. Water = i) Sp. heat at Sp. heat at Sp. heat con. pres. constant pressure constant volume Sp. heat con. vol. Air ... . 0-237 0-168 1-408 Oxygen 0-217 0-I5S 1-403 Nitrogen . 0-244 0-173 1-409 Hydrogen 3-409 2-406 I -417 Marsh gas . 0-593 0-467 — Etliylene . 0-404 0-332 1-144 Carbonic oxide . 0-245 0-173 1-416 Steam. 0-480 0-369 1-302 Carbonic acid . o-2i6 0-171 1-165 It is convenient to remember that the specific heats of combining or atomic weights of the elements are equal — Dulong and Petit's law. To this law there are few exceptions, and the permanent elementary gases, oxygen, nitrogen, and hydrogeiwobey it cdmost absolutely. As equal volumes of these gases represent me combining weights, it follows that equal volumes of these gases have the same specific heat. Taking the specific heat of air as the unit, the specific heat of hydrogen and oxygen gases is also unity. The compound gases do not obey the law so closely. Regnault's experimental method determined only the specific heat at constant pressure K^, and_ the specific heatat constant volume k„ is calculated from these numbers. COMBUSTION AND EXPLOSION iig Regnault also studied the effect of changing temperature on the specific heat of air and carbonic acid, and he concluded that within his experimental temperature range the specific heat of air remained constant, but that of carbonic acid increased materially. This is shown by the following table : Specific Heat at Constant Pressure with Varying Temperatures FOR Air and Carbonic Acid Regnault Air Carbonic Acid Temperature, °C. Sp. heat Kp Temperature, ° C. Sp. heat K/ From — 31 to + 10 to + 100 to + 300 0-2377 0-2374 0-2375 From — 30 to + 10 + 10 to + 100 + 10 to + 210 0-1843 0-2025 0-2169 Air obviously remains constant from — 31° to + 200° C, only the fourth decimal changes ; carbonic acid, however, changes from 0"i84 to 0'2i7, taking the nearest third decimal by nearly 18 per cent, in 200° C. Dr. Joly was the first to determine by a direct calorimetric method, his well-known steam calorimeter, the specific heat of gases at constant volume Kj, ; he found that the specific heat varied to some extent with change of density. In both air and carbonic acid he found increase of specific heat K„ with increased density, but with hydrogen specific heat appeared to decrease with increase of density. For air the specific heat at constant volume at a mean pressure of IQ'SI atmo- spheres and a mean density of 0'0205 was found to be 0'i72i. Carbon dioxide gave the following results : Specific Heat at Constant Volume with varying Pressures Joly Pressure in atmospheres Density K^ 7-20 ! O-OI1530 12-20 0-019950 16-87 1 0-028498 20-90 0-036529 21-66 , 0-037802 0-1684 0-1705 0-1714 0-1730 0-1738 The density in these figures is really specific gravity, water = i. Similar experiments with hydrogen proved that its mean specific heat at constant volume K„ was 2'402. 120 THE GAS, PETROL, AND OIL ENGINE Comparing these numbers found experimentally with the K„ values calculated from Regnault's K^ results at pressure over atmo- sphere, it is seen that for air the value k„ at atmospheric pressure is 0'i68, but at a mean pressure of ig'S atmospheres it becomes o'lya ; it increases by 2 "3 per cent. For carbonic acid Regnault's number K„ = O'lyz, Joly's maximum value is 0"i738 ; it is higher by 17 per cent. ; but Joly's numbers were taken at the lower temperature of 100° C, so that they agree sub- stantially with Regnault's values as given at p. 118. It is to be noted, however, that Joly's table shows that an increase from 7*2 to 2i'6 atmospheres increases K^ from 0"i68 to 0"i74, or 3'6 per cent. Joly's value K„ for hydrogen is 2 '402, which is in practical agreement with Regnault's 2 '406. Experiments were made by Mallard and Le Chatelier and by Berthelot upon gaseous explosions in a closed vessel at atmospheric initial pressure, from which they deduced values of K„ for temperatures up to and over 2000° C. for air, steam, carbonic acid and other gases, and these values indicated a large increase of specific heat for all the gases ; the present writer, however, with many others found it im- possible to accept their determinations, as he felt that no proof had been advanced that combustion had ceased before the measurements were made. He felt that no reliable determinations could be. made without avoiding combustion altogether as the means of heating the gases where specific heat was to be found. Messrs. Holborn and Austin have made a series of valuable experi- ments at the Reichsanstalt, Berlin, and at the British Association meeting of 1907 they gave the following tables for steam, carbonic acid, and nitrogen at temperatures from 110° C. to 1400° C. The experi- ments were made at constant atmospheric pressure like Regnault's, the gases were heated electrically, the temperature measured by a thermo- couple, and the heat quantity by a calorimeter. Mean Specific Heats at Constant Pressure with varying Temperatures FROM 110° TO 1400° C. FOR StEAM, CaRBONIC AcID, AND NiTROGEN Holborn and Austin Temperature, ° C. Steam H,0 no to 280 0-465 no to 400 0-467 no to 600 0"473 no to 800 0-482 no to 1000 0-494 no to 1200 0-510 no to 1400 0-532 • Temperature, " C. Carbonic acid CO. Nitrogen N no to 200 0-217 0-240 no to 400 no to 600 no to 800 0-229 0-240 0-250 0-242 0-247 0-251 iiQ to 1000 0-258 0-254 no to 1200 0-264 0-258 ' no to 1400 0-272 0-262 COMBUSTION AND EXPLOSION 121 From these experiments it is proved that the mean specific heat of steam at constant pressure from iio° to 1400° is greater than that between 110° and 280° C. by nearly I4'5 per cent. ; while with carbonic acid between 110° and 1400° C. and 110° and 200° the increase is 25'3 per cent, nearly. With nitrogen for the same temperature ranges as carbonic acid the increase is much smaller — it is only about 9 per cent. The value of y for steam at mean k^ = 0^532 is i'26, so that for that temperature K^, _- 0'532 _ I'26 0-422. For carbonic acid having mean K^ to 1400° = 0^272 the value of •)/ = i"2i and _ 0'272 _ I-2I 0-225. Comparing with Regnault's tables these values show very sub- stantial increases for H^O and CO2, and a small but appreciable increase for N. Earlier experiments of Holborn and Austin, using the same method, gave the corresponding values for air, nitrogen, and oxygen as follows : Mean Specific Heat at Constant Pressure with varying Tempera- tures FROM 20° TO 800° C for Air, Nitrogen, and Oxygen Holborn and A ustin Temperature, ° C. Air Nitrogen Oxygen 0-2240 0-2300 20 to 440 20 to 630 20 to 800 0-2366 0-2429 0-2430 0-2419 0-2464 0-2497 These values agree substantially with the later determinations. Holborn and Austin have calculated values of c^ at temperatures up to 800° as follows for CO 2, and they show corresponding values calculated from the Mallard and Le Chatelier experiments, as also from the later experiments of Langen. C, AT T° C FOR CO, T°C Hulborn and Austin C], at T° Langen Cj> at T° 0-2028 0-1980 100 o-2i6i 0-2100 200 0-2285 0-2220 400 0-2502 0-2450 600 0-2678 0-2690 800 0-2815 0-2920 Mallard and Le Chalelier C/atl° 0-1880 0-2140 0-2390 0-2840 0-3230 0-3550 122 THE GAS, PETROL, AND OIL ENGINE Langen's experiments were made by the explosion pressure method used by Mallard and Le Chatelier, but a very large volume of gaseous mixture was employed, and the means of measuring the rise and fall of pressure was much improved compared with the early apparatus. Mallard and Le Chatelier's value for 800° C. is nearly 26 per cent, above that of Holborn and Austin, while Langen's figure is only about 3"5 per cent, higher. Langen's numbers closely approximate to Holborn and Austin's, while Le ChateUer's depart considerably on both sides. The calculation of temperature of combustion can now be made. The amount of heat evolved from unit weight of a combustible is usually said to measure its calorific power, that amount divided by the specific heat of the products of the combustion is said to be the measure of its calorific intensity. The calorific intensity is indeed the theoretical temperature of the combustion : taking hydrogen first, unit weight evolves 34,170 heat units. But the water formed weighs 9 units (from formula H2O), and if its specific heat in the gaseous state were unity, the Supposed maximum temperature of combustion would be 44>_7. _ 27^5-6. But the specific heat is less than unity ; therefore the theoretical maximum will be greater. It is —i' 2_^- = 7Q0Q7. For certain reasons to be considered 9 X 0-480 later, no such enormous temperatures are ever attained by com- bustion. In the above calculation the latent heat of steam should first have been deducted, as it is included in the total heat evolved as measured by the calorimeter: it is 537 heat units. 34)170—537 gives the total heat available for increasing the temperature ; the amended calculation is ^^' ' ~„ - = 778q'4, still an exceedinErly 9 X 0-480 "-'-<' t> J' high temperature. Here we have taken Regnault's values as given on p. 118 for low temperatures. It is obvious that the real specific heat of steam through the longer temperature range must be greater. Assume Cp to be 0'532, the highest of Holborn and Austin's values, see p. 120, then the calculation becomes -^^' ' ~ ( i^^ = 7024'4. This is still a very 9 X 0-532 ^ '^ t y high temperature which has not been realised, so far as we know, by combustion or by any other method. Calculating the heat evolved by burning carbon in the same way, but omitting any deduction for the latent heat of carbonic acid (it does not affect the calorimeter, as it does not condense), the theoretical temperature produced by burning in oxygen is still higher, being 10,174° C. Burning in air the theoretical temperatures are lower, as COMBUSTION AND EXPLOSION - 123 the nitrogen present acts as a diluent, and must necessarily be heated to the same temperature as the products of the combustion. They are given as follows in ' Watts' Dictionary ' : Temperature produced °C. Calorific power ^ '^ ^ In oxygen In air Carbon 8,080 10,174 2,710 Hydrogen 34,462 6,930 2,741 These are the supposed temperatures burning in the open atmo- sphere, and therefore at constant pressure, the gases expanding doing work upon the air. At constant volume, that is, burning in a closed vessel so that the volume cannot increase but only the pressure, the temperature should be greater as the specific heat at constant volume is less. Allowing for that, the numbers become Theoretical Tempekatures of Combustion at Constant Volume Temperature produced ° C. In oxygen In air Carbon 12,820 — Hydrogen . . ... 9,010 4,119 Using the values given on p. 118 from Regnault and the thermal values at p. 117, we get 12 parts by weight of carbon burning in oxygen to COi evolves 12 x 8000 = 96,000 Centigrade heat units : this pro- duces 44 parts by weight of CO2, having a c^ value of 0'2i6. 12 X 8000 44 X 0.216 = 10,101° C. Taking Holborn and Austin's highest value of 0'272 for c^ of CO2, , 12 X 8000 o o r- we get = 8021 C. 44 X 0-272 For combustion of carbon in air, using Holborn and Austin's values, we get temperature of combustion in air 2321° C. For hydrogen in air, with the same number, we get 2776° C. Such temperatures as 4000° C. have never been produced by com- bustion, for many reasons, of which all save the most potent have been discussed by the earlier writers on heat. This is Dissociation. Dissociation Most chemical combinations, while in the act of formation from their constituent elements, evolve heat, and, as a general rule, the greater the heat evolved the more stable is the compound formed. The compound after formation may generally be decomposed by heating to a high enough temperature, heat being one of the most powerful splitting-up agencies known to the chemist. The nature of 124 THE GAS, PETROL, AND OIL ENGINE the decomposition varies with the compound. In many cases the process is irreversible, that is, although heating up will cause decom- position, cooling down again, however slowly, will not cause recom- bination. In some compounds, however, under certain conditions the process is reversible, and recombination occurs on slow cooling. Definition. — Dissociation may be defined as a chemical decom- position by the agency of heat, occurring under such conditions that upon lowering the temperature the constituents recombine. Groves found long ago that water begins to split up into oxygen and hydrogen gases at temperatures low compared with that produced by combustion. Deville made a careful study of the phenomena, and found that decomposition commences at 960° to 1000° C. and proceeds to a limited extent : raising the temperature to 1200° C. increases it, but a limit is reached. The amount of decomposition depending upon the temperature, for each temperature there is a certain proportion between the amount of steam and the amount of free oxygen and hydrogen gases present. If the temperature is increased, the propor- tion of free gases also increases : if temperature is diminished, the pro- portion of free gases diminishes. If the temperature be raised beyond a certain intensity, the water is completely decomposed : if lowered beyond a certain temperature, complete combination results. The same thing happens with carbonic acid, the temperature of -decom- position is lower. It is quite evident, then, that at the highest temperatures produced by combustion, the product cannot exist in the state of complete combination. It will be mixed to a certain extent with the free constituents which cannot combine further until the temperature falls ; cis the temperature falls, combustion will continue till all the free gases are combined. The subject, from its nature, is a difficult one in experiment, and accordingly different observers do not agree upon temperatures and percentages of dissociation, but all are agreed that dissociation places a rigid barrier in the way of combustion at high temperatures, and prevents the attainment of temperatures, by combustion, which are otherwise quite possible. With no dissociation, hydrogen burning in oxygen should be able under favourable circum- stances to give a temperature of o^r 6000° C, as has been shown. Deville's experiments upon the temperature of the oxyhydrogen flame, at constant pressure of the atmosphere, gave under 2500° C. The estimate was made by melting platinum in a lime crucible, with the oxyhydrogen flame playing upon the platinum, the crucible being well protected against loss of heat by lime blocks, so that the platinum could really attain the temperature of the flame ; when at the highest temperature, the molten platinum was rapidly poured into a weighed calorimeter, and the rise in temperature noted. From this was COMBUSTION AND EXPLOSION 125 calculated the temperature of the platinum. The experiment was dangerous and inaccurate, but it is the only serious attempt which has been made to determine the temperature of the oxyhydrogen flame at constant pressure. The highest temperature produced by hydrogen burning in oxygen has been determined by Bunsen, and also Mallard and Le Chatelier, for combustion at constant volume, that is, explosion. As the theoretic calculation shows, with no dissociation a tempera- ture of 9000° C. is possible. The highest maximum it is possible to assume from Bunsen's experiments is 3800" C. ; from Mallard and Le Chatelier's, 3300° C. The two sets of experiments are concordant. It is true the latter physicists do not attribute the difference wholly to dissociation, but they agree that part is due to this cause ; and that there is an enormous difference between the temperature actually got and that which should be possible if no limit existed, all are agreed. With air, Bunsen's figures show a maximum of about 2000° C, Mallard and Le Chatelier say 1830° C. ; the present writer has also made experiments with hydrogen in air, and finds the highest possible tem- perature to be 1900° C. The calculated maximum is 4119° C. The difference is not so great as with the true explosive mixture, which is to be expected, but all experiments agree in provingTthat there is a considerable difference. 126 THE GAS, PETROL, AND OIL ENGINE CHAPTER VI EXPLOSION AND COOLING IN A CLOSED VESSEL — EXPERIMENTAL INVESTIGATIONS The value of any inflammable gas for the production of power by explosion can be determined apart altogether from theoretical con- siderations by direct experiment. It is evident that the gas which for a given volume causes the greatest increase in pressure will give the greatest power for every cubic foot used, provided that the pressure does not fall so suddenly that it is gone before it can be utilised by the piston. Two qualities will be possessed by the best explosive mixture : (i) greatest pressure per unit volume of gas ; (2) longest time of maximum pressure when exposed to cooling. In the gas engine itself the conditions are so complex that the problem is best studied in the first instance under simplified con- ditions. The author has made a set of experiments upon many samples of coal gas mixed with air in varying proportions, to find the pressures produced, and the duration of those pressures ; igniting mixtures at atmospheric pressures and temperature, and also at higher temperature and initial pressures. He has made some experiments upon pure hydrogen and air mixtures in the same apparatus for comparison. The experimental apparatus used by him in the years 1884 and 1885 is shown at fig. 35. It consists of a closed cylindrical vessel 7 ins. diameter and 8 J ins. long, internal measurement, and therefore of 317 cubic ins. capacity. It is truly bored, and the end-covers turned so that the internal surface is similar to that of an engine cylinder ; the covers are bolted strjpgly so as to withstand high pressures. Upon the upper cover is placed a Richards indicator, in which the reciprocating drum has been replaced by a revolving one ; the rate of revolution is adjusted by a small fan, a weight and gear giving the power. The cylinder is fiUed with the explosive mixture to be tested ; the drum is set revolving, the pencil of the indicator pressed gently against it, and the electric spark is passed between the points placed at the bottom of the space. The drum is enamelled and the pencil is a EXPLOSION AND COOLING 127 blacklead one. The pressure of the explosion acts upon the indicator piston, and a line is traced upon the drum, which shows the rise and fall of pressure. The rising line traces the progress of the explosion ; the falling line the progress of the loss of pressure by cooling. The rate of the revolution of the drum being known, the interval of time elapsing between any two points of the explosion or. cooling curve is also known. That is, the curve shows the maximum pressure attained, the time of attaining it, and the time of cooling. Line b on fig. 36 is a facsimile of the curve produced by the explosion of a RevolvinsDrum Weight Fig. 35. — Clerk Explosion Apparatus of years 1884 and I mixture containing i vol. hydrogen and 4 vols. air. Each revolution of the drum was accomplished in 033 sec, so that each tenth of a revolution takes 0'033 sec. The vertical divisions give time ; the horizontal, pressures. In this experiment the maximum pressure produced by the explosion is 68 lbs . per sq. in. above atmosphere, and it is attained in 0-026 second. Compared with the rate of increase the subsequent fall is very slow. The rise occurs in 0'026 second ; the fall to atmosphere again]jtakes i'05 second, or nearly sixty times the other. It is in fact an indicator diagram from an explosion where the volume is constant, the motor piston being absent, and the only 128 THE GAS, PETROL, AND OIL ENGINE " a ■ S'S s aJ3 a cause of loss of pressure is cooling by the enclosing walls. The exact composition of the mixture, its uniform admixture, the temperature and pressure before ignition, are all accurately known. After studying explosions under these known conditions, it becomes easier to under- stand what occurs under more complex conditions, where the moving piston makes the cooling surface change, and where the expansion doing work also requires consideration. As the rapidity of the increase of pressure measures the explosiveness of a mixture, the time occupied from the commencement of in- crease to maximum pressure will be called the time of explosion. The explosion is complete when maximum pressure is attained. It does not follow from this that the combustion is complete ; that is another matter. The explosion arises from the rapid spreading of the flame throughout the whole mass of the mixture, which may be called the inflammation of the mixture. More or less rapid inflam- mation means more or less explosive effect, but not complete combustion. The complete burning of the gases present may not occur till long after complete inflammation. The terms combustion, ex-plosion, and inflammation will be used in this sense alone : Combustion burning ; complete com- bustion, the complete burning of the carbon of the combustible gas to carbonic acid, and the hydrogen to water. So long as any portion of the combustible remains uncombined with oxygen the combustion ime ( 8 ,g >8 S- S ° - - « ' " 1 ,' - A c \ \ c; lL \ t 1 — :: 3 \ ~s- " — — ^ . Wo U rt lU 2 "^ T3 ^ 3 >Sh ■aiaqdsoiujB aAoqB ■UT "bs lad -sqf ui ajnssajj is incomplete. Complete explosion, the attainment of maximum pressure. Time of explosion ; the time elapsing between beginning of increase and maximum pressure. Complete inflammation, the complete spreading of the flame throughout the mass of the mixture. Confusion has arisen through the indifferent use of these terms, which are really distinct and are not synonymous. GASEOUS EXPLOSIONS 129 With mixtures made with Glasgow coal gas the author has obtained the following maximum pressures and times of explosion : Explosion in a Closed Vessel. {Clerk) Mixtures of air and Glasgow coal gas Temp, before explosion 18° C. Pressure before explosion atmospheric. Mixture. Max. press, above atmcs. in pi unds per sq. in. Time of explosion Gas. I vol. I vol. I vol. I vol. I vol. Air. 13 vols. II vols. 9 vols. 7 vols. 5 vols. 52 63 69 89 96 0-28 sec. ©•i8 sec. 0-13 sec. o'07 sec. 0'05 sec. The highest pressure which any mixture of coal gas and air is capable of producing without compression is only 96 lbs. per sq. in. above atmosphere, and the most rapid increase is not more rapid than always occurs in a steam cylinder at admission. Many are still prejudiced against gas, compared with steam, because of the so-called explosive effect, and the fear that gas explosions may occasion pres- sures quite beyond control, like solid explosives. The fear is quite unfounded ; the pressure produced by the strongest possible mixture of coal gas and air is strictly limited by the pressure before ignition, and can always be accurately known ; and so provided for by a proper margin of safety in the cylinders and other parts subject to it. The most dilute mixture of air and Glasgow gas which can be ignited at atmospheric pressure and temperature contains ^^ of its volume of gas, and the pressure produced is 52 lbs. above atmosphere. The time of explosion is 0'28 second ; so slow is the rise that it cannot with justice be termed an explosion. It is too slow to be of any use in an engine running at any reasonable speed ; the stroke would be almost complete before the pressure had risen. The mixture con- taining I of its volume of gas is that with just enough oxygen to burn the gas. It is anomalous that the highest pressure is given with excess of coal gas. The rate of ignition also is greatest with that mixture. This agrees with the results obtained by Mallard and Le Chatelier, excess of hydrogen giving the highest rate of inflammation. Similar experiments were made with air and Oldham coal gas. VOL. I K 130 THE GAS, PETROL, AND OIL ENGINE Explosion in a Closed Vessel. (Clerk) Mixtures of air and Oldham coal gas Temp, before explosion 17° C. Pressure before explosion atmospheric Mixture Max. press, above atmos. in pounds per sq. in. Time of explosion Gas Air I vol. 14 vols. 40 0'45 sec. I vol. 13 vols. 51-5 0-31 sec. X vol. 12 vols. 60 0-24 sec. I vol. 11 vols. 61 0-17 sec. I vol. 9 vols. 78 008 sec. I vol. 7 vols. 87 0'o6 sec. T vol. 6 vols. 90 0-04 sec. I vol. 5 vols. 91 0-055 sec. I vol. 4 vols. 80 o-i6 sec. The highest pressure in this case is 91 lbs. per sq. in. above atmosphere, but the more rapid explosion is 0-04 second and 90 lbs. pressure, a little less pressure than is given by Glasgow gas but a slightly more rapid ignition. The mixtures are evidently more inflam- mable, as the critical mixture is rs volume of gas instead of j\ as with Glasgow gas. Although repeatedly tried, a mixture of i volume gas 15 volumes air failed to inflame with the spark. Hydrogen and air mixtures were also tested as follows : Explosion in a Closed Vessel. [Clerk) Mixtures of air and hydrogen Temp, before explosion . . . . . 16° C. Pressure before explosion atmospheric Mixture Max. press, above atmos. in pounds per sq. in. Time of explosion Hyd. I vol. 1 vol. 2 vols. Air. 6 vols. 4 vols. 5 vols. 41 68 0-15 sec. 0'026 sec. o-oi sec. The inferiority of hydrogen to coal gas, volume for volume, is very evident ; the highest pressure is only 80 lbs. above atmosphere, and the mixture requires \ of its volume of hydrogen to give it, while coal gas gives the same pressure with about yV volume. The hydrogen mixture, too, ignites so rapidly that it would occasion shock in practice, the strongest mixture having an explosion time of one-hundredth of a second. With gas the most rapid is four-hundredths of a second. THE BEST MIXTURES 131 The Best Mixture for use in Non-compression Engines From these tables can be ascertained the best gas and the best mixture for use in non-compression engines with cyHnders kept cold. duuuOuuu E-J-oo -0. 1 1 1 1 M 1 1 "saaqdsoiu^'E aAoqc •ui 'bs aad -sqi ui ajnssajfX •ajaqdsomjB SAOq-E 'bs asd ■sqj ui ajnssojj Take first Glasgow gas, and determine which mixture gives the best result. (i) Power of producing pressure. K 2 132 THE GAS, PETROL, AND OIL ENGINE Suppose one cubic inch of Glasgow coal gas to be used in each of the five mixtures, whose maximum pressures and times of explosion are given in the table on p. 129, the mixtures would measure respectively 14, 12, 10, 8, and 6 cubic ins. Let them be placed in cylinders of 14, 12, 10, 8, and 6 sq. in. piston area ; the piston will in each case be raised one inch from the bottom of its cyHnder. If the pressures upon the piston were the same, equal movements of piston would give equal power ; if therefore the mixtures gave equally good results the maximum pressure multiplied by the piston area will in all cases be the same. Multiplying 14, 12, 10, 8, and 6 by their corresponding pressures 52, 63, 69, 89, and 96 respectively, the products are 728, 756, 690, 712, and 576. These numbers are the pressures in pounds which each mixture is capable of producing with one cubic inch of Glasgow coal gas, cylinders of such area being used that the depth of mixture is in every case one inch. Proportion of Glasgow gas in mixture . i, ^, j^, J, >. Pressure produced upon pistons by one 1 2g_ ^jg_ ^^_ ^^^_ ^^^ ^^^ cubic inch J The best mixture is seen at a glance ; it is that containing one- twelfth of gas. The pressure produced by one cubic inch of gas is at its highest value 756 lbs., in a cylinder of 12 ins. piston area, and containing 12 cubic ins. of mixture. In modern gas engines the time taken by the piston to make the working part of its stroke is generally about one-fifth of a second. If the pressure in one mixture has fallen more, proportionally, in that time, then, although it may give the highest maximum, it may lose too rapidly to give the highest mean pressure. To find this cooling effect, find the pressure to which each mixture falls at the end of 0'2 second after maximum pressure ; it is in the different cases : Mixture containing gas • • • • n» A' w> s> 5- Time after beginning explosion (0-2 sec. | ^.^g_ ^.^g^ ^.^^_ ^.^^^ ^.^^ ^^^^ after max. pressure) . . . . "I Pressure in lbs. per sq. in. . . . 43, 48, 47, 55, 57. Press. X respectively by 14, 12, 10,^. \^^^^ ^^g_ ^^^^ ^^^_ ^^2_ and 6 . . . .... J The lower row expresses the relative pressures still remaining after allowing each explosion to cool for one-fifth of a second from complete explosion ; they express the resistance to cooling possessed by the mixtures. It is evident at once that the strongest mixtures cool most rapidly ; a higher temperature being produced, more of the heat of the explosion is lost in a given time. BEST MIXTURE FOR NON-COMPRESSION ENGINES 133 (2) Power of producing pressure and resisting cooling. To find the best mixture for producing pressure and resisting cooling, those numbers are to be added to the corresponding ones for maximum pressure : Proportion of Glasgow gas in mixture . j\, i, ^, |, |. Pressure produced upon pistons by one, ^^g^ ^^g^ g^^^ ^j^, 576. cubic inch gas f Pressure remaining upon pistons 0-2 sec. , g^^^ j^g_ ^^q_ ^^^ ^^^_ after complete explosion. . . . ' Mean pressure 665, 666, 580, 576, 459. The mean of the two sets gives numbers expressing the relative values of the mixture for producing pressure, and at the same time resisting cooling. The two weakest mixtures are best in both respects ; the low result given by the strongest mixture is due to the fact that excess of gas is present and it remains unburned, it proves how easily the consumption of an engine may be increased by even a slight excess of gas in the mixture. The two best mixtures ignite too slowly, but in the actual engine that is easily controlled, as will be explained later. The best mixtures are i volume gas 13 volumes air, and i volume gas 11 volumes air. With more gas the economy will rapidly diminish. The experiments with Oldham gas treated in the same way give the following results : Proportion of Oldham gas in , i_ 1 j_ _i_ j^ 1 1 11 mixture ' "' "' "' "^^' "' "' "" »' '' Pressure produced upon pistons , g^g_ ^^i, 780, 732, 780, 696, 630, 546, 400. by one cubic inch gas . . ) Pressure remaining upon pistons 0'2 sec. after complete explo- • 31, 40, 42, 44, 44, 47, 52, 50, 46. sion per sq. in ' Pressure per piston . . . 465, 560, 546, 528, 440, 376, 364, 300, 230. Mean pressure upon piston . . 532, 640, 663, 630, 610, 536, 497, 423, 315.- Here, too, the best mixture lies between one-twelfth and one-four- teenth of gas ; with less and more gas the result becomes worse and worse. Glasgow and Oldham gases seem to be very nearly equal in value per cubic foot for the production of power, as the pressure pro- duced from one cubic inch in the best mixture of each is very similar. The average pressures during o"2 second from complete explosion are exceedingly close, Glasgow gas mixture containing one-twelfth gas giving 666 lbs. pressure per cubic inch of gas, and Oldham gas for the same mixture and the same quantity giving 630 lbs. : Glasgow gas 134 THE GAS, PETROL, AND OIL ENGINE one-fourteenth mixture 665 lbs. pressure, Oldham gas 640 lbs. The hydrogen experiments give as follows : Proportion of hydrogen gas in mixture . ^ 1 f , Pressure produced upon pistons by one 1 „ , . . ■^, , , r r 1 > 287, 340, 280. cubic mch hydrogen . . . .J " ■^~' • Pressure remaining upon pistons 0-2 sec, after complete explosion per sq. inch . Pressure per piston 245, 195, 140. Mean pressure upon piston . . . 266, 267, 210. } 35. 39. 40. The best mixture with one cubic inch of hydrogen only gives a pressure of 267 lbs. available for 0"2 second, so that its capacity for producing power, compared with Glasgow and Oldham gas, is as 267 is to 665 and 640 respectively. Assuming the actual heat capacity of the mixtures to be equal, then to produce equal power with Glasgow gas nearly two-and-a-half times its volume of hydrogen is required. The idea is very prevalent among inventors that if pure hydrogen and air could be used, greater power and economy would be obtained ; these experiments prove the fallacy of the notion. Hydrogen is a poor gas to use alone in the cylinder of a gas engine ; it is useful in conferring inflammability upon dilute mixtures of other gases, but when present in large quantity in coal gas it diminishes its value per cubic foot for power. Pressures produced if no Loss or Suppression OF Heat Existed From the fact already mentioned in the last chapter, that the theoretical temperatures of combustion are never attained in reality, it will naturally be expected that the pressures produced by explosions in closed vessels will also fall short of theory. This is found to be the case. It has been observed by every experimenter upon the subject, beginning with Hirn in 1861, who determined the pressures produced by the explosion of coal gas and air, and hydrogen and air. He used two explosion vessels of 3 and 36 litres capacity ; they were copper cylinders with diameters equal to their length. He used a Bourdon spring manometer to register the pressure. He states that : (i) With 10 per cent, hydrogen mtroduced the results were : according to experiment, 3^25 atmospheres ; according to calculation, 5'8 atmospheres. (2) With 20 per cent, of hydrogen, the results were : according to experiment, 7 atmospheres, which is very much below the calculation. (3) With 10 per cent, of lighting gas introduced the results were : according to experiment, 5 atmospheres, i.e. much more than with the introduction of an equal volume of pure hydrogen. ACTUAL PRESSURES LESS THAN CALCULATED 135 He notices especially the low pressure produced by hydrogen as compared with lighting gases, but observes truly that this should not excite surprise — although the heat value of hydrogen is great, yet it is so when compared with equal weights of other substances — and that coal gas being four or five times as heavy as hydrogen, quantity is balanced against quality ; therefore volume for volume it gives out more heat. He considers that there is no difficulty in explaining the very con- siderable difference found between calculation and experiment, as the metal sides are at so low a temperature compared with the explosion, that the heat is rapidly conducted away, and the attainment of the highest temperature is impossible. Bunsen, in his experiments, observed the same difference, and so later did Mallard and Le Chatelier. The author's experiments fully confirm the accuracy of those observers. In no case, whether with weak or strong mixtures of coal gas and air, or hydrogen and air, is the pressure produced which should follow the complete evolution of heat. Thus, with hydrogen mixtures [Clerk's experiments) : Per sq. in. I vol. H, 6 vols, air, gives by experiment . 41 lbs. above atmosphere. The calculated pressure is . . . . 88'3 1 vol. H, 4 vols, air, experiment gives . 68 Calculated pressure is 124 2 vols. H, 5 vols, air, experiment gives . . 80 Calculated pressure is . . . .176 Without exception the actual pressure falls far short of the calcu- lated pressure ; in some manner the heat is suppressed or lost. That the difference cannot altogether be accounted for by loss of heat is easily proved ; the fall of pressure is so slow from the maximum that it is impossible that any considerable proportion of heat can be lost in the short time of explosion. If so large a proportion were lost on the rising curve, it could not fail to show upon the falling curve ; it would fall, in fact, as quickly as it rose. Again, the increase of pressure would be less in a small than in a large vessel, as the small vessel exposes the larger surface proportionally to the gas present. It is found that this is not so. Bunsen used a vessel of a few cubic centi- metres capacity, and got with carbonic oxide and oxygen true explosive mixture iO'2 atmospheres maximum pressure ; Berthelot with a vessel 4000 cubic centimetres capacity got lo'i atmospheres ; with hydrogen true explosive mixture, Bunsen 9'5 atmospheres, Berthelot, 9-9 atmo- spheres. All the difference, therefore, cannot be accounted for by loss before complete explosion. Mixtures of air and coal gas give similar results. 136 THE GAS, PETROL, AND OIL ENGINE The following are the observed and calculated pressures for Oldham coal gas [Clerk's experiments) I vol gas, 14 vols, air, experiment gives Calculated pressure is I vol. gas, 13 vols, air, experiment gives Calculated pressure is I vol. gas, 12 vols, air, experiment gives Calculated pressure is I vol. gas, 1 1 vols, air, experiment gives Calculated pressure is I vol. gas, 9 vols, air, experiment gives Calculated pressure is I vol. gas, 7 vols, air, experiment gives Calculated pressure is I vol. gas, 6 vols, air, experiment gives Calculated pressure is Per sq. in. 40 lbs. above atmosphere. 89-5 51-5 96 60 103 61 112 78 134 87 168 90 192 The results with Glasgow gas are so similar that it is unnecessary to give a table ; in no case does the maximum pressure account for much more than one-half of the total heat present. It is to be noted that these calculations assume the specific heats determined by Regnault to be true for high temperatures. Regnault's numbers, as given at p. 118, have been used for the calculations. As all of the deficit cannot have disappeared previous to complete explosion, it follows that the gases are still burning on the falling curve or that the specific heats have greatly increased. That is, the falling curve does not truly represent the rate of cooling of air heated to the maximum temperature, because heat is being continually added by the continued combustion of the mixture, or varying specific heat produces an effect similar to heat evolution by combustion. It may, however, be taken as completely proved by the complete accord of all physicists who have experimented on the subject, that for some reason nearly one-half of the heat present as inflammable gas in any explosive mixture, true or dilute, is kept back and prevented from causing the increase of pressure to be expected from it on the assumption of constant specific heat. Although differences of opinion exist on the cause, all are agreed on the fact ; they also agree in considering that inflammation is compete when the highest pressure is attained. Temperatures of Explosion With a mass of any perfect gas confined in a closed vessel the absolute temperatures and pressures are always proportional ; double temperature means double pressure. Temperatures t, t (absolute), T P pressures corresponding p, ^ ; then - = - (Charles's law). If explo- • TEMPERATURES OF EXPLOSION 137 sive mixtures behaved as perfect gases, the pressure before explosion and temperature being known, the pressure of explosion at once gives the corresponding temperature. It has been shown at page 113 that explosive mixtures do not fulfil this condition, but change in volume from chemical causes quite apart from physical ones. It follows, therefore, that these changes must be known before the temperature of the explosion can be calculated from the pressure. In the cases of hydrogen and carbonic oxide true explosive mixtures with oxygen, a contraction of volume is the result of combination. It comes to the same thing as if a portion of the perfect gas in the closed vessel was lost during heating ; the temperature, then, could not be known at the higher pressure unless the volume lost is also known. Suppose one-third of the volume to disappear upon cooling to the origina.' temperature the pressure would be reduced to two-thirds of the original pressure, and this fraction of the original pressure must be taken as />, = 10. As both steam and carbonic acid at temperatures high enough to make them perfectly gaseous occupy two-thirds of the volume of their free constituents, it follows that f^ must be taken as f p, wherever the temperatures are such that combination is complete. But here another difficulty occurs. Bunsen found that hydrogen and oxygen in true explosive mixtures gave an explosion pressure of 9"5 atmospheres. The calculated pressure for complete combustion and allowing for chemical contraction is 21 "3 atmospheres. It is evident enough that complete combustion has not occurred, but it is difficult to say what fraction remains uncombined. Yet unless the fraction in combination be known the contraction cannot be known, and therefore the temperature corresponding to the pressure cannot be known. Berthelot has pointed out that in a case of this kind the true tem- perature cannot be calculated, but it may be shown to lie between two extreme assumptions, both of which are erroneous : (i) Temperature calculated on assumption of no contraction. (2) Temperature calculated on assumption of the complete con- traction. Let the two temperatures be (i) t^ and (2) t. T' T 2 vols. H, I vol. O, explosion pressure 1 24.4.0° C ^8oq° C (absolute) 9-9 atmospheres. . . J 2 vols. CO, t vol. O, explosion pressure 1 2612° C 4140° C (absolute) iO'8 atmospheres . . I The lower temperature could only be true if no combination what- ever had occurred, which is impossible, as then no heat at all could be evolved ; the higher temperature could only be true if complete 138 THE GAS, PETROL, AND OIL ENGINE combination, and therefore complete contraction, occurred The truth is somewhere between these numbers. When the explosive mixture is dilute, the limits of possible error are narrower, because the possible proportion of contraction is less ; with hydrogen and air mixture in proportion for complete combina- tion, 2 volumes of hydrogen require 5 volumes of air. The greatest possible contraction of the 7 volumes is therefore i volume. If all the hydrogen burned to steam, the 7 volumes contract to 6 volumes. With more dilute mixtures the proportion diminishes. With a mixture containing i of its volume hydrogen, 10 volumes can only suffer contraction to 9 volumes. With y volume hydrogen, 14 volumes can contract to 13 volumes. The limits of maximum temperatures for those mixtures are as follows [Clerk) : I vol. H, 6 vols, air, explosion pressure , (absolute), 55-7 lbs. per sq. in. . .1 1 vol. H, 4 vols, air, explosion pressure . (absolute), 82-9 lbs. per sq. in. . . ' 2 vols. H, 5 vols, air, explosion pressure , (absolute), 94-7 lbs. per sq. in. . .1 T' 826° C. 1358° C. 1615° C. 909° 1539° c. 1929° c. The possible error is here much less than with true explosive mixtures ; coal gas is of such a composition that some of its con- stituents expand upon decomposition previous to burning, and so to some extent balance the contraction produced by the burning of the others. The possible error is therefore still further reduced. The composition of Manchester coal gas as determined by Bunsen and Roscoe is as below. The oxygen required for the complete com- bustion of each constituent is also given, and the volumes of products formed. Analysis of Manchester Coal Gas. {Bunsen and Roscoe) Amount required for complete combustion Products Hydrogen, H. Marsh gas, CHj . Carbonic oxide, CO . Ethylene, C,H| . Tetrylene, C4HS . Sulphuretted hydrogen, H.^S Nitrogen, N. Carbonic acid, CO^ Total .... vols. 45-58 34-9 6-64 4-08 2-38 0-29 2-46 3-67 wis. *-79 69-8 3-32 12-24 14-28 o'43 vols. 45-58, H,0 104-7, CO., and H.p 6-64, cof, 16-32, CO2 and HjO 19-04, CO2 and HjO 0-58, H,0 and SO^ 2-46 3-67 1 00 -GO 122-86 198-99, CO,, H.O &SO2 TEMPERATURES OF EXPLOSION 139 When burned in oxygen 100 volumes of this sample of gas require i22'86 volumes of oxygen, total mixture 222'86 volumes ; the pro- ducts of the combustion measure igS'gg volumes. Calculating to percentage, 100 volumes of the mixture will contract to Sg'^ volumes of the products. As 100 volumes of the mixture will contain 55'i volumes of oxygen, it follows that if air be used, four times that volume of nitrogen will be associated with it, that is, 55'i x 4 = 220-4. The strongest possible explosive mixture of this coal gcis with air containing 100 volumes of the true explosive mixture will be 320-4 volumes, and it will contract upon complete combustion to 309-8 volumes. One volume of this gas requires 6-14 volumes air for complete combustion, and 100 volumes of the mixture contract to 96-6 volumes of products and diluent, a contraction of 3-4 per cent. Dilution still further diminishes the change ; thus a mixture, i volume gas 13-28 volumes air, will have only half that contraction, or 1-7 per cent. From these figures it is evident that the limits of possible error in calculating temperature from pressure of explosion does not exceed, in the worst case, with coal gas and air 3-4 per cent., and in weaker .mixtures half that number. The fact that the whole heat is not evolved at the explosion pressure, and that therefore the whole con- traction does not occur then, further reduces the error. It is then nearly correct to calculate temperature from pressure without deduc- tion for contraction. This has been done for Glasgow gas and for the Oldham gas experiments by the author. Explosion in a Closed Vessel. {Clerk) Mixtures of air and Glasgow coal gas Temp, before explosion . Pressure before explosion 18° C. atmos. 14V lbs. ,.. ^ Max. press, above atmos. M''""'^^ in pounds per sq. in. Temp, of explosion calculated from observed pressure Gas I vol. I vol. I vol. I vol. I vol. Air j 13 vols. 52 II vols. 1 63 9 vols. 69 7 vols. 89 5 vols. 96 1047° C. 1265° C. 1384° c. 1780° C. 1918° c. 140 THE GAS, PETROL, AND OIL ENGINE Mixtures of air and Oldham coal gas Temp, before explosion 17° C. Max. press, above Temp, of explosion Theoretical temp. Mixture atmos. in pounds calculated from of explosion if all per sq. in. observed pressure heat were evolved Gas Air I vol. 14 vols. 40 806° c. 1786° c. T vol. 13 vols. 51-5 1033° c. 1912° c. I vol. 12 vols. 60 1202° c. 2058° c. I vol. II vols. 61 1220° C. 2228° c. I vol. 9 vols. 78 1557° C. 2670° c. ^ vol. 7 vols. 87 1733° c. 3334° C. I vol. 6 vols. 90 1792° c. 3808° C. I vol. 5 vols. 91 1812° C. I vol. 4 vols. 80 1595° c. Those temperatures calculated from maximum pressure, although not quite true are very nearly so, whatever be the theory adopted to explain the great deficit of pressure. It does not follow, however, that they are the highest temperatures existing at the moment of explosion ; they are merely averages. The existence of such an intensely heated mass of gas in a cold cylinder causes intense currents, so that the portion in close contact with the cold walls will be colder than that existing at the centre. There will be a hot nucleus of con- siderably higher temperature than that outside, but whatever that temperature may be, the increase of pressure gives a true average. It may be taken, then, that coal gas mixtures with air give upon explosion temperatures ranging from 800° C. to nearly 2000° C, depending on the dilution of the mixture. The more dilute the mixture the lower the maximurn temperature ; increase of gas increases maximum tem- perature at the same time as it increases inflammability. The author has made explosion experiments in the same vessel with mixtures previously compressed, and finds that the pressures produced with any given mixture are proportional to the pressure before ignition, that is, with a mixture of constant composition, double the pressure before explosion, keeping temperature constant at 18° C, doubles the pressure of explosion. Efficiency of Gas in Explosive Mixtures Rankine defines available heat as follows : ' The available heat of combustion of one pound of a given sort of fuel is that part of the total heat of combustion which is communi- cated to the body to heat which the fuel is burned ; and the efficiency of a given furnace, for a given sort of fuel, is the proportion which the available heat bears to the total heat.' FURNACE EFFICIENCY OF GAS 141 The gas engine contains furnace and motor cylinder in one ; nevertheless the efficiency of the working fluid is quite as distinct from the furnace efficiency as in the steam engine. Rankine's definition is quite true for the gas engine. The fuel being gas, the working fluid consists of air and its fuel and their combinations ; the available heat is that part of the heat of com- bustion which serves to raise the temperature of the working fluid ; the part which flows into it to make up for loss to the cold cylinder walls cannot be considered available. To be truly available it must either increase temperature, or keep it from falling by expansion. The heat flowing through the cylinder walls is a furnace loss, incident to the explosion method of heating. The experiments upon explosion in a closed vessel provide data for determining the furnace efficiency as distinguished from that of the working fluid. The proportion of heat flowing from an explosion to the walls in unit time wiU depend upon the surface of the walls for any given volume. The smaller the cooling surface in proportion to volume of heated gases, the slower will be the rate of cooling. There- fore to be applicable to any engine, the explosion vessel in which the experiments are made should have the same capacity and surface as the explosion space of the engine. The author's experiments are therefore only strictly applicable to engines with cylinders similar to his explosion vessel. Within certain limits, however, the error introduced by applying them to other engines is not large. Assuming the stroke of a gas engine (after explosion) to take 0-2 second, this may be taken £is the time during which the pressure of explosion must last if it is to be utilised by the engine. In a closed vessel the pressure falls considerably in 0"2 second : the average pressure may be taken as nearly indicating the available pressure during that time. The heat necessary to produce that pressure is the available heat ; and its proportion to the total heat which the gas present in the mixture can evolve is the efficiency of the gas in that explosive mixture. With Oldham gas the best mixture is (table, p. 133) i volume gas 13 volumes air ; the average pressure during the first fifth of a second is 51 lbs. per sq. in. above atmosphere. If all the heat present heated the air, the pressure should be 103 lbs. effective, so that the efficiency of the heating method is {j^g = 0-49. The strongest mixture which still contains oxygen in excess is I volume gas 7 volumes air, the average available pressure is 67 lbs. per sq. in. (all heat evolved would give 168 lbs.), the efficiency is rVs = 0-40 nearly. 142 THE GAS, PETROL, AND OIL ENGINE Calculated in this way the efficiency values for Oldham gas mixtures are : Prop, of Oldham gas in mixture i, i, i, i, i, |, } Heating efficiency . . . o'40, o'48, 0'50, 0'43, 0'46, 0"40, o"37. The furnace efficiency plainly diminishes with increased richness of the mixture in gas. These calculations, however, assume constant specific heat and completed combustion, and therefore include more than furnace loss. Time of Explosion in Closed Vessels The rates of the propagation of flame in explosive mixtures given in tables, pages 113 and 115, are true only where the inflamed portion is free to expand without projecting itself into the unignited portion. They are the rates proper for constant pressure. Where the volume is constant, in a closed vessel, the part first inflamed instantly expands and so projects the flame surface into the mass, compressing what remains into smaller space. To the rate of inflammation at constant pressure is added the rate of projection of the flame into the mass by its expansion and also the increased rate of propagation in the unignited portion by the heating due to its compression by the portion first inflamed. It follows that the rate continually increases, as the inflammation proceeds untU it fiUs the vessel. This is evident from all the explosion curves. The pressure rises slowly at first, then with ever-increasing rate till the explosion is complete ; thus the explosio]i curve for hydrogen mixture with air [ ^H ) shows an increase of 17 lbs. in the first 0'005 second, the maximum pressure of 80 lbs. being attained in the next 0'005 second. With the weaker mixtures the same thing occurs, rise of pressure, slow at first, then more rapid, and in some cases becoming slow again before maximum pressure. The time taken to get maximum pressure varies much with the circumstances attending the beginning of the ignition. If a considerable mass bejgnited at once, by a long and powerful spark, or by a large flame, the ignition of the weakest mixture may be made almost indefinitely rapid. Something very like Berthe- lot's explosive wave may result. This is due to the great mechanical disturbance caused by the rapid expansion of the portion first ignited ; the smaller that portion is, the more gently does the flame spread. A small separate chamber connected with the main vessel, if filled with explosive mixture and ignited, will project a rush of flame into the main vessel and cause almost instantaneous ignition. The shape TIME OF EXPLOSION 143 of the vessel, too, has a great effect upon the rate. Where it is cylindrical and large in diameter proportional to its axial length, ignition is extremely rapid, the flame is confined at starting, and is rapidly deflected by the cylinder ends, and so shoots through the whole mass. By so arranging the explosion space of a gas engine that some mechanical disturbance is permitted, it is easy to get any required rate of ignition even with the weakest mixtures. The maximum pressure is not increased by rapid ignition. Starting the ignition from a small spark, the time taken to ignite increases with the volume of the vessel. Berthelot has experimented upon this point with explosion vessels of three capacities, 300 cubic centimetres, 1500 cubic centimetres, and 4000 cubic centimetres. He finds time of explosion (he also takes 300- J ^ — 200- 100 - " TtME. , SECS. Fig. 38. — Explosion of Oldham Coal Gas and Air Mixture in Closed Vessel, with previous compression of 40 lbs. per sq. in. above atmos. maximum pressure to indicate complete explosion) of mixture 2 vols. H, I vol. 0, and 2 vols. N, in 300 cubic centimetre vessel, 0-0026 second ; and in 4000 cubic centimetre vessel, o-oo68 second. With mixture of carbonic oxide and oxygen, 2 vols. CO, i vol. O, smaller vessel, O'OiaS second ; larger vessel, 0'0i55 second. Mixtures with air were much slower. The conclusion, then, is obvious, that in large engines the time of explosion will be longer than in small ones. Later Experiments. Explosions in Closed Vessels The experiments on gaseous explosion in a closed vessel were began by the author in 1883, and in part communicated to the late Professor 144 THE GAS, PETROL, AND OIL ENGINE Fleeming Jenkin for the purpose of a lecture delivered by him to the Institution of Civil Engineers on February 21, 1884. The hydrogen and Glasgow coal gas curves were published in that lecture. Other experiments were made in 1885, and all the foregoing results were published in a paper read by the author on March 9, 1886. Fig. 39- — Clerk Explosion Apparatus of 1900 Mixtures compressed before ignition were also tested in the same vessel. Fig. 38 shows one of the curves published in a paper by the author read to the Society of Chemical Industry on January 29, 1886. Further experiments were made in 1900 with the author's new apparatus. EXPLOSION IN A CLOSED VESSEL 145 Clerk's Later Apparatus, 1900 The apparatus is shown at fig. 39 to consist of a closed cylindrical vessel 7 ins. diameter, 7 ins. long, internal measurement, 269 cubic ins. capacity, bored, turned, and supplied with strongly bolted covers as in the earlier device. A Richards indicator was mounted upon the upper cover, as shown, and provision was made for firing electrically. The rotating drum used in the earlier apparatus v/as dispensed with, and a long falling platform or slide was adopted. The object was to obtain a diagram on a long indicator card, so that no transfer from a drum to a tracing was required. The card was over 30 ins. long. The long slide carrying the paper was fitted accurately Fig. 40. — Outline of Cam Groove in Clerk Explosion Apparatus on V slides, and it was suspended by a pianoforte wire from a cam groove in a fljrwheel placed above it. The cam groove resembled the fusee of a watch, and the curve was so constructed that when a lever was operated the fl57wheel was let go and the weight of the slide forced it to rotate. For the first 3 ins. of the fall the motion of the slide accelerated, but then the pianoforte wire arrived at a part of the cam surface of smaller diameter than at first, and the diameter progres- sively diminished, so that as the flywheel accelerated the diameter diminished. In this way, after 3 ins. fall, the motion of the slide re- mained uniform until at the end of its travel a brake applied to the flywheel stopped its motion. This chronograph resembles the falling VOL. I. L 146 THE GAS, PETROL, AND OIL ENGINE bar chronograph used to determine the velocity of a bullet, but the fall of the slide is controlled by the fl}^wheel instead of being free. The '1.N30 93ati03a 'atjnx\/ti3dw3x o o o o c ti « 2 a I , , l_ •Ni -tis -aad sen '3anss3ad EXPLOSION IN A CLOSED VESSEL 147 time of fall is determined and checked by means of a tuning-fork operated electrically and actuating an electrical relay which traces a curve on the falling paper of 200 periods per second. Once the apparatus is calibrated by this tuning-fork, it can be used for a great number of successive experiments. The operations of charging the explosion vessel, measuring the gas, and so forth are the same as were used in the earlier device. On fig. 39, A is the explosion vessel with its indicator, which in this case had a metallic pencil, b is the falling slide, c the fljAvheel with its cam groove C|, shown on a larger scale at fig. 40. D is the piano- forte wire, E the trigger lever for letting go the slide, and f the wedge brake for stopping the flywheel at the end of its movement ; G is an electric contact arrangement, arranged to pass the electric spark through the explosion vessel at any period of the movement of the fly- wheel ; H is the clamping device to stretch the metallic paper and hold it on the falling slide ; e is the Richards indicator in position. This apparatus is very useful for performing a great number of experiments, as it requires no motive power to keep a drum in rotation, and the parts are so solidly constructed that derangement due to change in friction is easily avoided. Experiments at atmospheric pressure with London coal gas were made with this apparatus, and the diagrams of one set are shown at fig. 41. The analysis of the London coal gas used is given below the dicigrams. These explosions with London coal gas give the following maximum pressures, temperatures, and times of explosion : Explosion in a Closed Vessel. {Clerk 1900) Mixtures of air with London Coal Gas Temp, before explosion . Pressure before explosion 16° C. I4'8 lbs. per sq. in. Max. Pressure Temp, of Explosion Mixture above atmosphere from observed Time of Explosion in lbs. per sq. in. pressure Gas Air I vol. 12 vols. 4 — — I vol. II vols. 58 1150° c. 0-290 sec. I vol. 10 vols. 60 1155° C. 0-305 sec. I vol. 9 vols. 65 1270° C. 0-155 sec. I vol. 8 vols. 77 1475° c. 0-087 sec. I vol. 7 vols. 80 1565° c. 0-067 sec. I vol. 6 vols. 85 1660° c. 0-055 sec. I vol. 5 vols. 87 1710° c. 0-042 sec. I vol. 4 vols. 93 1830° c. 0-045 sec. 148 THE GAS, PETROL, AND OIL ENGINE Calculating the best mixtures as before, it is found that i gas to 11 of air is best both for maximum pressure and duration for ^ second. Massachusetts Institute of Technology Experiments In 1898 an apparatus somewhat similar to the author's was con- structed at the Massachusetts Institute of Technology at Boston, and TUNING FORK INDICATOR DISC PENCIL INDICATOR Fig. 42. — Massachusetts Institute of Technology — Explosion Apparatus many interesting experiments have been made with it. It consists of a cast-iron cylinder of 310 cubic ins. capacity, so that it closely corresponds to the Clerk apparatus of 317 ins. capacity. To in- troduce the mixture the explosion vessel is exhausted by a pump and BOSTON EXPLOSION EXPERIMENTS 149 scavenged by admitting fresh air ; this operation is repeated suffi- ciently to clear the cylinder from the products of the previous explosion. The cylinder is then exhausted down to a carefully measured pressure and coal gas is admitted to raise the pressure to atmosphere again. By this device any desired proportion of gas may be mixed with air within the cylinder. In the early Clerk experiments the record of explosion and cooling was taken upon a rotating drum ; in the Boston experiments a power-driven disc was adopted, and the line was traced by the indicator on the face of this disc. The disc at the same time receives a time tracing from a pointer attached to one arm of a tuning- !