JENGINEERIHG LIBRARJ THE AIRPLANE ENGINE 5K? Qraw-MlBook & 7m PUBLISHERS OF BOOKS FO R^/ Coal Age v Electric Railway Journal Electrical World ^ Engineering News-Record American Machinist ^ Ingenieria Internacional Engineering 8 Mining Journal ^ Power Chemical & Metallurgical Engineering Electrical Merchandising THE AIRPLANE ENGINE BY LIONEL S. MARKS, B.Sc., M.M.E. PROFESSOR OF MECHANICAL ENGINEERING, HARVARD UNIVERSITY MEMBER AMERICAN SOCIETY MECHANICAL ENGINEERS, FELLOW AMERICAN ACADEMY ARTS AND SCIENCES FIRST EDITION McGRAW-HILL BOOK COMPANY, INC. NEW YORK: 370 SEVENTH AVENUE LONDON : 6 & 8 BOUVERIE ST., E. C. 4 1922 T ' l~7 M 3 Engineering Library COPYRIGHT, 1922, BY THE MCGRAW-HILL BOOK COMPANY, INC. THE MA.PUE I'K K S S YORK MY WIFE JOSEPHINE PRESTON PEABODY 550891 3 Engineering Library COPYRIGHT, 1922, BY THE MCGRAW-HILL BOOK COMPANY, INC. THE MA.PLE PRESS YORK MY WIFE JOSEPHINE PRESTON PEABODY 550891 PREFACE This volume attempts two things: to formulate existing knowledge of the functioning of the airplane engine and its auxiliaries; and to present and discuss the essential constructive details of those engines whose excellence has resulted in their survival. The material here collected is largely new; very little of it could have been written before the war and only a small frac- tion was available for publication before 1919. It is based mainly on the researches and engine developments originating during the war and resulting from the war's urgencies. The researches have been carried out almost exclusively under governmental auspices; in the United States at the Bureau of Standards and at the Air Service experimental plant at McCook Field; in Great Britain at the Royal Aircraft Factory and the National Physical Laboratory; in France and Germany at equivalent institutions. Many of the results of these investiga- tions were published confidentially during the war in Reports of the Bureau of Standards; in Bulletins and Technical Orders of the Airplane Engineering Division of the U. S. Army; in Reports of the (British) Advisory Committee for Aeronautics; in Bulle- tins de la Section Technique de 1'Aeronautique Militaire; and in Technische Berichte. This material has now become avail- able and much of it has been published in the Reports of the (U. S.) National Advisory Committee for Aeronautics and in the technical press. Similarly, the constructive details of most of the existing airplane engines are now available, chiefly from descriptions of captured machines. The German and Austrian engines captured by the British were subjected to a technical analysis which has set a new standard in such matters. Not only were the engines and their auxiliaries tested exhaustively for performance but all the parts were minutely measured, loads and stresses calculated, and the metal analyzed for chemical composition. The French carried out similar analyses of German engines. The Germans published corresponding, though less detailed, analyses of vii viii PREFACE French, English and American engines. Since the war, the U. S. Air Service has also analyzed American and foreign engines and has published its findings in Technical Orders and Informa- tion Circulars. With all this material, the designer of the airplane engine has at hand more detailed precedent from which to depart than is available for other types of engine. The writer desires to acknowledge his indebtedness to Professor E. B. Warner and to Lieut. E. E. Aldrin for assistance in obtaining information; and to Mr. R. H. Taylor for assistance in reading proofs and in preparing the index. L. S. M. CAMBRIDGE, MASS. January, 1922. CONTENTS PAGE PREFACE vii CHAPTER I. POWER REQUIRED AND POWER AVAILABLE 1 II. ENGINE EFFICIENCIES AND CAPACITIES . . / 11 III. ENGINE DYNAMICS 40 IV. ENGINE DIMENSIONS AND ARRANGEMENTS 60 V. MATERIALS 114 VI. ENGINE DETAILS 122 VII. VALVES AND VALVE GEARS 151 VIII. RADIAL AND ROTARY ENGINES 176 IX. FUELS AND EXPLOSIVE MIXTURES 212 X. THE CARBURETOR 245 XI. FUEL SYSTEMS . 289 XII. IGNITION. ....... ^ . 295 XIII. LUBRICATION. 327 XIV. THE COOLING SYSTEM. . . , 344 XV. GEARED PROPELLER DRIVES 378 XVI. SUPERCHARGING . 386 XVII. MANIFOLDS AND MUFFLERS 416 XVIII. STARTING .....;. . . 424 XIX. POTENTIAL DEVELOPMENTS 434 INDEX . 445 IX THE AIRPLANE ENGINE CHAPTER I POWER REQUIRED AND POWER AVAILABLE Power Required for Flight. An airplane in flight is sustained by the lift of the wings. Consider the wing as a thin flat plate, Fig. 1. Four forces are acting: FIG. 1. Forces acting on a flat plate. 1. The weight, W, of wing and parts supported thereby, downward. 2. The thrust, T, or forward impulse due to the propeller. 3. The lift, L, of the air which acts in a direction perpendicular 1 2 THE AIRPLANE ENGINE to that of the plane with respect to the air, and produces sustentation. 4. The wing resistance or drag, D, measured perpendicularly to L, the component of the total force on the wing which opposes forward motion. At constant horizontal speed, L = W and D = T. As these pairs of forces are not acting in the same lines they give rise to turning moments and it is necessary for stability that L X m = T X n. Both L and D are created by the velocity of the plane; the direction of the latter is concurrent with (but opposite to) the path of flight. The former is directed at right angles with the path of flight. In horizontal flight L and D are vertical and horizontal forces respectively. Wing Characteristics. Flat Plates. A plate moving with respect to the air undergoes an approximately normal pressure F which is proportional to the density of the air and (within limits) to the square of the relative velocity. This pressure may be resolved into components L and D perpendicular and parallel to the path of flight. F 2 = L 2 + D 2 . The component L is useful, while D is objectionable. The pressure F depends on the area of the plate, but varies somewhat with its shape. The incidence, i, or angle between the plate and the flight path has a dominating influence on the resulting pressure; and particularly on the relation between L and D. Since L = F cos i and D = F sin i, L/F decreases and D/F increases, as i is increased from deg. For a given angle of incidence the following relations hold: L = K L dAV 2 D = K D dAV 2 where K L and K D are experimentally determined lift and drag coefficients respectively; d is the air density; A is the wing area; and V the relative air velocity. The usual units are L and D in pounds; d, relative air density in terms of normal density; A in square feet; and V in miles per hour. From the above equations T 7^- it is seen that T:= ^- That is, lift is obtained with minimum Lf J\-D TV- drag when^ is a maximum. &D It is found that more favorable ratios of K L /K D are obtained with curved wings, as in Fig. 2, than with flat plates. The angle POWER REQUIRED AND POWER AVAILABLE 3 of incidence of such a wing is arbitrarily defined as the acute angle between the wind direction and the lower chord of the wing. Figure 3 gives the values of K L) K D , and K L /K D or L/D for the wing or aerofoil section shown in Fig. 2, and shows a maximum value of L/D of 17 at an angle of incidence of 3 deg. Chord-- FIG. 2. Cross section of wing. An airplane consists not only of the wings which give susten- tation, but also of other members such as the fuselage, radiator, landing gear, and wing bracing. These give no aid in sustaining the plane but offer a resistance, the parasite resistance, which 0.0028 0.0024 0.0020 20 0.0016 16 :* 0.00 12 12 0.0008 8 0.0004 4 -0.0004 -nonnR ^- ^\ 0.0010 0.0008 0.0006 a M 0.0004 0.0002 o / / v / / f n x < / 1 l x \i I/ x A ^ % -* *-" ^ y ^6 w j /'^ ^ ^ , A ^> P \ is fixed by the clearance and may be Vc varied in a given engine by the use of pistons with heads of dif- 1200 1400 1600 1800 2000 Engine Revolutions per Minu-fce FIG. 12. Variation of compression pressure with engine speed. TABLE OF COMPRESSION PRESSURES, POUNDS PER SQUARE INCH Compression ratios n 3.5 4.0 4.5 5.0 6.0 7.0 1.25 59.9 70.6 81.9 93.2 116.5 143.0 1.30 63.7 75.8 88.3 101.3 128.4 156.9 1.35 67.8 81.2 95.2 109.8 140.4 172.9 1.41 73.1 88.3 104.2 120.9 156.4 194.3 ferent shapes. It is the chief factor in determining p c . The accompanying table is based on pb = 12.5. A high compression ENGINE EFFICIENCIES AND CAPACITIES 17 ratio increases power output and efficiency. It also increases the temperature at the end of compression since = (!l\ n T b t b + 460 " \pj A high value of t c may cause preignition and thus fixes a limit of compression ratio which must not be exceeded. Usual values are from 4.5 in hydroplanes to 5.6 for high-altitude land machines. Values of compression ratio are given for various engines on pages 66 to 70, where it is seen that high values result in increased power output per unit of cylinder volume. Values of six or higher may be used for high altitude work, but the engines will develop preignition if operated at full throttle near the ground. When operated on partial throttle the entering charge is of reduced weight, both because of lower pressure and because of dilution of the fresh mixture with a relatively greater weight of burnt gas remaining over from the previous cycle. Consequently the heat developed per cycle is less and the mean temperatures in the cylinders are reduced. Efficiency. If the working substance in the cylinder followed the laws of a perfect gas pv = RT and C v = constant and if the combustion were instantaneous and complete, the efficiency of the cycle would be equal to -i (i) where r is the compression ratio. It is here assumed further that the cycle takes place without any heat exchange between the working charge and the cylinder. The efficiency so found is the highest possible efficiency for an engine operating on the Otto cycle and could be attained only under the conditions stated above. It is sometimes called the air -cycle efficiency. Its value for various compression ratios is given in the following table : r=3.50 4.00 4.50 5.00 5.50 6.00 6.50 7.00 e = 0.40 0.43 0.45 0.48 0.50 0.51 0.53 0.54 The efficiencies actually obtained in airplane engines are seldom greater than 60 per cent of these values. For instance, with r = 5.5, the actual efficiency will not exceed 0.6 X 0.5 = 0.3 18 THE AIRPLANE ENGINE and will in general be between 0.25 and 0.3. This considerable discrepancy between the actual performance of the engine and the air cycle efficiency is due to a variety of causes, the principal of which are as follows : I. The theoretical cycle assumes that the total heat (lower heat value) of combustion of the fuel taken into the cylinder is utilized during explosion in heating up the working mixture. This is not actually the case for two reasons : (a) The whole of the heat of combustion is not evolved during explosion because combustion is not instantaneous, so that combustion will continue for part (or with incorrect mixture, for the whole) of the expansion stroke, thereby reducing the amount of heat available for conversion into work. Furthermore, complete combustion at the end of explosion is not attainable because chemical equilibrium requires the presence of a certain amount of hydrogen and carbon monoxide. Their existence is commonly ascribed to dissociation. The amount of heat suppres- sion from these causes is not considerable in a high-grade engine operating with gasoline and with a good mixture. (6) Some of the heat actually evolved goes to the cylinder walls by radiation and conduction. The total heat so going to the walls in airplane engines is from 25 to 30 per cent of the total heat of combustion. If the heat were abstracted from the burning mixture during the explosion it would result in a loss of efficiency of the same magnitude. The actual passage of heat from the mixture to the walls continues from the middle of compression to the end of exhaust. Throughout the whole of this time the gases are hotter than the walls. The heat flows in the opposite direction during the admission period and the first part of the compression, but the amount of heat thus flowing is small, as the temperature difference between the walls and the gases is small. Such heat as passes off during the explosion and the first part of the expansion stroke may be regarded as entirely lost to the engine; the heat flow to the walls near the end of expansion and during exhaust is no loss at all, as it is necessary to discharge the hot gases and it is immaterial, from the point of view of efficiency, whether the heat is carried away by the jacket water or in the exhaust gases. General experience would indicate that more than one-half of the total heat given to the jacket may be regarded as abstracted by radiation or conduction from the working substance during ENGINE EFFICIENCIES AND CAPACITIES 19 explosion and the early part of expansion. It should be noted in this connection that the heat of the jacket water includes most of the friction work between the piston and cylinder, which is a considerable fraction of the total friction of the engine. As the jacket heat is 25 to 30 per cent, the heat lost during explosion by radiation and conduction may be taken as not more than 12 to 15 per cent of the heat of complete combustion. II. The working substance is not a perfect gas and, in particu- lar, it is not true that the specific heat at constant volume is a constant. It is found on investigation that the gases (C02, N, H20 etc.) which are present in the cylinder after explosion have specific heats which increase considerably with increase of tem- perature. These specific heats follow the equation C v = a + bt where a and 6 are constants. The efficiency of the cycle is diminished as a result of this increase of specific heat. The immediate result is that the rise of temperature during explosion, for a given amount of fuel burned, is diminished and consequently the pressure, pd (Fig. 10), is lower than would be realized with constant specific heat. The expansion curve de is consequently lowered and the work of the cycle diminished. The efficiency with adiabatic expan- sion and compression but with variable specific heat is given by the expression, 1 where e is the air-cycle efficiency and Td and Tj> are the abso- lute temperatures at the end of explosion and beginning of compression respectively. The constants a and b have values of about 0.194 and 0.051 X 10~ 3 for the average working mix- ture. For the conditions customarily met in airplane engines E the ratio of the two efficiencies is about 0.80; in other words, 6 the theoretical efficiency with the actual working substance is only 80 per cent of that which would be attainable if these sub- stances were perfect gases. The actual working substance consists almost exclusively of nitrogen, water vapor and carbon dioxide. All three of these substances show a considerable increase in specific heat with rise 1 WIMPERIS, The Internal Combustion Engine, p. 85. 20 THE AIRPLANE ENGINE in temperature and the last two dissociate at high temperatures, especially at low pressures. The following tables 1 give mean specific heats at constant volume, and percentage dissociation. MEAN SPECIFIC HEATS AT CONSTANT VOLUME (IN B.T.U. PER DEGREE FAHRENHEIT) BETWEEN 200F. AND THE STATED TEMPERATURE Temperature, degrees Fahrenheit 930 1 ; 830 2,730 3,630 4,530 5,430 Nitrogen. . 185 188 196 205 214 225 Water vapor 0.350 0.385 0.425 0.468 0.540 0.623 Carbon dioxide 0.187 0.217 0.229 0.238 0.247 0.249 DISSOCIATION, PER CENT Pressure in atmospheres Temperature, degrees Fahrenheit 0.1 1.0 10 100 H 2 O 2,730 3,630 4,530 5,430 0.043 1.25 8.84 28.4 0.02 0.58 4.21 14.4 0.009 0.27 1.98 7.04 0.004 0.125 0.927 3.33 C0 2 2,730 3,630 4,530 5,430 0.104 4.35 33.5 77.1 0.048 2.05 17. 6| 54.8^ 0.0224 0.96 8.63 32.2 0.01 0.445 4.09 16.9 Calculation of the theoretical efficiency, taking into account both the variable specific heats and dissociation, shows that this 1 TIZARD and PYE, The Automobile Engineer, Feb., 1921. ENGINE EFFICIENCIES AND CAPACITIES 21 efficiency, with a mixture giving maximum efficiency, is repre- sented very closely by the equation E = 1 - (- 0.295 (2) The heat loss to the walls reduces the actual efficiency below the theoretical values. With the very best design of cylinder and optimum operating conditions the highest attainable indicated thermal efficiency is given fairly accurately by the equation E 1 - (3) A comparison of these efficiency values is given in the following table. There are added the best results obtained by Ricardo 1 on a special engine in which every known refinement was em- ployed with a view to raising the thermal efficiency. CYCLE AND ENGINE EFFICIENCIES Efficiency Compression ratio r Air cycle From equa- tion (2) From equa- tion (3) Ricardo's observed values 4.0 0.426 0.336 0.296 0.277 4.5 0.452 0.359 0.314 0.297 5.0 0.475 0.378 0.332 0.316 5.5 0.494 0.396 0.348 0.332 6.0 0.512 0.411 0.361 0.346 6.5 0.527 0.424 0.375 0.360 7.0 0.540 0.437 0.386 0.372 7.5 0.553 0.449 0.396 0.383 8.0 0.565 0.460 0.406 The difference between the air-cycle efficiency (constant specific heat and no dissociation) and the theoretical efficiency of the cycle using imperfect gases, with the properties given in the preceding tables, diminishes as the explosion temperature diminishes. In the limiting case in which there is no fuel in the charge and consequently no rise of temperature at explosion the two efficiencies become equal. The less the fuel in the charge, or, 1 Proc. Royal Aeronautical Society, 1920. 22 THE AIRPLANE ENGINE the weaker the mixture, the more nearly does the cycle efficiency approach the air-cycle efficiency. Calculations by Tizard and Pye show the cycle efficiency to vary with the mixture strength as in Fig. 13. The curve for correct mixture shows the efficiency when the air-fuel ratio is 258 /1\ 0. chemically correct; the equation to the curve is E = 1 f-J The 20 per cent weak curve is calculated for 20 per cent ex- cess of air which is usually about the limit of explodibility; the improved cycle efficiency in this case is verified by en- gine tests which generally show maximum indicated thermal efficiency with about 20 per cent excess of air. The 50 per cent weak curve represents a condition which cannot be attained in the normal Otto- cycle engine as it gives a non- explosive mixture; it can be realized by an injection of the fuel into the compressed air as in the Diesel cycle, or by hav- - & gtratified charge ^ the ,. , .,, , . Cylinder With an explosive mixture surrounding the ig- niter. In any case the reali- zation of the higher efficiency of the weak mixture will be attended by reduced engine capacity. Another point of importance is brought out by the curves of Fig. 13. The ratio of cycle efficiency to air-cycle efficiency in- creases with the ratio of compression; that is, we may expect to realize a larger percentage of the air-cycle efficiency as the compression ratio increases. The ratio of the efficiencies for a compression ratio of 4 is 0.685; with a compression ratio of 10 it is 0.735. This improvement is shown also in actual engines. The ratio of observed efficiency (Fig. 13) to the air- cycle efficiency rises from 0.65 at a ratio of compression of 4 to 0.685 at a ratio of compression of 7. 7 & 9 10 Compression Ratio FIG. 13. Calculated and observed thermal efficiencies with various strengths of mixture and compression ENGINE EFFICIENCIES AND CAPACITIES 23 III. The theoretical indicator diagram is not realized for still another reason. The admission and exhaust of the charge are attended by frictional resistance to the passage of the gas through the carburetor, inlet manifold, inlet valve, and exhaust valve. Moreover, as the flow of the gas is at high velocity, there must be a pressure drop to bring about this flow; with an inlet velocity of 250 ft. per second, this would amount to about 0.6 Ib. per square inch. The frictional resistance and velocity head cause a lowering of the admission pressure, a, a raising of the exhaust pressure, and the forming of the "loop" (Fig. 11) at the bottom of the indicator diagram. This loop represents the negative pumping or fluid friction work which the engine has to perform. Engine tests indicate that the pumping work increases rather more rapidly than the square of the engine speed; the actual amount of the work depends on the dimensions and arrangement of the engine. At 1,000 r.p.m. it will probably average about 4 per cent of the indicated work of an aviation engine. This means in an engine with 120 Ib. per square inch, brake m.e.p. that the mean height of the loop is 5 Ib. per square inch at 1,000 r.p.m. The pumping loss is a function of the gas velocity in the manifolds and through the valve ports. Its magnitude will vary from about 2 Ib. per square inch with a gas velocity of 100 ft. per second to 8 Ib. per square inch with a gas velocity of 200 ft. per second. The indicated work may properly be considered as being only the positive loop of the indicator card; the suction-exhaust loop is one of the engine friction losses. Minor factors affecting the area of the indicator card are the rounding of the "toe" of the diagram which results from the opening of the exhaust valve before the end of the expansion stroke in order to facilitate exhaust, and the departure of the expansion and compression curves from the theoretical adiabatic curves. IV. It is not usually practicable to determine directly the work done by the working substance in the engine cylinder or the indicated work. In all tests the power measured is the useful work or brake horse power which is what remains of the indicated work after some of it has been used up in overcoming the friction of the engine and in driving the water and oil pumps. Indicated work friction work = useful work Useful work , , , i -ZT. r = Mechanical efficiency Indicated work 24 THE AIRPLANE ENGINE Friction Losses. The mechanical losses in an engine may be divided into two groups: 1. The losses due to bearing friction and the driving of such auxiliaries as valve gears, oil and water pumps, magnetos, etc. 2. Piston friction. Tests by Ricardo show that to overcome the first group a mean effective pressure of from 1.5 to 3 Ib. per square inch is usually required the lowest figure applying to a large multi- cylinder engine. The distribution of these losses is about as follows : Bearings . 75 to 1 . 00 Ib. per square inch Valve gear 0. 75 to 0. 80 Ib. per square inch Magnetos 0. 05 to 0. 01 Ib. per square inch Oil pumps 0. 15 to 0. 25 Ib. per square inch Water pump 0.30 to 0.50 Ib. per square inch Total 2 . 00 to 2 . 65 Ib. per square inch Piston friction is the largest item of loss; its magnitude prob- ably results from the fact that the motion is reciprocating and 1000 1200 1400 1600 1800 2000 2200 Engine Revolutions per Min. FIG. 14. Mechanical efficiencies of airplane engines. that the film of oil on the walls is more or less carbonized by the high temperatures and consequently has a high viscosity. The magnitude is probably about 7 Ib. per square inch of piston area. The total loss from bearing friction, piston friction and fluid friction in the best ungeared engines is from about 10.5 to 14 Ib. per square inch of piston area, the lower figure referring to radial air-cooled engines and the higher to water-cooled engines. Taking 120 Ib. per square inch as the brake m.e.p., these values correspond to mechanical efficiencies of 92.0 and 89.5. Tests (Fig. 14) indicate that friction work increases more rapidly than the engine speed but not so rapidly as the square of the speed. Taking into account the losses enumerated, it is possible to arrive at a fair approximation to the actual efficiency of an air- plane engine. Consider for example a high-grade engine with ENGINE EFFICIENCIES AND CAPACITIES 25 a compression ratio of 5.5, using gasoline as fuel. For every 100 B.t.u. (lower heat value) of heat of combustion we may expect a heat suppression (Item La) of 4 B.t.u., leaving 96 B.t.u. developed. Of this quantity, 13 B.t.u. will go to the walls by radiation and conduction (Item 1.6) before it can be utilized, leaving 83 B.t.u. The theoretical efficiency for a compression ratio of 5.5 is 0.396 (equation (2) p. 21). The theoretical work of the cycle is 0.396 X 83 = 32.9 B.t.u. The actual indicated work is thus 32.9 per cent of the heat of perfect combustion of the fuel and this quantity is usually spoken of as the indicated thermal efficiency, or the thermodynamic efficiency, E t , of the engine. It measures the efficiency of the engine in converting heat into work. As previously stated, this indicated work is not readily meas- urable. The useful or brake work may be taken as 85 per cent (Item IV) of the indicated work, or in this case, 0.85 X 32.9 = 28.0 B.t.u. The thermal efficiency referred to b.h.p. is then 28.0 per cent. This quantity may be compared with the results of tests on a high-grade engine. Such tests may be expected to show the consumption of about 0.50 Ib. of gasoline per brake horse power hour. The fuel has a lower heat value of almost 18,500 B.t.u. per pound. The thermal efficiency referred to Work of 1 b.h.p. hour, B.t.u. 2,545 n n n is - Heat of combustion of the fuel, B.t.u. 0.50 X 18,500 = 0.275, which agrees very closely with the calculated efficiency. A reduction in any of the itemized losses will increase the final efficiency. Mean Effective Pressures. The mean effective pressure (m.e.p.) of a gas engine is that gas pressure on the piston which, if maintained constant for one stroke of the engine, would do as much work as is actually done in the two revolutions of the cycle. In aviation engines the m.e.p. is practically always obtained from the brake horse power and is called the brake m.e.p. It is given by the equation b.h.v. X 33,000 b.h.p. brake m.e.p. = ^r - = 1,083,000 ^: ^, , r v Vx-X-Xn d*XsXNXn 4 rt X 12 X 2 X where d is the cylinder diameter in inches, s is the stroke in inches, N is the revolutions per minute, and n is the number of cylinders. 26 THE AIRPLANE ENGINE The brake mean effective pressures usually given are computed from the b.h.p.; values range from 70 to 135 Ib. per square inch. The true m.e.p. in the cylinder is this value divided by the mechani- cal efficiency, E m . Torque and Power. If p = actual brake m.e.p., the average useful force exerted in the cylinders of a four-cycle engine is 7T 1 ) .nd z X T = 0.1964 pnd 2 Ib. This force is maintained while the piston moves during each revolution 2s in., or s -f- 6 ft. and the o work done in foot-pounds is 0.1964 pnd 2 X fi = 0.03273 pnd 2 s. Torque is the average turning moment and is numerically equal to the force continuously exerted at the propeller at 1 ft. radius. This is exerted, during each revolution, over a distance of 2?r = 6.2832 ft. The work being equal to that already computed, the torque in pound-feet is Q = 0.03273 pnd 2 s -r- 6.2832 = pnd 2 s -^ 192. For 3 = 7, d = 5, n = 12, p = 120; Q = 1,315 Ib-ft. The actual torque varies, but has this average value. 2sN If S = piston speed, feet per minute = -TTT' the b.h.p. is H B = *>| Sn -5- 33,000 = pd 2 Sn ^ 168,000 = Q X N -5- 5,250. Thus for 1,600 r.p.m., in the preceding example, H B = (1,315 X 1,600) *- 5,250 = 401. Capacity and Volumetric Efficiency. The weight of fuel mixture taken into the ideal engine (Fig. 10) is given by the gas equation b v a ) where Vb v a = A d 2 s is the volume of the mixture admitted and 4 T m is the absolute temperature of the external air. Actual engines do not draw in weights equal to that expressed by the above equation. The weight of mixture actually admitted is the difference between the weight present at the points b and 144 pv a, Fig. 10. The weight present at any point is w = -fwir* therefore the 'weight admitted is ~j>%r & III b HI a The ratio of the weight actually admitted to that which would be ENGINE EFFICIENCIES AND CAPACITIES 27 admitted to the ideal engine is called the volumetric efficiency, E v . < a Vg\ * 144 X 14.7 (V b - Vg) 1 nn I * ~D /TT 1 a' K 1 m Writing - = r, the above reduces to - M TJ 14.7(r - 1) V T b Taking t m = 100, r = 5, p b = 12, t b = 200, p a = 16, t a = 900 we find E v = 0.76. The volumetric efficiency is determined mainly by two factors, the temperature, T b , and the pressure, p b , at the end of admission. ,3.3 I 3.1 .ci -[2.9 " 200 400 600 800 1000 1200 1400 1600 R.P. M. FIG. 15. Volumetric efficiencies of hot and cold engines. The temperature of the mixture rises during admission as a result of the addition of heat from the hot interior surfaces of the cylin- der. Comparative tests of the volumetric capacity of an engine (1) when being motored over cold, and (2) in ordinary operation, show that the heating effect decreases slightly as the speed of the engine increases in consequence of the shorter time available for the transmission of heat. The pressure drop, however, increases continuously with increase of engine speed. Figure 15 1 gives curves of weight of charge taken into the cylinder per revolution for an engine of high valve resistance. The effect of the tempera- ture rise in reducing the volumetric efficiency is from 12 to 15 per cent and would be appreciably greater but for the evaporation of the fuel, which, through the abstraction of the latent heat of evap- 1 JUDGE, High-speed Internal-combustion Engine, p. 161. 28 THE AIRPLANE ENGINE oration, reduces the rise of temperature of the charge by 30 to 40F. The variation of volumetric efficiency of the Liberty-12 with engine speed is shown in Fig. 16. In the same figure there is also shown the volumetric efficiency of the Hispano-Suiza 300 engine, but in this case the volumetric efficiency is given as the ratio of the weight of air actually admitted to the weight of a volume of air equal to piston displacement and of the density of the air in the inlet manifold. Measured in this way the volu- metric efficiency is 95 per cent at 1,600 r.p.m.; this corresponds to 93 per cent when compared with air at room density. The broken line shows the volumetric efficiencies compared with air at room density. ume+ric Efficiency, Per C ^J OO TO .o- 11 ~fy 5 -C '***, 1 , . ? $ FIG. 16. Volumetric efficiencies of airplane engines. R.RM. The pressure drop during admission is much more variable in different engines than is the temperature rise. Its magnitude depends on the pressure drop through the carburetor and the size and arrangement of the manifolds and the inlet valve. In every case it will increase rapidly with increased engine speed; its actual magnitude will usually be small for low speeds. The pressure drops in the manifolds of two engines are given in Fig. 17. It will be seen that with wide-open throttle the pressure drop is nearly proportional to the engine speed. With an engine loaded with a propeller, change of speed is obtained only by varying the opening of the throttle valve; the manifold vacuum increases rapidly as the throttle valve is closed. The pressure in the cylinder is considerably less than that in the manifold because of the valve resistances. The volumetric efficiency in engines of good design will be from 80 to 85 per cent. With low speeds and other exceptionally favorable conditions, values as high as 90 to 92 per cent have been recorded. The maximum possible volume of charge admitted per ENGINE EFFICIENCIES AND CAPACITIES 29 cycle in the ordinary engine is the volume enclosed at the instant of valve closure less the clearance volume. The admission valve always closes past the dead center. If the closing angle is 45 deg. late the piston will have returned about 12 per cent of its stroke and the maximum possible volumetric efficiency will be 88 per cent. Occasionally the operating conditions and induc- tion pipe length may be such as to give more than atmospheric pressure in the cylinder at the instant of closure which would result in increased volumetric efficiency. s \ \ x j\ \3\ 1200 Hisparro -SuigaSOOHp. Liberty 6 2400 1400 1600 1800 2000 2200 Revolutions per Minu+e of Engine FIG. 17. Intake manifold depressions with full throttle and with propeller load. It should be noted that the capacity (as affected by volumetric efficiency) and thermal efficiency of an engine are not necessarily related to one another. The diminution in capacity of an engine resulting from heating of the entering charge, from a high carburetor or inlet-valve resistance, or from diminishing speed, may or may not result in a change in efficiency, and the change, if it takes place, may be either an increase or a decrease. A given engine may at one time be developing 200 h.p. ; at another time 250 h.p.; the efficiency may, however, be the same in both cases, although it tends to be lower for the lower h.p. because of the approximate constancy of engine friction, which makes the efficiency referred to the b.h.p. less at light loads. Units of Capacity. In determining the size of a projected engine, or in comparing the performance of existing engines, it is desirable to have some standard unit for measuring the specific capacity. The most common unit is the piston displacement in 30 THE AIRPLANE ENGINE cubic inches per brake horse power, or the brake horse power developed per cubic foot of piston displacement. The piston displacement is the displacement per stroke of one cylinder multiplied by the number of cylinders. As the horse power varies almost directly as the engine speed, the above units do not really lead to a satisfactory comparison of engines operat- ing at different speeds. For this purpose it is better to state the capacity at 1,000 r.p.m., deducing this capacity from the actual performance by the use of the assumption that horse power is proportional to engine speed. For example, the Lib- erty-12 engine, 5 by 7 in., develops 400 h.p. at 1,700 r.p.m. The piston displacement per cylinder is ^ X 5 2 X 7 = 137.4 cu. in. per stroke; the total piston displacement is 12 X 137.4 = 1,648.8 cu. in.; the piston displacement per brake horse power is 1,648.8 -5- 400 = 4.12 cu. in. The brake horse power per cubic foot of piston displacement is 12 3 -f- 4.12 = 420 h.p.; the piston dis- placement per brake horse power at 1,000 r.p.m. is 4.12 X 1 700 = 7.0 cu. in.; the b.h.p. per cubic foot of piston displace- ment at 1,000 r.p.m. is 420 X ~ = 247. ' The Hispano-Suiza engine, with 718.9 cu. in. displacement, develops 150 h.p. at 1,450 r.p.m. This corresponds to 4.8 cu. in. per horse power; or 6.96 cu. in. per horse power at 1,000 r.p.m. This last figure shows that the Hispano-Suiza and Liberty engines have practically the same capacities per cubic inch of piston displacement per minute. The fixed-cylinder radial air-cooled ABC Dragonfly nine- cylinder engine, with 1,389.3 cu. in. displacement, develops 310 h.p. at 1,650 r.p.m. This corresponds to 4.48 cu. in. per horse power, or 7.38 cu. in. per horse power at 1,000 r.p.m. The lower specific capacity of rotating-cylinder engines is illustrated by the nine-cylinder Gnome, with a piston displace- ment of 770 cu. in., which develops 104 b.h.p. at 1,200 r.p.m. This corresponds to 7.41 cu. in. per horse power, or 8.89 cu. in. per horse power at 1,000 r.p.m. The above figures may be regarded as characteristic of the different types. Tests of Performance. The results obtained on the test of an engine will vary greatly with a number of factors such as the air ENGINE EFFICIENCIES AND CAPACITIES 31 pressure and temperature, kind of fuel, type and dimensions of carburetor, temperature of jacket water and of lubricating oil, and condition of engine. For example the Liberty 12 has shown a brake horse power at 1700 r.p.m. which varies from 380 to 480 b.h.p. The following tests are reported under sea-level conditions: Engine Revolu- tions per minute Brake m.e.p., Ib. per sq. in. Brake horse power Friction horse power Me- chanical efficiency Hispano-Suiza-180 1,200 114.0 121.8 16.8 0.88 1,500 1,700 1,900 119.1 117.7 111.0 159.0 178.0 187.0 23.5 28.4 34.2 0.87 0.86 0.84 Liberty-12 1,200 118.0 295.0 27.6 0.91 1,400 1,600 1,800 2,000 119.5 119.5 117.6 104.0 348.0 398.0 442.0 433.0 38.3 49.1 65.4 88.0 0.90 0.89 0.87 0.83 A plotting of test results on the Liberty 12 is shown in Fig. 18. These figures illustrate the usual laws of performance. The 430 460 440 420 -380 J5360 I 340 320 300 / Me -A. E ffici *ncy / / * i - 1 ^ / I hi rmc j f f / ~j f- A / / / / 280 260 1200 1400 1600 1800 2000 R.P. M. FIG. 18. Performance curves of Liberty-12 engine. mean effective pressure, p, and consequently the torque, reach maximum values at some moderate speed. The power increases with increasing speed, but at a rate which diminishes after the 32 THE AIRPLANE ENGINE maximum value of p has been reached. If p were constant the power would vary directly with the speed. Cylinder cooling reduces p at low speeds; high resistance through ports and passages reduces volumetric efficiency and p at high speeds (Fig. 16). Maximum power is reached when the rate of decrease of p with engine speed is equal to the rate of increase of engine speed. Figure 79 gives results of trials on a 230 h.p., six-cylinder Benz engine. Here the failure of the power to increase proportionately to speed is clearly shown. Maximum mean effective pressure occurs at 1,050 r.p.m. and maximum power at 1,650 r.p.m. The fuel consumption rate is also shown. The economy is practically constant over the speed range 900 to 1,200 r.p.m. It will be observed that throttling the engine increases the fuel consumption per horse power. The full-line curves show the performance when the throttle is wide open and the engine is loaded until it assumes the desired speed. The broken lines show the performance when the engine had the propeller load only; in this case the engine speed can be reduced below 1,400 r.p.m. only by partly closing the throttle; speeds above 1,400 r.p.m. are not possible with the propeller load. Figure 143 is for a rotary-cylinder engine. In this case the effective horse power is the indicated horse power minus the engine friction loss and minus the windage loss; the rapid increase of windage loss when the engine speed increases makes the net horse power a maximum at about 1,250 r.p.m. Correction to Standard Atmospheric Conditions. The pub- lished results of engine tests may give either the actual horse powers observed or these horse powers corrected to some standard atmospheric condition. The latter is much preferable as it will permit an immediate comparison of engine performances. The best measure of capacity is the brake m.e.p. That engine which has the maximum m.e.p. is developing a horse power 'on the smallest piston displacement. But to make such direct com- parison the operating conditions must be the same or else the results must be corrected to allow for differences. Such correc- tions can be readily applied to differences in atmospheric pressure and temperature. It has been commonly assumed that the horse power developed is proportional to the density of the atmosphere. The density is proportional to the atmospheric pressure, and inversely proportional to the absolute temperature. If the standard conditions are 14.7 Ib. pressure (29.92 in. of ENGINE EFFICIENCIES AND CAPACITIES 33 mercury) and 32F., and the observed horse power is P at pressure p and temperature t, the corrected horse power is 14.7 460 + t P C =P P 492 In most of the published tests the correction to standard conditions has been made by use of this equation. Tests at the Bureau of Standards indicate, however, that the temperature correction in this equation is excessive and that more accurate results are obtained from the equation 14.7 920-M p 952 The m.e.p. is corrected in exactly the same manner. There is no method for directly comparing two engines which are using different fuels or mixtures of different strengths Influence of Strength of Mixture on Capacity and Efficiency. The effect of strength of mixture has been investigated by Berry 1 on automobile engines; his results are supported by the investigation of others on both automobile and aviation engines (see p. 260). For any constant engine speed and con- stant throttle opening, they show (Fig. 19) that the max- imum power is obtained with a comparatively rich mixture, and that for maximum effici- ency a weaker mixture must be used. As the throttle is closed the mixture for maxi- mum efficiency (Fig. 21) be- comes richer and at the lowest loads coincides with that for maximum power. The speed of the engine has no appreci- able influence on the variation of engine power with strength of mixture (Fig. 22). Maximum power (Fig. 20) is obtained with 0.08 Ib. of gasoline per pound of air, or 12^ lb. of air per pound of gasoline. Maximum efficiency is obtained with a mixture of 15 to 16 lb. of air per pound of gasoline at full load, but this 1 Trans. Am. Soc. M^Ji. Eng., 1919. 3 0.07 0.08 0.09 0.10 O.I I 0.12 0.13 Pounds of Gasoline per Pound of Air in Mixture FIG. 19. Variation of power and thermal efficiency with strength of mix- ture, at full throttle. 34 AIRPLANE ENGINE mixture must be made richer as the load diminishes and becomes Ib. of air per pound of gasoline at lowest loads. 0.05 0.06 0.07 0.08 009 0.10 0.11 0.12 0.13 0.05 0.06 0.01 0.08 0.09 0.10 0.11 0.2 0.13 Pounds of Gasoline per Pound of Air in Mixture FIG. 20. FIG. 21. FIG. 20. Variation of engine power with strength of mixture at constant engine torque and varying speed. FIG. 21. Variation of thermal efficiency with strength of mixture at constant engine torque and varying speed. 0.05 06 0.07 tt08 0.09 0.10 0.11 0.12 0.13 founds of Gasoline per Fbund of Air in Mixture 20 14 12 10 8 Ratio of Air to Fuel by Weight FIG. 23. FIG. 22. FIG. 22. Variation of engine power with strength of mixture at constant engine speed and varying torque. FIG. 23. Maximum thermal efficiencies of certain fuels with varying strength of mixture. Tests by Watson 1 on an automobile engine with gasoline, benzol, and wood alcohol as fuel show (Fig. 23) the variation in 1 Proc. Inst. Aut. Eng., 1914. ENGINE EFFICIENCIES AND CAPACITIES 35 efficiency of these fuels with strength of mixture. The com- parison given by these curves is not, however, complete, since the same compression ratio was used for all three fuels. With alcohol it is possible to increase the compression pressure con- siderably without danger of preignition and without producing excessive explosive pressures; with benzol the compression ratio can similarly be increased slightly. The efficiency with alcohol could probably be raised to at least 35 per cent by the use of a higher compression ratio. Influence of Air Temperature on Capacity and Efficiency. As already pointed out in the discussion of volumetric efficiency (p. 27) the temperature of the air admitted to the cylinder has 005 0.06 0.07 0.08 OX)9 0.10 OJ! 0.12 0.13 05 0.06 0.07 0.08 0.09 0.10 0.11 0.12 0.13 Pounds of Gasolene per Pound of Air in Mixture FIG. 24. FIG. 25. FIG. 24. Variation of engine power with strength of mixture at various air temperatures and full throttle. FIG. 25. Variation of thermal efficiency with strength of mixture at various air temperatures and full throttle. a considerable influence on the power developed. The tests at the Bureau of Standards show the diminution in power with increase in temperature to be proportional to half the increase in absolute temperatures, for the range of temperature from 4 to 120F. For example, if the absolute temperature of the air increases from 500 to 580, or 16 per cent, the power of the engine will decrease 8 per cent. Berry's tests on automobile engines (Figs. 24 and 25) show that this law does not hold for a higher temperature range. The decrease in power is roughly proportional to one-third the increase in temperature (Fig. 24). The efficiency (Fig. 25) is also seen to diminish with increase of air temperature but through a much smaller range. Tests by Berry with air at a temperature lower than 80F. 36 THE AIRPLANE ENGINE showed a rapid falling off in capacity and efficiency. These tests, however, were carried out with a commercial gasoline of low volatility as compared with the gasolines specified for airplane engines. In all cases maximum power is obtained with the lowest air temperature which will permit satisfactory dis- tribution and vaporization of the fuel. This temperature depends not only on the volatility of the fuel but also upon the manifold design. A" hot spot " between the carburetor and mani- fold (heated by the exhaust gases), on which the liquid spray from the carburetor impinges, causes vaporization of part of the fuel without heating up the air appreciably and is found to result in a better distribution of the mixture to the different cylinders and in improved engine operation when the air supply is cold. With this device it is possible to lower the temperature range of the entering air somewhat without a falling off in capacity or efficiency. Influence of Throttling on Efficiency. It is generally found that thermal efficiency tends to increase as the power is cut down 0.65 J 7: 0.60 0.55 0.50 0.45 half oad "800 1000 1200 1400 1600 1800 FIG. 26. Variation of specific fuel consumption with engine speed at various throttle positions. from maximum to three-fourths load. This phenomenon is prob- ably due mainly to improvement in the mixture at partial load. The carburetor is set to give maximum power for full throttle, which, as just shown, is obtained with an over-rich and conse- quently inefficient mixture. If the mixture becomes leaner at partial throttle, the economy will improve. What actually happens will depend primarily on the characteristics of the carburetor used. ENGINE EFFICIENCIES AND CAPACITIES 37 In Fig. 26 are given the fuel consumptions per brake-horse- power hour of a Liberty 12, at full, three-fourths and one-half loads. It will be seen that the efficiencies at full and three- fourths loads are substantially the same, but that there is a marked falling off at half load. The efficiency is again seen to increase with engine speed. Influence of Compression Ratio on Capacity. Tests of a 150- h.p. Hispano-Suiza engine at 1,500 r.p.m. with various compres- sion ratios show the maximum attainable brake horse power to have been as follows: Ratio of compression , . 4.7 5.3 6.2 Maximum brake horse power . . 160 . 165 . 169 . The percentage increase of power with increase of compression ratio is about the same as in the tests of a Liberty 12 engine, which at 1,600 r.p.m. gives the following results: Ratio of compression 4.9 5.5 % Maximum brake horse power 380.0 398.0 The Influence of Revolutions per Minute on Capacity is shown in all performance curves (see Figs. 48, 50, 53, etc.). The fall in 1000 1200 1400 1600 1800 2000 2200 2400 Revolutions per Minute of Engine FIG. 27. Variation of mean effective pressure with engine speed for airplane engines. brake m.e.p. (Fig. 27) causes the b.h.p. to go through a maxi- mum; the efficiency practically always increases with speed, but becomes a maximum before the engine reaches maximum horse power. Influence of Jacket-water Temperature on Capacity. Figure 28 shows the effect on the capacity of a Liberty 12 engine of varying the temperature of the jacket water. The amount of water circulated was constant at any given speed but the inlet temperature was varied, thereby giving the series of outlet 38 THE AIRPLANE ENGINE temperatures indicated. It will be seen that the power increases as the cooling-water temperature decreases to about 100F. At 200F. and 1,800 r.p.m. the power is only 417 h.p. while at 90 it is 436 h.p. 420 400 u |380 (X 4> 29. Typical heat balance of airplane engine. dicated horse power. The increase is primarily due to improve- ment in volumetric efficiency resulting from less heating up of the charge as it enters. This phenomenon is shown in Fig. 15. Heat Balance. Of the total heat of combustion of the fuel admitted to the engine cylinder, part is converted into brake ENGINE EFFICIENCIES AND CAPACITIES 39 horse power, part goes to the water jacket, part escapes as heat in the exhaust gases, and the rest is lost in various ways such as incomplete combustion and radiation and conduction from the engine. A typical heat balance for the Liberty 12 is shown in Fig. 29 in which the percentages are in terms of the higher heat value of the fuel. It will be observed that the heat distribution does not change notably with engine speed. CHAPTER III ENGINE DYNAMICS Turning Moment. The pressure on any single piston of a four-cycle engine is varying continuously throughout the cycle. t>vv 40 g \ k \ ^300 JS _J onn \ \ 1_ CV.(J * 100 X \ \, \ ---. ^ ^^ " . ' s ~~ = == - *ai 234567 Piston Travel, In. FIG. 30. Indicator card of the Liberty-12 engine. If the indicator card is as in Fig. 30, the resulting total gas pressure on the piston of a 5-in. diameter, 7-in. stroke engine for various 5E A-TohtI6a>. iPressu re on Piston 7000 i y B- Inertia Forces of 1700 Rpm. - Resultant Pressure along Cy It Hkfi x/s . 5000 3 4000 T 3000 2000 \ u p\ ^ . ^ A > 1000 * -1000 -2000 V <- ( ^ ^ / / ^ 'S / 1 ? B >/ \ / X ^r / s V / r B>\ ^i * Crank Angle Decj rees FIG. 31. Forces acting on the crank pin of the Liberty-12 engine. successive crank positions is represented by the curve A, Fig. 31. The pressure transmitted to the crankpin is modified, however, by the inertia of the reciprocating masses of the piston 40 ENGINE DYNAMICS 41 and connecting rod. During the first part of each stroke these masses are being accelerated; during the second part they are retarded. Hence the net useful force acting on the crankpin in a direction parallel to the cylinder axis is alternately less or greater than that shown by the curve A . Let W = Weight of reciprocating parts, pounds (complete piston and half of rod). n = Revo'utions per minute. a = Angle turned through by crank, starting from its uppermost position, degrees. r = Crank radius, feet (half the stroke of the engine). I = Length of connecting rod (center to center of pins), feet. Then the accelerating force at any moment, in pounds, is P a = 0.00034 Wn 2 r(cos a cos 2a), approximately, the + sign being used for the down stroke and the sign for the up 7* -^cos2a) i 7* stroke. The quantity (cos a -cos2a) is an approximation; a 2a + sin 4 a more correct expression is cos a - Values of this quantity for the range of ratios of I to r common in airplane engines are given in Table 1. The minus sign indicates negative acceleration from deg. to 180 deg., and positive acceleration from 180 deg. to 360 deg. Calculations made for the Liberty-12 engine give the results shown in Fig. 31. The indicator card (Fig. 30) is plotted for 18 per cent clearance and a brake m.e.p. of 123 Ib. per square inch, the exponents of the compression and expansion curves being taken as 1.32. The cylinder is 5 by 7 in., the connecting rod 12 in. long and the weights of reciprocating parts are: piston complete with pin, 4.838 Ib.; upper half of connecting rod, 1.225 Ib.; total, 6.063 Ib. The engine is assumed to make 1,700 r.p.m. The inertia forces. P. calculated from the preceding equation, are plotted as curve B, Fig. 31. The algebraic sum of gas pressures A and inertia pressures B is shown by the resultant pressure curve C. 42 THE AIRPLANE ENGINE TABLE 1. INERTIA FACTORS -^cos2a + sin 4 a cos a Crank angle, degrees / = 4 r '--3.75 r '-=3.5 r '- = 3.25 r - = 3.0 r Crank angle, degrees .2500 .2667 .2857 .3077 1 . 3333 360 5 .2426 .2590 .2778 .2995 1 . 3249 355 10 .2204 .2362 .2543 .2752 1 . 2997 350 15 .1839 .1986 .2155 .2351 1 . 2580 345 20 .1335 .1468 .1621 .1798 1 . 2006 - 340 25 .0702 .0817 .0948 .1102 1.1283 335 30 0.9950 1.0042 1.0149 1 . 0274 1 . 0423 330 35 0.9091 0.9158 0.9236 0.9328 0.9440 325 40 0.8140 0.8179 0.8225 0.8281 0.8349 320 45 0.7112 0.7121 0.7133 0.7149 0.7172 315 50 0.6026 0.6004 0.5980 0.5955 0.5929 310 55 0.4899 0.4846 0.4787 0.4719 0.4643 305 60 0.3751 0.3668 0.3573 0.3465 0.3338 300 65 0.2601 0.2490 0.2363 0.2215 0.2041 295 70 0,1468 0.1332 0.1175 0.0992 0.0776 290 75 0.0368 0.0211 0.0030 -0.0182 -0.0434 285 80 -0.0682 -0.0854 -0.1055 -0.1288 -0.1567 280 85 -0.1669 -0.1851 -0.2062 -0.2309 -0.2605 275 90 -0.2582 -0.2767 -0.2981 -0.3234 -0.3536 270 95 -0.3412 -0.3594 -0.3805 -0.4052 -0.4348 265 100 -0.4155 -0.4327 -0.4528 -0.4761 -0.5040 260 105 -0.4809 -0.4965 -0.5146 -0.5358 -0.5610 255 110 -0.5373 -0.5509 -0.5665 -0.5848 -0.6064 250 115 -0.5851 -0.5962 -0.6090 -0.6237 -0.6411 245 120 -0.6249 -0.6332 -0.6427 -0.6535 -0.6662 240 125 -0.6573 -0.6625 -0.6685 -0.6752 -0.6829 235 130 -0.6830 -0.6852 -0.6875 -0.6901 -0.6927 230 [135 -0.7030 -0.7021 -0.7009 -0.6993 -0.6970 225 140 -0.7181 -0.7142 -0.7096 -0.7040 -0.6972 220 145 -0.7292 -0.7225 -0.7137 -0.7055 -0.6944 215 150 -0.7370 -0.7279 -0.7172 -0.7047 -0.6898 210 155 -0.7423 -0.7310 -0.7178 -0.7025 -0.6843 205 160 -0.7459 -0.7326 -0.7173 -0.6997 -0.6788 200 165 -0.7480 -0.7333 -0.7163 -0.6968 -0.6738 195 170 -0.7492 -0.7334 -0.7153 -0.6944 -0.6700 190 175 -0.7498 -0.7334 -0.7146 l -0.6929 -0.6675 185 180 -0.7500 -0.7333 -0.7143 -0.6923 -0.6667 180 ENGINE DYNAMICS 43 The resultant pressure, P, curve C, acts along the axis of the cylinder. The force acting along the connecting rod, Fig. 32, is P E = P + COS 6. The component acting tangentially to the crankpin circle is PQ = PE sin (a + b) = P sec b sin (a + b)- Table 2 gives values of the tangential factor [sec b sin (a + &)] The angles a and 6 are connected by the equation sin b = j sin a. Consequently, n . /., PO = P sin a (1 H The torque or turning moment applied to the crank at any crank angle a is T = P Q r Figure 33 shows the torque variation for a single cylinder of the Liberty 12 engine. FIG. 32. Diagram Since the brake m.e.p. is 123 lb., the engine showing effect of obiiq- F uity of connecting rod. horse power per cylinder is 123 X ^2 X (~ X 5 2 ) X 850 33,000 = 36.3 The mean torque at the propeller must lead to the same result: 2irnT = 36.3 X 33,000, or, T = 112; that is, the mean torque per cylinder is 112 Ib.-ft. The mean torque at the crankpin as determined from Fig. 33 is greater than this by the torque required to overcome the frictional resistance of one cylinder, or one-twelfth of the total frictional torque of the engine. If the mechanical efficiency of the engine is 85 per cent, the indicated mean effective pressure is lo %& X 123 lb. per square inch and the mean torque per cylinder is 10 % 5 X 112 Ib.-ft. The total horse power for the 12-cylinder engine is 12 X 36.3 = 436 and the mean total crankshaft torque is 12 X 112 = 1,345 Ib.-ft. 44 THE AIRPLANE ENGINE TABLE 2. TANGENTIAL FACTORS sin (a + b) cos b Crank angle, degrees I - = 4 r l - = 3.75 r i-j - = 3.25 r L-. r Crank angle, degrees 0.0000 0.0000 0.0000 0.0000 0.0000 360 5 0.1089 0.1103 0.1119 0.1139 0.1161 355 10 0.2164 0.2193 0.2226 0.2264 0.2307 350 15 0.3214 0.3257 0.3305 0.3360 0.3425 345 20 0.4227 0.4281 0.4343 0.4415 0.4499 340 25 0.5189 0.5254 0.5329 0.5415 0.5515 335 30 0.6091 0.6165 0.6250 0.6314 0.6464 330 35 0.6923 0.7003 0.7098 0.7206 0.7333 325 40 0.7675 0.7761 0.7860 0.7974 0.8108 320 45 0.8340 0.8429 0.8529 0.8647 0.8786 315 50 0.8914 0.9001 0.9101 0.9219 0.9358 310 55 0.9391 0.9475 0.9572 0.9685 0.9819 305 60 0.9770 0.9847 0.9938 .0041 1.0167 300 65 .0046 .0116 1.0195 .0290 1.0401 295 70 .0223 .0282 1 . 0352 .0430 1.0524 290 75 .0303 .0349 1.0401 .0464 1.0539 285 80 .0290 .0320 1.0356 .0399 1.0452 280 85 .0186 .0202 1.0221 .0242 1 . 0268 275 90 .0000 .0000 1.0000 1.0000 1.0000 270 95 0.9739 0.9723 0.9703 0.9680 0.9656 265 100 0.9408 0.9376 0.9339 0.9296 0.9245 260 105 0.9016 0.8970 0.8916 0.8853 0.8780 255 110 0.8570 0.8511 0.8443 0.8364 0.8268 250 115 0.8082 0.8011 0.7930 0.7836 0.7723 245 120 0.7551 0.7473 0.7384 0.7278 0.7153 240 125 0.6989 0.6907 0.6811 0.6697 0.6563 235 130 0.6406 0.6320 0.6219 0.6102 0.5962 230 135 0.5801 0.5713 0.5613 0.5495 0.5356 225 140 0.5181 0.5094 0.4997 0.4882 0.4748 220 145 0.4549 0.4468 0.4375 0.4267 0.4140 215 150 0.3908 0.3835 0.3750 0.3652 0.3536 210 155 0.3263 0.3198 0.3124 0.3038 0.2936 205 160 0.2614 0.2559 0.2498 0.2425 0.2339 200 165 0.1962 0.1920 0.1872 0.1817 0.1751 195 170 0.1309 0.1280 0.1247 0.1209 0.1166 190 175 0.0654 0.0640 0.0624 0.0604 0.0582 185 180 0.0000 0.0000 0.0000 0.0000 0.0000 180 With the firing order used on the Liberty engine and with a Vee angle of 45 deg., the firing intervals between two cylinders on any one crank are 315 and 405 deg. of crank revolution. The turning moment on any one crank is obtained by superimposing ENGINE DYNAMICS 45 IUW \ \ oUU \ \ 5 60 -- | 400 c^?00 * . -200 -400 ( FIG. 33. Tu \ \ \ r, \ a r\ Me an \Jo rqu > i \ \ \ \ r\ \ \ 1?0 \ / \ \ I ^ ? \ 1 \ J \ J ) 100 200 300 400 500 600 700 Crank Angle, Degrees riling moment for a single cylinder of the Liberty-12 engine 100 200 300 400 500 600 700 Crank Angle, Degrees FIG. 34. Turning moment on each crank of the Liberty-12 engine. 46 THE AIRPLANE ENGINE two torque curves like those of Fig. 33 with a phase lag of 315 deg. Adding the ordinates of two such curves, as in Fig. 34, gives the total torque on one crank. The six cranks of this engine are spaced at angular intervals of 120 deg. The total torque on the crankshaft is obtained by superimposing six curves like the resultant curve of Fig. 34, with angular intervals of 120 deg., and taking the algebraic sums of ordinates at the various crank positions. This process gives the curve of Fig. 35. The torques and torque ratios are as follows: ONE ONE CYLINDER CRANK WHOLE ENGINE Maximum crankshaft torque, pound-feet 1 , 030 1 , 240 1 , 670 Ratio of maximum to mean torque 9.2 5 . 54 1 . 24 100 200 600 700 300 400 500 Crank Angle, Degrees FIG. 35. Torque at propeller end of the crankshaft of the Liberty-12 engine. The ratio of maximum to mean torque varies with the angle of the Vee. For 5- by 7-in. cylinders at 120 Ib. mean effective pressure, the following ratios hold for torques on one crank: T s . 45 DEO. 5.2 -2.7 7.9 60 DEG. 5.1 -2.3 7.4 75 DEG. 5.6 -1.7 7.4 90 DEO. 4.8 -1.5 6.3 Here = maximum torque -r- mean torque, Tz = minimum torque -f- mean torque, TZ = range of torque -f- mean torque. For the whole engine, ratios are as follows: 8-cylinder. 12-cylinder. VEE ANGLE 90 75 60 45 60 45 Ti 1.40 1.42 1.70 2.14 1.13 1.25 0.66 0.18 -0.13 -0.26 0.86 0.89 ENGINE DYNAMICS 47 In both tables negative signs indicate reversal of direction of tor- que. The minimum torque variation is seen to occur with equal firing intervals (eight-cylinder, 90-deg., and 12-cylinder, 60-deg. Vee engines). The great increase in this variation as the Vee angle of the eight-cylinder engine is diminished is very marked. A smooth curve of crankshaft turning moment, approximating as closely as possible to the mean torque line, is in every way desirable. This can best be obtained by the use of a plurality of cylinders with equal firing intervals. The greater the number of cylinders, the more uniform is the torque. The firing order of the cylinders is unimportant from this standpoint, as long as the firing interval is constant. The firing order is of the utmost importance, however, in relation to the balancing of forces and the stresses in the engine. Smoothness of running depends on the magnitude of the areas enclosed between the total torque curve and the mean torque line (Fig. 35). Areas above the line represent work done by the engine in excess of the resisting pro- peller torque and lead to acceleration; areas below the line represent a deficiency in engine work and a consequent slowing down. In ordinary engine practice a flywheel is used to absorb the excess and make up the deficiency without permitting excessive change in engine speed. In an airplane engine the propeller takes the place of a flywheel; its large radius of gyration enables it to absorb a considerable amount of excess work with only a very small increase in speed of rotation. Moreover, the resisting torque varies as the square of the angular velocity and hence increases notably with small increases in rotative speed. The influence of the number of cylinders on the variation of crankshaft torque is clearly shown in the following table. 1 The firing interval is constant in all cases. TABLE OF TORQUE VARIATION Number of cylinders. . . 1 2 3 4 5 6 7 8 9 10 12 16 18 Ratio of maximum in- stantaneous torque to mean torque 7.70 5.20 2.74 2.94 1.64 1.17 1.45 1.40 1.22 1.12 1.13 1.06 1.03 Relative values of the maximum torque at propeller 1.0 1.35 1.07 1.53 1.06 0.91 1.32 1.45 1.43 1.46 1.76 2.20 2.41 The relative values of the maximum torque are also given in the table, the value for a single cylinder being taken as unity. 1 From article by G. D. ANGLE, Aviation, Oct. 1, 1919. 48 THE AIRPLANE ENGINE It will be seen that the maximum torque at the propeller end of a 6-cylinder engine is less than the maximum torque exerted by a IWU 800 600 : / **-* ^ CQ 4000 ~ r [s 3000 '. 2000 S. 1000 t ^ 1000 .8 2000 ^ 3000^ 4000 / 1 r \ \ f / \ \i \v i \-jrt 1 f ^ i v, " 1 \ 1 / \ / ^ / / \ 1 \ / / -400 V J / ^. vv N ^ 1 x f ' s x / 1 : 300 10 18 } 360 540 720 Crank Angle, Degrees FIG. 36. Side thrust against the cylinder walls of the Liberty-12 engine. single cylinder and consequently the crankshaft must be as strong at the free end as at the propeller end. Side Thrust. Side thrust of the piston against the cylinder wall exists in con- sequence of the obliquity of the con- necting rod; it disappears at the two dead centers. Its magnitude, G (Fig. 32), is given by G = P tan 6. Since r sin a = I sin 6, G _ r sin a P ~ \/Z 2 -r 2 sm 2 a For the Liberty engine with indicator diagram, as in Fig. 30, the side thrust is as shown in Fig. 36, the maximum value reaching nearly 1,000 Ib. Change of sign indicates change of thrust from one side of the cylinder to the other. In sta- tionary cylinder engines side thrust is important only in relation to frictional wear. Offset Cylinders.-The side thrust during the expansion stroke, and the friction loss and ^ter wear, may be reduced by offsetting the cylinder: that is, by so locating it that its center line does not pass through the axis of the crankshaft. Minor effects of this arrangement are that FIG. 37. Diagram show- ing the effects of the obli- ENGINE DYNAMICS 49 the piston stroke is slightly greater than twice the crank throw and the mean speed of the piston is greater during the down stroke than the up stroke. This last point has the advantage of reducing the heat loss to the cylinder walls during the expansion period. In Fig. 37, if k = offset and x = distance from a point on the piston to a horizontal plane through the crankshaft axis (in a vertical engine), then x = r cos 6 + I cos a dx da -^r r sin I sin a -=, do bd d 2 x d 2 a , /da\ 2 = r cos o I sin a ~rr^ * I ( ~JT ) cos a db db \do/ . /r sin b k\ a = sin- 1 1 -j 1 da r cos b db I sin a I cos a-r sin b + r cos b-l sin a -j d 2 a _ dp db 2 = I 2 cos 2 a d z x , , r 2 cos 2 6 - r cos b I cos a ^ ^~ db 2 I 2 cos 2 a , , . da I cos a-r sin b + r cos b-l sin a--^- sn a cos a r 2 cos 2 6 , r 2 cos 2 6-tan 2 a = r cos o -- j r T sin o- tan a -- I cos a L cos a , r 2 cos 2 b ,., x . , , = r cos 6 T - (1 + tan 2 a) + r sm D-tan a. Z cos a v The acceleration of the reciprocating parts is equal to /27rn\ 2 d*x (w) * and the accelerating force is The side thrust is given by G = P tan a. In using the above equations it should be noted that the angle a is positive when the connecting rod swings away from the crankshaft, as in the position shown in Fig. 37, and becomes 50 THE AIRPLANE ENGINE negative on the other side of the vertical. The angle a is found for any value of b from the equation I sin a = r sin b k 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 560 Degrees Rotation of Crank from Top Dead Center FIG. 38. Effects of different degrees of offset on the side thrust in a single cylinder of the dimensions of the Liberty-12 engine. The results of an analysis of the Liberty engine with offsets of 0.5, 1.0 and 1.75 in. gives side thrusts as shown in Fig. 38. It will be seen that with an offset of half the crank throw (1.75 in.) the side thrusts during the exhaust and compression strokes are nearly as high as the maximum value reached in an engine with- out offset. The lowest maximum is obtained with an offset of 1 in. Rotary Engines. The turning moment in a rotary engine results entirely from side thrust on the cylinder walls. This thrust is due not only to the obliquity of the connecting rod, as with stationary- FIG. 39. Diagram of rotary engine. T j i A i J.T, cylinder engines, but also to thrust resulting from tangential acceleration of the reciprocating parts. The radial and tangential accelerations of these parts and the inertia forces resulting from them may be determined as follows : Take a single cylinder, as in Fig. 39, rotating about the shaft 0, while the connecting rod rotates about the fixed crankpin P. ENGINE DYNAMICS 51 The piston pin Q will move along the axis OX as that axis rotates about Q. If the length OQ = x, the point Q will undergo radial acceleration a R along OQ and also tangential acceleration a T at right angles to OQ. The magnitudes of these accelerations are given by the general theorems : da 2 and _ dx da d*a aT " **& / rod) and the connecting -sooo-J rod might be replaced by FlQ - 41. Forces acting along the cylinder , . , . axis of a rotary engine. a chain except that it is under compression before the engine has attained its full speed. The turning moment for a seven-cylinder Gnome engine of the dimensions given above is shown in Fig. 42. The maximum excess of power developed over the mean resistance (shaded 720 !&0 540 360 Degrees FIG. 42. Turning moment of a 7-cylinder Gnome engine. area) is 60 ft.-lb., or about Ml 20 of the total kinetic energy of the engine, and would be a much smaller fraction of the kinetic energy of engine and propeller combined. The earlier Gnome engines governed by cutting out the explosion on one or more of the cylinders according to the power requirements. The cutting 54 THE AIRPLANE ENGINE out of a cylinder has a serious effect on the uniformity of turning moment and results in a maximum deficiency of work done during the cycle of 280 ft.-lb. as compared with the maximum excess (or deficiency) of 60 ft.-lb. when all cylinders are functioning. The resultant force on the crankpin of the above engine at full load varies from 2,640 to 800 Ib. BALANCING The forces acting in an engine are of two kinds, in so far as their effect on stresses is concerned. The gas pressure in the cylinder, being exerted both on the cylinder head and on the piston, subjects the engine to equal and opposite forces which are inherently balanced, or, in other words, has no effect in displacing the engine as a whole. The inertia forces of the moving part in each cylinder are not inherently balanced. If the engine is to operate smoothly (without vibration) it must be so designed as to make these unbalanced forces counteract one another as far as possible. There are two kinds of moving parts to be considered: (a) the rotating parts, and (6) the reciprocating parts. Rotating Parts.*- A body of mass m (weight W = mg) revolving with angular velocity co radians per second (n r.p.m.) in a circle of radius r feet has an acceleration of co 2 r which gives rise to a centrifugal force F = mco 2 r = 0.00034 Wn 2 r Ib. acting radially. The rotating body is usually composed of the crank, crankpin, and the large end of the connecting rod. The resulting force F is an unbalanced force acting on the crankshaft. In multicrank engines there will always be a number of these forces, one acting at each crank. If there is more than one connecting rod attached to the crank (as in Vee and fixed radial engines) the revolving mass includes the large ends of all the connecting rods attached to the crank. In rotating cylinder engines the revolving mass includes the cylinders themselves but not the crank and crankpin. It is easy to balance the rotating parts. To accomplish this it is necessary (1) that the vector sum of these unbalanced forces should be zero and (2) that the sum of their couples about any (arbitrarily selected) plane should be zero. As the centrifugal forces on all the cranks are equal, the condition (1) is met, with any number of cylinders greater than one, by making the crank ENGINE DYNAMICS 55 angle intervals equal. Condition (2) can be met, with any num- ber of cylinders greater than two, if the cranks are spaced at equal distances apart along the shaft. There is an unbalanced centrifugal pressure on the crankpin due to the inertia of the big ends of the connecting rods, and on the main bearings due to the inertia of all the rotating parts. Reciprocating Parts.-K-The inertia force of the reciprocating parts has already been seen to be P a = 0.00034 Wn*r (cos a ^-cos If the connecting rod were infinitely long the expression would be P al = 0.00034 Wn 2 r-cos a. The second term in the brackets is due to the obliquity of the connecting rod and increases in value as I becomes shorter. It is customary to separate the two terms in a discussion of balancing, the quantity P al being called the primary inertia force, while Pan = 0.00034 Wn*j- cos 2a is called the secondary inertia force. It is found that the balance of the primary forces is much more easily achieved than that of the secondary forces. Since j is usually about J^, the magnitude of the secondary forces is only about one-quarter that of the primary forces so that secondary balance is not as important as primary balance. The conditions for balance of primary and secondary inertia forces are exactly the same as those for centrifugal forces. There is, however, this difference, that these forces act always in the direction of the line of stroke. The practice of balancing the primary forces by the use of counterbalancing masses attached to the crankshaft is inadmissible (where avoidable) because of the necessity of keeping the total weight as low as possible and also because the same result can be obtained by multicylinder construction. The following general results of an analysis of various possible cylinder arrangements may be stated, it being assumed that the reciprocating masses are the same in all cylinders: With cylinders in a line and equally spaced at intervals of Lft.: 56 THE AIRPLANE ENGINE For three-crank engines with cranks at 120 deg., the primary and secondary forces are balanced, but the primary and secondary couples are not. The maximum unbalanced primary couple is \/3'mco 2 rL. For four-crank engines with cranks at 180 deg., the primary forces are balanced but the secondary forces are not. The secondary forces in a vertical engine have a vertical resultant r 2 which is equal to 4mco 2 y- at each quarter turn (0 deg., 90 deg., 180 deg., etc.) and become zero at each eighth turn (45 deg., 135 deg., etc.). For five-crank engines with cranks at 72 deg., the primary and secondary forces are balanced, but the couples are not. The maximum unbalanced primary couple is 2.5 mco 2 rL; the maximum r 2 unbalanced secondary couple is 5mco 2 y- L. For six-crank engines with cranks at 120 deg., the primary and secondary forces and couples are all balanced. With opposed cylinder engines, with two cylinders and cranks at 180 deg., the primary and secondary forces are balanced; the primary couple is unbalanced with a maximum value of ma>VL. With Vee engines : For six cylinders, angle of Vee 120 deg., three cranks at 120 deg., the primary and secondary forces are balanced; the couples are unbalanced; maximum unbalanced primary couple is For eight cylinders, angle of Vee 90 deg., four cranks at 180, primary forces balanced, secondary forces unbalanced; primary and secondary couples balanced. Unbalanced secondary force acts always at right angles to the longitudinal plane of symmetry r 2 and has a maximum value of 5.6 mco 2 y For twelve cylinders, angle of Vee 60 deg., six cranks at 120 deg., primary and secondary forces and couples all balanced. With fixed radial engine of k cylinders with all the connecting rods on one crank, the inertia forces have a constant resultant k of ^wcoV, approximately, along the crank. It can be balanced ~L*yY} completely by a counterbalance mass of - at radius r, opposite ENGINE DYNAMICS 57 the crankpin. If there are two banks of cylinders, each with k cylinders and with cranks at 180 deg., the primary forces will balance but there will remain an unbalanced primary couple of <-mco 2 rL. z Rotating-cylinder engines have two sets of rotating masses: 1. The cylinders, which are perfectly balanced, and 2. The pistons and connecting rods, which have primary balance but not secondary balance. PERIODIC UNBALANCED FORCES V The results of the existence of periodic unbalanced forces in an airplane engine may be two-fold: (1) to move the engine as a whole, and (2) to distort the engine. The engine in an airplane is supported on wooden members which are flexible and are in no way the equivalent of a rigid foundation. If the engine moves as a whole it will flex its supports instead of moving the airplane as a whole. The maximum possible duration of the periodic variation of any unbalanced force is one engine revolution or about ^5 sec.; ordinarily it will be not more than one-quarter of a revolution or Hoo se c. This time is too short to permit the unbalanced force to produce appreciable deviation of the airplane as a whole although it may set up vibration in it. The crankcase of the airplane engine is always very thin and consequently flexible. Much of the unbalanced force may be taken up in producing distortion of the crankcase. This is very undesirable since it necessarily results in varying align- ments of the crankshaft bearings and consequent increase in shaft friction. - . Both the engine supports and the crankcase have their own natural periods of vibration. If the frequency of the distur- bances due to unbalanced forces in the engine is the same as the natural frequency of the vibration of either of these members, the amplitude of vibration will be greatly increased beyond the very small amount due to a single application of the unbalanced force. It is most important that the critical speed at which the vibration of the engine on its supports becomes a maximum should be avoided, as well as simple multiples of those speeds. The natural periods of vibration of such complex structures as engines, on their supports in an airplane, can not be calculated. If excessive 58 THE AIRPLANE ENGINE vibration is found at or near the designed speed of the engine the structures must be altered to change the natural period. Various devices have been used for neutralizing the unbalanced secondary forces in four- and eight-cylinder engines. A good example of these is the Lanchester balancer (Fig. 43), which has had considerable application to automobile engines. This consists of two exactly similar unbalanced cylinders located under the center main bearing and driven by a gear on the crankshaft. These cylinders revolve in opposite directions and their unbalanced weights are located so that their common center of gravity travels up and down in a vertical plane and so balances the displacement of the common center of gravity of the pistons, which falls in a plane at the middle of the piston travel when the pistons are on Mid-stroke FIG. 43. Lanchester balancer. their dead centers, but falls below it when the cranks have revolved 90 deg. past that position. It should be clearly recognized that complete balance of primary and secondary forces and couples does not ensure absence of vibration. Such complete balance means that the engine as a whole has no tendency to move, but there are always internal stresses, particularly those imposed by the opposing couples on the engine structure. If the periodicity of the application of these stresses coincides with the natural period of the structure (or some fraction of it) severe vibrations may be set up. In any case heavy bearing loads are likely to be imposed, especially in the center main bearing, through the action of opposing couples. For torsional oscillations of the crankshaft see p. 146. The following table gives values of inertia and centrifugal forces, resulting bearing pressures and other calculated quantities ENGINE DYNAMICS 59 for the six-cylinder 200-h.p. Austro-Daimler engine, the six- cylinder 270-h.p. Basse-Serve engine and the 12-cylinder 400-h.p. Liberty engine: INERTIA FORCES, BEARING LOADS, ETC. Austro- Daimler engine Basse- Selve engine Liberty- 12 engine Weight of piston complete with rings and piston pin, Ib. 4.18 188 6.187 211 3.838 1955 Weight of connecting rod complete Ib ... 4 84 9 00 2 9 and Weight of reciprocating part of connecting rod, Ib Total reciprocating weight per cylinder, Ib 1.66 5 84 2.25 8 437 6.35 1.225 5 063 Weight per sq. in. of piston area, Ib Length of connecting rod (centers), in 0.263 12.40 0.288 14 17 0.258 12 Ratio connecting rod length to crank throw 3.6:1 3.6:1 3.44:1 Inertia, Ib. per sq. in. piston area (top center) 63.8 80.7 117.0 Inertia, Ib. per sq. in. piston area (bottom center) 36.2 25 45.7 31 6 65.2 3 18 6 75 1 675 and Total centrifugal pressure, Ib 610 1 480 5.125 1,469 Centrifugal pressure, Ib. per sq. in. piston area Mean average loading on crankpin bearing, total from all sources, Ib. per sq. in. piston area Diameter of crankpin, in Rubbing velocity, ft. per sec 27.5 91.0 2.20 13.42 50.7 115.7 2.75 16.8 74.8 175.0 2.375 17.7 Effective projected area of big-end bearing, sq. in Ratio piston area to projected area of big-end bearing . Mean average loading on big-end bearing, Ib. per sq. in. 5.02 4.42:1 402.0 5.39 3.5:1 405.0 5.34 3.68:1 642.0 CHAPTER IV ENGINE DIMENSIONS AND ARRANGEMENTS Certain special requirements control the selection of the dimensions and arrangements of airplane engines. These are: 1. Minimum total weight of the engine and its accessories per brake horse power developed, or maximum power output per pound of weight. 2. Maximum fuel economy. 3. Compactness. 4. Freedom from unbalanced forces and from vibration. 5. Reliability. To obtain minimum weight per horse power, it is necessary that the engine should have minimum weight per cubic foot of piston displacement per revolution, and that it should be operated with maximum power per cubic foot of cylinder volume. The latter demands a combination of the maximum obtainable mean effective pressure with high speed of revolution. The mean effective pressure is constant at moderate engine speeds but falls off at high speeds in consequence of falling volumetric efficiency. Beyond a certain limiting speed the mean effective pressure will fall off more rapidly than the increase in engine revolutions per minute (see p. 37) and the engine power will decrease. This limiting speed should be made as high as possible by making the valve openings large and the inlet and exhaust manifolds short and of ample cross-section. All airplane engines must be multicylindered in order to give the necessary uniformity of turning moment and freedom from unbalanced forces. The weight of the engine per cubic foot of piston displacement per revolution will depend on the unit size of cylinder selected. Comparing two unit cylinders of exactly the same form but of different sizes, it will be found that the thickness of the cylinder walls will not have to be increased as rapidly as the cylinder diameter because the wall (for structural reasons) is always made thicker than the stresses demand, by an amount which does not vary much with the diameter. Conse- quently the weight of the cylinder per cubic foot of piston dis- 60 ENGINE DIMENSIONS AND ARRANGEMENTS 61 placement diminishes with increased size, and the same is true of most of the other engine parts. On the other hand, the weight of the engine per cylinder diminishes with increase in the number of cylinders in line. The engine consists of a number of exactly similar units (cylinder, running parts, section of crankcase, etc.) and certain approxi- mately constant weights such as ends of crankcase, pumps, magnetos, propeller hub and so forth. The addition of more cylinders will diminish the weight of the engine per cylinder. A further diminution in weight of the' engine per cylinder can be obtained by using the Vee or W arrangement of cylinders. A substantial saving in the weight of crankshaft and crankcase, per cylinder, results from these arrangements. Still greater saving results from the adoption of the radial arrangement. Considerations of torque and balance (see p. 56) indicate that the number of cylinders should not be less than six for an engine of moderate power (100 to 150 h.p.). For higher powers the choice is between more cylinders and larger cylinders. Engines have been built with as many as 24 cylinders, but it does not seem likely that this number will be much used or exceeded. Cylinder size can be increased by increasing either diameter or stroke or both. The diameter is at present limited to from 6 to 7 in. by the difficulty of keeping the piston cool. The heat given to the center of the piston has to travel radially to the walls, and it is necessary to increase the thickness of the piston as the diameter increases in order to give sufficient section of metal to carry the heat away. This results in a heavy piston and excessive inertia forces in the reciprocating parts. The engine stroke is also limited at present to 8 in.; increase in stroke beyond that limit increases the over-all height of the engine to dimensions which are difficult to accommodate without increasing the size of the fuselage. Furthermore, increase in stroke increases the weight of the engine more than does a corresponding increase in diameter. The small ratio of stroke to diameter which character- izes airplane engines is not as objectionable as it would be in other engines; it increases the ratio of water-jacketed surface to cylinder volume, but the percentage of heat lost to the jacket is nevertheless smaller than in other engines in consequence of the high engine speed and high mean effective pressure. The weight of the engine per cubic foot of piston displacement per revolution is also a function of the ratio of connecting rod 62 THE AIRPLANE ENGINE length to stroke. The smaller this ratio the less is the over-all height and the weight of the engine. The objection to a small ratio is the increase in magnitude of the secondary inertia forces which result from the obliquity of the connecting rod. These secondary forces can be perfectly balanced with certain arrange- ments of cylinders (see p. 56) and the objection eliminated. The ratio usually ranges from 1.5 to 1.7. High fuel economy is important primarily in its effect on the weight to be carried by the plane. For a five-hour flight, an engine weighing 2.5 Ib. per Horse power and using 0.5 Ib. of fuel per horse- power hour will have the same total weight of engine and fuel as an engine weighing 2 Ib. per horse power and using 0.6 Ib. of fuel per horse-power hour. The heavier and more efficient engine would ordinarily be the better of the two in respect to reliability and durability. To obtain high fuel economy the most important factor is the ratio of compression which should be as great as can be used without detonation or preignition. Compactness is important both in respect to frontal area and over-all length. The frontal area of large engines should, if possible, be of such form and dimensions as not to require any in- crease in the cross-section of the containing fuselage. Short over- all length is distinctly advantageous and permits the fuel tanks to be located close to the center of gravity of the plane. Freedom from vibration is necessary because the mounting of the engine is on elastic supports usually wooden longerons which leave the crankcase free to distort under internal forces. Such arrangements of cylinders as will eliminate or minimize unbalanced forces are desirable but they cannot be relied upon to prevent vibration. Even with the most perfect balancing, torsional vibration of the crankshaft may produce excessive vibra- tions at certain engine speeds; such speeds must be avoided. Engine Arrangements. Airplane engines have been built in a great variety of arrangements of which a number have survived. These may be classified as (1) radial, (2) vertical, (3) Vee, (4) W, (5) X. Radial engines (both fixed and rotary) are discussed in Chapter VIII. They have minimum weight per horse power and shortest over-all length, but they have maximum frontal area and until very lately have shown economy inferior to that of the other types. They are generally air cooled. The vertical engine (Figs. 8 and 9) has one row of cylinders. The Vee engine (Figs. 46 and 47) has two linear rows of cylinders; ENGINE DIMENSIONS AND ARRANGEMENTS 63 the " angle of the Vee" is the acute angle between the axial planes of the two rows. The W or \J/ engine (Figs. 72 and 73) has three rows of cylinders of which the central one is vertical and the other two form equal angles with the vertical. The X engine has four rows of cylinders arranged symmetrically about the vertical and horizontal planes but not necessarily with equal angles between the planes of the cylinder axes. In the radial engine (Fig. 136) three or more cylinders, constituting a group, have their axes intersecting at a point on a common shaft. The radial or fixed engine should not be confused with the rotary engine, in which the cylinders revolve about a stationary crankshaft. A radial engine may consist of more than one group or bank of cylinders, each group having its own common plane of cylinder axes, perpendicular to the axis of the shaft. Each cylinder of a four-cycle engine requires two revolutions or 720 deg. of rotation of the crankshaft to complete its cycle. 1234 / 2 3 4 5 6 FIG. 44. Four-cylinder engine FIG. 45. Six-cylinder engine The explosions in a cylinder occur every other revolution. If explosions are to occur in n cylinders at equal intervals, the interval expressed in degrees of crankshaft rotation must be 720 -T- n. Explosion always occurs when the piston is closest to the cylinder head. With a four-cylinder engine, the interval between explosions is 180 deg.; the crankpins lie all in one plane passing through the shaft axis, one possible arrangement being shown in Fig. 44. A six-cylinder vertical engine has the cranks 720 -f- 6 = 120 deg. apart: the crankpins lie in three planes intersecting at the shaft axis. One arrangement is shown by Fig. 45. Constancy of interval between impulses may be obtained with crank dispositions other than those shown in Figs. 44 and 45. For example, in Fig. 44, cranks 2 and 3 may be either in line or opposed. If the former, cranks 1 and 4 will be. in line and 180 deg. away from 2 and 3. If the latter, 1 and 4 will be opposed, and either may be in line with 2. If equal inter- 64 THE AIRPLANE ENGINE vals of time between explosions were the sole requisite, the number of possible crank arrangements with a 6-cylinder engine would be large. The actual crank arrangements are determined mainly by considerations of engine balance. In a Vee engine each pair of cylinders (in one plane) acts on a common crankpin. If there are n cylinders in <-> pairs, the crank interval is 720 -f- ~. The interval between explosions of the cylinders is 720 -5- n. If 6 is the angle of the Vee, any one crank moves through the angle 6 in the interval necessary for the two pistons actuating it to reach respectively their highest positions. As the explosions occur with the pistons in their highest positions, the explosion in a leading cylinder will precede that of its fol- lowing cylinder by an angle of 0, or 360 + 6, deg. according to whether both cylinders explode during the same revolution of the crank or the explosions occur in succeeding revolutions. The latter is always the case, because, if the second explosion occurred after the crank angle 6, the explosion pressure would be trans- mitted to a crankpin which was already subjected to the pressure of the gas expanding in the other cylinder of the pair and the crankpin would be unduly loaded. For equal explosion intervals the angle is equal to 720 -f- n. This leads to a 90-deg. angle for 8-cylinder and a 60-deg. angle for 12-cylinder Vee engines. These angles give the maximum uniformity of turning movement; other angles may be employed for special reasons but always at some sacrifice of uniformity of turning movement. The choice as between vertical, Vee and W arrangement is largely determined by the number of cylinders. It has been found that much trouble is experienced with a crankshaft having more than six cranks and the over-all length of the engine becomes excessive. Eight-crank engines have been built but they have not survived. The perfect balancing of the six-crank engine has made it a favorite. With six cylinders the vertical arrange- ment is almost universally adopted; it has the advantage of having minimum frontal area and consequently of being most easily accommodated in the fuselage. With eight cylinders the 90-deg. Vee engine with four cranks is generally accepted as the best arrangement although the wide Vee angle results in con- siderable over-all engine width. With 12 cylinders the usual arrangement is a 60-deg. Vee and six cranks. The 45-deg. Vee ENGINE DIMENSIONS AND ARRANGEMENTS 65 adopted in the Liberty engine results in decreased width of engine but increased height; it also results in a less uniform turn- ing moment on the crankshaft. The Warrangement for 12 cylinders is with three rows of cylinders and four cranks; this shortens the over-all length and decreases the weight but greatly increases the engine width, especially if a 60-deg. angle (which gives most uniform torque) is used between the rows. With 18 cylinders the W arrangement with three rows of cylinders and six cranks is used; the angle between the rows for most uniform torque should be 40 deg., which diminishes the over-all width as compared with the 12- cylinder W engine. Vertical, Vee and W engines are nearly always water- cooled in airplane practice. Air cooling has been successful only with low compression ratios. Engine Dimensions. Table 3 gives the general dimensions and arrangement of the principal American and foreign airplane engines. The horse powers given are generally the maker's rating, but they are naturally variable with the fuel, the car- buretor, the manifolding, the ratio of compression, and the engine speed. The ratio of compression is readily varied by changing the dimensions of the piston and for certain engines different pistons are supplied, according to whether the engine is to be flown at low altitude (as in seaplanes) or at higher altitudes. The dry weight includes carburetors, magnetos, and propeller hub. The weight per horse power naturally varies with the horse power and is given for the rated horse power. It is the product of two factors, the weight per cubic inch of piston displacement and the piston displacement per horse power. The piston displacement per horse power is the aggregate dis- placement volume of the cylinders per stroke divided by the horse power developed. It is an excellent measure of the degree to which the piston displacement is utilized. The aggregate displacement volume is equal to I X a X n cu. in. where I is the stroke in inches, a the piston area in square inches, and n is the , r , _, p X iXaXnXN number of cylinders. The horse power = 2 X 12 X 33 000 * where p is the mean effective pressure, and N is the revolutions per minute. The piston displacement per horse power = 33,000 X 24 -T- p X N; that is, it depends only on the mean effective pressure and the revolutions per minute. 66 THE AIRPLANE ENGINE iCOO C W OOO (N Oo C OC OOOO <*< 00 00 * O 00 O O O CO CO CO i-H CO OS O5 CO : : : : : : : ::: :^ :::: i r-i-^ : : : : i-* : : : : :^^ i^^ : : : ;OCOCOOOOOt^O O5 CO CO O5 M >H M CO CO IO CO rj* CO CO CO ^H 00 O5 i-l rH CO * CO ococoo--HC t^ OS Oi t >C O5 "5 O >C O5 W t^ 1C >C C O5 C (N 1C C X 1C C^ rjJ 1C 1C <* C C C CO >C "3 <0 ! rjl 1C 1C 1C* 1C CO CO t>-' N.' C CO 5 OOOO-*Tf(iCTj- 1^ i-H rH I-H t^ C >C t^ t^ t^ CO O 3 fi II- O OCOO OOOO OOcN "500OOCOOOOOlNOO(N O O OCO O OO O >C OO CO O O * l> M O O C rH O CO rH C^l rH CO rH * -^ CO O 1! t>- b OS * Tfl GO C5 O5 * b- Oi '- I >O(NJ>O5COO5GiTt-> CO i^ososas'o6ododo5oo' ot~t^ *** co' coco Tt< CO CO CO Tf (N O (N (N OS O \ (N (N OS 'OCO * CO O N CM CO M CO ^tl lH CO CD r-> CM CO COt^'O^'-l . -co -C CM -Tj< OCO : : : : : :OSCMtOOO5COrJ.'coco-HcocOTjHo6 TjJos t^ b- <350000O '~"~' O ^J^^2lH . "Qi."** ^.'rt**' ^ :::::::::: --2 :^ : : : : : : : >, I j :^4; ; j ; ; ; jl|| ; i iilHllll 3.2 S^ ' * 2'a'c tfc 9 M ff i ^a.S ^g Q l H -i o-S G " 68 THE AIRPLANE ENGINE OOOOO OOO^ l C l OOO k OO*O^OO'^O l OOOOtf3 l O | O l OO ^ r-i 1-1 CO CO CD COt>^^C5aoOi-HCOl^rMCCcOCOCOCOO<^iOt>t>l^r^OO I-H - I-H O M o > t, 0) O> OOOOO OOOOOOOOOOOOOOOOOOOOOOOO CO joooo g8oooooooooooooo< III 1--N.COOO OCOOCOOOC3Si-HOO-iN.iOOCOOJTtiCO ,-H r-l CO rjl * CO CO * CO rt< Tj< (N <*< CO CO i-H CO CO O O i-H i-H i-l OiOOOO O CO T(* CO Tf< rf Tj< ^ CO CO CO i-H (No o M o o <-H TH I-H r^ t^ 1-1 co co "5 "5 co *o co t^ r^ T-I co co co * ^' 10' 10 o ui co' TjJ 10' 10' Tti 10 10' CD' co o' co 10' 10' 10" 10' 10' 10' co * i is OOOOO 3OOOcoSa)MCC^J^*3CB63C3C'rt -j-^.S.S.S -(t-if-(^"*f"tj"if' ^-^^ r^ (^ GwClccnocy^cysas^^^cDonEHC^^^ t- 1 : : : ; : a a a rt a cco*r-^*r . > . . |M8f2 :-|j^ I? : : : I : * :J' Hill 1 ENGINE DIMENSIONS AND ARRANGEMENTS 69 CNOO5OO 00 I>CN O O 3 00 r-t O 00 ^ O5 3 O l> CN <-< O5 00 t>- t-l O O O5 3 O300COi-< i-l CO CO t^ IO * CN t^ t- CO "3 t^ CO O O5 I-H t^ t^ 00 Oi CO CO *O O 00 O * iH CO *' O CN *' CO CN CN CO "3* CO CO CO * CO >0 CN ^ CO* CO* CO* .U3r.00 00 00 ^ O i-i "3 I-H O * CN "tf O "3 Oi 00 CO "3 " CO* *' co' CD* O CO CO CO *' CO "3 C "3 U) CO l> * CO *' U3 Tj< 1C co' Tji Tjl -tf' rjl -^ CO "3t^(Nl^"3 COOOOOOt^COO CO O CO 00 00 (N Oi 3 Tj< C35 O 1-1 rftCN CO r-i O COCOi-HOOO 00 O I-H t-i O 00 * CN t- O rt< t^. O5 O> rH 00 CN O * O -^ f^ O O O5 ddddd ddo'dddo'r4ddddddddddo"ddddd d SoSCO 5oSS8coSOCN2l^CNS^HCOCO1c5(N(NCOCN l> "3 O5 Tf( O5 TJ< Tf * O "3 C U3 00 "3 "3 00 CO "3 t^ O 00 CO O5 CO O CO CO 00 "3 O5 "-< U3 . -CO -00 W CN(N -00 O : : : : : : : : : 'do *d *d *d ' -d 'do 'd d ..'... .10 . . .rt< . -CN OSi-HCN -00 CO -t>- 00 -COCOiOOOt^ -O CN I ' '. ' * Ii-i ' ' *CN -O CO cO--.O t^ CO O5 CO CO 00 O5 Oi CO OS O - g CN 00 O * CN O O CN O O O O O 5 O O O ifl O t>- O "3 O N- O O CN 00 O CN O CO 1C t>.' t>* N.' CN CN O CO -* O O CO CO Co' CO 10* t>* U3* N* O* 00 O 00* W5 CO CN O5 I o a ! .a ' ji^_r ' 3 Sai SS ": s !-"s s S 3- s iiiii lilil^ B Jllill1iiilli^ 2 11111 ! 70 THE AIRPLANE ENGINE ENGINE DIMENSIONS AND ARRANGEMENTS 71 CO*HOOCOO'*C O5 O -O -Oh- O5 CMrHOOt^O5>C05OOCO>CCCOO5COCOCMCO rH 00 CM -OO "* -OCO 00 HH CO rH rH CM CM rH CM CO CM CO CO CO rH CM CM CM' CM -COCMCM CM CM Tj< -CMCM rH OCMCOCOCOO5COi-H^t^iCCOCOCOrHTtC 1C O CM iJCCOICCOCOOO .OO^O -gOO -O -OO5 O CO CO CO 00 -^ * CD 00 "3 t^ 1C 1C O5 O5 CO 1C ] rH 00 O -COrHOO T}< 00 t>- OO do 'o'do'o'o'd -odd id -d ' *d d tHNl-I *l-IlHCMCC * ^< ^ C -^ Tfi .CO -CO ^ CO oo -oooooo -ooo -o -o 'd d C5CO>O -OOOOiCOOO OO *& rHCMrH ' rH rH CM* CM l-J rH I CM rH CM '. ' i rH ' i I '. '. '. '. '. i CM I rHrHCM ^ rH rH rH rH rH rH \ \ ^ rH rH rH CM rH CM CM ', ' \ '. \ ] ' [ | CM M CM O5 rH 1C CO ^ 00 O O *O t>- ^ CO O *C O O CO CO O5 00 O O ^ !C 00 I s * O CO O O -*OJ>- Tf C CO C35 t^ O O * 05 -ICCOOOJCM rH iH CM rHrn" rtCo" rH ' rH CM*" rH OOO ->COOOOOOOCOOOOOOOOCOOOO -OOOO-* O 1C O *C ^ CO t* O5 O ^C O l^ O ^* CM ^ O I s - O ^C O5 CO ^C CO CO ^ rH O rH ^ CM OrHrH .rHrHCMrHCM(NOOt^OrHCqCOrHrHCM^.rHr-IOOCM CM 00 CM I O> CM "-*hH c^ s : : : _: : : : :^ ill w ^^_ i3 COOQQQOQSSS-^caca^rtOflJ OQ^ CO rH rH 1-H t^ 00 1C O COO OrH OCO %% S OS 00 CO CO >O CO rH I 00 05 CO CO CN CO rH O * rH OOOCO -00 -O i^ -d - coos b. CO(N O OS OrHOO CO t>- OOCO -T}( US-& CNJ O CO rH 1C (N rH rH cd coic'od O CO i-H CM 1-H 1-H i-H (N * dCO rH O t^CO -00 i-l "5iCiC(NiJI rH t^(NCO -CN -OO ylinders power Number Rated ho 1? H rg a s 0>-Q sa -as a -- l: :-s fee tem: er assembly. . nd connectio connections. . weight water in engine. . or eade or ret h m a p u p m n rte ry of 00 PnO Ca I W Oil Fu Mi Sel To We ENGINE DIMENSIONS AND ARRANGEMENTS 79 The weight per cubic inch of piston displacement per stroke is an excellent measure of the success of the designer in keeping down weight. The actual horse power developed depends on mean effective pressure and revolutions per minute and is affected by factors outside the engine proper: the mean effective pressure is determined largely by the fuel used and the carburetor and manifold resistance; the revolutions per minute are limited by the desirable propeller speed in ungeared engines. Detailed dimensions of some of the most successful American and German engines are given in Table 4. Weights of engine parts are given in Table 5. External over-all dimensions for selected engines are given in Table 6. The maximum width occurs at the crankcase level TABLE 6. OVER-ALL DIMENSIONS OF ENGINES Type Name Horse power No. of cylinders Dimensions, inches Maximum Length Width Height Rotary Radial air-cooled Vertical 45-deg. Vee.... 60-deg. Vee. . . . 90-deg. Vee f Clerget 130 100 80 170 320 200 200 160 240 125 95 240 200 400 320 360 220 250 270 400 200 90 160 180 9 9 9 7 9 6 6 6 6 6 4 6 6 12 12 12 12 12 12 12 8 8 8 8 36.22 24.75 36.1 35.7 42.1 63.0 67.25 57.08 67.32 63.38 57.0 69.88 68.9 69.1 61.81 75.98 72.04 55.41 62.25 68.3 43.5 50.0 67.38 51.3 40.15 37.5 37.25 42.2 48.5 22.37 19.0 19.92 20.07 18.5 18.5 24.09 22.4 26.8 37.79 42.52 37.24 35.46 27.125 27.9 31.7 30.0 45.5 33.5 40.15 37.5 37.25 42.2 47.7 39.25 43.0 31.88 42.91 41.25 39.5 43.62 45.3 43.0 38.89 48.03 42.0 33.85 34.75 40.1 35.5 27.0 35.92 32.7 < Gnome 1 Le Rhone . . ( ABC Wasp ( ABC Dragon Fly Curtiss K6 Liberty 6 Beardmore Galloway Hall-Scott A5 Hall-Scott A7 Siddeley Puma ... . Austro-Daimler Liberty Sunbeam Cossack Rolls-Royce, Eagle 8. . Rolls-Royce, Falcon 3 Sunbeam Maori Packard Curtiss K12 Sunbeam Arab L Curtiss OX. Curtiss VX Hispano-Suiza 80 THE AIRPLANE ENGINE in vertical engines and at the cylinder tops in Vee engines. As the tops of the cylinders are often above the fuselage, the width of the fuselage is not necessarily controlled by the maximum width of the engine. AMERICAN ENGINES Liberty Engine. The Liberty engine is constructed either as a six-cylinder vertical, or a 12-cylinder Vee with an included angle of 45 deg. It has built-up steel cylinders, overhead valves and camshaft, and battery ignition. The cylinder units are the same in both constructions. Detailed dimensions are given in Table 4; weights are given in Table 5. Longitudinal and transverse sections of the 12-cylinder engine are shown in Figs. 46 and 47. The performance of this engine is shown in Fig. 48; the full throttle curves are from tests with a dynamometer load and are carried up to 2,000 r.p.m.; the propeller load curves are from tests of the engine equipped with its proper propeller and mounted on a torque stand. With the propeller used the engine runs at about 1,700 r.p.m. at full throttle at ground level. Maxi- mum power is obtained at about 1,850 r.p.m. and maximum economy at about 1,800 r.p.m. The brake mean effective pres- sure, mechanical efficiency, and manifold depression are shown in Figs. 27, 14, and 17 respectively. The gear trains for driving the camshafts and various accessories are shown in Figs. 49 and 50 for six- and 12-cylinder constructions respectively. Some of the details of this engine are discussed under the appropriate headings in Chapters VI and VII. At the rated speed of 1,700 r.p.m. the six-cylinder engine develops 232 h.p., and the 12-cylinder engine 425 h.p. The diminution in horse power per cylinder results from the lower mean effective pressure (see Fig. 27), which apparently results from lower volumetric efficiency. The weight per horse power falls from 2.45 Ib. for the six-cylinder to 1.99 Ib. for the 12- cylinder engine. Packard. This engine is built both as an eight- and 12-cylinder engine, with an included Vee angle in both cases of 60 deg. It is very similar to the Liberty engine in its general features but has smaller bore and stroke, a different method of cylinder construction (see Fig. 92) and an underneath carburetor with induction pipes through the crankcase. Detailed dimensions are given in Table 4; weights are given in Table 5. The per- ENGINE DIMENSIONS AND ARRANGEMENTS 81 82 THE AIRPLANE ENGINE formance of the 12-cy Under engine is shown in Fig. 50. With the propeller used in the tests the engine speed is 1,600 r.p.m. at full throttle; maximum power is at about 2,400 r.p.m., and maximum economy at about 1,800 r.p.m. The brake mean effective pressure and mechanical efficiency are shown in Figs. 27 'and 14 respectively. At the rated speed of 1,600 r.p.m. the eight-cylinder engine develops 192 h.p., and the 12-cylinder engine 280 h.p. The FIG. 47. Transverse section of Liberty-12 engine. horse power per cylinder remains constant. The weight per horse power falls from 2.82 Ib. for the eight-cylinder engine to 2.62 Ib. for the 12-cylinder engine. Hispano-Suiza. This engine is built in several sizes as an eight-cylinder Vee with included angle of 90 deg. In this country two cylinder sizes are built. The design is characterized by steel cylinder sleeves screwed into aluminum water jackets cast in blocks of four, by overhead valves and camshafts, and by ENGINE DIMENSIONS AND ARRANGEMENTS 83 450 400 i_ 3J 300 1 ' 250 o> 1 200 150 100 c 0.65 JC ap 0.60 to 8. 0.55 ^ Q50 0.45 ^ -^ "X, x x X ^ (X* full. /* Q Jx / X* / A / V 1 ft! ^ $'/ / y $ ^ r / \ \, ^ \ ? Fu ?/ rdfo2332"Hg.and32 F Observed Horsepower (Propeller) Barometer 30.5" Hg. *>yy i , 1700 18.00 1900 2000 2100 2200 2300 2400 Engine Revolutions per Minute FIG. 68. Performance curves of Bugatti engine. ENGLISH ENGINES Rolls-Royce. The Rolls-Royce engines are interesting as probably representing the highest grade of design and manu- facture in any country. General dimensions of the 12-cylinder, 60-deg. Vee "Eagle" and "Falcon" models are given in the following table, which shows how the power and efficiency of these engines have been improved by increasing the revolutions per minute and the compression ratio. Sectional views are given in Figs. 69 and 70. These engines have an epicyclic reduction gear concentric with the crankshaft (see p. 381) ; this gear is contained in a housing bolted to the front of the crankcase and is not shown in Fig. 69. The housing for the driving gears of the valve motion, magnetos, and other accessories is bolted to the rear of the crankcase. The upper part of the crankcase carries the bearings of the crankshaft; the lower part is merely a deep oil well. Each cylinder is fastened to the crankcase by four bolts. The engine is supported by arms which ENGINE DIMENSIONS AND ARRANGEMENTS 99 100 THE AIRPLANE ENGINE are bolted to vertical surfaces at the sides of the crankcase, thus permitting the ready adaptation of the engine to the airplane without the need of special engine bearers in the fuselage. FIG. 70. Transverse section of Rolls-Royce "Eagle." The cylinders are of steel with valve fittings welded to short nipples in the cylinder head and with welded jackets. The aluminum piston is of special design without a skirt at the middle third of its length (see Fig. 98, p. 135); this design transfers the gas pressures directly to the piston bosses and reduces the ENGINE DIMENSIONS AND ARRANGEMENTS 101 if i^ IO 00 CO ^* N* (N CO N f - Tf* Tjl CO rH rH OS OS CO OS OS ^ 1 ^ 11 tf CO O C? 3" 00 5O O OS 00 O CO O O O rH T* 00 II ^ ^0 ^^ 10 U2 TH * 10 .a . 