• — 60 LBS.-H Mixture : i vol. Boston gas, 9 vols, air Fig. 43. — Record from Explosion Apparatus Massachusetts Institute of Technology, Boston fork, which is kept in movement electrically. The mixtures are fired electrically. Fig. 42 shows a general view of the apparatus, and iig. 43 shows the record as taken on the disc. These diagrams are taken on the base of a spiral line, and, although mechanically convenient, this leads to increased work in reducing the observations. Fig. 44 shows a set of these diagrams developed upon a straight line base, so that they directly compare with the Clerk diagrams already discussed. 150 THE GAS, PETROL, AND OIL ENGINE The analysis of the Boston coal gas used was as follows : ' Analysis of Boston Coal Gas Massachusetts Institute of Technology Per cent. Carbonic qxide, CO 25-3 lUuminants 12-0 Carbonic acid, COj 1-9 Marsh gas, CH, 28-9 Nitrogen, N 3-0 Hydrogen, H . . 27-9 Oxygen, O . . . i loo-o The percentage of gas present is marked in the corresponding curve at fig. 44, and proportions by parts are also marked. Fig. 44. — Explosion of Gaseous Mixtures in a Closed Vessel. Boston Experiments The Clerk method of determii|ing the efficiency was used, and the same period of \ second from explosion and maximum pressure was chosen, so as to make the observations comparable with those of Clerk. As the vessel was of the same capacity, the comparison is very close. Two periods of ^ second were examined — the first timing from the moment of firing, and the second from the moment of maximum pressure. The following table shows the main results, comprising the examn a- tion of the cooling curve to I second from maximum pressure. BOSTON EXPLOSION EXPERIMENTS 151 Explosion in a Closed Vessel. (Boston Experiments) Mixtures of Air and Boston Coal Gas Temperature and Pressure before explosion Atmospheric Mixture Gas. Air Max. pres. lbs. per sq. in. Time of Explosion Area, sq. ins. Mean pres. lbs. per sq. in. Mean pres. gas ratio r 2 3 4 5 6 7 8 Sec I— 3 45 0-49 — 43 172 40 160 I— 4 86 0-08 1-88 62 310 46 230 I— 5 96 0-05 1-93 64 384 44 264 I— 6 88 0-05 [•93 64 448 46 322 I— 7 86 o-o6 1-93 64 512 48 384 I— 8 87 o-o6 1-83 61 549 46 414 I— 9 77 o-o8 1-86 62 620 46 460 I — 10 71 O'll 1-69 56 616 45 495 I — II 68 0-14 1-66 55 660 43 516 I — 12 39 o'33 0-98 33 429 30 390 I— 13 32 0-42 079 26 364 24 336 1—14 9 0'42 0-24 8 120 8 120 The only column which requires a word of explanation is Column 4. Area in square inches signifies the area under the cooling curve from maximum pressure to \ second after, and the mean pressure in Column 5 is taken from this area as determined by planimeter. This is a more accurate method than taking the mean of the maximum pressure and pressure \ second after maximum, as was done by the author. The final pressure after the I second from maximum is given at Column 7. The difference between the methods is not great, as wiU be seen by taking out the means of the second and seventh columns. Column 6 gives the numbers obtained by dividing the mean pressure on Column 5 by the gas ratio. This is what was done by Clerk, as already described. The numbers in Column 6, then, give not only the relative values of the different mixtures for maximum pressure and resistance to cooling in a closed vessel, but they enable Boston gas to be compared with Glasgow, Oldham, and London gases of the author's experiments. The Boston experiments show the best mixture to be i gas and II air, or -^V of gas, as shown by the author's experiments to be true for both Oldham and Glasgow. The comparable numbers for -^ for Oldham and Boston are respectively 630 and 660, so that the gases vary but little in power -producing value. The Boston experiments closely correspond in other results with the author's, as will be discussed later. Interesting experiments have also been made at the same Institution with mixtures of air and petrol vapour, using the same apparatus and the Clerk method of comparison. The two tables on p. 152 are similar to that just described. 152 THE GAS, PETROL, AND OIL ENGINE Explosion in a Closed Vessel. {Boston Experiments) Mixtures of Air and Petrol Vapour. Petrol sp. gr. 0-648 at 86° F. 0*2 sec. after maximum pressure Time of Max. press, in petrol vapour explosion. lbs. per sq. in. above Mean press. Mean press. Final pressure Area, sq. in. lbs. per sq. in. -i- vapour ratio I-5I 0-083 70 1-48 49-4 3260 34 I '64 o-ioo 73 I-S3 51-0 311O 36 179 0-090 71 1-43 477 2670 33 1-96 0083 76 I'SS 51-7 2634 35 2-17 0-058 70 1-45 48-4 2225 30 2-44 0-067 80 1-60 53-4 2190 36 2-56 0-07S 84 1-69 56-4 2200 40 i 2-63 0-0S9 86 171 57-0 2164 38 278 0-083 78 1-62 S4-0 1945 36 3-03 0-091 76 I -60 53'4 1760 38 3-23 0-083 77 1-62 54-0 1675 37 3-45 0-083 77 1-64 547 1587 37 3-85 0-075 66 1-50 50-0 1300 38 4-17 0-066 60 1-38 46-0 1104 35 476 0-066 56 1-32 44-0 925 33 Petrol sp. IX. 0-680 at 76° F. o'2 sec. after maximum pressure Percentage of petrol vapour in mixture Time of explosion. Seconds Max. press. in lbs. per sq. in. above atmosphere Area, sq. in. Mean press, lbs. per sq. in. Mean press. ~- vapour ratio Final pressure 1-32 0-167 52 1-28 427 3240 33 I -41 0-II7 62 1-42 47-3 3360 35 I-5I 0-109 64 1-45 48-6 2950 35 1-64 0-182 51 1-25 417 2540 32 179 0-109 67 1-53 51-0 2855 36 1-96 0-091 73 1-53 51-0 2600 36 2-17 0-082 76 1-56 52-0 2391 37 2-44 0-060 85 1-63 54-3 2225 36 2-63 0-058 85 1-62 54-0 2052 36 278 0-058 84 1-64 547 1970 38 3-03 o-o66 78 1-60 53-4 1760 37 3-23 0-067 83 :1 567 1760 38 3-45 o-ioo 75 53-0 1536 38 3-85 0-117 62 1-42 47-3 1230 35 4-17 0-I33 55 1-40 467 II21 38 476 0-210 35 1-02 34-0 714 32 These experiments show clearly that the best mixture of petrol and air with petrol of 0"648 sp. gr. is i'5i per cent, of petrol vapour in the air mixture and with petrol of o-68o sp. gr. 1-41 per cent. The mean pressure measuring the resistance to cooling and best pressure GROVER'S EXPLOSION EXPERIMENTS 153 is in the first case 49-4 lbs., and in the second 47'3 lbs. per sq. in. ; while the best mean pressure with Boston gas is 55 lbs. per sq. in. According to this comparison a petrol engine using these petrol samples should give less power than a gas engine of the same volume swept by the cylinder. As a rule it is found that petrol engines give rather higher mean pressures than gas engines. Some of the conditions in the petrol experiments appear to prevent a true comparison. Nevertheless the experiments are interesting and valuable, as but few explosion experi- ments have been made with petrol as yet. Grover's Experiments Interesting experiments were made by Mr. F. Grover in 1895 at the Yorkshire College of Science, Leeds. His explosion chamber was of I cub. ft. capacity, cylindrical, and internal dimensions about 8 ins. diameter, 34 ins. length. Mr. Grover states his object as follows : ' Experiments previously made upon gaseous mixtures have been directed towards the investigation of the actual pressures produced by the combustion of an inflammable gas, in the presence of oxygen or pure air only.' . . . ' The most complete practical contribution upon this subject has been afforded by the experiments of Mr. Dugald Clerk, which enabled him to estimate the most economical mixture to be used in a non-compression engine, but no account was taken of the effects of the products of combustion which are preserved in the cylinder of a gas engine, notwithstanding that early engines were constructed with a clearance volume of 60 per cent. To obtain some definite data upon this important subject the author (Mr. Grover) has carried out a series of experiments in the engineering laboratory of the York- shire College, Leeds.' Mr. Grover states that products of combustion when mixed with fresh charge have been generally supposed to reduce the pressure pro- duced by explosion, and therefore reduce the efficiency of the charge. Mr. Grover's experiments, however, lead him to the opposite con- clusion. He says : ' The experiments carried out by the author show that the presence of the products of combustion in certain mixtures actually raise rather than diminish the maximum pressure obtained.' This is a most important conclusion, if correct. But is it correct ? Mr. Grover's apparatus consisted of a thick cast-iron cylinder, flanged at both ends, as has been already stated, of i cubic foot capacity. The cylinder was bolted vertically to a column ; the charge was ignited by passing an electric spark at the upper part of the vessel. Temperature before ignition was measured by a thermometer enclosed in a wrought-iron case containing mercury, inserted into the gas 154 THE GAS, PETROL, AND OIL ENGINE chamber. The pressure changes were recorded by a Crosby indi- cator, the pencil of which was arranged to inscribe upon a con- tinuously revolving drum, 8 ins. in diameter, driven by clockwork. The speed of the drum was checked by a vibrating spring adjusted to make four complete oscillations per second. The recording apparatus is shown diagrammatically at fig. 45. In all the experiments Mr. Grover states that the volumes were measured by filling the cylinder with water and afterwards allowing the gas to enter as the water flowed out. The products of a previous combustion were retained to the desired extent in order to mix in the required proportions with the fresh charge. In some experiments the charge experimented upon consisted of gas and pure air only. In making the pure air experiments, the explosion vessel was filled with water to expel the products, the water was then run out to draw into the vessel half the total volume of pure air required, then the coal Fig. 45. — Grover Recording Apparatus gas was admitted, and lastly the remaining half of the air required to complete the charge. Mr. Grover states that in all experiments, ' no appreciable time was allowed for the diffusion of the mixture, it having been fired immediately after taking the temperature.' The coal gas used was taken from the service pipes of the Leeds gas- supply : the mean of three samples taken after the experiments were made gave the following analysis : Analysis of Leeds Coal Gas. (Grover) Constituents Volume per cent. Marsh gas 35'2 Olefines 4'2 Hydrogen 52-9 Carbon monoxide 6*5 Nitrogen o'l Carbon dioxide and oxygen i-i lOO'O GROVER'S EXPLOSION EXPERIMENTS 155 Mr. Grover, from air experiments, gave the results collected by the author in the following table : Explosion in a Closed Vessel Pressure and Temperature before explosion. . . Atmospheric Mixture Maximum pressure above atmospliere in lbs. per sq. in. Gas Air Grover, 1895 Clerk, ■ Oldham Gas Clerk, London Gas Massachusetts Inst. Boston Gas IS 14 13 12 II 10 9 8 7 6 lbs. 16 24 31 36 48 62 62 40 51-5 60 78 87 90 60 65 17 80 85 9 32 39 68 71 77 87 86 88 It will be noted that the explosion pressures produced in these experiments of Mr. Grover are much below those shown for similar mixtures by the experiments of Clerk and the Massachusetts Institute of Technology, which shall be called shortly the Boston experiments. To enable the comparison to be easily made the explosion pressures shown for similar mixtures of Oldham, London, and Boston coal gas are given in the last three columns. It wiU be observed that Grover's experiments agree with the Boston experiments ; for the mixtures con- taining jV and xV of gas the maximum pressures are very nearly the same, but the mixtures rV) ij ^"^^ t give explosion pressures 30 per cent, lower. Compared with the Clerk Oldham gas experiments, mixtures tV' T4) a-iid iV ^e 40 per cent, below, and mixtures t^J^j Jj and |_are 30 per cent, below, as in the Boston experiments. The three sets of experiments are compared by the curves shown at fig. 46, where explosion pressures in pounds per square inch above atmosphere are shown on the vertical scale, and the percentage of gas in total volume of mixture is shown on the horizontal scale. The curves show considerable discrepancies. The Clerk curve for Oldham gas, it will be observed, smoothly passes through aU the experimental points, and the Grover curve passes smoothly through all the points except the^- mixture; neglecting that point, the curve is a smooth one, resembling the Clerk curve, but at a lower pressure. The Boston curve is very irregular, and crosses both Clerk and Grover curves, but in its last six observations it agrees substantially with the Clerk curve. All these samples of coal gas have a lower calorific value of 156 THE GAS, PETROL, AND OIL ENGINE nearly 600 B.Th.U. per cubic foot, so that they should show sub- stantially the same maximum pressure for the same mixture. The Clerk curve for London gas is slightly below that for Oldham gas, the calorific value for London gas being a little lower. Why are the results to some extent divergent ? Mr. Grover's view is that his explosion pressures are lower than the author's because of his method of charging his cylinder by means of water, which leaves the whole of the interior surface in a moist condition. No doubt this explains some of the deficit ; the water film would exercise some cooling effect, but this could not produce anything approaching 30 per cent, or 40 per cent. The author's experience convinces him that maximum pressure is but little checked in this way except in the case of the weakest mixtures. This water film explanation cannot account for CLE W,0 .DHA » CA i BOS XOH~ :^ 1 ^^ 1 — ■ "^ -^ ^ ^ t^ ^ i-— -^ -CLE RK,L ONDO V ca: ^ / CRO fER,i 195, / ■^ ^ ^ / J ^ "^ 1 f^ ^ <; f / / PEHCENTACL OF CAS IH TOTAL VOLUME OF MIXTURE Fig. 46.— Comparison of Grover, Clerk, and Boston Experiments the departure of the lower three Boston experiments from Clerk's curve. In the author's view such differences arise entirely from imperfect mixing of the gas and air contained in the explosion vessel. It was found by him in the course of the 1885 experiments that unless gas and air be well mixed low maximun?gpressures would be found. It was also found that weaker mixtures could be fired when the mixture was not uniform, provided some strong mixture was situated near the igniting point. In the author's early experiments he care- fully experimented in order to determine this mixture point. In a paper ' On the Explosion of Homogeneous Gaseous Mixtures,' Inst.C.E., 1886, he states : ' To charge the explosion vessel, it is exhausted by an air pump, and a measured quantity of the inflam- mable gas admitted from a graduated glass measuring-vessel ; air is GROVER'S EXPLOSION EXPERIMENTS 157 then let in, bringing the pressure up to the atmosphere again. The inflammable gas, being admitted while the vessel is almost vacuous, diffuses throughout the whole space, and the air entering afterwards, the contents are mixed thoroughly. To make perfectly sure that mixture is complete, the vessel is allowed to stand for at least half an hour before the explosion. To check the results obtained in this manner, a mixture of gas and air in a separate gasometer was made, and introduced into the explosion vessel by repeated exhaustion and filling ; the results were precisely similar, and the first method was used in most of the experiments, being more accurate and much safer.' Clerk's London gas experiments were performed by the sepa- rate gasometer method, and here the effectiveness of mixing was undeniable. The London gas curves show the same higher pressures of explosion and give smooth curves. The method of charging used in the Boston experiments, in which the gas added was measured in by reducing the pressure within the vessel and then letting gas flow in to fill up to atmosphere, would permit of very indifferent mixing, especially at the weak mixtures, where the pressure differences were small. Where this difference was greater better mixing would take place. The lower pressure obtained in weak mixtures the author considers to be entirely due to bad mixing. Again, in both Boston and Grover experiments weaker mixtures were found to be ignitable than was found by the author, and this also points to bad mixing. Mr. Grover's experiments with exhaust gas admixtures were made as follows : (i) Water was admitted to the cylinder to discharge part of the products of a previous combustion. (2) Water was run out, so as to aspii-ate inwards half the volume of pure air required for the experiment. (3) Air-cock was shut, and coal gas was drawn in by the same method. (4) Water was entirely run out from the cylinder and the remaining air drawn in. (5) The mixture of gas, air, and products of previous combustion were ignited at once by the electric spark and the time-pressure diagram taken. Experiments made in this way showed that for equal dilutions made with air whoUy and products of combustion and air, higher explosion pressures were recorded when products of combustion were present. That is, assume the explosion vessel to contain ^^, y^j tV> tV' I'r of its total volume of coal gas by measurement than when products of a previous combustion formed part of the |f, |f, ff , yf , or |f forming the other contents, higher explosion pressures were obtained than when 158 THE GAS, PETROL, AND OIL ENGINE pure air only was used. The increase of -pressure obtained, however, became less and less as the proportion of gas present increased, and when mixtures containing -^ and ^ of gas were tested no increase was found, but a diminution instead. Mr. Grover's air and products of combustion mixtures gave results of which the most important are shown in the following table : Explosion in a Closed Vessel. {Grover 1895) Mixtures of Leeds gas, air, and products of cotabustion. Pressure before explosion. . . Atmospheric. Mixture - Leeds Gas Air Products of combus- tion Total Max. pressure lbs. per sq in. above atmo- sphere - Frac- tion 1^0 Per Cent. Per Cent. Per Cent. 6-2 6-2 6-2 6-2 93-8 88-8 68-8 63-8 none s 25 30 100 100 100 100 16 Maximum rise of pressure above that of pure mix. ture, 35 — 16 = 19 lbs. per sq. in. 22 34 35 6-6 6-6 6-6 6-6 93-4 88-4 68-4 63-4 none 5 25 30 100 100 100 100 24 Maximum rise of pressure above that obtained with pure mixtures, 36 — 24 = 12 lbs. per sq. in. 24 28 36 yi 7-1 7-1 7-1 7-1 92-9 87-9 67-9 62-9 57-9 none 5 25 30 35 100 100 100 100 100 31 27 30 34 37 Maximum rise of pressure above that obtained with pure mixtures. 37 — 31 = 6 lbs. per sq. in. 77 T7 T7 92-3 77-3 67-3 none IS 25 100 100 100 36 43 - 36 = 7 lbs. 42 43 The maximum increase found witl||products of combustion mix- tures was in the case of the weakest mixture, and the increase was 19 lbs. The increase diminished as the proportion of coal gas added ncreased, so that at yV coal gas the increase was only 7 lbs. This change is clearly shown below : Coal gas present .... Increase of pressure above corre- sponding gas and air mixture II \ 19 lbs. 12 lbs. 6 lbs. 7 lbs. 2 lbs. GROVER'S EXPLOSION EXPERIMENTS 159 The two mixtures containing j- and 7 of gas showed reduction. Coal gas present ... J } Deficit of pressure compared 1 with gas and air mixture . f 7 ^- 3 s. The cause of these effects is obvious enough. When a weak mixture, such as that shown in Clerk's experiments with London gas containing yV of coal gas and |f air, is ignited, the pressure only rises some 5 lbs. per sq. in., and although there is ample gas present if burned to pro- duce a much higher pressure, yet the gas does not complete its burning — on the contrary, the flame flickers out. If the products of com- bustion produced from such a mixture be analysed, some 30 per cent. or more of the gas originally put in is found to be stiU present as inflammable gas. This is found to be the case where the gas was care- fully mixed with the air in a separate gasholder, as was done in the Clerk experiments, and where gas is imperfectly mixed a stiU larger portion may escape combustion. If, then, to products of combustion containing a portion of inflammable gas be added a fresh portion of coal gas and a further supply of air, the mixture contains more gas than that calculated from the added part. This renders it more inflam- mable, and of course produces a higher pressure. Consider, first, Grover's experiment with ^i^ coal gas, or 6'2percent., and 93-8 per cent, of air ; the explosion pressure attained was only 16 lbs. per sq. in. above atmosphere. When the mixture consisted of the same percentage (6'2 per cent.) of added coal gas, 63'8 per cent, of air, and 30 per cent, of products of combustion, then a pressure of 35 lbs. was obtained upon explosion. If the 30 per cent, of products contained enough gas, unburned, to raise the total percentage of com- bustible present from 6'2 per cent, to 7 per cent., then, experimenting in Mr. Grover's manner, this pressure would be obtained. The pressure of 35 lbs. would be easily obtained, even allowing for some defects in mixing, as is shown by Clerk's experiments. The portion of the original gas present in the 30 per cent, would be 6-2 X 0-3 = 1-9 nearly, and 0-8 added to 7-2 would produce the neces- sary 7 per cent., so that this phenomenon is easily explained. As the mixture gets richer and richer in gas less and less gas remains unburned in the combustion products formed till the mixture containing i\- of gas is arrived at, when practically no further unburned gas remains to enrich the added fresh charge in inflammable material. The author accordingly concludes that it is incorrect to consider that the products of combustion — that is, the mixture of nitrogen, car- bonic acid, water vapour, and oxygen — in any way raise the pressures of gaseous explosion or act in any way different from that of a diluent, except in cases when they contain inflammable material remaining from the previous charge. i6o THE GAS, PETROL, AND OIL ENGINE The author feels it necessary to discuss Mr. Grover's results at some length, as many fallacious explanations have appeared of the action of the exhaust gases, proceeding on the assumption that- the gases are inert and uninflammable. Grover's Later Experiments Mr. Grover's later experiments on acetylene and air explosions are of a more useful and important kind. They were made between 189& Fig. 47. — Grover's later Explosion Apparatus and 1901, and the apparatus used c|w:rected many of the errors of his earlier devices. The acetylene was accurately measured from a separate gasholder, and so the introduction of water into the explosion cylinder was avoided. Mr. Grover's later apparatus is shown at fig. 47. It consisted of a vertical explosion vessel a, with covers bolted top and bottom. A rotating wing b was provided, operated by an outside handle on a properly packed spindle. This rotating wing served the double purpose of stirring the mixture first and then igniting by low-tension current. To cause ignition an internal contact breaker i GROVER'S ACETYLENE EXPERIMENTS i6i was coupled in series with a glow-lamp c to the electric light leads of the building. The glow of the lamp clearly indicated that a contact had been made within the explosion vessel, and, by quickly rotating the wing b, which had made contact at D, the circuit was interrupted TIME . SECS. Fig. 48. — Diagrams of Explosion Various mixtures of acetylene and air ignited at atmospheric pressure. { Grmier) and a powerful low-tension spark produced within. A Crosby indicator E was used for determining the pressure, and its pencil was held by a light spring to the revolving drum F driven by clockwork. V ( ^ y^ r- ^ ^ ■^ lA ^ k-c Kl TYLl NE & AIR '^ ^ > -^ > 5 -^ 5°° < >--*" ( >-^^ CO. \L-C, s s. MR ~---^ ^ An "> ^., s ■'° ~"--< )^^ 31 ^■\ ) 0- 1 RATIO OF AIR TO CAS BY VOLUME. Fig. 49. — Explosion in a Closed Vessel Maximum pressures recorded when exploding mixtures of acetylene and air, also coal gas and air. Initial pressure, I atmosphere ; initial temperature, 0° C. ( Gj'over) Mr. Grover timed the revolutions of this drum in an ingenious and effective manner. A watch G was mounted vertically on a small spur- wheel geared to the worm h rotated with the drum, and the arrange- ment rotated the watch contra-clockwise, so that at a certain speed VOL. I. M i62 THE GAS, PETROL, AND OIL ENGINE the centre-second hand of the watch became stationary. A small mirror was fixed at the centre of the hand, and by observing a distant object as reflected from its surface it was easy to see when the hand was quite stationary. By this device the rotation of the drum could be adjusted with great accuracy to a constant speed for the few seconds of the experiment. Three sets of experiments were made ; first, at atmospheric pressure ; second, at two atmospheres pressure absolute ; and third, at three atmospheres pressure absolute. Diagrams of acetylene explosions at atmospheric pressure are shown at fig. 48. The proportion of air to acetylene is marked on each. Temperature scales have been added to Mr. Grover's diagrams. Fig. 49 shows observations of maximum pressures plotted against the ratio of air to gas and coal gas and air mixtures shown as well. 250- r ^ 9 to \ -zooog S -1500 g b -1000 „■ -500 1 -0 H i n *tol- '-/ / --,j . ( < » ' » |)-^~-. > ^ ) AC ■TYU Wi AIR -1500 = V) a: -1000 g a: -500 £ £ 111 S ■^ ^\ bJ ! ? ? i L ^ ( ' r > V c I ■ a. \ -^ 5 < '^-J ) COAk -CAS &AI ? ^K '^-^ ~^. a ~~*^' 0- -0 RATIO OF AIR TO CAS BY V OLUME. Fig. 51. — Explosion in a Closed Vessel Maximum pressures recorded when exploding mixtures of acetylene and air, also coal gas and air. Initial pressure, 2 atmospheres ; initial temperature, o'' C. {G70ver) that some air had got into the gasholder from which he measured his acetylene; so that, instead of measuring pure acetylene from his gas- holder, he was in reality sending in a mixture of acetylene and air containing from 6 per cent, to 20 per cent, of air. He considers that 5 per cent, of air is probably present in all the mixtures exploded at an initial pressure of one atmosphere, and probably more at the higher pressures, so that the mixture plotted at fig. 49 as 15 air to i gas should in reality be i^y air to i of gas. At initial pressure of two atmospheres the weakest mixture which could be fired was 21 gas i air as measured by the gasholder. Assuming 10 per cent, of air in the gasholder, Mr. Grover gives the true mixture as 23'i to I. Four mixtures of coal gas in the proportion of 8 to i and 11 to i were fired at the same initial pressures, and the maximum pressure of £64 THE GAS, PETROL, AND OIL ENGINE the acetylene explosions was found to be from 1-5 to 27 as great as with corresponding mixtures of coal gas. Diagrams of acetylene explosions at three atmospheres initial J3ressure are shown at fig. 52, arranged as already described with reference to the two preceding sets of experiments. CO 360-1 / -vct: 1 — 1 1 — 1 — . p-_ l~~1 ___ ^_^ 1 ^,„ 7tol v: ■12- >to - -zooo^ 111 > s z d (n /I r / ^ I7t tol -ISOO u ^ / / ^ -< 19* tol -1000 °\ •1 p ^ ^ — — ' 30 0/ -500 S ft: u ' ■ ° "1 Fig. 52.— -Explosion Diagrams Oi various mixtures of acetylene and air ignited at 3 atmospheres initial pressure ; initial temperature, 0° C. {Graver) Fig. 53 shows observations of maximum pressure plotted against the ratio of air to gas, also coal gas and air mixtures. X w^ - ~~^ " . AC ■TY .EN ■ & AIR ~ i_ — _ '-, COi L-( •AS &A R _/_ -- - ft- II tZ 14 16 18 20 22 24 RATIO OF AIR TO CAS BY VOLUME. Fig. 53. — Explosion in a Closed Vessel Maximum pressures recorded when exploding mixtures of acetylene and air, also coal gas and air. Initial pressure, 3 atmospheres ; initial temperature, 0° C. {Gj-over) Mr. Grover's table of pressures is given below ; Mixtures of Acetylene and Air Exploded at Two Atmospherls Pressure. (Crovcr.) Initial Temperature, o'' C. Proportion of air to gas Maximum, pressu above atmosp Time of explosion f air 30 22 21 19*6 i7'5 i6'g i6'i 14*7 12* 3 i2'i 11^ 1 gas iiii 1 I I 1 1 I I ''^Xv"aSpTJ"."^''"^''■'"■| '46.97 -7 -x .46 .36 .59 ^6: 308 307 3=5 o'i5 o'o6 In these experiments the strongest mixture fired was ii-y air to I gas, and the weakest 30 to i. Stronger mixtures were not used, GROVER'S ACETYLENE EXPERIMENTS 165 because the limit of safety of the explosion vessel was being too closely approached. Mr. Grover points out the great rapidity of explosion of acetylene mixtures as compared with coal gas, and notes that the inertia of the parts of the indicator would cause a material lag, so that the more rapid explosions are more liable to error in the time of explosion. All the experiments at initial pressures higher than atmosphere were conducted as follows : The gas required for combustion was first measured at atmospheric pressure and then driven over into the cylinder ; all cocks were then closed, and compressed air was passed into the cylinder until the pressure gauge showed 15 or 30 lbs. as required. The mixture was then stirred and allowed to stand for ten minutes before ignition was attempted. Slight leakage was expe- rienced, but it was allowed for. As the difficulty of getting pure acetylene gas was found to be great, Mr. Grover determined the proportion of acetylene to air, by analysis of the products of combustion of the three-atmosphere explo- sions, and he believes these experiments to be more accurate than those at two atmospheres. Grover gives the following table of analyses of the products from the combustion of acetylene and air exploded at three atmospheres : Analyses of Products of Combustion of Air and Acetylene Mixtures. (Grover) Mixtures Constituents by volume per cent. Air Totals CO, CO N Steam. H,0 6-5 Acetylene 117 I I3'0 3-2 o-o 79 -o IOI-7 12-3 I 15-1 O'O o-o 78-0 8-2 IOI-3 I4-S I 12-4 o-o 2-5 78-9 6-2 lOO-I 16 I II-8 o-o 4-2 79-0 5-9 100-9 17-5 I lO-O O'O 5-1 79-6 5-0 997 21 I 8-6 0-0 8-1 78-9 4-3 99-9 22 I 9-0 O-o 8-6 79 -o 4-5 lOI-I 30 I 6-8 o-o 12-0 78-5 3-5 100-7 The column headed ' Mixtures ' has been calculated by Mr. Grover i66 THE GAS, PETROL, AND OIL ENGINE from the analysis, and the steam has been calculated also from the CO.;, and in one case CO. Mr. Grover gives the following values, which are useful in making calculations as to acetylene explosions : Calorific value (C2H2) acetylene. Lower value 1504 B.Th.U. (C2H2) „ Higher value 1558 „ for I cub. ft. at 0° C. and 147 lbs. per sq. in. absolute. I cub. ft. of acetylene (C2H2) ato°C. and 147 lbs. weighs 0-0725 lbs. I cub. ft. of acetylene (C2H2) requires I2'5 cub. ft. of air for complete combustion. Following Clerk's method of determining best mixture as to maxi- mum pressures produced, Mr. Grover gives the numbers below : Best Mixtures for Maximum Pressure — Acetylene and Air — AT I, 2, AND 3 Atmospheres Initial. (Grover) Air Gas Initial pressure i atmosphere . . Best mixture 13 i .. ,. 2 ,, . . ,, 15 I ,, ,. 3 .. ... „ 27 I Mr. Grover has calculated out the efficiencies of combustion for different acetylene mixtures, using the following specific heat values for mean specific heats from 0° to about 2000° C, as given by Mallard and Le Chatelier. Steam (HjO) = o-68 C,. Carbonic acid (COj) = 0'3o8 C^. Nitrogen (N) = 0-215 C^. He takes oxygen also at 0-308. Using these numbers he determines the efficiency of combustion as from 47 per cent, to 73 per cent., and he states that these efficiency values are higher for acetylene than has ever been noted for coal gas. With this deduction the author does not agree. Mr. Grover's experiments are very valuable for discussion, as no other acetylene experiments are available, and they are the more valuable because of Mr. Grover's caftdid and praiseworthy admissions and details of his difficulties. Petavel's Experiments Dr. J. E. Petavel, F.R.S., has made many interesting experiments on mixtures of coal gas and air compressed before ignition to pressures of over 1000 lbs. per sq. in., and resulting in explosion pressures of some 9000 lbs. per sq. in. To stand such pressures he has devised an apparatus of a very ingenious kind. It consists of a strong PETAVEL'S EXPERIMENTS 167 spherical steel bomb of 4 ins. internal diameter, which has therefore a capacity of 55i'9 cub. cm. = 0-0195 cub. ft. The indicator used is of a novel construction, and is thus described by Petavel in a paper published in the ' Philosophical Magazine ' in May 1902. Referring to fig. 54 he states : ' A cylindrical groove is cut half through the walls of the inclosure. The upper part p of the cylinder thus obtained represents the piston of our indicator, and the lower part s the spring. Under the pressure of the explosion the piston p will be forced outwards, a certain small amount corresponding to the elastic compression of the material of which the spring is made. This motion is transmitted to the exterior by the rod r. Fig. 54. —Diagrammatic Representation of the Petavel Recording Manometer ' The lever l supporting the mirror rests on the fulcrum F at " 3 " ; it is kept against the knife-edge " 2 " of R by the tension of the wire w. The wire w is of considerable length, and it is stretched to near its limit of elasticity. The lever l can therefore follow the small advance of the rod R without greatly diminishing the tension of the wire w. ' The mirror focusses a point source of light on to a rapidly revolving cylinder, thus recording on a magnified scale the motion of the piston p. It is not impossible that an indicator of this type would work in practice ; but the deflexion of the mirror, and therefore the scale of the records obtained, would be much too small. To increase the deflexions three modifications are necessary : the spring s must be made longer, the ratio of its cross-sectional area to that of the i68 THE GAS, PETROL, AND OIL ENGINE piston must be decreased, and the knife-edges ''2" and "3" brought closer together. ' In fig. 55 the design of the actual instrument is given, the lettering being the same as in the previous figure. ' By means of the thread u the gauge screws into the explosion chamber c being flush with the inside surface ; an air-tight joint is formed by the ring d pressing against a flat steel ledge. ^ The end of the gauge from D to E is a good fit in the walls of the explosion chamber, and the joint is thus protected from the direct effect of the explosion. ' The spring s, about 5 ins. in length, is tubular in shape. To nn nn TTi w vv LiJ Lfd I 13456789 10 „ Fig. 55.— Longitudinal Sections and Elevation of the Petavel Recording Manometer prevent any buckling it is made to closely fit the cylinder in which it is contained at two places, e, and e.2. The spring is fixed at the outer end z, being held in place by the nut k ; at the inner end it is free and supports the piston p. The ordinary U -leather is replaced by a leather washer attached to the piston by the screw C and to the fixed part of the gauge by the rim E. The end of the piston projects about an ' In the case of apparatus designed for gases under high pressures all joints should be made directly metal to metal, no packing being used. A joint thus made, if properly designed, is and remains absolutely air-tight. It can be made or broken in an instant, .and as many times as may be required. PETAVEL'S EXPERIMENTS 169 hundredth of an inch above the rim H, and it can therefore move back without straining the leather. ' Tlie mirror (not visible in the figure) is carried by the lever L. This lever is so designed that the knife-edges i, 2, and 3 are in the same plane, it being at the same time possible to bring the knife- TIME. SECS. Fig. 56. — Rise of Pressure during Explosion. (Petavet) Spherical enclosure capacity, 55i'9 c.c. Temperature of enclosure : before firing, 18° C. ; after firing, 24° C. Initial pressure, 77*3 atmos. (ii"36 lbs. per sq. in.). Maximum explosion pressure. Air _ ,;._ j^^jj^ Maximum pressure _ g.^g^ 646 atmos. (9,508 lbs. persq. in.). Ratio Gas Initial pressure edges 2 and 3 within one-hundredth of an inch of each other should so great amplification be found necessary. Up to the present, however, the distance has not been decreased below yV in-, the scale obtained with this distance being found satisfactory.' Fig. 57.— Fall of Pressure after Explosion. (Petavel) Spherical enclosure capacity, 551'g c.c. Temperature of enclosure : before firing, 21° C. ; after firing, 27° C. Initial pressure, 74 '38 atmos. (1,094 lbs. per sq. in.). Maximum explosion pressure, 6s4atmos.(9,6.81bs. persq. in.). Ratio ^ = s^r- Ratio Mj'™H51Pr555Hi = 8 -8. Gas Initial pressure The chronograph used is very simple ; it consists of a rotating drum carrying a photographic film. The drum is rapidly rotated by an electric motor, and it is enclosed within a light-tight box having a long, narrow slit (yj- in. wide) ^running its full length parallel to the 170 THE GAS, PETROL, AND OIL ENGINE axis of rotation. One of the filaments of an incandescent lamp is focussed on the slit at right angles to it, so as to form a sharp point of light which moves along the slit with increase of pressure. Figs. 56 and 57 show typical records obtained by Dr. Petavel from mixtures of air and coal gas fired at an initial pressure of about iioo lbs. per sq. in. Oxygen was in excess, as the residual gases contained about 3 per cent. No part of the instrument except the mirror frame moves more than one or two thousandths of an inch, so that inertia troubles are avoided. Thus, although in fig. 56 the pressure at 0'055 of a second after ignition is rising at the rate of over 1,000,000 lbs. per sq. in. per second, the curve turns sharply at a right angle and shows no sign of vibration at maximum pressure. Petavel gives the following table with regard to the explosion curve, fig. 56 : Rise of Pressure. (See fig. 56) Spherical inclosure 4 ins. diameter. Capacity 55i'9i c.c. = O'Oigs cub. ft. Temperature of inclosure before firing, 18° C. ,, ,, after products of combustion had cooled, 24° C. T ... , (77"28 atmospheres Initial pressure . . . . \" r ^^ ^ 1= 1,130 lbs. per sq. m. Ratio : Air to Coal-gas . . 6'0 ,, . 1 . ( 646-2 atmospheres Maximum explosive pressure . \ ^ on ^ ^ == 9,508 lbs. per sq. m. Ratio : maximum explosive pressure to initial pressure 8-4. „ ., , 63-8 atmospheres Residual pressure . . . . -^ „ ^ ^ »^ = 937 Ids. per sq. m. Analysts of Residue. Carbon dioxide 9'8 per cent. Oxygen 3-0 Nitrogen 87-2 „ loo-o PETAVEL'S EXPERIMENTS 171 Absolute Absolute Time in Reading In Pressure Time in Reading in Pressure seconds millimetres in lbs. per sq. in. seconds millimetres in lbs._ per sq in. o-ooo — II36 0-058 21-99 9508 O'OIO 0-77 1237 o-o6o 21-91 9477 0-020 I-S7 1549 0-062 21-91 9477 0-030 3-06 2130 0-064 21-67 9385 0-040 5-03 2898 0-066 21-63 9369 0-042 5-21 2968 0-068 2 I -60 9357 0-044 5-84 3204 0-070 21-56 9340 0-046 6-38 3424 0-080 21-22 9209 0-048 6-86 3612 0-090 21-07 9149 0-050 7-47 3850 o-ioo 20-87 9071 0-052 io-8o 5147 0-150 1973 8628 0-054 15-92 7143 0-200 18-61 8192 0-056 19-83 8667 And this table with regard to the cooling curve, fig. 57 : Fall of Pressure. (See fig. 57) Spherical inclosure 4 ins. diameter. Capacity SSi'Qi c.c. = 0-0195 cub. ft. Temperature of inclosure before firing, 21° C. ,, ,, after products of combustion had cooled, 27° C. Initial pressure Ratio : Air to Coal-gas . Maximum explosive pressure Maximum temperature . Ratio : maximum explosive | pressure to initial pressure) Residual pressure (74'38 atmospheres 1 = 1,094 lbs. per sq. in. 57I- (654-2 atmospheres 1 = 9,618 lbs. per sq. in. 2483° C. 58'98 atmospheres = 867 lbs. per sq. in. Analysis of Residue Carbon dioxide 11 "2 per cent. Oxygen 2-0 ,, Nitrogen 86-8 lOQ-o 172 THE GAS, PETROL, AND OIL ENGINE Absolute Tempera- Absolute Tempera- Time in Reading Pressure ture of Gas Time in Reading Pressure ture of Gas seconds in millims. in lbs. in degrees seconds in millims. in lbs. in degrees per sq. in. Centigrade * per sq in. Centigrade * o-o8 23-54 9618 2483 2-2 5-68 2979 S8i o-i 23-19 9487 2445 2-3 5-54 2927 566 0-2 20-55 8506 2164 2-4 5-32 2845 542 0-3 18-62 7789 1957 2-5 5-13 2774 522 0-4 16-95 7167 1781 2-6 5-00 2726 508 0-5 15-83 6751 1 661 27 4-83 2663 490 0-6 14-78 6361 1548 2-8 4-68 2607 474 07 13-79 5993 1445 2-9 4-51 2544 456 0-8 12-82 5632 1 341 3-0 4-35 2484 439 0-9 11-92 5298 1245 3-5 3-82 2287 382 i-o 11-08 4986 1156 4-0 3-33 2094 327 I'l 10-38 4726 1081 4-5 2-96 1968 291 I "2 972 4481 lOII 5-0 272 1878 265 1-3 9-20 4287 955 6-0 2-47 1785 239 I '4 8-68 4094 900 7-0 2-16 1670 205 i-S 8-16 390I 845 8-0 1-95 1592 183 1-6 7-68 3722 793 9-0 1-74 1514 161 17 7-30 3581 753 lo-o 1-54 1439 139 1-8 6-86 3418 706 ii-o 1-43 1399 128 1-9 6-54 3299 672 . I2-0 1-32 1358 106 2-0 6-23 3183 639 15-0 1-15 1295 98 2'I 5-95 3079 609 * The temperatures in this column are calculated in the usual manner from the pressure, allowing for the quantity of water-vapour formed. From the last table he has prepared the cooling curve shown at fig- 58. Petavel calls attention to the three following points shown by his experiments : ' I. The time required to reach the maximum pressure, namely, 0-058 second, is not far from that which would be required with the same mixture at atmospheric pressure. ' 2. The ratio of explosive to initial pressure has been increased. At or near atmospheric pressure the ratio for this mixture would be about 7 ; in the present case it is 8-6. This fact is due to three causes which work simultaneously, namely : (A) The departure of gases from Boyle's law ; (B) the relative decrease of thermal loss during the time occupied by the combustion ;%(C) the increase in the absolute temperature at which dissociation would take place. ' 3. The rate of cooling has greatly decreased.' Petavel also states : ' The quantity of heat dissipated per unit of cooling surface increases with the temperature interval and with the pressure of the gas, but not at the same rate as the latter. The heat developed, on the other hand, is simply proportional to the pressure. ' By increasing the pressure from i to 70 atmospheres we increase PETAVEL'S EXPERIMENTS 173 the heat generated in a given volume 70 times, but we do not increase the rate at which heat is dissipated in anything like the same ratio.' Royal College of Science Experiments These experiments were proposed by Professor Perry, and the main part of the apparatus was designed by Messrs. M'Diarmid and Mann, students of the Royal College of Science, South Kensington. Messrs. Leonard Bairstow, A.R.C.Sc, and A. D. Alexander, A.R.C.Sc, made \ ■ \ \ \ \ \ ^ - — ■ 0- TIME,SECS Fig. 58. — Coal Gas and Air. Rate of Cooling. {Petavel) Initial pressure . 1,094 lbs. Explosion pressure . . g,6i8 lbs. Ratio ^ .... 57 Gas ■p .. Explosion pressure g.g Initial pressure the experiments with the sanction and encouragement of Professor Perry. The work occupied two years continuously, and a paper was written by Messrs. Bairstow and Alexander, which was read at the Royal Society in 1905. Only part of the paper was published by the Royal Society, the part which was considered most interesting from the purely scientific standpoint, and the present author is accordingly indebted to Messrs. Bairstow and Alexander and the Royal Society for a copy of the complete paper with its numerous tables illustrating many experimental points. 174 THE GAS, PETROL, AND OIL ENGINE Much of the work is of moment to the engineer, and the author has extracted the parts of the greatest importance. The apparatus used is illustrated diagrammatically at fig. 59. A is an indicator of the Simplex type, mounted upon an explosion vessel and having a rotating drum operated mechanically and connected to a striker b^ which served to enable the apparatus to be timed to the beats of a metronome, so that in all experiments 42-5 ins. of diagram represented one second in time ; c is the driving cord which served to actuate the striker and drum ; d is a Bourdon gauge, which indicated the initial pressure before explosion ; E a temperature plug contain- ing mercury to get the initial temperature of the apparatus before firing ; F is a firing tube, to be described more fully later ; g is an KiG. 59. — Diagrammatic View of Royal College of Science Explosion Apparatus indicator tube, which also serves as a charging tube ; H is a perforated stirring plate, operated from the exterior by a rod and handle shown, used to mix the gases thoroughly before firing ; j, k is the gas- measuring device, to enable the gas to be measured at atmospheric pressure and discharged into the explosion vessel under pressure ; L is a mercury gauge ; M is a spa* ing key ; N the induction coil, and O a sparking plug. The explosion chamber is 18 ins. long by 10 ins. internal diameter, and its capacity is 1419 cub. ins. It is of cast iron, and strong enough to stand 1000 lbs. per sq. in. A pump p supplies air to an air reservoir r, and this air is admitted at the desired pressures to the explosion chamber by way of the indicator cock and suitable pipe connections. BAIRSTOW AND ALEXANDER'S EXPERIMENTS 175 It was found at an early stage of the experiments that mixing of the gas and air required special attention ; both maximum pressure and time of explosion varied greatly for identical proportions of gas, in accordance with the perfection or imperfection of mixing. Thus, with a mixture of 0"225 gas to i volume of air, the maximum pressure varied 30 per cent, when the gas was admitted to the air at the top of the cylinder, depending on the rate at which it was caused to flow into the air. In one experiment the gases were left for seventeen hours, in the hope that gaseous diffusion would produce a uniform mixture, but even then ignition could not be produced by a spark passed at the top of the cylinder ; this proved that the mixture then was too rich for ignition and had not diffused throughout the vessel. By adding air rapidly to the gases, however, very consistent results were obtained. It was found necessary to add the mixing plate h (fig. 59), and the method of charging used for the experiments was as foUows : The explosion cylinder was put into connection with the suction pump and the mercury gauge, and the pressure reduced below atmosphere to a known amount as measured by the mercury gauge ; the suction pump was disconnected and the cylinder put into connec- tion with the gasholder, and gas was admitted to bring the pressure within up to one atmosphere ; the mercury gauge was then discon- nected, and communication opened with the air reservoir till the pressure rose to 35 lbs. per sq. in. above atmosphere. The stirrer was then operated to thoroughly mix the whole of the gases. The pressure in the vessel was then reduced to the initial pressure required, and the reading of the initial pressure was made on the Bourdon pressure gauge, which was carefully calibrated. The initial temperature of the explosion chamber was then read, and aU the instruments were then disconnected. The recording drum was set in motion, the indicator was opened to the cylinder, and the spark passed within the next half- second. After the cooling of the products of combustion to nearly the initial temperature the pressure in the cylinder was read. The cylinder was cleared of the products of combustion by fiULng up with air twice and blowing off at the top and bottom alternately. When the experimenters had satisfied themselves that they always obtained a homogeneous mixture in the explosion vessel, they made preliminary experiments to discover if the position at which firing was started within the explosion chamber made any difference as to time of explosion and maximum pressure attained. For this purpose they used the firing tube f ; this tube was plugged at the lower end, but it had a pin-hole aperture in the side directed horizontally. When the gases were ignited within the firing tube by the spark, the position of the pin-hole determined the point within the vessel from which ignition started. Taking one mixture in this way and firing with the 176 THE GAS, PETROL, AND OIL ENGINE pin-hole at different distances from the upper cover of the explosion chamber, the explosion was found to be most rapid fired at the middle point of the chamber. IGNITION LtVEL WlTh COVER. H 200- --^ 100- / A BC^,„„u^ ■ ICNITION 12 INS. DOWN. / ^- 200- 100- I 005 TIME, sees. 1GNITI0W 3 ms. DOWH 0-O5 TIME.SECS. O-IO 0-1 200- IGNITION 15 INS. DOWN. /^'''^ 100- XK' 005 1 O-IO O-l TIME.SECS. L..,____ 200- IGNITION 9INS. DOWN. 100- y Fig. 60.— Explosion Diagrams from a mixture 01 i air and 0'I42 coal gas, with initia pressure 35 lbs. per sq. in. absolute and ignition started at varying distances from the upper cover. (Bainiow and Alexander) The time of explosion was O'oy second less when the mixture was ignited at the middle of the vessel 9 ins. from the cover, than when ignited I'a in. from the cover. Notwithstanding this difference in time the maximum pressure attained in both cases was the same. BAIRSTOW AND ALEXANDER'S EXPERIMENTS 177 The mixture was rich in gas — i air to 0-125 gas — so that the explosion was rapid. The authors conclude that, as maximum pressures coincide, the cooling loss during explosion must be small. In the case of a weak mixture, when ignited near the top, the maximum pressure of 137 lbs. per sq. in. absolute was attained in I '8 second; when ignited 12 ins. from the top, a maximum pressure of 180 lbs. per sq. in. absolute was attained in 0'4 second. The cooling action of the cylinder during the explosion period (i'8 second) prevents an increase of more than 30 per cent, of the pressure (137 lbs.) actually attained. The diagrams obtained with a rich mixture, i air and 0'i42 of gas — that is, very nearly 7 of air to i of gas — are shown at fig. 60. The initial pressure was 35 lbs. per sq. in. absolute, the mixture and initial temperature was the same in aU cases. The mixture was fired with the pin-hole, level with the inside of the upper cover, 3 ins., 6 ins., 9 ins., 12 ins., 15 ins., and 18 ins. down. The latter figure, of coarse, means that the pin-hole was at the bottom of the vessel. Two diagrams were simultaneously taken by two indicators ; one of the diagrams is shown in fuU lines, the other in dotted lines. X B indicates the point where the fuU-line diagram begins to be recorded and i C the same point for the dotted line. In the first diagram of the set o A marks the real point in time when the electric spark passes as determined by a spark in series, passing through the indicator paper as weU as the sparking points within the cylinder. The true time of explosion is obviously the time elapsing between the passage of the spark and the attainment of meiximum pressure. This is given in the following table, together with the maximum pressures attained : Depth of pin-hole in firing tube from top of vessel Time from spark to maximum pressure Maximum pressure of explosion ins. sec. o-i6o lbs. per sq. in. abs. 220 3 6 0-141 O'll/ 220 220 9 12 15 18 o-ios 0-123 0-140 0-145 223 223 218 218 From these experiments it is seen that explosion is most rapid when the mixture is fired at the middle of the vessel. The maximum pressure is then attained in 0-105 second after the moment of passing the spark ; fired at the top, this time is 0-160 second, and at the bottom VOL. I. N lyS THE GAS, PETROL, AND OIL ENGINE 0"i45 second. The maximum pressure, however, varies but little: 220 lbs. absolute with the top firing, 223 lbs. with the middle firing, and 218 lbs. with the bottom firing. Evidently it is best to fire at the middle position of the vessel in these experiments when the most rapid ignition is desired. Further experiments with the same composition of mixture proved that with top firing the indicator lagged behind the spark in indicating beginning of pressure rise by approximately 0'03 second — that is, three-hundredths of a second. By arranging, however, for four sparks in series to pass right through the axis of the cylinder the indicator lag was reduced to 0-017 second, or about half the lag given when one spark was used at the top of the cylinder. Similar experiments were made with a weak mixture, fired in the same manner at different points in the cylinder. The initial pressure was 50 lbs. per sq. in. absolute. The results are given in the following table : Depth of pin-hole in firing Time from spark to Maximum pressure of tube from top of vessel maximum pressure explosion ins. sec. lbs. per sq. in. abs. 2 1-9 152 4 i-i 168 6 0-8 178 8 0-8 173 10 0-65 189 12 0-50 197 14 0-55 197 16 o-6o 194 Here the shortest time of explosion was found when the igniting point was 12 ins. from the upper cylinder cover and the longest time was at 2 ins. below that cover. The shortest time of explosion was 0-5 second and the longest 1-9 second. The lowest maximum pressure was 152 lbs., and the highest 197. It is to be remembered that all these changes occur with identical mixtures, the only difference being the point of firing in the vessel. With this slow explosion the best point for firing is 3 ins. lower than the centre of the vessel, because convection is sufficiently rapid to*e comparable with the rate of com- bustion and the lower position assists the action of convection. In these experiments the indicator lagged behind the spark from 0-25 second at 2 ins. from top ignition to O'li second with the 12-in. position. With a similar weak mixture the four-spark method reduced the lag to 0-036 .second. An the experiments now to be described were made using four BAIRSTOW AND ALEXANDER'S EXPERIMENTS 179 sparks in series, passing through the axis of the cylinder and also using mechanical mixing by the plate. Two sets of experiments were made with varying mixtures — one set at an initial pressure of 55 lbs. per sq. in. absolute, and the other at 34-5 lbs. Fig. 61 shows the explosion and part of the cooling curves taken with varying mixtures of gas and air at an initial pressure of 55 lbs. per sq. in. Fig. 61.