10 ,0 10 10 5 ^i^S CO CO O CO CD CO CO O O >O f *> V * V V ., '" ||fi|g i3l O CO >O l^ IO 00 O CO OS 00 1 ^ S ' 1 ^ 00,00,0, 00000 " S3 iH W i-H OS 00 CD O O OS OS, OS, I fi " 3 J g '" OS CO rH IO CD CO CO 0000 o o o o 00 CO t>- O O H ^ ^ c^ ^c l]i CM N, CO CO CO O CM O O * CSI O O >O CD CO CO 0> 3 OJ P- 3 < o 1 o 1 t^ 10 ^H CO II COrtfCilNCOOOOJCO TjtCOrHOOTjJr-ICO d d d d o' i-< d d o S CO I o 1 I-H (N t~ 1C -^ O ^ 1 OJ 10 i-l CO CO -CO CO O i-l O O -OS d d d d d oi _|0 S|p 5 1 I I 1 I % 3 2 3 S % d d d d o* r-I ^ O O 00 Oi O i-I CO (N - (N 10 O T}< o co * g al rH (N i-H O O '00 d d d d d * TjJ 1C OS CO CO oo ^ l> CO (N * 8 Ml oo i-l CO O i-H CX| . Cfl i-l J> rH -1-1 d d d o o 'o 23! SS33SS d d d d d III 00000 il 1 55 i ulfill! tf pq 118 THE AIRPLANE ENGINE The materials that have been employed for the various parts of airplane engines are given in Table 8, 1 which also states the nature of the stresses to which the part is subjected and the inten- sity of the service which it has to perform. In addition, there is given in the last column the material recommended by Doctor Hatfield on the basis of an unusually wide experience in investi- gating the physical properties of the steels and the causes of failure of the parts of airplane engines. There is naturally much difference of opinion among engineers as to the best material to use for several of these parts, and practice is by no means standardized. Aluminum Alloys are largely used in airplane engines on account of their light weight. The specific gravity of pure alumi- or / ~ensi!eStrength,Ibpe i-o j 5i 3 or o i o o o c ff ^*** J- <& / / J ? S ' 0246 8 10 IE 14- Metal AlloLjed with Aluminum, Per Cent. FIG. 83. Tensile strength of aluminum alloys. 14 2 4 6 8 10 12 Metal Alloyed vv'ith Aluminum, Percent FIG. 84. Ductility of aluminum alloys. num is 2.56; of the common alloys about 2.8; of steel about 7.8. The important alloys are those with copper and with zinc. The effect of the presence of these substances on the tensile strength . MATERIALS FOR ENGINE PARTS Glossary of Terms used. A. H. Ni. Cr. = Air-hardening Nickel-chro- M. C. I. mium Steel. M. C. Steel = Aluminium Alloy. M. S. = Case-hardening Carbon Ni. Steel. Ni. C. H. = Cast Iron. = Carbon Steel. Ni. Cr. = Chromium Vanadium Steel. = High-carbon Steel. Ph. Br = 12 to 14 per cent tungsten Si. Cr. High-speed Steel. Steel C. = High-tensile Steel. T. Steel Al. C. H. C. C.I. C. Steel Cr. Van. H. C. Steel H. S. H. T. Malleable Cast Iron. Medium-carbon Steel. Mild Steel (0.15 carbon). Nickel Steel (3 per cent). Nickel Case-hardening Steel. (5 per cent Ni.). Nickel-chromium steel (3 per cent). Phosphor Bronze. Silicon Chromium Steel. Steel Casting. Tungsten Steel. 1 HATFIELD, loc. cit. MATERIALS TABLE 8 (Continued) 119 Part Nature of chief stresses in service Intensity of service Materials which have been or are used Materials recommended Cylinder H'gh temperature, Heavy or Al.; C. I.; Steel Al.; 80,000 Ib. tension and abra- sion. medium. C.; 100,000 Ib. C, Steel; Ni.; 3 per cent Ni. Cr. ; Steel Mixture. Steel. Cylinder holding- down bolts. Tension and bend- ing. Medium. Medium or M. S.; Ni. 3 per cent Ni. Cylinder liners High temperature and abrasion. Heavy. C. I.; Steel C.; Forged Steel. 100,000 Ib. C. Steel. Spark-plug body. . . High temperature. Medium. Brass; C. I.; M. S.; Stainless. Stainless. Spark-plug elec- trode. Very high tempera- ture. Heavy. T. Steel; Ni. Ni chrome Alloy or Stainless. Valve cages High temperature, various slight stresses. Medium. C.I. M. C. I. Valve-rocker roll- ers. Abrasion. Light to medium. C. H. C.; Ni. Cr. 5 per cent Ni. C. H. C. Valves High temperature, Heavy. H. S. Steels; 25 Stainless. tension, shock and abrasion. per cent Ni.; Stainless Steels; Ni. Cr.; 3 per cent Ni.; H. C. Steel. Valve guides High temperature Medium. AL; C. I.; C. Stainless. and abrasion. Steel; T. Steel. Valve seats Abrasion, shock Heavy. C. I.; M. C. 80,000 Ib. C. and high tem- perature. , Steel; Ni. Cr.; Ph. Br. Steel. Valve springs Torsion and bend- Light. Cr. Van. Steel; Cr. Van. Steel. ing. Si. Cr. Steel; H. C. Steel (oil hardened). Valve rockers .... Bending, shock and Medium. Ni. Cr.; C. H.; 5 per cent Ni. abrasion. Ni.; Bronze. C. H. ; 3 per cent Ni. Valve-rocker bear- ing. Abrasion and com- pression. Light. Ph. Br. and White Metal. Ph. Br. Water jacket Very slight. Light. Al. ; Copper; Steel Sheet. Pressed Steel; Sheet Steel. 120 THE AIRPLANE ENGINE TABLE 8 (Continued) Part Nature of chief stresses in service Intensity of service Materials which have been or are used Materials recommended Connecting rod. . . . Compression, ten- sion, bending and shock. Heavy. A. H. Ni. Cr.; Cr. Van. Steel; Ni. Cr.; Ni. Ni. Cr.; A. H. Ni. Cr. Connectin g-rod, big-end bearing . . . Abrasion and com- pression. Heavy. White Metal; Ph. Br. Connectin g-rod, little-end bearing.. Abrasion and com- pression. Medium. Ph. Br.; Gun Metal. Connectin g-rod bolts. Tension and bend- ing. Medium. Ni.; Ni. Cr.; Cr. Van. Steel. 3 per cent Ni. Piston pin Shear, bending and abrasion. Heavy. Ni.; 160,000 to 180,000 Ib. C.; C. H. C. Steel; C. Steel; Ni.; Ni. Cr.;T. Steel. 5 per cent Ni. C. H. Piston . ... Temperature, bend- Heavy Al C I Steel Al 80 000 Ib C ing and other stresses. Drawn or Pres- sed; Ni. Cr. Steel. Piston ring High temperature, Heavy C I H S Steel C I bending and abrasion. Bevel-gearshaft for overhead c a m- shaft. Torsion and bend- ing. Medium. C. Steel; Ni.; Cr. Van. Steel. 5 per cent Ni. C. H. Crankcase .... Bending and vari- Light C Steel- Al Al ous slight stresses. Ni. Crankshaft Torsion, bending, Heavy Cr Van Steel Ni Cr and shock. Ni. Cr.; Ni. Crankshaft journal bearing. Abrasion and com- pression. Medium. Ph. Br.; White Metal. Propeller reducing gears. Shear and bend- ing, abrasion and shock. Medium. Ni. C. H.; Ni.; Ni. Cr.;Cr. Van. Steel; A. H. Ni. Cr.; C. H. C. A. H. Ni. Cr. Cams Compression and Heavy. Ni C H C H 5 per cent Ni. C. abrasion. Cr. H. Cam housings Negligible. Light. Al. Al. Camshaft bearing. . Abrasion and com- pression. Medium. White Metal; Ph. Br. White Metal; Ph. Br.; Gun Metal. Camshaft Torsion and abra- sion. Heavy. Ni. Cr. C. H.; Ni.; H. C. Cr. Steel. 5 per cent Ni. C. H. Tappets Compression and abrasion. Heavy. C. H. C. hard- ened on wear- ing surface; Ni\ Cr. 5 per cent Ni. C. H. MATERIALS 121 of cast-aluminum alloy is shown in Fig. 83. It will be seen that copper increases the strength up to about 9 per cent, with 7 to 8 per cent of copper there is obtained a tough alloy of a tensile strength of 20,000 Ib. per square inch. With zinc the tensile strength increases up to about 35 per cent; with this alloy the tensile strength reaches 50,000 Ib. per square inch but it is very brittle and has a specific gravity of 3.3. The ductilities of the two types of alloy are shown in Fig. 84. With both copper and zinc present, higher tensile strength combined with fair ductility can be obtained; for example, with 2.75 per cent Cu and 7 to 8 per cent Zn a tensile strength of 28,000 Ib. per square inch and a ductility of 8 per cent. This alloy falls off rapidly in tensile strength with increase in temperature; at 570F. the strength is 9,500 Ib. per square inch. A small addition of manganese to a copper aluminum alloy increases the strength and maintains it better with increase of temperature. Forged-aluminum alloy is best represented by Duralumin, whose composition is 93.2 to 95.5 per cent aluminum, 0.5 per cent magnesium, 3.5 to 5.5 per cent copper and 0.5 to 0.8 per cent manganese. This material can be made into plates and tubes. The tensile strength is about 60,000 Ib. per square inch but can be increased by rolling to about 75,000 Ib. per square inch though with loss of ductility; the elongation of the alloy is 15 to 20 per cent. At 460F. the tensile strength is halved. Forged- aluminum is a good bearing metal. Both cast- and forged-aluminum alloy have a modulus of elasticity of about 10,000,000 Ib. per square inch or about one- third that of steel. The stiffness of a plate structure is propor- tional to the modulus of elasticity and to the cube of its thickness. An aluminum plate of the same weight as a steel plate would be nearly three times as thick; its stiffness would be about eight times as great as that of the steel plate. CHAPTER VI ENGINE DETAILS Cylinders. There is considerable variety in the design of modern water-cooled airplane-engine cylinders. In one impor- tant respect they are all in accord, namely, in the adoption of overhead valves, in which they differ from the common auto- mobile engine. They differ further from the automobile engine in that the cast-iron block construction is seldom used. Both of these changes have been made, primarily, in order to reduce the weight of the engine. Airplane-engine cylinders are formed either singly, or in blocks of two, three or four. The single cylinder is flexible and permits freedom of movement of the cylinder without putting strains on other parts of the engine. As the engine is not held rigidly it is desirable to give all its parts a maximum of freedom. The single-cylinder construction is used in the majority of existing engines and is always employed in those engines whose cylinders are all steel. Examples of this construction are the Liberty, Packard, Hall-Scott L, Curtiss K, Rolls-Royce, Napier, Renault, Lorraine-Dietrich, Fiat, Mercedes and Austro-Daimler engines. It is necessarily employed also in radial and rotary engines (see Chapter VIII). The block arrangement has a common jacket around the cylinders which may result in some slight decrease in weight of the jacket itself but will usually increase the weight of water in the jackets and give inferior water circulation. The major advantage of employing the block construction is that it permits the cylinders to be more closely spaced and thereby diminishes the over-all length (and weight) of the engine. In the Thomas-Morse and Sturtevant engines the cylinders are in pairs; in the Siddeley "Puma" in threes; in the Hispano-Suiza and Bugatti in fours. Another variation among airplane-engine cylinders is in the form of the cylinder head. In some engines (Hispano-Suiza, Napier) the cylinder head is flat and of the same diameter as the cylinder barrel. With two equal valves in the head, the maxi- mum possible external valve diameter is half the cylinder diame- ter minus half the thickness of the bridge between the valves. 122 ENGINE DETAILS 123 It is frequently found that this diameter is insufficient to give the desired opening for the admission of the mixture and results in a low volumetric efficiency. To remedy this the head may be made with sloping sides either without enlargement of diameter (Curtiss OX, Mercedes) or with enlargement (Liberty, Lorraine Dietrich, Austro-Daimler, Fiat). Another common method of meeting this difficulty is by the use of multiple valves. The two devices may be used together. The most important factor in determining the design of cylin- ders is the material employed; the following constructions are in use: (a) All cast iron, with cylinders either single (Curtiss OX) or in blocks. This construction leads to excessive weight. (6) Cast-iron barrel, head, and part of jackets, but with aluminum sides to the jacket (Bugatti). This is for block construction only. It reduces weight and makes the inside of the jacket accessible for machining and cleaning. (c) Cast-iron barrel and head, but with sheet metal jackets, either copper (Beardmore), steel (Benz) or monel-metal (Curtiss). (d) Steel barrels, aluminum heads and jackets. The steel barrel may be integral with its head (Hispano-Suiza, Siddeley "Puma") or only a cylin- drical shell (Sturtevant). The aluminum jacket may be complete (Hispano- Suiza, Sturtevant) or may use the steel barrel as the inner wall (Siddeley "Puma"). (e) Steel barrel, cast-iron head and steel jacket (Maybach, Benz). (/) All steel (Liberty, Packard, Hall-Scott L, Curtiss K, Rolls-Royce, Napier, Renault, Lorraine-Dietrich, Fiat, Austro-Daimler, Bass6-Serve, Maybach). This construction is used more than any other and appears to be displacing other constructions. The cylinder is either machined out of a solid forging or may be made from drawn steel tubing. The jackets are commonly stamped to shape in two halves along a plane through the cylinder axis, and are welded together and to the cylinder. The valve ports and guides offer some difficulties as compared with cast cylinder heads. Thickness of Cylinder Walls. If p = maximum gas pressure in the cylinder, pounds per square inch, d cylinder diameter, inches, t = wall thickness, inches, and s = allowable tensile stress, pounds per square inch, then t = pd/2s, gives the thick- ness of metal necessary to withstand the gas pressures. Taking s as 5,000 and 14,000 Ib. for cast iron and forged steel, the respec- tive wall thicknesses for a 5-in. diameter cylinder and for p = 500 Ib. are 0.25 and 0.09 in. With cylinders of small diameter the thickness may have to be increased over the calculated values to ensure sufficient stiffness and, in the case of cast iron, 124 THE AIRPLANE ENGINE to offset a possible lack of homogeneity. No allowance need be made for wear or reboring as the engines are essentially short- lived. The clearance space alone of the engine is subjected to the maximum explosion pressure; the pressures to which the walls are subjected become progressively less from the clearance space to the part of the cylinder at the lowest point reached by the top of the piston, below which point they become zero. In addition to the gas pressures the cylinder walls have to tie the cylinder head to the crankcase and shaft bearings and consequently have to withstand the maximum gas pressure exerted on the cylinder head. The thickness of wall required for this is given by pd t The lowest part of the cylinder consequently does not have to be more than one-half the thickness of the upper end. In many designs the cylinders are turned of diminishing thickness from head to crank end; in other designs (Mercedes, Lorraine Renault, Liberty, Curtiss K) the upper part of the cylinder is reinforced by stiffening ribs while the lower part is without such stiffening. The following table gives some cylinder thicknesses and calcu- lated stresses; a maximum gas pressure of 500 Ib. per square inch is assumed. Engine Diam- eter, inches Material Cylinder- head thickness, inches Cylinder- barrel thickness, inches Calculated maximum stress, pounds per square inch Benz, 230 5 71 Cast iron 2600 216 6,600 Liberty 12 5.00 Steel 0.1875 0.156 8,000 Curtiss, K-12 4 50 Steel 078 14,400 Hall-Scott A5a 5.25 Semi-steel 0.125 10,500 Austro-Daimler, 200 Renault, 400 5.31 4 92 casting Steel Steel 0.1970 0.138 0.110 9,620 11,200 Basse-Selve 6.10 Steel 0.2700 0.118 12,900 Maybach, 300 6.50 Steel 0.3100 0.110 14,800 The thickness of the cylinder head is determined mainly by considerations of stiffness. It is essential that the valve seats, which are located in the cylinder head, should be free from ENGINE DETAILS 125 deformation and this cannot be secured unless the heads are stiff. In cases where the integral cylinder head is backed by an aluminum casting (Hispano-Suiza, Napier) the thickness cannot be reduced because the heat transfer through the double thickness of metal is poor and the exhaust valve seat, which is the hottest part of the cylinder, must have sufficient thickness of metal around it to conduct the heat away and prevent warping from overheating and unequal expansion. The head thicknesses of a few cylinders are given in the table above. Steel cylinders are generally machined out of solid sheet forgings, but in the case of the Liberty engine the cylinders have been made from steel tubing J^ in. thick, whereby a great reduction is obtained in the amount of metal to be removed by machining. The valve seats are integral with the heads in steel cylinders but have to be cast, or otherwise fastened, into aluminum heads, With integral or non-detachable valve seats it is impossible to take out the valves without taking off the cylinder or taking out the piston. Detachable valve cages have been used, but these will always result in imperfect cooling of the valve seat. A com- promise adopted in the Beardmore and Austro-Daimler engines is to have a detachable valve cage (Fig. 85) for the inlet valve, which requires little cooling; the exhaust valve can then be dropped into the cylinder and withdrawn if necessary through the inlet valve seat opening. The valve ports and valve stem guides in all steel constructions are integral for each valve and are welded to the cylinder head. The guides are provided with bush- ings of steel or bronze. The valves must work very freely in these bushings and at the same time there must be no leakage of air or exhaust gas between valve stem and guide; the guides are made quite long. Many devices have been used for attaching the cylinder to the crankcase. The best method is to tie the cylinder, by through bolts, directly to the main bearing caps on the crankshaft on each side of the cylinder. With two long bolts on each side of the cylinder tying the cylinder flange to the main bearing caps, the gas pressure on the cylinder head is supported entirely 126 THE AIRPLANE ENGINE by stresses set up in the bolts and is not transmitted to the crank- case, which can consequently be made lighter. As the bolts are parallel to the cylinder axis and are symmetrically disposed around it, this method of attachment avoids all distortion. One pair of bolts is commonly made to serve for two adjacent cylinders by the use of dogs through which the bolt goes and which are supported equally on the flanges of both cylinders (Fig. 80). In the Bugatti engine (Fig. 67) the through bolts are supplemented by studs in the crankcase. It is only in engines with a single row of cylinders that the above method of attachment can be employed. In Vee and W engines other methods must be used; the most common is by studs in the crankcase which pass through the flange at the lower end of the cylinder (see Fig. 47). In the Curtiss K en- gines the steel cylinder is kept in place by the aluminum cyl- inder head, which is bolted to an aluminum jacket cast integral with the upper half of the crank- case (Figs. 87 and 56). Typical Cylinder Designs. Single cast-iron cylinders are sel- dom used; a typical example is in the Benz 230 engine of Fig. 77. The barrel and head are cast in one piece. Jackets are of die-pressed steel welded on and extending well down towards the base flange. They are provided with annular corrugations to take care of expansion. Plates welded in position in the jacket space above the crown of each cylinder deflect the water to the exhaust valve pockets. The cylinder barrels extend 0.39 in. below the base flanges into the crank chamber and are held down each by four bolts and four dogs. The cylinder walls taper from 6.5 mm. at the top to 5.5 mm. at the base. FIG. 86. Cylinder of Hispano-Suiza engine. ENGINE DETAILS 127 An example of the block cast-iron construction is shown in Figs. 66 and 67 (Bugatti). The cylinders are in blocks of four. Composite steel and aluminum constructions are made up in several ways. In the Hispano-Suiza engine (Figs. 86 and 51) the steel liner with integral head is threaded throughout the entire length of contact with the aluminum. The aluminum jacket is a block construction for four cylinders and completely surrounds the steel liners so that valve ports go through both the aluminum and the steel. For proper cooling it is necessary to FIG. 87. Cylinder of Curtiss K engine. have perfect contact of the two metals at the cylinder head as well as the barrel, otherwise warping of the steel head will occur and the valve seats will distort. There is no actual contact anywhere of water with the steel cylinder. The steel liners are attached to the crankcase by bolts. A different steel-aluminum block construction is employed in the Curtiss K engines. In this case the steel liners with integral heads are turned with stiffening flanges (Figs. 87 and 56) on the outside, with a packing retaining flange at the bottom and a central stud at the top. The upper end of the liner is of slightly enlarged diameter and is threaded on the outside. The alumi- 128 THE AIRPLANE ENGINE num. cylinder head is a block casting into which the six cylinders are screwed. To ensure intimate contact the threaded stud at the center of the liner head, which passes through the head casting, is drawn up by a nut. The head casting matches up with and is bolted to a flange on the upper end of the aluminum cylinder block, which is integral with the upper half of the crankcase. Jacket water is in contact with the liner below the combustion space only. Water tightness is obtained by packing between the lower liner flange and the crankcase. Still another construction is used in the Napier "Lion" engine (Fig. 73). This engine employs liners with integral heads, which are fastened to a four-cylinder aluminum-block head by four valve seats (inlet of bronze, exhaust of steel) in each cylinder. These seats are screwed into the cylinder head. The use of remov- able seats is in general objectionable as it means less perfect cooling of the seats than is possible with an integral construction. The barrel jackets in the latest design are separate and of pressed steel, made in two halves welded together and to flanges on the cylinders at the top and bottom of the water spaces. In the design shown in Fig. 73 a common steel water jacket is bolted to the head casting and makes a joint with each cylinder at its lower end by means of a rubber ring pressed against a flange on the liner by a large circular split nut. A steel-aluminum block construc- tion in which the liner has no head is employed in the Siddeley "Puma" engine (Fig. 8). In this case the cast in sets of three and the steel liners Aluminum water jackets The lower FIG. 88. Cylinder of Sturte- vant engine. aluminum heads are are screwed and shrunk into them. are also cast in threes and are bolted to the heads. joint between the aluminum jacket and the liner is made by a screwed gland, which squeezes a rubber ring against a shoulder on the outside of the liner. The valve seats are of phosphor bronze expanded into the aluminum head. ENGINE DETAILS 129 FIG. 89. Cylinder of Maybach engine. FIG. 90. Cylinder of Benz 300 engine. 130 THE AIRPLANE ENGINE Another construction in which a headless steel liner is em- ployed is the Sturtevant engine, Fig. 88. The aluminum cylin- ders are cast in pairs and are provided with closely fitting steel liners; the heads are also of aluminum with inserted valve seats and are in pairs. A peculiarity of this construction is that the aluminum cylinders are bolted direct to the crankcase; the steel liner does not transmit any longitudinal forces and is perfectly free to expand. A composite steel and cast-iron con- struction is used in the Maybach engine (Figs. 89 and 81). The steel barrel is screwed with a buttress thread into a cast-iron head and comes up against a soft brass washer which prevents water leakage. The water jacket is a machined forging and screws on to the cylinder head (see details, Fig. 89) where it is sweated in position with soft solder; the lower joint is kept watertight by a gland and rubber ring. The valve-stem guides have cast- iron bushings. In the Benz 300 engine (Fig. 90) the cylinder head is of cast iron and the rest of the cylinder is of steel. There are ports for two inlet valves and one exhaust valve. The steel liner screws into the cast-iron head and makes a watertight joint by bed- ding into cement in the small groove into which the top of the liner goes. The all-steel construction is exemplified in the Liberty cylinder (Fig. 91). This design has a bumped head and obliquely set valves. The jackets and valve ports are welded to the sheet cylinder. Details of this welding are shown for the very similar Packard engine, Fig. 92. In the Austro-Daimler engine (Fig. 93) the cylinders are all steel with pressed steel jackets and twin inlet and exhaust valves in the cylinder heads. The valve pockets are welded in position with the exception of one inlet valve (Fig. 85), which is detachable with its seat and guide. The cylinders taper from 4 mm. at the top to 3 mm. at the middle and increase again to 4 mm. at the bottom; the jackets are 1 mm. thick. The bottom FIG. 91. Cylinder of Liberty engine. ENGINE DETAILS W 131 W: w FIG. 92. Cylinder head of Packard engine showing welding, W, of the water jacket. P FIG. 93. Cylinder of Austro-Daimler engine. 132 THE AIRPLANE ENGINE of each jacket is flanged over and welded to a bevelled flange machined on the barrel. Pistons. One of the most important steps in the improve- ment of the airplane engine has been the general substitution of an aluminum alloy for the cast iron that, until very recently, was universally employed for the piston material. The piston must be as light as possible in order to keep down inertia forces and at the same time must be thick enough in the crown to conduct the heat away rapidly from the center to the circumference, where it is taken up by the cylinder walls. With cast-iron pistons, reduc- tion in weight has resulted in many troubles and especially in the burning and cracking of the piston head. Measurements by Hopkinson in a Siddeley engine show the cast-iron piston head has a temperature of over 900F. This is borne out by similar measurements on the pistons of air-cooled cylinders by Gibson. FIG. 94. Aluminum piston. FIG. 95. Cast-iron piston. With aluminum pistons this temperature is reduced to about 400F. and at the same time the piston is considerably lighter. The two pistons of Figs. 94 and 95 for a 100 by 140 mm. air- cooled engine have weights of 1.26 Ib. and 1.77 lb., or a reduction in weight of 29 per cent by the substitution of aluminum alloy for cast iron; these pistons show the temperature difference noted above. The lower temperature of the aluminum piston has other important results. It reduces the rise in temperature of the incoming charge during the suction stroke and thereby increases the volumetric efficiency of the engine, and it permits the use of a higher compression ratio without danger' of preignition and thereby increases both the capacity and efficiency of the engine. The horse power of an engine can be increased at least 5 per cent by the use of aluminum pistons. The piston should be designed with heat dissipation in mind as much as the other piston functions. The heat received at the center of the head must pass out radially to the circumference, and ENGINE DETAILS 133 this should be provided for either by adequate thickness of metal from center to periphery or by the provision of radial ribs which serve the double function of heat carriers and stiffening members. Furthermore, the thickness of the cylindrical wall must not be cut down behind the first ring as the heat has to flow downward from the crown. The friction between the piston and the cylinder is by far the largest item of mechanical loss (see p. 24) and should be kept as low as possible. Its high value results apparently from the partial carbonization of the lubricant clinging to the walls under the action of the high gas temperatures and of the slight but unavoidable leakage of burning gases past the piston rings. As a result, the viscosity of the lubricant is greatly increased. The FIG. 96. Slipper piston. extent of the friction depends upon: (1) The pressure of the piston against the cylinder walls, which governs the thickness of the film; the friction appears to be proportional to the average loading; (2) the area of the bearing surface; (3) the quantity of the lubricant on the walls; the friction increases with this quan- tity; (4) the temperature of the walls, which controls the viscosity of the lubricant. Of the means adopted to reduce this friction loss the most prominent is the cutting away of the piston skirt on the sides which do not support side thrust. This practice has the further advantage of reducing the weight of the piston. An example of such a piston is shown in Fig. 96. It will be seen that the piston-pin bosses in this design are supported by vertical transverse ribs, which pass within a distance from the center of the head of a little more than half of the radius of the piston. This is an important feature in keeping the piston head cool; the heat absorbed by the central portion of the head 134 THE AIRPLANE ENGINE can pass down these ribs to the gudgeon-pin bosses and to the skirt. This in turn permits the use of a thinner crown, especially as the stiffness of the crown is greatly increased by the support of the ribs. The thickness of the lubricant film is diminished by providing holes in the slippers through which excess oil is squeezed out. If both slippers are designed for the same inten- sity of pressure, then areas of the two slippers will be different. Such a design is shown in Fig. 97; the supporting ribs are no longer parallel. FIG. 97. Piston with unequal slippers. One of the important troubles with pistons is the slap which occurs when the side thrust is transferred from one side to the other. The amount of this slap depends on the clearance between the piston and cylinder. The cold clearance must be larger with aluminum than with cast iron as the expansion is much greater. The pistons of Figs. 94 and 95 have cold clear- ances of 0.026 and 0.020 in. respectively; the hot clearances for both are 0.008 in. The clearances should be greatest at the top and where the temperatures are highest and should diminish as the bottom of the skirt is approached. With aluminum pistons a ENGINE DETAILS 135 cold clearance over the top lands of about 0.005 in. per inch diameter is necessary. Such a piston will be noisy when cold. For cast iron the cold clearance should be 0.003 in. per inch diameter at the top, and 0.00075 at the base. The clearances for special engines are given in Table 4. In order to prevent piston slap, or the opposite danger of seizing when hot, the practice has arisen of insulating the piston skirt from the ring-carrying portion of the piston. This is most readily accomplished by the use of a piston with piston-pin bosses carried by ribs (Fig. 97) and with the skirt or slippers separate from the upper portion of the piston. A design of this character is shown in Fig. 98 ; in this case a complete skirt is used. Such constructions are satisfactory only for cylinder diameters up to about 5 in.; for larger sizes the piston cooling will be inadequate. The clearance necessary for the skirts of aluminum divided slip- per pistons is about the same as that re- quired for the normal cast-iron piston. Another method of reducing piston slap is by offsetting the wristpin by a small amount, usually not more than J4 m - The object of this construction is to cause the piston to tilt slightly about the piston pin and therefore to pass progressively instead of abruptly from one cylinder wall to the other. Such offset is shown in Figs. 96 and 97. The composition of the aluminum alloy used in German pistons is given below. These pistons are usually die castings and have a tensile strength of 28,000 to 31,000 Ib. per square inch and extension of 4^ per cent as against about one-half those quan- tities for sand castings. FIG. 98. Divided-skirt piston. 1 g S3 I N i i Silicon a H 2 I i Alumin Benz 230 h p 6 02 12 13 1 42 0.31 Tr. Tr. 80. 12 Austro-Daimler 200 h.p 7.67 1.33 1.32 0.52 2.21 Tr. 0.29 86.66 Basse-Selve 270 h.p 1.90 15.62 1.06 0.45 o 80.97 136 THE AIRPLANE ENGINE Piston weights (including piston rings and gudgeon pin) vary from 0.19 Ib. (Austro-Daimler) to 0.25 Ib. (Liberty) per square inch of piston area in aluminum construction; for cast iron the weight may exceed 0.42 Ib. (May bach). Typical Pistons. Figure 99 is the May bach cast-iron piston. Figure 100 is the Benz cast-iron piston with a thin crown and with __^ FIG. 99. Maybach cast-iron piston. a hollow conical steel pillar riveted to the piston head and resting directly on the middle of the piston pin. The small end of the connecting rod is cut away to avoid interference with this pillar. With this arrangement the gas pressure is transmitted in the line of the connecting rod and there is no bending moment on the wrist pin. The ribbed aluminum construction is shown in Fig. FIG. 100. Benz 230 cast-iron piston. 101 for the same engine. The top clearances for the two pistons are 0.02 and 0.03 in. respectively; the bottom clearances are 0.004 and 0.014 in. respectively. The weights are 6.72 Ib. and 4.90 Ib. respectively, complete with rings and setscrews but without piston pins; the saving in weight is 27 per cent. The ribless construction is typified by the Liberty engine piston shown in Fig. 102; oil grooves are provided on the piston skirt. ENGINE DETAILS 137 Piston rings are of dense gray cast iron, fully machine-finished, peened on the inner curved surface and exactly ground to size upon the outer curved surface. With very narrow rings semi- steel is used. The rings may be either of the concentric or eccentric types. The ends are commonly chamfered at an angle of 30 to 40 deg. but stepped ends are also used ; the gap when in the cylinder is about ^40 the diameter of the piston. Three FIG. 101. Benz 230 aluminum ribbed piston. rings are commonly used, but four rings are found occasionally. Two narrow rings are sometimes used in one groove. A scraper ring near the bottom of the skirt is sometimes used to clear excess oil from the cylinder walls; the same result is obtained by the use of perforations through the skirt. In some pistons (Figs. 99 and 100) the lowest ring acts as a scraper and has a groove below it through which small holes are drilled to the interior of the piston to drain away any excess of oil. 138 THE AIRPLANE ENGINE Piston or Gudgeon Pin. The piston pin is usually of steel, machined to size, case-hardened and ground. It is always hollow. It is most commonly fully floating, that is, it has bearing in the end of the connecting rod as well as in the piston bosses, with some end motion as well. The pin will then rotate and local wear will be avoided. The bushing in the connecting rod is also floating in many engines. On the cold motor the pin should be a mild driving fit in the bosses and a running fit in the connecting-rod bushing. When warm the aluminum bosses expand more than the bushing and the pin becomes free. With standard piston types, as in Figs. 99-102, the piston pin is comparatively long and is subjected to considerable bending stress. By the adoption of the slipper piston (Figs. 96 and 97) or the divided skirt (Fig. 98) the bosses are brought closer together and the pin is shortened. It can consequently be made of smaller diameter and lighter and if fully floating will show no wear. Connecting Rods. In vertical engines the connecting rods are of uniform (non-tapering) circular or I-section, with solid small ends and marine type big ends. An example of the circular section (Benz 230) is shown in Fig. 103; the I-section, assembled with the piston (Siddeley "Puma"), in Fig. 104. The small end is usually provided with a bronze bushing, although in the Maybach engine this is replaced by a perforated cast-iron floating shell 0.124 in. thick. The large end has a babbitted-bronze shell. The tubular rods used in several German engines are sometimes provided (Fig. 103) with a centered internal pipe for lubricating the small end. The Benz rod has a number of radial holes FIG. 102. Liberty ribless aluminum piston. ENGINE DETAILS 139 drilled in the big end to reduce weight and has the top of the small end cut away to permit direct application of the gas pressure load on the crankpin through a pillared piston as in Fig. 100. In Vee engines several connecting rod arrangements are used. In the Curtiss OX and V, Sturtevant and Thomas-Morse engines FIG. 103. Benz tubular connecting rod. the cylinders in the two rows of the Vee are staggered so that the connecting rods do not lie in the same plane. With this arrange- ment the big ends of each pair of cylinders lie side by side on the same crankpin. Such an arrangement results in an increase in the over-all length of the engine. 140 THE AIRPLANE ENGINE With the cylinders of each pair opposite- one another, as in the general practice with Vee engines, the connecting rods are in the same plane and special arrangements must be made to connect them both to the crankpin. Two arrangements are in use, the forked rod and the articulated rod. The forked rod is used in the Liberty, Hispano-Suiza, Packard, and Fiat engines. Each pair of rods consists of a plain rod and a forked rod. The forked rod clamps the big end bronze bushing; the plain rod works on the FIG. 104. Siddeley "Puma" I-section connecting rod. outside of this bronze bushing between the two forks of the forked rod. The rods for the Liberty engine are shown in Fig. 105. The bearing is prevented from rotating in the forked rods by dowel pins. In the Fiat engine the bottom ends of the fork are fastened together. In the articulated rod assembly a master rod is used and a short rod is attached to a pin which is held in the upper half of the big end of the master rod. Ordinarily the master rods are all placed on one side of the Vee, but occasionally (Renault) the master rods alternate with short rods. A good example of a ENGINE DETAILS o 141 /A FIG. 105. Forked connecting rods of Liberty engine. FIG. 106. Master rod of Benz 300 engine. 142 THE AIRPLANE ENGINE master rod is shown in Fig. 106 for the Benz 300, 60-deg. Vee engine. The rod is tubular; the pin for the small rod is held by two clamp screws. In the Renault engine (Fig. 107) the pin for the small rod is of the same diameter as the piston pin so that both ends of the small rod are alike. In W engines the articulated rod is most common. The Napier "Lion" uses a central master rod, on each side of which is mounted an articulated rod (Fig. 108) carried on pins fixed in lugs integral with the big end of the master rod. The main rod is of I-section; the side rods are tubular and carry bronze bush- ings at both ends. Each side-rod pin is tapered at one end, fits into a tapered hole in the corresponding lug, and is drawn up tight by a bolt screwing into the pin at the taper end. In the Lorraine W engines the two outer rods bear on the cylindrical outer surface of the big end of the master rod, the FIG. 107. Articu- lated connecting rods of Renault engine. FIG. 108. Articulated connecting rods of Napier "Lion." bearing slippers covering less than half the circumference. Two circular steel rings hold the two halves of the big end of the master rod together and hold the slippers of the outer rods to the outer surface of the master rod. ENGINE DETAILS 143 The articulated arrangement of connecting rods suffers from some disadvantages as compared with an arrangement in which the connecting rods are always radial to the crankpin. The short rod is materially shorter than the master rod usually at least 20 per cent shorter and consequently causes greater angularity of that rod and increased side thrust in the cylinder. The explosion pressures transmitted along the short rod do not act directly on the crankpin but impose stresses on the master rod which under unfavorable conditions may be serious. For example, with a 90-deg. Vee and with explosion pressure reached 30 deg. before dead center in the short-rod cylinder, the force acting on the master rod CD (Fig. 109) would be directed along the line AE. The reaction of this force at C can be readily found and, treating the master rod as a FIG. 109. Diagram of articulated con- cantilever loaded with this re- necting rods< action force at C and held on the crankpin, the stress at any section of the rod can be ascertained. Connecting rods up to the present time have always been made of steel. The use of forged aluminum rods would mate- rially reduce the weight of the reciprocating parts and the bearing pressure at the big end. Crankshafts. Crankshafts are made of alloy steel (nickel or chrome-vanadium) and are usually forged in one piece. An exception to this is the Bugatti eight-throw shaft which is made in two lengths. In airplane practice there are usually bearings on both sides of each throw; this gives great stiffness to a light shaft. The arrangement common in automobile engines of two or more throws between main bearings has been employed by the Sturte- vant, Thomas-Morse and Duesenberg engines and is still used in the Curtiss K engines (see Fig. 55). The long crankarms (between the first and second, and between the fifth and sixth cranks) of this engine have centers of gravity which do not coincide with the axis of the shaft (as in four-cylinder engines) and consequently produce an unbalanced moment about the crank axis. This can be balanced by a counterbalance weight between the two center crankpins which are in line. The addi- 144 THE AIRPLANE ENGINE tion of such a counterbalance weight sets up an undesirable bending moment on the long central crankpin and also puts more load on the two central main bearings, which have to resist the moments created by these unbalanced masses. To eliminate these objectionable conditions it is desirable to balance directly the masses of the two long crankarms. This is accomplished in the Curtiss K engines (Fig. 55) by applying balance weights directly to the long crankarms. To obtain the greatest possible balancing effect with the least weight, aluminum spacers are inserted between the steel balance weights and the crankshaft, the balance weights being held to the crankshaft by steel bolts. The crankshaft and pins are always made hollow and holes are drilled through the crank webs for oil passages connecting the hollow crankpins and journals. The open ends of the shaft and crankpins are plugged by screw plugs (Hispano-Suiza) or by discs or caps which are expanded or brazed into place, or in some cases (Liberty, Siddeley "Puma") are held in position by bolts which tie together a pair of caps. In this last case the bolts can be used to obtain rotational balance, the method being to use special bolts thickened in the middle. The caps in different con- structions are of duralumin, gun metal, steel and other metals and may be used with or without gaskets. Main bearings and crankpin bearings are nearly always of babbitted bronze. Occasionally a ball bearing is used at one end; at the rear end in the Hispano-Suiza; at the front end in the Fiat. In the Napier "Lion" with three cylinders on each crankpin and with a heavy big end, the main-bearing pressures are very high and roller bearings are used throughout. Double-thrust ball bearings are usual and are placed just behind the propeller hub. The pressure on main bearings is high and demands con- siderable oil circulation, not only for lubrication but also for cooling. The babbitt tends to flake off unless the bronze has been tinned before casting the babbitt, in which case a perfect bond can be obtained; mechanical holding of the babbitt by holes, dove- tails or screw threads is generally found unsatisfactory. The bearing pressure in a six-throw, seven-bearing crankshaft is greatest at the center main bearings because the crank throws on the two sides of it are in line so that the dynamic loads imposed on the two cranks are in phase. Consequently the center main bearing is often made longer than the intermediate bearings to ENGINE DETAILS 145 diminish the intensity of pressure. In Rolls-Royce and Fiat engines the center bearings are 60 per cent longer than the intermediate bearings. In the Liberty engine the maximum load on the center main bearing is 7,700 Ib. or 1,675 Ib. per square inch of projected area; the mean unit bearing pressure is 1,265 Ib. per square inch. As the rubbing velocity is 19.5 ft. per second, the friction work is F = f X 19.5 X 1,265 ft.-lb. per second, where / is the coefficient of friction. On the inter- mediate main bearings of this engine the maximum load figures out as 7,250 Ib. or 1,580 Ib. per square inch of projected area; the mean unit bearing pressure is 700 Ib. per square inch. The end main bearings receive loading on one side only and show a maxi- mum load of 4,025 Ib. or a maximum unit pressure of 815 Ib. per square inch and a mean unit pressure of 610 Ib. per square inch. The crankpin bearing pressure for the Liberty engine has a maximum value of 4,980 Ib. or 932 Ib. per square inch of projected area; the mean unit bearing pressure is 642 Ib. per square inch. The crankpin pressures used in the German engines are somewhat lower, ranging from a mean unit bearing pressure of 402 Ib. per square inch in the Austro-Daimler to 585 Ib. per square inch in the Maybach. The crankpin pressure is mainly due to inertia and centrifugal forces the gas pressures have com- paratively little effect. This is evidenced by the fact that the wear on crankpins bearings is on the side remote from the piston, that is, on the side subjected only to inertia and centrifugal forces. Crankshafts are subjected to stresses which vary rapidly in sign and magnitude and consequently are especially liable to fail from fatigue of material. The weakest point is generally at some place where there is a sudden change in cross-section and poor distribution of stress. It is particularly important that the fillets at the junctions of the crankpins and journals with the crank webs should be of adequate size. Tests to determine the desirable size of the fillet have been conducted recently in England; they show that the steel is materially weakened if the fillet is less than % in. radius. In the discussion of torque on page 47 it has been shown that the maximum torque at the propeller end of the crankshaft of a six-cylinder engine is actually less than the maximum value at the rear crankpin. Consequently there is no need for any increase in diameter of the crankshaft from rear to front in that 10 146 THE AIRPLANE ENGINE case. The free end of the crankshaft is subjected to much more severe conditions than the propeller end. The free end is, as it were, wound up when the maximum torque is applied to it and released when the torque diminishes. At certain speeds this alternate winding up and release may coincide with a natural period of vibration of the crankshaft, and in that case the shaft will vibrate excessively and the reciprocating masses attached to it will also vibrate and impart their vibration to the whole structure. Such torsional vibration could be reduced by the use of a flywheel on the free end. It has given much trouble in various six-cylinder engines and has been largely responsible for the failure of engines with eight crank throws. With six-cylinder engines of 200 to 300 h.p. the freely vibrating shaft has a frequency which is usually about 6,000 vibrations per minute; for four-cylinder engines this frequency is higher, and in single-crank radial or rotary engines it may be as high as 20,000. The period of vibration can be determined by striking a series of light blows at regular intervals; the vibrations will increase markedly when the frequency of the blows coincides with the natural period of the shaft. In a six-cylinder engine the impulses are three per revolution so that the dangerous speed for a shaft with vibration frequency is 6,000 per minute of 6,000 -f- 3 = 2,000 r.p.m. The next most dangerous speed would be 1,000 r.p.m. With eight-cylinder engines the dangerous speeds would be J^ an d M the vibration frequency. The addition of counterweights to the crankshaft as a means of obtaining rotary balance is of value mainly in reducing bearing pressures. The centrifugal force arising from unbalanced rotating weights acts radially from the center and produces con- siderable bearing pressures. Under airplane-engine conditions a lower total weight is obtained by omitting counterbalances and giving the bearing sufficient area and stiffness to support the centrifugal forces. With higher speeds of rotation the need for counterbalance weights increases. Crankshafts are drop forgings and are usually made in dies when the quantity warrants it; in other cases they are cut from large billets. The dies may be made of cast iron when the number of forgings required is small; for quantity production they are of steel. Strength of Shafts. Shafts should be designed for their strength in shear. For a solid circular shaft of diameter d in. ENGINE DETAILS 147 subjected to a bending moment M in pound-inches and a torsion T, also in pound-inches, and with a maximum permissible intensity of shearing stress at the outer surface of the shaft of / Ib. per square inch, d 3 = 5.1 For a hollow shaft of outside diameter d 2 and inside diameter the equation becomes d* = 5.1 FIG. 110. Propeller hub of Hispano-Suiza engine. With M = 24,500, T = 54,000, and / = 16,000, these equations give d = 2% in. and if d 2 is assumed to be 3 in. di is 2.275 in. Under these conditions the hollow shaft will weigh 56 per cent as much as the solid shaft. A hollow shaft of still larger outside diameter would be lighter and stiffer but would require larger and heavier bearings and would result in increased rubbing velocities at the bearings and increased friction. Propeller hubs in American practice are mounted on a tapered extension of the crankshaft. The Hispano-Suiza hub (Fig. 110) is a good example of standard practice. It is keyed to the engine shaft, which is given a taper of 1 in 10, and is threaded at the end to receive a long nut which is used for forcing the hub on the taper. The inner flange is integral with the tapered hub; the outer flange has splines which fit in grooves on the outer end 148 THE AIRPLANE ENGINE of the hub and permit axial movement of about 1 in. for adjust- ment to the thickness of the propeller, which is held between the two flanges. Eight bolts hold the flange and the propeller together. Rotation of the hub on the engine shaft is prevented by a key. The long nut is held in position by a locknut or a locking pin. The Benz engine (Fig. Ill) employs a hub which is bolted to a flange at the end of the crankshaft and has an outer flange which fits on the splined end of the hub. Crankcases. The crankcase has to serve several functions: it has to tie various parts of the engine together; it has to with- stand stresses due to gas pressures, and bending moments due to unbalanced forces; it contains the lubricating oil; it supports FIG. 111. Propeller hub of Benz engine. various auxiliaries; and it has to support the engine as a whole. The stresses in the crankcase are chiefly of the two kinds sug- gested above. The explosion pressure, acting on the cylinder head, puts the cylinder under tension and this should be supported as directly as possible by connecting members from the cylin- der to the corresponding lower main crankshaft bearings. In vertical engines this can be accomplished very satisfactorily by the use of through bolts from the lower cylinder flange to the lower bearings (see Fig. 112), using transverse webs in the upper crankcase as distance pieces. In Vee and W engines this construction is not possible and it is necessary to fasten the cylinders and the lower bearings to the upper crankcase, and transmit the tension through the transverse webs to the lower bearings. The lower half of the crankcase is sometimes cast as a unit with the lower bearings but this practice has nothing to recommend it. The assembly of cylinders and upper crankcase should sustain all the stresses due to gas pressures. ENGINE DETAILS 149 The unbalanced centrifugal and inertia forces acting through the main bearings subject the crankcase to bending moments which change continuously in direction and magnitude. It is necessary that the crankcase should have sufficient stiffness to withstand these bending moments without objectionable deflec- tions. For this purpose a webbed box structure has been found most satisfactory. It is easily possible to obtain the necessary stiffness by utilizing the upper crankcase only as a stressed member. The lower crankcase is preferably used only as an oil FIG. 112. Transverse section of crankcase. container and a support for oil pumps and other auxiliaries. This practice can be seen in the Hall-Scott L (Fig. 64), Bugatti (Fig. 67), Curtiss K and OX (Figs. 55 and 59), Napier "Lion" (Fig. 72), Maybach (Fig. 81), and Benz engines (Fig. 78). In other engines the lower crankcase carries also the lower halves of the end main bearings, as in the Hispano-Suiza (Fig. 51) and Sid- deley "Puma" engines (Fig. 9). These arrangements permit of easy accessibility. In the Liberty engine (Fig. 47) the lower crankcase has transverse webs and the crank chamber is divided into six separate chambers; a similar construction with double transverse webs is employed in the Fiat engine (Fig. 76). 