— Explosion and part Cooling Curves of various mixtures of gas and air at initial pressure of 55 lbs. per sq. in. absolute. [Bairstow and Alexander) The composition of the mixture is marked on each curve, and it will be seen that the mixture varies from a gas content of 6'36 per cent. to 127 per cent. The following table gives the numbers relating to fig. 61 : Explosion in a Closed Vessel. (Bairstow and Alexander) Initial pressure 55 lbs. per sq. in. absolute Initial temperature atmospheric Mixture containing from 6"36 to 127 per cent, gas Fraction of total volume occupied by gas Maximum pressure observed lbs. per sq. in. abs. Time to reach maximum pressure in seconds Maximum temp of explosion Per cent. 6-36 70 1-5 to 1-8 — 7-27 — — — 772 127 0-8 396 8-i8 140 0-8 465 8-63 208 17 823 8-88 239 i-o 790 9-09 255 07 1070 10-04 331 0-2 1470 10-09 350 0-1 1570 12-07 385- 0-04 1760 i8o THE GAS, PETROL, AND OIL ENGINE Fig. 62 shows the explosion and part of the cooling curves taken with varying mixtures of gas and air at an initial pressure of 34*5 lbs. per sq. in. absolute. Q^ -14-98 X f ~IZO I /^ -9-23 % / .^ 7-79% A / ^ '' f^ _ -^ TIME. SECS. Fig 62. — Explosion and part Cooling Curves of various mixtures of gas and air at initial pressure of 34*5 lbs. per sq. in. absolute. (Bairstow and Alexander) The composition of the mixture is marked on each curve, and it wiU be seen that the mixture varies from 779 per cent, gas to I4"98 per cent. The following table gives the numbers relating to fig. 62 : Explosion in a Closed Vessel. (Bairstow and Alexander) Initial pressure 34'S lbs. per sq. in. Initial temperature • 17° C. Mixture containing from 779 per cent, to I4'98 per cent, gas Fraction of total volume occupied by gas Per cent. 779 9-23 1072 12-05 13-5 14-98 Maximum pressure observed, lbs. per sq. in. above atmos. 97 193 216 245 263 272 Time to reach maximum pressure in seconds 1 0-8 0-24 0-13 0-07 0-05 0-04 Maximum temperature of explosion, °C. 540 1350 1540 1790 1930 2010 Two sets of experiments were made with varying initial pressures and constant mixtures. In both sets the initial pressure was varied from about 7 lbs. to 45 lbs. per sq. in. absolute. One mixture contained approximately 14-4 per cent, of gas, and the other 9-5 per cent. BAIRSTOW AND ALEXANDER'S EXPERIMENTS i8i Fig. 63 shows the explosion and part of the coohng curves of the stronger mixture. TIME. SECS. Fig. 63. — Explosion Curves of gas and air mixture containing I4'4 per cent, of coal gas at varying initial pressures between 7 and 45 lbs. per sq. in. absolute. (Bairstow and Alexander] The following table gives the numbers relating to fig. 63 : Explosion in a Closed Vessel. {Bairstow and Alexander) Initial pressure varied from yiS lbs. to 44'8 lbs. per sq. in. absolute Initial temperature approximately 22-5 C. Mixture containing 14-4 per cent, gas and 85-6 per cent, air Mixture. Initial Max. pressure Tims to Max. temp. pressure, lbs. per sq. in. Initial temp. observed lbs. per sq. in. reach max. pressure in of explosion Air Gas abs. abs. seconds °c. 0-169 44-8 33-5 348 0-042 2030 0-172 34-5 22 270 0-041 2040 0-170 247 21 189 0-041 1980 0-168 14-55 21 112 0-036 1990 o-i66 971 24-5 68 0-05 I8I0 0-166 7-18 24 47 o-io 1670 l82 THE GAS, PETROL, AND OIL ENGINE Fig. 64 shows the explosion and part of the cooling curves of the weaker mixture : INITIAL PI ^ESSURE ^ I 1 44-7 34-6 4_ 6_ 14-4 9-5 706 / / 100- ^ 3 1 / / y / 4 5 — 6 Fig. 64. — Explosion Curves of gas and air mixture containing 9-5 per cent, of coal gas at varying initial pressures between 7 and 45 lbs. per sq. in. absolute. (Bairstow and Alexander) The following table gives the numbers relating to fig. 64 : Explosion in a Closed Vessel. (Bairstow and Alexander) Initial pressure varied from 7-06 lbs. to 447 lbs. per sq. in. absolute. Initial temperature . . atmospheric. Mixture containing 9-5 per cent, gas and 90-5 per cent. air. Mixture Initial 1 Max. pressure Time to Max. temp. pressure, lbs. per sq in. Initial temp. "C. observed, lbs. per sq. in. reach max. pressure in of explosion Air Gas abs. abs. seconds °C. 0-103 447 16-5 238 0-33 1270 0-105 34-6 i8-6 185 0-35 1280 ; 0-104 247 ; 20-0 126 0-41 1220 1 0-107 14-4 ! 2I-0. 74 0-44 1235 0-104 9-5 S-0 46 0-50 II50 0-107 7-06 33 , 0-50 1 1 10 The cooling curves for the last two sets of experiments are shown at fig. 65. In this figure an arbitrary zero line of time is shown, so that all the curves begin at the same temperature at the small circles shown. The second circles on the cooling lines also show equal temperatures, and from these it is obvious that the higher the initial pressure of the BAIRSTOW AND ALEXANDER'S EXPERIMENTS 183 mixture the more slowly does the temperature fall. These curves are important and will be discussed later. 0S'il3d'SS1 '3iinSS3lld The composition of the coal gas used in these experiments is given by Messrs. Bairstow and Alexander as follows, from their analysis : i84 THE GAS, PETROL, AND OIL ENGINE Analysis of Coal Gas used in Bairstow and Alexander's Experiments Per cent, per volume Marsh gas, CH4 27-8 Hydrogen, H 42-8 Heavy hydrocarbons, Cj-.jHj,., . . 5-4 Carbonic oxide, CO 11 "5 Oxygen, O o-i Nitrogen, N i2-o 99-6 An analysis by the Gas Company of earlier date gives the heavy hydrocarbons at C^.8sH5.s3. The higher heating value of this gas as measured by a Dowson calorimeter was 682 B.Th.U. or 379 Centigrade heat units ; but determinations at other times showed 648 B.Th.U. or 360 Centigrade heat units higher value for i cub. ft. of the dry coal gcis measured at 0° C. and 30 ins. mercury. Bairstow and Alexander have calculated the proportion of heat accounted for at maximum temperature of explosion for mixtures con- taining 0'i84 gas to I of air (i volume gas, 5*3 volumes air), and 0-102 gas to I of air (i volume gas 9-8 volumes air), using the following values : 'Calorific value (higher), i cub. ft. of coal gas at 0° C. and 30 ins. mercury . . . . . = 379 Centigrade heat units. = 682 B.Th.U. Specific heat at constant volume of I cub. ft. of oxygen or nitrogen at 0° C. and 30 ins. mercury . = 0-0138 Centigrade heat units. „ „ Carbon dioxide = 0'02ii ,, „ ,, „ Steam = 0-0197 For the mixture i volume gas, 5-3 volumes air, the composition of the products of combustion was : CO2 = o-2i6 cub. ft. It= 1-677 „ O = 0-013 „ Water of saturation, HjO = o-oi8 „ Volume of steam formed, H2O = 0-498 „ In this experiment 0-382 cub. ft. of coal gas was burned from an initial pressure of 44-1 lbs. per sq. in. absolute, giving a maximum pressure when corrected for cooling during explosion of 403 lbs. per sq. in. absolute. HOPKINSON'S EXPERIMENTS 185 If the maximum pressure of complete heat evolution be calculated from the above figures, it should have been 591 lbs. per sq. in. absolute. That is, the heating value of the gas measured by the rise of pressure corrected for cooling is 657 per cent, of the calculated value, and the corresponding temperature attained is 2410° C. instead of 3660° C, as calculated. With the mixture containing i gas to 9'8 of air the rise of pressure is 80 per cent, of the calculated value. Bairstow and Alexander state that these values are higher than those previously noted, because a. The specific heat values adopted by them are somewhat greater. b. The actual maximum pressure obtained is greater. c. A cooling correction is applied. Bairstow and Alexander also state : ' The determination of the actual maximum temperature and the amount of heat developed at that point involves considerable diffi- culties.' Mr. D. Clerk discusses the effect of change of volume on the tem- perature, and says : ' It is evident that the limits of possible error in calculating temperature from pressure of explosion does not exceed, in the worst case, with coal gas and air 3-4 per cent., and in weaker mixtures half that number.' The 3-4 per cent, referred to is the difference between the volumes before and after explosion on the basis of no condensation of water vapour. This statement is fundamentally incorrect ; if the products of combustion were heated to such a tem- perature that carbon dioxide was completely dissociated into carbon monoxide and oxygen and the steam into hydrogen and oxygen, the volume would correspond to one 12 per cent, greater than the volume of coal gas and air before explosion, and the contraction which is really concerned in the calculation of temperature is therefore more than 15 per cent., and the uncertainty is proportionately increased.' The present author does not agree with Messrs. Bairstow and Alexander in their criticism of his statement of the case made in 1886. The question of the change of volume caused by dissociation of carbon dioxide and of steam is dealt with in the Report of the Committee of the British Association on Gaseous Explosions, which will be found in Appendix IV. Hopkinson's Experiments with a Large Explosion Vessel Important experiments have been made by Prof. Hopkinson, of Cambridge University, and published in a paper read before the Royal Society in February 1906. They are important in three respects : the explosion vessel was very large, its capacity was 6-2 cub. ft., and it 1 86 THE GAS, PETROL, AND OIL ENGINE was of dumpy cylindrical form, as shown in section at fig. 66. 59-5 cm. diameter (23-4 ins. diameter), 73 cm. long {2875 ins. length) ; pressure changes were measured by an optical indicator ; temperatures at various parts of the vessel were measured by fine platinum-wire resistance thermometers placed at different positions in the vessel. Experiments were made with two mixtures — i of gas to 9 of air ; and I of gas to 12 of air — that is, mixtures containing respectively 10 per cent, and 77 per cent, of gas. In all experiments it was endeavoured to saturate the mixture with water vapour. The experiments were all made with the initial pressure atmo- sphere : with a mixture of Cambridge coal gas having an average ' higher ' calorific value of 378 Centigrade heat units or 680 British thermal units per cubic foot at 0° C. and 760 mm. mercury. The analysis of Cambridge gas, together with oxygen required for combustion, steam, and carbonic acid produced is as follows : Analysis of Cambridge Coal Gas. (Hopkinson Experiment) _ 1 Per cent. [ by volume Oxygen required for combustion Steam produced Carbonic acid produced Hydrogen, H . . 47-2 Marsh gas, CH,, . . 35-2 Heavy hydrocarbons . ' 4-8 Carbonic oxide, CO . 7-15 Nitrogen, N . . , 5-4 Other gases . . 0-25 23-6 vols. 70-4 vols. 2 2 '6 vols. 3 -6 vols. 47-2 vols. 70-4 vols. i6'0 vols. 133-6 vols. 35-2 vols. 14-4 vols. 7-15 vols. loo-oo vols. 1 20 '2 vols. 5675 vols. By experiment, 100 volumes of gas require 576 volumes air for complete combustion. By experiment, 100 volumes of gas burned in 900 volumes air give 133 volumes steam, 57 volumes CO2, and 780 inert gases. By experiment, higher calorific value of i cub. ft. at 0° C. and 760 mm. mercury = 680 B.Th.U. or 378 Centigrade heat units. Referring to the section of the explosion vessel, fig. 66, A is the sparking-point, nearly at the cen^fe of the vessel ; B, c, and D are three platinum thermometers. B is close to the spark, c is 30 cm. from the spark, and d is i cm. from the walls of the vessel. Each thermometer consists of a coil of about 5 cm. of pure platinum wire of O'ooi in. diameter. Glass tubes carry stout copper leads to the platinum wires, and the . copper-platinum junctions are made within the glass tubes so as to be protected from the fliame. Each thermometer coU is placed in series with a storage cell and with a d'Arsonval galvanometer having a stiff HOPKINSON'S EXPERIMENTS 187 phosphor-bronze suspension giving a period of from -V to ^V of a second. The mirror of the galvanometer throws the image of a fine hole, illuminated by an arc lamp, on a revolving drum, carrying a photographic film. Knowing the resistance of the platinum wire when cold, its rise in temperature can be calculated. Usually two thermometers were in use at once and a record was obtained from both, on the same drum and in the same explosion, of the changes of temperature at two different points of the vessel. A record of pressure change was taken on the same drum. The indicator was of simple construction, con- sisting of a steel piston, which was forced by the pressure against a piece of straight spring held at the ends. The displacement of the Fig. 66. — Hopkinson Explosion Vessel spring tilted a mirror about a fulcrum, and the mirror cast an image of the above-mentioned fine hole on the moving film. The indicator had a natural period of about ^j-^^ of a second, and so was able to foUow the changes of pressure with great rapidity. The experiments were made as follows : Steam was blown into the vessel, so that the air should be as nearly saturated with moisture as possible. The vessel was then exhausted, say, to -^"^ of an atmo- sphere, and gas was admitted to bring the pressure up to atmosphere ; this gave a mixture of i gas to 9 air. The explosion vessel was allowed to stand from four to six hours to permit mixing by gaseous diffusion. The completeness of combustion was tested in a few cases by measuring the fall of pressure due to the condensation of steam formed by the explosion. The combustion was found to be satis- factorily complete. Experiments were made to determine • the lag of the platinum thermometer behind the temperature of the gases. i88 THE GAS, PETROL, AND OIL ENGINE From this short account it will be seen that Prof. Hopkinson's experiments were very carefully made, and all precautions taken to obtain accuracy. The only uncertain point in the present author's view is in the question of mixing — even a six hours' delay may not have completed diffusion. With the large vessel used, however, and the relatively small gas-inlet aperture, it is probable that the velocity of the entering gas was sufficient to obtain good mixing. The diagrams obtained seem to be those proper to a good mixture. Explosion of Rich Mixture, i Gas, 9 Air With 9 volumes of air to i of gas the maximum pressure of explo- sion varied from 76 to 82 lbs. per sq. in. above atmosphere, and was reached in 0-25 second after firing. Fig. 67 is a copy of the photographic diagram produced, with its pressure line and two temperature lines. 0:8 07 Ore . . .0:5, . , , . . . .t^-*. . . ■ 1 ■_■■■_ Jl A A A A A A A/ A :P^— lU ^W t{\fi\t{ \M[t\ /fl TTffltrT 1 L iMiJll: niMWWMM tliiwtr r T iirrm mm\mmm\i J'\l \l \l \l 1/ 1/ A/-W ^V^l/W H/ Hi H/ ^l/WWW 1 0-3 ^ |^?0;J D, 01 ° TIME.StCS. _ 1 t- ^ -80 1 \^j A [\ /{ [\ A-i B -.0| Ifi -40. a -20 X M/L.^ M mwm AMMM MKli X li^ =M. i= i: ■ TwWiTT mT -0 mMtrririiLii rW \f\f\i r\7-\7 ^TTM v^ hPvwTTir Fig. 67. — Explosion and Cooling Curve of gas and air mixture ( i gas and 9 air) at atmospheric pressure, with temperatures reached by platinum resistance thermometers, (ffjjtkmsim) The film is shown in two parts : A is the pressure curve, of which the explosion portion is shown on the lower part ; the zero point of the horizontal line of time indicates the point of passing the spark ; the line of cooling is continued on the second part, as indicated by the asterisk. B is the curve recording the temperature determined by the centre thermometer B (fig. 66), and b„ is the zero line for this curve, traced on the film immediately after the explosion by disconnecting the HOPKINSON'S EXPERIMENTS 189 thermometer circuit. The current flowing in the circuit is, therefore, after allowing for inertia effects, proportional to the ordinate of the curve B, reckoned from the zero line b^, and the resistance of the circuit is inversely proportional to that ordinate. D is the curve recording the temperature determined by the ther- mometer D (fig. 66), which is i cm. from the side. The galvanometer here has a stiffer suspension and a lower period of oscillation, so that it is less sensitive. Do is the zero line for this record. On this diagram we have, therefore, the means of determining by direct measurement the temperature changes occurring at the centre of the vessel close to the point of starting the explosion, the tem- perature at I cm. from the wall of the vessel at the same time, and the changes of pressure shown by the indicator, giving mean tem- perature of the whole mass of contained gas. Comparing first the indications of thermometer B with the explosion line of the pressure indicator at the point B|,the curve b rapidly departs towards the zero line Bo, showing a reduction in resistance, and con- sequently a rise in temperature. The galvanometer is thrown into violent oscillation by the rapidity of the temperature change. This indicates the contact of the flame which has been started on its course by an electric spark passed at the point a, about 2 cm. distant. This is taken as the beginning of the explosion, and it will be seen that it has had so far no effect on the pressure in the vessel as indicated by the pressure line A. Smoothing out the oscillations of the curve B, it will be seen that the temperature remains nearly constant, there is a very slow rise, till the point B2 is reached, when a further abrupt temperature rise occurs, so that the curve falls to the zero line Bo; it continues oscillating about that line for the remainder of the record. The point B2 is slightly before that of maximum pressure as shown by A, and corresponds to the sudden melting of the wire. With this mixture fired at the centre the wire was always found to melt. The comparison of temperature at the centre of the vessel with mean temperature as determined by pressure therefore ceases about 0'025 second before maximum pressure is attained. Consider now the curve D : it is not deflected till the point D , is reached, and here the temperature begins slowly to rise ; at Dj there is a sudden rise, just coi second before the centre wire melts, and 0"035 second before maximum pressure is attained. The temperature attained by this rise is maintained for nearly one second and then slowly falls. By taking the mean of successive maxima and minima. Professor Hopkinson has prepared the following table showing the temperature of the thermometer, B at the centre of the vessel reckoning time from the point Bi on the diagram (fig. 67) tUl the point of melting the platinum. I go THE GAS, PETROL, AND OIL ENGINE Explosion in a Closed Vessel. (Hopkinson) Initial pressure, 14-7 lbs. per sq. in. abs. Initial temperature, 20° C. Time from start of ignition-point E' Temperature (° C.l by platinum thermometer at centre of vessel Time from start of ignition-point B^ Temperature (° C.) by platinum thermometer at centre of vessel 0-008 0-024 0-041 0-057 0-074 0-09 560 995 II35 \ II65 II65 1225 0-107 0-123 0-140 0-173 0-26 1260 1275 1275 1400 1710 wire melts The maximum temperature shown on the curve D, Professor Hopkinson states, is about 1250° C, and occurs very nearly at the moment of maximum pressure. This estimate he says, however, may be a good deal wrong, because the diagram being on a smaller scale cannot be as accurately measured as B. As the result of a considerable number of experiments with this mixture always fired at the centre Professor Hopkinson considers that the distribution of temperature at the moment of maximum pressure is roughly as follows : Mean temperature (inferred from pressure) .... 1600" C. Temperature at centre of vessel, thermometer B . . . 1900 Temperature 10 cm. (4 in.) within wall, thermometer C . 1700 Temperature i cm. (o'4 in.) from wall at end, ther- mometer D iioo to 1300 Temperature i cm. (o'4 in.) within wall at side . . . 850 At the thermometers B, C, and D (he states) the gases can have lost but little heat at this period, and the temperature differences are mainly due to the different treatment of the gas at different places. At B it has been burned nearly at atmospheric pressure, and com- pressed after burning to about 6^ atmospheres absolute ; at D it has been compressed to about 6 atmosplftres as in a gas engine, and then ignited without any subsequent compression ; at the point i cm. within the wall at the side much heat has been lost, as this is the first point reached by the flame ; the gas here is ignited when the pressure is about 2 atmospheres ; its temperature rises instantly to 1300° C, and at once begins to fall. Half a second after maximum pressure the distribution is very different : convection has now had time to take effect. HOPKINSON'S EXPERIMENTS 191 The distribution of temperature is broadly as follows : Mean temperature inferred from pressure 1100° C. Mean temperature, exclusive of layer i cm. thick at walls, determined by long platinum wire from B to d . . . 1160 Temperature at centre of vessel, thermometer B . iioo to 1200 Here the temperature differences are much smaller than at the moment of maximum pressure. The mass of gas during cooling may be described as a hot core in which the temperature is approximately uniform, varied accidentally by currents, surrounded by a thin layer wherein the temperature falls to the temperature of the walls. Pro- fessor Hopkinson calculates that if such layer were ^ cm. thick, and if the fall of temperature were uniform, the mean temperature inferred from the pressure would fall short of that of the hot core by about the observed amount, that is, 60° C. Assuming both air and gas to be saturated with moisture, Professor Hopkinson calculates the composition of the products of combustion from the mixture used in the experiment — g of air to i of gas — to be as follows : Carbonic acid, CO 2 = 578 per cent, by volume Water, H2O = 15-80 Nitrogen and oxygen, N and = 78"5 ,, ,, 100 -08 per cent, by volume. This assumes that the CO 2 and H O occupy their proper molecular volumes at 20° C. The following are the most important of Professor Hopkinson's deductions with regard to this mixture : Velocity of Spread of Flame. — This is found to be roughly 150 cm. (59 ins.) per second between the thermometers B andD. Mallard and Le Chatelier's value for a mixture containing 17 per cent, of coal gas in a tube 2 to 3 cm. diameter was 125 cm. per second. Period, of Explosion when Flame fills Vessel entirely. — The explosion vessel is entirely filled with flame when the pressure reaches 70 lbs. per sq. in., although the maximum pressure attained later is 82 lbs. per sq. in. above atmosphere. Maximum pressure is attained in ^V of a second after complete filling of the vessel by flame. Maximum Temperature at Centre of Vessel partly due to Compression after Ignition. — The temperature at the centre of the vessel rises rapidly after ignition to about 1225° C, which it reaches within about ^V second. This temperature remains nearly stationary during the early part of the spread of the flame ; the pressure during this time 192 THE GAS, PETROL, AND OIL ENGINE remains nearly constant. After this the mass of gas near the centre, at about 1200° C., is compressed to 6-5 atmospheres, nearly adiabatically to a temperature of about 1900° C. Assuming a loss by radiation from the flame of about 15 per cent., then the specific heat of the products of combustion between 1200° and 1900° C. is 1*3 times that of air, and the average value of y (ratio of specific heats) is i'25 between 1200° and 1900° C. Temperature Differences would exist in a Gaseous Explosion ignited in a perfectly non-conducting Vessel. — Hopkinson's experiments dis- tinctly prove that temperature differences would exist after complete inflammation, even in an entirely non-conducting vessel. The highest temperature would be reached at the point of origin of the ignition due to the compression of the gases first heated by explosion to about 1200° C, and then compressed by the compression from the walls inward as the mixture near the walls inflames. TIME. SEC3. The successive portions of the pressure and temperature curves are numbered in the order of the corresponding revolutions of the drum Fig. 68. — Explosion and Cooling Curves of gas and air mixture (i gas and 12 air) at atmospheric pressure, with temperature recorded by platinum resistance thermometer. [Hopkinson) This is the most novel point found by Hopkinson's investigations. It was pointed out by the present writer in 1886 that a gaseous explosion in a cooling vessel consists of a hot core and a cool zone in contact with the walls ; but Hopkinson's point was not discovered, although, now that it has been proved, it is obvious that the pheno- menon should have been capable of prediction. Explosion of Weak Mixture, i Gas, 12 Air. — With 12 volumes of air to I of gas the maximum pressure of the explosion is about 50 lbs. per sq. in. above atmosphere, and it% attained in 2*5 seconds after the passage of the spark. Fig. 68 shows a copy of the photographed diagram produced with its pressure line and one temperature line. A large number of diagrams were taken with the platinum ther- mometer in all sorts of positions, but in this case only one wire was used, and it was placed 15 cm. (5 -9 ins.) from the spark and vertically below it. At first the temperature in the wire rises very slowly. HOPKINSON'S RECORDING CALORIMETER 193 More than 2 seconds after ignition (at a) it is only about 210° C, and this heating is almost entirely due to adiabatic compression. The flame reaches the thermometer at the point A, and the temperature rises in yV second to 1300° C, at which point (b) the pressure has attained its maximum value, about 25 seconds after the spark has passed. The temperature is steady for a while and then faUs, but there is no perceptible rise after the pressure begins to faU. Hopkin- son states that in a large number of trials he was unable to discover any Fig. 69. — Recording Calorimeter for Explosions. {Hopkinson) point at which inflammation occurred later than in this diagram, so that it is probable that all the gas is ignited at the time of maximum pressure. Hopkinson's Experiments with a New Recording Calorimeter FOR Explosions If the heat-flow from a mass of hot gases contained within a cylinder could be directly measured as the heat passes from the gas into the waUs, it would be possible to avoid many difficulties in the determination of the questions of varying specific heat, continued combustion, law of heat loss, and so forth. The apparatus would have VOL. I. 194 THE GAS, PETROL, AND OIL ENGINE the advantage of independence of specific heat determinations by combustion and other methods. Professor Hopkinson has attacked the difficult problem with a most ingenious apparatus, which he describes in the following way : ' It consists essentially in lining the explosion vessel as completely as possible with a continuous piece of copper strip and recording the rise of resistance of the copper strip during the progress of the explosion and the subsequent cooling. Knowing the temperature of the copper and its capacity for heat, the heat that has flowed into it from the gas may be calculated from the resistance.' Fig. 70. — Recording Calorimeter for Explosions. [Hopkinson) The explosion vessel is shown in longitudinal section in fig. 69, and the end of the vessel in elevation at fig. 70. ' It consists of a cast-iron cylinder A, i foot in diameter and i foot in length, on to which are bolted two end-plates b. The cylinder was first completely lined with woo* |-in. thick and the end-plates covered with pieces of cork. Thirty-nine turns of copper strip c of the quality used for electric lighting purposes, and of a high degree of purity, were then wound on the inside of the curved portion, a clearance of about ^ of an inch being left between successive turns. The strip was J-in. wide by ^^y-in. thick. The ends of this piece of strip were brought to terminals outside the vessel. The end-plates were similarly covered HOPKINSON'S RECORDING CALORIMETER 195 with parallel pieces of copper strip of the same dimensions, as shown in fig. 70, the ends being brazed to connecting pieces. The strips on the ends were electrically connected outside the vessel to the strip in the cylindrical part. The whole when put together formed an explo- sion vessel having a capacity of about 0-684 cub. ft., which (except for the uncovered portions on the ends where the cocks, &c., came through) was completely lined with an electrically continuous length of copper having an approximately uniform section of -r|^ of a square inch. For recording the pressure I used an optical indicator, consisting of an iron piston which was forced by the pressure against a piece of straight spring held at the ends. The displacement of the spring tilted a mirror about a fulcrum, and the mirror cast an image of a fine hole illuminated by an arc lamp on to a photographic film carried on a revolving drum. This indicator was repeatedly calibrated by dead \ ^^___^ ._— — — — — ,^ 'to 30- — _— -40 M- ^^ /%/ H^ -.>^ / ^ -60 10- r /^ fo B -80 " TIME FROM ICMITION. SECS Pressure before ignition (atmospheric), 14*6 lbs. per sq. in. Maximum pressure, golbs. per sq. in. above atmosphere. Temperature before ignition, 15° C. Maximum temperature of explosion, T760 ° C. Fig. 71. — Explosion and Cooling of mixture (gas 1, air 6-88 vols.), with copper- strip calorimeter, curve showing heat flowing to walls during explosion and cooling. [Hopkinson) weights, and I think its readings are to be trusted to within i per cent, of the maximum reached. The mixture was fired by an electric spark at the centre of the vessel, and it was at atmospheric pressure and temperature before firing. When the explosion takes place the copper strip is heated and its resistance rises, and since the current in it remains constant during the short time occupied by the cooling of the gas to ordinary temperatures, the potential at the terminals of the strip rises by an amount proportional to the increase of resistance or to the increase of temperature. Since the potential at the terminals of the resistance remains constant, except for the small disturbance due to the passage of the galvanometer current, the gal- vanometer deflection from the reading just before the explosion will be proportional to the rise of potential between the terminals of the strip or to the rise of temperature. The mirror of the galvanometer reflected on to the moving film an image of the same small hole as 196 THE GAS, PETROL, AND OIL ENGINE was used for recording the change of pressure, and a simultaneous record was thus obtained of the change of temperature of the strip and of the pressure in the vessel.' Fig. 71 shows a record of the change of pressure due to explosion and cooling, and the change of temperature of the copper strip which forms the calorimeter. Curve A is the pressure measured downwards from the atmospheric line a„ and curve b is the galvanometer deflection measured upwards from the zero line Bo. The galvanometer is thrown into oscillation by the rapidity of the first heat addition from the explosion. An arc using an alternating current was the source of light for illuminating the aperture ; hence white dots were shown in the photograph, and these are useful to identify corresponding points on the two curves. The following are the particulars of the experiment which gave the record, fig. 71. Mixture used, i gas, 6 -8 8 air Vessel, 0-684 '^"6' fi- capacity Cambridge coal gas 0'082 cub. ft. = 127 per cent. Air (includuig some water vapour) . . o-56s = 87-3 Total ..... 0-647 = loo-o These volumes are at standard temperature and pressure 0° C and 760 mm. Calorific value of the gas by Boys' calorimeter at 0° C. and 760 mm. per cub. ft. Higher value 670 British thermal units or 372 Centigrade heat units = 170,000 gramme calories per cub. ft. Pressure before explosion, 753 mm. = I4'6 lbs. per sq. in. Temperature before explosion, 15° C. = 288° C. absolute. Products of Combustion Carbonic acid .... CO.^ = 0'046 cub. ft. = 7-4 per cent. Steam Hp = o-ii8 = 18-9 Nitrogen and oxygen .... 0-462 = 73-7 Volume of products assumed to be] gaseous at 0° and 760° mm. . \ °'°^^ = ^°°'° Holborn and Austin's determinations of specific heat of carbonic acid, steam, nitrogen, and oxygen have been taken. Carbonic acid, CO2 = 107 calories per standard cub. ft., mean value between 15° C. and 545° C. Steam under same condition, 8-4 calories. Nitrogen and oxygen, 6'3 calories. These are specific heats at constant volume, aU in gramme calories. Allowing for a contraction of 3 per cent, for the combustion of HOPKINSON'S RECORDING CALORIMETER 197 Cambridge gas (i gas and 7 air) on the original photographic record, I mm. on the pressure diagram corresponds to a temperature rise ol 36°'6 C.J and on the galvanometer curve B a rise of i mm. is equivalent to a mean temperature rise of the strip of o°'83 C, or a heat quantity of 222 gramme calories. Fig. 71 is reduced from the original Glm, but the temperature and pressure are marked. ' Of the heat which passes into the copper^ some part is lost to the wooden backing behind it, and it is the balance only which is directly measured in the diagram. The percentage of heat so leaking out is a correction which increases from less than i per cent, o'l second after firing up to about 20 per cent, i second after firing. In order to deter- mine the amount of this correction, resource was had to a method of electrical heating. . . . ' In addition to the heat which passes into the copper and, via the copper, into the backing behind it, heat also goes into those parts of the walls which are not covered.' Professor Hopkinson estimates that the whole heat which the gas has lost exceeds that which has gone into the copper in the ratio fy-g-^, or 6 per cent. To test the accuracy of the new calorimeter. Professor Hopkinson considers the point on the cooling curve A, fig. 71, one second after ignition. Here the gaseous mass has fallen to a mean temperature of 545° C, and the corresponding point on the curve B shows the heat actually in the copper strip at the moment to be 7,850 calories. The heat passed through to the backing Hopkinson calculates 1,570 calories, or 20 per cent, of the heat in the copper. The total heat in and passed through the copper is thus 9,420 calories. Multiplying this by i'o6 gives 10,000 calories as the total heat-flow from the gaseous contents to the walls from the moment of ignition to the moment when cooling has reduced the gases to 545° C. Comparing this determina- tion with the total heat in the gas and the amount remaining when cooling from 545° to 15° C. by Holborn and Austin's figures, already given, he gets Total heat of combustion of 0'647 cub.i ^, „„„ „„i„„- „ ' 14,000 calories, ft. of the mixture . Heat of products evolved on cooling 1 from 545° to 15° C ) 3,807 10,193 gramme calories. That is, if Holborn and Austin's and other constants be accurate, the gas contents, in burning to maximum temperature and cooling from it to 545° C, should evolve, say, 10,200 calories. Hopkinson's instrument igS THE GAS, PETROL, AND OIL ENGINE shows 10,000, a satisfactory correspondence. Calculating, however, to 0-5 second from firing, the agreement is not so good. The tempera- ture reached is 840° C, and the total heat lost to the walls, as shown by the instrument, is 7,980 calories ; but from Holborn and Austin's figures it should be 8,820 calories, so that the instrument appears to register about 10 per cent, too low. Hopkinson accounts for this partly by possible error in Holborn and Austin's values, but finally comes to the conclusion that at half a second after firing part of the gas is still unburned. Fig. 72 is an interesting curve, taken from Professor Hopkinson's paper, which shows the heat loss from the gases to the copper strip in calories per square centimetre at different times from the moment of firing. The mean gas temperatures at the different times have MEAN CAS TEMPERATURES. DECREES CENT. 1300 1100 950 e£0 Ij ^-' TIME FROM ICNITIOH.SECS. Fig. 72. — Curve of Heat Loss per square cm. at different times from ignition, from Experiment, Fig. 71. (Hopkinson) been added. The maximum mean temperature of this explosion was 1760° C, and it fell to 840° C. 0-5 second after firing . An examination of this curve shows equal heat loss per degree fall from 1760° to 840° C. With regard to the rate of loss Professor Hopkinson states : ' The loss of heat begins about -^^ second after ignition, when the flame first comes into contact with the copper. At first the loss goes on at a very great rate, and by the time maximum pressure is reached (when the flame is in contact with the whole surface of the strip and losing heat to every part of it), about 1,700 calories, or 12 per cent, of the gross heating value of the gas, has passed into the walls. The rate of loss of heat at this point is about 10 calories per second per square centimetre, and the mean gas temperature is 1760° C. HOPKINSON'S RECORDING CALORIMETER 199 At 0'2 second from ignition the rate of heat loss is about 3^ calories per second, and the mean gas temperature is 1300° C. The mean temperature is reduced in the ratio 074 between these two points, the product of mean temperature and pressure is reduced in the ratio 0'55, but the rate of loss of heat at 0"2 second is only one-third of what it is at maximum pressure. ' CHAPTER VII EXPLOSION AND COOLING IN A CLOSED VESSEL — DISCUSSION OF DATA DEDUCIBLE The experiments made upon explosions of mixtures of coal gas and air have been given in some detail in the previous chapter, because it is desirable for the engineer to understand what has been done and the difficulties which have been met by different investigators. As will be seen, a considerable amount of experimental knowledge is now available for the purpose of developing the general principles underlying the use of gaseous explosions for producing motive power ; but there remain many points of difficulty to be disposed of before it can be said that our knowledge of the working fluid is complete. Some things, however, have been definitely settled. Holborn and Austin's investigations have placed it beyond doubt that the specific heat of steam and carbonic acid increases considerably with increase of temperature, and that a small increase occurs with oxygen and nitrogen. Nernst's investigations have proved that the dissociation of steam and carbonic acid at about 2000° C. is unexpectedly small. Using the new specific heat values, there is still a consider- able discrepancy between calculated and observed temperatures, so that the question of continued combustion has still to be con- sidered. It is not desirable to discuss these matters at this stage, because happily many general deductions may be drawn without relying on disputable theoretical points. It is proposed, then, to discuss the experiments from the point of view of the engineer- designer desirous of having some knowledge of the working properties of his working fluid. As the first tlmig required is an accurate con- ception of the losses to the enclosing walls, this will now be considered for mixtures of similar composition ignited at an initial pressure of one atmosphere. Explosion and Cooling. Initial Pressure Atmospheric. — Broadly, the experiments may be taken as establishing that when mixtures of similar composition are ignited either in large or small vessels at an initial pressure atmospheric, the maximum pressures attained are the same except where the ignition is slow and cooling loss is experienced DISCUSSION OF DATA 201 during the explosion period. The smallest vessel contained o'i5o cub. ft. and the largest 6*2 cub. ft. It may be that there is a small difference masked by this and mixing difficulties, but no such difference has so far been established. Although explosion pressures may be considered to be practically independent of dimensions of vessel, the rate of cooling varies greatly with dimensions. isoo x'^^ ~^^ \ sN^ ^ \; \^ -^ ^ \^ \_ --^ ■"" \^ "^ •^^ - ~ / N^ -- 2 ^> N.^ -~-, ^~- ^*^ ~~-~., 3 ^'^'^ X: — ^ 6 "-•^ 4 ^5 o- TIME. SECS. 1. Hopkinson's Large Vessel 2. Bairstow and Alexander 3. Hopkinson's Small Vessel 4. Clerk's Vessel, 1884, 1885 5- Boston Experiments 6. Clerk's Vessel, igoo Fig. 73. — Cooling after Explosion in Closed Vessels of different dimensions from temperature 1600° C. for 0'5 second Cooling from the same Mean Temperature in Vessels of Different Dimensions. — To enable the rate of cooling to be compared, cooling curves given by six vessels have been drawn ; they are : (i) Hopkinson's large vessel (2) Bairstow and Alexander's vessel (3) Hopkinson's small vessel (4) Clerk's vessel (5) Boston experiments vessel . (6) Clerk's vessel, 1900 6'2 cub. ft. capacity. 0-82 0-684 0-183 o-i8o 0-150 The curves start from the same maximum 1600^ C, at which point time is made zero; the curve is carried to 0-5 second, so that the cooling 202 THE GAS, PETROL, AND OIL ENGINE curves start at zero of time at 1600° C. and fall to various temperatures one half -second afterwards. Fig- 73 is a diagram showing these six cooling curves. The gases in vessel (i) have cooled in 0-5 second through about 530° C, whUe those in vessel (4) have fallen through approximately 1030° C. in the same time. Compare the temperature fall in 0-2 sec. (one-fifth of a second), as has been done in the case of the author's earlier experiments, so as to use a period of the order of the complete working stroke of an ordinary gas engine. At the end of 0-2 second the tem- perature fall in vessel (i) is about 270° €.; in vessels (4) and (5) about / y / J / y / '^ / > y / ^ ^ y ^ y — • ^ i> ^ y --— ' ^ ^^ i-^ / ^ 0- z^^ " 8OO 1000 iZOC MEAN TEMPERATURE. DECREES CEHT. 1. Hopkinson's Large Vessel 2. Bairstow and Alexander 3. Hopkinson's Small Vessel 4. Clerk's Vessel, 1884, 1885 6. Clerk's Vessel, 1900 Fig. 74. — Cooling after Explosion in Closed Vessels of different dimensions ; curves showing fall of temperature in jLth second for equal mean temperatures 740° C. It is obviouSjthen, that the ^j^mperature is falling much more slowly in the large vessel than the small one. Owing to the rapid fall of temperature in the small vessel as compared with the large one, the mean temperature during the periods compared is lower in the small vessel. To compare heat loss more accurately it is necessary to measure the temperature fall in each vessel for a much shorter period, and to so arrange matters that temperature falls are compared not only for equal times but also for equal mean temperatures. Fig. 74 has been drawn to make this comparison possible. The DISCUSSION OF DATA 203 fall of temperature in -^ second has been plotted against the mean temperature in that period of time. In the following table the fall of temperature in the different vessels for ^V second for mean temperatures 1450°, 1400°, 1300° and 1150° C. have been taken from these curves. Fall of Temperature in different Vessels for i second at differing Mean Temperatures Initial pressure Atmospheric Vessel Temperature falls in ^g sec. from mean temperature No. Capacity 1450° C. mean temp. 1400° C. mean temp. 1300° C. mean temp. 1150° C. mean temp. cub. ft. I 6-2 80'' c. 68° C. 52° C. 38° c. 2 0-82 I27°;C. 114° C. 93° C. 65° c. 3 0-684 166° C. 153° c. 131° C. 100° c. 4 0-183 372° C. 327° C. 257° C. 182° c. 5 0-180 372° C. 327° C. 257° C. 182° c. 6 0-150 238° c. 216° c. 184° C. 138° c. From tliis table it is evident that for equal mean temperatures the temperature fall is less in the large vessels than in the small. It would be expected that, if a mass of gas be in contact with a cold surface at a given temperature and a given density, it would lose heat to that cold surface at the same rate wliether it be contained in a large or small vessel. This, however, assumes that the film in contact with the walls is of the same temperature in aU sizes of vessels. If this be true, then the heat loss per square foot expressed for hot mixtures of the same chemical composition should be the same for identical tem- peratures and times. It is therefore necessary to compare the heat losses above given as to quantity of heat. In order to avoid using specific heat values or continued combustion values, it is desirable to make the comparison in terms of temperature fall. As 1° C. temperature fall represents a larger absolute loss in a large as compared with a small vessel, the author will adopt as a unit of heat quantity that amount of heat which is given out by the fall of I cub. ft. of the glowing gas at constant volume by 1° C. The gas to be measured in the first instance at atmospheric pressure and a standard temperature. Call this heat quantity o-we cubic foot degree. The capacity of vessel (i) is 6-2 cub. ft., so that the heat quantity lost by cooling for g^ of a second at a temperature of 1450° C. is 80 x 6-2 — 496 cub. ft. degrees. With vessel (5) 238 x 0-150 = 35-7 cub. ft. degrees, the total heat 204 THE GAS, PETROL, AND OIL ENGINE lost in the large vessel during the time was 496 cub. ft. degrees, and in the small 357 cub. ft. degrees. The internal surfaces of the different vessels are approximately : Number Capacity Internal surface cub. ft. sq. ft. I 6-2 17-3 2 0-82 S-02 3 0-684 4-33 4 0-183 1-79 5 0-180 179 6 0-150 1-60 If now the heat quantities be divided by the surface we get the number of cub. ft. degrees lost in ^ second per square foot exposed. This has been done for the mean temperatures below : Cubic Foot Degrees lost in different Vessels for 55 second, per Square Foot of Exposed Surface at differing Mean Temperatures Initial pressure atmospheric Vessel Cub. ft. degrees No. Capacity 1450° mean temp. 1400° mean temp. 1300° mean emp. 1150° mean temp. I 6-2 28-6 24-4 18-6 13-6 2 0-82 20-7 i8-6 15-2 IO-7 3 0-684 26-2 24-2 207 15-8 4 0-183 38-0 33-4 26-3 18-6 5 o-i8o 37'4 32-9 25-8 18-3 6 0-150 22-3 20-2 17-2 12-9 In the experiments compared the mixtures of coal gas and air were practically the same, i gas, 9 of air ; for vessels (i), (3) and (4) the mixtures (2), (5) and (6) were richer of about i gas to 7 of air. The Hopkinson experiments (i) and (3) were made with the same gas at Cambridge, and they show very similar heat losses per square foot as measured in cub. ft. degrees. Jhe Bairstow and Alexander (2) and Clerk 1900 (6) values seem low, and the Clerk and Boston values (4) and (5) high. There is reason to believe, however, that the early explosion experiments of Clerk and the Boston experiments were not so reliable as the later experiments of Clerk, Bairstow and Alexander, and Hopkinson ; and if the results from vessels (4) and (5) be re- jected it will be noted that the other heat -loss values closely approach each other. Further experiments must be made to eliminate the uncertainty as to continued combustion. Meantime useful deductions DISCUSSION OF DATA 205 may be drawn as to the effect of dimensions on temperature fall by using the values suggested. Assume, for example, a series of five cubical vessels with the sides respectively o'5, i, 2, 3, and 4 ft. Their cubic contents will be respectively : 0-125, I, 8, 27, and 64 cub. ft., and their surfaces will be 1-5, 6, 24, 54, and 96 sq. ft. Assume each cube to be filled with a mass of glowing gas at a mean temperature during ^\ second of 1300° C. Take 20 cub. ft. degrees as the heat loss for each square foot, then the heat lost by the respective vessels will be 30, 120, 480, 1,080, and 1,920 cub. ft. degrees. Dividing these numbers by the contents of the vessels we have as the temperature fall in each vessel 240°, 120°, 60°, 40°, and 30° C. Assuming 0° C. to be the lowest available point for cooling the gas at 1300° C, then these temperature fall values in percentages of the total range give i8'4 per cent., 9-2 per cent., 4-6 per cent., 3-08 per cent., and 2 '3 per cent. Assume the exposure to 1300° C. to last ^ second instead of 1^, then these numbers must be multiplied by four. Multiply them : 73 '6 per cent., 36-8 per cent., i8'4 per cent., 12.32 per cent., and 9*2 per cent. These are the relative losses incurred during the period of the stroke of an ordinary gas engine by these masses of gas. From these figures it is evident that loss by cooling may be reduced indefinitely by increasing the dimensions of the engine, so that when volumes of flame, such as 64 cub. ft., are used in an engine, which is the order of dimensions of many large engines now, heat loss may be rendered practically negligible even with a non-compression engine. Taking the mean temperature 1150° C, and assuming 15 cub. ft. degrees to be the temperature fall per square foot per ^ second, we get the following numbers as the temperature falls in the respective vessels in ^ second : 180°, 90°, 45°, 30°, and 22 '5° C. The percentages of 1150° C. are as follows : 157 per cent., 7-8 per cent., 3'9 per cent., 2'6 per cent., and i'95 per cent. Obviously a smaller proportion of the total heat present is lost by working at a mean temperature of 1150° C. ; the fall of 150° C. in the mean temperature effects a considerable saving. The temperature fall in a given time for any vessel can be readily calcvdated if the heat loss per square foot exposed surface be assumed the same for all vessels ; let c = capacity in cubic feet ; s = surface exposed in square feet, and t = cubic foot degrees heat loss per square foot for the particular given time, and x = temperature fall for the time in the vessel, St 2o6 THE GAS, PETROL, AND OIL ENGINE For a cubical vessel with a side of a feet, the capacity is a' and the surface exposed is 6a', so that a' a and for a spherical vessel with a diameter of d feet the capacity is 0'5236i^ and the surface exposed is ^•i^ibd^, so that 3-1416 i^ f _ 3 -1416 it _ 6i AS o'5236i^ 0-52:^6 d That is, for cubes and spheres the temperature fall for equal circum- stances is inversely proportional to the side of the square or the diameter of the sphere. This is also true for similar cylinders and other vessels of varying dimensions. The temperature fall in similar engine cylinders of different diameters may be taken as inversely proportional to the diameter. Explosion and Cooling. Initial Pressure above Atmosphere. — The experiments appear to indicate that for similar gaseous mix- tures the maximum temperature attained upon explosion increases slightly with increase of initial pressure ; but as the rate of ignition also appears to increase slightly with increasing initial pressure, it may be that the diminished heat loss during the explosion period is sufficient to account for the difference. In the Royal College of Science experiments, for example, a mixture containing i volume of gas and 5-3 volumes of air ignited at an initial pressure of 44'i lbs. per sq. in. gave a maximum temperature of 2400° C, while in Petavel's experiments a mixture of i gas and 57 volumes air, ignited at an initial pressure of 1,094 lbs. per sq. in. gave a maximum temperature of 2483° C. Calculated on the same basis, with the same allowance for chemical contraction. Clerk's explosion experiments at atmospheric initial pressure give a maximum temperature of 3100° C. It may therefore be taken for the present that higher initial pressures do not increase the maximum temperatures of explosion to any extent sufficient to introduce a material error, if the time of explosion be sufficiently small. « The Royal College of Science vesseiwas 0'82 cub. ft. capacity, while Petavel's vessel was a sphere of O'oigS cub. ft. capacity, and maximum initial pressure on the former experiments was 55 lbs. absolute and in the latter over 1000 lbs. absolute ; so that our knowledge extends over a wide range of capacity. It may be taken, therefore, that, given sufficiently rapid ignition, maximum temperature is little if at all affected by dimensions of vessel with such mixtures. The rate of cooling, however, varies greatly with dimensions and initial pressures. DISCUSSION OF DATA 207 Cooling from the same Mean Temperature in the same Vessel with different Initial Pressures. — The Royal College of Science experi- ments made by Messrs. Bairstow and Alexander are the only experi- ments of a sufficiently complete character to enable deductions to be drawn as to the changes in the cooling curves produced by varying initial pressures. From their pressure cooling curves (see fig. 65) the writer has prepared the temperature time cooling curves shown at fig- 75- These curves show the effect of varying density on cooling from a common temperature of about 1720° C. ^~ ( ^mA . PRE SSUf iES I A 2^3 4-8 4-5 .BS.j \BS. if \, 3^Z 4_U 4-7 ■55 " ^^ \ w ,^ \ \ x ^ •\^ \ \- X \ \ N N, ^ \ X \ ^^ ^; \ N \ ^z ^ ■~3 Ji 600 TIME. SECS. Fig. 75. — Cooling from the same mean temperature in tlie same explosion vessel with different initial pressures. After explosion of mixture containing I vol. gas to 5-95 vols. air. The four curves are for initial pressures of 14-55 lbs., 247 lbs., 34-5 lbs., and 44-8 lbs. per sq. in. absolute ; all start from the common temperature of 1720° C. It is evident from these curves that the higher the initial pressure the slower is the rate of temperature fall. Take the fall of temperature in 0-5 second, the highest density gives a fall of about 480°, while the lowest gives 715° C. This is shown in the table on the following page. 208 THE GAS, PETROL, AND OIL ENGINE Temperature Fall in the same Closed Vessel for different Initial Pressures Common maximum temperature 1720° C. Initial pressure Temp, fall in 0*5 sec. lb=. abs. 44-8 34-5 247 14-55 480° c. 530° c. 630° c. 715° c. As one-fifth of a second has been used as a period of the order of a gas engine stroke, fig. 76 has been prepared to show the fall of tem- perature from 1720° C. which occurs in ^ second with the different densities. The temperature fall values are plotted against initial pressures. Temperature Fall in the same Closed Vessel for Different Initial Pressures Common maximum temperature 1720° C. Initial pressure Temp, fall in o'2 sec. lbs. per sq. in. abs. IS 39 45 480° c. 335° C. 285° C. As the mean temperatures vary considerably during 0-5 and 0-2 second, the curves shown at fig. 77 have been prepared, which give tem- perature falls in ^ second for different mean temperatures with varying initial pressures. The initial pressures are marked upon the figm^e. In the following table the faU of temperature in the same vessel ignited at different initial pressures is given for -^ second for mean temperatures 1600°, 1500°, 1400°, 1300°, 1200°, 1100°, and 1000° C. Fall of Temperature in the same Vessel for i sec. at different Mean Temperatures and different Initial Pressures Vessel. Bairstow and Alexander. 0-&2 cub. ft. capacity Initial pressure Temperature falls in ^V second from mean temperatures 1600° C. 1500° c. 1400° c. 1300° c. 1200° c. iioo°.C. 1000° C. lbs. abs. 44-8 34-5 247 14-55 74° C. 85° C. 139° c. 185° c. 63° c. 72° c. 109° c. 140° c. 54° C. 61° c. 85° c. 109° c. 46° c. 52° c. 68° C. 87° c. 39° C. 44° C. 54° C. 71° C. 32° c. 37° C. 44° C. 59° C. 25° C. 30° c. 34° C. 49° C. DISCUSSION OF DATA 209 Here it is evident that increEise of initial pressure causes considerable diminution of temperature fall in ^V second for equal mean tempera- tures. The incresise from practically i to 3 atmospheres diminishes the temperature fall to less than half in the 1600° and 1500° C. values, and to half in the lower values. The curves at fig. 78 show temperature fall for the different mean temperatures 1600°, 1500°, 1400°, 1200°, 1100° and 1000° C, plotted against initial pressures in atmospheres. The curves have been extended to 55 lbs. abs. to give an idea of the temperature falls in the region of such densities, which are now common in gas and petrol engine practice. Further experiments are necessary to obtain accurate \ \ \ ■^ ^ -_^ INITIAL PRESSURE. LBS- PER. SQ. IN. Fig. 76 Temperature fall in Jth sec. in the same closed vessel for different initial pressures. Common maximum temperature, 1720° C. data, but these numbers materially aid us in giving some quanti- tative accuracy to our reasoning upon internal-combustion motor problems. They show that an increase from i to 4 atmospheres density the temperature fall diminishes to less than one third at the mean temperatures 1600° and 1500°, and to less than one half at the lower temperatures. The proportional heat loss incurred at higher initial pressures is therefore less than at lower pressures, so that it is economical in a gas engine to increase the mean density of the charge. The absolute VOL. I. p 210 THE GAS, PETROL, AND OIL ENGINE N \ \^ \ ^ \+ ^ +\. \ \ \ \ INITIAL PRESSURE 1 74-55 ? 24-70 3 34-50 \ \ \+\ +\ +\ \\ \ \ \ \\ « V \ \\\ > « > ? 3 < b -iN3D saaaoaa -S03s o^/i mvj aanivbadwai DISCUSSION OF DATA 211 heat flow per square foot is greater, however, with higher pressures, as will be readily seen as follows : Take the temperature fall at 44-8 lbs. (practically 3 atmospheres) at 1200° C. as 40° C. This would mean 40 x 0-82 = 32"8 cub. ft. degrees at atmosphere ; but there is three times the weight of gas cooling, so that the heat quantity is given by 32'8 x 3 = 98-4 cub. ft. degrees. The surface of the vessel is 5-02 sq. ft. : '^~= I9'6 cub. ft. degrees per square foot of surface per .