150 THE AIRPLANE ENGINE The transverse webs are often cut away in places for lightness. Aluminum alloy is universally used for crankcases. The lower crankcase serves as an oil sump. The earlier practice of keeping a considerable body of oil (wet sump) in the bottom of the crankcase is now being superseded by the dry sump which is. kept drained by a scavenger pump or pumps. Wet sumps are shown in the Benz engine (Fig. 78), Hispano-Suiza (Fig. 51), Hall-Scott L (Fig. 64), and Curtiss OX engines (Fig. 59). In the last case the sump is separated by drainage plates from the rest of the crankcase. The advantage of the dry sump is in avoiding drowning the cylinder with oil in case the engine operates momentarily upside down or in any posture approximat- ing that position. The drainage point of the sump is usually in the middle, but in some cases the scavenger pumps take oil from both ends (Liberty, Napier "Lion 7 ')- Oil cooling is carried out in the lower crankcase in the Austro-Daimler engine by casting outside cooling ribs running longitudinally along the bottom of the crankcase and attaching a sheet of aluminum in such a way as to form an air duct along the whole underside of the engine. In the Basse-Serve engine an oil cooler with air tubes is fastened to the bottom of the crankcase but has no direct com- munication with the inside. In the Curtiss K engine (Fig. 55) oil cooling is effected inside the crankcase by the jacket water on its way to the pump; this arrangement serves also to heat up the oil quickly after starting the engine and puts the lubrication system into normal operation earlier than would otherwise be possible. German airplane engines usually have provision for air cooling of the crankcase. In the Benz engine (Fig. 77) the support- ing webs for six of the main bearings form air passages trans- versely across the engine. Two of these serve as air intake passages to the two carburetors, which are thereby supplied with heated air. In addition the lower crankcase is traversed by 18 aluminum tubes, 30 mm. in diameter; air is scooped into the tubes through an aluminum louvered cowl on one side of the engine and discharged through a reversed cowl on the other side. CHAPTER VII VALVES AND VALVE GEARS Location of Valves. The diversity of valve locations which characterizes automobile practice is not found in modern airplane engines. L-head and T-head arrangements have been supplanted by overhead valves which permit the simplest form of cylinder head and watex jacket, shortest and most direct passage of the gases, and, with overhead camshafts, a considerable simplifica- tion in the valve gear and a reduction in the number of rubbing or contact points. The valves may either be seated in a flat head in which case the stems are parallel to the cylinder axis or the head may be domed, in which case the valve stems are inclined to the cylinder axis. Valve Lift. Valves are always of the poppet type with bevelled seats. They are opened by cams which operate either directly or indirectly; they are closed by springs. The valve (Fig. 113) has a face which is usually about 25 per cent wider than the seat on which it closes and a stem which passes through a long guide (often provided with a bushing) and which connects with the valve by a rounded fillet. The bevel of the valve face is usually 45 deg. but 30 deg. is sometimes used; The width of the valve face must be small to ensure gas tightness and is usually about one-fourth the lift of the valve. The width of the valve seat is usually less than 0.1 in. The free area for the passage of the gas through the fully opened valve may be taken approximately as irdh where d is the smallest diameter of the bevelled valve face and h is the lift. This area should not be greater than the free open- ing through the valve seat. Neglecting the area occupied FIG. 1 13. Typical airplane engine valve. by the valve stem, irdh = d 2 , or h 151 gives the lift which 152 THE AIRPLANE ENGINE makes the gas passage area equal to the free opening through the valve; usually h varies from one-fifth to one-sixth of the outside valve diameter. Values will be found in Table 4. With a valve lift of one-quarter of its diameter the gas flow for a given pressure drop is found by experiment to be about 67 per cent of the flow through an unobstructed port 1 ; with a lift of one-half the diameter this is increased to from 80 to 90 per cent. These " coefficients of efflux" are found to be the same for all pressure drops, and for valves of different sizes at equal lifts expressed in per cent of their respective diameters. The experi- ments were carried out with continuous flow which presumably would give results differing from those actually occurring under the operating conditions of intermittent flow. The earlier in- vestigations of Lucke 2 indicate coefficients of efflux lower than those given above ; the variation with lift is probably of the right order of magnitude. The volume of gas passing the inlet valve per unit of time is approximately equal to the piston displacement in that time; the volume of gas passing the exhaust valve is from two to three times as great. If the mean piston speed during the suction stroke is s in. per second and the piston diameter is D in. the mean gas velocity V in feet per second past the valve is given by ~D 2 -s = 12V 1%> and 1 in.), gave the results shown in Fig. 118. It is noteworthy that the per- formances of these three valves were practically identical up to 1,600 r.p.m., above which speed the smallest valve showed inferior results and the intermediate size valve showed best results. It is seen by comparing Figs. 117 and 118 that valve area alone is not important; it is necessary to know the lift-diameter ratio also. The highest curve in Fig. 118 is obtained with a lift of 23.6 per cent of the diameter; in Fig. 117 the highest curve has a lift- diameter ratio of 18.7 per cent, but it is probable that still better results would have been obtained with a higher lift. Valve Materials. Inlet valves in airplane engines under normal operation may reach temperatures of over 1,100F.; exhaust valves may go to 1,600F. or higher. The heat received by the head of an exhaust valve is dissipated in three ways: (1) by conduction down the stem to the guide, (2) by direct radiation from the back surface of the head and (3) by direct conduction from the face to the valve seat. The last of these is by far the most important. To be effective it is essential that the valve should have good metallic contact with its seat through- out the whole of the explosion stroke. If, through valve warping or the lodging of scale on the seat, there should be any leakage of gas past the valve, there will be rapid heating at the place where the leakage occurs and the valve will burn away at that place. Another prolific cause of valve burning is persistent preignitions in the cylinders; it is found that valves which stand up satis- factorily under normal operation fail very rapidly when per- sistent preignitions occur; with such preignitions the temperature of the exhaust valve may rise to 2,100F. If the exhaust ports are so designed that the exhaust gases play directly on the neck of the valve this may become highly heated and may actually supply heat to the valve head instead of taking it away; in such a case overheating of the valve is likely to occur. It is also impor- tant that the valve guides should be efficiently water-cooled and 160 THE AIRPLANE ENGINE should not project into the exhaust pocket so as to be heated directly by the exhaust gases. A final cause of overheating the exhaust valve is the use of an overrich mixture which may be still burning during the exhaust stroke. An interesting suggestion for valve cooling is the use of a hollow stem into which is put a small amount of mercury before plugging. The liquid mercury in contact with the hot center of the valve head is vaporized and is condensed again in the upper part of the stem. The mercury thus acts as a heat carrier abstracting from the valve head its latent heat each time it is vaporized. The vapor pressure of mercury at 820F. is 50 Ib. per square inch. The principal types of valve failure are (1) elongation of the stem, (2) distortion of the head, (3) cracks in the valve face, (4) wear of the stem, (5) wear of the foot, (6) burning of the head, (7) scaling and (8) breaking due to self -hardening. Elon- gation of the stem results either from the use of a steel of insuffi- cient strength at the working temperature or from overheating of the stem. Distortion of the head occurs usually when proper heat treatment has not been given to the valve forging before machining; in other cases unequal heating or softening under the action of high temperature may be the causes. Cracks come usually from cracks in the steel from which the valves are made; they are fairly common and are dangerous as they may result in the breaking away of a section of the valve. Wear on the valve stem occurs usually in rotary engines which produce a side pressure due to the inertia of the valve. Wear of the valve foot results from the hammering of the tappet or the wipe of the cam; it is diminished by hardening the foot or by the use of a cap. Burning is due to overheating. A steel to be satisfactory for exhaust valves in airplane engines should have the following properties as stated by Aitchison. 1 1. The greatest possible strength at high temperatures. 2. The highest possible notched bar value (resistance to impact). 3. The capacity of being forged easily. 4. The capacity of being manufactured free from cracks. 5. The capacity of being easily heat-treated. 6. The least possible tendency to scale. 7. The ability to retain its original physical properties after repeated heatings for prolonged periods. 8. Freedom from liability to harden on air cooling. 1 The Automobile Engineer, Nov., 1920. VALVES AND VALVE GEARS 161 9. Freedom from distorting stresses after heat treating. 10. Hardness to resist stem wear. 11. Capacity of being hardened at the foot. 12. Reasonable ease of machining. The best steels for exhaust valves are in five classes: 1. Tungsten steel with not less than 14 per cent tungsten and about 0.6 per cent carbon. 2. High chromium steels (stainless steel) with about 13 per cent chromium and about 0.35 per cent carbon. 650 150 850 950 Temperature , Decj. C- FIG. 1 19. Resistance of valve steels to scaling. 3. Steel containing from 7 to 12 per cent chromium and about 0.6 per cent carbon. 4. Steels containing about 3 per cent nickel. 5. Ordinary nickel-chromium steels. Of these steels the first four are superior to the last. The nickel-chromium steels are difficult to manufacture free from flaws, they tend to harden during the running of the engine, they scale rapidly and they show no superiority at high temperatures over the other steels. The relative resistances to scaling are shown in Fig. 119, from which it is apparent that stainless steel is superior to the others. The tensile strengths of these steels at higher temperatures are given in the following table. 11 162 THE AIRPLANE ENGINE ULTIMATE STRENGTH OF VALVE STEELS, POUNDS PER SQUARE INCH Steel Temperature, degrees Fahrenheit 1,300 1,650 High tungsten, high carbon 39,600 19,700 High tungsten, low carbon 34 700 14,100 High chromium, high carbon 33,800 16,800 High chromium, low carbon 27 100 10,700 Low chromium, high carbon 41,400 16,800 Low chromium, low carbon 38 , 000 15,800 3 per cent nickel, high carbon 25,800 10,100 3 per cent nickel, low carbon . . 21,000 8,700 Nickel chromium 23,500 10,100 The 3 per cent nickel steel is much cheaper than the others but is markedly inferior in tensile strength at high temperatures and consequently should be used only on inlet valves or for the exhaust valves of rotary engines. The high chromium (stainless) steel is highly resistant to scaling and, if of low carbon content, is readily machined but is not easy to forge and is liable to cracks. High tungsten steel retains its strength best of any steel at high temperatures and is fairly resistant to scaling. Exhaust valves which are liable to be subjected to unusually high tempera- tures should be of tungsten steel; for more moderate tempera- tures stainless steel will be more durable. Monel-metal valves have been used, and although they have stood up well under test on the Hispano-Suiza engine, they have failed rapidly on the Liberty engine. Valve Operation. The valves of modern airplane engines are mechanically operated; automatic action, which is found in some automobile engines and in a few of the earlier airplane engines, must always result in lowered volumetric efficiency and capacity. Actuation of the valves is by means of cams acting either directly or indirectly. The camshafts may be placed near the base of the cylinders and operate the valves through push rods and rocker arms, or overhead camshafts may be used acting on the valves directly or through rocker arms. VALVES AND VALVE GEARS 163 A good example of push-rod operation is shown in the Benz 230 engine (Fig. 78), which has separate camshafts for the inlet and exhaust valves, both located in the crankcase; a similar arrangement is used in the May bach engine (Fig. 81). In Vee engines the usual practice, where push rods are employed, is to have a single camshaft, located in the angle of the Vee, inside the crankcase, carrying inlet and exhaust cams for both rows of cylinders; the Curtiss OX and V2 engines (Figs. 60 and 62) show this arrangement, which is also used on the Benz 300 (Fig. 131). In recent years the tendency has been to do away with push rods and to use overhead camshafts. This last arrangement reduces the weight and complexity of the valve gear, and, in consequence of the smaller number of joints involved, makes for better maintenance of the valve timing. There may be either (1) one camshaft over each row of cylinders acting directly on the valves as in the Hispano-Suiza (Fig. 51), or (2) one camshaft acting directly on one set of valves and indirectly through a rocker arm on the other set as in the Siddeley "Puma" (Fig. 8), or (3) one camshaft acting indirectly through rocker arms on both sets of valves as in the Liberty (Fig. 47), Basse-Selve (Fig. 129), Bugatti (Fig. 67), Fiat (Fig. 76), Rolls-Royce (Fig. 70), Mercedes, Lorraine-Dietrich, Renault, Austro-Daimler, or (4) two camshafts acting directly on the two sets of valves as in the Curtiss K (Fig. 56) and Napier "Lion" (Fig. 73). Cams. The shape of the cam depends on the desired valve movement and on the form and location of the cam follower. FIG. 120. Individual cam. (a) (b) (c) FIG. 121. Cam forms. The cams are usually integral with the camshaft, which gives maximum security and accuracy of location, but sometimes they are fastened by taper pins to the hollow camshaft, as in Fig. 120; this arrangement permits more satisfactory hardening of the cams and the replacement of a worn cam but is less secure and may become slack. The cams are sometimes made with convex flanks as in Fig. 1216, or with flat surfaces tangential to circular 164 THE AIRPLANE ENGINE arcs as in Fig. 121a, or with flanks that change from concave to convex and a top which is concentric with the camshaft as in the constant acceleration cam of Fig. 121c. The cam follower may be flat as in Fig. 122a, rounded as in Fig. 1226, or a roller as in Fig. 122c, and it may be fixed on the end of the valve plunger or it may be mounted on a radius rod as in Fig. 122cL With a flat follower a convex flanked cam is used; tangential and constant acceleration cams are used with the other types of follower shown in Fig. 122. The work which a cam has to do is in three parts. (1) It must overcome the difference of gas pressures on the two sides of the valve. This pressure difference is important only in the case of the exhaust valve, which just previous to opening has a pressure FIG. 122. Cam followers. of about 60 Ib. per square inch (gage) on one side and atmospheric pressure on the other. With a valve 2% in. diameter the pres- sure difference is nearly 300 Ib. at the instant of opening and falls rapidly to a negligible quantity. (2) The valve spring is operat- ing at all times to keep the valve closed; the compression on the spring is usually about 50 Ib. when the valve is closed (see table 4) . The cam must do work in compressing the spring. (3) The moving parts from the cam to the valve, including the valve and spring, must be accelerated and work must be done in giving them the necessary acceleration. The force required to acceler- ate these parts is determined by the design of cam and follower and by the masses that have to be accelerated. For maximum volumetric efficiency of the engine the valves should open promptly, should remain wide open as long as possible and then should close promptly. If a valve is to be opened in a given time (number of degrees of crankshaft rotation) VALVES AND VALVE GEARS 165 the force required to accelerate the moving parts will be kept a minimum by making the acceleration constant and thereby keeping the accelerating force constant. During the opening the moving parts must be first accelerated and then brought to rest; the deceleration is accomplished by the valve spring and, sometimes, if a push rod is used, by an additional spring acting on the push rod. Smooth action will be obtained when the deceleration is constant and has the same value as the accelera- tion. The cam does not necessarily do any work at all during the decelerating period. The acceleration and the force required to produce it are readily calculable. Suppose the valve to move from the closed to the fully open position in 60 deg. of crankshaft rotation and that the moving parts are accelerating uniformly for half that period, or 30 deg. ; and are decelerating uniformly for the next 30 deg. Then at 1,500 r.p.m. the time, t, available for acquir- 60 30 1 ing maximum velocity is - X TT = sec. If the lift is 0.5 in. the distance, d, moved in this time is 0.25 in, and the acceleration, a, is given by d = ~at 2 , or a = 3,750 ft. per second A per second. If the weight, w, of the moving parts is 1 Ib. and if all of the parts move with the same velocity as the valve, the force required to accelerate the parts will be a = 110 Ib. y The force exerted by the cam will be greater than this by an amount equal to the valve spring compression and, at the instant of opening the exhaust valve, by the gas-pressure difference. With the numerical values given above the force exerted by the exhaust cam at the moment when the valve begins to open must be 110 + 300 + 50 = 460 Ib. As there is always some tappet clearance to permit expansion of the valve stem without forcing the valve to lift, this maximum force occurs a short time after the cam has come into action. The force will diminish rapidly as the gas pressure in the cylinder falls but will tend to increase later with increasing spring compression. During the decelerat- ing period the spring pressure would have to be greater than 110 Ib. to bring the valve to rest in 30 deg. of crank rotation. During the closing of the valve the gas pressure difference is absent, the acceleration is due to spring action and the deceleration is brought about by pressure on the cam. 166 THE AIRPLANE ENGINE The forces exerted by the cam, which have just been considered, are the radial forces, R, acting along the push rod, or, in the case of overhead camshafts, at right angles to the outer end of the rocker arm. The actual pressure, N, between the cam and its follower acts normal to the surface of contact and will be greater than the radial force throughout the accelerating period. The relation between these two forces is indicated by the triangle of forces in Fig. 1226. The side thrust, S, may be trouble- some. The quicker the opening of the valve the greater will be the acceleration force, R, and the greater will be the ratio of both N and S to R. In other words, the normal pressure and the side thrust increase much more rapidly than the radial force. The side thrust is particularly objectionable with valve plunger guides as in Figs. 1226 and c; with the arrange- ment of Fig. I22d, or with the cam operating directly on the rocker lever, a rapid operjing of the valve can be obtained without trouble from side thrust. A good example of the constant acceleration type of cam with roller follower is shown in the Maybach engine, Fig. 123. The displacement, velocity, and acceleration curves both for inlet opening and for exhaust closure are given in Fig. 124; it will be seen that the valves open and close rapidly, and remain full open for considerable periods of time. The velocity of the exhaust valve when closing increases uniformly for 60 deg. of crank rotation, then decreases but not quite uniformly for the next 46 deg.; the inlet valve on opening has acceleration which increases for about 48 deg., when maximum velocity is attained, and then comes to rest after 60 deg. more of uniform deceleration. Valve displacements for the whole cycle are shown in Fig. 125; it will be seen that the valves are wide open for considerable fractions of the stroke. The tangential cam is used in the Liberty engine (Fig. 130) operating on a roller at the end of the rocker arm. With this type of cam, the center of curvature of the highest part of the cam cannot coincide with the center of the camshaft (as in the constant acceleration cam), and consequently the valve cannot FIG. 123. May- bach valve gear. VALVES AND VALVE GEARS 167 stay at its wide-open position. The actual valve lifts are shown in Fig. 126 plotted against crank position, and in Fig. 127 plotted against piston position. The valve opening is not so good as in the Maybach engine, but the forces required to acceler- Exhaust DISPLACEMENT CURVE Inlet Crankshaft Degrees 120 110 100 90 80 70 60 50 40 30 20 10 10 20 SO 40 50 60 "70 80 90 100 110 120 BO 140 150 *~X/7Q 14 >>s 1 , i ^-* [yCQ A\ 1 ^f ^ , ijch&usrr 1 V ELOC1T> \ N ' CURVE I Inlc *I 1 Q> - A ^ \f) V - >^>y U) l_ rxS Kfe ^ s Jr **+^e* - 4 u /*> - -7 - 4- uji^rn j _^ 5hl rns^: ^>n^ i **"*" _ s -n i r>^ - - n 14 - Exhau&r ACCELERATION CURVE InJat $ 1600 -- ::::E 1U - ^ --1600^ ZX/JC +, * r _ _ _ _ son h ** * uu g_ ij o J - T fe 800 T\ jr - 800^ ^^ * 1600 ^- "1600 \ \ 120 HO 100 90 80 70 60 50 40 30 20 10 10 20 30 40 50 60 70 80 90 100 110 120 130 MO 150 Crankshaft Degrees FIG. 124. Displacement, velocity and acceleration curves for the valves of the Maybach engine. ate the moving parts are less in consequence of the longer time available for opening or closing the valve; with symmetrical cams this time is one-half the total time the valve is open. In the Maybach engine the intake valve is open for 223 deg. of crank rotation and the valve, while opening, is accelerating for t'xfa 140 m 180 MO 120 110 100 90 80 70 60504030 304050 60 70 80 90 100 IK) 120 130 MO 180 220 240250 Diagram of Exhaust and Inlet Port Opening in Relation to Piston Position FIG. 125. Valve openings of Maybach engine. 29 deg.; in the Liberty engine the inlet valve is open for 215 deg. and the valve, while opening, is accelerating for 54 deg. of crank rotation. As the acceleration is inversely as the square of the time taken to lift the valve through a given distance, the force 168 THE AIRPLANE ENGINE required to overcome the inertia of the moving valve parts would be 3.5 times as great for an engine using 29 deg. of crank rotation for the valve acceleration as for the same engine, with the same revolutions per minute, using 54 deg. of crank rotation. Valve Springs. The function of the valve spring is to deceler- ate the valve moving parts during the latter half of the valve opening and to accelerate them during the first half of the valve 0.30 ui c s^~ - ^ ^ _vj / \ c$ CJ / \ vi ! / EXHAUS. foil Opening * 236 \ t< . 1^ 4- <*- 'Jj 0.10 0.02"-"^ Clearance 0.40 0.30 0) c "0.20 t J] 0.10 0.015** Clearance - ? 1 \ Vj^ gf ? 8 i 1 ^ ' 9 i i \ u> I i V 1 k/ \ ^ -^0 I&O 200 240 280 320 ^~ / S~ N A / \ / \ vj cs v< I \^ >^ 0) | 40 30 20 10 Gross M.E.P. 95 90 75 70 800 1300 900 1000 1100 1200 Revolutions per Minute FIG. 140. Performance curves of Le Rhone 80. overrun them one complete revolution in 10 revolutions of the engine. Each cam has five lobes. Each inlet and exhaust valve should be opened once only in two revolutions of the engine, and this will be accomplished when the engine overruns the cam one- fifth of a revolution. As the engine overruns the cams one- tenth of a revolution each engine revolution it is evident that two revolutions of the engine are required to complete the opening and closing of all the valves. The action of centrifugal force on the valve-actuating* rod causes it to press continuously against the valve rocker lever. At low speeds of revolution this force may not be sufficient to open the exhaust valve at the desired time and consequently an exhaust cam is necessary to push the valve rod out. At high speeds the operation of both valves can be taken care of by the 188 THE AIRPLANE ENGINE RADIAL AND ROTARY ENGINES 189 inlet cam if it is properly shaped for that purpose. The valves are brought back to their seats by spiral springs at low speeds; at high speeds centrifugal force closes the valves. Performance curves for the 80-h.p. Le Rhone are given in Fig. 140. Clerget. The Clerget rotary engines are built with 7, 9 or 11 cylinders. The 130-h.p., nine-cylinder engine shown in longi- tudinal section in Fig. 141 is 120 mm. bore, 160mm. stroke, makes Cam Gear Box Exhaust Tappet Guide: Inlet Tappet. Locking Hut. ExhaustCam. LockincjNut. -Oil Hole. Inlet Cam. Cam6egr Cover..--' FIG. 142. Cam gears of B.R. 2. 1,250 r.p.m., weighs 381 lb., develops 135 h.p. and has a compres- sion ratio of 4. Its points of difference from the previous engines include the use of an aluminum piston, tubular connecting rods, inlet and exhaust valves operated by means of separate cams, tappets and rocker arms, and a double-thrust ball race which is a pure thrust bearing and distinct from the combined thrust and radial bearings of the other engines. The inlet and exhaust cam plates are driven at nine-eighths the engine speed by separate internally-toothed gears mounted 190 THE AIRPLANE ENGINE inside and keyed to the cam-gear case. These mesh with external gears mounted eccentrically on the crankshaft; the cams are attached to these external gears. This arrangement is the reverse of that used on the Le Rhone engine. The cam plates overtake the engine once in eight revolutions. Each cam plate has four lobes so that in eight revolutions each tappet will be lifted four times, or once in two revolutions. A sectioned per- spective view of the similar cam-gear box of the B.R.2 rotary engine is shown in Fig. 142. The four cams on each gear are simply rearward extensions of every fourth tooth. The valve timing differs in some respects from that of the Gnome and Le Rhone. At top center of the suction stroke no 1150 1200 1250 1300 Revolu+ions per Minute FIG. 143. Performance curves of Clerget 130. 1350 both exhaust and inlet valves are open. The inlet opens 5 deg. before top center; the exhaust closes 5 deg. past top center. The inlet remains open till 58 deg. past bottom center (or a total of 153 deg.) and compression begins. Ignition is at 25 deg. before top center and exhaust begins 68 deg. before bottom center. The carburetor is located at the rear end of the hollow crank- shaft. Fuel is injected under air pressure through a jet which is controlled by a needle valve. The air supply is controlled by a cylindrical throttle valve. Equal movement of both throttle lever and needle valve lever controls air supply only. Operation of the throttle lever alone controls both air and fuel. The charge -entering the crank passes to the annular inlet chamber at the rear of the crankcase and then by the separate inlet pipes to the cylinders. The connecting rod assembly is similar to that of the Gnome engine (see p. 203). RADIAL AND ROTARY ENGINES 191 The performance curve for this engine (Fig. 143) is typical of rotary engines. The effective horse power goes through a maximum at 1,250 r.p.m. but is very flat for a considerable range of speed. The rapidly increasing difference between the effective horse power and the indicated horse power is due to the rapid increase in the air-churning resistance. B.R.2 Engine. The British B.R.2 engine is one of the largest air-cooled rotary engines. In general construction it is similar to the Clerget. The cylinders are of aluminum with steel liners and steel head. The cylinder diameter is 140 mm., stroke 180 mm., compression ratio 5.01, brake horse power 230 at 1,300 r.p.m., weight dry 498 lb., weight per brake horse power dry FIG. 144. Thrust box of B.R.2. 2.16 lb. The cam gear for this engine is shown in Fig. 142 and follows exactly the same principle as the Clerget cam gear. The thrust box contains two ball bearings and a thrust bearing which differs from the Clerget in having one row of balls only. This single-thrust bearing is adapted both for pusher and tractor use as indicated in Fig. 144; a very small clearance is left for the travel of the crankcase along the crankshaft when changing from pusher to tractor. Double Rotary. The double-rotary engine has cylinders revolving in one direction while the crankshaft revolves in the other direction. The effective speed is the sum of the two speeds so that the power of an engine in which both cylinders and crank- shaft revolve at 900 r.p.m. is the same as that of a radial or rotary engine of the same dimensions operating at 1,800 r.p.m. Such a speed is permissible in radial engines but would give 192 THE AIRPLANE ENGINE excessive air-churning resistance in a rotary. There is no reason why even higher effective speeds up to 2,400 r.p.m. may not be practicable with this type, if the volumetric efficiency of the engine can be maintained and if the cylinders can be kept cool enough. In any case this arrangement leads to a combination of high engine speed and low propeller speed with consequent high propeller efficiency. It has important advantages over all other types in (a) the possibility of the elimination of unbalanced Carburetor FIG. 145. Longitudinal section of Siemens-Halske double rotary. gyroscopic effects, which is an advantage for maneuvering, and (&) the elimination of the unbalanced turning moment exerted by the engine on the plane. This unbalanced turning moment is a constant, though small, power drag on the plane and its elimination is a distinct advantage. The only engine of this type which has been in production is the Siemens-Halske 11-cylinder engine (Fig. 145), which was brought out in 1918 and develops 200 h.p. at 900 r.p.m. of both cylinders and crankshaft, or a virtual speed of 1,800 r.p.m. RADIAL AND ROTARY ENGINES 193 The single propeller is mounted on a nose attached to the revolv- ing crankcase; the torque of the crankshaft is transmitted to the crankcase by securing a bevel wheel to the crankshaft and a similar gear, facing it, to the crankcase and mounting an inter- mediate pinion between the two on a stud which is fastened to the stationary cylindrical housing at the rear of the engine. The carburetor is mounted on a stationary hollow extension of the crankshaft. The combustible charge is drawn in through the hollowcrank shaft to the crankcase and goes from the annular inlet chamber at the rear of the crankcase to the individual inlet pipes. The inlet and exhaust valves are operated through two cam plates which are loose on the crankshaft and are rotated through double reduction gears from an internal gear attached to the crankcasing. The engine is supported by steel rods both before and behind the cylinders. The weight of the engine complete is 427 lb., which at 240 maximum h.p. gives a weight of 1.78 lb. per horse power. The fuel economy is as good as with stationary engines and is much better than with other rotaries. Other designs of double rotaries with two propellers (right- and left-hand respectively) forward of the cylinders, attached to the crankshaft and crankcase respectively, have not passed the experimental stage. The efficiency of a pair of propellers close together but operating in opposite directions has been found to be but little inferior to that of a single propeller. There is consequently the possibility of the development of a satisfactory double-rotary engine on the lines indicated. Radial Engines. In a fixed radial engine the cylinders are stationary and the crankshaft revolves. Three to eleven cylin- ders can be accommodated in a single row or bank, but two rows with a two-throw crank must be adopted if a larger number of cylinders is desired or if it is necessary to cut down the over-all diameter. The two-throw crank eliminates the need for counter- balance weights but increases the length of the engine and intro- duces difficulties in the air cooling of the rear row cylinders. Radial engines offer certain special construction problems. The most important are the balancing of the masses at the crank- pin and the avoidance of excessive pressures on the crankpin. It is possible to operate radial engines at speeds as high as those used in vertical and Vee engines, but special care must be taken to prevent the overheating of air-cooled cylinders and the over- is 194 THE AIRPLANE ENGINE loading of the crankpin. There is no fundamental reason why the mean effective pressure and economy of radial engines should not be as good as those of any other type. A B C. The development of air-cooled fixed radial engines has been carried on in England more -than elsewhere. The FIG. 146. Sectional outlines of A B C " Dragonfly." Radial engine. ABC engines, built by the Walton Motors Co., have the following general characteristics : Type name Gnat Wasp Dragonfly Number of cylinders (copper-coated steel fins) . 2 7 9 Bore, inches 4 75 4 75 5 5 Stroke, inches 5 5 6 25 6 5 Normal brake horse power Revolutions per minute Oil consumption, pints per hour . . Gasoline per brake horse power hour, pints 45 1,800 1.7 56 . 200 1,800 4 56 340 1,650 7 56 Weight of engine, dry, pounds 115 320 600 Weight per brake horse power, pounds 2.3 1.6 1 765 Over-all diameter, inches 42 7 50 5 RADIAL AND ROTARY ENGINES 195 The Wasp and Dragonfly engines have each two exhaust valves and one inlet valve per cylinder. Their engines use the master- rod connecting-rod assembly with roller bearings (see p. 207) and have counterbalance weights. Sectional views of the Dragonfly engine are shown in Fig. 146. Cosmos. The Cosmos Engineering Co. has fixed radial engines with the following characteristics: Type name Lucifer Jupiter, direct drive Jupiter, geared Mercury Hercules, geared Number of cylinders . . . Number of rows 3 1 9 1 9 1 14 2 18 2 Bore, inches 5.75 5.75 5.75 4 375 6 25 Stroke, inches Normal brake horse power 6.25 100 7.5 400 7.5 450 5% 315 7.5 1,000 Brake mean effective pressure, pounds per square inch 113 113 Revolutions per minute. Propeller speed, revolu- tins per minute 1,600 1,650 1,850 1 200 1,800 1,750 1 150 Weight of engine, dry, pounds 220 636 757 587 1 400 Weight per brake horse power, pounds 2.2 1.59 1 863 1 4 Weight per brake horse power at maximum power, pounds 1 413 Over-all diameter, inch. 52.5 52.5 41.625 The Jupiter engine has two exhaust and two inlet valves; the Mercury engine has two exhaust and one inlet valve. Performance curves for the Jupiter engine are shown in Fig. 147. It will be seen that the brake mean effective pressure reaches a maximum of 117 Ib. per square inch at 1,700 r.p.m. A special feature of the Jupiter engine is the method of con- veying the explosive charge to the cylinders. There are three independent carburetors at the rear of the engine discharging into the cover of the annular inlet chamber which forms the rear of the crankcase. This chamber (Fig. 148) contains an alu- minum spiral casting which fits closely into the chamber. The 196 THE AIRPLANE ENGINE 560 540 520 500 480 460 440 420 ^400 1380 cti 360 340 320 300 280 260 240 220 200 147 s .--- ^ ~~* *~ _ < i ^ - . ^ IdU CL" z: y > / / / A / / > ? / / ': / / / * DOO 1200 1400 1600 1800 2000 2200 R.P. M. Performance curves of Cosmos "Jupiter" radial engine. 2 ^(^S 3 FIG. 148. Induction chamber of Cosmos "Jupiter.' RADIAL AND ROTARY ENGINES 197 casting constitutes a three-part spiral. The carburetors dis- charge into the spaces marked X, Y and Z respectively. The FIG. 149. Longitudinal section of Salmson radial engine. space X is part of the spiral marked AAA, so that the mixture drawn into X will flow along the spiral groove AAA. This 198 THE AIRPLANE ENGINE groove is opposite the inlet pipes for cylinders 2, 8 and 5; similarly the middle carburetor will supply cylinders 3, 9 and 6. This arrangement gives the mixture a clean sweep from the carburetor to the cylinder and isolates the cylinders in three groups so that should one carburetor fail to act properly there would still be six cylinders in normal action. Exhaust Valve ---fn/ef Valve FIG. 150. Transverse view of Salmson radial engine. Salmson. The Salmson (Canton-IInne*) engine is a good example of the water-cooled fixed-radial engine. Figure 149 shows a longitudinal section of a nine-cylinder engine; Fig. 150 is a transverse view of the same engine. The general dimensions of the engine are: bore, 125 mm.; stroke, 170 mm.; ratio of compression, 5.3; weight of engine without water or radiator, RADIAL AND ROTARY ENGINES 199 474 lb.; weight of water in jackets 20 lb.; power at 1,500 r.p.m., 250 h.p.; weight dry per horse power, 1.89 lb.; gasoline con- sumption per horse-power hour, 0.507 lb.; oil consumption per horse-power hour, 0.077 lb. The variation of brake horse power with engine speed is shown in Fig. 151. The cylinders are steel forgings 3 mm. thick; the jackets are of sheet steel welded to the cylinders. The inlet and exhaust valves are symmetrically located and are both 62.5 mm. dia- meter; they are held to their seats by rat-trap springs. The connecting-rod assembly is of the master-rod type (see p. 203) with ball bearings on the crankpin. The crankshaft is of the 300 Brake Horsepower ro ro r ro I 8 S si J s* " s^ ^ s s* s / ^ / " S / s / / / / S ^/ / 1300 1400 1500 1600 IKX Revolutions per Minute FIG. 151. Performance curve of Salmson radial engine. built-up type with counterweights. The valves are operated through push rods and rocker arms from a cam sleeve which is revolved on the crankshaft at one-fourth the engine speed by means of an epicyclic gear set. There are three pairs of cams on the cam sleeve, each pair at opposite diameters in its own plane. In each of the three planes are the cam followers of both valves for three cylinders. All the valves will be opened twice in one revolution of the cam sleeve. Consequently in two revolutions of the engine, or one-half revolution of the cam sleeve, each of the valves will have been operated once. The inlet valve opens at top center and closes 55 deg. after the bottom center, or at about 16 per cent of the^re turn stroke. Ignition occurs about 30 deg. before top center. The exhaust opens 65 deg. before bottom center and closes at top center. 200 THE AIRPLANE ENGINE The water circulation is shown in Fig. 152. A centrifugal pump taking water from the bottom of the radiator discharges it through two pipes into the heads of the two lowest cylinders. The top and bottom of each jacket is connected by pipes to the tops and bottoms respectively of the adjacent cylinders. The water is finally delivered from the top of the highest cylinder to the radiator. The carburetor (Zenith) discharges through long vertical pipes into the annular inlet chamber at the rear of the crankcase and thence through separate inlet pipes to the individual cylinders. Wafer Screen . Thermometer, -Radiator FIG. 152. Water circulation in Salmson radial engine. The exhaust passes from each cylinder into a sheet metal exhaust duct which encircles the engine, discharges at the sides of the fuselage, and is stream-lined to serve as a cowling for the engine. Details of Radial and Rotary Engines. Air-cooled cylinders are either made from solid steel, as in the Gnome, Le Rhone and Clerget rotary engines, or they are composite with steel barrel or liner and aluminum alloy head. All-aluminum cylinders have been tried with fair success but there is doubt of their durability; they are no lighter than the other types and their considerable longitudinal expansion increases tappet clearances and alters valve timing to a greater extent than with other constructions. RADIAL AND ROTARY ENGINES 201 EXHAUST The satisfactory operation of an air-cooled cylinder depends on keeping down its temperature. When overhead valves are used, this temperature is highest in the middle of the head. With open exhaust as in rotary engines and in some radial engines there is not much difficulty in arranging for adequate cooling of all parts of the cylinder. With overhead valves it is essential to make the cylinder head of the best available conductor (see p. 346), which in practice turns out to be an aluminum-copper alloy. The valve seats and the working surface of the cylinder barrel must be of some harder material. When an aluminum head is used the valve seats should consist of rings of steel or bronze, cast or expanded into position. Bronze seats, in consequence of their high coefficient of expansion, are less likely to come loose than steel seats. One type of construction is shown in Fig. 153. An aluminum casting forms the head and surrounds the greater part of the steel liner, which is shrunk into the casting at about 300C. Cylinders of this type have given excellent results, but the differ- ence between the coefficients of ex- pansion Of the Steel and aluminum FIG- 153. Aluminum air-cooled cylinder with steel liner. tends to cause separation of the liner and casing at working temperatures and a film of oil may work in between them. With cylinders below 4 in. in diameter there is little trouble. A shrinkage allowance of about 1 in 600 should be made. The expansion trouble can be overcome by the use of bronze liners if a bronze sufficiently hard to resist wear is developed. The holding-down bolts go through lugs in the aluminum casting. Screwed-in liners have not given good results owing to the impossibility of maintaining adequate contact between liner and casing. If the contact is good when cold, the difference of expansion when hot causes contact at points only. 202 THE AIRPLANE ENGINE The best method of composite construction is one with an aluminum cylinder head into which is cast or screwed a steel barrel with its own cooling fins (Fig. 154). This construction is mechanically sound and has been used successfully with cast-in barrels for sizes up to 6 in. in diameter and with screwed-in barrels up to 5^ in. in diameter. The length of the screwed portion should be about one-fourth of the cylinder diameter. The holding-down bolts grip a ring integral with the steel barrel and thereby avoid the breakages of holding-down lugs which have been rather frequent with the construction of Fig. 153. In an- other type of construction the barrel and head are formed ^of steel in one piece and an aluminim cap embody- ing the inlet and exhaust ports is bolted to the cylinder head. Tests of all-steel cylinders such as are used in Le Rhone and Clerget engines, with cylinder diameters rang- ing from 4 to 6 in., show that the all- steel cylinder gives very appreciably higher fuel consumption and lower brake mean effective power than does the aluminum-headed cylinder. 1 A 5M by 6H- m - steel cylinder with one aluminum inlet and two cast-iron exhaust ports bolted to it was changed (1) by having an aluminum cap bolted to its head and (2) by having the original head cut off and an aluminum head cast on to the same barrel. Tests showed that under maximum load conditions at 1,450 r.p.m. and in a wind of 82 miles per hour the aluminum headed cylinder gave 15 per cent more power than either of the others. The fuel consumption was 26 per cent less than that of the steel cylinder and 20 per cent less than that of the capped cylinder. A capped steel cylinder is usually not much better than the normal steel cylinder; however well fitted initially, " growth" and distortion of the aluminum impair the contact after a few hours' running. 1 A. H. GIBSON, Inst. Aut. Eng., Feb., 1920. FIG. 154. Steel air-cooled cyl- inder with aluminum head. RADIAL AND ROTARY ENGINES 203 The largest all-steel air-cooled cylinder tested by Gibson was 6 by 8 in. With a compression ratio of 4.48 and in a wind of 75 miles per hour this cylinder developed 115 Ib. brake mean effective pressure on a fuel consumption of 0.68 Ib. per brake- horse-power hour at 1,250 r.p.m., and 105 Ib. brake mean effective pressure at 1,600 r.p.m. An aluminum-headed cylinder of the same dimensions developed under the same conditions 121 Ib. brake mean effective pressure on a consumption of 0.56 Ib. per brake horse power per hour. Cylinder distortion may arise from the fact that the cooling air blast is directed against one side of the cylinder. Such distortion is negligible when the blast is directed on the exhaust side. This side is normally the hottest and needs most cooling. Tests on a 5^-in. aluminum cylinder with the blast on the exhaust side showed a maximum temperature difference between the front and back of the barrel of 58C., and a mean difference of 19C. With the blast on the inlet side the maximum tem- perature difference was 180C. and the mean 120C. In spite of this the cylinder, which was fitted with an aluminum piston of only 0.025-in. clearance, gave no sign of binding, showing that even in this extreme case the distortion was not serious. With longitudinal fins and a comparatively uniform distribu- tion of air flow, the distortion is not noticeably less. The exhaust side will be the hottest and the temperature will be less uniform than with circumferential fins and a free blast on the exhaust side. Furthermore, longitudinal fins do not stiffen the cylinder as strongly against distortion as do circumferential fins. Connecting-rod Assembly. The problem of connecting seven or nine big-ends to a single crankpin is usually solved either by the "articulated or master rod" assembly or by the "slipper" assembly. The master rod assembly is used on the Gnome and Clerget rotaries and on most of the radials. Details of the assembly, as installed in the Salmson engine, are given in Figs. 155 and 156. The big end of the master rod encircles the crankpin, holds the wristpins for all the short rods, and carries the outer races of the ball bearings. It will be seen that this construction shortens the effective length of all rods except the master rod; that the axes of the short rods pass through the crankpin only twice in the revolution; and that the obliquity of the short rods is considerably greater than that of the master rod. 204 THE AIRPLANE ENGINE The slipper type of assembly is used in the Le Rhone and Anzani engines. The crankpin carries on ball bearings (Figs. FIG. 155. Articulated connecting-rod assembly. 157 and 158) two thrust blocks each of which has three annular grooves lined with bearing metal. The two discs are fastened ^ FIG. 156. Section through articulated connecting-rod assembly. together with the annular grooves opposite one another. The big ends of the nine connecting rods are provided with slippers RADIAL AND ROTARY ENGINES 205 \ 'IG. 157. Section through slipper FIG. 158. Assembly of slipper type connect- type connecting-rod assembly. ing rods. FIG. 159. Diagram of rotary engine with slipper'type connecting-rod assembly. 206 THE AIRPLANE ENGINE each of which is turned with the same radius of curvature as one of the annular grooves. Three connecting rods act on each groove and consequently there are three designs of slipper. The slippers for the middle and outermost grooves are slotted to avoid contact with the connecting rods for the innermost and middle grooves. The arrangement is shown in outline in Fig. 159. The plan of the slippers in Fig. 160 shows the slotting to prevent interference with adjacent connecting rods. The slipper assembly is considerably heavier than the master- rod type and consequently is better adapted to rotaries than to radials. It has the advantage that the connecting rod is of maximum length and conse- quently of minimum angularity and also that the thrust (or ten- sion) of the rod always passes through the center of the crank- pin. Furthermore a large bear- ing surface is provided at the thrust block which is easily FIG. 160.-Pro.jected views of slip- l ubricate d by the oil thrown off from the ball bearings. Dynamical Comparison of Radial and Rotary Engines. The fixed-radial engine presents the special problem of a large mass rotating with the crankpin and consequently large centrif- ugal force. The inertia forces of the reciprocating parts are additive to this. The result is a considerable total pressure on the crankpin, which is relieved somewhat by the gas pres- sures during the explosion strokes. Roller or ball bearings are necessary at the crankpin if high speeds of rotation are to be maintained. Balancing of the primary inertia forces of a single-crank fixed- radial engine is readily effected by a mass, approximately equal to half the mass of all the reciprocating parts, used as a counter- balance opposite the crankpin at crankpin radius. The counter- balance weight will add 7 to 10 per cent to the weight of the engine and can be avoided only by using two rows of cylinders and a double-throw crank. In the last case there is an unbal- anced primary couple. Balancing the centrifugal and inertia pressures on the crankpin has been accomplished in an ingenious manner in the latest design of Cosmos " Jupiter" engine. Two bob- weights are suspended on the outer sides of the master rod; RADIAL AND ROTARY ENGINES 207 their other ends are connected to the main crankshaft balance weights through hardened blocks working in slots machined in the bob- weights. The bob- weights serve not only to relieve the pressure on the crankpin but also as part of the weight necessary to balance the engine as a whole. The general arrangement of these bob- weights is shown in Fig. 161. In rotary engines the pistons and connecting rods rotate about a stationary crankpin with an angular velocity which is variable. With the master-rod type of connecting-rod assembly there is FIG. 161. Balanced connecting rod of Cosmos "Jupiter" engine. some lack of centrifugal balance at the crankpin, but it is usually negligible. The connecting rods are subjected to centrifugal tensional loading; the pressures on the pins at the ends of the rods increase as the square of the revolutions per minute. With the same connecting rod loading a fixed-radial engine may run at approximately twice the speed of a rotary engine with the same moving parts; before that speed is reached, however, the crankpin loading of the radial becomes excessive. Ball and roller bearings for crankpins of radial engines offer special problems. The bearing rotates as a whole and presents 208 THE AIRPLANE ENGINE conditions of loading quite unlike those of stationary bearings. Considerable investigation of this matter was made by the British Department of Aircraft Production 1 as a result of the failure of both caged and uncaged bearings. An analysis of the situation showed that with cageless bearings the balls are crowded away from the center of rotation, by centrif- ugal force, and rub against one another. The balls rotate usually at about 2,500 r.p.m. about their own centers; the points that touch are always moving in opposite directions and the abrasion is considerable. When a cage is used the centrifugal force on the cage and the balls causes a displacement of the cage until the bearing load on the balls nearest the crank center due to the cage wedging between them is equal to the total centrifugal load on the cage. This causes a heavy abrasive action between the balls and the cage. For successful operation it is necessary to have a cage which will carry independently the rubbing loads on each ball due to centrifugal force. To accomplish this (1) the cage must be strong enough to take the independent loads from the balls without distortion, (2) sufficient bearing surface must be provided at the surface of location of the cage to carry safely the total centrifugal load, (3) sufficient bearing surface must be provided between the balls and the cage to prevent wear on the cage, (4) the cage must be made of a metal of minimum abrasion, and (5) all surfaces must run with a continuous flow of oil. To meet these requirements a cage as in Fig. 162 may be used. This type of cage must be definitely located and not displaceable by centrifugal force for more than a few thousandths of an inch. The design shown is made in two halves with eight hemispherical holes with 0.10 in. clearance for the balls. The outside circumference is turned in a flat V to avoid the actual ball path, and on either side of the V is a true cylindrical surface about %Q in. wide. The cage is of phosphor bronze and fits the outer ball race with a clearance of 0.005 to 0.007 in. This bearing proved entirely satisfactory on the crankpin of a 10- cylinder, 115 by 150-mm. radial Anzani engine developing 150 h.p. at 1,300 r.p.m. The details of a satisfactorily located cage for rollers for the crankpin for a 320-h.p. nine-cylinder radial engine making 1,700 1 J. B. SWAN, The Automobile Engineer, July, 1919. RADIAL AND ROTARY ENGINES 209 r.p.m. are given in Fig. 163. The cage is located on the roller track, which in practice works out advantageously in polishing the track and keeping it free from foreign matter. fT &*&#***?*' view of R.H.HQIF Sec-Kon A-A-A FIG. 162. Located cage for ball bearings. Details of successful and unsuccessful ball and roller bearings are given in the following table. At speeds above 1,600 r.p.m. and with a radius of rotation above 2*^ in. cageless bear- *&& _ r _____ \---~ ? i ~r - \ if 1 " 1 ^ .fc 6 $ tf+aoio \ fcv / "^ ^ Q '*s K "^ ^ H %?*? 1 *; j 1 ^ -e \ 1 ^ \ ^ 7 V '//////k*. JK I > L _ :..... 4. 1" Pitch a/iarn. of Rivets FIG. 163. Located cage for roller bearings. ings will not run satisfactorily if the balls or rollers are larger than % in. in diameter and the inner race larger than 1.5 in. in diameter. 14 210 AIRPLANE ENGINE u fc ^ u ij o O O CO o CO \ llj 11 1 OS 00 1 .2 S M " 0 10 CO 10 10 CO CO CO CO 11 b- b- CO CO IO IO fl CM CM ^ T Tf * 0) (H OQ V 10 CO o 10 o o w a *H rH I-I e of engine H -d-d T3 03 S3 03 ft ft ft " Is Is ^ ^ * V V ? V % \ K r-* 1 111 HI l-H i^ iH w & . t II O O O OS r-l 05 11 1C CO CO 1 W II sss II H CO l-l 0) Bj * 8 i "o 1 H 3 -3 -3 - 111 RADIAL AND ROTARY ENGINES 211 The crankshaft in a radial engine will be solid or built-up according as the big end of the connecting rod has a plain bearing or a ball or roller bearing. The plain bearing is likely to give trouble in view of the heavy loading of the bearing except for low-speed engines, and can be used only with high-pressure forced lubrication; ball or roller bearings are very generally used. The built-up crank necessary with ball or roller bearings may be either a two-web shaft with equal loading on front and rear bearings or an overhung crank with a drag crank for driving auxiliaries. The former type is the more desirable. With the overhung crank the main balance weight is on a single-crank web, which leads to an unbalanced couple, and also, the diameter of shaft and bearings has to be made greater. Valve Operation. Apart from the design of multilobed cams and their driving gears, the valve operation of radial engines does not present any special problems. In rotary engines the effects of centrifugal force on the push rods and tappets have to be met. Counterweights have sometimes been used on the valve side of the rocker arm, but their use increases the load and wear on the cam profile. The necessity for keeping down the over-all engine diameter is likely to result in the selection of an unfavorable type of valve spring and an undesirable reduction in the length of the valve stem guide; failures have been frequent when volute valve springs have been used. Lubrication. Rotary engines are always wasteful of oil, using almost Ho lb. per brake-horse-power hour. There is no return of surplus oil to the pump, which consequently has to determine the amount of oil used. Plunger pumps are always used discharging directly to the main bearings, big end and cam gear, and relying largely on centrifugal force for the lubrication of wristpins and cylinders. In radial engines either a plunger pump 'or a gear pump may be used with a dry sump. There is danger of over-oiling the lower cylinders. Most of the oil goes direct to the crankpin and is distributed thence by centrifugal force to the bearings, connect- ing-rod assembly, and cylinders. The oil consumption of radial engines runs from about 0.02 to 0.04 Ib. per brake-horse-power hour. CHAPTER IX FUELS AND EXPLOSIVE MIXTURES The properties desired in an airplane engine fuel are as follows : 1. It must have a high heat of combustion per pound. This determines the cruising radius for a given weight of fuel, since efficiency does not vary appreciably with the fuel. 2. It must have a high heat of combustion per cubic foot of explosive mixture if it is to develop high horse power per cubic foot of piston displacement. Alcohol and gasoline have about the same heats of combustion per cubic foot of explosive mixture but very different heats of combustion per pound of fuel. The heat of combustion is nearly constant for all the available fuels. 3. It must be able to withstand high compression without preignition or detonation. 4. It must vaporize readily (preferably with little or no pre- heating of the air) upon admixture with air and should be com- pletely vaporized at the beginning of explosion. For good distribution it should be completely vaporized upon reaching the admission manifold, but this is not usually attained. 5. Combustion should be complete, leaving no solid residue in the cylinder. 6. The fuel and products of combustion must not be corrosive. 7. The explosion rate must be neither too rapid (as with hydrogen, acetylene and ether) nor too slow. 8. The bulk of fuel and the weight of the container must be low. This eliminates gaseous fuels. The liquid fuels which meet the above conditions best are: (a) certain hydrocarbons, which form the constituents of gasoline and of certain coal tar products, and (b) the alcohols. The hydrocarbons under consideration may be divided into two main groups, saturated and unsaturated. The latter term is here applied to the behavior and not to the composition of the sub- stance. The saturated hydrocarbons are again subdivided into the aliphatic or acyclic group, and into the aromatic or cyclic group. The hydrocarbons all form series in which the members differ from each other by the addition of CH 2 . The members of the groups of most importance are listed in Table 9, together with 212 FUELS AND EXPLOSIVE MIXTURES 213 || - 1 2^ CO CO 00 Ni-100 8 "5 M 3* -4J X> J m 1 - a O Tj T^H t^ O rH co r~ ^ . < OOOiO lO O lO o >> 3 ocot^t^ < OJ ^0^0 00 iS ^">,^ S^ ft bers of the pe; Bthylene grou n ^ ^ "w 1 a i to o _ d M Group fon ( s t i I f 1 w| CnH 2 (Naphthe w e O W e aomers of th naphthalen< o 1 si 1- h o g-o Ii i 1 o '"& P C 1C C ,_(,_(,-( ^ CO l-H fH Tt W d e o n o o o d d dddcodddd i-l O OCO oo i-sj ?; 1 1 1 j s s i ^i s ^ 1 2-f II II iitfl 216 THE AIRPLANE ENGINE 3 111 s ~r>' M m 00 CSIrHrHrHrHiHTHTHiHOOOiHrHrHrHOOOrHrH r lOr-lT}(TjHTj(Tf(OOOTjO C^ CO CO CO i-l rH (N l-f o 10, i FIG. 1 x^ ^ J ^ ^ / '' 4 / '''$ ? x / / VV'/ 0| '/ / // 1 1 tj- / .t>.t^osoooo doooddddo' 2 S .z * M OOOOOOOO i i ;1 230 THE AIRPLANE ENGINE 14.7 reduced pressure will be TT~TB X 197 = 202 cu. ft. This quantity, and similar quantities for other air temperatures and other fuels, are given in Table 14. The vapor coexists in the same space with the air (202 cu. ft.) and as this volume is greater than the volume which the saturated vapor of heptane occupies (103 cu. ft.) the vapor cannot be saturated in a chemically correct mixture; the vapor will be superheated. At the temperature of 40F. the air volume is seen to be 194 cu. ft. and that of the saturated vapor of heptane 185 cu. ft.; as these are approximately equal the vapor will be practically saturated. At temperatures lower than 40F. the volume of the air will be less than the volume of the saturated vapor and in that case part of the fuel will necessarily be in the liquid form. An excess of air above that chemically necessary will lower the temperature at which liquid must begin to appear; an excess of fuel will raise that temperature. With a less volatile fuel such as octane it will be seen by inspection of the table that a higher temperature (a little under 80F.) will be necessary if the fuel is to be in the vapor form. With benzol the temperature is well below 40; with methyl and ethyl alcohol between 70 and 80F. It should be noted that the temperatures of the table are the 'temperatures after the vapor is formed. In the carburetor, the latent heat of vaporization of the fuel is taken from the air and the liquid fuel, with the result that the temperature of the mixture falls below the temperature of the entering air and fuel, unless heat, equal to the latent heat, is supplied from the jacket water or exhaust gases. If the latent heat of the fuel is 135 B.t.u., the specific heat of the liquid 0.45 and the specific heat of air at constant pressure 0.241, the fall in temperature AT 7 for a mixture of 1 Ib. of fuel with 15.1 Ib. of air is given by the equation 135 = A7X0.45 + 15.1 X 0.241) or AT 7 = 33F. If the fuel is just saturated at 40F., the entering temperature of the air and fuel would have to be at least 40 + 33 = 73F. to permit all the fuel to be vaporized, if no heat is supplied to the mixture from outside. The following table 1 gives data of a similar nature for various fuels. Column 3 gives the temperature of the fuel-air mixture at 1 From KUTZBACH, Technical Note No. 62. National Advisory Committee for Aeronautics, 1921. FUELS AND EXPLOSIVE MIXTURES 231 which the vapor of the fuel is just saturated; the mixture is supposed to be chemically correct and the pressure of the mixture is atmospheric pressure. In column 4 is given the fall in tempera- ture of the air and liquid fuel required to supply the latent heat for complete vaporization. 1 The initial temperature of the air must be at least equal to the sum of the quantities given in columns 3 and 4; this sum is given in the last column. SATURATION TEMPERATURES OF AIR-FUEL MIXTURES Fuel Boiling point, deg. F. Saturation temperature of the fuel mixture, deg. F. Drop in temperature due to evaporation, deg. F. Minumum temperature of the air for complete vaporization, deg. F Hexane 158 54 54 Benzene . 176 23 54 77 Ethyl alcohol 172 72 198 270 Decane Naphthalene 320 428 108 198 63 72 171 270 It is evident that the temperature of the air-fuel mixture with decane or naphthalene as fuel is so high as to reduce considerably the volumetric efficiency and the power of the engine if all the fuel enters in the vapor form. Gaseous Explosions. In an airplane engine making 1,800 revolutions per minute, the duration of the explosion should not be greater than the time of one-sixth of a revolution or }{ go second. The possibility of employing a gasoline engine depends on the possibility of carrying out the explosion process with a high degree of completeness in this extremely short time. Explosion is a chemical reaction attended by the liberation of a considerable amount of heat. It is a combustion process. Combustion results from the chemical union of a fuel with oxygen and this union may take place either (1) at the place where the two are brought into contact as with the ordinary gas burner, or (2) in an intimate mixture of the two, as in a bunsen burner or in a gas engine cylinder. Explosive reaction can take place only with an intimate mixture. The reaction in an intimate mixture is not necessarily explosive ; for example, no explosion occurs in the bunsen burner. An 1 Some of these values are calculated by Kutzbach from values of latent heat which are apparently too high. 232 THE AIRPLANE ENGINE explosion is always self -propagating : that is, if part of the mixture is ignited the combustion will jgpread throughout the mass of the mixture. The term " explosion " is commonly reserved for the case where the velocity of such propagation is high; but there is no definite line of demarcation between explosion and slow burning. The velocity of propagation of combustion in an explosive mixture depends on the kind of fuel, the amount of oxygen present, the amount of inert gases present, the temperature, pressure, and a number of other factors. The strength of the explosive mixture is the most important factor. No explosion is possible if the ratio of air to fuel exceeds certain limits. Bunte 1 has found the explosive limits for various air-fuel mixtures at atmospheric pressure and temperature as given in the following table : EXPLOSIVE LIMITS OP AIR-FUEL MIXTURES ^Fuel Ratio of air to gas by volume Theoretical ratio of air to gas by volume Lower limit, air in excess Upper limit, gas in excess Carbon monoxide 5.06 9.58 7.06 28.8 11.6 23.4 24.3 15.4 35.7 36.7 40.7 .0.33 0.50 0.49 0.91 4.23 5.84 6.32 6.81 12.0 14.4 19.4 2.4 2.4 2.4 11.98 5.7 14.4 14.4 9.63 28.41 36.0 37.5 Hydrogen. . . Water gas Acetylene . . . Coal gas Ethylene .... .... Alcohol Marsh gas. Ether Benzene Pentane Burrell and Gauger 2 give explosive limits of air-gasoline mixtures as 66 and 16 (ratio of air to gasoline vapor by volume). The above results were obtained with mixtures at ordinary atmospheric pressures and temperatures. They show that a self-propagating combustion is possible with most fuels where 1 The Engineer, March 28, 1902. 8 Technical Paper 150, U. S. Bureau of Mines. FUELS AND EXPLOSIVE MIXTURES 233 there is a considerable excess present either of air or of fuel. These limits are considerably extended as temperature and pres- sure increase. For example, at 600C. it is possible to explode a mixture of CO with 12 times its volume of air, as compared with 5.06 times at atmospheric temperature. The presence of carbon dioxide in place of some of the excess air diminishes the explosive limits. The temperature to which part, or all, of the mixture must be brought to initiate an explosion is called the ignition temperature. This varies with the fuel, strength of mixture, the volume or mass of the mixture heated, the temperature and dimensions of the containing vessel, and the method of ignition. A weak spark, although it has a temperature much higher than the ignition temperature, may fail to cause an explosion. It may start combustion at the place where it passes, but the heat loss by convection, conduction and radiation may be in excess of the heat of combustion and the flame will fail to propagate. A sufficient duration of spark is also necessary. A flame may ignite a mixture that cannot be exploded by a spark, because it gives, initially, so large a volume of flame that the radiation loss to the containing vessel does not cool it below the ignition temperature. If the whole mass is raised in temperature simultaneously (as by adiabatic compression) the ignition temperature will be less than when part of the mixture only is heated. This ignition temperature, with adiabatic heating of fuel-air mixtures, is from about 1,200F. for hydrogen to about 1,700F. for carbon monoxide. With the usual gas engine fuels it falls between those limits, the value depending on the hydrogen and the neutrals present. The ignition temperature has great importance as it determines the permissible ratio of compression, and thereby, the limit of efficiency in the engine. Compression must stop just short of that temperature at which ignition will occur. Any means for increasing the cooling of the mixture during compression (such as improved water jacketing) will permit a greater ratio of com- pression. Local heating of the mixture, as by carbon deposit, may result in preignition. Combustion once started in an explosive mixture may either die out or be propagated. If it once starts to propagate itself, it is likely to continue and there will result an explosion. The velocity with which the combustion is propagated increases 234 THE AIRPLANE ENGINE progressively in all true explosions. In the case of a bunsen burner the velocity remains constant and the combustion is not explosive. The flame in that case is stationary, but as the gas is moving the flame is really moving relative to the gas, in the opposite direction and with the same velocity. If the velocity of the gas is diminished too much by partly closing the gas supply, the flame will shoot back, i.e., the flame will travel more rapidly than the gas. The flame remains at the mouth of the burner under considerable variations of gas velocity in the burner because the velocity of the mixture decreases rapidly as it issues from the burner, so that there will be some place, close to the burner, at which the gas velocity equals the velocity of flame propagation. The flame will remain stationary at that place. The cooling effect exerted by the metal burner also reduces the flame propagation velocity. If the velocity of the gas which will just keep the flame away from the burner is measured, it will give a rough indication of the velocity of flame propagation in the mixture. The results will not be very accurate because of cooling and diluting influ- ences of the atmosphere. Experiments of that general nature show that, at atmospheric temperature and pressure, for H and 0, the velocity of propagation is about 115 ft. per second, and for CO and O about 4J- ft. per second. This is for the combining proportions, which give approximately maximum velocities. With H and air the velocity drops to about 10 ft. per second at 212F. For gasoline-air mixtures, at atmospheric temper- atures, velocities of about 3.5 ft. per second and for alcohol about 3 ft. per second are realized. These results apply only to linear propagation at atmospheric pressure. In a closed vessel, such as a gas engine cylinder, the conditions are quite different. The propagation, starting from a point, is spherical; the increase of temperature results in increase of pressure and as the flame spreads the unburned portion will be compressed adiabatically and will increase continually in pressure and in temperature. As the temperature increases the rate of propagation will increase. The velocity of propagation will then be continually accelerated. The flame, moreover, is carried forward bodily by the expansion of the burned portion. Experiments on explosions in closed vessels have determined the time required to reach maximum pressure with various mixtures exploded in vessels of various shapes. If the maximum FUELS AND EXPLOSIVE MIXTURES 235 distance from the ignition point to the boundary of vessel is divided by this time, the quotient gives a measure of the average rate of flame propagation. With illuminating gas at atmospheric temperature, in a tube J- m - m diameter and with 7J^ in. travel of flame, this varies from 5 to 24 ft. per second, according to the strength of the mixture. It increases rapidly with increased initial temperature; in some cases as the tenth power of the absolute temperature (= 1,000-fold for doubled temperature). With the largest existing gas engines (using blast-furnace gas) the available time for a good explosion is about ^ sec. an d the maximum distance the flame must travel is about IJ^j ft.; this gives a mean velocity of 11 ft. per second. The addition of a third igniter has sometimes increased the capacity 20 per cent and shows that the speed limit has been reached. Blast-furnace gas consists mainly of CO, which, at low temperature, has a velocity of propagation not greater than one-third that of gasoline. With an airplane engine at 1,800 r.p.m., the time for explosion is about Hso second; if the flame travels 2 in. the mean velocity will be 33 ft. per second. By increasing the ignition lead, still more time might be provided ; the speed of the airplane engine is not yet limited by the velocity of propagation of the explosion. Alcohol is slower so that alcohol engines could not be run as fast as gasoline engines if the rate of propagation of the explosion should ultimately determine the limit of speed, instead of valve areas and inertia effects as at present. The observed velocities of propagation in actual engines are higher than those which experiments with closed vessel indicate. This results from another factor, turbulence. The velocity with which the gases enter gas-engine cylinders is very much higher than the velocity of propagation of flame. With 1-lb. drop of pressure into the cylinder and no frictional resistance the velocity of the entering air would be about 350 ft. per second; with J4 lb., 175 ft.; with ^{Q Ib. about 120 ft. per second. This gas velocity causes turbulent conditions which cannot be quieted down by the time explosion starts. The propagation is not spherical but is by currents and eddies of burning gas which carry flame to all parts of the vessel more rapidly than is possible with spher- ical propagation. Recent experimental work bears this out. Dugald Clerk found, in a common gas-engine cylinder, that after quieting down turbulence, the explosion takes nearly three times as long as when the usual conditions exist. Experiments in 236 THE AIRPLANE ENGINE closed vessel without stirring gave the time of explosion as 0.13 sec.; with vigorous stirring the time required was only one-sixth as long. Detonation. During explosion in a closed vessel the advancing flame sphere sends off compression waves which travel through the unburned mixture with the velocity of sound in that medium. If the vessel is of sufficient dimensions the increasing velocity of the flame and the continuously increasing pressure and tempera- ture of the unburned mixture will result in the formation of a wave in which the pressure will be such as to bring the mixture (adiabatically) to the ignition temperature. In that case the wave will cause combustion as it moves on. The velocity of this wave will be greater than that of sound because the process is not merely one of wave transmission but of chemical reaction also. Investigations of the explosive wave show velocities of the order of magnitude of 3,000 to 6,000 ft. per second and pressures of 1,000 to 2,000 Ib. per square inch. These pressures are destructive to engines and should be avoided. The "detonations" or "pinking" which are both felt and heard in engine cylinders under certain conditions of operation probably indicate either the generation of an explosive wave or breaking down of the fuel with the liberation of free hydrogen, which explodes with extreme rapidity. In such cases the com- bustion is notably incomplete, the exhaust containing much free carbon, and the power and efficiency of the engine fall off. Fuels consisting of paraffins have a low-ignition temperature, and are readily detonated. Fuels belonging to the aromatic group have higher ignition temperatures and can be used with higher compression pressures without detonation. The maximum pressures to which fuels can be compressed without serious detonation have been determined by Ricardo, 1 who used for that purpose a variable compression engine with compact combustion space, central igniter, and other features making for maximum capacity and efficiency. His results, including the corresponding indicated mean effective pressures and thermal efficiencies, are given in Table 15. The data for toluene, xylene, and acetone are for a compression ratio of seven, which gives a compression pressure well below their detonation pressures; it was not considered desirable to go above that com- pression ratio for hydrocarbon fuels on account of the excessive 1 The Automobile Engineer, Jan. and Feb., 1921. FUELS AND EXPLOSIVE MIXTURES 237 -838 n COT) !!?! sa 5 S A .2 sill* a ill K CO r-t CO 00 -H *O O rfrH OiOOJ -^CCOO i-HO W5OOO ^O5O O C CD5 ir? cod cd^cci cs't^d co CD' co'ic co CO "-^ ^^^ COCOCO CO ^C ^ kC rH 5O OCO O O >CiCO Ous Oo oo ^' cococo ' t^.^. o'eo tN O i-iCDOO t^t I-HCN AA i-Ht^ C5OO OJOOO5 AA OOOiOOON -O .2.2 8J o . .0 ooo^o *3 t>. O Oi 1C CO CO CO >C 00 "5 t-~ OOO COO """cO^H i-H^l t^ ' O5O505 rH ooooooo -o oo O ' * I>O S. |S: II . "5 c-S" ftft ;"'-' "-^ftO 08 c OC i-i 00 M C O> N IN O5 t~tt-t-t^t>.t^t^r^oooo coco oooooo t^t^t^ t>- t^ ooooooooooo oo ooo ooo o o coo oo o-< 50 n di line n dis o a II c Ha 0) 0) mill O)E-i>4 OWW O wS S IU f? a ^ *s ,s ^s, v ^ V ff. / S ^ ^ / / X ^ ^ 2 / Xc D s S 4 Ju 7 x x s <: c n s X 20 18 16 14 Ratio of Air to Gasoline FIG. 173. Composition of the exhaust gases from a gasoline engine. Still another method is available if actual air and fuel measure- ments are impracticable. The investigations of Watson, on automobile engines, have shown that the composition of the exhaust gases varies in a regular manner with the strength of the mixture of air and gasoline admitted to the cylinder. His results are shown graphically in Fig. 173. With a chemically perfect mixture of about 14.5 parts of air to one of gasoline the exhaust gases contain about 13 per cent of C0 2 by volume and about 0.5 per cent each of 2 and CO. If the air is present in excess (weaker mixture) there is more free 2 and less C0 2 in the exhaust; if the mixture is richer, the free 2 disappears and the amount of CO increases while the C0 2 decreases. All that is necessary for the test is an Orsat or other volumetric gas-analysis apparatus and the determination of the C0 2 and 2 content, or in case no 2 is present, the CO 2 and CO content. CHAPTER X THE CARBURETOR An ideal explosive mixture arriving at the intake manifold of an engine should have the following characteristics: (1) it should be homogeneous throughout, (2) it should be of the composition or strength to develop maximum economy under each condition of engine operation, and (3) it should permit of the development of the maximum possible power. In a stationary constant-speed engine, in which engine torque alone is variable, these results might be approximated by the use of an injection valve, under the control of the governor, spraying finely atomized fuel into the current of air going to the cylinders. In an automobile engine, with both engine torque and speed variable, this simple injection method cannot give satisfactory results. In the airplane engine with the three main variables of torque, speed and air density, the | ToEnglne problem is even more com- plicated. For such engines the explosive mixture is formed by the use of a carburetor. A carburetor is a device in which part or all of the air going to the engine passes through a restricted passage, thereby acquiring velocity with consequent fall of pressure; the fuel is sucked into the current of air in an amount which varies with the pressure drop. In the simplified standard form of carburetor shown in Fig. 174, air flows through the restricted " choke," C, and creates a partial vacuum. Gasoline is maintained at a constant level in the float chamber by the action of the float, F, which controls the position of the needle valve, V, past which the gasoline enters. As the float chamber is open to the atmosphere the level of 245 Air FIG. 174. Diagram of simple carburetor. 246 THE AIRPLANE ENGINE gasoline in the nozzle or jet, J, will be the same as that in the float chamber so long as the engine is not operating. The dis- charge orifice of the nozzle is placed higher than the gasoline level in the float chamber to prevent overflow of the gasoline into the air passage when the engine is standing in such position as to incline the carburetor at a moderate angle to the position shown in the figure. When air is drawn through the carburetor, increasing reduction of pressure at C, resulting from increasing velocity of the air, will give an increasing head on the gasoline and will cause an increasing weight flow of the fuel. The mixture of air and fuel will be of constant strength if the weight of gasoline discharged by the jet is directly proportional to the weight of air flowing through the choke. The actual strength of the mixture whether constant or not is controlled by the size of the gasoline jet. A carburetor built as in Fig. 174 would not discharge a mix- ture of constant strength for all rates of air flow, nor is such constancy desirable. It is common experience that the mixture delivered to the engine should be richer at very light loads (idling) than for heavier loads, and also that it should be richer for maximum power than for maximum economy. A satis- factory carburetor should vary the strength of the mixture so as to maintain the desired strength under all conditions of operation of the engine. A study of the action of a carburetor requires a knowledge of the laws of flow of gases and liquids through such passages as are found in carburetors. The more important results of experiment on such flow are given in the following pages. Theoretical Flow of Air through a Constricted Tube. When air is flowing steadily through a tube whose cross-section varies, the weight and the total energy passing each section of the tube per second are constant. Let W = Weight of air passing in pounds per second. p = The air pressure in pounds per square foot absolute. P The air pressure in pounds per square inch absolute. v = The specific volume of the air in cubic feet per pound. V = The velocity of the air in feet per second. T = The absolute temperature of the air in degrees Fahrenheit. 7 = The internal energy of the air per pound in foot-pounds. A = The cross-section of the tube in square feet. a = The cross-section of the tube in square inches. q Gravitational acceleration = 32.16 feet per second per second. THE CARBURETOR 247 The total energy passing any cross-section with unit mass of air is the sum of the internal energy I, the displacement work pv, and the kinetic energy V 2 /2g at that section. Assuming no heat transfer through the tube the total energy at 1 (Fig. 175) can be written equal to that at 2. /i + Pi v, + = 7 2 + p 2 v 2 + * 2 (1) FIG. 175. Venturi tube of optimum proportions. Air is practically a perfect gas. If the expansion is without eddies or friction and without transfer of heat (adiabatic) /0 v n- 1 _ specific heat at constant pressure _ ~ specific heat at constant volume Furthermore, with adiabatic expansion Substituting from equations (2) and (3) in equation (1) there may be obtained the equation TV_Zi 2 = ^JL r (f*\^*\ (4) 2g " 2g n-l plVl I \pj J The weight flow past any section is equal to the volume passing that section per second divided by the specific volume, or, V Since the weight flow is constant at all sections and substituting from equation (3) 248 THE AIRPLANE ENGINE Substituting this value of FI in equations (4) and (5) F* = and Pi/ - (7) W = A, ^ Pl n-1 1 - p XPi We 5 (8) If the section ^.i is taken just outside the tube where the cross- section may be regarded as infinite and the air velocity zero, these last equations become (9) and (10) In the use of equation (10) the specific volume, v\ t has to be deter- mined from the known pressure, p\, and absolute temperature, TI, by the gas equation, p&i = RTi = 53.347Y Furthermore, it is practically most convenient to deal with pressures in pounds per square inch, P, and with areas in square inches, a. Substitut- ing the numerical values of g and n, substituting P and a for p and A, and substituting 53.347 7 i/pi for v\, equation (10) becomes W //PA 1 - 422 /P 2 \i. Vw IK) This equation is difficult to use when the desire d g weight flow, W, is known and the pressure drop is required. The curves of Fig. 176 are based on this equation. The ordinates are weight flows per square inch of area per minute, and the abscissae are pressure drops measured in inches of water. One pound per square inch equals 27.70 inches of water. An initial temperature of 60F. THE CARBURETOR 249 (520 absolute) is assumed. The separate curves are for the initial pressures, PI, marked on them. Equations (8) and (10) have a limit to their range of applica- tion. If air, initially of pressure pi and specific volume v it flows through a tube the smallest section of which is A 2 , the weight flow varies with the pressure, pz, at that section. The weight 20 40 60 80 Pressure Drop in Inches of Water FIG. 176. Weight flow of air. 100 120 flow will be found from the equation to reach a maximum value as Pz diminishes to a certain critical value, and then will apparently diminish as pz is still further reduced. This critical pressure occurs when - - pi n 250 THE AIRPLANE ENGINE That is, the maximum weight flow will occur when the pressure at the smallest cross-section of the tube is 53 per cent of the initial or maximum pressure. It is found by experiment that the pressure at the smallest cross-section is never less than this amount, and that it remains at that exact value so long as the pressure on the downstream side is equal to or less than that pressure. The weight flow through a frictionless tube is deter- mined by the area of the smallest cross-section and cannot be increased by decreasing the pressure on the downstream side of that section below the critical pressure. In equations (8) and (10) p% can never have a value lower than 0.53pi. For very small pressure ranges, for example when (pi p 2 ) is equal to or less than 1 per cent of p if the expansion of the air resulting from the pressure drop is so small as to be negligible and the flow may be assumed to follow the simpler laws of flow of incompressible fluids. In this case I\ = 7 2 , and vi = v 2) and equation (1) becomes V 2 2 V, 2 If the initial velocity is zero y 2 (12) or V is proportional to \/p 1 p 2 , and since the weight flow is proportional to V (with constant cross-section and constant air density), W = Kpi - p* (13) With air at atmospheric pressure, the error resulting from the use of this equation would be about 2.3 per cent for 1 Ib. per square inch pressure drop, and is roughly proportional to the pressure drop for small pressure drops. Equation (12) shows that the velocity and therefore the weight of air flowing is pro- portional to the square root of the pressure drop so long as the pres- sure drop is small. The actual flow of air through a constricted tube is found to be less than the amount indicated by equation (10). Actual flow is always accompanied by frictional resistance and the formation of eddies. The ratio of the actual flow to the theo- retical flow of equation (10) is called the coefficient of discharge of the tube and its value has to be determined by experiment. The choke of a carburetor is usually of the general form shown THE CARBURETOR 251 in Fig. 175, that is, it consists of three parts: (1) a converging entrance; (2) a throat; and (3) a diverging discharge; such a tube is generally described as a venturi tube. With equal areas at the entrance and exit, the pressure drop from the entrance to the throat would be entirely regained at the exit, if the air flow were frictionless and eddyless. In actual carburetors, the pressures at discharge will be less than that at entrance, and the difference will depend on the velocity of the air, the "stream- lining" of the passage, the degree of obstruction offered by the gasoline jet, and the weight of gasoline carried by the air. The total pressure drop in the venturi is important in determining the volumetric efficiency and capacity of the engine; to develop maximum power the charge should enter the cylinder with the maximum possible density. Loss of pressure in the carburetor is a direct source of loss of power in the engine. The results of published tests on comparatively large venturi tubes, with straight axes, and without obstruction at throat or entrance, show discharge coefficients /varying from 0.94 to 0.99 for cases where AI = A S) and A 2 is equal to or less than 0.5 A\. In these tubes it is found that, for minimum friction and eddy loss, the included angle for the converging entrance should not exceed 30 deg., and the diverging discharge tube should have an included angle between 5 deg. and 7.5 deg.; these should be joined to a short cylindrical throat by well rounded junctions. Figure 175 shows a venturi tube of these optimum proportions. Such optimum proportions are generally not practicable for airplanes. Considerations of space available make it necessary to modify the entrance by curving its axis, and force the adoption of larger included angles. Furthermore, the air passage is obstructed by the gasoline jet and its supporting bosses, and, in many cases, by the throttle valve. All these factors will cause a diminution in the discharge coefficient and an increase in the pressure loss. An investigation at the Bureau of -Standards 1 gives data on certain carburetors which were designed for the Liberty engine. The air passages of these carburetors are shown in Fig. 177. The tests were made with various air densities (corresponding to different altitudes), and both with and without fuel admission. Figure 178 shows the coefficient of discharge; Fig. 179 the ratio of the exit to the entrance pressure, 1 P. S. TICE: National Advisory Committee for Aeronautics, 4th annual report, pp. 608-615. 252 THE AIRPLANE ENGINE for both carburetors, with air of 750 mm. pressure and with various weights of air flowing. Figure 180 shows the pressure recovery ratio for the Zenith carburetor, with various air densities, and both with and without fuel admission to the air. The conclusions derived from these tests are as follows: B FIG. 177. Zenith (A) and [Stewart- Warner (fi) carburetors. 1. The coefficient of discharge for the carburetor passages tested has an almost constant and maximum value for effective throat velocities greater than about 150 ft. per second. 20 30 40 P,-P 2 ( Inches of Water) 50 FIG. 178. Venturi discharge coefficients for Zenith (A) and~Stewart-Warner (.B) carburetors. 2. The value of the coefficient of discharge for the carburetor passages tested lies between 0.82 and 0.85, under service con- ditions. These values are probably typical of reasonably well formed passages of similar type. THE CARBURETOR 253 3. The coefficient of discharge for carburetor passages of this type is apparently only slightly modified as a result of consider- ^ ^ >< .ft"* X \ "S %, tC ^ *J ^ -P o a: Foe ^ \ \ \ \ E \ 1 \ 0.96 Q 02 04 a )& 0. i 0. 16 0. 20 Air-Lb. per Sec. per Sq. In. of Throat Area FIG. 179. Pressure drops at partial loads in Zenith (A) and Stewart- Warner (B) carburetors. able changes in passage form, with respect to angles of entrance and exit. 1.00 0.99 a: 0.98 0.91 0.96 "X 5 ^v A = Air only at 75 B=Air 55 C-Air 2,1 D= Air and fuel at 15 c s Asr ** " * 5i f =A/ - r 31 Omm. o -Q M \ N S^ ^ \ \ L X NX > ^ \ \ S3 ' ^< V \\ \ \\ \\ V \\ V \ \ \ s \ \ \ \ \ ^ \ s 1 \ \ \ \ 0.04 0.08 0.12 0.16 0.20 Air - Lb. per Sec. per Sq. In. of Throat Area FIG. 180. Pressure drop through a Zenith carburetor as affected by air density and the injection of fuel. 4. The coefficient of discharge for a carburetor passage is practically unaffected by wide variations in atmospheric density 254 THE AIRPLANE ENGINE (less than 1 per cent maximum variation between the density limits of 0.075 and 0.035 Ib. per cubic foot). 5. The coefficient of discharge for a carburetor passage is practically unaffected by the introduction of fuel to the air stream (fuel discharge introduces irregularities not to exceed plus or minus 1 per cent). 6. The pressure loss in the carburetor outlet changes with the turbulence or internal motion of the air stream. 7. The pressure loss in the carburetor outlet changes with the quantity of fuel admitted to the air stream, and with the method of dividing the fuel by spraying. Pulsating Flow. The previous discussion relates to steady flow of air through the choke. In actual operation the flow is pulsating; each carburetor usually supplies three or four cylinders. With a maximum of four cylinders the carburetor will be supply- ing one cylinder only at any instant. The flow of the air through the carburetor is determined by the velocity of the piston in the cylinder to which the air is going. As this velocity is zero at the ends of the stroke and a maximum at midstroke, the variation in velocity of flow through the carburetor would be considerable were it not for the steadying effect of the intake manifold. The volume interposed between the carburetor and cylinder acts as an equalizing device and cuts down the pressure pulsations at the exit of the carburetor. Tests made in England and at the Bureau of Standards 1 show that for a given weight of air flowing under pulsating discharge the coefficient of discharge of the carburetor (as determined from pressure measurements at the throat), the pressure recovery ratio, and the strength of the mixture are practically the same as for steady flow. The Flow of Fuel through a Nozzle or Jet. The flow of a liquid through an orifice is given by the expression V = C\/2gh , where V is the velocity of flow, C a coefficient, and h the head under which the flow is occurring. This expression becomes where W = Weight of liquid discharged in pounds per minute. a = Area of passage in square inches. s = Specific gravity of the liquid (referred to water at 60F.) h = Head or pressure drop across the jet expressed in inches of water. 1 National Advisory Committee for Aeronautics, 4th annual report, p. 616. THE CARBURETOR 255 The coefficient C includes losses due to skin friction, fluid friction, contraction, and end effects. Its value varies with the head, h, with change in shape of the entrance to the jet, with change in ratio of length, L, to diameter, D, of the passage, and with the viscosity of the fuel. Investigations by Tice 1 on the flow through jets show the in- fluence of these different factors on the value of C. The effect of the alteration of the shape of the jet entrance from square to chamfered is shown in Fig. 181. The diameter and length are the same for both jets. The major effect of the chamfering is to reduce the contraction of the stream in the entrance, in this case, at heads above 2 in. in water. While the coefficient, gxu c .0.4 O.Z a__a Squar re Chamfered EFFECT OF CHAMFERING ENDS OF PASSAGE Includedanqle of chamfer = 60 Depth f " = 0.008"' Diameter of passage = 0.040 Length - 0.4016' Lenqth + depth =10.04 I ^ I L_J 1 1 L_ 8 12 16 eo Head =h (Inches of Water) FIG. 181. Discharge coefficients of square and chamfered jets. C, has considerably higher values with increase of h with the entrance chamfered in this way, it will be noted also that its value varies through wider limits. Chamfering has the very practical advantage in carburetor manufacture, that the angle and depth of the chamfer, within comparatively wide limits, have an almost negligible effect on the discharge; while, on the other hand, small departures from truth in the making of sharp square edges result in wide variations in the discharge. This, together with the great difficulty of producing duplicate parts hav- ing square edges free from burr, practically rules out the square edge for carburetor metering passages. Within the range of metering passage diameters used in general carburetor practice, it is found that the value of C increases with increase of D (Fig. 182). The effect upon C of change in the ratio L:D is brought out in Figs. 183 and 184. In the former, C is plotted against h for l Loc. cit., p. 603 256 THE AIRPLANE ENGINE several values of L:D with D a constant. In Fig. 184, C is plotted against L:D, each curve being representative of a constant value for h. i.o o.8 0.6 0.4 0.2 EFFECT ON-C OF CHANGE IM-0 -^-D SUBSTANTIALLY CONSTANT Submerqed Orifice-Chamfered Ends D'* L" L^D TC I A =0.0327 0.4041 12.36 23.25 B* 0.0350 11.54 2380 C = 0.0373 " 10.84 24.10 D = 0.0395 10. 23 24.70 = 0.0310 0.0150 0.484 27.55 f = 0.0357 0.42024.45 12 16 20 24 Head = h( Inches of Water) 28 FIG. 182. Influence of diameter on the discharge coefficients of jets. 1.0 o0.8 JC 0.2 EFFECT ON-C OF CHANGE IN-L+D WITH D CONSTANT Submerged Orifice -Chamfered Ends D" L" L*D TC - A = 0.0351 0.406 11.31 21.10 B= " 0.200 5.60 26.60 . C = " 0.100 2.80 23.80 D= 0.015 0.42 24.40 l I I I I I 1Z 16 20 24 Head = h (inches of Water) 28 FIG. 183. Influence of the ratio of length to diameter on the discharge coeffici- ents of jets. VALUE OF "L-rD AT VARIOUS HEADS Submerqeol Rassaqe -Chamfered = 0.0357" | 10 Diameter 14 4 6 L/D = Length FIG. 184. Influence of the liquid head on the discharge coefficients of jets. A change in temperature, T, affects the discharge from a passage in two ways through its influence on the density, s, and through THE CARBURETOR 257 the change in fluidity. For ordinary variations in T, the change in s is comparatively small and has very slight influence on the discharge. The curves A, B and C, in Fig. 185, for gasoline discharged from a jet at three temperatures, expresses the order 0.8 0.6 0.4 r ' c% ,B & *. &= ->H p^ o ,2 ^^, ^ EF FECT ON -C OF CHANGE IN-T Gasoline Flowinq jmerqed Orifice -ChamYered End ID" L" L-D TC = 00344 0.407 11.83 9.85 0. >. mo o. = * 29.80 0. Free Orifice -Square Ends = 0.042 0.005 0.119 24.50 = 0.020 0.0,05 0.250 2+50 & & f bu A t / B C W tf9 D 4.00 12 14 -16 Head = h (Inches of Water) ZO 24 FIG. 185. Influence of temperature on the discharge coefficients of jets. of magnitude of the effect upon C of change in fluidity resulting from change in T. These results are for a comparatively long passage, in which this effect is much greater than with the smaller values for L:D found in carburetor practice. The curves D and 500 i 0400 c -4- 300 !!r200 -4- 'i5 C 100 pd ^ -V . X X 2 r x ^J^ ^ oiS> X X GAS NO. SP. 6R. X ^ V ^ x *y X / =0.t80 2=0.694 3=0699^ X s ^1 p J X $ X 4=0.702 5=0.726 ^ ^r s ^ <^ ? ^x s 6=0.722 7=0.7/7 & J^O^- ^" ^x^ .X X 9=0.748 10--0.813- g^ fo- > ^x R ELATIONSHIP BETWEEN FLUID1P AND TEMPERATURE FOR everal samples of special aviati isoline and one commercial grade f "p ^ ^ ^ cj-^rl gc J (9) '"" . o-" J ) 20 40 60 80 100 120 140 Temperature, Deg. C. FIG. 186. Variation of the fluidity of liquid fuels with temperature. E are for sharp-edged orifices; they show great constancy of C with variation both of h and of T 7 : a change in T from 24.5C. to 4C. shows no appreciable change in C at any value of h. 17 258 THE AIRPLANE ENGINE The fluidity of a liquid is the reciprocal of its viscosity. The variation of the fluidity of aviation engine fuels with temperature has been investigated by Herschel, 1 who finds the results shown in Fig. 186. The value of C for a jet will increase as the tem- perature and, therefore, the fluidity of the fuel increases. There is no fixed relation between the densities and fluidities of different fuels; a change of fuel will ordinarily result in a change in C. Mixture .Characteristics of a Carburetor with Constant Air Density. It has been shown by equation (13) that, for moderate pressure drops in the choke, the theoretical air flow, W, is sensibly proportional to the square root of the pressure drop, and with a constant coefficient of discharge this means that the actual air flow follows the same law. It has been further shown that with discharge through a sharp-edged orifice the flow of the fuel follows the same law. Consequently, it would seem possible to construct a carburetor in which the air-fuel ratio would remain constant for moderate air flows. Actual carburetor constructions do not, however, employ sharp-edged orifices on account of the production difficulties already mentioned. Furthermore, the air flow does not increase as rapidly as the square root of the pressure drop for example, in Fig. 176, with air initially at 14.7 Ib. per sq. in. pressure, as the pressure drop in- creases from 10 in. to 40 in. of water, the weight flow instead of doubling increases only from 6.56 to 12.6 or 1.92 times. At the same time, using the stand- ard form of chamfered jet, as shown in Fig. 181, the coefficient of discharge in- creases and thereby in- creases the flow of fluid more than two fold. The mixture will therefore increase in richness as the load increases. To offset this increase, the structure of Fig. 174 is modified in all com- mercial carburetors. These modifications are extremely diverse in character and can be such as to produce a constant mixture 10 FIG. 187. Variation of mixture strength with load in Zenith, Stewart- Warner and Stromberg carbureters. Bureau of Standards Technologic Paper No. 125. THE CARBURETOR 259 or almost any desired variation of mixture with load. Some of these constructions will be considered later. The results of tests on three special airplane carburetors at standard air density, shown in Fig. 187, are characteristic of the methods of variation of the air-fuel ratio with the load in actual carburetors. The Zenith carburetor shows a very constant mix- ture; the other two show enrichment of the mixture with dimin- ishing load, a characteristic exactly opposite to that of the simple carburetor of Fig. 174. The actual value of the air-fuel ratio depends on the size of the fuel orifice and is not characteristic of the type of construction. Mixture Characteristics of Carburetor with Variable Air Density. When the air density changes, as a result of change of air pressure and temperature during the ascent of an airplane, a new disturbing element is introduced into the behavior of the carburetor. With level flight and wide-open throttle, the engine speed may be assumed, as a first approximation, to be constant at all altitudes; this is not the case since the engine speed may fall off as much as 10 or 12 per cent. The volume of air passing through the carburetor is equal (approximately) to the piston displacement of the engine per unit of time and may also be assumed to be constant. The weight of air, W, taken in will then be proportional to the air density, D. At any altitude x, W. Di W a - D where o indicates ground condition. With a sharp-edged fuel nozzle of constant coefficient of discharge, the weight of fuel dis- charged, w, is proportional to the square root of the pressure drop at the carburetor throat (equation 14) and this is, approxi- mately, proportional to the air density, D (equation 12). This may be written: w If R is the air-fuel ratio , then w' that is, the strength of the mixture varies inversely as the square root of the air density. As the air density is proportional to its 260 THE AIRPLANE ENGINE pressure and inversely as the absolute temperature, T, this becomes Ro R* On going from the ground to an altitude of 30,000 ft., where the air density is 40 per cent of ground density, the air-fuel ratio would fall from 20 to 14, or the strength of the mixture would be enriched X 100 = 43 per cent. The strength of mixture desired can only be determined by engine tests. Such tests have been carried out on Hispano- Suiza and Liberty engines at the Bureau of Standards. l The best 2.4 o.oi oxn o.o5 Air Density in Lb. per Cu. Ft. 0.03 FIG. 188. Influence of air-fuel ratio on brakejn.e.p. at various air densities. FIG. 189. Influence of air density on maximum power and maximum thermal efficiency. mixture to use depends on whether maximum power or maximum economy is wanted. The curves of Fig. 188 show how the brake m.e.p. varies with the mixture ratio at air densities, D, from 0.075 to 0.025. As these results are for constant engine speed, they also show the method of variation of the horse power developed. It will be seen that maximum power, P, is obtained with an air-fuel ratio of 15 at all air densities. Maximum economy (minimum fuel consumption per brake horse-power hour) is obtained at- the points crossed by the curve M; it is seen that at ground level (D = 0.075) the most economical air-fuel ratio is 23, and that this value diminishes (richness in- creases) as the air density decreases. The maximum economy curve runs very near to the limit of explosibility, this limit requiring an increasingly rich mixture as the compression pressure 1 P.' S. TICE: Nat. Adv. Comm. Aeronautics, 4th Annual Report, p. 624. THE CARBURETOR 261 diminishes. The air-fuel ratio for maximum economy is given approximately by R = 106D + 15 where D is the air density. In Fig. 189 the same data are re- plotted to show the variation of brake mean effective pressure, and of fuel consumption per brake horse-power hour, with the air density, both at maximum power, P, and maximum economy, M. It is not possible for a carburetor operating with wide-open throttle to give both maximum power and maximum economy without some kind of manual control since the demands for these two conditions differ only in the amount of fuel supplied. The condition of maximum economy alone is important, except for war purposes. For a flight of several hours' duration the com- bined weight of an engine and its fuel consumption will be less for a larger engine operating at maximum economy than for a smaller engine operating at maximum power and developing the same total power. The carburetor should be devised to give maximum economy at full throttle, with a manual control to increase the fuel supply so as to give maximum power if desired. Economy of operation at low loads is unimportant in heavier than air machines since this condition of operation is not possible for other than very short periods. In lighter than air machines the economy at low loads may be of more importance. At full throttle the most economical air-fuel ratio varies from 23 at the ground to 19 at half -ground density; for operation at partial loads these figures must be reduced. It is not desirable to operate an engine with the mixture giving maximum economy because this mixture is o close to the limit of explosibility that slight changes in condition might result in exceeding that limit. Since the economy changes but slowly with change of mixture in the neighborhood of the optimum value, it is the practice to operate with smaller mixture ratios; a value of 20 at the ground is seldom exceeded. The optimum mixture at partial loads may be presumed, as at full load, to be fairly near to the upper explosive limit. This limit changes with the load as a result of change in compres- sion pressure and temperature and of change in the percentage of diluting residual gases present. The compression pressure exerts considerable influence on the explosive properties of a weak mixture, necessitating the use of a stronger mixture as the 262 THE AIRPLANE ENGINE load diminishes. The temperature at the end of compression does not change much since the ratio of temperatures at the beginning and end of compression is a function of the ratio of compression which remains constant; it may be presumed that the tempera- ture effect is negligible. The- effect of charge dilution on ex- plosibility has been investigated for mixtures of air with methane and with natural gas, the diluting agent being C02. 