}^ second. This is a heat flow which is about 50 per cent, greater than that which occurs with an initial pressure of one atmosphere for the same mean temperature. IWrTIAL PRESSURE. LBS. PEB 80. IH. AB8. Fig. 78. — Temperature fall in terms of density for given mean temperatures From this it follows that, although the percentage heat loss is diminished, a greater heat flow takes place per square foot of metal exposed, and therefore the metal is strained to a greater extent by unequal expansion. If now the values given at p. 208 be multiplied by °- ^ = 0-163, we get heat loss per square foot expressed in cub. ft. degrees at the particular initial pressures. The value of the cub. ft. degree will, however, be proportional to the initial pressure. The table on p. 212 shows the values so obtained for 44-8 lbs. initial pressure and 14 '55. 212 THE GAS, PETROL, AND OIL ENGINE Cub. Ft. Degrees Lost in the same Vessel for i second per Square Foot OF Exposed Surface at Differing Mean Temperatures and Initial Pressures The value of the cub. ft. degree in this case being proportional to the initial pressure. Capacity of vessel 0-82 cub. ft. Surface exposed 5-02 sq. ft. Initial pressures Cub. ft degrees loss from square foot per ^^ second 1600° C. 15CX)'' C. 1400° C. 1300° C. 1203° C. :ioo° C. 1000° c. lbs. 44-8 14-55 I2-I 30-2 10-3 22-8 8-6 17-8 7-5 14-2 6-3 II-4 5-2 9-6 4-1 8-0 To realise the effect of increased density, make the same com- parison as has been done at p. 205 for atmospheric initial pressure, using the five cubical vessels of 0'5, i, 2, 3, and 4 ft. side, or o'i25, i, 8, 27, and 64 cub. ft. capacity. Take the mean temperature 1300° C. and assume loss per ^\j second per square foot to be 10 cub. ft. degrees instead of 7 '5, as given by the table ; then the temperature faU in each vessel will be 120°, 60°, 30°, 20°, and 15° C. Assuming 0° C. to be the lowest available point for cooling the gas at 1300° C, then these temperature fall values in percentages of the total range give— 9-2 per cent., 4-6 per cent., 2'3 per cent., 1-54 per cent., and I '15 per cent.; and for an exposure of ^ of a second— 36-8 per cent., 18-4 per cent., 9-2 per cent., 6-i6 per cent., and 4'6 per cent. That is, the cooling loss may be reduced at 1300° C. to 4-6 per cent, of the total temperature range if the vessel be cubical and 64 cub. ft. capacity with the moderate compression ratio of 3. Taking a mean temperature of 1100° C. and assuming the loss to be 5 cub. ft. degrees, the percentage %sses for ^ second are — 18-4 per cent., 9-2 per cent., 4-6 per cent., 3'o8 per cent., and 2 '6 per cent. With an initial pressure of three atmospheres it is therefore possible to make proportional heat loss practically negligible in a vessel of 8 cub. ft. capacity, which corresponds to the cylinder capacity of an engine well within the range of present dimensions. These experiments demonstrate that, so far as economy is con- DISCUSSION OF DATA 213 cerned, heat loss may be rendered negligible without having recourse to incandescent walls or non-conducting linings, which for a long time have formed the favourite device of the inventor who imagines that the greater part of the heat loss of an engine is through the enclosing walls. It is also evident from the curves that the rate of reduction of heat loss diminishes with increased density, so that little increased heat saving would follow greatly increased density. CHAPTER VIII EXPLOSION AND COOLING IN A CYLINDER BEHIND A MOVING PISTON In the preceding chapter it has been shown that fall of temperature per unit surface in a closed vessel depends upon — Temperature difference ; Time of exposure ; Density of heated gas ; and probably, to some extent, Capacity of containing vessel when the vessel is small. The larger vessels appear to give a smaller temperature fall per unit surface exposed in unit time, and this opens up many interesting but complex questions as to the effect of the history of the particular ex- plosions upon the rate of cooling. When a vessel is very large and the duration of cooling is consequently long, it appears probable that a surface film of the metal attains a much higher temperature than is commonly supposed possible. Bairstow and Alexander consider that after a high temperature explosion of, say, 2000° C. maximum temperature a film upon the surface rises some hundreds of degrees and affects the subsequent rate of cooling. This conclusion they base upon Mallard and Le Chatelier's and Petavel's cooling curves , both show a discontinuity which they explain in this manner. In the author's experiments, to be described in this chapter, he also finds that the interior surface of the cylinder and piston must attain relatively high temperatures compared with the temperatures of the water-jackets. It cannot, therefore, be accepted without proof that the conditions of cooling are the same in an engine cylinder behind a moving piston as in a closed vessel of constant volume. If it be assumed that the engine cylinder presents merely a succes- sion of density, temperature, surface, and capacity changes, then the temperature falls to be expected can be determined from the foregoing data. To do this, however, it will be necessary to prepare curves of mean temperature, mean density, mean surface, for equal time-intervals in order to determine the temperature fall to be expected in a given cylinder ; but at best this method assumes uniformity of conditions which are known to differ in many points. For example, the movement of the piston keeps the hot gases in motion, and whether COOLING IN ENGINE CYLINDER 215 this motion be turbulent or not depends on many circumstanceSj such as shape of combustion chamber and absolute velocity of motion of the piston. Accordingly the author, some years ago, addressed himself to the problem of measuring the temperature fall due to cooling in the engine cylinder itself. This is by no means a problem easy of solution, as the work done on the piston by the gases introduces complications difficult of elimina- tion. No attempt had previously been made to determine a cooling curve for a cylinder having in it a moving piston. The method of experiment adopted by the author was as follows : The engine selected for the first experiments had a cylinder of 14 ins. diameter and 22 ins. stroke ; the exhaust and inlet valve levers were supplied with longer pins than usual, so that the rollers mounted on these pins could be moved into or out of the range of the exhaust and inlet valve cams. When each roller was caused to slide to one end of its pin, the cam passed clear of it and the lever was not operated. When at the other end of the pin, the roller engaged with the cam and the lever operated in the usual way. A spring and trigger gear was so arranged that the rollers could be put out of range of the cams at any required instant. By this contrivance the engine could be run in its normal way in accordance with the Otto cycle either at a light or heavy load, and any given explosion could be selected for the purpose of the experiment by operating the trigger at the proper moment. It was thus possible to run the engine at its normal speed under the usual propelling explosions, and to select at any given moment any par- ticular charge, move the rollers out of the range of the cams imme- diately the charge entered, and so obtain an explosion and expansion stroke in the usual manner, with the usual charge. When the exhaust period was approached, however, the exhaust valve remained shut, and accordingly the hot exhaust gases were retained in the cylinder and compressed by the return stroke of the piston into the combustion space at the end of the cylinder. The energy of the flywheel was sufficient to keep up the rotations of the engine, with but little fall in speed during the short period of observation. The piston was thus caused to move to and fro, alternately compressing and expanding the hot gases contained in the cylinder. An indicator card taken of such an initial explosion and expansion and the subsequent series of compressions and expansions is given at fig. 79 ; ab is the ordinary compression line indicating the compression of the charge before explosion, & c is the usual explosion line, and c A the usual expansion line after explosion. At A, however, instead of the pressure falling to the atmosphere by the opening of the exhaust valve, as the exhaust valve remains closed no escape of the hot products of combustion is possible, and accordingly the return of the piston 2l6 THE GAS, PETROL, AND OIL ENGINE produces the compression line ab ; the next outward movement of the piston produces the expansion line bc, followed by the compression line C D ; expansion line d e ; compression line E F ; expansion line F G ; compression line G H ; expansion line H i, and so on. In this diagram the successive compression and expansion lines have continued to be traced until the fall of pressure due to cooling brings the contents of the cylinder at the outer end of the stroke below atmospheric pressure, when the outer atmosphere opens the valves against the pressure of their springs, and so the experiment terminates. It will be observed that cooling is proceeding during the tracing of all these lines ; had no cooling occurred or any particular expansion and compression stroke, the compression line would lie on the top of the expansion line. Con- f Fig. 79. — Clerk diagram of explosion and alternate compression and expansion of fiot gases in engine cylinder sider, for example, the moment represented by the point B when the piston is at the extreme inner end of its stroke ; then the combustion space is filled with hot gases at a temperature and pressure corre- sponding to the point B ; after a complete expansion and compression, one out- and one in-stroke of the engme, a complete revolution of the crank, the piston is again at its innermost position, and the whole of the gases are again contained in the combustion chamber at a pressure and temperature marked by the point d. This point d is lower than b, and as the weight of the gaseous contents has not changed it foUows that the temperature at d is lower than at b. The points b, d, f, h, and j thus indicate the temperatures of the gases at the same volume at intervals of one revolution of the engine. If the engine be running at 120 revolutions per minute, then the COOLING IN ENGINE CYLINDER 217 temperatures represented by the points d,f,h, and j give the successive temperature falls suffered by the contents during successive revolutionSj each lasting 0"5 second. In the same way the successive tempera- tures of the gaseous contents at the out ends of the stroke are given by the pressures at A, c, e, g, and i. Call the successive temperatures at the out end i,-,, '01 > ^02, ios, and ^ \ 1 ^- \ 1 T^ 1 1 1 1 1 — ' r -' — 1 — 1 ' — I — 1 — 1 M — Aoo eoo eoo MEAN TEMPERATURE. DECREES CEHT. DOUBLE COMPLETE STROKES. Fig. 83. — Temperature fall for double strokes plotted against mean temperatures during double strokes for diagram, fig. 81 time, has resulted in a heat flow to the walls, causing a temperature drop of 275° C. The second, third, and fourth strokes, although they occupy the same time and expose the same surface, yet expose the gases at a lower and lower mean temperature, so that the heat loss, and therefore temperature fall, becomes less and less as the gases approximate more and more to the temperature of the enclosing walls. To reason on these values it is necessary to plot the respective temperature falls against the mean temperatures existing during the double stroke in which the loss is incurred. This has been done at fig- 83, where the mean temperatures of exposure during complete double strokes is plotted against the temperature falls due to such exposure. The full vertical lines marked on the figure indicate the mean tem- perature of the pairs of expansion and compression lines b c + c d, COOLING IN ENGINE CYLINDER 221 DE + EF, FG+ GH^and HI + I J, while the dotted vertical lines repre- sent the mean temperatures of the single expansion strokes B c, d e, FG, and HI. The temperature falls given are those incurred in a double stroke at the particular mean temperature, whatever it may be, so that if the temperature fall in a single stroke is required, those temperature fall values must be divided by 2. Consider the expansion lines. From the figure it is seen that the mean temperature of the expansion line BC is about 930° C, and the temperature fall due to heat flow from the gas in two strokes at this mean temperature is 320° C. — that is, in a single stroke the temperature fall would be 160° C. Now, as the engine is running at 160 revolutions 4. — Clerk diagram arranged for calculating temperature fall at ~ of the piston stroke per minute, each revolution takes 0'375 second and each stroke o-i87 second, so that this is the temperature fall in this engine during a complete stroke at 930° C. mean temperature of o'i87 second duration. At 1000° C. mean temperature the temperature fall would be about 180° C, and at 1200° C. about 260° C. By this method it is possible to determine the law of temperature fall, and so to calculate the tem- perature fall which would truly represent the heat loss on the explosion expansion line. It is true that there is some turbulent motion in the gases during explosion and for some time after, but this method appears to give values which are very close to the truth, as will be seen later. To get the heat loss temperature fall on the explosion and expan- sion line of an ordinary indicator diagram, it is only necessary to 222 THE GAS, PETROL, AND OIL ENGINE calculate the mean temperature in time on the line, and the tempera- ture fall may be deduced from such a diagram as fig. 83. The examination of the expansion line B C has shown that its mean temperature is about 930° C, and its temperature fall due to heat flow through the cylinder walls 160° C. in the completed stroke. The question now arises, In what manner is this temperature fall incurred ? Is it due to a uniform loss throughout the stroke, or does the tem- perature fall more rapidly at one part of the stroke than at another ? We know from the previous chapter that increase of surface exposed increases heat flow and that increase of density also increases flow, but it would be difficult to predict what would occur throughout the stroke from the closed vessel experiment. The question, however, can be answered by the author's method of experiment. The cooling curves, which have been already drawn (figs. 82 and 83), are those proper to complete revolutions. In order to answer this, it is necessary to deduce cooling curves for part of the stroke only, and for this purpose the first /V of the piston stroke has been chosen. Looking at fig. 84, it will be seen that a vertical line marked j% inter- sects the various compression and expansion lines; the partial com- pression lines have been marked respectively ic, 2c, 3c, 4c, &c., and the partial expansion lines have been marked le, 2e, ^e, ^e, &c. Starting from the vertical j-^ line, at A', the partial compression line IC terminates at the point B, and the expansion line le passes from b to the -ny line at the point b'. In passing from a' to b' the hot gases have been compressed to b and expanded to b', and the temperature has fallen from a' to b' ; the corresponding temperature fall, however, does not represent all the temperature faU due to heat flow, because the work done by the piston in the gas during the compression a' b is greater than the work done on the piston by the gases during the expansion b to b'. This difference has disappeared as heat, and the temperature fall equivalent of the area a' be' must be added to the apparent temperature fall a' — b' to get the true value. By treating the other compression and expansion lines 2c, ze; 3c, 3^ ; 4c, 45, &c., the cooling curve shown at fig. 85 has been prepared as the mean of three experiments at full load cards under the conditions of fig. 81. ^ From the cooling curves for the whole stroke and fV stroke, it is now possible to consider the division of temperature fall between the first 0-3 and the last 07 of the stroke. Consider first the expansion line b c . From fig. 83 it wUl be found that during the whole stroke its mean temperature in time is 930° C, and the temperature faU during the stroke due to heat flow from the hot gases to the colder walls is 155° c. COOLING IN ENGINE CYLINDER 223 From fig. 85 it will be found that the part b b' of this expansion line — that is, xe — has a mean temperature in time of 1190° C, and the tem- perature fall for this part is ^ g^ = 92'5°. N^ \ s. ai+oi \ \ SI -^ \ \ \ sz+oz \ \ 9Z --^ \ se+3e \ \ 3£ ■■\ \, \ V 3t ""\ \ g ■^. '1N33 S33i!33a 3>iOdis aisnoQ °i/e iivj stmivuadwaj. That is, while the temperature fall due to cooling on the whole stroke is 155° C, of that total fall 92°-5 C. is incurred in the first 0-3 of the piston stroke. In this particular line 59'7 per cent, of the 224 THE GAS, PETROL, AND OIL ENGINE temperature fall due to cooling loss is incurred in the first 0-3 of the stroke, and only 40-3 per cent, in the last 07 of the stroke. Comparing the two portions of the expansion line d e in the same manner, it is found that 59-3 per cent, of the total fall is incurred on the first 0"3 and 407 per cent, in the last 07 of the stroke. 80 o o o o n Q O if ■1N3D S33aD3a '(310NIC) 3>IOdiS 3nOHM 'IIVJ 3aniva3JH3J. COOLING IN ENGINE CYLINDER 225 In these two expansion lines the temperature fall may be taken as 60 per cent, in the first 0-3 and 40 per cent, in the last 07. Comparing the two parts of f g in this manner, it is found, however, that 48'5 per cent, temperature fall is incurred in the first 0'3 and 5i'5 per cent, in the last 07. Arranged in tabular form, the particulars for these]^three expansion lines are as follows : Mean temp. Temp, fall due in time to heat loss 'I 3l Complete expansion line B c . . 930° C. 155° C. Partial expansion line bb' Complete expansion line, D E Partial expansion line dd' . . 900° C. 55° C. 560° C. 65° C. 700° C. 3i°-2 C. Complete expansion line F G Partial expansion line F f' Lines i and 2, division of temperature fall 60 per cent, in first y^^ ; 40 per cent, in last ^'^ of stroke ; and line 3, say, 50 per cent, in first jV ; 5° per cent, in last /^ of stroke. The division is practically the same in the first two lines, but the ratio alters in the third, where the mean temperatures fall enough to approximate more closely to the temperatures of the valves and piston surface. Where this occurs the mean temperature of the enclosing waUs is so much higher at y^j-stroke as to affect the law of cooling. That this is so will be seen from experiments which give approximate determinations of the actual mean temperatures of the cylinder waUs at different parts of the stroke. These figures clearly show that in these expansion curves at the higher mean temperatures a larger temperature fall is incurred at the first 0*3 of the stroke as compared with the remaining 07, notwith- standing the fact that a greater surface is exposed in the latter part. The difference between the distribution in the first and third lines shows, however, that the distribution must not be taken for granted — it will vary with many circumstances. From these curves it is possible to reason upon the explosion expansion line in order to arrive at an approximate value of the temperature fall distribution throughout the stroke. By deahng with the card shown at fig. 81, and measuring the mean temperature in time of the explosion and expansion line be A, see fig. 84, then that of the explosion and partial expansion line b c A^ using curves figs. 85 and 86, the following values will be obtained : Mean temperature on explosion expansion line = 1250° C. Temperature fall corresponding . . . . = 257° C. Mean temperature on explosion and partial ex- pansion line to y^ = 1410° C. Temperature fall corresponding . . -f- = 128° C. VOL. I. Q 226 THE GAS, PETROL, AND OIL ENGINE That is— Temperature fall on first 0-3 of stroke= 138, say 50 per cent, and „ „ last 07 ,, 129, say 50 per cent. In this case the temperature fall is equally divided : half is incurred in the first 03 and half in the last 0*7 of the stroke. If the ignition had been more rapid, however, so that the maximum pressure was attained before the piston had moved out appreciably, then the dis- tribution would have been different, because the mean temperature of the explosion expansion line for the first ^s of the stroke would have been higher. Every variation in the explosion curve will produce a corresponding variation in the distribution of temperature fall. To compare the effect of the increased surface due to the outward position of the piston only, it is necessary to compare the temperature falls incurred in equal times and equal temperatures for the f ^^ piston movement and the whole piston movement of a complete out stroke. To do this a considerable number of diagrams have been measured for 120 revolutions per minute with the engine cold — that is, no load upon it, and only ignitions sufficient to keep it running at the speed while the water was freely passed through the jacket at a tempera- ture of about 13° C. — and also for 160 revolutions with the jacket hot, water leaving about 80° C, while the engine carried a load of 50 BHP. The curves shown at fig. 87 give the temperature falls incurred per second for different mean temperatures calculated in time. The par- ticulars are fully given under the diagrams. Consider, first, the curves a a' ; these correspond to the conditions in the engine cylinder when the engine is running at 120 revolutions per minute without any load, so that very few ignitions keep it in motion, and while the water jacket is kept cold by running water at 13° C. freely through it. Under these conditions the end of the piston and the interior surface of the valves wiU be but little heated by the explosions, and the interior surfaces of the cylinder and combus- tion space waU will tend to fall beH^een the explosions to the water- jacket temperature of 13° C. The curve a is that for the complete strokes, while a' is for the partial first Vlr of the stroke. Taking a first, it appears that the rate of temperature fall per second for a mean temperature of 1300° is 1460° C. ; that is, that a complete stroke of the engine during which the mean temperature in the cylinder was 1300° C. would lose heat to the walls at a rate sufficient to produce a temperature fall of 1460° C. per second. Of course the stroke does not last a second, but only COOLING IN ENGINE CYLINDER 227 a quarter of a second, so that the real temperature fall under these conditions would be '-^-O = 365° C. It is better, however, to express the temperature fall in degrees per second. On this curve the following mean temperatures — 1200", 1100°, \ \, ^^ x> V "to ■--«^ ^ X x^ \ \ V \ \ <\ :? \ ^ -^ ^ \ \ V \ > ^\ \ \\ ^^^ \^ ''A ^ ~ 2^ IS. i %S Q. O is §3 s.o S OJ y *" is a 8 <" ^■5 — ' O s 3'» ^ a. ■XN3D saawDsa oas 'aaa nvj aanivaadnai Q 2 228 THE GAS, PETROL, AND OIL ENGINE 1000°, 900° — give respectively 1250°, 1080°, 920°, and 780° C. temperature fall per second. It will be observed that the curve is concave, so that the rate of temperature fall increases with the increase of temperature. This curve, when prolonged to the zero line of temperature fall, cuts it at the temperature of 65° C, which means that when the gases within the cylinder fall to the mean temperature of 65° C, no further heat loss occurs to the enclosing walls — that is, the mean temperature of the enclosing walls for the complete stroke of the engine must be 65° C. The water outside does not succeed in keeping down the mean temperature of the interior surface, including valve surfaces and piston end, to its own temperature of 13° C. This curve thus gives an interesting indication of the temperature of the walls. Taking curve a' in the same way, its general slope is greater than a, and it cuts the zero line at 165° C, so that, although its rate of change is greater, it does not cross the curve a until a mean temperature of 610° C. is attained. Up to this point the temperature fciU is less for a given mean temperature, above it the temperature fall rapidly increases with increasing mean temperature, so that the rate of fall is con- siderably greater in a' than in a at the higher mean temperatures of 1200° and 1300° C. The fact that the intersection with the zero line is at 165° indicates that the mean temperature of the cylinder walls is much higher for the inner /^ of the stroke than for the whole stroke. This was to be expected, because the proportion of piston end, valve surface, and other un jacketed surfaces becomes greater as the piston moves in — that is, the water- jacketed cylinder surface is covered as the piston moves in, so that the ratio changes, and therefore the mean temperature rises. Apart from this, however, it was to be expected that the surface temperature of the jacketed and unjacketed parts would be highest at the combustion chamber end, where it is exposed to the maximum temperature. This curve, then, indicates that the mean temperature of the ~p,j surface during the fV period is 165° C, while the mean tem- perature of the whole cylinder surface during the whole stroke period is only 65°, and this with the water in the water-jacket at 13° C. Compare now the two curves % a' at the mean temperatures following : Mean temperature . . . 1300° C. 1200° C. iioo°C. 1000° C. 900° C. Temperature fall per sec, curved 1460 1250 1080 920 780 Temperature fall in 0-25 sec, I ^g^ ^^^ 270 230 195 curve a i Temperature fall per sec, curve a' 1970 1570 1285 1060 870 The rate of loss of temperature is less in curve a by the following percentages, calculated on the a values for the successive temperatures : COOLING IN ENGINE CYLINDER 229 nearly 26 per cent., 20'5 per cent., 16 per cent., 13 per cent., and 10 per cent. This clearly shows that, notwithstanding the diminutions of surface exposed in a given time at t:% stroke, as compared with whole stroke, the absolute temperature fall rate is increased. This is probably due to the fact that the mean surface exposed diminishes more slowly than the mean density increases, so that the economical effect of increased density, discussed in a previous chapter, is not realised. The existence of some turbulent motion may also mask the other effects. This clearly shows that it is necessary to determine cooHng in the cylinder with the moving piston, as the conditions are too little known to be entirely predicted from explosions in closed vessels of fixed capacity. This becomes even more evident when the curves b b' are studied. These curves b b' are taken while the engine is running at a load of 50 BHP. Here the explosions were almost consecutive and the water-jacket temperature was 80° C, so that the interior surfaces, both jacketed and unjacketed, were much hotter than in curve a a' ; as before, curve b is for whole stroke and b' for first ^Ij of stroke. Curve b cuts the zero line at 190°, indicating the mean temperature of the enclosing walls during the whole stroke. The condition of the internal surfaces is different — 190°, as compared with 65°, a very large increase in temperature. Curve b' cuts the zero line at 400°, also a much higher temperature. The continued explosions and the hot- water jacket have produced a very considerable change upon the surface temperature of the enclosing walls. Here also the curve b for the whole stroke is less steep than b' for the -[% stroke, but, owing to the high wall mean temperature at f'V, they do not cross tUl the temperature of 1100° C. is passed. Below this temperature the tem- perature fall is less in b' ; above that temperature it is greater. Comparing b b' as a a' have already been treated at the same mean temperatures : Mean temperature .... 1300° C. 1200° C. 1100° C. 1000° C. 900° C. Temperature fall per sec, curve 6. . 1580 1340 1130 950 780 Temperature fall pero'iS/sec.curveb 290 250 211 177 146 Temperature fall per sec, curve 6' . 1650 1370 1125 920 730 For 1300° and 1200° C. the temperature fall is greater in b' than in b, but only 4 per cent, and 2 percent, respectively, calculated on b' -, at 1100° C. the temperature fall is practically equal ; at 1000° b' is less by about 3 per cent, on 6', and 900° C. is nearly 7 per cent, less, also calculated on the value of b'. It is thus seen that for the practically interesting range of mean temperatures 1300° to 900° C. the temperature fall given by the two curves varies but little. That is because of the high average 230 THE GAS, PETROL, AND OIL ENGINE temperature of the interior surface. The point of intersection of the curves b and b' is raised to such a high value that the actual heat loss about these temperatures remains nearly constant, although the two curves are undoubtedly different in their slope. The temperature faU on b is also given for o'iSy second, as this is the period of one stroke of the engine at i6o revolutions per minute. These numbers are therefore the temperature falls incurred during one forward stroke due to heat flow to the walls under working conditions. Compare now the curves a' b', and it will be seen that they are fairly parallel one to the other. Their general slope is very similar, so that, except at the lower end, they could almost be superimposed. It appears as if the difference in absolute value of the temperature fall for given mean temperatures were due mainly to the difference between the enclosing wall temperatures. These experiments prove conclusively that cooling behind a moving piston depends largely on the varying temperature of the water jacket, and stiU more upon the varying mean temperatures of the cylinder walls according to the condition of load and water circulation. And although general laws may be deduced from closed vessel experiments of fixed volume at varying initial pressures, yet the problem in the working engine is so complex that it is desirable to make many direct determinations on actual engines as to explosion and cooling in cylinders of varying dimensions before endeavouring to deduce any general formulje. It is useful, however, to compare the temperature falls so observed with those in closed vessels of initial atmospheric pressure given in the previous chapter at the working condition of heavy load, and for that take the curve b full stroke at the following mean temperatures, reducing the temperature falls to ^V second instead of i second, and calculating the cubic foot degrees per square foot at the temperature of charging as given below : Mean temperature . . . . • 1400° C. 1300° C. 1150° C. Temperature fall in i sec 93"5 79 61-5 Cubic foot degrees per sq ft. surface, piston assumed 1 j_| full out ' The capacity of the cylinder and combustion space when the cylinder is fuU out is 2-41 cub. ft., and the surface exposed is ii-2 sq. ft., so that cubic foot degrees are obtained by multiplying the respective temperature falls by ?4- = 0-215. The temperature fall values obtained from the cooling curves of Hopkinson's large vessel of 6-2 cub. ft. capacity (see p. 204) in cubic foot degrees for the same three mean temperatures are: 24-4 -i8'6 and I3'6. COOLING IN ENGINE CYLINDER 231- so that the temperature falls here given closely resemble those which would have been obtained in a closed vessel of 2"4i cub. ft. capacity exposed for the same time to the cooling walls. It is to be noted that the absolute value of the cubic foot degree used in the moving piston experiments is less than the cubic foot degree in the closed vessel experiments, because the mean temperature of the engine charge is 95° instead of 16° C, which makes the value for the former ^73 + — = o'7q that of the latter. 273 + 95 DENSITY. LBS. PtH. SQ. IN. Fig. 8S. — Proportional heat flow at different densities and temperatures, calculated from Bairstow and Alexander's Experiments This does not, however, affect the reasoning as to relative cooling, but applies only where absolute rates of heat flow are to be compared. It has been shown that absolute heat flow to the enclosing walls increases with the density of the gas exposed to unit surface for unit time, but for any given space the flow increases at a slower rate than the density, so that at higher densities the temperature fall rate is less than at low densities. Broadly, the absolute heat flow increases from I to i'5 when the density increases from i to 3 at temperatures of 1200° and 1300° C, such as are common mean temperatures for the working stroke of a gas engine. 232 THE GAS, PETROL, AND OIL ENGINE To understand the effect of the moving piston on the cooling curve, it is desirable to formulate approximately the effect of density upon absolute heat flow through the walls. This can only be attempted on a rough approximation in the present state of experimental knowledge, but it is useful to attempt it. The curves shown at fig. 88 have been prepared from the table at p. 208, as deduced by the author from Bairstowand Alexander's experi- ment. The temperature faUs there shown are those occurring in the explosion vessel at varying pressures before ignition, so that the actual heat flow for a given temperature fall is greater at the higher initial pressures. To compare the absolute heat flow at these temperatures and pressures, it is necessary to convert the numbers by taking density into account for the pressures 247, 34-5, and 44-8 lbs. absolute ; the temperature falls corresponding have been multiplied by -^^-1 , -24 5 44-8 ^4'55 i4'55 3-iicl — — ^. Taking the lower pressure, I4"55 lbs. per sq. in., as the atmospheric pressure at the time of the observations, the numbers so obtained give the absolute, not the relative, heat losses. They are plotted against density in pounds per square inch. It will be observed that the results through the observations for iioo", 1200°, 1300°, 1400° and 1500° C. are represented by straight lines, and that these lines are all nearly parallel to each other. , The observations require repetition to eliminate some discrepancies, but broadly they are properly represented within the limits of accuracy of the experiments. These values of temperature faU multiplied by density are repre- sented approximately by the formula td=t+ {d—T.) 25, Where t = temperature fall at the mean temperature during ^^ second ignited at atmospheric pressure; d = density, in atmospheres and 25 the value of a constant ; td = the required value at the density d in atmospheres. The heat flow in cubic foot degrees per square foot at atmospheric pressure and temperature of the experiments is found from this by the formula H/= (i!+(^-i)25 )f fc _^ {d -i)25 c s *' s s where h/= heat flow per square foot exposed to walls in cubic foot degrees at atmospheric pressure and temperature; t = temperature fall in vessel in ^jV second at atmospheric density and temperature; d = density before ignition in atmospheres ; c = capacity of vessel in cubic feet ; s = surface exposed in square feet. COOLING IN ENGINE CYLINDER 233 Bairstow and Alexander's vessel was of 0*82 cub. ft. capacity and 5"02 sq. ft. internal surface exposed. Assume that it is desired to compare heat flow at 3 atmospheres initial pressure with i atmosphere. It will be observed that the expression ^ ~ ' — ^— does not include t, so that it is independent of that temperature fall, where (^ = 3, c = 0-82, and s = 5-02 (d - I) 25 c ^ g.^ s t for 1000° C. is 49, so that U _ 49 X 0-82 _ g s 5-02 The loss per square foot at 1000° at atmospheric pressure is 8, see table p. 212, and at 3 atmospheres it is i6-i. At 1300° C. atmospheric density loss 14-2; at 3 atmospheres 14-2 + 8-i = 22'3. Assuming the formula to hold at 6 atmospheres, the value for 1300° C. is 14-2 + 20-2 = 34-4 cubic foot degrees. It is now possible to consider the effect of the moving piston, and it is desirable to take as a case the cylinder of the National Gas Engine Company's engine which has been discussed. This cylinder is 22 ins. stroke and 14 ins. diameter. When the piston is full out, the total interior surface is ii'2 sq. ft.; when the piston is full in, the surface enclosing the explosion space, including the surface of the piston end, is 4-5 sq. ft. The cylindrical surface swept by the piston is 67 sq. ft. Assume the compression ratio to be g and the density of the charge when fully compressed 6 atmospheres. If the piston remain fuU out, and a charge of gas and air be fired at atmospheric pressure, the piston remaining fixed whUe cooling goes on'; our experiments prove that the heat flow for 1300° C. mean temperature during ^V second wUl be I4'2 cub. ft. degrees per square foot. That is, the cylinder fuU of gases will lose I4'2 x ii'2 = 159 cub. ft. degrees in ^V second. If the piston remain fuU in, and a charge of gas and air be fired at a density of 6 atmospheres, the piston remaining fixed while cooling goes on, our numbers show that the heat flow for 1300° C. mean temperature during ^/d second wUl be 34*4 cub. ft. degrees per square foot. That is, the cylinder fuU of gases will lose 34-4 x 4-5 = I54"8 cub. ft. degrees, but little less than that lost to the larger surface in the same time. If the increase or diminution of surface exposed in an engine cylinder followed the inverse law of the diminution or increased flow due to density, then, so far as surface and density change affected the matter, the same absolute heat loss would be suffered by a charge from equal mean temperature at every point of the stroke. Surface always increases with diminution of density, so that the changes cancel 234 THE GAS, PETROL, AND OIL ENGINE out to some extent, but the laws of surface change vary with the proportions of the cylinder diameter and stroke, and also with the configuration of the explosion space, so that engines must differ in this respect. Diagrams could be constructed, however, for any engine on a time base dividing up to the period of the stroke into ten parts, and mean surface, mean density, and mean temperature could be calcu- lated for each part, so that heat flow could be calculated for the whole stroke in any particular case. The data available from closed vessel experiments are not yet sufficiently complete and concordant to enable such calculations to be carried further with advantage to the engineer. The Clerk diagram, when more fully studied and applied, will enable full information to be ultimately obtained. So far, the measure of heat quantity used in the present and two preceding chapters is the cubic foot degree, and no attempt has been made to deduce its value in any standard heat unit or in foot-pounds. Not only is its value unknown, but the question as to variation with temperature is so far left open. If a gas be compressed without gain or loss of heat from volume F„ to F, , and the temperature rises from T„ to T, , so that the work done upon the gas is W , then the mean specific heat C.^ of the gas per unit volume at o° and 760° mm. at constant volume between the tempera- tures is : C = - ^- where if/^ is a constant depending on the quantity of the gas in the cylinder. This is also true of expansion as well as compression. The dynamical value of the rise or fall of 1° C. for 1 cub. ft. of the gas will be given by the same formula : where W is the work done by or on the gas in foot-pounds, V is the volume in cubic feet, and D^ is the dynamical value in foot-pounds. It is evident that the method of oj^ration described in this chapter affords the means of determining the heat loss on expansion or com- pression lines, and so permits the temperature fall or rise, due to work done, to be determined at any temperatures. In this way the author has experimented on the products of combustion contained within a gas-engine cylinder, and has deduced the values of D^ for that working fluid at different temperatures. The experiments were numerous, and many difficulties were en- countered, for the full discussion of which the reader is referred to the APPARENT SPECIFIC HEAT OF WORKING FLUID 235 paper describing the experiments read by the author before the Royal Society in 1906.' It is sufficient here to give the values obtained for the working fluid, which was of the following composition Steam (assumed gaseous) Carbon dioxide . Oxygen .... Nitrogen .... II-9 volumes 5-2 ,, 7-9 J 75-0 n loo-o volumes Table of Apparent Specific Heats (Instantaneous) in Foot-Pounds PER Cubic Foot of Working Fluid at 0° C. and 760 mm. Temperature Specific heat at constant volume Temperature Specific heat at constant volume °C. 100 200 300 400 500 600 700 ft. -lbs. 19-6 20-9 22-0 23-0 23-9 24-8 25-2 257 "C. 800 900 1000 IIOO 1200 1300 1400 1500 ft. -lbs. 26-2 26-6 26-8 27-0 27-2 27-3 27-35 27-45 Table of Mean Apparent Specific Heats in Foot-Pounds per Cubic Foot of Working Fluid at o°C. and 760 mm. Temperature Specific heat at constant volume Temperature Specific heat at constant volume °c. — 100 — 200 0—300 0—400 0—500 — 600 — 700 — 800 fi.-lb<. 20-3 20-9 21-4 21-9 22-4 22-8 23-2 23-6 °C. — 900 — 1000 — IIOO — 1 200 — 1300 — 1400 — 1500 ft. -lbs. 23-9 24-1 24-4 24-6 24-8 25-0 25-2 These tables enable a definite value to be given to the cubic foot degree of gas-engine mixture, which has been dealt with in this chapter, and they show that for various reasons the apparent specific heat values increase considerably with temperature. These numbers will be used later in this work, together with the other data arrived at in this chapter. ' Proceedings of the Royal Society , A, vol. 77, 1906, p. 499. CHAPTER IX THERMAL AND MECHANICAL EFFICIENCY OF THE DIFFERENT TYPES OF GAS ENGINE IN USE In the previous chapters the different types of gas engine have been considered as air engines — that is, engines in which the working fluid is atmospheric air of constant specific heat throughout the temperature range, and the theoretic efficiencies under varying conditions have been calculated and compared on this assumption. The phenomena of gaseous explosion have also been studied, and the properties of the working fluid, the result of a gaseous explosion, have been discussed. The way is now clear for the study of the results obtained from the engines in practice, and in this chapter it is proposed to discuss, so far as is possible, tests made by observers independently of the engine constructors. It is proposed to deal with standard tests mainly. Type I The most important engines of this type which have been in public use are those of Lenoir and Hugon and their mechanical arrangements have been sufficiently described in the Historical Sketch. As they have long disappeared from practice, it is undesirable to deal with their results at any length. It is sufficient to say that Prof essor Tresca's experiments made in Paris with a Lenoir engine nearly fifty years ago showed an indicated efficiency of only 4 per cent. That is, the engine received, say, 100 heat units as inflammable gas, and the indicator diagrams obtained showed that only 4 heat units of the hundred appeared as work done by the expanding gases upon the piston. For the con- ditions of the diagrams obtained by Tresca, and later diagrams obtained by Slade, it is found that a corresponding air-engine diagram without heat losses would give iy$ per cent-down to i2'6 per cent, efficiency, so that the losses in the Lenoir engine were undoubtedly very great. The theoretical air-engine efficiency was poor, and its realisation in practice was still poorer. Later experiments by Professor Tresca on the Hugon engine showed somewhat better results. Experiments were made by the author in 1884 upon a Bischoff engine, and the indicated thermal efficiency was found to be somewhat lower than 4 per cent., Professor Tresca's number. THERMAL AND MECHANICAL EFFICIENCY 237 Type ia The most important engines of this type were those of Barsanti and Matteuccij 1857, and Otto and Langen, 1866 ; they are found fully described in the Historical Sketch. The Otto and Langen engine attained considerable success in practice, and the author made a number of tests with it in 1885. The engine tested was of 2 HP ; its cylinder was I2'5 ins. diameter, and its longest observed stroke was 40'5 ins. ; working at the rate of 28 ignitions per minute, the indi- cated power was 2" 9 horse ; the consumption of Oldham gas was 24-6 cub. ft. per IHP hour. The BHP was 2 horse, and the brake consumption was 36 cub. ft. per BHP hour. This did not include the consumption of the gas-supply to the igniting flame, which was 12 cub. ft. per hour. The thermal efficiencies were as follows : Indicated thermal efficiency = 16 per cent. Brake thermal efficiency = 11 „ The largest engine built of this type was of 3 HP, and it is probable that its brake consumption was about 30 cub. ft. of Oldham gas per BHP hour, which would be equivalent to : Brake thermal efficiency . 13 per cent. Professor Tresca tested a half-horse engine at the Paris Exhibition in 1867 ; it gave 0"456 brake horse and consumed Paris gas at the rate of 44 cub. ft. per BHP hour. The gas consumed by the igniting flame is not included in this figure. It was a great matter to realise so much as 16 per cent, of the total heat supplied in indicated work and 11 per cent, in brake work at the early date of 1867 ; but the cumbrous nature of the engine and its noisy action prevented its general adoption, although its historic interest is great. The theory of the action of this engine is some- what fully discussed in the earlier editions of this book, but its present interest does not warrant further mention here. Type II In engines of this kind compression is used previous to ignition, but the ignition is so arranged that the pressure in the motor cylinder does not become greater than that in the compressing pump. The power is generated by increasing volume at constant pressure. Engines of Type II are therefore : Engines using a mixture of inflammable gas and air compressed before ignition and ignited in such a manner that the pressure does not increase, the power being generated by increasing volume. 238 THE GAS, PETROL, AND OIL ENGINE These engines are truly slow- combustion engines ; in them there is no explosion. The two engines which found some practical use between 1876 and 1880 were those of Brayton and Simon. They are sufficiently described in the Historical Sketch. The only engine of Type II now in public use is that of Diesel, and its use so far has been confined to oil engines. As it uses the Otto or four-cycle system of mechanical operation, it is proposed to discuss its thermal and mechanical efficiency after the Otto cycle engines in the consideration of Type III. Early tests were made, however, by the author on a Brayton engine, which will now be described. The author's test was made at the Crown Ironworks, Glasgow, on February 21 and 22, 1878. The engine was made by the New York and New Jersey Ready Motor Company. The motor cylinder was 8 ins. diameter and the stroke 12 ins. ; the pump cylinder was also 8 ins. diameter, and the stroke 6 ins. The details of the engine tested are shown at figs. 9, 10, II, and 12 in the Historical Sketch. Diagrams were taken from both pumps and motor by a well-made Richards indicator. At the same time the brake was applied to the fly-wheel, fully loading the engine ; readings were taken at regular intervals. The revolutions were recorded by a counter. The petroleum used was measured in a graduated glass vessel. The results are as follows : Test of Brayton Petroleum Engine 1878. {Clerk) Petroleum consumed during one hour . . . i -378 gallon Mean speed of engine 201 revs, per minute Mean brake reading ... . 4-26 HP Mean pressure, power cylinder ... 31 lbs. per sq. in. Mean pressure, air pump ... . 27-6 lbs. per sq. in. Piston speed, motor . .201 ft. per minute Piston speed, pump . . 100-5 ft- P^r minute Power indicated in motor 9-49 HP Power indicated in pump . 4-10 HP Available indicated power . • 5 "39 HP The power by the dynamomqger is 4-26 horse ; therefore the mechanical friction of the engine is 5-39 — 4-26 =I'I3 horse. Consumption of petroleum . . . 0-255 galls, per IHP per hour Consumption of petroleum . . . 0-323 galls, per actual HP per hour Figs. 89 and 90 are diagrams from the motor and pump, which are fair samples of those taken. It will be observed that considerable throttling occurs in entering the motor cylinder ; the pump pressure is higher than the reservoir pressure, and the motor pressure is lower. THERMAL AND MECHANICAL EFFICIENCY 239 so that a double loss has bean incurred. The principle of the engine is so good that the author anticipated better results. Great improve- ments could be obtained by reproportioning the valves and air- passages ; they are in this engine much too small and cause needless resistance and loss. The maximum pressure in the motor cylinder is 48 lbs. per square inch, which remains steadily till the inlet valve shuts at four-tenths of the stroke : the pressure then slowly falls as the gases expand, the exhaust valve opening at about 10 lbs. per square inch above atmosphere. r M Ji-S US fiO Si 23 42 H Mean pressure 30'2 lbs. per sq. in. 8 ins. dia. cylinder. Stroke 12 ins. 200 revs, per mln. , Fig. -Brayton Petroleum Engine. Motor Cylinder The average available pressure upon this diagram is 30' 2 lbs. per square inch. The air pump shows a maximum pressure of 65 lbs. per square inch, the reservoir pressure being 60 lbs. The average resistance is 27-6 lbs. per squai'e inch ; as the pump is half the stroke of the motor and equal to it in area, the pressure to be deducted is 27*6 ' ~ = 13-8 and 30'2 — I3'8 = i6'4. The actual available pressure actuating the engine is therefore only i6'4 lbs. per square inch. The / -^\ 63 , 63 / 60 J — TT /7 f7 23 33 ^ y Mean pressure 27"6 lbs. per sq. in. 8 ins. dia. cylinder. Stroke 6 ins. 200 revs, per min. Fig. 90. — Brayton Petroleum Engine. Pump Cylinder effect of the clearance in the pump cylinder is noticeable upon the diagram ; the air-inlet valve does not open till one-tenth of the down- stroke is completed. The theoretical efficiency of this type, with a maximum tempera- ture of 1600° C, compression of 60 lbs. per square inch above atmo- sphere, and motor cylinder of twice the pump volume, is 0-30. 240 THE GAS, PETROL, AND OIL ENGINE But the diagram is imperfect in many ways. Using the mixture it does, the diagram should show a maximum temperature of 1600° C. at least ; in reality, the highest temperature is only 840° C. The fiame is entering the cylinder at an actual temperature of 1600° C. during the whole period of admission, but the convection has so greatly increased by the mixing effect of the entering current that greater cooling results ; accordingly, when the gases are fuUy admitted and the inlet valve is closed, the gases have only a temperature of 840° C. instead of 1600° C. After admission ceases, the expansion line from 45 lbs. to 10 lbs. pressure is far above the adiabatic, indeed it is iso- thermal, the combustion is proceeding and the sinall igniting flame also is helping to sustain the temperature. It is therefore quite evident that the loss of heat is much greater than that occurring during explosion in equal time. The specific gravity of the petroleum was 0'85, therefore the weight of one gallon is 8"5 lbs. As 0'255 gallon is burned per IHP per hour, this amounts to 8'5 x 0'255 = 2"i6 lbs. of liquid fuel per IHP per hour. One pound gives out 11,000 heat units, and for one horse- power for one hour 1,424 units are required ; the actual indicated efficiency is therefore -^-^y-y = '4 7- _ o-o6 nearly; that is, 6 per cent, of the 2-i6 X 11,000 23,760 whole heat given to the engine is accounted for by the power developed in the motor cylinder. If there were no losses of heat to the cylinder, or losses by throttling during the inlet and transfer of the air from the purap to the motor, or loss of heat from the reservoir to the atmosphere, then the efficiency of this type of engine would be 30 per cent. These losses in practice reduce it to 6 per cent. The cycle is a good one, and under other circumstances is capable of better things, but it is unsuitable in this form for a cold-cylinder engine. Cooling and undue resistance are the main causes of the great deficit. The gases entering the cylinder as flame in passing through the inlet chamber expose a large surface to the action of the water jacket ; the entering currents also impinge against the piston, causing more rapid circulation than ordinary Convection. Both causes intensify the cooling action of the cylinder walls. In the engine tested by the author the communicating pipes and the motor admission valve were much too small ; a considerable loss of pressure resulted ; although the reservoir pressure was 60 lbs., that in the cylinder never exceeded 48 lbs. above atmosphere, showing a loss of 12 lbs. per square inch from undue resistance. To enable this engine to realise the advan- tages of its theory considerable modifications in its arrangements ai'e required. THERMAL AND MECHANICAL EFFICIENCY 241 Type III Engines of this kind resemble those just discussed in the use of compression previous to ignition, but differ from them in igniting at constant volume instead of constant pressure ; that is_, the whole volume of mixture used for one stroke is ignited in a mass instead of in successive portions. The whole body of mixture to be used is introduced before any portion of it is ignited ; in the previous type the mixture is ignited as it enters the cylinder, no mixture being allowed to enter except as flame. In Type III the ignition occurs while the volume is con- stant ; the pressure therefore rises ; it is an explosion engine in fact, like the first type, but with a more intense explosion due to the use of mixture at a pressure exceeding atmosphere. The most obvious means of applying the method is that suggested by the Lenoir engine. The addition of a pump taking mixture at atmospheric pressure, compressing it into a reservoir from which it passes to the motor cylinder at the increased pressure, seems a simple matter. The igniting arrangements would act as in the original. As the gases are under pressure, the piston would take its charge into the cylinder in a smaller proportion of the forward stroke, and so more of the motor stroke would be available for useful effect. The diagram such an engine should produce is seen at fig. 26, p. 74 ; the shaded part is the available portion, the other part is the pump diagram. The theoretic efficiency of such an engine is as good as the type can give. The patent list shows that it was the first proposed after Lenoir. Many such engines have been attempted and have given very good results economically, but the difficulties of detail are considerable, the greatest being the necessity for the intermediate reservoir. Million's patent 1861 proposes to do this, the present author also constructed one of this kind in 1878, and later one was made by Mr. Atkinson. The difficulties, however, are too great to allow the success of small motors on the plan. Mr. Otto, the first to succeed with the free piston engine, was also the first to succeed in adapting compression in a reliable form. In the third type are included all engines having the following characteristics, however widely the mechanical cycle may vary : Engines using a gaseous explosive mixture, compressed before igni- tion, and ignited in a body, so that the pressure increases while the volume remains constant. The power is obtained by expansion after the increase of pressure. In the Otto or four-cycle gas engine, the first to combine the VOL. I. R 242 THE GAS, PETROL, AND OIL ENGINE compression principle with a simple and thoroughly efficient working cycle, the difficulties of compression are overcome in a strikingly original manner. To the engineer accustomed to the steam engine, ttie main idea seems a bold and indeed a retrograde step. The early gas engines were moulded more upon the steam engine model and were to some extent double acting. The Lenoir and Hugon both received two impulses for every revolution, the Brayton was single acting, and the Otto is only half single acting. The steam engine in its advance passed from single to double acting, and then to four and even more impulses per revolution. The gas engine in its progress has in this respect moved backwards, beginning with double action and then going back. The gain by this arrangement, however, has completely justified the retrogression. A single cylinder serves alternately the purposes of motor and pump ; during the first forward stroke of the piston, the valves are in such positions that gas and air stream into the cylinder from the beginning to the end of the stroke, the charge mixing as it enters with whatever gases the space may contain ; the return stroke then com- presses the unif . :m mixture into the space, and when the piston is full in, the pressure has increased to an amount determined by the relative capacity of the space. Meantime the charging valves have closed to the admission of gas and air, to permit of the compression, then ignition is caused when the compression is completed, the compressed charge ignites, and the pressure rises so rapidly that maximum is attained before the piston has moved appreciably on its forward stroke (second stroke) ; the piston is thus under the highest pressure at the beginning of its stroke and the whole stroke is available for the expansion. This is the motive stroke. At the end of it, the exhaust valve opens and the return stroke is occupied in driving out the burned gases, except that portion remaining in the space which cannot be entered by the piston. These operations form a complete cycle, and the piston is again in the position to take in the charge required for the next impulse. The cycle requires two complete revolutions, or four single strokes : First out stroke. Charging cylinder with gas and air. ,, in ,, Compressing the cliarge into the space. Second out stroke. Explosion impelling piston. ,, in ,, Discharging burned gases into atmosphere. Thermal Efficiency of the Four-cycle Engines. — The Otto type engines are termed four cycle because the necessary operations require four piston strokes to complete the power-producing process. The indicated thermal efficiency is determined by the proportion THERMAL AND MECHANICAL EFFICIENCY 243 of the total heat given to the engine, which appears as work done by the expanding gases upon the piston. The brake thermal efficiency is determined by the proportion of the total heat given to the engine, which appears as work given out by the engine for overcoming external resistances. In the early engines, of 100 heat units given to the engine in the form of coal gas 16 heat units appeared within the cylinder as work done by the expanding gases on the piston, so that the indicated thermal efficiency was 16 per cent. Such efficiencies were given by the Otto engines 'from 1877 to 1882. At the present day (1909) thermal efficiencies as high as 35 per cent, to 37 per cent, are readily obtainable. The indicated thermal efficiency has risen from 16 per cent, to 37 per cent, in thirty years ; even higher efficiencies have been obtained experimentally, but 37 per cent, may be taken as above ordinary practice at the present time. The following table of the results obtained by many experimenters will make this advance quite clear : Indicated and Brake Thermal Efficiency of Four-cycle Engines FROM 1882 TO 1908 No. Mechanical efficiency Per cent. 1 87-6 2 84-2 ^ 86-1 4 80 -g ■5 87-3 6 82 7 87 B 83 Q 817 10 85-5 n 77-1 12 87 '5 Names of Year experimenters Slaby. 1882 Thurston . 1884 Society of Arts . 1888 Society of Arts . 1888 Kennedy . 1888 Capper 1892 Robinson . 1898 Humphrey 1900 Witz . 1900 Inst. Civil. Eng. 1905 Burstall . 1907 Hopkinson 1908 Dimensions of engine "Dia^ meter Stroke 6"-75 X i3"-7 8"-5 X 14" 9"-5 X 18" 9'''02 X 14" 7"'S X is" X 18" X iB" X 36" X 5S"'i3 X 22" X 24" X 21" 8"-5 Indicated Brake thermal thermal efficiency efficiency Per cent. Per cent. 16 14 17 14 '3 22 18-9 21 17 21 i8-3 22-8 17 '4 28-7 25 31 25-7 28 22-9 35« 29-9 41 -5 1 32 36-8 32-2 Type of engine Deutz Crossley Crossley Griffin (6-cycle) Beck (6 cycle) Crossley National Crossley Cockenll National Premier Crossley * The value 35 per cent, is deduced by the author from the Inst.C.E, Committee's values, t This value is, in the author's view, too high ; probably due to indicator error. It wiU also be observed that the brake thermal efficiency has in- creased from 14 per cent, to 33 per cent. ; the 29'9 — that is, 30 per cent, of the National Company's engine — ^represents their ordinary practice at moderate compression. It has been shown in Chapter III that if the working fluid had been air, having a specific heat constant through the range, thermal efficiency would increase with increase of compression — that is, with diminution of the compression space in proportion to total volume behind the piston. Taking this ratio as I _ compression space volume r volume swept by piston + compression space volume R 2 244 THE GAS, PETROL, AND OIL ENGINE for the different values used in these engines and arranging the value of ^- in the same order as the table with the indicated thermal efficiencies r E| below as follows No- I 234 56789 '. ' I II I I I 1 — 2-66 2-66 3 '5 3'2 3 3'4 5'i7 5 — 5 '34 7*22 6-37 16% 17% 22% 21% 21% 22-8% 28-7% 31% — 35% 41-5% 36-« Here it will be observed that, considered broadly, increased thermal ef&ciency is associated with diminution of the volume of the com- pression space, although it by no means follows that equal compression ratios give equal thermal efficiencies ; in some cases even the thermal efficiency may be higher with the lower compression. These variations are to be expected from the great variations in the designs and dimensions of the twelve engines investigated. Broadly, however, other things being equal, increase in compression by diminution of compression space volume increases thermal efficiency. If now these indicated thermal efficiencies be divided by the air standard efficiencies, as calculated in Chapter III corresponding to the various values of -, it will be seen that they have a well-defined relationship to that standard. The values are again arranged as follows : No. r 2 '66 2'66 3*5 3 '2 3 3 '4 5'J7 5 — 5 '34 7"== 6-37 Indicated thermal efficiency 16 17 22 21 21 22-6 28-7 31 — 35 4l'5 36-8 Air standard efficiency . 33 33 39 37 36 39 43 47 — 49 55 S3 Indicated thermal efficiency Air standard efficiency o'48 0*51 0*56 0-57 0-58 0-58 o*6o 0-66 — 0-71 0-75 o'69 Diameter of cylinder . 6"-7S 8"-5 9"-S 9"-02 7" -5 8"-5 10" 26" — 14" t6'' ii"'5 It will be seen that in the older engines the indicated thermal efficiency was 0'48 and o"5i of that which an engine using air as the working fluid would have realised had heat losses and other imper- fections of operation been entirely suppressed. That is, the Otto engines tested by Professor Slaby and Thurston gave half the indicated thermal efficiency which they woul^ have done had they been perfect air engines performing exactly the same mechanical cycle of operations. This ratio gradually increases as the years pass, until the value becomes 0'75 in Professor Burstall's experiments. This value, however, the author considers to be too high. Professor Burstall's indicator appears to have erred in some way in this test, and the highest correct value appears to be 071, that deduced by the author from the Institution of Civil Engineers' experiments on a National gas engine of 14 ins. diameter cylinder and 23 ins. stroke. THERMAL AND MECHANICAL EFFICIENCY 245 From the foregoing it is obvious that indicated thermal efficiency increases with increasing compression : this fact clearly emerges from the examination of these twelve tests. It would be better, however, to determine this by experiments on the same engine in which the only variation made is a change in the volume of the compression space. Professor Eugen Meyer, of Berlin, has made experiments of this nature with an engine of 7'8 ins. cylinder diameter and ii-8 ins. stroke, in which he arranged to vary the compression from 40 to 80 lbs. per sq. in. above atmosphere by means of a variation in the length of the connecting rod. The combustion space is thus changed in volume as required. The following table has been calculated from Professor Meyer's experiments : Indicated Thermal Efficiency with varying Compressions cal- culated FROM Professor Meyer's Experiments Compres- Indicated Air Indicated Revolutions sion thermal standard efficiency per pressure Dimension of engine, &c. r efficiency efficiency air standard minute above atmosphere per cent. per cent. lbs. 1 4 T 25 44 0-58 257 80 Engine 7 -8 ins. diam. 24-4 42 0-58 249 75 by 11 -8 ins. stroke. Compression 40 to 3 21-4 37 0-58 251 50 80 lbs. per sq. in. 27 i8-8 33 0-57 225 40 above atmosphere In Meyer's experiments it is evident that the indicated thermal efficiency bears a practically constant ratio to the air standard efficiency proper to the particular compression between 40 and 80 lbs. per sq. in. above atmosphere, or a variation of the compression ratio - of from — to ? of the total volume behind the piston. r 27 4 When the indicated thermal efficiency is 25 per cent., 24*4 per cent, and 21*4 per cent, of the total heat supplied to the engine, it is in the three cases 0-58 of that which would have been given if air alone had been the working fluid and it had suffered no change in its specific heat. When the indicated thermal efficiency is i8-8 per cent., it is 0-57 of the air standard. This small change may be due to the fact that the engine was running at 225 revolutions per minute during the test instead of about 250 revolutions, as in the other tests. In these experiments the change from 40 lbs. to 80 lbs. compression raised the efficiency from i8'8 per cent, to 25 per cent., an improve- ment of 33 per cent, on the lower heating value of the gas. The 246 THE GAS, PETROL, AND OIL ENGINE examination of these and the other tests referred to shows most clearly that indicated thermal efficiency increases with increasing compressionj and that the ratio with reference to the air standard in a good, well- designed engine is now from 0'58 to 071 of the corresponding air standard efficiency. It was pointed out by the author many years ago that for equal compression ratios the indicated thermal efficiency increased with increase of engine dimensions. This is clearly shown by the following table from the 1896 edition of this book, and the discussion which accompanied it. ' Comparison of thk Actual and Theoretical Efficiencies OF Otto Engines of Different Dimensions Engine cj Under Relative capacity Tlieoretical efficiency Actual indicated efficiency Ratio of ac- tual and ideal efficiency Nearly equal|7"diam.x 15" stroke . compression 1 „ . 1 ij diam. X 21 stroke ^^ , , f9i" diam. X 18" stroke. Nearly equal J '^°'"P^^^^i°'^l 14" diam. X 25" stroke. I 377 I 2-97 0-428 0-428 0-40 0-41 0-25 0-275 0-2I 0-277 0-25 o-42i = °-5« 0-428 -°^4 0-2I 0-41 -°'" 0-277 „... 0-41 ~.^ From these numbers it is evident that efficiency for equal compression increases considerably with the dimensions of the engine. ' There is, however, a limit to this increase of efficiency with in- creased dimensions. ' The increase in the efficiency of the larger engines as compared with the smaller using the same proportion of compression space is due to the diminished proportional loss of heat from the gases of the explosion to the inclosing metal walls, and it is always found that in larger engines the expansion curve tends more and more to rise above the adiabatic line. With a maximum temperature of explosion of about 1600° C. it is found by ex^riment that the actual increase of temperature due to explosion accounts for about from 0-6 to 0-7 of the total heat of the gas present ' ,- there is therefore heat enough present in a gas engine of ordinary proportions, if none be lost, to keep up the temperature during expansion performing work to the maximum 1600° during the whole expansion stroke. The increase in dimensions if carried to an extreme could therefore only reduce the loss to insig- nificant relative proportions, and in such a case the mass of incan- ' Assuming constant specific heat and air as working fluid. THERMAL AND MECHANICAL EFFICIENCY 247 descent gas might be considered to lose no heat whatever to the walls of the cylinder. ' Assume an air engine in such a case : the total volume of the stroke plus clearance space being i cub. ft. ' Assume the engine to have a compression space of 0'275 of the whole cylinder volumCj as in the test made by the author on Crossley's Otto scavenging engine. Then the diagram and results would be as shown in fig. 91, where the temperature of explosion is 1600° C. ' From this it wiU be seen that while 0-409 is the efficiency for adiabatic expansion, then 0'346 is the efficiency for isothermal expan- sion ; from this, then, it appears that, allowing for the known property of the suppression of heat in a gaseous explosion, the utmost efficiency Efficiency of adiabatic compression and expansion = o'4o9. Efficiency of adiabatic compression and isothermal expansion = o"346. Fig. gi. — Theoretical Diagram, comparing adiabatic and isothermal expansion possible for an engine using coal gas, having a compression space of 0'275 of the total cylinder volume, and expanding to the same volume as existed before compression, is 0*346, so that the efficiency actually attained in practice is —^-^ = 0'8o or 80 per cent, of the possible.' In November 1903 a Committee ' of the Institution of Civil Engineers was formed to consider and report to the Council on the Standards of Efficiency of Internal Combustion Engines. Professor Unwin was the chairman, Pfof. Dalby the honorary secretary, and the ' Report of the Committee on the Efficiency of Internal Combustion Engines, vol. clxii., session 1904-1905, part iv., and vol. clxiii., session 1905-1906, part i. 248 THE GAS, PETROL, AND OIL ENGINE other members were Sir Alexander B. W. Kennedy, Professor Callendar, Professor Ashcroft, Captain H. R. Sankey, and Messrs. Hayward, Wilson, and Dugald Clerk. This Committee published a very careful and complete report, recommending the adoption of the air standard ; they considered that the standard engine of com- parison should satisfy the following conditions : ' (i) The reception and rejection of heat should take place as nearly as may be in the same way as in the actual engine. ' (2) There should be no heat losses due to conduction, radiation, leakage, or imperfect combustion. ' (3) The data for the numerical evaluation of the standard should be ascertainable by simple measurements. ' (4) The expression for the efficiency of the standard should be a simple one.' Also, that the standard working fluid to operate in the engine be air, which is to be assumed as obeying perfectly the laws of gases. The specific heat is to be assumed as constant for all working tempera- tures, and the value of y, the ratio of specific heats at constant pressure and at constant volume, is to be taken as T4. The standard thus recommended is practically that which has been used in all the editions of this book. The members of the Committee were satisfied that, given the ideal efficiencies of a number of engines calculated in this way from the formula — ©" that the actual indicated thermal efficiencies obtainable with good gas engines bore a ratio which varied from 0-5 to p-y. The author has discussed the Committee's report elsewheTe''^ as follows : ' The Committee considered the correspondence of the air standard with actual indicated results to be firmly established by these and other tests ; but they were not satisfied with the existing knowledge as to the variation of the ratios between the ideal and the actual efficiencies, with varying dimensions of the engines of the same ideal efficiency. It was felt that knowledge was wanting as to the change of ratio, when dealing with engine^of widely different dimensions. The author accordingly arranged to place at the Committee's disposal three gas engines of exactly the same type, having the same value of i/r and of quite different dimensions. The cylinders of the engines were respectively 5'5 ins., 9 ins., and 14 ins. in diameter, and the powers were respectively 6, 24, and 60 IHP. All three engines were arranged in one room and supplied with the same gas, and apparatus of the same ' " On the Limits of Thermal Efficiency in Internal Combustion Engines," Pro- ceedings, Inst. C.E., vol. clxix. page 157. THERMAL AND MECHANICAL EFFICIENCY 249 type was used for testing them. The tests provedj however, that there was some variation of ratio between the relative eificiencies. Taking the mean mechanical efficiency of the full-load trials as 88 per cent., and calculating the indicated power from the brake power, it was found that the relative efficiencies were respectively 0'6i, 0'65, and 0'6(^. The change of relative efficiency due to change of dimensions was fully proved, and it was clearly shown that by bearing these small changes in mind the engineer could obtain a close approximation to the best actual indicated efficiency to be expected from any given compression used in a gas engine between these widely varying limits. The experi- ments, in fact, clearly proved the great utility of the air standard for the purpose of comparing different engines with different compression spaces, and predicting the thermal efficiency to be expected from each, assuming the combustion space to be properly proportioned, and the valve actions to be properly performed. The experiments also proved in a quite definite way that which had been long believed by gas-engine builders — namely, the small gain in economy which can be attained by a large engine as compared with a small one. It is some- what surprising to note, however, that between 6 HP and 60 HP increase of dimensions gives an advantage of only about 12 per cent. ' In recommending the air standard, the Committee were well aware that in proposing air as the standard working fluid with the properties defined in the report, they were using a fluid which differed in its properties from the actual working fluid. If the actual working fluid had the same properties as the ideal, then the relative efficiency numbers given above would truly represent the ratio of the heat con- verted into work to the heat which could have been converted under the ideal conditions. In the small engine, for example, the relative efficiency number o'6i would have meant that 0"6i of the heat which the actual working fluid could have converted into work under ideal conditions was converted into work, and with the large engine the pro- portion was 0"69. That is, the margin to be worked on for improve- ment was, in one case 39 per cent., and in the other case 31 per cent. It was known from certain investigations that this was not so — in fact, that the possible efficiency of the actual working fluid under ideal con- ditions was not so high as the number given by the air standard. Had the properties of the actual working fluid been known, it would have been possible to calculate ideal efficiencies using the known properties. Enough was known, however, to show that the actual properties of the working fluid in the engine were by no means simply ascertained, and it was felt better to adopt a standard capable of definite expression, from which the actual best efficiencies could be deduced by a multiplier found experimentally. This multiplier, the relative efficiency, thus includes not only the actual imperfections of the engine cycle, but the 250 THE GAS, PETROL, AND OIL ENGINE variations between the actual properties of the working fluid and those of the ideal air assumed as a standard.' In the Committee's trials considerable difficulty was experienced in obtaining the true indicated power by the use of the indicator, although every cEire was exercised. The difficulty was partly due to variation in the diagram itself when the engine was adjusted to give its best brake efficiency, and partly to imperfections of the indicator itself, which have undoubtedly been accentuated in recent years by the great rise in compression and explosion pressures. The report accordingly states : ' The values for the indicator horse-power, and consequently those for the mechanical efficiency, are probably not very accurate, because the indicator diagrams vary, and the mean of the limited number taken in a trial is not the true mean. It is at least probable that the mechanical efficiency was more nearly constant for the three engines than the figures in the table indicate.' Further on the report also states : ' It would be desirable, but for one circumstance, to calculate the relative efficiency only from the indicator horse-power. But it appears that in the case of gas engines, and especially gas engines governed by hit-or-miss governors, the indicator diagrams do not give as accurate results as is generally supposed. The diagrams vary much more than those of a steam engine with a steady load, and the mean indicator horse-power, from the diagrams taken in a trial, may, it appears, differ a good deal from the real mean power . . . ' Notwithstanding these difficulties the Committee succeeded in arriving at a satisfactory accurate determination of the mechanical efficiency of each of the three engines by two methods which dispensed with accuracy in the indicator, or, rather, minimised the effect of inaccuracy. These methods are described in the report as follows : ' The mechanical efficiency values obtained from comparison of IHP and BHP of three Ashton engines are obviously incorrect ; the three full-load tests show : Engine L R X Mechanical efficiency . . . 0-90 o-8o 0-94 t ' There appears to be no reason why the R engine should show so low an efficiency as o'So, and none why L and X should be so high as 0-90 and 0'94. Fortunately the observations made supply means of calculating these efficiencies by two other independent methods : ' (i) By adding IHP without load to BHP at full load, assuming the friction as determined by the indicator to be the same when the engine is running without load as it is when the engine is fully loaded. THERMAL AND MECHANICAL EFFICIENCY 251 ' With engine L, Test 4, No Load shows that O'gS IHP maintains the speed of the engine at 291 revolutions per minute ; reducing this to 258'9 revolutions per minute, the fuU-load speed, the IHP necessary to drive the engine without load at 258-9 revolutions is found to be 2 58-9 X 0-9 6 ^^.gj^p 291 -^ ' The full load at 258-9 revolutions per minute is 5-2 BHP. The IHP is 5-2 + 0-85 = 6-05 and the mechanical efficiency is -? — = o-86. 6-05 ' Calculated in this way the respective values of the mechanical efficiency are : Engine L R X Mechanical efficiency . . 0-86 0-866 0-888 ■ (2) By calculating from the full-load and half-load values ,of the BHP and the gas consumptions, assuming friction to be constant from half to full load. ' The values required are : BHP at full load. BHP at half-load. Gas per hour at fuU load. Gas per hour at half-load. ' The BHP and the gas at half-load must be taken at the same number of revolutions per minute as in the fuU-load trial. Gas per hour at full load— Gas per hour at half -load _ra5„orTHPimnr BHP at fuU load- BHP at half-load ^aspermt- nour ,, , ■ 1 re • Gas per IHP hour Mechanical etticiency = -^ £ — ttyt^ ,- Gas per BHP hour ' Calculated in this way the mechanical efficiencies are : Engine L R X Mechanical efficiency . . . o-8i 0-83 0-84 ' The mean values by (i) or (2) are : Engine L R X Mechanical efficiency . . 0-835 0-848 0-864 ' Method (i) depends on the accuracy of the indicator ; but an error of, say, 5 per cent, only introduces an error of that amount in the friction value itself. In calculating mechanical efficiency from the total indicated power an error of 5 per cent, on the total may readily amount to 20 per cent, on the friction, While by method (i) it is limited 252 THE GAS, PETROL, AND OIL ENGINE to the 5 per cent, on the friction value calculated. Method (2) gives the mechanical efficiency without reference to the indicator, and it only assumes that the diagrams remain constant at the lighter load and that friction is constant between full and half-loads. ' It seems clear, as has already been stated, that however carefully indicator diagrams are taken in gas-engine trials, they do not furnish as accurate a value of the mean indicator horse-power as has been generally supposed.' If, then, the values 0'84, 0-85, and 0'86 be taken as the mechanical efficiency of the three engines ' L,' ' R,' and ' X,' a very close approxi- mation to the truth will be obtained. Fig. 92.— Sections of ' L ' Engine. (National Gas Engine Co., Ltd.) Institution of Civil Engineers' Committee Tests Engine ..... Brake thermal efficiency, per cent. The brake thermal efficiencies of the Committee, which were undoubtedly accurate, were as follows : . L R X 26T 28 29-9 Dividing these values by o"84, 0-85, and o'86 respectively, we get : Engine ... . • . . L R X Indicated thermal efficiency, per cent. 31 32-9 34-8 The air standard efficiency for the three engines is : Engine . . . . L R X Air standard efficiency, per cent. 49 '6 49-6 49 Dividing by the respective indicated thermal efficiencies, we get : Engine L R X Relative indicated efficiency . . 0'625 0-662 071 THERMAL AND MECHANICAL EFFICIENCY 253 It may be taken as established, then, from the experiments of the Committee, that when three engines of different dimensions are adjusted to give their best economical result in gas consumption per brake horse- power, that the efficiency ratio relative to the air standard varies from the value 0-62 to 071, the cylinders being respectively 5-5 ins., q ins., and 14 ins. diameter. The conclusions thus support those of the author already mentioned. As these experiments of the Institution of Civil Engineers are of great importance, it is desirable to describe the engines experimented ...^..^^.^^..■ FiG. 93. — Scctions of R ' Engine. (National Gas Engine Co., Ltd.) Institution of Civil Engineers' Committee Tests '^^^'^ '' with and some of the other results obtained. The three engines designated L, R, and X are shown in longitudinal section and part horizontal section at figs. 92, 93, and 94. Some particulars are marked under the figures. They are described as follows in the Committee's report : ' The three engines tested were of the standard four-stroke cycle tj^je, built by the National Gas Engine Co., Ltd., for cylinders of 5'5, 9, and 14 ins. diameter respectively. The arrangements of the engines were identical in respect of the proportions of the combustion space, the mechanism for the admission of the charge, and the exhaust valves. In each case the charge was admitted by means of an inlet valve opening into a central port at the end of the combustion 254 THE GAS, PETROL, AND OIL ENGINE space. Behind the inlet valve was a gas valve, and within the inlet port the electric igniter was arranged, operating on the low-tension principle, with a Simms-Bosch magneto-instrument. The three engines were each provided with two electrical igniting arrangements, the L engine having the second igniter placed within the cylinder in a plug above the exhaust valve. The R and X type engines had the second igniter placed at the side of the combustion space, and not at the top. These second igniters were introduced, as it was intended at one time to make experiments with ignition inside the cylinder instead of in the port. These experiments, however, were not made by the Committee. The inlet valve in each case was operated from the cam on the usual two-to-one shaft, which runs along the side of the cylinder, and the gas valve was operated by a similar cam placed close to the main supply- valve cam. The engines were governed on the hit-or-miss method by the action of a centrifugal governor on a small block, which engaged with a pecker. The exhaust valves were placed in the bottom of the combustion space, and were operated Fig. 94. — Sections of ' X ' Engine. (National Gas Engine Co;, Ltd.) Institution of Civil Engineers' Committee Tests also from the two-to-one shaft by levers. The exhaust gases, after passing the exhaust valve, traversed | water-cooled passage, the water- jacket of which was included in the jacket circulation of the engine. This is clearly seen in the section of the exhaust calorimeter, fig. 95. It will be noticed that with this arrangement heat is abstracted from the exhaust gases after opening the exhaust valve, before these gases arrive at the exhaust calorimeter. ' In the large engine, that is, the X type, the starting was accom- plished by means of a hand-pressure pump, which communicated with a plug above the exhaust valve. This plug contained a valve which 256 THE GAS, PETROL, AND OIL ENGINE V- ','St ;on(,.J allowed the access of mixture to the cylinder from the hand-pump. The engine was started by placing the crank conveniently over the centre, pumping in a mixture of gas and air behind the piston, to a pressure slightly above that of the atmosphere, shutting the inlet valve in the plug, and tripping the magneto-shield of the Simms-Bosch instrument. The electric spark then passed between the separated surfaces within the cylinder, and the engine started. ' The Simms-Bosch magneto was of a well-known type, having a fixed armature, fixed permanent magnets, and movable shield. The shield was withdrawn and tripped by the operation of an adjustable pin rotating on a collar on the two-to-one shaft. This pin can be adjusted to vary the time of firing the charge as may be required. ' It is needless to describe the operation of the engines. They act on the ordinary four-stroke cycle, and there was nothing in the mechanical arrangements which calls for special description. They are the well-known arrangements of the Fig. 96.— Details of National Gas Engine Co., as applied to engines of Gas^ Calcfrfmeter ^'^^ ^^^^^ '"^^^ Selected by the Committee. of fig. 95 ' It may be stated, however, that in order to secure results comparable scientifically, the com- pression spaces were carefully adjusted to have as nearly as possible similar proportions in the three engines. Care, too, was taken that the internal surface of the combustion spaces should be smooth and clean, in order that no unknown element should enter into the observations due to irregularities in the castings. In the small engine, care was taken that the working parts were aU very free, in order that no undue friction should affect the results. Dimensions of Engines {(Captain Sankey and Professor Dalby) Designation of engine Clearance volume in cubic centimetres . Clearance volume in cubic inches { i cubic inch | = i6'387 cubic centimetres) . H . . J Diameter of cylinder . . . inches Stroke .... . . . ,, Area of cylinder square inches Volume displaced by piston-stroke . cubic inches Total volume of cylinder . . ,, ,, Clearance. Percentage of total volume . Circumference of brake-drum .... feet ,, rope ... . ,, Effective circumference of brake . . . ,, Diameter of air-orifice in the measuring trunk, inches L R X 850 3.920 12,680 52 239 774 5-502 9-00 14-008 10-00 17-03 22-00 23-78 63-62 154-1 237-8 1,083 3.390 289-8 1.322 4,164 17-94 1 8 -08 18-59 1 9-19 1 8 -208 19-84 o-oi 0-230 0-295 9-200 18-438 20-141 4 8 12 THERMAL AND MECHANICAL EFFICIENCY 257 ' The three engines were adjusted as to gas-supply so that ea(-h engine should run as nearly as possible at its most economical load, ui ^ S P X o q H to U j= "bii s rt O O -3 With more gas the three engines could each give considerably more power, but such power would partake somewhat of the nature of an overload, and the consumption per brake HP would not be quite so VOL. I. s 2=;t THE GAS, PETROL, AND OIL ENGINE low. In all the tests made, the igniter used was that in the admission port." Fig. 97 shows a diagrammatic arrangement of the apparatus used l'>^ -,, THERMAL AND MECHANICAL EFFICIENCY 259 e in the testing of each engine. A novel feature in connection with this is the method adopted for measuring the air-supply to the engine by means of an anemometer placed in a box which forms an enlarged continuation of the air-suction pipe. A novel form of exhaust-gas calorimeter, designed and fitted by the Company, was used to extract the heat from the exhaust gases ; this was arranged on the principle first used by Professor Bertram Hopkinson, and is given at figs. 95 and 96. The water measurements were made in tanks, which were calibrated by the Committee. All the weights, spring balances, and thermometers used were calibrated. Fig. 98 is a plan showing the general arrangement of the engine- room, and fig. 99 is a photograph of the room with the engines in position. Fig. 100 is a photograph of part of the large or 'X' engine, showing an optical indicator of the author's design as arranged for experiments such as those indicated in an earlier chapter. The report proceeds : ' The measurement of air quantity was made by means of a small anemometer, placed inside a wood trunk B, fig. 97, which was attached to an extension of the air-suction pipe a of the engine. A large gas-bag, the details of which are shown in fig. loi, was placed on the connecting pipe in order to keep the flow of air at the anemometer as uniform in speed as possible. The fan of the anemometer was placed close to the air inlet to the box, and its indications were read from without through a glass window let into the top of the box. The air inlet was circular, and was cut out of a piece of sheet-iron so that the size could be easily adjusted to determine a rate of flow past the anemometer which 1 pk 26o THE GAS, PETROL, AND OIL ENGINE would give a suitable rate of rotation to the fan. The pressure and temperature of the entering air were measured by a barometer and a thermometer x,, whilst the hygrometric condition was taken by means of a wet- and dry-bulb thermometer. ' At the engine trials, readings of the anemometer were observed which gave the number of lineal feet of air passing the anemometer. These had to be reduced by some means to cubic feet of air. The method devised for doing this was simple and effective. When the engine was quite cold, it was driven by an electric motor through a belt placed on the flywheel. The gas-supply was cut off and the engine Anemometer Calibrating Trials Number of test Diameter of orifice Anemo- meter reading. Velocity of air Revolutions of engine per minute Volume displaced by piston per stroke Ratio of eflfective to total stroke Volume of air tbrough orifice per minute Volume of air corresponding to anemometer unit 5 6 II 13 1 * 2 * 3* 18 19 4* 5* Ins. 4 8 12 Ft. per min. 293-3 268-0 LU_ 227-3 118-2 185-5 268-5 357-S 193-8 214-3 138-0 166-0 285-0 263-0 2I4-I 142-5 84-8 125-5 163-0 142-2 155-6 165-0 193-0 Cub. ft. 0-1376 0-6273 1-962 0-886 0-920 -1' . 0-892 0-756 I-ooo I-ooo 0-977 0-927 I-ooo 0-988 0-977 Cub. ft. 17-26 16-65 Mean 59-9 33-8 53-2 78-7 99-9 Mean 129-3 152-6 102-3 118-3 Mean Cub. ft. 0-0588 0-0621 0-0605 0-263 t 0-285 0-287 0-293 0-279 0-281 0-667 0-712 0-741 0-713 0-708 * These tests Nvere made with special cams for the valve-gear so that the engine acted as a suction pump without compression t This result is rather anomalous. allowed to draw in air through the orifice used in the trial, the cylinder in this way serving as a calibrating chamber. In some subsequent tests made by Mr. Dugald Clerk on the engine 'R,' the cams were changed so that the cylinder became a single-acting air-pump without compression, and two of the orifices were separately calibrated on this engine at the mean speed of the actual trial. The dimensions of the box are given in fig. 97. ' In these anemometer calibrating trials the number of revolutions of the engine, the anemometer readings, and the duration of the run were observed. Indicator diagrams also were taken. It was assumed THERMAL AND MECHANICAL EFFICIENCY 261 that the proportion of the cylinder fiUed with air at atmospheric pressure at each stroke was given by the length on the indicator diagram between the points at which the pencil-line crossed the atmospheric line. This distance may be termed the effective stroke. From these results the cubic feet of air per unit reading of the anemometer for each orifice used were obtained. ' The hot gases were cooled by passing them through an exhaust- gas calorimeter. The construction of the calorimeters used in the trials is shown by the working drawings, figs. 95 and 96. It will be observed that the calorimeter is water- jacketed right up to the con- nection with the engine exhaust flange. The water is led in at a, and after passing through the jackets is led to the rose, R, fig. 96, through the small orifices of which it spurts out to meet the stream of exhaust gas. The water and gas then find their way out, passing various obstructions to ensure mixing and abstraction of heat from the gas, until finally the gas, cooled down to about 90° F., escapes at G, and the water escapes at e, having about the same temperature as the escaping gas, although sometimes it was higher and sometimes lower. The general arrangement of the calorimeter is shown in fig. 97, from which it wiU be seen that the water was brought into the calorimeter directly from the water main, and was led to measuring tanks placed outside the testing-room. These tanks were fitted with gauge glasses and scales graduated in feet and decimals of a foot, and were caUbrated by pouring in weighed quantities of water. The calibrations obtained were : Lbs. of water per ft. No. I tank for exhaust calorimeter . . . 68o'82 No. 2 ,, ,, ,, ,, . . 679-42 ' In the design of these calorimeters care should be taken that the thermometer measuring the temperature of the escaping cooling water is placed so that the bulb is completely immersed in the water. As the pipe does not always run fuU, a pocket should be formed in the pipe to ensure this condition. Care should also be taken that the g£is itself does not carry away water mechanically suspended in it. The drawings, figs. 102, 103, and 104, show a form of calorimeter in which special attention is given to these two points, and which was used in some trials made by Mr. Dugald Clerk subsequent to those made by the Committee. The results showed that, although the calorimeters used by the Committee were not provided with these safeguards, the quantities concerned were measured without appreciable error. ' In the reduction of the observations on the calorimeter, several minor points have to be observed. In the first place,the water flowing out of the calorimeter, and which is measured in the measuring tanks, 262 THE GAS, PETROL, AND OIL ENGINE Q S 1^ THERMAL AND MECHANICAL EFFICIENCY 26s is not exactly the quantity which enters it. The gases cooled to about go° F. are at this temperature like a sponge as regards the absorption of water-vapour, and as they have been in intimate contact with the water during the whole time of the passage through the calorimeter it may be assumed that they wiU pass out completely saturated with moisture at the exhaust temperature. Now the amount of water required to saturate the exhaust gases may be computed as though the gas was entirely air without introducing serious error. . . . ' This amount is not, however, all abstracted from the water entering the calorimeter, for it will be observed that the air entering the engine cylinder brings in with it a definite weight of moisture, and the combustion of the gas produces a definite weight also. Hence the quantity actually abstracted from the cooling water is the difference between the quantity required for the saturation of the exhaust gases to- I'/h hol-dA- . 'i- fwles SECTION B B -SECTION C C SECTION D D . Fig. 103. - Details of fig. 102 and the sum of the water produced by combustion and that brought in by the air.' Radiation of heat from the engine surface is difficult to determine, and it is dealt with as follows : ' The Committee include under the heading of radiation the radiation from the hot surfaces of the cylinder, trunk piston, &c., and the radiation from the bearings of the engine due to friction. In other words, the term includes the heat lost by direct radiation from the hot surfaces, together with the heat corresponding to the difference between the indicator horse-power and the brake horse-power. The difficulty of finding the indicator horse-power exactly led the Com- mittee to define radiation in this way. ' The method of measuring this quantity was to run the engine light and then to adjust the number of explosions per minute so that the temperatures maintained on the thermometers measuring the tem- peratures of the water to and from the jackets were resjijectively equal to those at a full-load trial. In this way the surface temperatures, 264 THE GAS, PETROL, AND OIL ENGINE allowing a very small quantity of water to flow through the jackets in order to obtain similar conditions to those during the normal working of the engine, were kept approximately the same in the two trials, although the inner temperatures would be lower because of the fewer explosions. ' The method is of course only an approximate one, but it gives some idea of the losses from this cause.' Very full and interesting details of measurements and corrections are given in the report, and the reader is referred to it for further study ; but some of the most generally interesting results are given in the following table : Engine . L R X B-akeHP 5"z 20*9 5^'7 Reva. per miniite . 258-9 203*6 165-8 Cyl. diameter and itroke 5 '-S X 10" g" X 17" n" y 22'' lias per BHP hour at [^ working temp, and pressure) i6'87 cub. ft. 15-84 cub. ft. i4'9 cull. ft. Lower calorific value of gas at t working temp, and pressure f 566 B.Th.U. 567 B.Th.U. 574 B.Th.U. Ratio by vulume of air to gas ) air 9'i5 air 9-17 air 8 '21 in charge . . I g.is ~ I gas 1 gas ~ I Ratio by vohinie of air + ex- 1 haust to gas in charge . j air + exhau.it lo'i air + exhaust __ 9*75 air + exhaust 9*3 gas ~ I gas I gas I It is to be noted that the brake power given was found to give the most economical result per brake horse in each engine, and it will be seen that the proportion of air to gas in the L and R tests was prac- tically as 9-15 to I, while the ' X' engine used the somewhat stronger mixture of 8-2i to i air to gas. The author has calculated the total air + exhaust gases Q'3 engme as ^° = ^-^^ gas I observed that the Committee determined by direct methods all the values necessary to enable a complete balance-sheet of the disposal of heat to be prepared for each engine. The coal gas used was measured by meter, its heating value was determined by Junker calorimeter, and it was also analysed chemically. This enables the total heat given to each engine in the form of chemical energy to b^ccurately known. The heat leaving the engine by way of the hot exhaust gases was accurately known from the readings of the exhaust calorimeter. The heat leaving by the water jackets was also accurately determined. Total radiation from the engine was also determined, but here the method used was rough, and the total radiation value cannot be considered as better than a fair approximation. dilution in the case of the It wiU be ■M- f^ h^Ut'-T^^^^-^^ Fig. 104. — De- tail of Nozzle, fig. IC2 THERMAL AND MECHANICAL EFFICIENCY 265 The brake power of the engine was very accurately determined ; the values for brake power are the most reliable of aU. By adding together all the values so determined, it should be possible to account for all the heat given to each engine. The following balance-sheets are given for the full-load tests in the report : Institution of Civil Engineers' Committee Tests — Heat Balance-Sheet. Full Load Designation of engine L , R Exhaust waste Jacket waste Radiation BHP ... 35-3 23-5 7-6 267 40-0 29-3 lO-O 28-3 93-1 107-6 39-5 25-0 7-3 29-8 ' I0I-6 Although every care was exercised, the experiments do not account for all the heat given to the small or ' L ' engine ; of 100 heat units given to the engine only 93'i units have been found — a loss of 6'9 per cent, has been incurred in some way. Analysis of the exhaust gases was made to find if all the gas had been burned and combustion appeared to be complete, so that the deficit was not accounted for in this way. In the tests of the larger engines ' R ' and ' X ' too much heat has been found by y6per cent, on ' R' and i'6per cent, on the 'X' engine. The test of the large engine comes most nearly to the 100 heat units, and this greater accuracy would be expected as the heat quantities measured were greater and error more easily avoided. By reasoning upon other figures found in the report the author has discussed this balance-sheet, and has arrived at the conclusion that the following adjusted balance-sheet more truly represents the actual disposition of heat given to the three engines : Institution of Civil Engineers' Committee Tests — Heat Balance-Sheet OF the three Engines adjusted by Clerk Designation of engine L R X Exhaust waste . Jacket waste and radiation . IHP . 34-1 34-1 31-8 37-1 29-6 33-3 39-9 25-4 347 lOO-O 1 00-0 loo-o Here the indicated power is given instead of the brake power, so that the friction of the engine is no longer included under radiation. 266 THE GAS, PETROL, AND OIL ENGINE The reasons for this adjustment will be found in the paper already referred to, read by the author before the Institution of Civil Engineers in 1907 ; they need not be dealt with here. In all gas-engine balance-sheets as hitherto presented the jacket loss is over-estimated. The jacket item in the balance-sheet should represent the heat flow from the hot gases in the engine cylinder during explosion and expansion, but a considerable proportion of heat finds its way into the jacket water after expansion is complete ; that is, heat comes from the exhaust gases upon discharge, which should appear under exhaust waste. The adjusted balance-sheet above is erroneous in this respect : too much heat appears under jacket waste and radiation, and too little under exhaust waste. There are several reasons for this. One is that when the exhaust valve opens, the hot gases discharging round the valve impinge violently upon the water-jacketed metal beneath the valve before their course is turned, and then pass through a water- jacketed space before reaching the water- jacketed passage included in the calorimeter space. As gases in violent motion impinging against a metal surface lose heat very rapidly, it follows that some of the heat which should appear in the exhaust calorimeter appears in the water jacket. Further, when the gases in the cylinder have fallen to atmospheric pressure, the cylinder still remains filled with hot gases, often at a temperature of over 1000° C. During the exhaust stroke of the engine piston these gases are in contact with the sides of the cylinder, and are flowing through the combustion space and valves, so that a good deal of remaining heat also passes into the water jacket. The friction of the piston, too, generates heat, and this heat either flows into the water jacket or disappears in radiation and conduction from the piston-bottom to the external atmosphere. This difficulty cannot be effectively met by any of the ordinary methods hitherto used, and for the purpose of arriving at a more accurate division the author has continued experiments upon the large ' X ' engine of the Committee by his new diagram method described in the last chapter. The work of the Institution of Civil Engineers' Committee has proved beyond doubt that the air st#idard furnishes a reliable measure of the varying efficiency of different engines ; that the actual indicated thermal efficiency may be readily deduced from the theoretic thermal efficiency by the use of a multiplier which varies to a small extent with the dimensions of the engine ; that in the best modern engines of varying dimensions this multiplier varies between 0"62 and 071 ; that the ordinary indicator does not give so reliable a result as is com- monly supposed ; that it is more difficult to obtain accurate indicated power from the gas engine than is usually assumed ; that this difficulty THERMAL AND MECHANICAL EFFICIENCY 267 may be overcome by combining in use the brake and indicator for light loads ; that the anemometer may be used to determine air-supply to the engine ; that radiation may be approximately determined by the new method described, and, finally, that in the three engines tested brake efficiencies of 26-7, 28'3, and 29^9 per cent, are readily obtained, corresponding to indicated thermal efficiencies of 3i'8, 33"3, and 347 per cent. These appear to be the highest thermal efficiencies obtained and authoritatively vouched for up to the date of the Com- mittee trials. These results are of the utmost importance and supply a distinct step in the development of the science of the internal-combustion motor. It has been pointed out in the preceding discussion that the air standard efficiency is higher than the efficiency which could be obtained from the actual working fluid, knowing its properties as we now know them. In the preceding chapter the author has described a new method of determining the cooling curve in the actual operation of the piston in the gas-engine cylinder, and has shown that not only can a cooling curve be deduced giving the temperature fall due to cooling to the walls, but the dynamic value of this fall can be deduced in foot-pounds per cubic foot of working fluid. It wiU be at once recognised that this method supplies a means of arriving at a balance-sheet of the gas engine by indicator measurements only; it does not require any gas measurements or heat-flow measurements whatever ; it only requires diagrams of the type shown at p. 26S to be able to arrive at a balance- sheet without the immense labour of the usual methods of testing. A test of this kind was made by the author on the ' X ' engine of the Institution of Civil Engineers' Committee report at the brake load of 50 HP. Three diagrams were taken in the manner described in Chapter VIII. These diagrams are given at fig. 105. The mean temperature in terms of time was determined for each explosion and expansion stroke ; the temperature fall due to cooling to the walls v/as taken, and the dynamic value of the temperature fall was obtained from the table of apparent specific heats given at p. 235. The temperature of the gases at the end of the expansion was measured by the pressure at that point with the piston in the full-out position. From this temperature and the initial temperature of the charge the heat carried away by the exhaust gases was determined by using the apparent specific-heat table referred to. Then the indicated work was obtained from the positive loop of the diagram — that is, the area between the compression line and the explosion and expansion line. The volume of air entering the engine was determined by ane- mometer and the gas by gas meter, the temperature of the external air ^^ f n, . X ■ engine at a brake-load of 5° HP FK>. .05.-ae.U ^^Z^^i:\Zx.^.s %. minute THERMAL AND MECHANICAL EFFICIENCY 269 Heat Balance-Sheet for ' X' Engine prepared from Clerk Diagrams, Fig. 105. (Clerk) Card No. 22 Heat flow during explosion and expansion Heat contained in gases at end of expansion . Indicated work 28,900 Total heat i.e.. 104 B.Th.U. Card No. 23 Heat flow during explosion and expansion ''Heat contained in gases at end of expansion . Indicated work ... ... Total heat i.e., 106 B.Th.U. Card No. 24 Ft. -lbs. Per cent. 12,480 I5'4 39,800 49-0 28,900 35-6 81,180 loo-o Ft.-lb'. Per cent. 14,000 17-0 40,500 49-3 27,700 337 82,200 lOO-O Ft. -lbs. Per cent. 13,100 i6-o 40,600 49-5 28,260 34-5 81,960 lOO'O Heat flow during explosion and expansion Heat contained in gases at end of expansion Indicated work '28,260 Total heat i.e., 106 B.Th.U. All the values have been taken in foot-pounds. Taking card No. 22, for example, the temperature fall due to cooling is equal to 12,480 ft. -lbs., the heat contained in the gases at the end of the expansion is equal to 39,800 ft. -lbs., and the positive loop of the indicator diagram shows that the work done on the piston is 28,900 ft. -lbs. Now, if we add together these three items we should get the total heat given to the charge for one stroke of the engine ; this amounts, it would seem, to 81,180 ft. -lbs., which is equivalent to 104 British thermal units. Cards 23 and 24 vary slightly from this, but they show each 106 B.Th.U. Now, if the combustion is nearly complete at the end of the stroke the heat present found in this way should be equal to the heat evolved by the gas known to be present in the charge. The gas present in the charge is known to be approximately 0'i83 cub. ft. at the working temperature of the measuring meter, and its lower calorific value was 574 B.Th.U. per cubic foot. The heat of combustion of the gas is therefore 0-183 X 574 = 105 B.Th.U. It is thus seen that the approximation is very close. The indicator 270 THE GAS, PETROL, AND OIL ENGINE has been able by the new method of application to account for the heat present in the charge. The distribution of the heat varies to a small extent in the three diagrams, so that it is better to consider the mean result. The mean balance-sheet of these three from cards Nos. 22, 23, and 24 is as follows : Mean Balance-Sheet, Cards Nos. 22, 23, Heat flow during explosion and expansion Heat contained in gases at end of expansion , Indicated work AND 24 Per cent. . :6-i 49-3 34-<5 lOO'O These indicator diagrams, together with the apparent specific heat values given in Chapter VIII., thus enable the total heat evolved in the cylinder per stroke, and its distribution in indicated work, necessary exhaust loss, and heat flow through the cylinder walls, to be determined from the diagram only. Compare this balance-sheet with that deduced from the Com- mittee's trials on p. 265. Committee's trials New diagram trials Heat flow during explosion and expansion . Heat contained in gases at end of expansion Indica ted work . ... 25-4 39-9 347 i6-i 49-3 34-6 loo-o lOO-O The indicated work is pra.ctically the same in both trials and the sum of the other two items is the same also, but the distribution is different. Less heat flows through the cylinder walls as determined by the author's new method, and the exhaust gases contain more heat than the Committee's calorimeter trials show. The ordinary trials show 9'3 per cent, too much heat as passing through the cyUnder-walls, and practically the same amount too little appears in the exhaust calorimeter. That is, i8'8 per cen^of the total heat remaining in the hot gases at the end of the expansion passes into the cylinder water- j acket during the flow through the exhaust valve upon the first opening and while the piston is making its exhaust stroke. Or, roughly, the true heat flow on explosion and expansion is about 0-63 of that usually measured by water jacket and radiation. This seems to be a quite reasonable portion of the total heat, such a portion as experience would lead one to expect. These new diagram trials afford, in the author's view, a more accurate heat distribution balance-sheet than has yet been obtained THERMAL AND MECHANICAL EFFICIENCY 271 in any engine, from which can be deduced the ideal efficiency of the working fluid. If it be assumed that in this experiment the whole of the heat loss, say 16 per cent., is incurred at the beginning of the stroke before the attainment of maximum temperature, then the total heat dealt with during expansion would be 100 — 16 = 84, and the thermal efficiency had there been no loss would have been ^^— = 0'4i. 84 That is, the best efficiency which this working fluid could give for the particular compression ratio — would be 41 per cent. Had air been the working fluid, it has been already pointed out, the ideal efficiency would be 49 per cent., so that the possible efficiency from the actual working fluid is considerably less than that possible from air. The heat loss is not all incurred at the beginning of the stroke. It has been shown in the last chapter that about 50 per cent, of the total heat loss is incurred during the first -i\ of the stroke, and the remain- ing 50 per cent, in the last i^. Even 41 per cent, ideal efficiency is too high, and the accurate method of deducing this number depends on knowing the true adiabatic expansion line of the working fluid. Assuming the apparent specific heat to be the true specific heat, this line can be calculated from the tables on p. 235 by a method described in Appendix II. The curve shown at fig. 106 has been so calculated for expansion from I volume to 10 volumes. The curve is calculated on the assumption that a mass of gas which would occupy 5 volumes at a pressure of 14-7 lbs. per sq. in. at 100° C. is heated to 1830° C. at volume i, and then expands adiabatically. The temperatures are marked on the diagrams. Any other adiabatic between the temperatures on the line shown may be calculated from the diagram. Calculating for the temperature conditions of the Clerk diagrams, shown at fig. 105, the ideal efficiency of the actual working fluid there used is found to be 39'5 per cent. From this it appears that the air-standard efficiencies are too high. The actual properties obtained from these experiments only give 39-5 per cent, efficiency, while the air standard gives 49 per cent. As the engine balance-sheet shows 347 indicated efficiency 34^^ = 0-878 ; that is, the actual enigne has converted 88 per cent, of the heat which it possibly could convert into indicated work. This shows that forgiven expansion the best engines have approached very closely to the theoretical realisation of their cycle. The complete suppression of all heat losses due to conduction, &c., on the explosion expansion strokes 272 THE GAS, PETROL, AND OIL ENGINE could only increase the indicated power from 347 to 39-5 — that is, improve it by about 13 per cent. This at once explains why so little difference has been found in the economy obtained from large engines as compared with small ones. In order to check the new method by a test of an entirely different engine, experiments have been made in the author's laboratory upon a Stockport gas engine, 6^ ins. diameter by 13 ins. stroke, giving about 5 HP. The engine is too small, and the heat loss is too large, to enable an accurate determination of specific heat to be made from that engine. The mixture used was slightly different, and the coal gas was h830°( 300- \mo' C. tso- \-l6i o°c. ¥ \ S \ i \ i 150- r«50° 60CXX) / 'y Dit4ald Clerk'a r esiJta= <■ 1 m-'c tCotiL- (?