1 Some of the results of this investigation are plotted in Fig. 190. It is seen that with 20 per cent C02, a mixture of natural gas and air cannot be made to explode at atmospheric pressure and temperature; 246 8 10 12 14 16 18 20 22 Carbon Dioxide * Per Cenf by Volume FIG. 190. Influence of carbon dioxide dilution on the explosibility of a mixture of natural gas and air. as the percentage of C0 2 diminishes the upper and lower limits recede until with no CO 2 present we have the lower limit with 5.2 per cent and the upper with 11.6 per cent of natural gas present. The figures for a gasoline-air mixture are probably not very different. With higher pressures and temperature the explosibility limits will be changed but the method of variation will be the same. The amount of dilution of the charge by residual gases can be calculated approximately if the temperature of these gases is assumed. The amount of such dilution will vary with the load since the residual gases fill the clearance at exhaust pressure and are approximately constant in weight at all loads. The amount of such dilution, d = W r /W c (where W r = weight of residual gases and W c = weight of fresh charge), is shown in ELEMENT: Bureau of Mines Technical Paper No. 43; "The Influence of Inert Gases on Inflammable Gaseous Mixtures." THE CARBURETOR 263 Fig. 19 1, 1 which also shows the corresponding compression pressures. These curves are for a ratio of compression of 5.5 and must be regarded as approximations only. The pressure at the end of compression is well above atmospheric pressure and as the temperature is probably about 1,100F. absolute the dilution can be carried further than indicated in Fig. 190 without exceeding the explosive limit. The pressure and dilution of the charge at partial loads are such as to demand a richer mixture 1300' 0.8 0.6 0.4 Load under Throttle 0.2 FIG. 191. Influence of compression pressure on charge dilution at various air densities and loads. than at full load if a satisfactory explosion is to be obtained. It seems probable that the air-fuel ratio for maximum economy does not fall below 15 for any operating condition that is likely to be met; that is, the maximum-economy mixture approximates to the maximum-power mixture as the air density and load decrease. With this in mind the performance curve for carburetors under partial loads can be examined. It would appear that constancy of mixture ratio under varying load is not desirable, and that 1 P. S. TICE, loc. cit., p. 634. 264 THE AIRPLANE ENGINE a carburetor should show enrichment of the mixture with diminishing load. Performance of Representative Carburetors. Several carbu- retors have been investigated at the Bureau of Standards 1 to 20 0.08 0.07 0.06 0.05 0.04 0.02 Air Density in Pounds per Cu. Ft. FIG. 192. Variation of air-fuel ratio in Zenith carburetor. ascertain the variation in air-fuel ratio with variation (1) of air density and (2) of load. Three of these carburetors are considered here. The Zenith carburetor, A, Fig. 177, which is 0.8 0.6 0.4 0.2 Load under Throttle FIG. 193. Variation of air-fuel ratio in Zenith carburetor. described in detail on page 272, has two jets, of which one is operating under constant discharge head to compensate for the natural enrichment of the mixture with increase of load which would take place if the other or main jet alone were used. The i P. S. TICE, loc. tit., pp. 620-636. THE CARBURETOR 265 Stewart- Warner carburetor, B, Fig. 177, has the throttle in the intake (anterior) and compensates for load changes by reducing 10 0.08 0.03 0.07 0.06 0.05 0.04 Air Density in Pounds per Cu. Ft. FIG. 194. Variation of air-fuel ratio in Stewart- Warner carburetor. the air pressure in the float chamber as the load increases by means of a passage connecting the choke discharge to the float chamber. The Stromberg carburetor, C, is described in detail 28 10 1.0 0.8 0.6 0.4 Load under Throttle FIG. 195. Variation of air-fuel ratio in Stewart- Warner carburetor. on page 278. The results of the investigations are exhibited in Figs. 192 to 197. For each carburetor there is shown the varia- 266 THE AIRPLANE ENGINE tion of air-fuel ratio with constant throttle opening and variable air density, and with constant air density and variable throttle opening. The absolute values of the air-fuel ratio are unimpor- 0.07 0.06 0.05 0.04 0.03 Air Density in Pounds per Cu. F+. FIG. 196. Variation of air-fuel ratio in Stromberg carburetor. tant in this connection since they are controlled by the size of the fuel jet, which can be readily changed; the method of variation of that ratio may, however, be considered as characteristic of each type of carburetor. In the Zenith and Stromberg carbu- 18 12 B 8 0.0758 D= 0.0498 D = 0.030! 0=0.0704 1.0 0.8 0.6 0.4 Load under Thro-Hfle 0.2 FIG. 197. Variation of air-fuel ratio in Stromberg carburetor. retors, the need for an additional altitude control device is obvious; the mixture ratio at full load varies from 19 to 10.5 in the Zenith (Fig. 192) and from 15.5 to 9.5 in the Stromberg THE CARBURETOR 267 (Fig. 196) as the air density diminishes from 0.07 to 0.03. The enrichment is considerably in excess of that which has been shown (Fig. 188) to be necessary. With load variation at constant air density, the mixture is practically constant in the Zenith car- buretor (Fig. 193), but enriches with diminution of load in the other two (Figs. 195 and 197) ; it has previously been shown (p. 263) that such enrichment is desirable. Altimetric Compensation. The importance of maintaining an economical mixture at high altitudes is attested by general experience in the air. British tests, to ascertain the advantages of a special altimetric control of the carburetor, have shown with water-cooled engines an increase in endurance from 4 to 4^ hr., and in ceiling from 19,000 to 21,000 ft.; with air-cooled cylinders an increase in endurance from 6^2 to 6% hr., of ceiling from 15,000 to 18,000 ft. and of speed from 84 to 92 miles per hour. In addition to this there is less fouling of the spark plugs, the cylinders keep cleaner, and there is less danger of stalling the engine. Viscous Flow Carburetor. It has been shown (p. 259) that with a standard simple carburetor with sharp-edged fuel orifice, the air-fuel ratio varies as the square root of the air density with full throttle and constant engine speed. There is a possibility of making this ratio constant, under varying air density, by sub- stituting for the sharp-edged orifice a capillary passage in which the flow is entirely viscous. The laws of viscous flow are com- plicated, 1 but, with velocities below those of turbulent flow, it is approximately true that the velocity of flow is proportional to the pressure head. In that case, referring to page 259, we have W* = D* = w x W Do Wo W x W or - - = - , that is, the air-fuel ratio remains constant with W x W ' varying air density. Carburetors have been built embodying the above principle, the viscous flow being obtained by the use of long capillary tubes, or by flow between flat discs or cones as in Fig. 198. At partial loads the fuel supply will fall off in proportion to the decrease in pressure head instead of in proportion to the square root of the pressure head and the mixture will consequently be too weak at low loads. Load control is obtained by raising or lowering the 1 See HERSCHEL, Bureau of Mines, Technologic Paper 100. 268 THE AIRPLANE ENGINE disc (or cone) of Fig. 198 and thereby changing the width of the capillary passage; this can be done by interconnection with the throttle lever. The principal objection to this type of carburetor is that the fuel flow varies with the fluidity of the oil and this varies both with the grade of oil used and with its temperature (Fig. 186). A further difficulty is sluggishness in response to quick opening or closing of the throttle valve. \ FIG. 198. Diagrams of viscous flow carburetors. Altimetric Control. The only practical method at present available for adjusting the air-fuel ratio to the desired value at all air densities, as well as at all throttle positions, is by the use of an additional or altimetric control. A carburetor may be designed so as to give correct mixtures for varying load or for varying air density but it cannot satisfactorily meet both con- ditions, since, with the same weight of air flowing the weight of fuel will be different in the two cases. For example, the weight flow of air at half load at the ground will be the same as at full load at an altitude where the air has half ground density; the pressure drop and the fuel flow will, however, be different in the two cases and therefore the air-fuel ratio will be different. If the carburetor is designed to give correct mixture at all alti- tudes at full load there would have to be added to it a load con- trol (preferably connected with the throttle valve) which would enrich the mixture at partial loads. The other method of pro- cedure is, however, usual; the carburetor is designed to give cor- rect mixtures at full and partial loads, and an altitude control is installed to permit a diminution in the fuel supply at higher altitudes. This control is nearly always manually operated but it can be made automatic without much complication. A diminution of fuel supply can be brought about either THE CARBURETOR 269 (1) by diminishing the size of the fuel orifice, or (2) by controlling the pressure head under which the fuel is flowing. The former is most readily accomplished by the use of a needle in the jet; the latter is the method generally employed because it is less sensitive in adjustment and turns out to be more robust as a structure. n A B FIG. 199. Altitude control by regulation of the float-chamber pressure. Schematic diagrams of some of the more promising methods of altitude control are shown in Figs. 199 and 200. l For the control of the float-chamber pressure, Fig. 199, the top of the float chamber must be provided with a vent, a, to the atmosphere, and a connection, b, to some place where the pressure is less than atmospheric. The control valve may be in either of these passages. ABC FIG. 200. Altitude control by regulation of the jet discharge pressure. The nozzle outlet pressure can be controlled in several ways. The position of the outlet relative to the air passage can be changed, either by shifting the choke, as in Fig. 200A, or by shifting the outlet. The amount of air passing the outlet can be reduced by the use of an auxiliary air valve located at a point 1 National Advisory Committee for Aeronautics, 4th annual report, p. 637. 270 THE AIRPLANE ENGINE beyond the fuel outlet, as in C. A third method is to admit (or bleed) air to the fuel jet past the metering orifice, as in B, thereby reducing the pressure head on the orifice. The structures involving a small plug valve controlling an air stream (Figs. 199 and 2005) are the simplest and most easily produced. Their regulation is comparatively direct and involves small forces and a minimum of parts; furthermore, they adapt themselves readily to automatic control. For such reasons, these methods are the ones usually encountered in service. The objection to them is that they do not permit of one setting for all loads at any given air density but require adjustment for each throttle position, if maximum economy is to be maintained. The method of Fig. 200A is structurally clumsy and would complicate the carburetor considerably. The method of Fig. 200C, using a balanced auxiliary valve, would offer little resistance to operation and little complication. Moreover, the mixture should be satisfactory at partial loads without further manipula- tion. The auxiliary valve would have to be large to give com- plete compensation up to one-half ground density. A simple calculation shows that for this range the area of the auxiliary port must be approximately 1.5 times that of the carburetor throat. Manual operation of the altitude control is extremely unde- sirable. The operation should be continuous as the plane changes its altitude or speed and can at best be only intermittent with manual operations. Moreover, the pilot has no definite means of knowing how far to move the control but must rely chiefly on the engine tachometer readings. He can find the maximum power position but not the more important maximum economy position. As he is already burdened with a large num- ber of controls it is much better to make the altimetric compensa- tion automatic. The simplest automatic operating device is an aneroid bellows. A sealed flexible-walled chamber will expand under reduced pressure and under increased temperature, that is, it will respond to change in air density. If correction for pressure only is desired, the bellows can contain a spring under compression and can be exhausted before sealing (see Fig. 212). Such devices can only operate satisfactorily if the resistance which they have to overcome is small and if the method of control is such as not to disturb the compensation at partial loads. THE CARBURETOR 271 Atomization. The preceding discussion has concerned itself with the metering or mixture-making characteristics of carbure- tors. Other qualities which are of importance are (a) the degree of atomization of the fuel and the homogeneity of the mixture; (6) the pressure drop through the carburetor at wide-open throttle; (c) satisfactory idling performance; (d) acceleration. All carburetors, in order to be acceptable, must be satisfactory not only in mixture making but also in these other characteristics. Favorable conditions for fine atomization of the fuel are high velocities of the air and, to a minor degree, of the fuel. The air velocity is always much greater than that of the entering fuel and the atomization is largely due to the high relative velocity of the air. This is particularly marked if the fuel is not discharged in the axial direction. The use of an anterior throttle, as in Fig. 177 B, by increasing the air velocity at the jet improves atomiza- tion at partial loads. The admission of air before the fuel outlet but past the orifice (see Fig. 2005) is a further favorable condition. Good atomization may be impaired by the impinging of the mixture on obstacles such as a butterfly throttle valve, placed centrally above the jet (see Fig. 177 A). Here again an anterior throttle has an advantage. The mixture will impinge on the inlet manifold and the valves before getting into the cylinder, but it is better to have such actions take place as far away from the mixing point as possible. Best results have been obtained with a long pipe leading from the carburetor to the manifold, giving more time for vaporization and the formation of a homogeneous mixture before the mixture is taken into one or other branch of the manifold. Pressure drop through the carburetor has been touched on in page 251 in the discussion of the discharge characteristics of the air passage. Its importance is solely in affecting the maximum power output. Idling. An engine requires a richer mixture at lighter loads. When the engine is cold a still richer mixture is necessary. None of the carburetors in use on airplanes will give a satisfactory idling mixture without the use of some auxiliary device. This consists of a fuel discharge above the throttle which utilizes the high vacuum above the closed throttle to suck in the necessary amount of fuel. Acceleration. It is of importance that the mixture should respond rapidly to sudden changes in load. If the throttle valve is opened suddenly, the greater density and inertia of the fuel 272 THE AIRPLANE ENGINE tend to make the mixture too weak, with the result that the engine will back-fire or misfire. To avoid this, it is common to have an auxiliary supply of gasoline which, at partial loads, collects near the fuel outlet and is drawn on first when the throttle is suddenly opened, keeping up the strength of mixture until the regular flow is established. Certain special conditions have to be met with by an airplane carburetor as a result of manoeuvres of the plane. The changing inclination of the plane will change the hydraulic head at the jet unless it is placed at the center of the float chamber. With the usual non-concentric arrangement of parts (see Fig. 177) it is desirable to have the float chamber placed in advance of the jet as this will give a greater hydraulic head and richer mixture on climbing and will cut down the fuel supply on descent or diving. It is necessary to see that the gasoline does not overflow from the jet when the plane is resting on the ground. The action of the float and float valves during a dive must be examined. The usual float, guided by a central spindle which is normally vertical, will go out of action during a dive, with the probable result of flooding the carburetor. Special float mechanisms are desirable and have been devised. In case of flooding during a dive, the air horn or intake pipe should be so arranged that gasoline can- not spill out into the fuselage. As the air horn is usually facing forward to get the advantage of the increased air pressure due to the relative wind velocity, such spilling will occur unless the air intake pipe is led upward before being turned forward. The usual dual carburetor has one float chamber, and one air intake to the two chokes. A dual air intake pipe is to be recommended as reducing the risk from back-fire, by making each group of three or four cylinders a separate unit so far as carburiza- tion is concerned. With a common air pipe, back-fire may cause the engine to stop; with double intake, back-fire into one intake will not interfere with the operation of the cylinders fed from the other intake, the engine continues to run and the flame in the back-firing intake is drawn up into the engine, reducing the risk of fire. Furthermore, a dual intake increases engine power by diminishing the -resistance to air flow. CARBURETOR CONSTRUCTION Zenith. The carburetor which has been used most for air- plane engines is made by the Zenith Carburetor Co. In this THE CARBURETOR 273 carburetor, an attempt is made to maintain constant mixture strength at varying throttle positions by the use of two jets or nozzles, one of which, the main jet, acts in the usual way, while the other, the compensating jet. delivers an amount of fuel which is entirely independent of engine speed and load. This arrange- ment was devised by Baverey in 1906. The main jet alone would give a mixture which is at all times too weak, but which becomes richer as the engine speed and load increase; the compensating jet alone would give a mixture which is at all times too weak but which becomes weaker still as the engine speed and load increase. The two jets working together tend to compensate one another, and, if properly proportioned, will give a mixture of fairly constant strength under varying speed and load. This is shown in Fig. 193. In this case, the jet sizes are No. 140 for the main jet and () FIG. 201. Diagram showing action of the Zenith carburetor. No. 150 for the compensating jet, the number indicating the cubic centimeters of water discharged per minute under a 12-in. head. The discharge for the compensating jet is under a con- stant head of 2 or 3 in. of water; the main jet discharge is under the variable head due to the pressure drop at the throat of the venturi, which depends on the size of the throat and may amount to 40 in. of water in usual designs. The arrangement of these jets is shown diagrammatically in Fig. 201, in which a shows con- ditions at rest, and b at full throttle. The main jet, G, is located as usual; the compensating jet,/, discharges into the well, J, which empties into a nozzle, H, concentric with the main jet, G. When at rest, the levels in the float chamber, the wells, and the nozzles G and H, are the same. On opening the throttle, the capacity of the nozzle, H, is so much greater than that of the jet, 7, that the well, J, is kept drained and both air and fuel are sucked up the 18 274 THE AIRPLANE ENGINE nozzle, H. As the pressure in the well, J, is atmospheric, the dis- charge through / is due to the hydrostatic head of the liquid in the float chamber and is therefore constant. The well, J, serves also as an accelerating well, giving a body of fuel immediately available on opening the throttle from the idling position. At low speed, when the throttle valve, T, is nearly closed, the suction at the throat is not sufficient to draw in any gasoline and it enters only through the idling device. This device, shown diagrammatically in Fig. 202a, consists of the idling tube, M, within the secondary well, P, which is inserted in the main well, J, into which the discharge from the compensating jet, /, occurs. The well P is provided with a small metering orifice at the bottom through which gasoline can enter from J, and with small air FIG. 202. Diagram showing (a) idling device and (6) altitude control of the Zenith carburetor. holes at the top. The idling tube, M , terminating opposite the throttle valve, is subjected to a very strong suction whenever the throttle is nearly closed and discharges gasoline from the well P. This gasoline meets the air passing with great velocity through the small opening around the throttle valve and forms the idling mixture. As the throttle is opened, the vacuum at the throttle diminishes while that in the choke increases, so that discharge through M ceases and that through G begins. The altitude control of the Zenith carburetor is shown diagram- matically in Fig. 2026. It is of the type illustrated in Fig. 199A. The float chamber is open to the air through screened air inlets. The well J is in open communication at its top with the float chamber. A passage, P, from the float chamber to the choke discharge, is fitted with a stop cock, L, which is manually operated by the pilot. This cock is closed at the ground and is opened THE CARBURETOR 275 gradually as higher altitudes are reached; it should be opened as far as is possible without appreciably diminishing the revolutions of the engine. The actual construction of a Zenith carburetor is shown in Fig. 203. Gasoline enters the float chamber through D and the needle valve seat, S. As soon as it reaches a predetermined height the metal float, F, acting through the levers, B, and the collar, Nj closes the needle valve, C, on its seat S. From the float chamber the gasoline flows (1) through the compensating jet, /, FIG. 203. Section of Zenith carburetor. into the bottom of the well, J, and then through the channel, K, to the cap jet, H, which surrounds the main jet, G, and (2) through the channel, E, to the main jet, G. The idling tube, M, is inside the secondary well,P, and discharges through the passage, R, to an opening (not shown) opposite the throttle valve. The altitude control valve, Y, is a tube which is shown communicating with the choke discharge; the other communication to the float chamber is not shown. It is operated by the lever X. As in other carburetors, a single float chamber is used to supply two air chokes if the engine has six or eight cylinders. One air horn 276 THE AIRPLANE ENGINE or intake commonly serves the two chokes of a duplex carburetor, but it has been found that greater engine power can be obtained if separate intakes are used. Tests of special Zenith carburetors for the Liberty engine showed maximum power developed with separate air intakes about 4 in. long. 1 The special feature of the Zenith carburetor which has recom- mended it is the absence of all moving parts. It is general experience that auxiliary air valves, metering pins, and other moving devices will stick at times and cause irregularity of action. For maximum reliability and fool-proofness the com- pensating device should be fixed. The Claudel carburetor, which has been used very extensively for airplane engines, especially in Europe, is now being made in this country. Like the Zenith, the compensation for load and speed is made without any moving parts. A general view is shown in Fig. 204. The fuel discharges into the choke from a diffusor which is shown assembled in Fig. 2056. The diffusor has four concentric tubes, the air tube e, guard tube d, diffusor tube c, and idling tube a. The main jet is in a small plug screwed into the bottom of the diffusor. Air at atmospheric pressure enters the bottom of the air tube, passes over the top of the guard tube (which prevents the fuel from overflowing when the engine is at rest), then goes through such holes in the diffusor as are above the fuel level, and out through the nozzle holes to the throat of the venturi. The fuel is at the level shown when the engine is idling or at rest. As the throttle is opened, the suction in the diffusor increases, thereby lowering the liquid level in the diffusor bore and uncovering progressively a series of air-bleed or compensating holes. Through these holes the air rushes into the ascending column of fuel and atomizes it as it leaves the nozzle holes at the top. At maximum load the diffusor is practical- ly emptied and all the air-bleed holes are in action, cutting down the effective head on the fuel. The compensation is by control- ling the jet outlet pressure along the lines indicated in Fig. 2005. Any desired kind of compensation can be obtained by appropriate design of the size and location of the air-bleed holes. The diffusor acts also as an accelerating well. When idling the diffusor is out of action and all the fuel goes through the cen- 1 Bulletin, Experimental Department, Airplane Engineering Division, U. S. A., Jan., 1919. THE CARBURETOR 277 tral idling tube, mixed with some air entering compensating holes from the air tube. FIG. 204. Section of Claudel carburetor. Holes Compensating Hofes Guard Tube Oj i _zj AirTub* Bore Guard Tube PJ 1/3 (d) FIG. 205. Details of Claudel diffusor. (e) The throttle is a cylindrical or barrel throttle, bored out so as to form a smooth continuation of the venturi when it is wide open. It offers no resistance at maximum load and consequently leads 278 THE AIRPLANE ENGINE to maximum volumetric efficiency and power. As the idling tube projects into the throttle space, the throttle is slotted out wide enough to pass around it. To diminish the area through this slot when the engine is idling a screw, c, extends into the air space. Advancing the screw lessens the air area and enriches the idling mixture. Figure 206 shows the idling position. Another feature of this carburetor is the sliding air cone, A (Fig. 204), which is controlled by an external lever. When the cone is raised to contact with the venturi, it shuts off all air supply and puts maximum suction on the diffusor. This greatly enriches the mixture and is advantageous for starting in cold FIG. 206. Idling device of the Claudel carburetor. FIG. 207. Section of dual Claudel carburetor. weather. The same device is used for altitude control. The venturi used in airplanes is larger than is necessary at the ground. At low elevations the air cone is kept in a raised position in order to increase the suction in the diffusor to the amount necessary to give the desired mixture. As elevation is gained the air cone is gradually lowered, thus compensating for the natural increase in richness. A cross-section through the diffusors and throttle valves of a duplex Claudel carburetor as used on the Hispano-Suiza engine is shown in Fig. 207. The Stromberg carburetor, Fig. 208, although structurally very different, uses the same general method of compensation THE CARBURETOR 279 for speed and load as the Claudel. The special features of this carburetor are the float mechanism and the double venturi. The float (Fig. 208) is spherical or cylindrical (with horizontal axis) and is hinged as shown with the pivot toward the tail of the Metering Nojj/e Air Horn Drain FIG. 208. Section of Stromberg carburetor. plane. With this mounting, the float is in action during all ordinary manoeuvres .of the plane (Fig. 209), that is, it keeps the needle valve closed with a moderate amount of gasoline in FIG. 209. Diagram showing Stromberg float chamber in different orientations. the chamber. If the plane goes upside down the weight of the float will close the valve. With the arrangement of Fig. 208 the main jet will overflow into the air inlet during a steep dive with closed throttle. A duplex carburetor arranged as in Fig. 280 THE AIRPLANE ENGINE 210, with the float between the two discharge jets, leaves no possibility of such leakage of fuel. Large venturi'' tube Accelerating well- Metering nozzle- Altitude control tube'' Main gasoline channel 'Float ----Air horn drain connect/on FIG. 210. Section of dual Stromberg carburetor. The diagrammatic sketch (Fig. 211) shows the metering jet, E, discharging into channel, A, with air-bleed holes, D, through which air at atmospheric pressure enters from the outer channel, B. The outer channel is also the accelerating well. The fuel and the atomizing air are dis- charged radially into the choke through a ring of small holes, located at the throat of a small venturi tube. This small venturi is concen- tric with a larger venturi and discharges at its throat. The discharge pressure of the small venturi is considerably below atmos- pheric pressure and the depression is still greater at the throat of the small venturi. This results in very high velocity for that por- tion of the air supply which passes through the small venturi, giving good atomization of the fuel without having to make the whole air supply acquire a very high velocity. This arrangement gives a small total pressure drop in the carburetor, and consequently high volumetric efficiency of the engine. The idling device is a miniature carburetor with discharge just above the closed throttle. The idling tube connects directly FIG. 211. Dia- gram showing load control of Stromberg carburetor. THE CARBURETOR 281 with the main jet passage and has a fuel nozzle discharging into a mixing chamber where it meets air entering through holes which are controlled by a needle valve. The discharge nozzle into the main choke is a slot of which more is exposed as the throttle moves from its closed position. The increased opening of the slot increases the suction in the mixing chamber, and sucks up more fuel as the throttle begins to open. With still further opening the suction at the main discharge nozzle increases while that at the idling nozzle decreases. There is a throttle position at which fuel discharges through both, but with still further opening the idling nozzle goes out of action. FIG. 212. Diagram showing automatic altitude control attached to Stromberg carburetor. Altitude compensation is effected by controlling the pressure in the float chamber. An arrangement for automatic control is shown diagrammatically in Fig. 212. The aneroid chamber, A, which has been exhausted before sealing, is compressed by the joint action of the air pressure and the spring B. As the air pressure diminishes the aneroid expands compressing the spring and raising the valve C. The valve point is slotted and offers a decreasing aperture for the admission of air as the valve rises. Air is sucked through this slot by the action of the venturi at D, and as the only air vent from the float chamber is into the pipe E, the pressure in the float chamber will vary with position of the valve C. An additional manual control is a necessary safety device. The design in Fig. 210 is especially adapted to a 90-deg. Vee engine and, as previously pointed out, permits a position of the 282 THE AIRPLANE ENGINE float chamber between the two carburetor outlets which largely eliminates the disturbing factor of changing inclinations of the plane. The carburetor barrels are water-jacketed for high altitude service. The main fuel nozzles are in an annular groove around the small venturi. The altitude-control suction is through the small axial tubes shown terminating at the throats of the small Venturis and consequently give the maximum possible suction and range of action of the control. The altitude control has a partial connection with the throttle in such way that the FIG. 213. Sections of Miller carburetor. mixture is enriched during the latter part of the closing of the throttle. The Miller carburetor has been used on the U. S. Bugatti engine. It is of the multiple-jet type in which load compensation is effected by bringing more jets into action as the throttle is opened, the sizes of the jets being designed to give correct mix- ture at all loads. The jets are air-bled, giving compensation for varying speed. The jets are held in a narrow holder (Fig. 213) and discharge across a diameter at the throat of the venturi. The drill sizes for the Bugatti engine are No. 76, which is the idling jet, No. 76, No. 75, No. 71, No. 68, No. 57, No. 53. The corresponding diameters in inches are 0.020, 0.021, 0.026, 0.031, THE CARBURETOR 283 0.043, 0.0595; the areas consequently increase very rapidly. These jets come into action progressively as the throttle is opened. Each jet has four small air holes just above the metering orifice; air enters at atmospheric pressure through a Ke-in- hole near the top of the jet holder and passes down around the outside of each j et to the air holes. The gasoline flows from the float chamber to the lower Ke-in- hl e in the jet holder. The idling jet is the first in the holder. The throttle valve is of the barrel type bored out to give a venturi form when wide open. The stop for the idling position is seen in the figure. Altitude compensation is obtained by varying the pressure in the float chamber, the air space of which is at all times in direct connection with the venturi. A manually- 6as Fbssage FIG. 214. Sections of Master carburetor. operated valve controls the size of the free air connection to the top of the float chamber. The Master carburetor is also of the multiple-jet type, but differs from the Miller in that the jets are all of the same size and are not air-bled. The throttle is of barrel type (Fig. 214) with an opening that is curved so as to uncover the jets pro- gressively as the throttle is opened. An air damper controlled by the pilot restricts the venturi opening and consequently enriches the mixture when desired for starting. The number of jets is usually from 14 to 21, which demands extremely small metering orifices. The Ball and Ball carburetor (Penberthy Injector Co.) is of the single metering orifice, air-bled type. The float is spherical in a spherical chamber. The venturi throat, A, (Fig. 215), has the main nozzle tubes, B, connecting through the annulus, C, 284 THE AIRPLANE ENGINE with the passage, D, and the mixing chamber, E. The metering jet, Fj is at the bottom of the nozzle, G, and the fuel overflows through the four air holes, H, into the chamber, E, which connects to the outside air through the passage, M, and the air orifice, N. Gasoline arrives from the float chamber at J. The idling jet, P, connects through the passage, 0, with the mixing chamber, E, and discharges just above the closed throttle. An auxiliary air valve, S } is sometimes used to reduce the strength of the mixture at heavy loads. FIG. 215. Section of Ball and Ball carburetor. The altitude control is by variation of the pressure on the dis- charge side of the main jet. This is accomplished by substitut- ing a larger valve-controlled opening for the air orifice, N ; opening this valve increases the pressure on the discharge side of the main jet and weakens the mixture. The carburetor used on the (German) Basse-Selve engine is simpler and lighter than any of the types previously discussed. The float (Fig. 216) is annular, and concentric with the choke, thereby reducing the possibility of overflow of gasoline from the main jet when the carburetor is inclined. The main jet is THE CARBURETOR 285 formed by a hole drilled in a tube which is screwed diagonally into the water-jacketed body of the carburetor and lies across the choke tube. The jet tube is open at its lower end and projects into the bottom of the float chamber. The idling jet is formed by a second tube of small diameter inside the jet tube. This idling tube is also open at the bottom and is drilled radially with a small hole just below the main jet. It communicates with the mixing chamber just above the throttle by a passage drilled in the carburetor body. Altitude compensation is by varying the air pressure in the float chamber. FIG. 216. Sections of Basse-Selve carburetor. The float chamber is made of pressed sheet steel of very light gage. The needle valve (Fig. 216) is acted on directly by the float without the intervention of levers. The carburetor of the Bayerische Motoren Werke engine has some noteworthy features. It consists of three carburetors with a common float chamber (Fig. 217). Each of these car- buretors has a separate discharge pipe leading to a common induction manifold. The central carburetor has both idling and main jets; the outer two have main jets only. There are five throttle valves arranged in two systems with independent control. The main system has three throttles, one to each carburetor. The secondary system, which is an altitude control, has valves on the outer carburetor only. The action is as follows: When the main throttle is opened slightly, the side throttles remaining closed, the idling jet (center carburetor) alone is in action; mixture from the center carburetor 286 THE AIRPLANE ENGINE alone reaches the cylinders. As the throttle is opened further the main jet of the center carburetor comes in action and supplies the whole mixture until the throttle is half open. After this, the two side carburetors, which are controlled by slotted links, begin to open. The normal continuous ground level full-power operation is at the point where the side jets are just about to begin to discharge. So long as the secondary throttles remain in their closed position with relatively small passages past them, a compara- tively rich mixture is supplied by the side carburetors. As altitude is gained the secondary throttles are opened and give increased power while keeping the mixture of the desired strength. Adjusfmerrf- ibr 5lon forming Main Jets FIG. 217. Sections of B.M.W. carburetor. An entirely different type of carburetor is used on the Maybach engines on large German dirigibles. These have been designed to dispense with the use of a float chamber and to work in con- junction with a gasoline-pump system. The construction is shown diagrammatically in Fig. 218. The throttle valve, J, is of the rotary-barrel type and admits carbureted air from N and fresh air from L. The throttle lever is interconnected with the sliding shutter, K, controlling the air that flows past the jets, and with a rotatable cover, P, regulating the size of the jets. Fuel from the gasoline pump enters an upper vessel, A, by the pipe, B. The level in this vessel is kept constant by an overflow pipe, C, which conducts the excess fuel back to the supply tank. An air vent fitted with a baffle plate is provided at F. The fuel passes THE CARBURETOR 287 from A through a strainer, M, to the vessel, D, whence it is sucked through the orifice, H, into the induction pipe. Excess fuel in D overflows and joins the excess from A in the pipe C. At the top of vessel D two holes are drilled the main and idling jets. These orifices are controlled by the eccentrically-mounted cap, P, which is rotated through interconnection with the throttle FIG. 218. Diagram of Maybach carburetor. lever. The fuel has a constant liquid head equal to the difference in levels between the liquid in A and the level of the orifices; in addition it is subjected to the suction in the passage above H. In the idling position, L is open slightly (Fig. 219), K is closed, and the idling jet only is uncovered by P. The throttle-lever Angular displacement of control lever. FIG. 219. Action of the Maybach carburetor. quadrant is marked with the positions "idling," "low speed," "full power," and "altitude." As the throttle is rotated from the idling position, which demands a rich mixture, the shutter K opens but the fuel opening does not increase much till the "low speed" position is reached; the fuel discharge increases in con- sequence both of increased fuel orifice and of the increased 288 THE AIRPLANE ENGINE suction at H. The "full power" position is not maximum power but is the maximum at which it is desirable to operate the engine at ground level. The fuel orifice is nearly wide open at the full- power position. With further opening of the throttle the fresh- air inlet L opens more, thereby preventing the enrichment of the mixture which otherwise would occur at high altitudes and maximum power. It is evidently possible to design the dimensions and the interconnections of the three orifices G, N and H in such way as to give any desired mixture to an engine operat- ing at ground level and at maximum power at various altitudes. Partial loads at high levels are not provided for. This method of meeting the carburetor problem is un- desirable because of the com- plexity of the design and the practical impossibility of mak- ing the varying fuel orifices of the desired dimensions. This particular carburetor is very heavy and offers a large air resistance, thereby reducing the volumetric efficiency and power of the engine which it supplies. A very simple type of carburetor is used on the rotary Le Rhone engine. The air-fuel mixture enters the rotating crankcase through a stationary hollow crankshaft. The screened air supply is controlled by a throttle which is in the form of a shutter (Fig. 220) carrying at its lower end a long metering pin which controls the size of the fuel jet. The pressure at which the fuel arrives at the orifice is controlled by a by-pass valve; this serves to con- trol the mixture when altitude or load is changed. The inherent mixture control is irregular and uneconomical with a device of this nature. FIG. 220. Section of LeRhone car- buretor. CHAPTER XI FUEL SYSTEMS The following statement of the requirements of the fuel system of an airplane engine is abstracted from the "Handbook of Instructions for Airplane Designers" prepared by the Engineering Division of the U. S. Air Service. There should always be more than one means of supplying fuel to the engine. Main-feed System. Gravity feed should be used throughout if it is possible to maintain a sufficient head with the airplane at maximum angles of flight. It has been found that a head of 18 to 30 in. is required for satisfactory operation of current types of carburetor. Unless it is possible to maintain a sufficient head by gravity, pumps must be installed to supply gasoline from the main tanks to the engines. Pressure in supply tanks is not permitted on fighting planes. The main fuel pumps should have a capacity at least 50 per cent greater than the maximum requirement of the engines. Two pumps, other than hand pumps, are desirable, either of which can supply sufficient fuel. They should have automatic pressure regulation to eliminate the use of relief valves or other means of adjusting the pressure at the carburetor. The gasoline pressure at the carburetor must always be at least 1 Ib. and the system should be so adjusted that this pressure can never rise above 3 or 4 Ib. as a result of change in position of the airplane. Pumps capable of a discharge pressure higher than 4 Ib. should have relief valves connected between the suction and discharge, so adjusted as to limit the maximum discharge pressure to 4 Ib. The fluctuation of pressure at the carburetor, due to pulsations of the pump, should not be over 25 per cent. Where air pressure is used, the power air-pump must be capable of keeping a pressure of 2 Ib. on the tanks at the ceiling of the airplane and both spring- and manually-controlled relief valves should be furnished, the former set to relieve at 4 Ib. per square inch. Pumps should preferably be located below the lowest point in the supply system. If they are located higher than the bottom of the main tanks, means must be provided for admitting gasoline from the auxiliary supply to the suction side of the pumps. It should be possible for the pilot to make use of this connection during flight. A non-return valve must be installed to prevent this gasoline from going into the main tanks instead of the pumps. Pumps which do not require glands are preferred, although satisfactory glands will be accepted ; in case glands are used, they must be so located that any leakage from the glands will be drained to a point outside of the fuselage. Pumps should preferably be connected to and driven by the engine. Auxiliary-feed System. The auxiliary-feed system supplies gasoline to the engine in case of failure of the main supply; this auxiliary system should 19 289 290 THE AIRPLANE ENGINE be such that fuel can be supplied to the engine in the shortest possible time never to exceed a period of 10 sec. from the time the pilot starts to make use of the auxiliary system. For emergency use, gravity tanks are best, but must not be used if a head of 12 in. in level flight is not obtainable; they should have sufficient capacity to operate the engines for 30 min. at an altitude of 10,000 ft. with wide-open throttle and should be so connected to the system that they can be shut off and used for reserve or emergency only. They must be so constructed or connected that they can be entirely emptied with the airplane inclined at maximum angles of flight. An over- flow pipe from the gravity tank returns any excess gasoline to one or more of the main tanks; this overflow must be so constructed that there can be no gasoline trapped in it when the airplane is in normal flying position. Unless there are three means of delivering fuel to the engine, such as two engine or wind driven pumps and a gravity tank, a hand gasoline pump must be provided which will permit the pilot, while controlling the airplane, to pump, without undue exertion, sufficient fuel from the main supply at proper pressure for the operation of all engines at full throttle. The capacity of this auxiliary system must be such that the pilot will not need to operate the pump during more than one-third of the time. Tanks should be of tinned steel and of such thickness that the tank will stand 5 Ib. per square inch pressure on the inside without undue distortion. Flat surfaces are to be avoided. Wherever the width or length (horizon- tally) of a tank is greater than 12 in., a splash plate for reinforcing purposes must be installed at least every 12 in. ; wherever the height of a tank is greater than 18 in., a splash plate for reinforcing purposes must be installed at least every 18 in. All seams, including the connection between the splash plates and the walls of the tanks, should be riveted and soldered. Copper or soft iron rivets must be used throughout; the exposed parts of the rivets to be tinned in case iron rivets are used. Drains leading to a point outside the fuselage must be installed in the bottom of each main tank. Fillers must be conveniently located on each tank, and in such a position that t*he entire tank can be filled while the airplane is on the ground. A removable screen must be installed at the point of filling of each main tank, and also, if practicable, in the gravity tank. Vents must be located at the highest point on all tanks, usually in the filler tube, except on wing gravity tanks where the overflow pipe shall act as a vent. Line and Carburetor Strainers. A line strainer, with removable screen and bowl, must be installed between the tanks and pumps, located as low as possible and in such a position as to be readily accessible for draining and cleaning. The strainer screen should be of brass, bronze or copper of about 100-mesh and 0.005-in. diameter wire, and should have at least 1 sq. in. for each 6 gal. which must pass through per hour. Each carburetor should be provided with a strainer having a readily removable screen of brass, bronze or copper of about 50 mesh and approxi- mately 0.009-in. diameter wire and having an area of at least 2 sq. in. Service Pipes and Connections. Service pipes should be % in. outside diameter where flow is 30 gal. per hour or less; % in. outside diameter where flow is between 30 and 60 gal. per hour; and % in. outside diameter where FUEL SYSTEMS 291 flow is between 60 and 100 gal. per hour. All vent and air tubes should be Y in. outside diameter. Wall thickness should be 0.028 in. for ^-in., 3^ 2 in. for %-in., and % 4 in. for ^-in. and %-in. outside diameter. All service pipes, or tubing, should be seamless and of annealed copper, soft enough to withstand vibration. At all points where the tubing is connected to solidly mounted objects, such as pumps or tanks, flexible connections must be provided. The tubing must be properly protected at points of possible chafing. Sharp bends are not permitted. Tube fittings are to be of brass or bronze. Carburefo. r Filler Cap -^*A mall compartment trf/owinq into I rqer compartment - Main Tank fe= r Lock wire on f cock hand fe '* Primer Valve with spring -fa automatic- al fycfose 'Distributing valve acce&ible top/i/a Strainer ' Drain Cock %Drain of Fuselage FIG. 221. Fuel system for a single-engine airplane with gravity feed. Multi-engine Installations. When more than one engine is used, each should have its own gasoline system, consisting of pumps, main tanks, gravity or reserve tank, distributing valve and other apparatus required for a single engine system. A cross connection with shut-off valve should be provided so that any engine can take fuel from the tanks of the other engines, and unless two pumping units, not operated by hand, are provided for each engine, a cross connection should be provided so any engine may receive fuel from the pumps of the other engines. Priming Devices. A priming system should be installed on every engine, with the priming pump mounted in the cockpit in an accessible posit'on. 292 THE AIRPLANE ENGINE Typical arrangements of the fuel system are shown in Figs. 221 and 222. Figure 221 shows a system in which the carburetor is near the bottom of the fuselage so that gravity feed can be em- ployed. The auxiliary tank is most conveniently and simply made a portion of the main tank. A pump system with auxil- iary tank incorporated in one of the wings is shown in Fig. 222. Filler Cap -., (Jravity Tank Primer OverflowsightgJass readi/y visible topi/of Pump priming valve fo pilot this priming connection may be omitfad if pumps a re Mm tanks and if the gravity tank atalltime can provide \ sufficient head of the ' carburetor ' Yalve with spring \ to automafr'ca/fy \ close I Lock Wire on cock handle Distributing valve \ accessible to pi /of \ g Drain FIG. 222. Fuel system for a single-engine airplane with pump feed. Pumps. A simple form of air pump, used in the Hispano- Suiza engine, is shown in Fig. 223. It is operated by a cam on the camshaft which gives the pump its compression stroke; the return stroke is by the action of the spring. A cup leather on the piston acts as a suction valve on the return stroke. The Mercedes pump (Fig. 224), which is driven from the end of the camshaft, takes in air through ports uncovered by the piston near the end of the suction stroke. A relief valve is incorporated in the pump. Fuel pumps are made in many forms and are driven either from the engine or by small windmills. Sliding vane and gear FUEL SYSTEMS 293 FIG. 223. Hispano-Suiza air pump. To Tank AirlnlefPo. FIG. 224. Mercedes air pump. t FIG. 225. Maybach fuel pump. 294 THE AIRPLANE ENGINE pumps (see p. 338) are often used and differ from the oil pumps only in smaller capacity. The compact duplex reciprocating pump of the Maybach engine (Fig. 225) is driven from a crank on the end of the oil-pump shaft through a yoke with a sliding bushing. Any leakage of gasoline past the plungers is into the crank chamber, which is filled with lubricating oil under pressure. Another method of avoiding the use of glands past which fuel leakage might occur is the employment of castor oil as the dis- placing medium. In the Benz engine (Fig. 227) the fuel pump is driven by worm gearing from the end of the inlet camshaft. ...