<«V^ /' 40000 J y-- U-f/uU Hll-^CUi ^ y^ • ,.-■■ , ■ "^ ( ^^ -<^ 'C^J9.(j. ir.) ( ^ ^ ^' 400" 800° lecjo" 1200" Temperature C. Fig. 113. — Internal energy curves for weakest and strongest mixtures used. (Hopkinson) temperatures. The values given by Clerk for a mixture of intermediate composition are also shown. The ideal engine efficiencies for the two mixtures can be calculated from these curves by the method given by the author in the discussion on Mr. Dugald Clerk's paper before the Institution of Civil Engineers.' The ideal efficiencies corresponding to mixtures containing respectively 8"8 per cent, and ii"4 per cent, of coal gas, calculated by this method, are 42-4 and 39-4 per cent, respectively. For mixtures of other compositions the efficiency will foUow a straight line law sufficiently nearly for present purposes, and this straight line is shown dotted in fig. iii. It is worth noting that the two straight lines on that figure, if produced, would cut the line corresponding to a ' " On the Limits of Thermal Efficiency in Internal Combustion Engines," Pro- ceedings Inst. C.E., vol. clxix. page 157. VOL. I. U 290 THE GAS, PETROL, AND OIL ENGINE zero gas consumption, at 50'6 per cent, and 52'6 per cent, respectively. The air-cycle efficiency for this engine is 52-2 per cent. In other words, if it were possible to burn weaker mixtures — say, by using stratification — and if the actual and ideal efficiencies continued to bear a linear relation to the gas consumption, these efficiencies would tend to become equal to one another and to the air-cycle efficiency with a very small gas consumption. The ideal efficiency ought, of course, to approximate to the air-cycle efficiency when the charge is greatly reduced ; the close agreement in the other case is no doubt, to some extent, accidental.' ' Without laying too much stress on the absolute values ' of the real and ideal efficiencies shown in fig. iii, it is apparent, from the ratios that they bear to one another, that, while much of the supe- riority of the weaker mixtures is to be ascribed to increase of specific heat, that cause is not sufficient to account for the whole of the effect. Comparing the actual with the ideal efficiency, it wUl be seen that for a mixture containing 8-5 per cent, of coal gas the ratio — usually called the efficiency ratio — is oSy, but when the proportion of coal gas is increased to ii'O per cent, it is only 0"83 ; the weaker mixtures, in addition to giving a higher ideal efficiency, come nearer in practice to realising that ideal. This is due to the fact that the percentage of heat lost to the walls during expansion is less with small gas charges than with large. The difference is sufficient to counterbalance an influence tending the other way, viz. the more rapid combustion of the stronger mixtures. This has been established by a series of experiments directed to that end.' The author agrees with Professor Hopkinson as to the causes of the improvement with weak mixtures — it is undoubtedly due to diminished specific heat and diminished heat loss to the enclosing walls ; indeed, in the paper to which Professor Hopkinson refers, the author gives a table showing the change in ideal efficiency owing to the change in apparent specific heat, the maximum temperatures being taken as 1600° and 1000° C. respectively. At a compression ratio of ^, the author's values show an improvement of over 2 per cent, due to this cause alone. The author differs from Prof. Hopkinson in his calculation of the ideal efficiencies proper to the two mixtures : the ideal working fluid in the author's view should be taken as at the same maximum tempera- ture as is attained by the actual explosion, not at the temperature which would have been attained had combustion been complete at the beginning of the stroke. In this way Hopkinson takes the ideal maxi- mum temperature of the strong mixture to be 2210° C. and that of the weak mixture 1940° C, temperatures somewhat above those ' The absolute values of the efficiencies are affected by any errors in the calorific value of the gas or in the indicator calibration ; and may all be wrong in any experiment by as much as I per cent. But the relations between the efficiencies with different strengths of mixture will be unaffected by these errors, since they are based upon m^ssurgrnqnts with the same indicator an4 the same gas, THERMAL AND MECHANICAL EFFICIENCY 291 which would be given by the strong and weak mixtures used. The error introduced, however, for this particular purpose is too small to affect the accuracy of Hopkinson's conclusions. In order to determine the proportion of the improved efiF.ciency which is due to reduction of wall loss, Hopkinson has made experiments on the wall loss, which he discusses as follows : ' As pointed out by Mr. Clerk, the ordinary method of determining wall loss by the amount and rise of temperature of the cooling water does not give an accurate notion of the loss of heat occurring in the expansion stroke. Much of the heat in the cooling water passes into the walls after release, and should therefore in a proper heat balance be credited to exhaust. In a true heat balance the measured items must be the work done and the energy contained in the gases at the end of expansion, the heat loss during expansion being obtained by . difference. Such a heat balance has been formed for the weakest and strongest mixtures used in these experiments. ' The energy at the end of expansion is in part thermal, and in part the chemical energy represented by unburnt gas. For the calcula- tion of the first item the data required are : ' (i) Temperature of gas at end of expansion ; ' (2) Quantity of gas present ; ' (3) Its internal energy as a function of its temperature. ' The quantity of gas present is known from the suction tempera- ture and suction pressure ; it may be taken as i'o6 standard cub. ft. per explosion in full-load running with a medium jacket. The internal energy is given by the curve, fig. 113. The temperature at the end of expansion can be inferred from the pressure of the gases at release. For measuring this the indicator was fitted with a large piston giving an open scale. A series of consecutive tests were made, the gas charges being alternately about ci and 0'i3 cub. ft. per explosion. In each test the gas charge was measured by gas-holder as described above, and the release pressure was determined simultaneously either by photo- graphing the diagram or by reading it off in the telescope used with the indicator. The calorific value was also determined during the course of the experiments. The following table gives the mean of the results obtained in a series of such tests which show very good agreement : A (weak mixture) B (strong mixture) Gas per explosion as measured by holder . Gas used per explosion (std. cub. ft.) . . Percentage of coal gas present in cylinder contents. Pressure at release (lbs. per sq. in. absolute) Pressure at end of expansion (lbs. per sq. in. absolute) Temperature at end of expansion (absolute C.) 0-1007 0-095 8-5 52 45 1180° 0-1294 0-122 II-O 57 49-5 1290" 292 THE GAS, PETROL, AND OIL ENGINE From these experiments and his internal energy curves, fig. 113, Hopkinson calculated heat balance-sheets for the weak and strong mixtures A and B. The following table shows these balance-sheets, to which has been added the balance-sheet of the ' X ' engine of the Institution of Civil Engineers' trials as determined by the author's new method, see p. 269. BALANCE-SHEET EOR CrOSSLEY II J X 21 INS. ENGINE WITH WEAK AND Strong Mixture, together with Balance-Sheet of ' X ' Engine BY Clerk's Method Hopkinson's experiments Clerk's experiment X engine Crossley A (weak mixture) Crossley B (strong mixture) Indicated work . Heat in exhaust . Heat loss in expansion Per cent. 37 42 21 Per cent. 33 39 28 Per cent. 34-6 49-3 i6-i 100 100 loo-o The exhaust gases were discharged into the exhaust calorimeter and instantly cooled down by water jets j the gases were collected and analysed. In five analyses at full load the percentage of fuel dis- charged unburned varied from o-2 per cent, to 1-5 per cent., so that it may be taken that the results above shown are not materially affected by imperfect combustion. The saving of heat due to the weaker mixture is thus seen to be 7 per cent., that is, 7 per cent, less of the total heat given to the engine passes through the walls during the explosion expansion stroke in the case of the weak mixture. Professor Hopkinson considers that this 7 per cent, would add 2 per cent, of the total heat to the work area, so that the 4 per cent, difference between the weak and strong mixture is due one-half to change in specific heat and one-half to saving in heat loss to the walls. Hopkinson has compared thftheat loss to the walls by the total energy method above described with the heat loss determined by the ordinary method of measuring the heat carried away by the jacket water for the two mixtures weak and strong, having charges respectively O'l and 0-13 cub. ft. of gas per explosion, and he finds that, measured thus, the heat loss is 27 per cent, and 33 per cent. Adding to this the radiation loss from the hot engine surface, he finds the total losses amount to 30 per cent, and 36 per cent. The values compare as follows : THERMAL AND MECHANICAL EFFICIENCY 293 - A (weak mixture) B (strong mixture) Heat loss on expansion Heat loss by jacket and radiation Difference .... Per cent. 21 30 Per cent. 28 36 9 8 This shows that, determined by the ordinary jacket method, too much heat is trapped to the extent of 9 per cent, and 8 per cent, respectively of the total heat of the gas present. This corresponds exactly with the difference shown by the Clerk method and the Institution of Civil Engineers' Committee method described at p. 270. The two balance-sheets of the ' X ' engine are given below : Committee's trials Clerk's new diagram trials Differences Heat flow during explosion and expansion Heat contained in gases at end of expansion Indicated work Per cent. 25-4 39-9 34 7 Per cent. I6-I 49-3 34-6 Per cent. + 9-3 - 9-4 + o-i lOO'O 1 00-0 o-o The ' X ' engine trials thus show the difference between the jacket method and the Clerk new diagram method to be 9^3 per cent, of the total heat, corresponding to Hopkinson's experimental difference of 9 per cent. It may thus be taken as fairly well established that the heat absorbed from the discharging exhaust gases by impinging on the water jacket round the exhaust valve, and the heat passing on the exhaust stroke, is in the case of weak mixture about 9 per cent, of the total heat. Hopkinson states truly that no satisfactory method of determining radiation has yet been proposed, but he arrives at an approximation by determining the heat taken away by the water jacket when the exit temperature is 70° C, as compared with the engine running under exactly the same full-load conditions with the exit temperature at 40° C. He finds the difference to be between 2 per cent, and 3 per cent, of the total heat-supply ; that is, between 2 per cent, and 3 per cent, of total heat-supply less finds its way with the jacket hot and jacket cold. This, of course, assumes that the heat flow from the hot gases to the walls does not vary with the exit temperature of the water jacket. This the author fears is an assumption which cannot be truly made 294 THE GAS, PETROL, AND OIL ENGINE even when the diagram shows but small differences between jacket hot and jacket cold. Hopkinson's figure of 3 per cent., however, cannot be far out, as the true radiation from the 'X' engine was determined by the Committee method as 2'4 per cent. Hopkinson has made a number of interesting experiments with this Analysis of Exhaust Gases from Crossley iiJ x2i ins. Engine under DIFFERENT CONDITIONS OF RUNNING. ALSO GaSES FROM BoYS' AND Junker Calorimeters Test No. 13 Nature of Gas Analysed Exhaust from Boys' Calorimeter — =7-1 gas Exhaust from Junker Calorimeter — = 9'6 gas Engine Exhaust. Full load. Gaso'i204(ii'25percent.) Engine Exhaust. Full load. Gaso"i228(ii"46percent.) Engine Exhaust. Full load. Gas O'io02 (9-37 per cent.) Engine Exhaust. Full load. Gas 0-1320 (12-3 percent.) Engine Exhaust. Half load. Gas 0"i2i2 (11 '3 per cent.) Engine Exhaust. Half load. Gas 0-1199 (11 -2 percent.) Engine Exhaust. Full load. Gas 0-13 (i2-2 per cent.) . Same samples as 9 Engine Exhaust. Full load. Gas o'l (9"4 per cent.) Engine Exhaust. Half load. Gas o'loS (lO'O per cent.) Engine Exhaust. Half load. Gas o'i285 (i2'i per cent.) Quantity of Gas used Litres 1-53 174 1-5 1-94 1-63 1-5 1-5 i-S I'O 172 2-0 1-56 I 1-9 Steam Mg. 0-8 3-0 4-0 2-3 1-4 2-8 2-8 I '3 1-5 I '2 3-3 Per cent. 0-3 1-8 07 0-8 27 27 07 0-5 o'5 4-2 67 CO. Mg- «„■;. -0-4 0-S 1-2 0-2 1-3 0-6 57 3-2 O'l 07 6-8 2-3 2-8 07 2-5 1-8 6-3 3-5 0-3 -o-i I-S 5-6 2-9 Per- centage of unburnt coal-gas 1*0 1-5 0-4 i-i 0-6 4"2 3-0 0-4 0*2 0-6 5-5 4-5 engine running about half load with the hit-and-miss governor, and so scavenging freely. He finds from exhaust gas analysis that at half load a considerable quantity of the gas supplied escapes unburned. He made four analyses of the exhaust when the engine was missing every other stroke, and found 4*2 per cent., 3'2 per cent., 5-4 per cent, and 4'5 per cent, of the total gas unconsumed, the average of the four experiments being 4-5 per cent. This is a most interesting and THERMAL AND MECHANICAL EFFICIENCY 295 unlooked-for effect of scavenging, and the deficiency is fully proved because of five trials with half load, in which aU quantities were determined necessary for a balance-sheet, he found an average deficit of 10 per cent, of the total heat on the higher calorific value. Hopkinson also determined the percentage of unburned gas leaving calorimeters of the Boys and Junker type. The table on p. 294 gives the result of these important experiments. Many other interesting and valuable facts are given in this important paper,' to which the author would refer the reader. Professor Hopkinson is to be congratulated on adding very mate- rially to our exact knowledge of the subject by his careful and ingenious study of a 40 HP Crossley engine. Professor Burstall's Experiments on a Premier Gas Engine OF 16 INS. diameter Cylinder and 24 ins. Stroke, using Mono Gas and Various Compression Ratios.^ The Gas Engine Research Committee of the Institution of Mechanical Engineers was originally formed in the year 1897, and Professor Burstall, of Birmingham University, was appointed reporter, to conduct the experiments and report to the Committee. The present constitution of the Committee is as follows : Chair- man, Sir Alex. B. W. Kennedy. Members : Messrs. Fielding, Humphrey, Burstall, Dugald Clerk, and Captain Sankey. Reporter : Professor Burstall. Two reports have been already presented — in 1898 and 1901 — but these were concerned with a small engine of 6 ins. diameter and 12 ins. stroke. Professor Burstall's present experiments were made on a specially modified Premier engine operated with Mond gas at compression ratios varying from -— ^ to ^—. The smaller compression space thus gave compression of over 200 lbs. per sq. in. Professor Burstall thus describes the object of the tests : ' The tests were undertaken to determine in the first place the thermal efficiencies based on the indicated horse-power at various compressions, having regard to the richness of mixture, and in the second place to formulate if possible the law connecting efficiency and compression. Thus at each compression it was proposed to run a series of trials with different mixtures, which was done by using a number of different mixing valves in which the ratio of the air and gas ports varied. Had the composition of the gas throughout the tests been uniform this would have been a simple matter, but as the producer plant was in general worked at a fairly light load, it was ' Institution of Mechanical Engineers, 1908. ^ Institution of Mechanical Engineers, 1908. 296 THE GAS, PETROL, AND OIL ENGINE impossible to ensure beforehand that the composition of the gas should be exactly what was required for the particular valve employed. The calorific value of the gas aimed at throughout the tests was 160 B.Th.U. per cubic foot (lower value). ' In the first instance a number of preliminary tests were run with various ratios of compression in order to obtain some idea of the engine generally and of the degree of accuracy with which experiments could r \ ^ :\ Fig. 114. — Indicator diagrams. Institution of Mechanical Engineers' experiments. (Bursiall) be made. As was only to be expected in a novel engine a considerable number of changes had to be made, particularly in the governing mechanism. New air and gas valwes, with an entirely new arrange- ment of cams for driving them, had to be constructed, so that the engine as used for the trials differed very materially from the ordinary Premier engine, to such an extent, indeed, that it should be discussed entirely on its own merits. ' During the preliminary trials several types of indicators were used, including the Wayne rotary which was employed on the former trials, when using coal gas. This, however, was found to be so sensitive to the small quantities of dirt present with the producer-gas that its THERMAL AND MECHANICAL EFFICIENCY 297 use had to be abandoned, and the Crosby indicator was used instead. All the diagrams were taken by Mr. J. F. Gill, the reporter's assistant, the greatest care being exercised, in view of the fact that the indicated power was the most important measurement to be made. The barrel of the indicator was driven by a steel wire connecting the reduction gear on the piston to a bell crank about a foot above the indicator, from which it was driven by a very short string, thus getting rid of any errors due to the string itself. The diagrams were taken on smooth surface writing paper with a 6 H drawing pencil sharpened to a fine point, and before taking each diagram the indicator piston was lubricated. Three diagrams are reproduced on fig. 114. The spring employed was calibrated under steam pressure against a standard gauge, which itself had been cahbrated against a dead-weight tester. The whole of the diagrams for a given test were measured up by the method of ordinates, and the mean of these ordinates plotted on squared paper. In this way it was found that a regular curve passed through all the points, and that any oscillations in the expansion Unes caused by inertia cancelled out. The consistency of the diagrams is shown by the two trials F 2 and F 5, in which the indicated power is very nearly the same, while both the calorific value and quantities of gas widely differ, and, at the same time, the thermal efficiency of the engine has the same value. The power required to drive the engine itself was estimated from electrical data to be about 22 horse-power at full load, and from similar data the power empty to be about 20"8 horse-power. In all cases the indicated horse-power is calculated from the actual mean effective pressure on the piston. Reproductions of a number of diagrams were given in the paper. Losses due to back pressure and suction, and also the work required in the scavenger cylinder, are given in the tables. ' The compression of the engine was varied by inserting packing pieces at the big end of the connecting-rod, so that the compression space was always cylindrical in form, thus removing one of the defects of the former set of trials, in which the compression was varied by bolting a junk ring to the back of the piston. The variation in the mixture was obtained by using gas valves having different-sized ports for both air and gas, and also by throttling down the gas at the inlet. Although this latter method is very much more limited than could be desired, owing to the fact that it only alters the quantity of gas, it will be highly desirable in an engine of this kind to have the air for suction separate entirely from that for scavenging, so that both the air and the gas can be throttled during the progress of running the tests. This, however, was not foreseen at the time when the engine was constructed, and although the change could have been made, it would have involved a very considerable amount of alteration, which would have very much delayed the publication of the results.' To Met Vahfe Seal Jo Inlet Vulm •r Service ""lb Exhaust Uzh'e- h/jrutio/i Phig 300 THE GAS, PETROL, AND OIL ENGINE The details of the engine experimented with eire thus described : ' The engine chosen was that designated by the letter " O," capable of giving 150 horse-power at a speed of 170 revolutions per minute, the size of the cylinder being 20 ins. in diameter by 24 ins. stroke. In order to enable the engine to run at a compression pressure of 200 lbs. persq. in., with charges which were estimated to give an initial pressure of 600 lbs. per sq. in., the diameter of the cylinder was reduced to 16 ins., and, at the same time, in place of using the standard admission valve on the top and exhaust at the bottom, an entirely new breech end was constructed, with the admission and exhaust valves horizontal, care being taken that the interior of the cyUnder should have a per- fectly flat end, hke the cylinder of a steam engine. The cross sections and side elevation are illustrated in figs. 115 and 116, and a photograph is shown on fig. 117. General Data Particulars of engine Ins. Mm. Diameter of piston .16 4o6'4 Diameter of differential piston 19 482*6 Stroke 24 609-6 Test Clearance volume Clearance surface Ratio of compression Cub. in. Litres Sq. in. Sq. cm. A 682 II-I8 619 3992 8-07 B 726 1 1 -go 632 4076 7-65 C 776 1272 643 4147 7-22 D 833 13-65 657 4238 6-79 F 927 15-19 682 4399 6-20 J 1084 17-77 701 4521 5-45 Q 1436 23-52 807 5205 4-36 ' The engine was so constructed that it could be worked on any one of the three known systems of governing — namely, {a) keeping the quality of mixture constant, and varying the amount ; (&) keeping the quantity of mixture constant, and varying the amount of gas ; (c) hit-and-miss, or cutting out of charges. The engine was originally arranged to work on system {a), but during the whole of the experi- ments it was arranged to work orilfeystem (&). As the tests are all at full power, the difference between the two is extremely small. ' The working of the engine is as follows : Starting with the suction stroke, the combined air-and-gas valve is opened to a pre- determined point by a pivoted lever under the control of the governor and a positively driven pecker block, actuated by the half -speed shaft, the governor thus controlling the opening of the air-and-gas valve. The mixture, after passing through this valve, enters through the breech end into an annular casing, which contains the inlet valve, and li THERMAL AND MECHANICAL EFFICIENCY 301 then into the cylinder itself. After shutting the inlet valve the usual sequence of compression, explosion, expansion, and exhaust follows, but about half-way along the exhaust stroke a second valve, called the scavenger valve, lying alongside the mixing valve, is opened from the lay shaft, and allows a current of cold air from the differential piston to enter into the motor cylinder. This serves the double purpose of clearing out the exhaust products, and at the same time cooling the inner surfaces. During the idle stroke of the engine this scavenging charge is simply compressed and expanded in the passages leading up to the mixing and scavenger valves. In order to prevent, as far as possible, any possibility of pre-ignitions occurring through hot surfaces, every part of the engine exposed to the flame is water- jacketed, and, in order to estimate the amount of heat rejected through each of these surfaces, the water services are taken from separate measuring-tanks, and discharged without admixture with water from any other surface. The temperatures of discharge were in each Ccise measured by ther- mometers placed in the outlet pipes. The number of separate services are as foUows : ' (i) The barrel. This includes the water round the liner only. ' (2) The breech, which includes the water supplied only to the flat end of the cylinder end. ' (3) The piston ' (4) The inlet valve, inlet casing, and exhaust valve. ' As the engine was intended not only for experimental purposes but also to take its due share of the general work of the university, it was direct-coupled to a 55 -kilowatt direct-current dynamo, built by the Westinghouse Company to supply current at from no to 125 volts. ' The engine works entirely with producer-gas made from bitu- minous fuel, supplied from a 500 horse-power producer of the Mond pattern, made by the Power Gas Corporation. After the gas has passed through the sawdust scrubbers, and entered the power house, it passes into a standard wet gas-meter, having a capacity of 8000 cub. ft. of gcis per hour, made by Messrs. Braddocks, of Oldham. On issuing from the meter the gas passes through a Stott governor to the engine. Without the governor it was found that the suction of the engine produced violent oscillations of the water in the meter, which, besides affecting the accuracy of the readings, threw it out of order. Before commencing the tests the meter was calibrated against a standard meter by Messrs. Braddocks, and the water-level was kept constant throughout the whole of the experiments by running water continually through the meter. ' Ignition. — The engine was at first fitted with the ordinary make- and-break ignition, working from a no-volts circuit. This worked quite satisfactorily until the compression pressure was raised to 140 lbs. 302 THE GAS, PETROL, AND OIL ENGINE per sq. in., when premature ignitions resulted from the heating of the ignition points. This was more particularly the case when rich charges were used. As it would have been very difficult to use the magneto in the cramped position in which the ignition plug lay, high-tension ignition seemed to offer the most favourable opportunity of success, and of such the best seemed to be the Lodge ignition, which is so well known that a brief description will be sufficient. It consists of the usual induction coil and trembler, the primary circuit of which is made and broken by a contact on the lay shaft. This allows the timing of the spark to be readily adjusted while the engine is running, a matter of very considerable importance in an experimental engine. The secondary circuit is led to one of the coatings of a Leyden jar, the other coating of which is connected to the insulated central stalk of the ignition plug, one side of the high-tension circuit being, as usual, con- nected to earth. When the primary circuit is completed and the trembler in motion, a current passes through the secondary circuit, charging the Leyden jar and inducing another current on the opposite coating. When the potential is sufficiently high the coil discharges across a pair of outside points, and the Leyden jar itself discharges across the spark gap in the engine cylinder. According to Sir Oliver Lodge, the inventor, this " B " spark, being very rapidly oscillatory, is unaffected by water, dirt, or oil in the engine cylinder, and the reporter's experience of this system, extending over some two and a half years, is that it gives no trouble, provided the batteries which supply the primary current are always kept charged. ' Considerable trouble was experienced with the ignition plugs, par- ticularly owing to the lack of mechanical strength in the earlier forms, as the sparking head was apt to shear away from the centre stalk. This was prevented by using a steel tube both screwed and brazed on to the head, which gave so little trouble that one plug was in use con- tinuously for more than twelve months without cleaning or alterations. ' When, however, the compression was above i6o lbs. per sq. in., it was found that the central stalk of the ignition plug, which of necessity had to be insulated, became very hot, and the charge was apt to be fired from the hot metal. For the higher compression, therefore, the centre stalk was mfcde hollow, and a stream of oil dis- charged direct on to the firing head in order to cool it, which entirely cured any premature ignition from this source. That this precaution was absolutely necessary was noticed during the running of the " A " tests, when once, the supply of oil having failed for a few minutes, distinct signs of premature ignition could be observed on the diagrams. 'Results of the Tests — It is not proposed to analyse at any length the results which are given in the tables, as it would take up not only a great amount of space, but might possibly lead to misapprehension on THERMAL AND MECHANICAL EFFICIENCY 303 account of having to employ constants, the values of which are not agreed upon by all authorities. ' In considering the thermal efficiencies, it will be noticed that for each compression there is a particular mean pressure which gives the highest economy for that compression. This pressure appears to range between 85 lbs. and 95 lbs. for all the compressions, with the tendency to increase as the compression goes up. Higher mean pressures than these caused the efficiency to fall off, as is shown in tests Qi, F2, F5. The rise in efficiency with the compression is very marked from the Q to the F trials, rising from a minimum of 28 per cent, up to a maximum of 39 per cent. After this point the efficiency increases comparatively slowly, reaching the maximum of 43 per cent. on the C trials, and then diminishing to 39 per cent, on the highest compressions of all. This result does not accord with the usual belief that economy increases with compression, when a suitable mixture is used. The cooling action of the walls, however, affects the result materially. Consider the contents of the cylinder at the end of com- pression. The gas is confined in a space 16 ins. in diameter, at the highest compressions about 3} ins. long, and at the lowest compressions about 6g ins. long, the gas being entirely surrounded by water-cooled surfaces. This being the case, the leakage of heat during compression will be greater proportionately at the high than the low compressions, because the higher compression is accompanied by a higher density and by a temperature difference between the walls and the charge, and this more than compensates for the reduction of the area of surface ex- posed to the gases. Hence, after some definite compression is reached, further compression will result in a loss of economy and not a gain. For this particular engine the most economical compression pressure is apparently 175 lbs. per sq. in. ; but, of course, the particular com- pression that will give the highest economy will vary according to the design of the clearance spaces, but it does not seem to be probable to get a design which will give better results than in the engine experi- mented upon. In order to obtain higher thermal efficiencies by the aid of higher compressions, it would be necessary to increase the stroke of the engine in proportion to its diameter. In the particular engine experimented upon the stroke is one and a half times the diameter. If the stroke were twice the diameter, it might be possible to employ a higher compression pressure. In this way the disc of hot gas might still be kept fairly thick, but, of course, such a method would mean slower speeds of rotation for a given piston speed, and thus it is quite probable than the lower speed of rotation might produce prejudicial effects, which would more than counterbalance the gain due to heat losses. Very high mean pressures, extending to some 114 lbs. per sq. in., were proved to be very decidedly uneconomical, the economy 304 THE GAS, PETROL, AND OIL ENGINE falling from 39 per cent, to nearly 32 per cent, in the cases of F4 and F5. Such mean pressures were not used on the C, B, and A trials, because during the preliminary trials it was found that the economies were not good, and, moreover, that the maximum pressures produced — some 600 lbs. — were too high for safety. In the actual tests the maximum pressure that was allowed was 550 lbs., and this was only rarely reached. The heat rejected to the cooling water does not repre- sent the whole of the heat lost to the walls, because the scavenger charge carries some portion of heat from the interior walls of the cylinder, and that heat is thrown into the exhaust. Hence, the values found for the heat rejected into the jacket water are lower than those which are generally obtained for non-scavenging engines. ' It will be noticed that in most cases the high thermal efficiencies coincide with the lowest percentages of heat lost in the cooling water ; as, for instance, in the trials Ci, D5, F4, while in the case of the A test at the very highest compressions of all, the quantities of heat rejected are greater than those in the B and C trials, thus corroborating, to some extent, the conclusion that it is possible to carry the compression too high in a particular engine. ' The thermal efficiencies throughout are computed by taking the heat values from the Junker calorimeter, which was kept running throughout the entire period of the running of the tests. The heat rejected to the different portions is as given in the tables, and it will be noticed that the barrel takes, roughly speaking, 50 per cent, more heat than the breech, that the piston takes about half the heat of the breech end, and the valves about three-quarters that of the breech. The explanation of the breech end taking more heat than the piston, although their surfaces exposed to the hot gases are the same, is the fact that the exhaust pipe passes through the watered breech end, and therefore abstracts heat during exhaust. ' Analysing the exhaust gases, it is possible to calculate the ratio of air to gas, and then, by assuming that the scavenger charge entirely removes all the products of combustion, to calculate the temperature at the end of the suction stroke. This has been done for the trials in which the mechanically operated sampling valve was used, and it gives the ratio of the air to gas varyi% from iS to 3-6, and suction tem- peratures varying from 44° C. to 121° C. (111° F. to 250° F.). Further details are not given, because this method of deriving the value of ratio of air to gas from chemical analysis is subject to so many experi- mental errors as to render the figures so obtained not of much value, but probably for weak charges the suction temperatures will not be far removed from 50° C, rising to 100° C. for the rich charges. The maximum temperature calculated from these values was about 2000° C. (3632° F.), dropping to as low as 1200° C. (2192° F.) in all the econo- mical runs of the tests given in the C trials. THERMAL AND MECHANICAL EFFICIENCY 305 ' The whole of the experiments appear to point conclusively to the fact that the most economical mean pressure is very considerably below the maximum which can be obtained, and that the highest economies are obtained with a comparatively low maximum temperature. Both these results imply that the engine should not only be subjected to lower pressure, but to lower temperatures as well, and thus many of the difficulties which arise in large engines from rich charges might be avoided, and the maximum pressures kept down to quite reasonable limits. ' This, of course, only applies to the indicated power, and the con- clusions as to the brake horse-power would be widely different. If, however, the engine is constructed to work only with these moderate pressures and temperatures, the whole of the working parts might be very much lightened, and thus a good mechanical efificiency obtained with the very moderate mean pressures. ' The question of the liability to premature ignitions, of gas con- taining larger or smaller percentages of hydrogen, was borne in mind throughout these experiments, but in every case of premature ignition which occurred — and many such cases occurred — with compressions higher than 160 lbs. per sq. in., it was traced to dirt or carbonised oil in the cylinder, or to some part having got overheated, and such premature ignitions took place equally with a weak as with a rich gas. ' The reporter is of opinion that, as far as premature ignition goes, the compression might be made a great deal higher than any which have been used during these experiments, but in view of the fact that the economy falls off after a certain point, there does not seem to be any useful object gained in going to any higher compression.' The table on the next page gives the most important results ; for other interesting tables the reader is referred to the original paper. In this research Professor Burstall arrives at two main conclusions. First, that for any compression the highest economy is obtained when the mean effective pressure lies between 85 lbs. and 95 lbs. per sq. in. Second, that thermal efficiency increases from a compression ratio - of from —V to - ■"■- , and falls off at the higher compressions —^- 4-36 7-22 "^ ^ 7-65 and^i^-. 8-07 It is quite true that lower maximum and mean temperatures are the most economical, and under some circumstances that, of course, means that lower mean pressures are economical in the same engine as compared with high mean pressures. It by no means follows, how- ever, that the limits always lie between 85 and 95 lbs. per sq. in., although no doubt they did so in the special circumstances of Burstall's VOL. I. X 30& THE GAS, PETROL, AND OIL ENGINE M.E.P. on sca- venger reduced to area of piston VD CO ts* t^ 'O V^ 1" bob b boo n M M O ■* 1 wo M u-)u-ivo CO in -^ M op t^ p\'p vpt-vroi-iTt-Ownni-ip'Oc^ tI-'O »P c^ Cd (J <-* OsmO '^'-' '^^^ (^ M t-i -^M tv.oooo ■^O\'^^0oovpop CvlCv)Mwi-iNM«MNHint-HMw«cOWMC4MM w -LnO CT\0 ■^'O CTimoc mO mO\« t^tnO'O M NOO rooo 'O f^ rooo 00»O 0^0\f^)l-lCO rOM U11-. MOO t-iVOOO vo" vo" vo" 'o" in\o''vb"y5''o"'o'\o'''o"'o' t^'o" tC tC tC i-C d\oo co~ Thermal effi- ciency average ■^ CO ^^ '^ 0\ ii-1 O c^ o\ "-• o ^ -^ o p CO :=f ^ ro p r^ -J 39- ••5S "1 1 ONOO -^M COT:J-Ti-N\0 lx(^iH t-^Ooo ^0*0 -^"^ 00 00 O\oo 0\ O^ 0\ O I-* >-.\b t^. ^nsb O 'O ^v ri O « tx 0\^ O N CO ro •"^'O CO « 'O ■* O W oo oo 00 00 00 a\ O\oo O\0 MOochOi'-tootHoo O OsO rsvoo 1"" bj} CL . N P co« pco^ ^-i r^pr^v^p^r^ r^p*?* r r* P r P u-i ^s. u-> to u-i M o ■^oo c>NtH cow o '-' Tf-coou-i Tj-vb On ^^■l 00 Th O O vo T^ CO fooo CO 0\^ i-f O 0\Tj-o Thoo m cso O 1000*0 rou-iN 0\ThOco O COM coo iocom lo-rtM In. cococOfOCOCOM rOTtrococOCOTj-cOCOCOTJ-rocON (N n II top. . Vio^coN N w M M 't~> b M b\ONb\a\(>o a\«> ^ *b vb is--: 01 M "H M N loco lo 0^\0 "^00 K. On M On tJ- M Tfco tN. t^ w tsLO ON tN tN.VO tN >J-)\o lou-icOM row N CONOOOOOO Spark Dis- tance _of piston per cent. from end of stroke u-i -^ Tt-VO 'O ON « oo N ro coco ioMls.OtN.MOOOO^ N W rococoM « M NcOmM comMi~ip1 ih m N-^M bbbbbbbbbbbbbbbbbbbbbb Position of crank when contact was made, in degrees before dead centre CO ■>!l-'0 CO O\oo T^^ N -^t lo o t\oo r^ -^ t^^vc vo On N co ^O "■>"0 ^ ^ Ti- Lo Ti- lo u^KO "O^O comcoiococo-^tNU^ H2i w N WcOw escort" fO- and C^ be the specific heats at constant pressure and constant volume respectively in foot-pounds per cubic foot at 0° C. and 147 lbs. persq. in. pressure ; since Cj, — C^ is equal to the work done by a cubic foot of the working fluid in expanding against atmospheric pressure 147 lbs. per sq. in., while the tempera- ture is raised through 1° C. ; and since the working fluid, assumed to be a perfect gas, expands through j^g part of its volume for a temperature rise of 1° C, C^_C„ = HZiLi44ft.-lbs. 273 whence^ = ?^= I +MZJLI44 C^ 273 X c„ = 1+^' (I) In order to obtain the adiabatic it is necessary to know approximately the mean temperature on the part of the curve being constructed. It is therefore necessary, in drawing an adiabatic between any two given volumes, say, V„ and V,, to assume values of 7, and so obtain an approximation to the adiabatic, the temperatures on this approximate adiabatic being then taken in order to select 'the proper value of y for the final curve. It is convenient, in constructing the adiabatic, say, between temperatures 1600° C. and 1000° C, to divide the volume into a number of parts, say six, and in the first instance assume that the mean temperature on the curve decreases regularly between the volumes taken. Numbering the ordinates i, 2, 3, 4, and 5, on the assumptions made, the tempera- ture at V„ is 1600° C. At ordinate i the temperature will be 1500° C. There- fore it is required to draw between the volume V„ and the volume at ordinate i the adiabatic with a specific-heat value equal to the mean value between the temperatures 1600° and 1500° C. From equation i given above, and the specific-heat Table, it is seen that the value of y for the temperature range 1600° to 1500° C. is I f 776^j.2325_ 27-5 The point where the adiabatic cuts the ordinate i can therefore be found from the equation p^i .2925 -constant. Thus a first approximation is obtained to the adiabatic between the volumes V„ and V,. From the pressure at ordinate i obtained in this way, the tempera- ture on the adiabatic at i can be got, and, if necessary, a further approximation to the value of 7, which should be taken between the volumes considered. The ■3pecific-heat changes, however, are not sufficient to make it usually necessary to 336 THE GAS, PETROL, AND OIL ENGINE go further than a single approximation. To get a further point on the adiabatic, the process above described is repeated for the volumes V„ Vj, taking, as starting- point, the known point on the ordinate at volume i, and so on. Thus the adiabatic can be found for any temperature range within the temperatures given in the Table of specific heats. APPENDIX III CALCULATION OF EFFICIENCY WITH VARYING SPECIFIC HEAT OF WORKING FLUID Assuming that the apparent specific-heat values are true specific heats, the ideal efiiciency of any engine can readUy be calculated. With varying specific heat, however, the efficiencies will alter as the values of maximum temperature are altered, and it is therefore necessary, in calculating Fig. 120 efficiency, to take definite values of maximum temperature, as well as a definite ratio of clearance space to total volume of cylinder. The value of 7 for the compression line may be assumed to be 1-37. In the figure let OA be the clearance volume, OB, total cylinder volume, D, point of maximum temperature, DE, adiabatic through D, FC, compression line. % Then it is known that the efficiency, 17, _ Heat added from C to D - Heat discharged E to F Heat added from C to D _ ( Tp — Tc) C^ (CD) ~ (T e — Tp) Cp(EF) (To Td — Tc C7, (CD) from which the values of t}, given in the Table, can readily be obtained. BRITISH ASSOCIATION COMMITTEE REPORT 337 ' APPENDIX IV GASEOUS EXPLOSIONS First Report of the Committee, consisting of Sir W. H. Preece (Chairman), Mr. Dugald Clerk and Professor Bertram Hopkinson (Joint Secretaries), Pro- fessors Bone, Burstall, Callendar, Coker, Dalby, Dixon, Hele-Shaw, Smithells, and Watson, Dr. Harker, Lieut. -Colonel Holden, Dr. Petavel, and Captain Sankey, appointed for the Investigation of Gaseous Explosions, with Special Reference to Temperature. British Association, Section G, Dublin 1908. General Scope of the Report To engineers the investigation of gaseous explosions is chiefly of interest because of its bearing upon the theory of the internal-combustion engine. The Committee have hitherto considered it mainly from this point of view, conceiving that a limited interpretation of their reference would be necessary if their labours were to lead to any result within a reasonable time, and that the limitation adopted should be determined by the fact that the Committee was initiated by the Engineering Section. On the other hand, the work has been by no means entirely, or even mainly, of a practical as distinct from a purely scientific character. Many questions of a kind that might properly engage the attention of the Chemical and Physical Sections have been raised and discussed, and full scope has been given to the varied skill and knowledge possessed by the different members of the Committee, among whom, in addition to engineers, there are several whose interests are mainly in the direction of pure science. The test of practical value or interest has only been applied for the purpose of selecting from among the large number of questions arising in connection with explosions those which are proper subjects for investigation by this Committee, not with the idea of limiting such investigation to the practical aspect of these questions. Seven meetings of the Committee have been held, and they have been ex- cellently attended. At each meeting one or more notes written by members of the Committee have been presented and have formed the basis of the discussion . The following is a list of these notes : No. 1. Generallntroduction Dugald Clerk. No. 2. Dissociation of Steam and Carbonic Acid . . Dugald Clerk. No. 3A. Measurements of Internal Energy of Gases up to 1400° C. B. Hopkinson. No. 33. Explosion Pressures as a Means of Determining the Energy Function of Gases . . . . B. Hopkinson. No. 4. Dissociation and Specific Heat of Steam and Carbonic Acid, and Comparison of Gas- Thermometers at very High Temperatures . J. A. Harker. No. 5. The Temperature of the Walls of a Gas-engine Cylinder E. G. Coker. No. 6. The Deviation of Actual Gases from the Ideal State, and the Experimental Errors in the Determination of their Specific Heats . . H. L. Callendar. The essential feature common to the operation of all gas-engines is the conversion of a mixture of inflammable gases, by combustion or explosion, info VOL. I. Z 338 THE GAS, PETROL, AND OIL ENGINE a mass which consists in all practical cases of a mixture of steam, carbon dioxide, nitrogen, and excess oxygen. The performance of the engine depends primarily on the change in pressure or volume, or both, resulting from this chemical transformation, and on the properties of the products of the transformation after they are formed. It depends in only a secondary degree on the nature of the chemical process and on the velocity with which it takes place. These matters, important though they must be in any investigation of explosions and in the theory of the gas engine, are not of the first importance. The foundation must be a knowledge of the properties of the gases enumerated above at the tempera- tures occurring in the gas engine — that is, between 1000° and 2500° C. This Report therefore consists mainly of an analysis of the present state of knowledge on this subject, together with suggestions as to the directions in which further research may be undertaken with the object of advancing it. It will be found, however, that as the mechanism and the velocity of combustion must be taken into account as disturbing factors when applying our knowledge of the gases in the theory of the gas engine, so they enter into many of the experiments on which that knowledge is based, and some discussion of them is indispensable in any criticism of these experiments. Thermodynamic theory shows that the physical properties of a gas in chemical equilibrium are completely specified when — ■ (i) The relation between the pressure and volume at constant temperature is known, and (2) The internal energy per unit volume is given as a function of the tempera- ture and the density. The energy of a gas per unit of mass at temperature 6 is usually defined as A (9 - e„), where 8„ is the standard temperature from which energies are reckoned, and k the mean specific heat at constant volume between the tempera- tures 9, and 8. The second of these data is therefore equivalent to a knowledge of the specific heat in terms of the temperature and density. This form of state- ment is probably more familiar ; but, for reasons given later on, it is in many ways less convenient than that based upon the energy function. For those gases with which we have to deal it may be assumed that the first relation is that given by Boyle's Law. Experiment and theory alike point to the conclusion that deviation from this law only occurs when the density of the gas departs widely from its normal value, and that it is diminished by high temperature. In the gas engine the density of the gas rarely exceeds ten times that of the atmosphere, a point at which the deviation from Boyle's Law in air (at 100° C.) is only about one-half per cent.' It is usual to make the further assumption that the product pv is proportional to the absolute temperature 8. A detailed examination of the grounds of this assumption forms the subject of a section of this Report. At this point it is only necessary to notice that, if it be true, then the internal energy is a function of the temperature only, and is indepeiment of the density. If, on the other hand, the perfect gas law does not hold, then the true relation between pv and B can be deduced from a knowledge of the internal energy, which is in that case a function both of the temperature and of the density. The properties of the gases with which we have to deal are therefore completely defined when the energy has been tabulated as a function of the temperature and the density. So far as the present state of knowledge goes, the energy is to be expressed in terms of temperature only ; but an important part of future ' See Witkowski, PAH. Mag., vol. xli. (1896), p. 309. BRITISH ASSOCIATION COMMITTEE REPORT 339 investigation must dcul with its dcptntluucc on tlie density either by dnuet measurement or by a determination of the relation between /jw and 6 at high temperatures. The prediction of the temperature reached in combustion, which must be the starting-point of any investigation of explosions, also rests primarily upon a knowledge of the energy function. For, subject to corrections for loss of heat, incomplete combustion, and work done while combustion proceeds, the thermal energy of the mixture of steam, CO2, &c., after combustion is equal to the chemical energy of the gases from which that mixture was formed. The latter can be accurately inferred from the composition of the combustible gases, and, the thermal energy being thus known, the temperature can be calculated from a table of the energy function. The pressure or volume changes resulting from combustion can be deduced from the temperature by the use of the p — v — S relations, which again ultimately depend upon the form of the energy function. A table of this function at high temperatures is therefore the first datum necessary for the investigations entrusted to the Committee, and is the principal subject of this Report. Before proceeding to the discussion of this physical question, however, it is well to say something further about its bearing on practical engineering problems. The first requisite for predicting the performance of a gas engine is to know the rise of temperature and the consequent rise of pressure produced by the explosion. The importance of this need not be insisted upon ; it is not only the principal factor in the mean pressure developed, it also determines in large measure the mechanical design of the engine and the necessary strength of its parts. The part played by the energy function in the calculation of this rise of pressure has been indicated in the last paragraph. In proceeding further to analyse the indicator diagram given by the engine with the object of accounting at each point for the heat which has been put in, a knowledge of this function is again required. The heat accounted for on the diagram is the work which has been done plus the heat contained in the gas. The latter item can be calculated from the temperature if the energy function be known. The balance unaccounted for, which it is usually the object of such investigations to find — whether in the steam engine or the gas engine — is the heat which has been lost to the walls or has been suppressed owing to incomplete combustion. In fact, the internal energy of the gases at high temperatures plays much the same part in the analysis of gas-engine phenomena as does the total heat of steam in investigating the working of the steam engine. Again, from a table of internal energy, it is possible to predict the pressure changes resulting from any series of operations such as occur in the gas engine, one item in which is an explosion subject to certain hypothetical conditions which cannot be realised in practice, though they can be indefinitely approached. An ideal diagram of this kind, corresponding to the cycle of operations which is most usual in present-day gas engines, can, for example, be constructed for any given combustible mixture on the assumption that the combustion is instan- taneous and complete at the in-centre ; that there is no loss of heat in compres- sion, explosion, or expansion ; and that during expansion the gases are at all times in thermal and chemical equilibrium. These conditions can never be completely realised, but can in theory be approached asymptotically by improve- ments in design carried on within certain defined limits— namely, that the degree of compression and the nature of the mixture are to be unaltered. For example, the heat loss may be reduced by increasing the size of the engine and altering the z 2 340 THE GAS, PETROL, AND OIL ENGINE nature; of (ho cylinder walls, and the attainment of thermal and chemical equilibrium may be promoted by reducing the speedy Such an ideal cycle is, in fact, precisely analogous to the Rankine cycle of the steam engine, in that it takes account of the actual physical properties of the working substance, but leaves out of account such non-essential imperfections as heat loss to the cylinder walls. It represents an ideal which the real engine may approach indefinitely but can never attain ; and the closeness of the approach is a true measure of the perfection of the engine. The ideal cycle which has hitherto been used in discussing the performances of gas engines is the well-known air cycle. This is based upon a special assump- tion as to the form of the energy function — namely, that it is a linear function of the temperature at high, as it is known to be at low, temperatures. The specific heat of the working substance is taken to be constant and equal to 19 foot-pounds per cubic foot, or 4-8 calories per gramme molecule. In the state of ignorance as to the real form of the energy function which prevailed until quite recently, this assumption was as good as any other, since it was impossible to say that the value of the energy derived from it was further from the truth than any other value which might be assigned to it. So fax as was known, the differences between the indicator diagram of a real engine and the corre- sponding air-cycle diagram might have been wholly, or almost wholly, due to what have been called above ' non-essential imperfections ' — that is, to heat loss and to incomplete combustion. In other words, there was no conclusive evidence that the air cycle was not for practical purposes a true ideal cycle in the sense defined above and equivalent to the Rankine cycle for the steam engine. Under these circumstances its extreme simplicity made it the best available standard of comparison for judging the performance of a real engine. Recent researches, however, on the properties of the gases at high temperatures have definitely shown that the assumption of constant specific heat is erroneous, and have given sufficient information about the magnitude of the error to show that it is of material importance. They have shown that the air cycle cannot be regarded as equivalent to the Rankine cycle in the steam engine, inasmuch as it does not take account of the properties of the actual working fluid, but pos- tulates a hypothetical fluid which has no real existence. It is as though in the theory of the steam engine the total heat of the steam were to be taken as equal to its latent heat, the sensible heat of the water being neglected. This assump- tion would lead to a simpler formula for the ideal efficiency for the steam engine, but it would be erroneous in the same way and to about the same extent as the air-cycle formula for the gas engine.' The closer approximation to the real cycle which is made by taking account of the actual properties of the working fluid — in the steam engine the total heat of the steam instead of only the latent heat, in the gas-engine the true value of the energy instead of that based on the assumption of constant specific heat — though it leads to some complication of formulas, gives compensating advan'ro.ges of real practical value. It shows the engineer what are the limits to the improvements which can be effected by changes of design or increase of size, and it enables him to judge whether it is 1 If the sensible heat of the steam can be neglected in comparison with its latent heat, the Rankine cycle reduces to the Carnot cycle, with efificiency — 1 ^, for no heat is then neces- sary to warm the water from the condenser to the boiler temperature, and the whole process becomes reversible. The efficiency of the Carnot cycle usually exceeds that of the corre- sponding Rankine cycle by about one-eighth part. BRITISH ASSOCIATION COMMITTEE REPORT 341 better that the lines of development should proceed in such directions or in the direction of radically modifying the cycle of operations. Measurement of the Internal Energy or Specific Heats 0/ Gases at High Temperatures The results of most experiments on the energy of gases have been expressed in the form of tables or formulae giving the specific heat (referred to unit mass of the gas) in terms of the temperature. It would appear preferable for most purposes to exhibit them in terms of internal energy per unit volume. That is the form most convenient for purposes of thermodynamic calculation, and it has the further advantage that it expresses the actual quantity measured. In nearly all the experiments on the specific heats of gases the increase of energy in unit volume associated with a large rise of temperature is measured ; and in most the lower Umit of temperature is near that of the room. The rate of change with temperature, of the energy so determined, is sometimes called the ' true ' or ' instantaneous ' specific heat, and sometimes ' thermal capacity.' The Committee are of opinion that a definite name should be given to this important quantity, and they suggest the name ' volumetric heat,' which if adopted should include in its significance that the measurement to which it relates is made at constant volume, and is referred to unit volume of the gas. The term ' specific heat ' could then be restricted to its usual meaning, which refers to unit mass of the substance. Convenience of calculation is promoted if the unit of volume taken is that corresponding to the gramme molecule under standard conditions, which is sufficiently nearly the same for each of the gases under consideration and equal to 22-25 litres.' In this report internal energy and volumetric heat are expressed as calories ^ per 22-25 standard litres ; and the zero of temperature from which the energy is reckoned (except where otherwise stated) is taken to be 100° C, in order that steam may be included on the same basis as the other gases. The results are conveniently exhibited as curves in which the energy is the ordinate, and the excess of the temperature over 100° C. is the abscissa. The slope of such a curve represents the volu- metric heat C, and the ordinate divided by the abscissa for any temperature represents the mean volumetric heat from 100° C. to that temperature, here denoted by C. The experimental work done on this subject may be divided into three classes : (i) Constant-pressure experiments : Regnault, Wiedemann, Witkowski, Lussaua, Holborn and Austin, Holborn and Henning. The gas is heated from an external source in these experiments, and is at atmospheric pressure. 1 The volumes of the gramme molecule for the several gases are : Hj . . . . 22'24 1 CO . . 22'2I N, . . 22-28 ' CO2 . . 22-o8 O.J . . . . 2222 , in litres at o° C. and under a pressure of 760 mm. of mercury. It may be noted here that i calorie per gramme molecule is equivalent to 3-95 foot-pounds per cubic foot. '' There is some difference in the energy value of the calorie according to the temperature at which it is measured. The difference between the ma.ximum and minimum value over the range 0° to 100° C. amounts to about i per cent. This is of no importance for the purposes of this Report, except in one or two places ; but where it is necessary to be so precise the calorie at 15° C. — namely, the quantity of heat required to warm i gramme of water from 14^° C. to 134° C— is meant. 342 THE GAS, PETROL, AND OIL ENGINE (2) Experiments in which both volume and pressure are varied, the gas being heated by compression. The recent experiments of Clerk and the deter- minations of the velocity of sound in hot gas by Dixon and others belong to this class. (3) Constant-volume experiment. To this category belong the explosion experiments of Mallard and Le Chatelier, Clerk, Langen, Petavel, Hopkinson, and others, and Joly's determinations with the steam calorimeter. In the explosion experiments the gas is heated by internal combustion. (i) Constant-pressure Experiments The constant-pressure experiments have been carried to a temperature of about 1400° C. The gas under atmospheric pressure flows steadily through a heater and then through a calorimeter, where it is cooled. The temperature just before entering and just after leaving the calorimeter and the quantity of heat evolved per gramme molecule of the gas are measured. This quantity of heat less the work done in the contraction, which is 1-98 times the fall of tempera- ture, is the change of internal energy corresponding to that fall. Regnault applied the method to air, H,, CO, CO^, and other gases over the range 0-200° C. Wiedemann ' repeated Regnault's experiments with some modifications of the apparatus. On account of the small range of temperature these experiments must be regarded as only giving the slope of the internal energy curve at the origin ; but as they give this with an accuracy at least equal to that with which the ordinate is known at higher temperatures, they are of considerable import- ance in constructing the curve. The following table shows the values of the mean volumetric heat C. over the range o°-ioo C, found by these two observers for air, H, and CO. Witkowski's value for air, by the same method, is in exact agreement with Regnault's : - H, CO Air Wiedemann .... 4-84 4-81 4-90 Regnault . . . 4-84 4-88 4-86 These results give a good idea of the accuracy attained in these experiments. In both sets the different observations ranged about i J per cent, above and below the mean in each determination. Later work shows that the value of C for air is probably about i per cent, greater over the range o to 200 than over the range o to 100. Regnault was unable to detect this difference, though he looked for it. The volumetric heat of air has also been determined by Joly by means of the steam calorimeter. He found the specific heat of air at constant volume for the range 10° to 100° C. and at a pressure of about twenty atmospheres to be 0-172. There were distinct signs of an increase of specific heat with density, and, assuming this to follow the linear law given b^ Joly, the specific heat at normal density would be o-i/ij, equivalent to 4-93 calories per gramme molecule. Professor Callendar points out, however, that this is based upon Regnault's number for the latent heat of steam, which is of doubtful accuracy, and that more prob- ably Joly's determination when reduced to the 15° calorie should be 0-1732, or 4-98 calories per gramme molecule. According to some unpublished experiments by a constant-pressure method, which have been made by Mr. Swann in Pro- fessor Callendar's laboratory, and in which it is believed that some sources of systematic error inherent in the earlier experiments of this type have been avoided, the volumetric heat of air is 5 -o These results are distinctly higher than 1 Ati:ia!eK der Physik, 1S76, vol. clvji. ^ Phil. A/a»: vol. xlii. (1896), p. 5. BRITISH ASSOCIATION COMxMITTEE REPORT 343 those obtained by Wiedemann and Regnault, but the difference is of no import- ance for the present purpose except as an indication of the possibility of systematic errors in tlieir method of experiment which may become important when it is applied to higher temperatures. It may be taken as fairly certain that the volumetric heat of air at ioo° C. is within 2 per cent, of 4-9. In the case of CO.^ the results obtained by Wiedemann and Regnault were ; Increase of internal energy w R to 100° C. 710 680 to 200 1510 1490 100 to 200 800 810 The first two rows of figures represent practically the quantities actually measured in these experiments.' The third is obtained by difference from the first two, and is therefore affected with a greater probable error than either. The result of the two sets of experiments may be summed up by saying that the volumetric heat of COj at ioo° C. taken as equal to the mean volumetric heat between o° and 200° C. is between 7'45 and y55, and that its rate of increase with temperature is between O'OOp and 0'OI3, or roughly one six-hundredth part per ° C. The specific heat of steam at constant (atmospheric) pressure in the neighbourhood of 100° C, according to Regnault, is o'48, equivalent to 6-64 volumetric heat, and subsequent observers have shown that this value is at least as accurate as Regnault's value of the specific heat of air. From Joly's experiments with the steam calorimeter, when corrected accord- ing to Callendar for the error in Regnault's value of the latent heat of steam, the specific heat of CO^ between 10° and 100° C. and at a pressure of 12 atmospheres is 0'i72 and it increases by about 0'25 per cent, per atmosphere. Assuming this law of increase to hold between one atmosphere and 12 atmospheres, the mean specific heat at normal density for the range 10° to 100° should be o-i666, and the volumetric heat should be y$, which is again decidedly greater than the values obtained by Wiedemann and Regnault. According to a recent determina- tion by Swann, the result of which has been communicated to the Committee by Professor Callendar, the volumetric heat of CO^ at 100° is 776 — again materially higher than Regnault and Wiedemann (7'5). Holborn in conjunction with Austin carried the constant-pressure determina- tions for air and CO^, up to 800° C.^ The gas was heated electrically and the temperature was measured with a thermo-couple. Similar measurements on steam were made by Holborn and Henning, who subsequently carried the determinations for the three gases up to 1400° C Holborn and Henning express the results of all these experiments in alge- braical formulae representing the mean specific heats of CO^, air, and steam re- spectively over the range o-9 in the case of the first two gases, and ioo-9 in the case of steam. From these formulae the full-lined curves in fig. 121, exhibiting the internal energy, have been constructed. The actual observations are also shown in the same figure. Each of these observations represents the mean 1 The lower limit of temperature in Regnault's measurements was io°, the upper limits were 100° and 210° respectively. In Wiedemann's work the lower limit was 25° and the upper 100^ and 200'^ respectively. The results have been reduced to the forms here given on the supposition that the volumetric heat between 0° and 25° is 6'6, and at 200'' 74. For the small additional range of temperature required these figures are certainly as accurate as the experiments. ' 2 Jl'iss. Ahhavdlungen der Phys. Techn. Rcichsanstalt, 1905. ' Ann. d. Phys., 23, 1907, p. S09. 344 THE GAS, PETROL, AND OIL ENGINE of a large number of experiments, in some cases as many as thirty. Tlie results of the individual experiments in such a group ranged about 2 or 3 per cent, above and below the mean. These casual errors would no doubt cancel out to a great extent in taking the mean, which, apart from systematic errors inherent in the method of experiment, is probably correct within about 2 per cent. This degree of accuracy is not sufficient to enable any deduction to be made as to the manner of variation of the volumetric heat beyond a rough estimate of its average rate of increase over the whole range of experiment. The following lecoo- / / / • tl a i 1 / /xoi y /\ y^STEt M u CL in I i X / > y" ^ MR y /y y ^ y ^ C^ r*- ^ ^ 600 800 1000 TEMPERATURE. DECREES CENT. Fig. 121. — Curves showing internal energy at different temperatures from Holborn and Henning's experiments are the values of the volumetric heats of air, steam, and CO^ at 100°, 600°, and 1 100° respectively : — - 100° 600° 1100° Increase 100-1100° c y ^ • y c y 0-9 Air 4'9 6-6 1-404 5'2 1-38 5-75 1-345 Steam 1-30 6-85 1-29 8-5 1-24 1-9 CO, 7-5 1-26 9-95 1-20 ii'i i-i8 3-6 The corresponding values of 7 are also shown : 7 = i + BRITISH ASSOCIATION COMMITTEE REPORT 345 The values at 100° are derived from the experiments of Wiedemann and Regnault. Those at 600° and iioo° are based on the specific-heat values given by Holborn and Henning ; in other words, they are obtained by drawing tangents to the curves, fig. lai. The error at these higher temperatures may be double that of the internal energy, or, say, 4 per cent. The figures show that the volu- metric heat of air increases by about 0-0009, that of steam by 0-0033, S'Hd that of CO2 by 0'0036 per degree Centigrade over the range ioo^--iioo° C. There is no evidence that the rate of increase is other than constant in the case of air ; but there can be no doubt that the average rate of increase between 100° and 1 100° in COj is less than half the rate of increase between 0° and 200°, as determined by Wiedemann and Regnault. There is also distinct evidence in these and other experiments that the rate of increase of the specific heat of steam becomes greater as the temperature rises. {2) Clerk's Experiments ' These cover about the same range of temperature as Holborn and Henning The gas used was the products of an explosion in a gas engine, and therefore Fig. 122 consisted of a mixture of CO,,, steam, and air. It was first expanded in the ordinary course after the explosion, and was then heated by compression on the next in-stroke of the engine, the valves being kept closed for this purpose. On the next out-stroke the gas was again expanded, then compressed again, and so on, the valves remaining closed and the engine running on its own momentum. An indicator diagram was taken of the whole operation. The change of internal energy in any portion of a compression stroke (e.g., b c in fig. 122) is equal to the work done less the heat lost to the cylinder walls ; in an expansion stroke (c d) it is the work done plus the heat lost. The work can be obtained from the indicator diagram with an accuracy which is only limited by the indicating appliances. The change of temperature can also be calculated from the indicator diagram subject to a knowledge of the temperature at one point. Errors in the latter, however, do not greatly affect the results found for internal energy or volumetric ' i'/vc. Hoy, Soc. A, vol. Ix.wii. 346 THE GAS, PETROL, AND OIL ENGINE heat, because the figure for the quantity of gas present is affected by these errors in such a way as to cancel out the error in temperature interval. The loss of heat comes in as a correction on the work done and was estimated by a comparison of the compression line and the immediately following expansion line (b c and c d, fig. 122). The calculation is based on the assumption that the total heat loss from the hot gases during any given portion of a stroke is the same in expansion and compression if the mean temperature be the same. In the first compression the temperature of the gas rose to about 1 100° C. (at the point c, fig. 122). During the first three-tenths of the following expansion stroke (c d), the temperature fell to about 700° C. The work done in this part of the expansion was measured and the heat loss determined as above was added. Thus the change of internal energy corresponding to the temperature change iioo°-700° is obtained. The average volumetric heat over this range is within tlie errors of experiment equal to the volumetric heat at the mean temperature of 900° C, which accordingly is by this method determined direct instead of by difference, as is necessarily the case when (as in Holborn and Heuning's experi- ments) the whole internal energy change associated with complete cooling of the gas is measured. In view of the great difference in the method of experiment a comparison of Clerk's results with those of Messrs. Holborn and Henning is of great interest. Clerk's measurements extended to 1450° C, but those above 1200° C. were based on the first expansion Ime after the explosion when the method for getting heat loss would be of doubtful application, and when, moreover, combustion may have been incomplete. It will be better, therefore, to confine the comparison to temperatures of 1200° and below. The following table exhibits the internal energies of the mixed gas with which Clerk experimented calculated from Holborn and Heuning's figures, together with the energy calculated from Clerk's values for the mean volumetric heat. The energies are, as usual, reckoned from 100° C. ; and the energies of an ideal gas with a constant volumetric heat of 4-9 are added for comparison. Temperature Holborn and Henning Clerk Ideal Gai 400 Boo 1200 1580 3840 6285 1720 4300 7040 1470 3430 5 390 It will be seen that Clerk's results are throughout about 10 per cent, higher than the others. The difference between the energy of the real and of the ideal gas, the discovery of which is the true object of these experiments, is about twice as great in the one case as in the other. It does not seem possible to account for so large a discrepancy by ordiAry experimental errors. It must be due either to some systematic error inherent in the method of experiment in one or both cases, or to a difference in the conditions of experiment giving rise to a real difference of internal energy. Professor Callendar has favoured the Committee with a note dealing with the constant-pressure experiments. He is of opinion tliat the results obtained by Regnault's method are too low, and that at the higher temperatures reached by Holborn and Henning the error may possibly amount to as much as lo per cent. In all these experiments there is a considerable How of heat from the heater to the calorimeter. This, of course, has to be deducted from the heat registered in the BRITISH ASSOCIATION COMMITTEE REPORT 347 calorimeter in order to find that which has been given up by the hot gas. All the experimenters by this method have determined the amount of this correc- tion by observations when the gas was not flowing, or have compensated it under the same conditions by radiation from the calorimeter, making, in either case, the assumption that the amount of heat conducted is the same whether the gas be flowing or not. Professor Callendar is of opinion that this heat-flow is, in fact, much less when the gas is flowing. He considers that even the values obtained by Regnault may be as much as 2 or 3 per cent, too low, and he supports this contention by reference to the work of other experimenters (some of which has been alluded to above) and by theoretical considerations. As this type of error is likely to increase greatly with rise of temperature a systematic error of even 2 per cent, in Regnault's results, if established, would give reason to suspect that the experiments at high temperatures may be subject to errors of real importance for the present purpose. If there be systematic error in Mr. Clerk's work it seems most likely that it lies in the estimate of heat-loss. The total heat-loss in the first partial compres- sion and expansion line in the diagram (b c d, fig. 122) is estimated from the fall of temperature and from the net work done (area b c d) in the double operation, and amounts to, roughly, half the work done in expansion. This loss has to be divided between compression and expansion, and Mr. Clerk divides it on the assumption that if the mean temperature in compression and expansion were the same the heat loss would also be the same. The mean temperature in expansion is, in fact, rather less than in compression, and the heat loss calculated in this way is correspondingly smaller, but the difference on this account is not very great, and the result is, roughly speaking, that the loss is equally divided between the two operations. Thus the correction to be added to the work clone in expansion in order to get the total loss of energy of the gas is about 25 per cent, of the work, or 20 per cent, of the energy change. Professor Hopkinson has dealt with this point in a note which he communicated to the Committee, and he is of opinion that, relative to the mean temperature, the heat loss is really much greater in compression than it is in expansion. He supports this view by reference to some experiments which he has made on the compression and expansion of a charge of cold air in a gas engine which was motored round with the gas cut off. The specific heat of air being known, the loss of heat in any part of the compression or expansion stroke can in this case be independently estimated from the diagram. He found that while in the latter half of the compression stroke the heat lost to the walls amounted to a considerable fraction of the work done, some part of this loss was actually restored to the gas during the first half of the succeeding expansion, and this notwithstanding the high temperature of the air, which in expansion, as in compression, was much above that of the walls. An estimate of the thermal capacity could, of course, be obtained from this diagram by the application of Mr. Clerk's method, and it would lead to a result considerably in excess of the truth. Mr. Clerk has himself tried this same experiment of compressing and expanding air, and he also has found that the resulting value of the specific heat of air is too high and that the air takes in heat during expansion. Professor Hopkinson thinks it possible that the heat lost during the partial compression line in Clerk's diagram may be more than twice as great as the loss during expansion. If this were so, the correction for heat loss in expansion would be less than 16 per cent, of the work done instead of 25 per cent., and the true change of energy would be less than that calculated on the assumption of equal heat loss in compression and expansion by 7 per cent, or more. 348 THE GAS, PETROL, AND OIL ENGINE It will be seen that the errors believed to afEect each method o£ experiment are in such a direction as to account for the divergence of the results ; and it is quite probable that when these errors are completely allowed for, the discrepancy will largely disappear. Meanwhile the internal energy of the products of com- bustion in the gas engine at I200° C, if taken as the mean of Clerk's and of Holborn and Henning's results, must be regarded as subject to a possible error of about 5 per cent. Under these circumstances it does not seem necessary to discuss the possi- bility that there may be a real difference between the energy values obtained by the two methods due to the different conditions of experiment. It may be poiated out, however, that Clerk's gas was at the maximum temperature from fifteen to twenty times as dense as Holborn and Henning's. This difference in the condition of the gas is such that a comparison of the results obtained by the two methods, when freed from experimental errors, will be of great interest and importance. (3) Explosion Experiments If a combustible mixture of gases be fired in a closed vessel impervious to heat, and if suflicient time elapse to aUow of the attainment of complete thermal and chemical equilibrium, the internal energy of the products of com- bustion after the explosion will be equal to the chemical energy before explosion. The latter is capable of accurate measurement. The temperature reached after explosion can be inferred from the pressure, assuming the gaseous laws to hold. The pressure can also be measured without difficulty and with considerable accuracy. In the study of explosion pressures we have therefore a very convenient and simple means of getting the internal energy function at high temperatures provided that it is possible to make the necessary corrections for deducing from the pressures observed in a real explosion the pressure reached in an explo- sion under the ideal circumstances postulated above. Moreover, the gaseous laws on which the temperature estimations are based can themselves be checked, and if necessary corrected, by comparison of the pressures reached by mixtures of the same composition but of different densities. Thus explosion experiments are capable of furnishing a complete account of all the thermal properties of gases at the temperatures reached by combustion, subject always to the possibility of making the corrections referred to above. The difiiculty of finding these corrections is, however, very great, and in consequence of the uncertainty which prevails even as to their order of magnitude, the large amount of work which has been done on explosion pressures gives but little definite information as to the specific heats of gases. Nevertheless, it is to the study of explosion pressures that we owe such knowledge as we possess of the energy function at the tempera- tures which prevail in the gas engine, and it is to work on these lines that we must look in large measure for extension of mix knowledge. A full discussion of what has been done already must therefore form an important part of this Report. Let H be the calorific value of the mixture before combustion, let h be the heat lost at some point A on the record (taken on a revolving drum) connecting the pressure and the time (fig. 123). The energy in the gas is then n — h. The gas at this point is, however, certainly not in thermal equilibrium, and is probably neither in chemical equilibrium nor at rest. If therefore the loss of heat were suddenly arrested at A, the pressure would change owing to the more or less gradual attainment of equilibrium in aU three respects. The equilibrium value of the pressure would be reached asymptotically, as shown by the dotted line. BRITISH ASSOCIATION COMMITTEE REPORT 349 When cqtiilihrium has been attained the energy of the gas is all thermal and equal to h — A, and the temperature can be calculated in the ordinary way from the pressure. The problem, therefore, is first to find or estimate the heat loss /; which ha,s occurred at some point on the explosion record, and then to find or estimate by how much the equilibrium value of the pressure, if there were no further heat loss, would differ from that shown on the record. This change of pressure, marked p on the diagram, is due partly to the com- bustion of the gas remaining unburnt at A and partly to the equalisation of temperature by convection. It may also be due to some extent to the damping down of the motion of the gas set up by the explosion. The sooner the point A is taken, the less will be the loss of heat ; but the greater, on the other hand, will be the departure from equilibrium conditions. The principal workers in this field, Mallard and Le Chatelier and Langen, assumed that the latter might be neglected if the point a were taken at the point of inflexion on the falling curve, and they estimated the loss of heat by prolonging Fig. 123 this curve backwards, as shown. They assumed, in fact, that the pressure given by the point b was that which would ultimately have been reached had the explosion taken place in a vessel with walls impervious to heat. It is difficult to justify this procedure on a priori grounds ; the only satisfac- tory justification is to show, by independent evidence, that it leads to correct results. The main object of this section of the Report is to examine such evidence as there is of this kind, to point out the defects in it, and to suggest experimental methods by which they could perhaps be remedied. In so far as the heat loss and the departure from equilibrium are dependent on surface phenomena, a definite estimate of their amount can be obtained by a comparison of explosions of the same mixture in vessels of different sizes. Many years ago Berthelot tried this experiment, firing hydrogen and oxygen, in explosive proportions, in vessels of 300 c.c. and 4000 c.c. respectively. It is stated that the pressure reached was very nearly the same, which would show that such part of the cooling and other corrections as depends on the surface of the vessel is small in the case of this mixture. 350 THE GAS, PETROL, AND OIL ENGINE Materials for a more accurate comparison are to be found in the extensive researches of Mallard and Le Chatelier, and of Langen. The French experimenters worked with a cylindrical vessel 17 cm. x 17 cm., whereas Langen used a sphere 40 cm. diameter. volume second case. The following table shows the results obtained in two instances, in each of which the composition of the mixture was practically identical in the two sets of experiments : — The ratio ^"^ ^'^^ was 2-3 times as great in the first as in the Mixture Observer Ho Cooling correction 2 vols, air | 1 vol. (H., + 0) ] 2 vols, air "| I vol. (CO + 0)1 Mallard and Le Chatelier, Langen Mallard and Le Chatelier, Langen 7-40 7-50 7-50 7-50 14 per cent. 8 „ 7i .. 8 IXo is the pressure reached in the explosion in atmospheres after correcting for cooling in the manner described above, when the initial temperature is 0° C, LANCENS CURVE QUANTITY OF INERT CAS PER. UNIT VOL. OF ( H^+O). Fig. 124. — Comparison of maximum pressures corrected for cooling from Langen and Mallard and Le Chatelier's experiments. H^O + O + Inert Gas The cooling correction, or excess of the pressure at B (fig. 123) over that at a, is shown in the last column. Figs. I24and 125, which are taken from Langen's paper, show a comparison between the curve adopted by Langen, as representing the results of his experiments, and Mallard and Le Chatelier's observations. On the whole, the agreement between the two sets of experiments is very fair. BRITISH ASSOCIATION COMMITTEE REPORT 351 and the deviations are not such as to suggest that any very great error has been made in estimating such part ol the corrections for lieat loss or for unburnt gas as depend on the surface of the vessel. If, for example, Langen were, on the average, 4 per cent, out from this cause, Mallard and Le Chatelier would be 9 per cent, out, and would differ by 5 per cent, from Langen. Differences of that amount do occur, but they do not seem to be systematic. Further experiment of the same kind on vessels with a greater difference of size but of similar geometrical form is, however, desirable. The question remains, how far the corrections really are surface corrections. This appears to the Committee to be the most important question of a general character awaiting solution in connection with gaseous explosions when regarded as a means of investigating the properties of gases at higher temperatures. It *?i ' ^^ @ ■^-"-^ ° QUANTITY OF INERT CAS PER. UNIT VOL. OF lcO-»o ). Fir. 125. — Comparison of maximum pressures corrected for cooling from Langen and Mallard and Le Chatelier's ex- " periments. CO + O + Inert Gas. will be convenient to discuss each of the corrections enumerated above with special regard to this question. Loss of Heat. — That much of the heat loss goes on by direct conduction to the walls, and is, therefore, a surface phenomenon, is obvious. But there is reason to believe that the loss by radiation, which certainly exists in any flame, is practically important. (a) Measurements of the temperature reached in an explosion by means of ix platinum thermometer, under circumstances which render very improbable any loss of heat by conduction from the gas whose temperature is measured, show that that temperature is considerably lower than is to be expected from the heat of combustion of the gases and the specific heat of the products. 352 THE GAS, PETROL, AND OIL ENGINE Professor Callendar pointed out, in the discussion of tliese experiments, tliat there was probably a good deal of radiation, and stated that he had found that an ordinary Bunsen flame might radiate up to 15 per cent, of its heat.' (6) Recent experiments, in which the loss of heat during an explosion was directly measured by finding the rise of temperature of the walls, showed that in a certain coal-gas explosion it amounted to about 12 per cent, of the whole heat at the moment of maximum pressure. Estimated by Mallard and Le Chatelier's extrapolation method, the loss was at most 5 per cent.' The prevailing opinion seems to be that most simple gases cannot be made to radiate by direct heating. If this be so, the radiation must take place in the act of combustion. It seems very probable that when, say, hydrogen and oxygen combine a certain part of the energy of combination passes into the form of internal vibrations of the steam molecule, and that a large proportion, if not all, of this part is ultimately radiated away. If this be the case a definite proportion of the heat produced in combustion is always lost, and a comparison of explosions in vessels of different sizes would not reveal this loss. Thermal Equilibrium. — When an explosive mixture of gases is ignited in a closed vessel the effect of the change of pressure during the progress of the flame from the point or points of ignition is to raise the temperature round about those points much above the mean temperature, and, on the other hand, the temperature attained at those places which are last reached by the flame, and where the gas is compressed before instead of after ignition, is much below the mean. Even in a vessel whose walls are in pervious to heat the difference of temperature between the points first and last inflamed might amount to 700° C. at the moment of maximum pressure.' In a real explosion the cooling effect of the walls causes the temperature to range from perhaps 300° or more above the mean (as shown by the pressure) right down to the wall temperature at points close to the metal. The existence of large temperature differences in the gas close to the walls of an engine cylinder was first experimentally demonstrated by Professor Burstall with the aid of platinum thermometers. If the volumetric heat of the gas were constant the equalisation of these temperature differences by convection and conduction, could it take place without loss of heat, would cause no change of pressure. The volumetric heat is, however, not constant, but may quite possibly be 50 per cent, greater in the hottest than in the coldest part of the mass. The attainment of thermal equili- brium must, in fact, cause a change of pressure, and would contribute to the correction which has been designated p (see fig. 123). The amount of the change might be the subject of rough calculation, taking an assumed distribution of temperature and assuming values for the volumetric heat. Such a calculation in the present state of knowledge would only be of value as showing the possible order of magnitude of the quantity sought, and the assumptions made could therefore be of a character to make the calculation fairly simple. More accurate knowledge both of temperature distribution and of thermal capacity will enable greater accuracy to be attained in the estimation of this correction, which will be of such a kind that a method of successive approximation can be pursued, the revised values of thermal capacity resulting from its application being applied to a more accurate calculation of the correction if necessary. The temperature variation set up by the cooling action of the walls is a surface phenomenon, and as such the correction which it necessitates can probably 1 Froc. J?.S., A, vol. l.xxvii. p. 400. 2 /ij^ ^^ y^i jxxix-. p. 1.(7. 5 Hid. A, vol. Ixxvii. p. 389. BRITISH ASSOCIATION COMMITTEE REPORT 353 be determined and eliminated by experiments with vessels of different sizes. The variation caused by the change of pressure during the period of inflammation is not of this character ; and the necessity for a large correction on this account is quite consistent with the observations of Berthelot, or of Mallard and Le Chatelier and of Langen. In these experiments the maximum pressure reached in the explosion was measured, and at the time of maximum pressure very large differences of temperature are known to exist at a. distance from and quite independent of the walls. Soon after maximum pressure, however, the temperatures at points remote from the walls are equalised to a large extent by convection currents. There then remains only the layer of gas near the walls to be considered in this connec- tion. If, therefore, the measurements be postponed until a long enough time has elapsed to admit of this internal equalisation, the correction becomes of the surface kind, and can be dealt with by the method appropriate to corrections of that type. But in that case the heat lost will be too large a quantity to admit of rough estimation ; it must be directly measured. Chemical Equilibrium. — The view that chemical equilibrium is not attained until some time after the moment of maximum pressure was first put forward by Clerk in 1885, who then expressed the opinion that the greater part of the so-called ' suppression of heat ' in explosions was to be ascribed to this cause. On the other hand. Continental writers have almost completely ignored it. For example, Langen makes practically no reference to this in his paper. It can hardly be doubted, however, that in many explosions, especially of weak mixtures, a considerable amount of the energy is in the chemical form at the moment of maximum pressure. On the other hand, it seems probable to the Committee that the amount of unburnt gas at this moment in such experiments as those of Langen was not such as very greatly to affect the results. This belief is based on the supposition that the incomplete combustion is due to the cooling action of the walls. It seems probable that very shortly after the attainment of maximum pressure, that is, within a time small compared with that required to reach maximum pressure, the transformation of the chemical energy into thermal form is everyivhere complete except in a thin surface layer where this transforma- tion is retarded by the cooling action of the walls. If this view be accepted, the correction of the results for incomplete com- bustion is of the nature of a surface correction, and can be determined by com- paring the pressures reached by the same mixture when exploded in vessels of different sizes. In the discussion of this important matter the Committee have derived great assistance from the experience of Professors Dixon and Bone, who have made a special study of the velocity of chemical action in gases. These gentlemen are of opinion that though such action may be of great complexity, involving in many cases several successive molecular operations, yet, if it is not retarded by the presence of cold foreign bodies, it will generally be completed within a period which, for the purposes of gas-engine theory, may be regarded as negligibly small. In the simple case of the explosion of hydrogen and oxygen they consider that the complete transformation of the mixed gases into steam at any given point is complete within a time measured by the interval between molecular collisions. When the action is more complicated, as in the explosion of carbon monoxide and oxygen in the presence of water, or in the combustion of hydro- carbons, the period will be larger, but will still be measured by thousandths of a second. Some direct evidence that incomplete combustion in an explosion is mainly, VOL. I. A A 354 THE GAS, PETROL, AND OIL ENGINE if not entirely, a surface phenomenon is to be found in Hopkinson's measure- ments of the temperature at points within a large explosion vessel by means of a platinum thermometer. A photographic record of the resistance of a fine platinum wire immersed in the gas showed that when the flame reached it the temperature rose in less than jLth of a second from 20° C, which was the temperature of the unburnt gas, to about 1250° C, which was that of the burnt gas, and that it remained at the latter figure quite steadily except in so far as the increase of pressure in the vessel caused it to rise. In other words, there was no increase of thermal energy except that due to work done upon the gas from outside.' The mixture was one part of coal gas to nine parts of air — a slow burning mixture — and the time taken to reach maximum pressure was about a quarter of a second or at least ten times that required for combination of the gases at any one point. It is true that the vessel was of rather large size — about 6 cubic feet capacity — but, on the other hand, owing to the fact that the platinum wire extended over about I cm., so that the flame took an appreciable time completely to envelop it, it is probable that the period of j'jth of a second, given above, is a superior limit which greatly exceeds the actual time taken to effect the combination at any one point. On the other hand, it cannot be doubted that combustion must be greatly retarded in the neighbourhood of the cold metal walls ; and there is nothing to show that this surface retardation is not sufi&cient to account for all the phenomena of delayed combustion. A simple calculation based upon the rate of flow of heat per square foot into the metal of a gas-engine cylinder (which is roughly known from measurements of the heat carried away by the jacket water) shows that the mean temperature of the exposed surface at points separated by an inch from the cooling water cannot exceed quite a moderate value. Probably about 200° C. is a superior limit for the cylinder liner. Similar calculation of a still rougher kind, but still sufficiently accurate to give the order of magnitude of the quantity sought, shows that the fluctuation above and below the mean in the course of a cycle is very unlikely to exceed 20° C. The latter conclusion has been confirmed by some experiments made by Professor Coker with a preliminary account of which he has favoured the Committee. Measuring the cyclical variation of temperature of the inner surface of a 12-h.p. gas-engine cylinder by methods similar to those adopted by Professors Callendar and Nicholson in their well-known work on the steam engine, he found that the maximum was only 7° F. in excess of the mean. The direct measurements by Professor Hopkinson of the temperature of the walls of an explosion vessel lined with copper strip also lead to the conclusion that it is quite moderate. This cold metal must obviously profoundly affect the combustion in its neighbourhood. In a layer of gas of appreciable thickness the combustion will be of a smouldering character, depending upon the velocity with which the unburnt gas in contact with the walls can diffuse into the hotter rMjions at a distance from them, and so be brought to the ignition temperature. This layer being cold and highly com- pressed might account for a considerable fraction of the heat, though its actual thickness may be only a few tenths of a millimetre. It would appear probable that the continued burning which undoubtedly goes on after the time of maximum pressure in many explosions, and probably also occurs during the first portion at least of the expansion stroke of a gas engine, is mainly of this character.'' 1 Proc, R.S., A, vol. Ixxvii. p. 387. * Professor Bone is doubtful whether 'smouldering combustion ' plays so considerable,! part in gaseous explosions as is here suggested, BRITISH ASSOCIATION COMMITTEE REPORT 355 Motion of the Gas. — In many explosions intense vibratory motions of the gas are set up. The effect of these sometimes appears with a quick-period indicator as a rapid variation of pressure. It is a question of some importance how these motions affect the mean pressure shown by a gauge. The damping down of the motion which occurs in consequence of viscosity of course only means that the motion becomes distributed among the molecules in a random way, instead of following a definite arrangement. The total kinetic energy remains the same. But it is not certain that the mean effect on a pressure gauge of the molecular impacts wiU be the same. This is a question which might be considered by someone to whom the methods of the kinetic theory of gases are familiar. It is of course not a surface phenomenon. Results of Observations. — The temperatures reached in these explosion experiments range from about 1300° up to 3000° C. Temperatures of below 1 500° are, however, obtained by the use of weak mixtures, involving slow burning and large cooling corrections, and but little reliance can be placed on the results. Langen made very few observations on mixtures giving a lower temperature than 1500°, and takes that as the lower limit of the range of temperature to which his observations apply. The extreme upper limit of the constant pressure experiments is 1400°. The temperature of 3000° C. is about that reached in the explosion of hydrogen and oxygen in their combining proportions. This is much above the mean temperature ordinarily reached in the gas engine, the upper limit of which may be put at about 2000° C, though it is probable that 2500° or more is occasionally reached locally. Langen, however, places the upper limit of the application of his formulae at 1700° C, on the ground that there is dissocia- tion of the CO2 at higher temperatures than that. There does not seem to be much reason for this limitation, for the effects of dissociation (provided that equilibrium is attained) are indistinguishable from those of increasing specific heat, and should be included in the change of energy. Dissociation may give rise to errors in the temperature, measurement, but there is reason to suppose that the dissociation which occurs in the CO^in steam at a temperature of 2000° C. is too small to cause any material change of volume, though it may mean consider- able absorption of heat. The formulae given by Langen as representing the results of his observations are as follows : Air C = 4-8 -I- 0-0006 t CO2 0=67 + 0-00260 t Hpc = 5-9 -I- 0-00215 ' where S is the mean thermal capacity over the range o to t° C. The explosion pressures predicted by the use of these formulae agree well with the observed pressures except in the case of mixtures of CO and air, where they are a good deal too high. In the other cases the maximum deviation is about 4 per cent. Mallard and Le Chatelier represent their results by formulae which differ greatly from the above in the case of CO^ and H2O, though the formula for air is the same. This discrepancy must be due in some way to the method of reduc- tion adopted, for, as already pointed out, the explosion pressures reached with mixtures of the same composition are very nearly the same. Taking Langen's values, the following table exhibits the energy of the various simple gases, and of the mixture on which Clerk experimented, at i6oo° and 2000° respectively. The energy of the same gases at 800° and at 1200° based on Holborn and Henning's and on Clerk's results is also given for comparison. The A A 2 356 THE GAS, PETROL, AND OIL ENGINE results are given in calories per gramme molecule. To reduce to foot-pounds per cubic foot multiply by the factor 3 -96. 800° 1200° 1600° 2000° Clerk Holborn and Henning Clerk Holborn and Henntng Langen Langen Air ... . — 3570 — 5840 8700 1 1 500 CO, .... — 6460 — 10880 17000 23300 H.O .... — 4670 — 7930 14400 19900 Gas-engine Mixture . 4250 3840 6900 6340 9800 13200 Ideal Gas * 3430 5400 7350 9300 • C = 4-9- The results for the gas-engine mixture are plotted on fig. 1 26, on which points obtained by Mallard and Le Chatelier's formulae are also shown. The energy of gas-engine mixture at 1400° according to Clerk, Holborn and Henning, and Langen respectively would be as follows : — Clerk .... Holborn and Henning Langen. 8300 7700 8300 It will be seen that the agreement between Clerk and Langen is close, both being about 8 per cent, higher than Holborn and Henning. But it is to be observed that this temperature is just outside the range of all three sets of experiments. The Committee are of opinion that values of the energy obtained from explo- sion records are not subject to any very great errors on account of heat loss by conduction to the walls of the vessel, nor on account of incomplete combustion, but that they are affected by errors of quite unknovTn amount due, first, to heat radiated, and secondly, to the want of thermal equilibrium at the time when the pressure is measured. For the purpose of testing the first of these conclusions, it is very desirable that further experiments should be made on explosions in vessels of greatly different size but of similar form. The opinion entertained by the Committee that incomplete combustion is a surface phenomenon, on which this conclusion as to the validity of the method is based, also requires further confirmation. As regards the second conclusion, further experiment on the actual amount of heat radiated by burning gas is urgently required, and also experiments to confirm or negative th^ffect of the nature of the wall surface upon the pressure reached in an explosion. The effect of want of thermal equilibrium can be determined up to a point by calculation ; but before such calculation can be usefully made, it is desirable that further information should be obtained as to the temperature distribution after an explosion, especially in the neighbour- hood of the walls. It should not be difficult to get an idea of this sufficiently accurate for the purpose by means of platinum thermometers. The most hopeful way, however, of making use of explosions to give definite information as to the properties of gases would appear to be directly to measure the heat lost in the explosion, as if this be done it is possible to defer the pressure BRITISH ASSOCIATION COMMITTEE REPORT 357 measurement until such time as equilibrium conditions, except those that depend on the surface of the vessel, have been attained. The Measurement of Temperature In all the experiments for the determination of the energy function which 8 \ \ \ \ \ \ \ \ \ \ 4 II i . \ 4 '^ \\ \ \ \\ \ \ \\ 1. \ ^ 3 < ■3 ' 1 1 f ° .a ^aTfi5TTOH3Wwvu5a3dSiiaoTVD^oa3N3 O .J OJ p; 3 S g « O ^■^ o have been described above, the measurement of the temperature is ultimately based upon the pressure or volume changes of the gas. In the constant-pressure experiments of Uolborn and Henning the temperature of the gas before entering 358 THE GAS, PETROL, AND OIL ENGINE the calorimeter was measured by means of a thermo-couple which had been compared with a constant- volume nitrogen thermometer up to 1600° C. In the explosion experiments the mean temperature of a gas is inferred from its pressure. Similarly, in the analysis of the gas-engine diagrams, the gas is itself the thermometer. The mean temperature at any point is taken as proportional to the product pv, and the actual temperature at one point in the cycle (a know- ledge of which is necessary for getting absolute values) is obtained either by estimating the quantity of gas present in the cylinder or by direct measurement with the platinum thermometer, as was recently done by Callendar and Dalby. The temperature scale so obtained is probably sufficiently definite, at any rate for the purpose of gas-engine theory, since the mixture to which it is applied does not vary very greatly in composition, and always consists mainly of nitrogen. It is not so certain, however, that this scale agrees with the absolute thermodynamic scale ; and the question of the possible amount of the deviations at temperatures of 1500° and over is of great importance in connection with the present inquiry. So far as the Committee are aware, the experiments of Joule and Thomson still remain the only comparison between the various gas scales and the thermo- dynamic scale : this comparison only extended to about 200° C, and is, of course, of no application to the problem now under discussion except in so far as it gives an idea of the differences to be expected at higher temperatures. What it really shows is that thermometers constructed of the more permanent gases are all so closely accordant with the thermodynamic thermometer as to lead to the belief (as a matter of induction, and quite independently of the kinetic or any other theory) that there is really some definite cause tending to make a gas, as such and apart from its composition, obey the law ?- = constant. It would appear that the small deviations from this law, sometimes one way and sometimes the other, which are observed must be due to disturbing causes depending on the nature of the gas, whose influence may be in either direction and is of very various amount, but is at low temperatures small compared with the tendency to obey the perfect gas law. This view being accepted, there is a strong presump- tion that, if a number of thermometers constructed of different gases be com- pared at high temperatures and be found to agree fairly well, then they all agree with the thermodynamic scale at least as well as they agree with one another. It is upon the agreement between different gas thermometers that our belief in the measurement of temperature is really founded, and, so far as it goes, the foundation seems to be sound. The nitrogen thermometer has been used with an iridium bulb up to 1600° C.,' but no other gas has been taken above 1100° C. At the latter temperature the differences between thermometers constructed of hydrogen, nitrogen, and air are quite negligible for the present jjfcrpose. Rather more deviation has been observed in CO„ ; bu t having regard to the small percentage of this gas which is ordinarily present in the gas-engine mixture, it is not likely that temperatures up to 1 100° C, calculated in the usual way from the indicator diagram, will differ much from the true temperatures on the thermodynamic scale. 1100° C. is, however, not much above the lower limit of the gas-engine range ; as to what goes on in the upper part of that range we have little or no evidence. When considerable deviation from the gas laws at high temperatures is observed in the case of any gas, it is usually ascribed to dissociation. For ' Holborn and Valentiner, Ann. d. Phys. xxii. (1907), p. y. BRITISH ASSOCIATION COMMITTEE REPORT 359 example, if comparison be made of two constant-pressure thermometers filled respectively with hydrogen and with iodine vapour they will be found to agree up to about 1000° abs. ; the iodine thermometer will then begin to read higher than the hydrogen thermometer until, when the latter reads about 1800° abs., the former will read about double that amount. When the gases are hotter still, the temperature shown on the iodine thermometer will continue to be double that shown on the other. In this case the departure between the two thermometers is accompanied by a change in the absorption spectrum of the iodine vapour ; and the whole phenomenon is expressed by saying that the iodine molecule has been split up or dissociated. In the case of a compound gas this dissociation sometimes takes the form of an actual separation of the con- stituents, which can be detected by diffusion. A great deal of experimental work has been done with the object of ascertaining to what extent the gases COj and steam split up at high temperatures. These gases are constituents in most gas-engine mixtures, and if they dissociate to any considerable extent there will be a corresponding effect upon the pvS relations of the mixture of which they form a part. So far, however, there is, in the opinion of the Committee, no conclusive evidence that either steam or COj is dissociated to an extent which is material for the present purpose. Slight traces of dissociation have undoubtedly been found in both cases, but the method of experiment is such as to leave it doubtful how far these have been conditioned by the nature of the walls through which the dissociated gas is diffused. It must be observed, moreover, that CO^ and steam usually form only a small part of the mixture in the gas engine, and that therefore a considerable amount of dissociation of these gases would be necessary to produce much effect upon the pressure of the whole. Again, such dissociation, if it occurs, must have an effect upon the energy of the gas out of all proportion to the effect which it has upon its temperature. Take, for example, the case of a mixture formed by the explosion of CO and air and containing 10 per cent, of COj, the remainder being nitrogen. If, by heating, one tenth part of the CO^ be split up into CO and oxygen, the resulting change of pressure of the whole mixture will be only one two-hundredth part ; but this amount of dissociation could only be effected by the absorption of an amount of heat of the order of 10 per cent, of the total heat of combustion of the gas. In other words, the mean specific heat of the mixture, as determined by the explosion, would be roughly 10 per cent, lower than if there had been no dissociation. Any considerable departure from the gas laws in such a mixture, if it be ascribed to dissociation at all, must therefore be put down to dissociation of the nitrogen, which might conceivably occur at 2000° C, just as iodine vapour is dissociated at a much lower temperature. It does not seem likely, however, that if nitrogen dissociates its splitting up would be accompanied by any visible change in its physical properties, such as is observed in the case of iodine. The phenomenon in this case would be rendered evident only by the departure from the gas law, and possibly by absorption of heat. It would appear, therefore, that our knowledge of thermometry at these temperatures is more likely to be advanced by direct experiments on the rela- tion between the pressure or volume and the temperature than by looking for other evidences of dissociation. The difficulty in carrying the comparison of different gas thermometers to very high temperatures has hitherto lain in the absence of any material sufficiently refractory to withstand such temperatures and at the same time sufficiently impervious to the gas. Dr. Harker, to whom the Committee are greatly indebted for much information upon this subject. 