-- Pressure Re fief Vafve To Carbure-hors Top of Gasoline ' Tank Tachometer Drive Gasoline Delivery to Pressure ffeservo/r in Main Tank and Hand Pump and Auxiliary Tanks Outlet Check Yalve ' Gasoline Supply from Main and Auxilfiaru Tanks FIG. 226. Benz pressure reservoir. FIG. 227. Benz fuel pump. The lower portion of the cylinder is near the bottom of a chamber containing castor oil and the reciprocation of the piston produces a rise and fall of the castor oil in the annular space around the cylinder. The castor oil acts like an annular piston sucking in gasoline as its level falls and discharging it as the level rises. As the speed of the pump is slow (worm-gear reduction 10.75 to 1) it is necessary to keep the discharged gasoline under air pres- sure during the suction stroke and this is accomplished by the use of a pressure reservoir (Fig. 226) located in the main fuel tank; the pressure reservoir also serves to damp out pressure pulsations. CHAPTER XII IGNITION Ignition is produced by the passage of an electric arc through the explosive mixture, at a time which varies somewhat with operating conditions, but in airplane practice is about 25 to 30 deg. before dead center on the compression stroke. At the operating speed used in airplane engines the moving electrode of the make-and-break system is impracticable. The spark passes between stationary electrodes and is incorporated in "spark plugs" which are screwed into the cylinder head. For the production of the electric arc the following pieces of apparatus are necessary. 1. A source of electric energy; this may be a primary or secondary (storage) battery, or more usually, a magneto. 2. As the potential required to cause arcing is very large the low potential current generated in a battery or low-tension magneto has to be transformed into a high-potential current by the use of an induction coil; this is usually incorporated in the magneto. 3. The current from a single source has to be sent in succession to each of several cylinders; this is accomplished by the use of a distributor which is located in the high-tension circuit. 4. The distributor connects up the circuit to that cylinder in which ignition is next to occur and maintains that connection throughout a short period. The actual timing of the ignition within that period is controlled by a timer, breaker, or interrupter located in the low-tension circuit. Other minor but essential elements will be discussed later. Electric ignition systems utilize electro-magnetic phenomena. An electric current is induced whenever a conductor is moved through a magnetic field or when the magnetic field around a conductor is varied. The intensity of the induced current is proportional to the rate at which the conductor cuts the lines of magnetic force and to the number of coils cutting the lines of force. 295 296 THE AIRPLANE ENGINE The simplest kind of electric ignition system is shown in Fig. 228; B is a source of current, N 1 a coil of wire surrounding an iron core (forming an electric magnet), S is a switch, timer, or other device for breaking the circuit at any desired moment. The magnetic flux is represented by the arrowed lines. If the switch, S, is opened the current falls to zero and N' is surrounded by a diminishing magnetic field; if S is closed N' is surrounded by a rising field. In both cases self-induction occurs in the coil N f and a current is generated in it, whose mag- nitude depends on the rate at which the magnetic field through N f changes and on the number of turns in the coil. The FIG. 228. inductance phenomena on closing and on opening S n are quite different. On closing the cir- cuit, current can flow only after the switch is actually closed and the flow is opposed by the resistance of the circuit (which is small) and the self-induction pressure. When, however, S is opened, an air gap of great resistance is in- troduced into the circuit with the result that the current diminishes very rapidly and therefore establishes a high electromotive force by self-induction in N'. This electromotive force is sufficient to overcome the resistance of the small air gap formed at the instant of breaking contact and an arc is estab- lished across the gap. The resistance of the arc is considerably less than that of the air gap so that the current may continue to flow for a short time across a considerable arc. The more rapid the opening of the gap, the longer will be the arc. The energy for the arc is almost entirely the magnetic flux through the coil N' and is of comparatively small magnitude. If an additional or secondary coil N" be . wound concentric with the primary coil N', JL+ \ ^]" "k \" as in Fig. 229, the same magnetic changes ^- B c> will occur in both coils and an electro- motive force will be generated in N" p IQ 2 29. High-ten- which is proportional to the number of sion or jump-spark igni- turns in the coil. When the number of turns is very large, a high tension, sufficient to jump an air gap such as ab } will be produced. A coil wound as in Fig. 228 is called an inductance or spark coil. A coil with a double winding as in Fig. 229 is called an induction coil or jump-spark coil. IGNITION 297 The formation of an arc results in the vaporization or burn- ing of the metal of one of the points between which the arc springs and results in deterioration of that point. To reduce the arcing at S, a condenser is shunted around it. The condenser consists of two conductors separated by insulating material; it is usually made of a large number of sheets of very thin metal, such as tin foil, separated by thin paraffined paper sheets. Every other sheet of metal extends to one side and the balance to the other. All the sheets of one side are connected to one terminal and the remainder to another. By connecting the condenser across the switch, S, (Fig. 229) the energy which would otherwise go into the formation of an arc is absorbed in the system. T FIG. 230. Circuit diagram of battery ignition system. The schematic arrangement of a battery ignition system for a four-cylinder engine is shown in Fig. 230. The primary circuit includes the battery, B, switch, S, primary winding on the induc- tion coil, I, the interrupter, breaker, or timer, T, which breaks the primary circuit whenever ignition is required, and the condenser, C, shunted around the timer to prevent arcing. The secondary circuit consists of the secondary winding of the induction coil, the distributor, D, and the spark plugs, p,p,p,p,; a safety spark gap, G, (see p. 310) is shunted on this circuit. The revolving arm of the distributor, D, establishes contacts successively with the four spark plugs in any desired order; the interrupter, !T, breaks the primary circuit and the current thereby generated in the secondary circuit arcs across the spark plugs. The circuits are grounded as indicated. The Magneto. Most airplane engines at the present day have magnetos as sources of electric current. A magneto differs 298 THE AIRPLANE ENGINE from a dynamo or electric generator in having permanent mag- nets in place of electro-magnets for the fields. In Fig. 231 is shown the action of an armature type magneto, consisting of pole pieces, N, S, which are permanent magnets, and an armature, AB, consisting of core and end pieces, revolving between the shoes of the pole pieces. The clearance ("air gap") between armature end pieces and magnet shoes is only about 0.005 in. A coil is wound on the armature core, one end of the coil being grounded; the other end is carried away, insulated, through a collector ring and brush. As the armature revolves (being driven from the engine shaft) the lines of magnetic force take the successive directions indicated by the long arrows. The magnetic circuit is NABS for positions I and II. In the vertical position flux through the core ceases, and no current is generated FIG. 231. Armature type magneto. in the coil. As the armature passes the vertical position, the circuit reverses to NBAS. This continues for 180 deg. more, when the original direction of flow is restored. The strength of the magnetic field influencing the armature coil is greatest at horizontal positions of the armature; but the rate of change of field strength is greater near the vertical positions, where the direction of magnetic flux is reversing itself. The air gap (in a construction like Fig. 231) is then large, so that the maximum effective rate of change occurs shortly after leaving positions II and V. Hence at these positions, twice in every revolution of the armature, the induced current reaches a maximum value, and is capable of producing a vigorous spark. Figure 232 shows the method of variation of the induced current with magneto position. Starting at position II, Fig. 231, the magnetic flux begins to diminish and has completely reversed itself by the time position III is reached. The duration of this period depends on the width of the armature end pieces. From position III to V there is practically no induced current. IGNITION 299 A high-tension magneto differs from that just described in that it has both primary and secondary coils wound on the same armature. Both coils link with the same magnetic circuit and therefore the armature becomes an induction coil and replaces the separate induction coil which would otherwise be necessary. Figure 233 shows a high-tension mag- neto. The two windings are shown with one terminal grounded to the machine. The primary coil is short- circuited by the contact at the inter- rupter, at M, until the proper moment, when it is opened suddenly and the induced high-tension current goes through the distributor to one of the spark plugs. The magneto and inter- rupter must be properly synchronized so that the break occurs when the primary e.m.f. is a maximum. The switch, when closed, short-circuits the primary circuit and thereby prevents the building up of a high-tension current in the secondary circuit, and so shuts off the ignition. Of the elements shown in Fig. 233 the condenser and inter- rupter are usually incorporated in the actual construction of the Spark P/ugs EL 3E Position of Armature FIG. 232. Induced current in armature type magneto. Condenser . "Ground" FIG. 233. Circuit diagram of high-tension magneto ignition system. magneto. The distributor may also be incorporated when the magneto speed is one-half the engine speed. The ordinary construction of a magneto with- revolving armature gives sparks at 180-deg. intervals corresponding to the positions 300 THE AIRPLANE ENGINE (II and V, Fig. 231) of maximum induced current. With Vee type engines it may be necessary to have unequal time in- tervals between sparks; for example, with a two-cylinder 45-deg. Vee engine the sparks instead of occurring at 180-deg. rota- tion of the armature should occur alternately at 157^-deg. and 202 J^-deg. intervals. This unequal interval can be obtained in various ways. In one of the constructions of the Bosch Magneto Co. the armature end-piece is cut away on opposite I H EL FIG. 234. Magneto with unequal firing intervals (Bosch). sides of each half of the core so as to increase the air gap and the tips of the pole shoes are also cut away on diagonally opposite halves of the two poles so as to make the positions of maximum induced current (II, Fig. 231) come earlier. The construction and operation are illustrated in Fig. 234. The large air gap, B, effectively cuts off the lines of force. Maximum induced current will occur shortly after the armature has left the trailing pole tips C - D (position II) and also after the armature has left the trailing pole tips E-F (position IV). These two positions FIG. 235. Inductor magneto. are made less than 180 deg. apart as a result of cutting away tips of the pole pieces at E and F. The magneto with revolving armature has to be provided with insulated moving wires, collector rings, brushes, and moving contacts to convey the induced current from the armature to the stationary conductors. To avoid this complication a rotor or inductor type of magneto, with stationary windings, is often used. Figure 235 shows a construction with a rotating element, IGNITION 301 or inductor, consisting of two cylindrical segments of soft iron; all the rest of the magneto is stationary. The magnetic condition of the armature core depends on the position of the inductor. In the positions A and C the segments form a magnetic bridge between the magnet poles and the heads of the armature core; in 90 180 Rotation of Inductor - > FIG. 236. Induced current in inductor magneto. these positions the magnetic flux is a maximum. In passing the positions B and D the magnetic lines are abruptly changed in direction and a vigorous induced current is set up. The reversal takes place four times per revolution of the inductor and succeed- ing reversals give current in opposite directions. This inductor magneto can give twice as many ignitions per revolution and consequently has to be rotated only half as fast as the rotating armature type of magneto. All the elec- trical connections are stationary. Typical current curves are shown in Fig. 236. Another construction of inductor magneto is shown in Fig. 237. The rotor is a steel shaft carrying two laminated soft-iron arms with a space between them which is occupied by the stationary winding (not shown) . The arms project on opposite sides of the shaft and are of such radius as to give the smallest practicable air gap between them and the pole shoes. The magnetic flux in the position shown is from N to R, then back along the shaft to the other arm and so to S. When the inductor has rotated 180 deg. from the position shown, the flux will be from N to the rear arm, then forward along the shaft to R and S. The flux through the shaft is reversed twice every revolution and induces a cur- rent in the winding around the middle length of the shaft. FIG. 237. Inductor magneto with two arms. 302 THE AIRPLANE ENGINE The Dixie magneto uses a different type of inductor. The rotor, Fig. 238, consists of two revolving wings, N and S, sepa- rated by a bronze center-piece, B. Each wing is always in contact with one pole of the magneto (Fig. 239) and consequently keeps the polarity of that pole. The rotor is surrounded by the field structure, shown in Fig. 240, which carries laminated pole exten- sions on which the winding with its core is mounted. As the FIG. 238. Rotating element of Dixie magneto. rotor revolves the direction of magnetic flux through the core changes twice every revolution. Construction of Magnetos. The constructive features of a Bosch high-tension magneto of the rotating armature type are shown in Fig. 241. The armature rotates at engine speed and gives two electrical impulses per revolution. The distributor is geared to the contact breaker and rotates at half its speed. The end of the primary winding is connected to the brass FIG. 239. Diagrammatic outline of Dixie magneto. FIG. 240. Field structure of Dixie magneto. plate, 1. In the center of this plate is screwed the fastening screw, 2, which serves, in the first place, to hold the contact breaker in its position, and, in the second, to conduct the primary current to the platinum screw block, 3, of the contact breaker. Screw, 2, and screw block, 3, are insulated from the contact breaker disc, 4, which has metallic connection with the armature core. The platinum screw, 5, goes through the screw block, 3. Pressed against this platinum screw by means of a spring, 6, is the IGNITION 303 contact-breaker lever, 7, which is connected to the armature core and to the beginning of the primary winding. The primary winding is therefore short-circuited as long as lever, 7, is in contact with the platinum screw, 5. The circuit is interrupted 19 Longitudinal Sec+ion Rear View FIG. 241. Constructive features of Bosch high-tension magneto. when the lever is rocked. A condenser, 8, is connected in parallel wi-th the gap thus formed. The end of the secondary winding leads to the slip ring, 9, on which slides a carbon brush, 10, which is insulated from the Primary Wina/incf Secondary Winding Frame Con-Fact Breaker Disk FIG. 242. Wiring diagram for Bosch high-tension magneto ignition system. magneto frame by means of the carbon holder, 11. From the brush, 10, the secondary current is conducted to the connecting bridge, 12, fitted with a contact-carbon brush, 13, and through the rotating distributor piece, 14, which carries a distributor carbon, 15, to the distributor disk, 16. 304 THE AIRPLANE ENGINE In the distributor disc, 16, are embedded four metal segments, 17. During the rotation of the distributor carbon, 15, the latter makes contact with the respective segments, and connects the secondary current with one of the contacts. The contact breaker is fitted into the rear end of the armature spindle, which is bored out and provided with a keyway. The short-circuiting and interrupting of the primary circuit is effected by means of the contact-breaker lever, 7, and the fiber rollers, 19. As long as the lever, 7, is pressed against the contact screw, 5, the pri- mary circuit is short-circuited, and the rocking of the levers by the fiber rollers, 19, breaks the primary circuit; at the same moment ignition takes place. The distance between the platinum points, when the lever is lifted on the fiber rollers, must not exceed 0.5 mm. (approximately J^o in.). This distance may be adjusted by means of the screw, 5. FIG. 243. Bosch interrupter-distributor. Another type of Bosch interrupter-distributor is shown in Fig. 243. This is connected at 1 to the engine cam shaft and carries a cam, 15, which has as many lobes as there are cylinders to the engine; four lobes are shown. The interrupter lever, 16, is held against this cam by the spring, 17, and when the rubbing plate of the lever is between two lobes the platinum points, 18 and 19, are in contact; the contact is interrupted at the passage of each lobe and the primary circuit is thereby broken. The distributor rotor is on the same shaft and carries a rectangular brass tube in which is located the carbon brush, 11. This brush sweeps the cylinder cavity in the distributor body, 7, and the con- tacts, 8, which are as numerous as the cylinders. The central carbon brush, 10, keeps contact with the rectangular brass tube. Adjustment of the interrupter is by rotation of the whole dis- tributor through the timing arm, 6. IGNITION 305 Permanent magnets are of steel alloyed with 5 per cent of tungsten or with chromium. All parts at which sparks may occur (breaker, distributor, safety gap) should be enclosed to reduce the fire risk. The distributor speed is half the engine speed on all stationary four-cycle engines; the distributor may either be incorporated in the magneto or be driven direct by the cam shaft. The Dixie distributor for an 8-cylinder engine is shown in Fig. 244. The rotor carries two carbon brushes, which in an 8-cylinder engine are 180 deg. 45deg. = 1 35 deg. apart, and in a 12-cylinder engine are 180 deg. - 30 deg. = 150 deg. apart. These brushes are not in the same plane of revolution. In the plane of the outer brush Rotor Carbon Brushes Collector Brushes FIG. 244. Dixie distributor for 8-cylinder engine. are located four metal segments embedded in the insulating distributor block; the other four segments are located immedi- ately behind the first four in the plane of the second carbon brush. Contacts are made each 45 deg. of rotation of the distributor rotor. The collector brushes are in continuous contact with the secondary circuit and with the carbon brushes. When rubbing contact is employed in a distributor a deposit of carbon from the brush will be left on the distributor block which must be cleared off periodically. To avoid this a gap distributor is sometimes used with a nickel point and a small air gap (from 0.01 to 0.02 in.) across which the current arcs. The use of a gap distributor has the additional advantage of increasing the secondary voltage when the spark plug has its resistance lowered either by carbon deposit or high temperature, 20 306 THE AIRPLANE ENGINE and thereby giving a spark under conditions in which it would otherwise fail (see p. 310). The spark advance in airplane engines is generally fixed at about 30 deg. A slightly greater advance is desirable at high altitudes, but the advantage from its use is so slight and the complication of an additional control so undesirable that spark control is not used in airplane practice. Spark adjustment is obtained by adjusting the breaker. A corresponding adjustment of the magneto is sometimes provided. Figure 245 shows the adjustable driving gear arrangement of the magneto of the King- Bugatti engine. The bevel gear on the magneto shaft is fitted Packed with Soft' Grease FIG. 245. Adjustable driving gear for Bugatti engine magneto. . on a taper with a key. The gear has eight key ways, so spaced that the magneto timing may be set within 1J^ deg. of any desired position. The gear which adjusts the spark advance has four internal spiral grooves sliding over splines on the sleeve, which is keyed to the driving shaft, but may be moved along the shaft by a lever. Movement of this sleeve revolves the magneto driving gear with relation to the shaft-driving gear. Adaptation to Engine. In four-cycle engines ha vingn cylinders, there are ~ sparks necessary per engine revolution. If these are supplied by a magneto giving m sparks per magneto revolution, the ratio of magneto speed to engine speed is ^. Since m v aries from one to four (the interrupter may work only once per IGNITION 307 revolution even though conditions are right twice), the speed ratio varies from -' to -. The great majority of magnetos give 2 o two sparks per revolution, and run at 7 times engine speed. A multiple-spark magneto runs at relatively low speed. In an engine with equal ignition intervals, one magneto may serve any number of cylinders. Thus a two-spark magneto for 15 cylinders would run at 3% times engine speed: the same magneto for 9 cylinders runs at 2^ times engine speed. With unequal ignition intervals, a special magneto (see page 300) or a plurality of magnetos must be employed. The cycle of operations of a jump-spark ignition system 1 can be considered as consisting of five periods. For quan- titative values a typical magneto may be considered to have the constants given in the following table. CONSTANTS OF TYPICAL MAGNETO Primary turns (N\) 160 Secondary turns (N z ) 8,000 Ratio of turns (n) 50 : 1 Primary resistance (Ri) 0.5 ohm Secondary resistance CR 2 ) 2 , 500 ohms Primary inductance (Z/i) . 015 henry Mutual inductance (M) 0.74 henry Secondary inductance (L 2 ) 36 henrys Primary condenser (CO 0.2 microfarad Secondary (distributed) capacity (C) 50 micro-microfarads Normal speed of operation 2,000 r.p.m. Primary current at break (7 b ) 4 amperes Maximum current in spark . 075 amperes Breakdown voltage of gap 5 , 000 volts Sustaining voltage of gap 600 volts Period I includes the building up of current in the primary winding as a result of either the impressed voltage from a battery or the voltage generated by the rotation of a magneto armature. During this period the breaker or interrupter is closed and the armature rotates from the position of maximum flux to the position where the interrupter opens. For the typical magneto this corresponds to about 100-deg. rotation and lasts 0.008 sec. at 2,000 r.p.m.; the current builds up to 4 amperes. Typical 1 See Report 58, 5th Annual Report, Nat. Adv. Comm. Aeronautics. 308 THE AIRPLANE ENGINE curves for armature flux are shown in Fig. 246, in which A shows the flux with open primary circuit and is due to the permanent magnetos only. B and C give the total flux under normal operat- ing conditions at 500 and 2,000 r.p.m. respectively. Period II is the very short period (about 0.00002 sec.) extend- ing from the opening of the interrupter to the breakdown of the spark gap in the engine. During this period, the magnetic energy of the coil is in part transferred into electrostatic energy and charges the condenser and the capacity of the secondary circuit. The primary current flowing into the condenser against 50 50 100 150 200 250 300 Angle of Rotation (Degrees) FIG. 246. Typical curves for armature flux of magneto. a constantly increasing e.m.f. will decrease at a constantly increasing rate. The decrease in magnetic flux resulting from this decrease of primary current generates an e.m.f. in the sec- ondary windings which in turn sends a charging current into the distributed capacity of the secondary circuit. If the spark gap were not present this process would continue until the magnetic energy had been entirely converted into electrostatic energy; the maximum voltage which would be reached in the typical magneto would be about 70,000 volts. As a result of loss of energy due to resistances, eddy currents in the iron core, etc., IGNITION 309 this maximum voltage is greatly reduced. The curves of Fig. 247 show the rate of rise of secondary voltage as calculated; (A) with no energy loss except that in the resistance of the wind- ings, (B) with the usual eddy currents. It is assumed that the spark gap does not break down and there is no arcing at the interrupter points. Period III is the very short period (about 0.00005 sec.) begin- ning at the instant at which the spark gap breaks down and lasting until a steady arc is established. When this gap has broken down it affords a conducting path into which the charged secondary capacity discharges. As the secondary is now short- circuited by the arc the current increases rapidly. The energy 40,000 20 40 60 80 100 120 140 Time After Break'YMiliionthsof a Second) FIG. 247. Typical curves for secondary voltage of magneto. discharged into the gap during this time is about 0.002 joule, which is just about sufficient to ignite the explosive mixture (see p. 312). Period IV extends from the establishment of the gap to the extinction of the spark. During this time there is a steady discharge which lasts for a considerable time (0.003 sec. for a 5-mm. spark gap in air). The cessation of the arc is due usually to the exhaustion of the energy supply, although occasionally it may be extinguished by the closing of the interrupter if the r.p.m. is very high or the spark gap very short. Period V covers the remainder of the cycle during which time both circuits are practically free from current. Of these periods, II is the most important, as it determines whether or not a spark passes at all; the distributed capacity of the secondary circuit is of great moment in determining the maximum voltage in this period. While 5,000 to 6,000 volts 310 THE AIRPLANE ENGINE is usually required to jump the gap, it may be much increased by oil films on the points and a cold cylinder. If the spark plug is fouled with a conducting film of carbon some of the energy will be drained by leakage during Period II. A typical oscillograph showing the variations of the primary and secondary circuits is given in Fig. 248. The maximum current delivered to the secondary circuit is usually from 0.05 to 0.10 amperes. A safety spark gap is sometimes shunted on the secondary circuit (Figs. 230, 241, and 242) to prevent the formation of excessive voltages, and the consequent possible break- ing down of the insulation, in case the secondary cir- cuit is open. This would occur when a spark plug is being tested out of the cyl- inder and is not grounded. 0. 050 ' Secondary Current 4 Amp. Primary Current 275 Second FIG. 248. Typical oscillograph of primary and secondary currents in magneto. The width of the safety gap is from %Q to % in., the higher value being used for high compression in the engine. The air surrounding the safety spark gap becomes ionized and ozone is liberated. If the air is confined the ozone will rust adjacent steel parts and slowly decompose organic insulating materials. A rotary safety gap with one electrode on the gear driving the distributor and the other integral with the distribut- ing metal electrode is used sometimes; the air is churned up and expelled through a suitable gauze window. A series or subsidiary spark gap is frequently used in order to maintain sparking even when the spark plugs are fouled. The series gap is placed in the connection between the plug and the magneto. 1 Investigations at the Bureau of Standards show that it is possible, by the use of a series gap, on an average ignition system, to spark a plug having a resistance lowered to only 4,000 ohms by fouling. At least 100,000 ohms insulation resistance is ordinarily necessary at the plug if a series gap is not used. For example, with secondary current limited to 0.08 1 For elementary theory and results of tests see Report 57, bth Annual Report, Nat. Adv. Comm. Aeronautics. IGNITION 311 amperes and insulation resistance of 50,000 ohms the maximum voltage across the air gap is 0.08 X 50,000 = 4,000 volts. This is not sufficient; 6,000 volts is usually required. The efficacy of the series spark gap is well shown in Fig. 249, giving the results of some tests by Young and Warren. 1 The four resistances indicated were put in parallel with the spark plug to simulate different degrees of fouling. With each resistance the length of the main gap was varied while the series gap was kept constant. The curves show the maximum length of the main gap at which sparking oc- curred. With a parallel resistance of 58,000 ohms the main gap could be increased from 0.9 mm. to 3.4 mm. as the series gap was increased from zero to 0.02 in. The voltage across the main gap was also measured and was found to in- crease from 2,200 to 3,600 by the introduction of a 0.02-in. series gap, when the shunted resistance was 112,000 ohms. The series gap is sometimes in- tegral with the plug; sometimes at the plug but not integral with it; sometimes in the distributor. The amount of the gap should be variable to suit the degree of fouling of the plug. For maximum effectiveness the gap should be at, or integral with, the spark plug. These desider- ata are conflicting. It is not practicable to adjust subsidiary spark gaps at each spark plug, while the engine is operating; it is easily possible to have a single adjustable spark gap at the distributor, but it cannot be adjusted to suit the different degrees of fouling in the different cylinders which it serves. The effect of temperature and pressure on sparking voltage has been investigated at the Bureau of Standards. 2 Sparking voltage is a linear function of the density of the gas and depends on pressure and temperature only as they affect the density. For a typical spark plug with 0.5 mm. gap the sparking voltage 0.03 0.04 0.06 v Length of Auxiliary Spark Gap in Inches FIG. 249. Effect of auxiliary spark gap on length of main gap. Process of Ignition," The Automobile Engineer, March, 1920. 2 See Report 54, 5th Annual Report, National Advisory Committee on Aero- nautics. 312 THE AIRPLANE ENGINE in air varies from 2,800 volts at atmospheric density to 9,400 volts at a density five times as great. Other measurements indicate that the sparking voltage in an explosive mixture of gasoline in air is about 10 per cent less than in pure air and that the change in voltage is proportional to the percentage of gasoline present. Figure 250 shows the observed sparking voltage for plugs with different gaps; No. 1, 1.8 mm. (0.071 in.); No. 2, 1.2 mm. (0.047 in.); No. 3, 2.2 mm. (0.086 in.); No. 4, 0.5 mm. (0.020 in.). The voltage required for a spark plug set at 0.5 mm., in an aviation engine of moderate compres- sion, is about 6,000 volts. 22 20 18 !" 5 14 1234567 Density Refer-ed to Air at 1 Atm, and 273 Decj. Abs.^ent. FIG. 250. Sparking voltages for plugs with different gaps at various air densities. The sparking voltage is not affected appreciably by the material of the electrodes but is diminished by the use of finer points. The dimensions of the points are, however, determined by considerations of mechanical strength and durability. The " fatness" of a spark has no influence at all on the power developed in an engine. If the current is sufficient to charge the plug and its connections to the sparking potential, the maximum engine power will be developed. The energy repre- sented by that condition is usually about 0.002 joule; the energy per spark varies from 0.03 joule in battery systems up to 0.16 joule in the more powerful magnetos. The excess of energy above that necessary for ignition has no discernible effect on the power developed. Battery Ignition. In a storage cell, electrical energy is stored as chemical energy but returns to electrical energy when the cell IGNITION 313 is connected to supply an external circuit. The desired voltage is obtained by connecting a sufficient number of cells in series, forming a storage battery. The chemical reaction in a lead cell may be expressed by the following equation: Charge Pb0 2 + Pb + 2H z SOt = 2PbS0 4 + 2H 2 Positive and Positive plate Negative plate Electrolyte negative plates Electrolyte Discharge Discharge results in the formation of lead sulphate; charging restores the plates to their original conditions of lead sponge (negative) and lead peroxide (positive). The voltage of a fully charged idle cell is 2.05 to 2.10 volts. Discharge lowers the voltage in proportion to the current flowing. Complete discharge is reached at 1.7 volts, at the normal dis- charge rate fixed by the manufacturer. The capacity of a battery is expressed in ampere-hours at normal discharge rate; the capacity increases as the discharge rate decreases. The maximum discharge rate falls as the temperature decreases. The acid or electrolyte is an aqueous solution of density 1 .255 resulting from the addition of 1 part of sulphuric acid of sp. gr. 1.84 to 4}i parts (by volume) of distilled water. The strength and density of the solution fall as discharge progresses; when the density falls below 1.2 the cell needs recharging. Self-sustaining battery systems require a generator for recharging. The system may then be regarded as a generator system on which the battery " floats." The generator furnishes low-tension direct current, and must have a commutator and brushes. In starting, or at low speed (up to 650 r.p.m.), ignition current comes from the battery. At some definite speed, say 650 r.p.m. of the engine, the generator begins to supply the ignition current. Its rate of delivery is then considerably in excess of that needed for ignition and the surplus goes to the battery. The recharging rate has a maximum value of 10 amperes when the battery is nearly discharged. The delivery voltage of the generator may be automatically controlled by a potential regulator. The outlines of the Liberty engine battery ignition system 314 THE AIRPLANE ENGINE are shown in Fig. 251. Figure 252 shows the circuit diagram. It includes a low-voltage generator in connection with a storage Left Distributor Generator I Gen. Arm. Gen. Field Gen Fiefct Gen. Arm. Bcrff. FIG. 251. Liberty engine ignition system. battery of light weight and small liquid content. Through a switch, the current is sent to two distributors which are mounted Non-inductive Resistance Reverse Coil Resistance Unit. Blades Separately Insulated IR * IR Left Distributor Right Dis+ri bu-hr FIG. 252. Liberty engine circuit diagram. on the camshaft housings and are direct-driven by the camshafts. Right-hand distributors supply distributor-end plugs and left- IGNITION 315 hand distributors supply propeller-end plugs, there being two plugs to each cylinder. The entire system, exclusive of plugs and wiring, weighs 35 lb., and consists of generator, battery, switch, voltage regulator, and two distributors. The generator is shown in Fig. 253. It is four-pole, shunt- in. high, and weighs lb. It Brush. Armature - Filial Coils. Lower End Housing. Generator Armature Terminal 'Commutator Upper End Housing wound, 4> in. diameter, is mounted with its shaft vertical on top of the crankcase between the two rows of cylinders at the rear end. The arma- ture shaft extends down- ward into the crankcase, where it is driven by the auxiliary gearing which also drives the vertical shaft. Its speed is 1.5 crankshaft speed. The end housings of the gen- erator field frame are of cast aluminum. Gearing from the armature shaft forms the tachometer drive. The upper hous- ing contains the ball bearing for the shaft, which is the only bearing in the generator proper. This housing also sup- ports the four brush holders: two positive brushes, grounded to the frame. The ground side of the field is grounded through the voltage regulator, (Fig. 252), the generator voltage being determined by the amount of current flowing through the field, which in turn is controlled by the regulator. The armature has 21 slots and is wave-wound. Insulation between commutator segments is slotted down ^2 in. below the surface of the copper bars. The shaft is hollow, and ground through the bearing. The maximum generator voltage is 10 to 10J^ volts. A current of 5 to 6 amperes may be carried without overheating. , Splined End of Armature Shaft FIG. 253. Liberty engine generator. insulated, and two negative, 316 THE AIRPLANE ENGINE The voltage regulator keeps the voltage constant at all speeds above 650 r.p.m. of the engine. It weighs 1J Ib. and is mounted in a cast aluminum cup on the back of the dash behind the switch. It consists of a soft-iron core over which a pivoted iron armature is so mounted as to be normally held away from the core by an adjustable tension spring. When so held, the generator field current passes through a tungsten contact point on the armature to the ground. The core carries three windings (Fig. 252). The voltage winding is of fine wire leading from the positive terminal of the generator armature to the ground. The generator voltage is impressed on this winding. Increase in this voltage increases the core magnetism and opens the contact gap of the regulator armature. This cuts off the direct flow of field current and decreases the armature voltage of the generator. The reverse winding is superimposed on the voltage winding and is also of fine wire, wound in a reverse direction. The non-inductive winding consists of resistance wire wound so as to produce no magnetic effect on the core and so as to be itself free from induction due to changing core-flux. These two windings are connected in parallel from the regulator armature contact point to the ground: they form a permanent high-resistance ground for the field current when the contact is open. The reverse winding rap- idly destroys residual magnetism and enables the spring again to close the contacts, which in regular operation vibrate rapidly. The battery is designed for light weight and no leakage. It is 7 by 4 by 5^ in. and weighs 10)4 Ib. It can provide 3 amperes for 3 hr., which is sufficient energy for dual ignition on 12 cylin- ders. It floats on the line and normally supplies current only when the engine speed is under 650 r.p.m. The ammeter shows whether the battery is charging or discharging. Charging is automatic at speeds above 650 r.p.m. The hard-rubber battery jar has four compartments or cells, each cell (Fig. 254) con- taining 3+ and 4 plates, burned to connecting straps and separated by perforated rubber with wood. Plates are 3 by 3 in., and rest on %-in. bottom ribs. Above the top of each plate is a flat sealing or baffle plate of hard rubber. The top of each cell is further sealed by a rubber cap through which the lead terminal posts extend. These also are sealed by gaskets or by burning. IGNITION 317 In normal position, the electrolyte completely fills the plate compartments. When the battery is turned upside down, the electrolyte seeps through small holes to the compartment which is normally above the plates. This compartment is of such capa- city as to hold all of the fluid in the cell, without danger of over- flow at the vent plug. The switch, Fig. 252, is located on the dash and weighs 1 Ib. It is built on a Bakelite base. The circuits are controlled by Top Connector ,.-~ Venj- PlucfWasfrer- y - Positive Term/no f / / Negative Termtna/ VintPtucf ^ Positive and / Negative Plate Assembly Combined Wood ana? Rubber Separator FIG, 254. Section of storage cell. two aluminum switch-levers operating spring-bronze contact fingers which connect with contacts molded in the base. The left lever supplies the lef.t distributor, the right lever the right dis- tributor. The engine is started on one distributor, and both levers are switched on only when the speed is above 650 r.p.m. (To start on both distributors would require that the battery supply two sets of plugs and would also waste battery current through the generator.) Resistance coils are mounted on the back of the switch in series with the distributor circuits. These prevent an excessive flow of current should the switch be left on with the engine idle. The switch has four external connections: 318 THE AIRPLANE ENGINE positive battery, generator armature, and two to distributors. Two 12-cylinder distributors (Fig. 255) are used, each supply- ing one plug in all cylinders. Each weighs 5J-^ Ib. and is 7% in. diameter by 5% in. high. They are mounted one on each of the overhead camshafts. The transformer coils and breaker mechanism are incorporated. The Bakelite distributor head forms a cover for the breaker mechanism and a seal for the coil. The moving contact is through a soft carbon brush bearing on terminals molded in a hard-rubber track. Breaker Rotor Arm ^ Ffo-for Brush Low Tension Lead to Sw/Jr-h FIG. 255. Liberty engine distributor. FIG. 256. Liberty engine breaker mechanism. The breaker mechanism is operated by a 12-lobed cam, having lobes spaced 22^ deg. and 37^ cleg. (12-cylinder engine). Tungsten contact points are used. Two main-circuit breakers a and 6, Fig. 256, connected in parallel, are provided and are timed to operate simultaneously; the duplication is a precaution- ary measure. The auxiliary-circuit breaker, c, is provided to prevent the production of a spark when the engine is " rocked" or turned backward. This auxiliary breaker is connected in parallel with the other two through a resistance unit (Fig. 252) which reduces the amount of current flowing through it and is so timed that it opens slightly before the other two when the engine is turned in a forward direction. The opening of the main IGNITION 319 breakers then results in the production of a spark. When the engine is turned in a backward direction the two main breakers open first and no spark is produced due to the fact that the current continues to flow through the coil through the auxiliary breaker but in diminished quantity due to the resistance unit. By the time the circuit is opened at the auxiliary breaker the intensity of the magnetic field of the coil has weakened to such an extent that no spark is produced. The whole breaker mechan- ism may be revolved to advance or retard the spark. Spark Plugs. The conditions to which the spark plug is subjected in aviation engines are difficult to meet. The require- ments are: 1. The maintenance of a gap having a breakdown voltage of about 6,000 volts. 2. The maintenance of an insulation resistance of at least 100,000 ohms. 3. Practically complete gas tightness. These conditions must be maintained under pressures of 500 to 600 Ib. per square inch while immersed in a medium which alter- nates, 15 cycles per second, in temperature between 50 and 2, 500 C. and in an atmosphere which tends to deposit soot, and possibly oil, on the surface of the insulator. The inner end of the insulator and central electrode may have an average temperature of 900C. while the body of the insulator, well up in the shell, is in contact with a jacket containing water at 70C. For successful operation the insulating surface must remain clean, the insulator must not fracture or disintegrate under the varying temperatures and no part of the plug must become hot enough to cause preignitions of the charge. The shell is of steel with standard thread. The S. A. E. stand- ard plug dimensions are: outside diameter, 18.2 mm.; pitch diameter, 17.22 mm. 0.02 mm.; root diameter, 16.09 mm.; 16.9 threads per inch; 1.5 mm. pitch. In order to keep down its temperature it is often made with fins for radiating heat. Investi- gations at the Bureau of Standards 1 have shown that brass shells average from 90 to 270F. hotter than steel shells. The most common insulating materials are mica and porcelain. Other materials used are fused quartz, steatite and molded mater- ials such as Bakelite and Condensite. Porcelain of the highest grade is an excellent material except for its brittleness and conse- 1 Report 52, 5th Annual Report, Nat. Adv. Comm. for Aeronautics. 320 THE AIRPLANE ENGINE quent liability to fracture either from temperature or mechanical effects. Mica, while free from this trouble, is more likely to foul in consequence of its rough surface. Fused quartz is free from both the above objections. The insulation resistance of these materials diminishes with rise in temperature. At a temperature of 900F. the order of merit of the insulating materials is mica, quartz, steatite, porcelain. Porcelain plugs which show a resistance of millions of ohms when cold may have a resistance of only 100JOOO ohms at 900F. The most common source of failure in spark plugs is fouling. l It causes more than 50 per cent of spark plug troubles and is most serious at high altitudes. Fouling is due to the deposit of a layer of carbon and causes a short circuit. The carbon deposit results from either or both of two causes: (a) The chilling of the flame by a cool portion of the plug and the consequent incomplete combustion; this effect is particularly common when the mixture is overrich and is of frequent occurrence at high altitudes with an imperfectly compensated carburetor, (b) The decomposition of lubricating oil which is splashed on heated portions of the insulator. The oil itself is an insulator and when it wets a layer of soot in the plug it makes it an insulating layer. Such a deposit chars and becomes more and more con- ducting. The oil acts as a binding material and also increases the rate of deposition of soot since the carbon particles in the flame adhere to it readily. The conduction through the deposit seems to take place through a narrow path where the oil film between the particles has been broken down by electric stress. Such fouling causes misfiring. A method of attempting to reduce the deposit of carbon is to keep the insulation at such high temperature that all carbon deposit is burned off. This can be accomplished by making the insulator with petticoats, ridges or other projections, but there results the danger of preignition, particularly in high compression engines or in those that are not well cooled. Another method is to shield the insulator with a metal baffle plate which protects it from oil spray, but if this is done the flame does not get good access to the insulator and any deposit which has formed will have very little opportunity of being burned away. The use of a series gap (p. 310) is useful in maintaining firing after the plug is fouled. 