35o THE GAS, PEtROL, AND OIL ENGINE believes, however, that he is now in possession of a material which will satisfy both of these conditions up to a temperature of 1800° C, and he has suggested that an attempt should be made to compare thermometers constructed with nitrogen, with COj, and with argon up to that temperature. If the nitrogen and argon thermometers are found to agree, then, by reason of the great difference in the constitution of these gases, it is almost certain, as explained above, that each agrees with the thermodynamic scale. If, on the other hand, they do not agree, then the presumption is in favour of the argon thermometer, because this gas is supposed to be monatomic and to be incapable of dissociation. The Com- mittee venture to express a hope that a research on these lines will be commenced and carried to a conclusion. They believe that the results obtained will be of very great importance in the investigation of explosions and in the theory of the gas engine, and it seems to them an inquiry eminently fitted for the National Physical Laboratory. The comparison of gas thermometers is, however, not the only way in which the problem of thermometry at high temperatures may be attacked. Another method, and one that is more satisfactory in some ways because it is more fundamental, is to investigate the dependence of the energy upon the density of the gas. As pointed out at the commencement of this Report, any inter- dependence between energy and density at a given temperature must be accom- panied by a corresponding deviation from the perfect gas law, and investigation of change of energy with density must be the ultimate basis of gas thermometry. The Joule-Thomson experiment was, of course, of this character. Since then Joly has determined the change of specific heat of COj at pressures ranging up to the critical pressure. But these determinations refer only to temperatures of the order of 100° C. As was pointed out at the commencement of the section of this Report dealing with explosions, the corresponding measurement at very high temperatures can be very easily made when once the various corrections necessary to determine internal energy by explosion experiments have been satisfactorily performed. It is only necessary to compare the pressures reached in explosions of mixtures identical in composition but of different density. Should the pressures after explosion, when corrected, be proportional to the pressures before explosion, then the energy is independent of the density, and we have proof that the gas law holds up to the temperature reached by the combustion. On the other hand, a departure from the proportionality would imply a corresponding departure from the gas laws, the amount of which could be calculated. Mallard, Le Chatelier, and Langen have made very careful comparisons of this kind, and they have found that the actual maximum pressures reached in the explosions are in many cases very approximately pro- portional to the pressures before explosion. Petavel has found that this pro- portionality is not much altered even when the density of the gas is increased seventy times. This may be regarded to some extent as evidence that there is no very great difference between the gas scale and the thermodynamic scale at the temperatures of 1700°, or more, which were reached in these experiments. But it must be observed that this inference is subject to the same limitations as the determinations of internal energy based upon these experiments. It cannot be regarded as having a secure foundation until the various doubtful questions in regard to heat loss and delayed combustion, which have been raised above in this connection, have been satisfactorily determined. The Committee think that they can usefully continue their work in the direc- tion of suggesting, and to some extent organising, research on the lines which CALLENDAR'S APPENDIX TO B.A. COMMITTEE REPORT 361 have been foreshadowed in this Report. Research of this kind is expensive, and the Committee are of opinion that their work would be greatly facilitated if they had some funds at their command. They therefore recommend that they be reappointed, and ask for a grant of lool. APPENDIX TO REPORT By Professor H. L. Callendar, M.A., LL.D., F.R.S., on The Deviation of Actual Gases from the Ideal State, and on Experimental Errors in the Determination of their Specific Heats I. The equation pv = Rfl, where 6 is absolute temperature, is the charac- teristic equation of a fluid which (i) obeys Boyle's Law at all temperatures, and (2) has the difierence of its specific heats constant and equal to R. The specific heat at constant volume or pressure may vary in any manner with temperature, provided that the difference of the two is constant ; but both specific heats must be independent of the pressure or density. For the majority of common gases or vapours (excluding those which poly- merise, like sulphur) the deviations from Boyle's Law, as measured by the defect (R9/P — v) of the actual volume from the ideal volume, at moderate pressures (say up to ten atmospheres) are to a first approximation a function of the tempera- ture only, and diminish rapidly with rise of temperature. On this assumption, tables of correction for the gas thermometer have been independently calculated by Callendar ' and D. Berthelot ^ for various gases when employed in the usual manner. The corrections are very small, and agree very closely, though calcu- lated on slightly different assumptions. The differences are much too small to be taken into account in gas-engine experiments. In dealing with a mixture of gases and vapours at high temperatures, the method of procedure is necessarily somewhat different from the case of the gas thermometer, and the tabulated corrections do not apply. The effective temperature of the mixture is calculated from the value of the product pvj'R, assuming that the composition of the mixture is known, and that the constant R has the same value per gramme molecule for each of the constituents as for an ideal gas. The errors involved in this niethod will be small, and will diminish with rise of temperature, provided that the constituents do not dissociate or polymerise. The experimental evidence at present available with regard to dissociation would indicate that the error of this assumption is certainly less than I per cent, for a gas engine mixture at 2000° C, if the composition of the products of combustion is known. Effective Temperature and Effective Specif c Heat 2. Since the temperature of a mass of gas, when exploded in a closed vessel or in the cylinder of a gas engine, is far from uniform, and since the actual distribu- tion of temperature is necessarily somewhat uncertain, it is evident that the varia- tion of the specific heats of the constituents with temperature cannot be certainly ^ Phil. Mag., Januaiy 1903. '* Trav. ei Mdiii. Bur. Int., Paris, 1903. 362 THE GAS, PETROL, AND OIL ENGINE deduced from a knowledge of the heats of combustion and the effective tempera- ture, even apart from difficulties inseparably connected with the determination of the cooling corrections. It is possible, however, by explosion experiments to deduce values of the apparent or effective specific heats which, in so far as they approximate to the conditions actually existing in the gas engine, may be of greater practical utility than the true specific heats would be if they could be independently determined. The method of Dugald Clerk, in which the specific heat is directly determined from the work done on the charge after ignition, appears to be particularly appropriate for this purpose. It is well known that the values of the specific heats deduced from explosion experiments are generally higher than those deduced by more direct methods, and it has been customary to explain the discrepancy largely by possible errors inherent in the explosion method. Such errors undoubtedly exist, and require careful investigation, but in arriving at a decision it is most important to subject other experimental methods to an equally close scrutiny. Experimental Errors in the Determination of the Speeijic Heats of Gases by the Constant-pressure Method 3. Apart from errors in the measurement of the temperature of the gas and of the calorimeter, which are not likely to be serious at low temperatures, there is an important source of error in this method, as applied by Regnault and sub- sequent observers, which has generally been overlooked. In Regnault's experi- ments, the rate of gain of heat from the heating vessel by the calorimeter was observed before and after the experiment proper, while the gas was not passing through the connecting tube, and was assumed to be the same whether the gas was passing or not. The correction amounted, when the heater was at 200° C, to between 4 per cent, and 5 per cent, of the heat supplied by the gas. The effect of the gas current would certainly be to change the temperature gradient in the connecting tube in such tx manner as to diminish the heat con- ducted from the heater during the passage of the gas. The error from this cause cannot be exactly determined, but would probably amount to between 2 per cent, and 3 per cent, in Regnault's experiments at 200° C, and would have the effect of making the values as determined by Regnault too low. The true variation of the specific heat of water was unknown in Regnault's time, and he was also unable to correct his thermometers accurately to the absolute scale. These considerations introduce minor uncertainties which might amount to as much as I per cent, on the result. The specific heat of air considered as a, mixture of perfect diatomic gases, taking the calorie at 20° C. as equivalent to 4-180 joules, should be 0-2405. Since air is not a perfect gas the actual value must be somewhat greater than this. Regnault's value, 0-2375, is evidently too low. E. Wiedemann obtained the valuac-2389 by a method similar to Regnault's. This value is probably affected by a similar error. J. Joly measured the mean specific heat of air at constant volume, and at densities 7 to 22 times normal, by the method of the steam calorimeter, between 10° and 100° C. This method has the advantage of avoiding the majority of the sources of error above mentioned. Joly's value for air at constant volume, when reduced to the calorie at 20° C. and to normal density, would be 0-1732, wliich corresponds to a value 0-2419 for the specific heat at constant pressure at a temperature of 55° C This is a far more probable value than Regnault's, but 1 Callendar, Phil. Mag., January 1903, p. 76. CALLENDAR'S APPENDIX TO B.A. COMMITTEE REPORT 363 it must be observed that the extrapolation of the experiments to atmosplieric pressure involves some uncertainty. The specific heats of air and CO2 at atmospheric pressure and at temperatures of 20° and 100° C. have recently been determined by Swann at the Royal College of Science by the continuous electric method previously employed by Callendar ' in the case of steam. In this method the actual specific heat at any point is determined by observing the rise of temperature produced in a steady current of gas at the required temperature by supplying a measured quantity of electric energy. This method is better adapted than Regnault's for determining the variation of the specific heat, because it gives the actual specific heat over a small range (about 5°) at the required point in place of the mean specific heat over a large range. It has also the advantage that systematic errors may be more completely eliminated. The values obtained by Swann for the specific heat of air at atmospheric pressure in terms of the calorie at 20° C. equivalent to 4'i8o joules were S = 0-2415 at 20° C, and S = 0-2425 at 100° C. His value at 55° C. is in very good agreement with that deduced above from Joly's experiments by the constant-volume method. Adopting a linear formula, we have for the specific heat at any temperature, t between 0° and 100° C. S, = 0-2413 (i + 0-00005O (Swann). Holborn and Austin '' and Holborn and Henning '■" extended Regnault's method for the determination of the mean specific heat to temperatures up to 840° C. In working at these high temperatures the difficulties of the method are greatly increased. They found it necessary to employ electric heating and to connect the heater to the calorimeter by a porcelain tube in order to diminish conduction. The temperature of the hot gas was observed with a thermo-couple near the entrance to the calorimeter. The time of flow was about three minutes in each experiment, and the corrections were estimated by observing the rate of change of temperature of the calorimeter before and after each observation. There appeared to be some doubt whether the couple would give the true mean temperature of the gas flow, especially as the time of flow was so short. For this and other reasons the authors do not lay great stress on the accuracy of the absolute values of the specific heats obtained, but consider that the ratios or relative values, and the rates of increase with temperature, are more likely to be correct than the absolute values, because the various sources of error which they discuss are more likely to be eliminated in the relative values. The value found for the mean specific heat of air over the range 1 1 5° to 270° C. by Holborn and Henning was 0-2315, which is about 5 per cent, smaller than the probable value over this range. For the rate of increase of the mean specific heat they gave the formula : So,i = S„(i ^ o-oooo4() (Holborn and Austin), but considered that the rate of increase shown by their experiments was within the limits of probable accuracy of their work, and that it could not be regarded as certainly established that there was any increase over the range of their experiments. * P/vc. I^.S., 1900. ^ Siiz. Akad. IViss., Berlin, 1905, p. 175, 5 Wied. Ann. 18, 1905, p, 739. 364 THE GAS, PETROL, AND OIL ENGINE Later experiments by Holborn and Henning ' with a platinum heating tube, extending to 1400° C, were made by a similar method, except that the gain of heat by the calorimeter from the heating tube was partly compensated by surrounding the calorimeter at 115° C. with a, jacket maintained at a. much lower temperature. This compensation was found necessary at high temperatures in order to prevent an excessively rapid rise of temperature of the calorimeter ; but although it reduces the apparent magnitude of the correction for gain of heat by the calorimeter, it does not diminish the actual amount of heat trans- ferred and does not reduce the uncertainty of the correction. The magnitude of the effect at high temperatures may be judged from the fact that it was found necessary in the experiments at 1400° C. to maintain the jacket at as low a temperature as 40° C. by passing a stream of cooling water through it in order to prevent the calorimeter rising above 115° C. when no gas was passing. Under such conditions the calorimetric corrections become so uncertain that the probability of systematic errors must increase considerably with rise of tem- perature. If the method gives a probable error of 5 per cent, in defect over the range 1 1 5 ° to 270° C. it does not seem at all impossible that the error may amount to 10 per cent, over the range 115° to 1400° C. The rate of increase of the mean specific heat of nitrogen at atmospheric pressure between 840° and 1340° C, shown by the later experiments, was about double that found in the earlier series. Both series of experiments could be represented within the limits of probable error by the linear formula S„,( = O'2350(i -I- 0'000o8<) (Holborn and Henning). It appears probable, however, that the value of the specific heat at 0° C. given by the formula is too low, and that the rate of increase is not uniform, but increases with rise of temperature to some extent in the case of nitrogen. Specific Heat of CO2 4. Similar remarks apply to the determination of the specific heat of CO2 by the same methods, but the case of CO.^ is of special interest on account of the rapid variation observed at ordinary temperatures. The following table gives the specific heats of CO2 according to different observers at 0° and 100° C. : Temperatuie RegnauU Wiedemann Swann Holborn 0° 100° o'i870 0-2145 0-1952 0-2169 0-1973 0-2213 0-2028 O-2161 Increase . 0'0275 0-0217 0-0240 0-0133 The value of the mean specific heat at constant pressure from 10° to 100° C. deduced from Joly's experiments at constant volume is 0-2120, which is nearly 5 per cent, higher than Regnault's value at this temperature, but agrees as closely as can be expected with that found by Swann. The variation of .the specific heat with density observed by Joly agrees very closely with that calculated by Callendar- from the experiments of Joule and Thomson on the cooling effect in expansion through a porous plug. ' Wied. Ann. 23, 1907, p. 809. ^ Phil. Mag., January 1903, p. 78. CALLENDAR'S APPENDIX TO B.A. COMMITTEE REPORT 365 The rate of increase of the specific heat between 20° and 100° C. observed by Swann is nearly a mean between the rates given by Regnault and Wiedemann, but is much larger than that found by Holborn and Henning, or deduced by Langen from explosion experiments. It is probable that the variation is not linear, but that the rate of increase diminishes with rise of temperature, as indicated by Mallard and Le Chatelier's formula, which would make the specific heat a maximum at 1700° C. The latter formula differs from Holborn and Austin's by more than 20 per cent, at 800° C. The explanation appears to be partly that Regnault's value for the rate of increase at 100° C, adopted by Mallard and Le Chatelier, is too high, but chiefly that Holborn and Austin's values, as already explained in the case of air, are systematically too low, and that the error increases with rise of temperature. Specific Heat of Steam 5. Regnault's value o%y$ for the specific heat of steam at atmospheric pressure over the range 125° to 225° C. was obtained by taking the difference between the total heats of steam, superheated to these temperatures, as observed by condensing the steam in a calorimeter. Since the difference, corresponding to 100° superheat, is only yjth of the total heat measured in either case, it is evident that the method might give rise to large errors. For this reason many writers have preferred to deduce the specific heat of steam theoretically in various ways from Regnault's value of the rate of change of the total heat of saturated steam — namely, 305 cal. per 1° C. — which, however, really involves the same source of error in an aggravated form. Thus Zeuner gives S = 0'568; Perry,' S = 0'3o6 at 0° C. to 0-464 at 210° C. ; Grindley = 0-387 at 100° C. to o'66s at 160° C. A direct measurement of the specific heat of steam by the continuous electric method gave S = 0-497 ''■t 1°^° C.* Subsidiary experiments, in conjunction with Professor Nicolson,' by the throttling calorimeter method enabled the varia- tion of the specific heat with pressure to be calculated. These gave the formula S; = 0-478 + 0-0242 p (373/8)< 3 (Callendar) where p is the pressure in atmospheres. The approximate constancy of the limiting value 0-478 of the specific heat at zero pressure over the range 0° to 200° C. was verified by calculating the corresponding values of the saturation pressure, which were found to agree accurately with Regnault's observations over the whole range. The theory was also verified by a measurement of the ratio of the specific heats of steam by Makower,* which gave values 1-303 to I -307, agreeing closely with that deduced by Callendar. The experiments of Lorenz," and of Knoblauch and Jacob and I.inde ' afforded a remarkable verification of the theory of the variation of the specific heat with pressure. They found the specific heat at t atmo. to be practically constant over the range 100° to 300° C, but their value, namely, 0-463, is decidedly lower than Regnault's. Holborn and Henning," in their experiments on the specific heat of steam at atmospheric pressure, improved Regnault's method by employing an oil calorimeter at 1 10° C. so as to avoid condensing the steam in the calorimeter. I Sleam Engine, 1899, p. 582. - Phil. Trans. 1898. ' Callendar, Proc. R.S. 1900. '• McGill College, 1897. ■■ Phil. Mag., February 1903. " Forsch. Ver. Deut. Ing. 21, 1905, p. 93. '' Loc. cit. pp. I and 35, 1906, p. 109. ' Ann. Phvs, xviii. 1905, p. 739. 366 THE GAS, PETROL, AND OIL ENGINE They determined the ratio of the specific heat of steam to that of air by passing currents of air and steam in succession through the apparatus under similar conditions, and obtained the following values of the ratio for different intervals of temperature : Temperature Interval . iio°-270° iio°-440° iio°-620° iio°-820° Ratio/Steam: Air . i"940 i'958 i'946 i'998 In their subsequent series with a platinum heating-tube at higher tempera- tures they obtained the following ratios : Temperature Interval . ii5°-826° ii5°-n8o° II5°-I324° Ratio/Steam : Air . I'goo i'973 2 '003 The second series appears to make the ratio about 5 per cent, lower at I io°-820° than the first, which suggests the possibility of constant errors depend- ing on the type of apparatus employed or on the velocity of the gas-current. The experiments of Callendar and Swann would make the ratio 2-05 at 100° C, which is higher than any of the values obtained by Holborn and Henning at 1400° C. Holborn and Henning point out that their results at 1400° C. cannot be reconciled in the case of steam and CO, with any of the results of explosion methods. They are 6 per cent, to 13 per cent, lower than Langen's, which are among the lowest. But having regard to the fact that the constant-pressure method which they employed appears to give results so much lower than Joly's or Callendar's methods at ordinary temperatures, and that the experimental difficulties increase so greatly at higher temperatures, it does not seem at all improbable that a considerable part of the discrepancy is to be attributed to systematic errors of the constant-pressure method. On the Cause of the Variation of Specific Heat 6. It appears from theory that the energy of translation and rotation of the molecules of an ideal gas should vary in direct proportion to the product pv. The internal energy of vibration of the molecules, however, which is related to the absorption or emission of radiation must vary by the Stokes-Kirchhoff law in relation to the full radiation of a black body at the same temperature. According to Planck's formula, which has been verified over a very wide range, the energy of full radiation corresponding to wave-length L in full radiation, varies with the temperature according to the expression E = CL-s(e'/"- i)-i L the value of the constant c is 14,700 if L is measured in microns, 11, or million ths of a meter. The energy of vibration of a molecule which is in equilibrium with full radiation at any temperature wJU depend on the extent to which its free periods of vibration respond, as mdicated qualitatively by its absorption spectrum. Those periods which respond very strongly may produce an appre- ciable effect on the specific heat. It happens, for instance, that COj has a very marked absorption band at 1 5M, nearly, which can be detected even when the gas is present in small quanti- ties in the atmosphere. So far as this particular mode of vibration is concerned, the specific heat would increase most rapidly at ordinary temperatures, which is actually observed to be the case with CO,. According to Planck's formula, the effect of any mode pf vibration would be a maximum when 9 is infinite, and CALLENDAR'S APPENDIX TO B.A. COMMITTEE REPORT 367 would then contribute the term CjcU to the mean specific heat ; but for L ^ 15M the effect would have already reached within about 10 per cent, of the possible maximum at 2000°. According to Wien's original formula E = CL-Se-c/L» L which holds very well for short wave-lengths and low temperatures, but appears to fail when L9 is large, the energy E would reach a finite limit CL-s when fl is infinite, and the specific heat for L = 1 5/i would reach a maximum when fl = 500°- This does not appear to agree so well with the changes of specific heat actually observed. In the case of steam it appears that there are no equally well-marked absorption bands corresponding to strong natural periods of vibration, in the range of the heat spectrum available for investigation. The very high dielectric constant of water for short electric waves has been taken to indicate that there is a period of marked resonance very low down in the spectrum in the unexplored field between the shortest electric waves and the longest heat waves hitherto obtainable. This might account for the relatively high value of the specific heat of water and steam at ordinary temperatures. It must be remembered, however, that the absorption spectrum is very complicated, and difficult to investigate beyond the limits of photography. Moreover, it is very difficult to deduce, except in a qualitative manner, the relative intensities of the energy correspond- ing to each absorption band. An absorption band may appear strongly marked in a thick layer of absorbent, which really corresponds to a very small amount of energy. For this reason no quantitative estimation of the effects of vibration of the molecules on the specific heat is possible at the present stage of know- ledge, but it is important to bear the possibility of such effects in mind as a guide for future investigation. 368 THE GAS, PETROL, AND OIL ENGINE APPENDIX V Adiabatic and Isothermal Compression of Dry Air {Professor R. H. Thurston, Journal of Franklin Institute, 1884) One hundred volumes of dry air at the atmospheric mean temperature of I5°'5 C. and 147 lb. per square inch undergo change of volume without loss or gain of heat. The temperatures and volumes corresponding to various pressures are given. Also the volumes at the various pressures if the temperature remained constant at I5°'5 C. Absolute pressure in lbs. per sq. in. 147 15-0 20-0 25-0 30-0 35-0 40-0 45-0 50-0 55-0 60 -o 65-0 70-0 75-0 80 -o 85-0 90-0 95 'O [OO-O 125-0 150-0 175-0 200-0 225-0 250-0 300-0 400-0 500-0 600-0 700-0 800-0 900-0 1000 2000 Temperature of Volume at Volume if compression in temperature and temperature constant Centigrade degrees pressures preceding at 15° -5 C. 15-5 100-0 17-26 98-58 98-00 42-60 80-36 73-50 64-76 68-59 58-80 82-10 60-27 49-00 98-38 54-01 42-00 113-86 49-13 36-75 126-54 45-18 32-67 138-96 41-93 29-40 150-53 39-19 26-73 161-38 36-84 24-50 171-61 34-80 22-62 181-29 33-02 2I-00 190-49 31-44 19-60 199-26 30-03 18-38 207-66 28-77 17-29 214-71 27-62 16-33 223-45 26-58 15-47 230-91 25-63 14-70 264-66 . 21-88 11-76 293-91 19-22 9-80 319-87 17-23 8-40 343-31 15-67 7-35 36471 14-41 6-53 411-57 13-38 5-88 420-34 11-75 4-90 480-76 9-58 3-90 531-21 8-17 2-94 574-93 , 7-18 2-45 603-74 • 6-44 2-IO 648-80 5-86 1-84 680-86 5-39 1-63 710-49 5-00 1-47 929-67 3-06 0-74 INDEX TO THERMODYNAMICS OF THE GAS, PETROL, AND OIL ENGINE ABE Abel, Sir Frederic, on gun-cotton explosions, ii6 Acetylene and air mixtures, 160-166 proportions of, 162-164 Actual and ideal engines of the third type, 327-331 Adiabatic compression, 234 — line, 65, 66 — for gas-engine working fluid, 272 calculation of, 335 — and isothermal expansion dia- gram, 247 compression, table of, 368 Air engines, Ericsson's, 52 Joule on, 52 Rankine on, 52 ■ Stirling's, 52 Wenham's, 52 Air-engine diagram corresponding to Lenoir engine, 236 Air and acetylene mixtures, 160-166 gas mixtures, 129, 130, 135, 136-140, 170, 171, 179-182, 186, 188, 196 — and petrol mixtures, 152 — standard, 102, 244-250, 252, 266 — supply, measurement of, 259-261 Analysis of coal gas : Boston, 150 Cambridge, 186 Leeds, 154 London, 146, 184 Manchester, 138 Analysis of products of combustion, 165, 170, 184, 196, 235 from Crossley engine, 294 VOL. I. Apparatus : Bairstow and Alexander's, 174 Boston, 148 Clerk's, 127, 145, 215 Grover's, 153, 154, 160 Hopkinson's, 187, 193, 194 Petavel's, 167, 168 Area exposed, heat loss per unit, 205 Atkinson, 34, 35, 241 — ' Cycle engine,' 35 — ' Differential engine,' 35 Available heat, definition of, 140 Bairstow and Alexander's appara- tus, 174 -experiments, 173-185, 201- 204 Balance-sheet, heat, Clerk on adjust- ment of, 265-273 Balance-sheets, heat. Clerk's tests, 265, 269, 270, 293 compared, 270, 273, 292, 293 Hopkinson's test of Crossley engine, 292 Institute of Civil Engineers tests, 265 ■ of Stockport engine, 273 Barnett's compression engines, 7-10, 59 — igniting cock, g Barsanti and Matteucci's engine, 11, 13 Beau de Rochas on compression, 18, 59 Benier suction-gas producer, 32 B B 370 THE GAS, PETROL, AND OIL ENGINE BER Berthelot's explosion wave, 142 Berthelot on gaseous explosions, 120, 135, 137 — on time of explosion, 143 — and VieiUe on flame propagation, 116 Bischoff gas engine, tests of, 236 efficiency of, 236 Blast furnace gas used in gas engines, 33 Boston apparatus, 148 — experiments, 201-204 — explosion experiments, 148-153 Bousfield on the Otto theory, 41, 44 Boyle's law, 63 ■ deviation of gases from, 172, 361 Brayton gas engine, 20, 58 — petroleum engine, 23, 27, 28 efficiency of, 240 friction of, 238 tests of, 238-240 pump, 23, 24 Brown's gas vacuum engine, 3 applied to a boat, 5 vehicle, 5 Bunsen corroborates Davy's ex- periments, 113 — on explosion, 43, 135, 137 — on highest temperature of com- bustion, 125 — on velocity of flame propagation, 113 Burstall, Prof., experiments on a Premier engine with varying com- pression, 243, 244, 295-319 — conclusions on experiments, 305 Calculated examples of efliciency of the types, 83 — and observed pressures, 135, 136 Calculation of adiabatic lines with varying specific heat of working fluid, 335 — of efficiency with varying specific heat, 336 Callendar, Prof., appendix by, to Brit. Assoc, report, 361 Calorific intensity, 122 — power, 122 — value of coal gas, 156, 184, 186, 196 acetylene, 166 Calorimeter, exhaust gas, 255, 256 — experiments : Andrew's, 117 Favre, 117 Holborn and Austin, 120-123 Hopkinson's, 193-199 Calorimeter experiments : Joly, 119 Silberman, 117 — recording, 193 Cambell Gas Engine Co., 35 Carbon, heat evolved by, 122 — temperatures produced by, 123 Carnot cycle, 81 ideal diagram of, 93 Cayley's air engine, 52, 57 Cecil, Rev. W., experiments on pressures of explosions, 2, 3 explosion vacuum engine, 2 Charles' law, 62, 136 "Chemical equilibrium, 353 Clausius on specific heat, 118 Clerk, Dugald, 34, 241 — cycle, 34, 35, 60, 322-327 — — engine, 322-324 exhaust timed by piston, 34 — ■ indicated and brake thermal efficiency, 326 ■ — explosion experiments, 126-148, 345 -apparatus, 127, 145 and cooling experiments with a moving piston, 214-235 analysis of working fluid used, 235 — indicator diagrams, 216-218, 221, 268, 273 optical, 259 — ' James Forrest ' lecture, 309 — on adjustment of heat balance- sheets, 265-270, 273 ■ — on suppression of heat, 135 — on theoretic efficiencies, 36 — test of Bischoff engine, 236 Brayton engine, 238-240 National engine, 267 Otto and Langen engine, 237 Cockerill blast-furnace gas engine, ^ 33> 47 Combining volumes, iii — weights, no Combustion at constant volume, 123 — available heat of, 140 — Bunsen on highest temperature _ of, 125 • — continued, 136, 200 — definition of, 128 — heat evolved by, 117 — incomplete, 137, 159, 330 — oxygen required for complete, 138, 186 — products of, 112, 138, 165, 170, 184, 191, 330 effect of, on explosions, 153, 158, 159 — — ratio of specific heats of, 192 INDEX 371 COM Combustion products, volume of, 138 — space, effect of shape on time of explosion, 143 — temperature of, 122, 123 — ■ and explosion, 109 Comparison of actual and theoretical efficiencies, 246 Clerk, Boston, and Grover results, 155 experiments in a closed vessel, 201-213 heat-balance sheets, 270, 273 Langen and Mallard and Le Chatelier's results, 350, 351 — • — Otto and Lenoir diagrams, 40 Compression, Barnet on, 7 — Beau de Rochas on, 18 — Burstall on, 294-319 — Clerk on, 36 — cycles, further calculations on, 92 discussion of, 81 — economy due to, 36, 37 — effect of, on efficiency, 36, 86, 90, 243-246 — engines, Atkinson's, 35 Barnett's, 7, 8, 10 Brayton's, 20, 29 Cambell, 35 Clerk's, 35 — ■ — Million's, 17 Otto's, 29 Siemens', 18 Stockport, 35 Tangyes (Robson's), 35 — Fleeming Jenkin on, 36 — ■ Meyer on, 245 — Million on, 17 — of gases after ignition, 191 — ■ Schmidt on, 16 — Siemens on, 18 Constant efficiency, 83, 107 — pressure cycle, 82 ■ experiments, 342 rates of inflammation at, 113, 115 ■ — temperature cycle, 81 — volume cycle, 82 Contraction, effect of dilution on, 139 — of volume, in, 137-139 Cooling, cause of low efficiency in Brayton engine, 240 — curves, 173, 179, 183, 188, 195, 201, 202, 207, 219, 220, 223, 224, 227, 350 — .effect of, on highest temperatures possible, 135 size of vessel on, 201 surface on, 141 wall temperature on, 225 CUR Cooling, explosion and, behind a moving piston, 214-235 ■ — ^ in a closed vessel, I26-C99 — from the same mean temperature curves, 202, 207 — rate of, 136, 172, 201, 202 — resistance of mixtures to, 132, 133, 151, 152 Critical proportion of gas in mixture, 112 Crossley Bros., 48 — Otto engine, efficiency of, 243 frictional losses of, 277 improvements in design of, 33 mechanical efficiency of, 277 pumping losses of, 279, 280 — ■ thermal efficiency of, 282- 295 tests of, 243, 275-280 — pressure-gas producer, 32 ■■ use of bituminous fuel in, 33 — suction-gas producer, 32 Curves, explosion cooling, 128, 131, 150, 156, 162, 169, 176, 179-182, 188, 195, 201, 202, 207, 209, 216- 218, 220, 221 Curve of cooling from same mean temperature, 202, 207 (Langen), 350 — of heat loss (Hopkinson), 198 — of internal energy of gas engine mixture, 357 — of maximum pressures, corrected for cooling (Langen), 351 — ■ of mechanical loss in Daimler engine, 28 r — of proportional heat flow at dif- ferent densities, 231 — of temperature fall for different initial pressures, 209, 210 ^ of cylinder contents at inner end of stroke, 219 — of temperature fall in terms of density, 211 in time, 227 — ■ showing increase of thermal efficiency and heat loss with increasing compression, 313 indicated thermal efficiency in terms of strength-of mixture, 287 internal energy" at different temperatures, 344 ^ ■ — ■ — for weak and strong mixtures, 289 relation between gas consump- tion and mean pressure, 285 — • — thermal efficiency for varying compression ratios, 310 372 THE GAS, PETROL, AND OIL ENGINE CYC Cycle, Carnot's, 8i — Clerk's, 34, 60, 323-327 — ■ constant efficiency, 86 pressure, 82 temperature, 81 volume, 82 — Otto's, 35, 60 — two, engine, thermal efficiency of, 321 Cycles of action, 56-61 Cylinder, explosion and cooling in aj behind a moving piston, 214-235 — • influence of size of, on heat loss, 205 — walls, effect of mean temperature of, on cooling, 225, 230 Daimler, 30 — connection with Otto, 31 — petrol engine, 30 Hopkinson's tests of, 280- 282 mechanical efficiency of, 280-282 Davy, Sir H., on inflammability, 112 Delamere-Deboutteville's blast-fur- nace gas engine, 33 Density, effect of, on heat flow, 231 loss, 211, 212 ■ specific heat, 119 Determination of charge weight in internal-combustion engines, 332 Deutsche Gas-Kraft-Gesellschaft, 48 Deutz, Gasmoteren Fabrik, 48 Deviation of actual gases from Boyle's law, 172, 361 from the ideal state, 361-365 Deville on dissociation, 124 Diagram from recording calorimeter, 195 Diagrams : Brayton, 25, 239 Clerk, of alternate compression and expansion in a cylinder, 216-218, 221, 268, 273 of explosion in a closed vessel, 128, 131, 143, 146, 149, 161, m 162, 164, 169, 176, 188, 192 ideal air engine, of constant pressure type, 94 temperature type, 93 volume type, 97 Institute of Mechanical Engineers' experiments, 296 Otto and Lenoir compared, 40 Diesel oil engine, 31, 238 ^ compression ratio of, 93 EFF Diesel oil engine, definition of indi- cated horse-power of, 319 — — — ignition by temperature of compression, 31 thermal and mechanical effi- ciency of, 319 Differences of temperature in a closed vessel, 192 Differential engine, 35 Dilution of mixtures, 112, 129 Discussion of differences between actual and ideal engines of the third type, 327-331 Burstall's experiments, 307- 319 Dissociation, 123-125 — definition of, 124 — Deville on, 124 — Groves on, 124 — Nernst on, 200 — Petavel on, 172 Distribution of heat loss to jacket water, 266, 316 — of temperatures in a closed vessel, 190-193 Dowson pressure gas-producer, 32 — producer, effect of, on gas-engine development, 32 — suction producer, 32 Dulong and Petit' s law, 118 Economy, Clerk on causes of, 35, 37 — compression source of, 36, 37 — effect of jacket temperature on, 105 mean temperature on, 205 — of Otto and Langen engine, 20 Efficiency, absolute and relative, 290 — calculated examples of, 83 — Clerk on, 36 — conclusions on, 102 — curve of indicated thermal, in terms of strength of mixture, 287 — definition of, 63 — distinction between theoretical and practical, 108 — formulae, 67 summary, 80 — indicated and brake thermal, of two-cycle engines, 326 thermal, 243 — influence of initial temperature on, 107, 108 of jacket temperature on, 105 — Jenkin on, 36 — mechanical and thermal, of Diesel engine, 319 — of acetylene gas in explosive mixtures, 166 INDEX 373 EFF Eflficiency of gas in explosive mix- tures, 140, 142, 150 — of imperfect heat engine, 66 — of perfect heat engine, 65 — table of ideal, 274 of theoretic, calculated, 99 thermal, of the three symmetrical cycles, 92 — ■ thermal, of the two-cycle engines, 321-327 and mechanical, of the different types, 236-331 for varying compression, 310 — Witz on, 37 Efficiency of Bischoff engine, 236 Brayton engine, 240 Clerk engine, 326 Crossley engine, 243 Diesel engine, 3 1 9 Hugon engine, 236 Lenoir engine, 236 Otto and Langen engine, 237 Equilibrium, chemical, 353 — thermal, 352 Ericsson's air engine, 52 fuel used, 53 Exhaust gas calorimeter, 255, 256, 258-264 Experimental errors in the deter- mination of specific heat by con- stant pressure methods, 362 Experiments : Bairstow and Alexander's, 173- 185 Boston, 148-153 Clerk's, 126-148, 214-235, 267-274 Grover's, 153-166 Holborn and Austin's, 37, 41, 44, 344 Hennmg, 341 Hopkinson's, 185-199, 275-295 Institute of Civil Engineers', 247- 268 Mallard and Le Chatelier's, 114, 116, 120, 130 Petavel's, 166-173 Experiments in closed vessels com- pared, 200-213 Explosion, Bunsen on, 43, 113 ■ — chemical reactions of, in, 112 — definition, 128 — experiments, 348 — highest possible temperatures of, 125 — Hirn on, 43, 113 — Mallard on, 43 — pressure produced by, 2, 129, 130. 134-136, 139, 140, 147) 151, 152 -- temperatures of, 136 FIR Explosion, time of, in closed vessels, 128-130, 142, 147 Explosions in a closed vessel : Andrews on, 117 Bairstow and Alexander on, 173- 185 Berthelot on, 120 Boston gas, 1 48-1 51 Clerk on, 126 Glasgow gas, 129 Grover coal gas, 153-160 Grover acetylene, 160-166 Hirn on, 134 Hopkinson coal gas, 185-199 hydrogen, 130, 134, 135 Langen on, 122 Mallard and Le Chatelier on, 120, 135 Oldham gas, 130 Petavel coal gas, 166-173 petrol vapour, 152 Explosions, gaseous, 337 — with initial pressure above atmo- sphere, 206 — with initial pressure atmospheric, 200 Explosion and cooling in a closed vessel : Bairstow and Alexander on, 173- 185 Boston experiments, 148-153 Clerk on, 126-148 curves, 128, 131, 150, 156, 161, 162, 164, 176, 179-182, i88, 192, 195, 201, 202, 207 discussion of data deducible from, 200-213 Grover on, 153-166 Hopkinson on, 185-199 Petavel on, 166-173 Explosive mixtures : Berthelot and Vieille on, 116 Bunsen on velocity of flame in, 113 Davy, Sir H., on, 112 definition of, 109 efficiencies of gas in, 140 flame propagation in, rates of, 113- 115, 125, 142 inflammability of, 112, 113 Mallard and Le Chatelier on, 114- 116 oxygen required for, in properties of, 109-112 temperatures produced by, 139, 140 volume of products of, 112 First compression gas engine, 7 — working gas engine, 2 374 THE GAS, PETROL, AND OIL ENGINE FLA Flame propagation : Berthelot and Vieille on, ii6 Bunsen on, 113 Clerk on, 128 Hopkinson on, 191 influence of combustion space on rate of, 143 Mallard and Le Cha teller on, 114- 116 rates of, 113, 115, 142, 191 Formulae : efficiency, 80, 248 heat flow, 232, 233 specific heat, 234 dynamical value, 234 Free piston engines : Barsanti and Matteucci, 11, 13, 61 Gillies', 61 Otto and Langen's, 20, 30, 61 Type lA, 60 Wenham's, 61 Frictional losses of Crossley engine, 277 Furnace engine, Cayley's, 52 Wenham's, 53 — loss in engine cylinder, 141 Gas, coal, analysis of : Boston, 150 Cambridge, 186 Leeds, 154 London, 146, 184 Manchester, 138 Gas, coal and air mixtures, best proportions of, 131-134 consumption, measurement, 283 • efficiency of, in mixtures, 140- 142, 150 specific heat of, 184 Gas engine, the, method, 51 — engines classified, 35, 56 compound 30 gases used in, m heat losses in, 103 the working fluid in, 51, 102, 200 two-cycle, thermal efficiency of, 321 A Gas producers : Benier's suction, 32 Crossley bituminous plant, 33 Crossley's suction, 32 Dowson's pressure, 32 Dowson's suction, 32 Mond's pressure, 32 National Co.'s suction, 32 suction producers, fuel for, 33 reduced cost of power due to, 23 HEA Gases, compound, used in gas engines, in — heat flow from, 193-199 Gaseous explosions, 129, 130, 135, 136, 337 Gay-Lussac's laws, in Gillies' free piston engine, 61 Grover's apparatus, 153, 154, 160 — experiments, 153-166 Groves on dissociation, 124 Hautefeuille's, Abb^, engine, i Heat, added, 136 — available, definition of, 140 — balance-sheet. Clerk on adjust- ment of, 265-273 Clerk's experiments, 265, 268, 269 compared, 270, 273, 292, 293 Crossley engine, 292 Institute of Civil Engineers' tests, 265, 270 — engines, imperfect, 66 perfect, 63 — evolved by carbon, 122 combustion, 117, 184 hydrogen, 117 — ■ — in explosions proportional to pressure, 192 — flow from gases, 193-199, 203- 213, 231 curve of, 195 — loss compared, 201-213 — ■ • — division of, 222, 225 effect of strength of mixture on, 293 — loss of, 351 proportional, at different densities and temperatures, 231 rate of, 198 temperature fall due to, 221, 225 to walls, 200, 204, 205, 212 experiments compared, 201 influence of size of vessel on, 203 ■ unit of, 203 — of compression, 65 — specific, of gases, 118, 119 — suppression of, 134, 136 — unit of, 117, 203 Heat losses in gas engines, 103 conditions for minimum, 105 influence of average tempera- ture on, 104 Heating efficiency of gas in mixtures, 140, 142, 150 INDEX 375 HIR Hirn on explosion, 43 — on suppression of heat, 134 Holborn and Austin's experiments, 120-123, 341 — and Henning's experiments, 341, 345 Hopkinson, Prof. B., experiments in a closed vessel, 185-199, 201- 204 — experiments on Crossley engine, 275-280 on Daimler engine, 280-282 thermal efficiency of Crossley engine, 282-295 — on Burstall's experiments, 314 distribution of temperature in a closed vessel, igo-193 — • — heat losses, 104 Hopkinson, Dr. John, experiments on stratification, 45 Hornsby Akroyd oil engine, 31 — hot chamber ignition, 31 Hugon's engine, 19, 57 water injection in, 20 Huyghen's gunpowder engine, i Hydrogen, no, 112 — engine driven by, 2 — heat evolved by, 117 — mixtures, 113, 115, 130, 134 — pressures produced by, 134 — specific heat of, 118, iig — temperature produced by, 122, 123 highest possible, 125 — unsuitable for use in engines, 134 Impulse every revolution engines : Atkinson's, 35 First successful, 34 Robson's, 35 Imray on Otto theory, 37, 41, 44 Incomplete combustion, 137, 159, 330 Increase of specific heat, 136, 200, 343 Indicated and brake thermal effi- ciency of two-cycle engines, 326 — power, definition of, 276 — thermal efficiency, 243 Indicators, optical, 259, 276, 283 Inflammability, 112 — Davy on, 112 — effect of dilution on, 112, 113 — ■ pressure on, 112 temperature on, 112 Inflammation, definition of, 128 — complete, 136 Initial pressure, influence of, on tem- perature fall, 207-209 Institute of Civil Engineers' tests, 247-267 diagrammatic arrangement of apparatus, 257, 258 Mechanical Engineers' Gas Engine Research Committee, 295 report, 295-306 Internal energy, curve of, at different temperatures, 346 of, for gas-engine mixture, 357 Isothermal line, 65, 66 — and adiabatic expansion diagram, 247 Ideal efficiencies, table of, 274 — air-engine diagrams, 73-102 Igniting arrangements : — Barnett's cock, 9 — ■ Diesel by temperature of compres- sion, 31 — Hornsby hot chamber igniter, 31 — Lenoir's electric, 14 — Newton's incandescent tube, 11 Ignition, compression of gases after, 191 — Lodge high tension, 302 — premature, 302-305 — rate of, effect of combustion space on, 143 effect of ignition point on, 175-178 effect of, on temperature fall distribution, 226 Imperfect heat engines, 66 efficiency of, 66 — mixing of gas and air, effect of, 156, 159 Jacket, water, use of, 54, io6 most economical temperature of, 105, 106 James Forrest lecture by Clerk, 309 Jenkin, Prof. Fleeming, lecture by, on Clerk's experiments, 144 on compression, 36 on efficiencies, 36 Joly's calorimeter experiments, 119, 342, 362 Joule's hot air engine, 52, 57 KORTING, 34, 48 — engine, 324, 325 indicated and brake thermal efficiency, 325 uses Clerk cycles, 35, 48, 324 Langen on specific heat, 121, 122 Large gas engines, 47 Law, Boyle's, deviation of gases from, 172, 361 376 THE GAS, PETROL, AND OIL ENGINE Law of temperature fall, 221 Laws, gaseous : Boyle's, 63, 361 Charles', 62, 136 Dulong and Petlt's, 118 Gay-Lussac's, in Lebon's engine, 2 Lenoir electric ignition, 14 — engine, 14, 16, 57 efi&ciency of, 236 Tresco's tests of, 236 — road carriage, 16 Limits of maximum temperatures, 137, 138, 185 — of thermal efficiency, paper on, by Dugald Clerk, 289 Lodge high-tension ignition, 302 Longridge's test of Diesel engine, 319-321 Loss in gas engines, 103-108 frictional, 277, 280 furnace, cylinder, 141 — ■ heat (Crossley) , 294 pumping, 205, 279, 280 relative, 205 — of heat, 351 curve of, 198 to walls, 200 influence of size of vessel on, 203 Mallard on explosion, 43 — ■ and Le Chatelier's experiments, 114-116, 120, 135 Massachusetts Institute of Techno- logy (Boston) experiments, 148-153 Measurement of internal energy of gases, 341 — of temperatures, 357 Mechanical efficiency, 243, 250 of Crossley engine, 279, 280, 319 of Daimler engine, 280-282 of Premier engine, 319 — loss, definition of, 276 of Crossley engine, 278 of Daimler engine, 281 — thermal' and, efficiency of Diesel engine, 319-321 m and efficiency of gas engines, 236-331 Meyer, Prof., on compression, 245 Million, 59, 241 — gas engine, 17 — on compression, 17 Mixing, effect of imperfect, 156, 157, 159, 175 — good, essential, 175 Mixtures,' best for non-compression engines, 131-134, 151 OPT Mixtures, carbonic oxide, 135 — combustible, relative volumes of, with oxygen and the products of their combustion, 330 • — critical proportion of gas in, 112 — curve showing effect of varying, on thermal efficiency, 287 — dilute, 112, 133, 138, 140, 178, 192, 287-295 — efficiency of gas in, 140-142, 151 — explosive, true, 109-112 — gas engine, internal energy of, 357 — hydrogen, 113, 115, 130, 134 — limits of maximum temperatures produced by, 137, 138 — of acetylene and air, 160-166 — of Glasgow gas, 129 — of Oldham gas, 130, 136 — of petrol and air, 151, 152 best, 152 — rich explosions of, 188 — saturated with moisture, 187 Mond gas, Burstall's experiments with, 295-320 — pressure-gas producer, 32 National Gas Engine Co., 48 engine testing-room, 259 — engine efficiency of, 243 heat balance-sheets of, 265 tests of, 248-267 ' X ' type, 259 — suction-gas plant, 32 Nernst on dissociation, 200 Newton's engine, 11 — incandescent tube igniter, 1 1 Non-compression engines : Barsanti and Matteucci's, 11, 13 Gillies', 61 Hugon's, 19, 57 Lenoir's, 14, 16, 57 Otto and Langen's, 20, 30, 61 Street's, i Wenham's, 61 Wright's, 5 Non-compression engines, best mix- ture for, 131 Oechelhauser engine, 48, 325 indicated and brake thermal efficiency of, 326 Oil engines : Brayton's, 23 Priestman's, 31 Optical indicators : Clerk's, 259 Hopkinson's, 275, 283 diagrams from, 276, 288 INDEX 377 OTT Otto, 29, 59, 241 — cycle, 35 thermal and mechanical effi- ciency of, 243 — engine, economy of, due to com- pression, 36, 37 — proposed compounding, 30 — ' Silent ' engine, 29 — theory, 37 discussion of, 42-47 Otto and Langen free piston engine, 20, 30, 61 economy of, 20 efficiency of, 237 tests of, 237 Oxygen required for complete com- bustion, 138 for explosive mixtures, in Oxyhydrogen flame, temperature of, 124 Papin's experiments, i Perfect heat engines, 63 Petavel, Dr., apparatus, 167, 168 — experiments of, 166-173 — on h'eat losses, 104 Petrol and air explosion experiments, 151, 152 — engines mean pressure in, 153 Daimler's, 30 — mixtures, increase in volume of, on combustion, 330, 333 Premature ignition, 302-305 Premier scavenging engine, 49 Burstall's tests of, 243, 244 efficiency of, 243, 244 Pressure and temperature, 64, 136 — heat evolved proportional to, 172 — highest, possible without compres- sion, 129, 130 — initial, influence of, on tempera- ture fall, 207, 2og — maximum, effect of imperfect mix- ing on, 156 of products of combustion on, 153, 158 — mean, effect of varying mixture on, , 28s in petrol engines higher than in gas engines, 153 — produced by explosion, 2, 129, 130, 134-136, 139, 140, 147, 151, 152, 155, 158, 169-172 effect of ignition point on, 177 if no loss existed, 134-136 proportional to initial pressure, 140 Priestman oil engine, 31 spray vaporiser, 31 SLO Pumping losses Crossley engine, 279, 280 — water, i, 4, 15 Radiation loss, measurement of, 263 Rankine on air engines, 52 — on available heat, 140 — on science of thermodynamics, 62 Rate of cooling, 136, 198, 201, 202 • — Petavel's curve, 173 flame propagation, 113, 115, 142 heat loss, 198 reduction of, 213 temperature fall, 228 Ratio of air and exhaust to gas in charge, 264 — of compression, effect of, on effi- ciency, 243 experiments on varying, 243-319 specific heats, 118, 121, 192 Recording calorimeter, 193 record from, 195 — thermometer, 186 Regnault on specific heats, 118, 136, 341 Relative indicated efficiency, 252 Report of the British Association Committee on Gaseous Explosions, 337 Institute of Civil Engineers' Committee, 247-266 of Mechanical Engineers' Gas Engine Research Committee, 295 Residual exhaust, effect of, on maxi- mum pressures, 153, 158, 159 Resistance of exhaust and air dis- charge, 279 Robson, 34, 35 Sankey, Captain, 295, 337 — on Burstall's experiments, 313, 314 Scavenging, effect of, 295 Schmidt on compression, 16, 59 Schottler on stratification, 44, 46 Siemens, 59 — on air engines, 52 — on compression, 18 Simon steam gas engine, 26, 58 Slaby, on Otto theory, 37-41 Slade's test of Lenoir engine, 236 Slow combustion, Bousfield on, 41, 44 Imray on, 41, 44 Otto on, 37 — — Slaby on, 39-41 378 THE GAS, PETROL, AND OIL ENGINE SPE Specific heat, Berthelot on, 120 — — effect of density on, 119 temperature on, 121 effective, 361 formulcE, 234 Holborn and Austin on, 1 20, 341 Holborn and Henning on, 341- 345 increase of, 136, 200, 343 instantaneous, 235 Jolyon, 119 Langen on, 121, 122 Mallard and Le Chatelier on, 120 mean apparent, 235 measurement of, at high tem- peratures, 341 of dry air, 102, 118, 119, 342 gases, 118-121 at constant pressure, 118, 119 volume, 118-121 Regnault on, 118, 136, 341 varying, adiabatic line for, 271 calculation of, 335 Wiedmann on, 341 Sterne and Co., first builders of Clerk cycle engine, 35 Stirling's air engine, 52 fuel used, 53 Stockport Co., 35 — gas engine, tests of, by Clerk, 272, 273 Stratification, 44, 290 — Bousfield on, 41, 44 — Clerk's experiments on, 45, 46 — experiments on, 40, 41, 45 — fallacy of, 44-47 ^ Hopkinson, Dr. John, on, 45 — Imray on, 41, 44 — Lenoir on, 16 — Otto on, 29, 37 — Schottler on, 44 — Slaby on, 39, 41 Street's gas-engine pump, i Suppression of heat, 134, 136 ■ Clerk on, 135 Him on, 134 Surface film, effect of, on cooling, 214% — heat loss per nnit, 205 — influence of, on heat loss, 204, 205, 226 Table of apparent instantaneous specific heat, 235 mean specific heat, 235 — of dimensions of National gas engines tested, 256 — of engine tests, 243 Table of heat loss, 204 per square foot of surface, 212 — of ideal efficiencies, 274 — of indicated and brake thermal efficiency of two-cycle engines, 326 — of results of Bairstow and Alexan- der's experiments, 177, 178 of Boston experiments, 151, 152 of Burstall's experiments, 306 of Clerk's experiments, 129, T30, 135, 136, 139, 140, 147 of Grover's experiments, 155, 158 of Petavel's experiments, 170-172 temperatures (Hopkinson), 190 • falls, 203, 208 due to heat loss, 225 theoretic efficiencies calculated in the examples, 99 thermal efficiencies for the three symmetrical cycles, 92 ■ thermal efficiency of Crossley engine with varying mixture, 282 with varying compres- sions, 245 Tangye-Robson engine, 35 Temperature and pressure, 64, 136 — • average, of working fluid, effect of, on heat loss, 104 — • curve of, of cylinder contents at inner end of strokes, 219 — distribution of, in a closed vessel, 190-193 — effect of, on specific heat, 11 9-1 21 — effective, 361 — fall curve for double strokes, 220, 223 for single strokes, 224 — • of, in terms of density, 211 of, in time, 227 showing effect of initial pres- sure, 209, 210 due to heat loss, 221, 222 effect of mean temperature on, 220 effect of wall temperature on, 225, 230 — • — influence of initial pressure, 107, 109 of size of vessel on, 205 law of, 221 table of, 203, 204 — highest possible, 125 — ■ effect of cooling on, 135 — in hot air engines, 52, 53 — initial, effect on efficiency, 107, 108 INDEX 379 TEM Temperature, initial, effect of residual exiaust gases on, 107 — maximum (Hopkinson), 191 — measurement of, 357 — of combustion, 122 — of explosion, 136-140 diagram, 188 limits of error in calculating, 139, 185 limits of maximum, 137, 138 measurement of, by plati- num thermometer, 186 table of (Hopkinson) ,190 Temperature produced by carbon, 123 by hydrogen, 122, 123 Tests of gas engines : Bischoff, 236 (Clerk) Brayton, 238-240 (Clerk) Crossley, 275-280 (Hopkinson) Daimler, 280-282 (Hopkinson) Diesel, 319-321 (Longridge) Hugon, 236 (Tresca) Lenoir, 236 (Tresca) National, 267-275 (Clerk) — , 247-267 (Inst. Civil Eng.) Otto and Langen, 237 (Clerk), (Tresca) table of, 243 Thermal and mechanical efficiency of Diesel engine, 321 — efi&ciency, curve showing, for vary- ing compression, 310 indicated, 236-331 influence of compression on , 244 -of Daimler engine, 282 of the two-cycle engines, 321- 327 table of indicated, with varying mixture, 284 • — equilibrium, 352 — lines, 65 Thermodynamics of the internal-com- bustion engine considered as an air engine, 62 Thermometer, platinum resistance recording, 186 record from, 188 Thomson, Sir W., on air engines, 52, 57 Thwaite on blast-furnace gas, 33 Time of explosion, 128-130, 147, 151, 152, 162-164 acetylene mixtures, 165 Bairstow and Alexander on, . 176-178 Berthelot on, 143 effect of ignition point on, 176-178 VEL Time of explosion, Hopkinson on, 191 in closed vessels, 142 Petavel on, 170-172 Tresca' s experiments on Bischoff engines, 236 on Lenoir and Hugon engines, 236 on Otto and Langen engine, 237 Two-cycle engines, efficiency of, 321- 327 Types of gas engines, 35, 37, 56 heat losses in, compared, 104 ■ relative theoretic effi- ciencies, 36 — thermal and mechanical efficiency of the : Type I, 236 — lA, 237 — II, 237 — Ill, 241 Type, first description of perfect cycle, 56 — second description of perfect cycle, 57 — third description of perfect cycle, 59 — lA, description of perfect cycle, 60 — I : Hugon engine, 19 Lenoir engine, 14 — 2 : Brayton engine, 20 Simon engine, 26 — 3 : Atkinson engine, 35 Clerk engine, 34 Otto engine, 29 Stockport engine, 35 Tangye engine, 35 — lA : Barsanti and Matteucci engine, n-13 Gillie's engine, 61 Otto and Langen engine, 20 Unit of heat, 117 — — — loss, 203 proportional to initial pressure, 211 Unburned gas, 159, 331 Vacuum engines, i, 2, 3 Velocity of flame propagation, 113 • — Berthelot and Vieille on, 116 Bunsen on, 114-116 38o THE GAS, PETROL, AND OIL ENGINE VEL Velocity of flame propagation, Clerk on, 128 Hopkinson on, 191 Mallard and Le Chatelier on, 114-116 Volume, change of on combustion, 330 — contraction of, in, 137-139 — increase of, with certain hydro- carbons, 330-331 — ratio of surface to, effect on cool- ing, 141 WRI Volume, relative, of combustible mixture with oxygen and the products of their combustion, 330 Wenham's air engine, 52 — free piston engine, 61 Wiedemann on specific heat, 341 Witz on theoretic efficiencies, 37 Wright's engine, 5 'K^t>;fe^r^S;^^^§:^*-^;5&^ r* ^r^^^ir^-^ ^^^^^ ^^i^^ci v>'^v>f »*>^''*'^>s