1 Report 52, 5th Annual Report, Nat. Adv. Comm. for Aeronautics. IGNITION 321 Fouling with oil, either in the form of a surface film over the electrodes or as a drop between the points, will often prevent firing. The breakdown strength of oil is several times that of air, and the voltage required may easily exceed that which the ignition system is capable of delivering. The trouble is intensi- fied if the insulation of the plug is at the same time reduced by a layer of soot, thereby diminishing the maximum voltage which the system can develop. The oil trouble usually occurs on starting but may also be met when the plane is recovering from a long glide during which the engine is turning over slowly and pumping oil into the cylinders. It may sometimes be identified by the sparking of the safety gap (see p. 310). It seems to occur most often when the form of the electrodes is such as to drain the oil away by capillary forces. The only real remedy is to keep down the amount of lubricating oil going to the cylinders at starting and during glides. Cracking the insulator is one of the most common causes of spark plug failure. The thickness of the insulator is usually so great that a clear crack may not interfere with ignition, but after a while the cracks become filled with carbon and form a conducting path. Cracking may result from several causes. The high temper- ature gradient from the hotter inner end of the insulator to the relatively cold shell and the consequent unequal expansion is a frequent cause. Such cracks are most likely to occur at a shoulder or other place where there is a sudden change in diam- eter. Cracking may also occur if the metal parts of the plug are so arranged that their relatively greater expansion produces pressure on the insulator. The mechanical vibration of the engine as a whole may break the porcelain; such breakage often occurs in the outer portion of the porcelain. There is also considerable breakage from accidental mechanical injury such as striking the plug with a wrench. Mica plugs are free from this trouble. If porcelain is used it should combine high mechanical strength, low modulus of elasticity, low coefficient of thermal expansion and high thermal conductivity. The porcelain may also be made in two or more pieces-permitting the innermost porcelain to heat and expand considerably, while the outer pieces are cooler. The passage of a spark through the joint between the pieces is prevented by a 21 322 THE AIRPLANE ENGINE wrapping of mica around both the shell and the electrode. It is very difficult to make such plugs gas-tight. In plugs in which the central electrode is cemented in the porcelain, the differential expansion of the two can be taken care of only if the electrode is kept of small diameter. Breakage by mechanical vibration can be reduced if the insulation is cushioned by a considerable thickness of asbestos or other packing material between the shoulder of the insulator and the bushing. One-piece plugs in which the edge of the shell is crimped over the shoulder of the insulator are especially liable to cracking at the edge of the shell. (b) (c) (d) FIG. 257. Types of spark plug construction. A minor cause of failure is the change in the width of the spark gap either through warping of the wires or corrosion. Warping occurs only when the wires are relatively long. Cor- rosion is very slow with the alloy commonly used (Ni 97 per cent, Mn 3 per cent), although a chemical reaction between the cement and the metal of the electrode may sometimes cause the tip to drop off. Gas leakage is an evil in that it causes a rapid heating of the plug if its amount is considerable; such leakage is usually a matter of workmanship rather than of design of the plug. There are two joints to keep tight, that between the central electrode and the insulator, and that between the insulator and the shell. The general methods of construction are shown in Fig. 257. In the screw bushing, a, the insulator has a shoulder, one side of which is seated on a shoulder in the shell while a bushing is screwed down inside the shell on the opposite side. A gasket IGNITION 323 of brass, copper, asbestos, or some soft heat-resisting material is used and can be placed on either side of the shoulder of the insulator, or on both. With a mica insulator it is possible to dispense with the gasket. To relieve the insulator from the mechanical strain resulting from differential expansion of the shell and the insulator such constructions as those shown diagrammatically in Fig. 258 may be used. In a the shell A and sleeve B are made of different metals and B is of such length as to maintain constant pressure on the washer. The expansion of the central electrode is compensated in a similar manner. In b the steel clamping nut is very thin and flexible. Expansion of the central electrode is provided for by a strong spring washer under the nut B. A r-S (a) (6) FIG. 258. Special spark plug constructions. The crimped shell, b (Fig. 257), is most common (Champion, Titan, etc.) and is formed by forcing the top edge of the shell over a gasket, which rests on the upper side of the shoulder of the insulator. These plugs cannot be disassembled. The taper fit, c (Splitdorf), is used with a mica or steatite insulator. The mica does not stand well the pressure exerted on it during assembly. If a thin steel jacket is placed over the taper it protects the mica and is flexible enough to form a gas- tight fit with the shell. A molded-in insulator, d (Anderson), consists of glass which has been forced between the central electrode and the shell while in the molten state. It adheres to both electrode and shell and is gas-tight. The shape and arrangement of the electrodes seem to have little effect on the operation with the exceptions already noted of the greater liability to fouling with oil of plugs in which 324 THE AIRPLANE ENGINE the side wall of the shell forms one of the electrodes. The variation in breakdown voltage with the shape of the tips is slight. With fine wires any oil film at starting is burned off rapidly but there is greater liability to preignition. With a central electrode consisting of a disc (Fig. 257 c) the danger of short-circuiting with carbon is increased while the likelihood of complete fouling with oil is diminished and greater protection is afforded to the insulating material back of it. The location and number of spark plugs used are important in their effect on engine performance. An effort should be made to reduce as much as possible the distance through which the flame has to be propagated. The greater that distance, the greater is the liability to detonation. If the distance is shortened, a higher compression may be employed without detonation. Where a single plug is used its location should be in the center of the head. Multiple spark plugs are desirable, not only to insure ignition in case of failure of one plug but also because they permit higher compression pressures. With a constant com- pression pressure the power is increased by increasing the number of plugs; for example, in a 5J^ by 6^ -in. four-valve single- cylinder test engine with a compression ratio of 5.4 and at 1,800 r.p.m., the brake horse power was increased 5.7 per cent by the use of two spark plugs and 11.1 per cent by the use of four plugs. A typical wiring diagram is shown in Fig. 259, which shows the wiring of a 12-cylinder Vee engine equipped with two mag- netos for regular operation, and a starting magneto. The regular magnetos also have radio connection. The relative advantages of battery and magneto ignition have been much debated. The performance of the engine is not affected by the source of primary current. With battery ignition the engine is started by turning it with the current on but fully retarded. With magneto ignition the engine is pulled over a few turns with no current on so as to fill the cylinder with an explosive mixture, and a starting magneto is then operated by hand, giving a shower of sparks in one cylinder; the starting magneto is arranged to be fully retarded. The element of danger in starting the engine is eliminated by this latter method of- operation. The battery requires more attention than the magneto, particularly if the engine is to stand idle for a while; if it runs down there remains no means for starting the engine. IGNITION 325 The battery system is also more complicated than the magneto system but it lends itself better to irregular explosion intervals; the magneto must be of special type in this case (see p. 300). The generator in the battery system requires attention to keep commutator and brushes in good condition; there is no corre- sponding attention necessary with the magneto. The total weight of the magneto system, including dual magnetos and a starting magneto, is somewhat greater than that of a battery Right Distributor- Left Distributor. PropelterEncf 'Right Distributor Left Distributor Left Magneto . Radio .^ Connect/on -*. Radio Connection Righf Magneto Left Distri butor Atro Switch f^\ Right Distributor. HIT Starting ..' Magneto, FIG. 259. Typical wiring system for 12-cylinder Vee engine. system. A magneto system is easier for the pilot to operate; there is only one switch handle to control and no ammeter to be watched. The battery system has distinct advantage when current is required for an electric starter, lights and other uses. The firing order adopted in actual engines is given below. 8-cylinder 90-deg. Vee 1L, 1R, 2L, 2R, 4L, 4R, 3L, 3R (Curtis OX5, VX; Sunbeam Arab) 1L, 4R, 2L, 3R, 4L, 1R, 3L, 2R (Hispano-Suiza) (counting from propeller) 12-cylinder 60-deg. Vee 326 THE AIRPLANE ENGINE 12-cylinder 45-deg. Vee: 1L, 6R, 5L, 2R, 3L, 4R, 6L, 1R, 2L, 5R, 4L, 3R (Liberty) (counting toward propeller). 1L, 2R, 5L. 4R, 3L, 1R, 6L,5R, 2L, 3R, 4L, 6R (Rolls-Royce Falcon and Eagle) 1L, 1R, 5L, 5R, 3L, 3R, 6L, 6R, 2L, 2R, 4L, 4R (Sunbeam Maori, Cossack) 9-cylinder Rotary: 1, 3, 5, 7, 9, 2, 4, 6, 8 (Gnome, LeRhone, BRl, BR2, Clerget) 6-cylinder Vertical: 1, 5, 3, 6, 2, 4(Beardmore, Galloway, Siddeley, Austro- Daimler, Mercedes, Fiat) 7-cylinder Radial: 1, 3, 5, 7, 2, 4, 6 (A. B. C. Wasp) 9-cylinder Radial: 1, 3, 5, 7, 9, 2, 4, 6, 8 (A. B. C. Dragonfly) CHAPTER XIII LUBRICATION All rubbing surfaces in an engine should be lubricated. The most important of these surfaces are the cylinder walls, main bearings, crankpins, piston pins, and camshaft bearings, but there are also numerous other parts to be lubricated. The coefficient of friction of a bearing with good lubrication, moderate pressures and high speeds is practically independent of the materials composing the rubbing surfaces, but is proportional to the viscosity of the oil, to the rubbing speed and to the area; it is independent of the pressure. For high pressures and low speeds these laws do not hold; for velocities from 100 to 500 ft. per minute the coefficient decreases about as the square root of the velocity, for velocities from 500 to 1,600 ft. per minute it decreases about as the fifth root of the velocities, while above 1,600 ft. per minute it is practically constant. With high pres- sures the coefficient of friction increases. In an airplane engine the loads on the principal bearing surfaces are variable, going through a cycle of changes every two revolu- tions of the engine, and varying from a maximum to a low mini- mum. For example, in the Liberty engine the total force on the crankpin varies from 4,980 to 1,500 lb.; on the intermediate main bearing from 7,250 lb. to 800 lb. ; on the end main bearing from - 4,025 lb. to zero; on the center main bearing from 7,700 lb. to 2,500 lb.; the piston side thrust from 930 lb. to zero. Further- more, the direction of the force changes in these principal bearing surfaces, so that the portion of the bearing which at one instant is supporting maximum pressure is later relieved of all pressure. This intermittent application of the load is favorable to good lubrication and permits the use of maximum pressures greatly in excess of- what would be possible with continuous loading. The oil film which is squeezed out by the application of the maxi- mum pressure is replaced during the reduction or reversal of the pressure. The maximum load per square inch of projected area is great- est on the piston pin, which has a diameter considerably less than 327 328 THE AIRPLANE ENGINE the crankpin; in the Liberty engine this is 2,580 Ib. per square inch as against 932 Ib. on the crankpin; 1,675 Ib. on the center main bearings; 1,580 Ib. on the intermediate main bearings; 815 Ib. per square inch on the end main bearings. The rubbing speeds of the main bearings of airplane engines range usually from 16 to 20 ft. per second; the rubbing speeds at the crankpins will be somewhat lower. The total friction work at any bearing is proportional to the product of the mean total load by the rubbing speed. The limit- ing factors for a bearing for continuous operation are the mean pressure per square inch of projected area and the rubbing speed. The product of these two is a good index of the service of the bearing. In the Liberty engine this load index is 13,500 Ib.-ft. sec. for the crankpin; 24,670 for the center main bearing; 13,650 for the intermediate main bearing; and 11,900 for the end main bearings. The permissible pressure on the bearing depends on the viscos- ity and therefore on the temperature of the oil. The temperature tends to rise and must be kept down by oil cooling. Tempera- tures of 160F. and higher are common. The lubrication of the cylinder offers problems quite different from the lubrication of the rest of the engine. The side thrust pressures are moderate; the piston speed is high, reaching 2,000 ft. per minute, or 33 ft. per second, and the maximum speed (at mid-stroke) is about 52 ft. per second. The friction work under these conditions would not be a serious charge against the engine if the oil film could be maintained in good condition, but as pointed -out on page 24 the viscosity of the oil film on the cylinder walls is greatly raised by carbonization. The oil film on the walls and around the piston rings has to serve another purpose besides acting as a lubricant; it acts as a seal to prevent the blowing of the gases past the piston. For this purpose high viscosity is useful. In starting cold, in idling with overrich mixtures, and in cold weather, a certain amount of liquid fuel will meet the cylinder walls and, being perfectly miscible with the mineral lubricating oil, it will dilute the oil film and the thinned oil will then run down the cylinder walls and dilute the oil in the crankcase. This phenomenon, which is very common in automobile practice, is not so usual in airplane engines because of the higher volatility of the aviation fuels. The ordinary gasoline of commerce has a high end point, which means that it has kerosene constituents LUBRICATION 329 which will not vaporize during admission and will be deposited in the liquid form on the cylinder walls. With aviation fuel the dilution when it occurs is by more volatile elements which tend to vaporize out from the hot body of oil so that the dilution of the crankcase oil is not cumulative as with automobile engines. The friction at the bearings of an engine results in heat which has to be taken away as fast as it is generated if the bearings are not to rise in temperature. Much of this heat is conducted through the metal to cooler parts of the engine but part is carried away by the oil itself. For this purpose a large flow of oil is desirable. The amount of oil circulated in the Liberty engine is about 12 gal. per minute and the temperature rise may average about 10F. This corresponds to a heat abstraction of about 45 B.t.u. per minute. If the bearing friction work is taken as 1% Ib. per square inch of piston area (see p. 24), the correspond- ing heat generated will be about 200 B.t.u. per minute. The volume of oil circulated in this case is not sufficient for complete cooling of the bearings. Viscosity. The oil which gives minimum temperature rise of the bearing is the best to use, other factors being equal. An oil of lower viscosity will cause greater friction because it will squeeze out and allow a closer contact of metallic parts and increase metallic friction and wear. An oil of higher viscosity results in increased fluid friction of the oil. With a complete film of oil, the oil flows like a pack of playing cards sliding over each other, the outer layers adhering to the surfaces and not sliding with reference to them. The actual fluid friction F t in pounds, is given by the equation PXAXV 5,760 X t where P is the absolute viscosity in poises; A is the rubbing area in square inches; V is the rubbing velocity in feet per second; and t is the oil film thickness in inches. This formula indi- cates that the friction diminishes as the thickness of the film increases. The measurement of viscosity has been standardized in this country and is determined by the Saybolt Universal Viscosimeter in which the fluid to be analyzed flows through a tube 0.1765 cm. diameter, 1.225 cm. long, under an average head of 7.36 cm., from a vessel 2.975 cm. diameter. The time in seconds required for 60 c.c. of oil to flow through the tube is the viscosity in seconds 330 THE AIRPLANE ENGINE Saybolt. The absolute viscosity is obtained from the equation 100 P = where G is the specific gravity of the oil and S is the Saybolt viscosity. One of the most troublesome features of lubrication is the decrease in viscosity of oils with rise in temperature. It is found that if the logarithm of the Saybolt viscosity is plotted against the temperature (Farenheit), on cross-section paper, the points lie close to a straight line this is a purely accidental relation. The viscosities of the principal American lubricating oils at temperatures of 100, 150 and 212F. are given in Table 16. It will be noticed that the very heavy oils fall off most rapidly in viscosity as the temperature is raised so that whereas the range in Saybolt viscosities of the oils tabulated at 100F. is about 11 to 1, at 212 it is only 3 to 1. The desired physical characteristics of a lubricating oil are as follows : (a) Body sufficient to prevent metallic contact under maximum pressure and maximum temperature. (6) Lowest viscosity in keeping with the above conditions. (c) Capacity of resisting high temperatures without decomposition. (d) Fluidity at minimum temperatures. (e) High fire test. (/) Freedom from oxidation. (g] Freedom from corrosive action on metals. The standard specifications for lubricating oils for airplane engines adopted by the U. S. Army and Navy are given below. Grade 1 is the Navy specification, Grade 2 the Army. The specifications are also different for summer and winter use. Flash Point. The flash point of Grade 1 shall not be lower than 400F.; for Grade 2 not lower than 500F. Viscosity at 210F. shall be within the following limits: Grade 1 (summer) ........................... 90-100 sec. Grade 1 (winter) ............................ 78- 85 sec. Grade 2 .................................... 125-135 sec. Pour Test. Grade 1 not above 45F. for summer, or 15F. for winter. Cold Test. Grade 2 not above 35F. LUBRICATION 331 TABLE 16. PROPERTIES OF REPRESENTATIVE AMERICAN LUBRICATING OILS FOR USE IN INTERNAL COMBUSTION ENGINES Kind of oil Physical properties Baume grav. Flash, deg. F. Deg.F. burn Deg. F. chill Viscosity, seconds Open cup Closed cup 100 F. 150 F. 212 F. Havoline: Light .... 25.9 25.0 25.6 28.1 21.8 26:3 23.3 21.1 25.8 27.6 26.0 28.9 24.7 29.1 24.9 29.3 24.3 25.4 21.3 20.9 19.3 26.2 27.1 26.3 27.6 24.7 24.7 26.2 26.1 28.6 27.6 15.0 370 385 395 370 360 500 370 370 460 360 375 430 465 400 390 420 395 385 335 350 356 455 450 435 440 440 460 410 465 415 485 380 395 410 380 360 470 380 585 360 370 445 425 410 400 430 410 380 340 350 360 450 445 435 430 450 460 460 ... 430 450 455 420 420 580 425 430 540 410 430 505 535 470 450 495 470 450 380 400 420 535 530 520 515 520 540 480 550 475 550 33 34 46 24 41 6 8 46 20 23 34 58 26 32 40 SI at 35 SI at SI at 10 39 38 38 34 28 27 33 44 32 46 173 237 361 167 330 1,640 221 300 926 140 289 340 1,583 181 243 316 219 300 205 301 495 795 814 517 513 413 474 329 334 1,196 1,270 66 80 111 66 97 397 74 87 243 60 95 108 356 71 81 103 77 103 69 85 119 212 222 149 151 135 134 107 355 108 300 305 42 46 54 44 49 122 45 46 86 41 50 55 110 45 47 54 46 51 42 46 51 78 80 63 64 55 58 52 111 52 100 90 Mobiloil: "E" Light " A " Medium "B" Heavy Arctic Lt Med Arctic Medium "BB" Med. Heavy Monogram: Light Medium Heavy Ex. Heavy Perfection: "A" Light "C" Heavy Socony: Zero Texaco: Light Medium Heavy Veedol: Aero No. 1 Aero No. 2 Aero No. 3 Aero No 6 Zero Heavy Zero Extra Heavy Wolf's Head: Heavy No 8 332 THE AIRPLANE ENGINE Acidity. Not more than 0.10 mg. of potassium hydroxide shall be required to neutralize 1 gram of Grade 1 oil. Emulsifying Properties. The oil shall separate completely in 1 hr. from an emulsion with distilled water at a temperature of 180F. Carbon Residue in Grade 1 shall not be over 1.5 per cent; in Grade 2 not over 2.0 per cent. Precipitation Test. When 5 c.c. of the oil is mixed with 95 c.c. of petroleum ether and allowed to stand for 24 hr. it shall not show a precipitate or sediment of more than 0.25 c.c. The oil shall not contain moisture, sulphonates, soap, resin or tarry constituents. The Flash Test shows the temperature at which the vapor from a sample, heated in an open cup, will ignite. It has some relation to loss by evapora- tion. The open cup is 2^ in. diameter, 1% 6 in. high and is filled to within % in. of the top. It is placed on a metal plate and heated so that its tem- perature rises not less than 9 nor more than 11F. per minute. A test flame ^2 in. in diameter is passed across the top of the cup, taking 1 sec. for the passage, at every 5F. rise of temperature of the oil. The flash point is the temperature at which a flash appears at any point on the surface of the oil. Drafts must be avoided. The Viscosity Test has been discussed on page 329. The Pour Test indicates the temperature at which a sample of the oil will just flow. The oil is placed in a glass jar \Y in. diameter and 4 to 5 in. long to a depth of about Y in. and the jar is corked. It is then placed in a freezing solution and at each 5F. drop in temperature it is taken out and tilted. The pour test is 5 higher than the temperature at which the oil will not flow when the jar is placed in the horizontal position. The Cold Test has a similar purpose to the Pour Test but in this case the oil is first frozen and is then stirred with a thermometer until it will run from one end of an ordinary 4-oz. sample bottle to the other. The temperature reading at that time is the Cold Test. Carbon residue is obtained by heating 10 grams of oil in a porcelain crucible placed inside two iron crucibles with covers. It is heated so as to maintain a vapor flame of specified length and heated further after the vapors cease to come off. After cooling the weight of carbon residue in the porcelain crucible is determined. Reclaiming Oil. The oil used in a stationary airplane engine not only becomes diluted by the heavier constituents of the fuel but also becomes dirty by the accumulation of free carbon from the cylinder walls, of metal particles worn off the bearing surfaces, and of other solid impurities. The oil usually has to be changed between the fifth and twentieth hour of flying service; most oil is not used more than 5 hr. It can be reclaimed by allowing it to LUBRICATION 333 stand 30 hr. in a tall bucket, decanting off the upper two-thirds, filtering and warming to 150F. A more thorough process is to put the oil with some water and soda ash in a steam-jacketed tank, raising the temperature to 212F., forming an emulsion and obtaining a precipitation of carbon, iron, and dirt after a period of rest. The steam drives off the 2 or 3 per cent of diluent coming from the volatile aviation gasoline. A recovery of 85 per cent is possible in this way and the reclaimed oil is at least as good as new oil. A centrifugal oil cleaner has been tried on the Liberty engine with considerable success. This consists of a spun copper bowl, 5 in. diameter, rotating at IJ-^ crankshaft speed; the centrifugal force at 2,550 r.p.m. is about 45 times that of gravity. The oil is led into the center of the bowl and is thrown out at the top. Examination of the contents of the bowl after a run show that it collects metal particles, sand, carbon, rubber and other solids. Its use should increase considerably the periods between changes of oil and should prolong the life of all bearing surfaces by preventing their abrasion by solid particles in the oil. Castor oil is employed in rotary engines in which the gasoline is admitted to the crankcase on its way to the cylinders. Mineral oil (petroleum) cannot be used in this case as it is miscible with gasoline and its use would result in a thinning of the lubricant and a wastage of fuel. Methods of Lubrication. The splash system of lubrication often employed in automobile engines is not satisfactory for airplane engines on account of the high loading at which they are operated and also because of the extreme variations in engine orientation during flight. For the last reason also a wet sump is undesirable since it will deluge some of the cylinders during such airplane evolutions as a nose dive and may result in trouble from excess of oil in the cylinder. Consequently the modern airplane engine is provided with a pressure oiling system and a sump, which is usually kept dry by a scavenger pump. The normal lubricating system is as follows. The scavenger pump or pumps take oil from the sump and deliver it to the external oil tank. The pressure pump takes oil from the oil tank and discharges it into a distributing main from which branches go to each of the main bearings. Oil enters some of the hollow main journals through small holes which register with corresponding holes or channels in the bearing and there- 334 THE AIRPLANE ENGINE by with the branch oil pipes. Usually, alternate journals are rilled with oil and the crank cheeks on the two sides of these journals are drilled to connect with the hollow crankpins and thereby permit lubrication of the crankpins through appropriate holes in them. The oil-containing journals and all the crankpins have closed ends (see p. 144). The lubrication of the piston pin is sometimes carried out by oil pipes running along the connecting rod and registering every revolution with the oil hole in the crank- pin; in other cases the oil thrown out by centrifugal force from the crankpin is relied on to lubricate the piston pin as well as the cylinder wall. The camshaft is lubricated from an oil pipe from the end of the distributing main, connecting with it usually by an annular groove in the front main bearing. The camshaft is hollow and acts as an oil carrier discharging oil through small holes at each bearing. The oil escaping from the bearings lubricates the cams and returns to the crankcase over the distributing gears, meeting there the oil escaping from the main bearings, crankpins and cylinders. The lubricating system of the Liberty engine follows the lines indicated above. The cylinders, pistons and wristpins are lubricated by oil spray from the crankpin. A double scavenging pump at the rear of the engine (Fig. 261) keeps the two sumps drained and returns the oil to the outside oil tank. In the Hispano-Suiza engine (Fig. 51) a dry sump is also used. A single gear-type scavenging pump is driven directly from the rear of the crankshaft; an eccentric sliding-vane pressure pump is mounted on the vertical water- pump shaft and rotates at 1.2 times the crankshaft speed. The vane pump forces the oil through a filter to the main oil pipe in the lower crankcase. There are four oil holes in each crank pin through which oil goes to the bearing and is thrown off to lubricate the cylinder and wrist pin. A small hole is provided in the leading face of each cam to lubricate the cam and its follower. In the Curtiss K-12 (Fig. 55) the pumps are located in the lower part of the crankcase and are driven through a horizontal shaft. There is a triple-gear scavenging pump and two pressure pumps arranged in a unit with the spiral driving gears and surrounded by a filtering screen. The oil is supplied to the main bearings in the usual way and is then conveyed to the crankpins through small tubes built into the crankshaft. The oil is cooled LUBRICATION 335 by a temperature regulator through which the jacket water circulates on its way from the radiator to the pump. The Curtiss OX engine uses a wet sump (Fig. 59) covered by an oil-pan partition. A single-gear pump driven from a beveled gear on the crankshaft at the propeller end sucks oil from the sump and discharges it into the rear end of the hollow camshaft whence it goes through tubes to the crankshaft bearings. In this engine there is a continuous closed oil passage from one end of the crankshaft to the other. The cylinder is lubricated by oil spray. The oil-pan partition has a half-inch hole at its center for the return of oil to the sump. The Hall-Scott L-6 engine (Fig. 63) uses a wet sump and has scavenger and pressure pumps mounted as a unit inside the lower crankcase and driven through an inclined shaft from a bevel gear which is at the rear of the crankshaft. The system is otherwise similar to the Liberty engine. An oil sight gage (Fig. 64) shows the level in the sump. Splash plates in the lower case pre- vent excessive splash from the dipper action of the connecting rods. In the Napier "Lion" engine three oil pumps are combined as a single unit at the extreme rear of the engine ; they are of the gear type and are driven at half engine speed. The suction pumps draw the oil away from the two ends of the lower crankcase by two separate steel pipes (Fig. 72) and discharge to the tank through a common pipe. The pressure pump delivers to both ends of the crankshaft and to the three camshaft casings. There is a con- tinuous passage for oil through the crankshaft. In the Fiat -650 engine (Fig. 76)there are scavenger pumps at the two ends of the lower crankcase and a pressure pump di- rectly under the rear scavenger pump. They are driven by bevel gears from the horizontal tubular shaft inside the crank casing and are mounted in ball bearings as well as the horizontal shaft and the spur gear which drives it. The oil drawn from the main tank is discharged into a copper main cast into the crankcase. The main bearings are fed from copper branch pipes. In other respects the system is normal. In the Benz-230 (Fig. 77) a wet sump is used and the triple oil pump is submerged in the sump. The main pressure pump A (Fig. 260) draws oil from the reservoir in the sump and discharges it to the main bearings through a distributing main and branch pipes. The supply to the piston pins is through small pipes 336 THE AIRPLANE ENGINE inside the tubular connecting rods. Fresh oil is fed into the sump by the small suction pump, B, from the oil tank while the correct working oil level is maintained in the reservoir by the pump, C, whose curved suction pipe (see Fig. 77) terminates at the desired oil level. All return oil passes over the transverse air pipes in the lower crankcase and is cooled by them. In the Maybach engine (Fig. 80) scavenger pumps are mounted at both ends of the crankcase and the pressure pump is placed behind the rear scavenger pump. All three are operated by the same horizontal shaft driven by spur gearing from the front end of the engine shaft. The oil is discharged into an Return Pipe to Main Dtlivtry Pips Tank from /^\ fromTankto PumpB PumpC-... Oil L eve fin Sump. K. JoTank From Tank From Sump ToTank / fB I Q. ; r?,. FromSu. Delivery to Main Bearings From Sump to PumpC To Sump A- Pump Supply ing Main Bearin9 From Sump B- Pump Supplying C-Pump Sump from Returning Oil Tank Oil from Sump To Tank From Sump Supply to fv Pump-A Sump from Pump-B D- End View of Pump Cover. FIG. 260. Triple-gear pump of Benz engine. external oil main on the upper crankcase and past individual screens into branch pipes drilled through the transverse webs into the main bearings. The oil thrown off from these bearings is caught in aluminum scoops bolted to both ends of each crank- pin, is carried by centrifugal force into the hollow crankpin and thence through radially bored holes to the bearing. The piston pins are oiled through the internal pipes in the connecting rods. Baffle plates bolted to the upper crankcase just below the cylin- ders prevent an excess of oil reaching the cylinders. Relief Valves. All pressure-feed systems are provided with a relief valve on the discharge side of the pump. This is a spring- loaded valve as in Fig. 261 and is set for the maximum allowable pressure. The oil pressure is high in starting especially in cold LUBRICATION 337 weather when the viscosity of the oil is very great. The normal operating oil pressures after fully warming up are about 25 to 30 Ib. per square inch in Liberty engines, 50 to 60 Ib. in Curtiss engines, 40 to 65 Ib. in Hispano-Suiza engines. In cold weather it is best to drain off the oil after a flight and to fill up with hot oil before starting. The location of oil grooves in the bearings is a matter of con- siderable importance on which there is much divergence of practice. The actual pressure on the oil film will vary from zero at the ends of the bearings and at the split of the bearing to possibly as much as 10,000 Ib. per square inch in the center of the loaded area at the moment of maximum loading. As the oil pressure does not exceed 50 Ib. per square inch it is obvious that oil cannot be forced in at the place of maximum loading unless the pressure at that place falls below 50 Ib. per square inch during some part of the cycle. There should be no oil grooving length- wise in the middle of the most loaded half of a bearing; such grooves are channels of escape for the oil and may result in such thinning of the film as to increase the friction and, possibly, to cause seizing. The most heavily loaded part of the main bearing is usually the middle of the lower cap and it is at this place that the oil usually enters. The grooves should then be two helical grooves intersecting at the oil hole at the center of the bottom half of the bearing and running to the split but not too near the ends of the bearing. Short helical grooves in the upper half of the bearing may start at the split opposite the lower grooves but should not go more than half-way up each side. Similar groov- ing should be provided at the crankpin bearing, which also is most heavily loaded at the lower half. . - Wherever practicable it is desirable that the oil should enter at the place of minimum average bearing pressure. It is the cyclical variation in the loading that makes possible the proper lubrication of the heavily loaded bearing surfaces of aviation engines. The oil pressures employed are in themselves not nearly adequate to support the loads but are 'required to overcome the viscous and frictional resistance to the flow of the oil to the va- rious bearings and also to ensure that the oil channels will clear themselves of small obstructions. The amount of oil circulated per minute is determined not only by the lubrication needs but also, as pointed out on page 329, by the extent to which the oil is used as a cooling medium. For 22 338 THE AIRPLANE ENGINE lubrication the amount should probably be some function of the total projected areas of the main bearings and crankpins; expressed in this way the use of oil varies from 0.1 to 0.5 Ib. per square inch of projected bearing area per minute. In terms of the power delivered by the engine the oil circulated varies from 0.025 to 0.15 Ib. per horse-power minute. The oil consumption of an engine as usually measured is the amount of oil which has to be added to the system to make up for oil burned or otherwise used up during the engine operation. This quantity varies from 0.02 to 0.05 Ib. per horse-power hour in stationary water-cooled engines but may go as high as 0.15 Ib. in rotaries. Oil Pumps. The great majority of airplane engines use gear pumps both for scavenging and pressure pumps. A simple gear pump consists of a power-driven spur gear meshing closely into an exactly similar driven gear, the gears being enclosed in a casing with the minimum working clearance above and below the gears and also around them except where they mesh. The oil inlet is on the side where the gears separate; the discharge is on the side where the gears meet. If there were no leakage the oil carried from the inlet to the discharge side would be equal to the space between the teeth, but some of this is brought back to the inlet side since a tooth going into mesh does not fill the space between adjacent teeth and consequently does not displace all the oil content of that space. The capacity of each gear can be taken approximately as equal to half the annular space between the roots and tips of the gear, or, for the two gears, as equal to the whole of the annular space of one gear. As the width of this annular space increases with decrease of the number of teeth (decrease of pitch) it is evident that, for a given pitch diameter, capacity can be increased by decreasing the pitch. Two scavenger gear pumps are sometimes combined into a triple-gear pump as in the Liberty engine. In this case the driving gear is central and the inlets to the driven gears are on opposite sides of the pump. The driving gears for scavenger and pressure pumps are usually placed close to one another and driven by the same shaft. In the Benz engine (Fig. 260) three such gear pumps are mounted on the driving shaft and function as indicated in the figure. The Liberty oil pump is driven at one and one-half times crankshaft speed and has a capacity of 1.9 gal. per minute at normal speed. LUBRICATION 339 The pump consists of a double scavenging pump with three gears, A, B and C, Fig. 261, drawing oil from the two sumps and Plan Vi'ew of) Oil Pump Engine Sump-Oilfo Tank s~fs*L' from Engine Oil from Engine Sump to Upper Pocket - ' Gears in Lower Pocket , If 'n Upper Pocket .. Oil from Lower / Poctef Under Pressure Oil from Tank Dram Plug Cnoss-sec-hon of Liberf^-12 Oil Pump FIG. 261. Gear pump of Liberty engine. a pressure pump immediately below it giving an oil pressure of 35 to 50 Ib. per square inch at engine speeds from 1,500 to 1,800 r.p.m. The pressure gears, A', C', (Fig. 261) are immediately 340 THE AIRPLANE ENGINE below the gears A and C of the scavenging pump. The gear A' is driven from the pump shaft and the upper train is operated through a vertical-shaft connection from C f to C. The pressure pump draws oil from the tank through a copper pipe, the oil passing through the large lower strainer and entering the gear housing at M ; it is discharged through the passage NOP to the distributing manifold. The pressure relief valve is shown in 1800 2000 Rev per Min of Engine FIG. 262. Performance curves of gear pump of Packard-180. Fig. 261; the spring is usually set for a pressure of 50 Ib. per square inch. The discharge from the relief valve goes to the suction side of the pressure pump. The oil from the rear sump entering the scavenging pump after passing through the upper strainer is drawn through E and dis- charged to the outlets F and G, which are connected by a passage in the pump body. From G } the oil goes through the passages K FIG. 263. Sliding-vane pump. and L to the oil tank. Oil from the front sump is taken through an internal pipe and the passage HIJ and is also discharged at F and G. The volumetric efficiency of the gear pump can be made very high probably up to 90 per cent; the over-all efficiency is 50 to 60 per cent. Tests of the pressure pump of the Packard 180-h.p. engine give the results shown in Fig. 262. It will be seen that the slip is very low since the discharge curves are almost the LUBRICATION 341 From Sumo FIG. 264. Plunger pump of Basse-Selve engine. Oil Tank Main Pressure Pumps A Scavenger Pumps B Delivery of Pumps A .. Suction of Pumps A - Defivery of Pumps B Suction of Pumps B Gravity Feed from Oil Tank Oil Radiator Pulsation Damper FIG. 265. Oil system of Basse-Selve engine. 342 THE AIRPLANE ENGINE same with free discharge and with discharge against pressures which vary from 48 Ib. at 1,200 r.p.m. to 58 Ib. at 2,000 r.p.m. It may be further noted that the slopes of the discharge curves up to 1,600 r.p.m. are such as to go through the zero of ordinates; that is, the pump discharge is directly proportional to its r.p.m. The eccentric sliding-vane pump (Fig. 263) is used occasionally but is being displaced by the gear type. The sliding vanes are To De/ivery Main To Oil Tank From Sump FIG. 266. Plunger pumps of Salmson engine. pressed out by springs in a slotted cylinder which is mounted eccentrically in a cylindrical casing and is power-driven. Plunger pumps are used in some of the German engines, being operated by eccentrics or by scroll cams. In the Basse-Selve engine, (Fig. 264) the oil pump is driven by a worm gear. The pump consists of two double-acting steel plungers which work vertically in aluminum barrels and make in effect four pumps. FIG. 267. Action of LeRhone oil pump. The plungers are rotated by the worm gear and are simultan- eously reciprocated by the action of the scroll cam cut in the spindle and operated by a hardened steel roller working on a pin screwed into the pump body. At each stroke of the two plungers oil is drawn from the cooler tank t6 one of the inner pump cham- bers and is discharged from the other inner pump chamber into the delivery main. At the same time oil is sucked out of one of the engine sumps into one of the outer pump chambers and is LUBRICATION 343 pumped from the other outer pump chamber into the cooler tank. A diagrammatic view of the system is shown in Fig. 265. An entirely different arrangement of plunger pumps is used in the Salmson engine (Fig. 266). In this case two plungers are used, the larger one being the scavenging pump. The plungers are pivoted on a crankpin rotated by worm gearing. The pump bodies are pivoted and oscillate through a small angle as the crankpin rotates and the plungers make their strokes. The connections to suction and discharge are made when openings in the oscillating pump bodies register with appropriate openings Side View FIG. 268. Mounting of oil tank. in the pump casing. The method of action of a single pump of this type, used on the Le Rhone engine, is shown diagrammatically in Fig. 267. The port, P, in the oscillating cylinder comes alternately opposite the intake port, /, and the delivery port, D. The method of mounting the oil tank and of connecting it to the engine in the USD-9A airplane is shown in Fig. 268. The tank is slung from the engine sills, L, and is located just below the crank case, C. The tube, A , is the oil return pipe from the scaven- ger pump to the tank; the supply pipe from the tank to the pressure pump is shown in dotted lines. The front sump con- nects by a pipe, B, to a Y-shaped air-pressure-relief pipe inside the tank, the upper ends of which extend nearly to the top of the tank. This acts as a pressure-relief vent for the tank. The oil is cooled by longitudinal air pipes, H. CHAPTER XIV THE COOLING SYSTEM All airplane engines are ultimately air-cooled. The only option is as to whether the air shall be applied directly or indi- rectly. In the latter case, the heat is removed by water which is then cooled by the air. Indirect (water) cooling offers two advantages: (1) the transmission of heat from the cylinder is more rapid to water than to air, and (2) the ultimate cooling surface (in the radiator) can be made much greater than is possible at the cylinder. Direct-air cooling has the advantage of reduced total weight and diminished vulnerability; a bullet hole through the radiator or jacket will put the whole engine out of action while a hole through one cylinder of an air-cooled engine may put that cylinder alone out of action. Air cooling in airplanes is used almost exclusively in rotary and radial engines. In vertical or Vee engines with several cylinders in line, the cooling problem is much more difficult, although it has been met by the use of suitable cowling to direct the air on to the different cylinders. With the rotary and radial types the motion of the plane and the location of the engine in the slip stream of the propeller ensure an adequate flow of air for cooling; with the multi cylinder vertical or Vee type it is some- times necessary to add a fan to improve the air circulation. The resistance or drag of air-cooled cylinders is considerable but has not been determined experimentally in a satisfactory manner. In the rotary engine there is, in addition to the drag, the resistance due to churning which reduces directly the b.h.p. of the engine. This resistance increases so rapidly with speed that it is not found desirable to operate rotary engines at speeds in excess of about 1,400 r.p.m.; the increase in indicated power which results from increased speed is largely used up in over- coming the increased air resistance. The total work done by an air-cooled engine in overcoming air resistance is probably greater under ordinary conditions of operation than the total work done by a water-cooled engine in overcoming the drag of its radiator. Until quite recently, air-cooled cylinders were at a great dis- advantage both in fuel economy and in power developed per 344 THE COOLING SYSTEM 345 unit volume of piston displacement, but recent constructions have put the air- and water-cooled engines very nearly on a par in these respects. They have always had an advantage in weight per horse power. The heat which has to be removed from the cylinder in order to keep the temperature of the cylinder within the limit which permits satisfactory operation of the engine is usually about equal to the heat equivalent of the work done in the cylinder, or 42 B.t.u. per brake horse power per minute. This quantity will increase or decrease with change in operating conditions; its limits are apparently between 30 and 60 B.t.u. per brake horse power per minute. The cooling surface of the modem air-cooled cylinder almost invariably takes the form of a series of fins. Tests on the rate of heat dissipation from such surfaces, made by the British Advisory Committee for Aeronautics, 1 show that, for wind speeds between 20 and 60 miles per hour, the heat loss for a given material is independent of the roughness of the surface. Steel shows 5 to 10 per cent greater heat dissipation than aluminum or copper. Aluminum is improved about 10 per cent by a coating of stove enamel. Throughout the usual range of cylinder diameters the heat dissipation for copper fins in a parallel air blast at the ground is given by H = [0.0247 - 0.0054(i--.V^- 4 ^ ' 78 where H is the heat dissipated in B.t.u. per square foot of fin surface per minute per degree Fahrenheit difference between the mean fin temperature and the incoming air temperature; I is the length of the fins in inches; p is the pitch in inches measured from surface to surface of adjacent fins; and V is the wind speed in miles per hour. With tapering fins p should be taken at the mean height of the fin. The heat dissipation, depending on the weight of air brought into contact with the cylinder, is proportional to (dF) * 73 , where d is the air density. Shape and Size of Fins. The fin which gives the maximum heat-loss per unit of weight is one having slightly concave surfaces and a sharp tip; a plain triangular fin is only very slightly less efficient. The best proportions for such a fin depend on the conductivity of the material and on the wind speed. For a speed of 40 miles per hour, the following table shows the best 1 A. H. GIBSON, Institution of Automobile Engineers of Great Britain, 1920. 346 THE AIRPLANE ENGINE proportions for fins of aluminum alloy (conductivity = 0.38 C.G.S. units) and steel (conductivity = 0.12 C.G.S. units) and also for rectangular copper fins (conductivity = 0.90 C.G.S. units). Bottom breadth, centimeters 0.025 0.05 0.1 0.2 0.3 0.4 0.5 Length, centimeters: Aluminum . . . 2.0 2.9 3.5 4.1 4.5 Steel 1.1 1.5 1.8 2.1 ?, 3 Copper . . . 1.6 2.3 3.3 4.8 If such a fin be truncated until the tip breadth is one-fifth of the bottom breadth, the lengths become 80 per cent of those given above. The heat dissipation is about 0.88 time and the weight 0.96 time as great as 'for the complete triangular fin. Since the heat dissipated from a fin of given shape varies directly as the length of the fin, while the weight varies as the square of the length of the fin (other things being equal) cooling fins should be as short as possible, a large number of short thin fins being used in preference to a smaller number of longer and thicker fins. While in practice this is to be borne in mind, many other factors besides that of weight have an important bearing on the best size of fin to be adopted. Thus, in a thin steel cylinder, or in a cylinder of cast iron or cast-aluminum alloy, the circumferential ribs add greatly to the strength and resistance to distortion. Comparatively deep and heavy fins have a greater effect in this direction than a larger number of similar but smaller fins giving the same cooling. Again, as the number of fins in increased, the pitch is correspondingly diminished. This diminution in pitch reduces the air flow between the fins to an extent which may, with very small pitches, render the fins prac- tically useless for cooling purposes. In a cylinder of cast iron or of aluminum alloy, foundry difficulties put a definite limit to the minimum pitch of the fins. On the barrel itself a somewhat smaller pitch may be adopted than on the head,' or the barrel fins may be turned out of the solid if desired. On account of the complicated form of the cylin- der head and ports, however, it is difficult to machine their cooling fins, and the length of many of the cores necessitates the THE COOLING SYSTEM 347 pitch being made fairly large. The minimum practical pitch of fin for such cylinders, having a diameter of from 4 in. to 6 in., is about 8 to 9 mm. or about <^LG m - Foundry difficulties also prevent the casting of a fin having a tip less than about 0.5 mm. in thickness, or a root thickness less than about Z/10, so that an aluminum fin 1 in. long would not have a root thickness less than 0.1 in. For steel cylinders with fins turned out of the solid, the pitch may with advantage be cut down to about } in. on a cylinder of 3 in. or so in diameter, but there appears to be little to be gained by reducing the pitch beyond this point. The mean fin temperature depends on the total amount of heat which has to be dissipated and its value depends on many factors. Determinations of the actual temperatures of the cylinder walls, pistons, and exhaust valves show no definite differences between well designed air-cooled and water-cooled engines; the air-cooled cylinder may be, and often is, the cooler of the two. The influence of engine speed on the wall temperature is shown in the following table giving the temperature at a point on the side of the combustion space of an aluminum air-cooled cylinder operating at maximum load. R.D.m 800 1,000 a , 200 1,400 1,600 1,800 B.h.p 10 2 12 8 15 4 18 19 7 20 6 Temperature, degrees Cen- tigrade 100 103 124 123 136 138 The compression ratio is more important than engine speed in determining the wall temperature. There is usually a definite compression ratio giving minimum wall temperature; variations on either side increase that temperature. The following table gives tests of a 100 by 140 mm. aluminum air-cooled cylinder with varying compression ratio. The brake mean effective pressure and fuel consumption of this engine are noteworthy. Compression ratio 4 6 5.0 5.4 5.8 6.2 6 4 Brake mean effective pressure, Ib. per sq. in 116.1 119.3 122.0 125.0 129.0 123.0 Fuel, pounds per brake horse power per 530 507 490 475 0.480 520 Mean barrel temperature, f Top degrees Centigrade \ Bottom 180 105 170 95 157 89 154 85 183 110 212 135 348 THE AIRPLANE ENGINE The increased temperatures in the last two columns result from preignitions which were occasional with compression of 6.2 and frequent with 6.4. Increase in cylinder diameter diminishes slightly the heat loss to the walls per brake horse power; the experimental evidence suggests a decrease of about 3.5 per cent for 10 per cent increase in cylinder diameter. As the ratio of cooling surface to b.h.p. varies inversely as the diameter in similar cylinders, the ratio of cooling area to heat given to the walls decreases as the diameter increases. The temperature difference between an air-cooled cylinder and the cooling air in a given wind may be taken as inversely proportional to D- 6 , where D is the cylinder diameter. The air-fuel ratio has considerable influence on the heat transmitted to the cylinder walls. The cylinder is hottest with the weakest mixture capable of sustaining maximum load; or, approximately, with an air-fuel ratio of 13.5. Further weakening of the mixture makes a cooler cylinder on account of the reduction in brake horse power and in the heat loss per b.h.p. At the same time it gives a hotter exhaust valve. The last point is brought out in the following table giving test results for a 100 by 140 mm. air-cooled aluminum cylinder. Air-fuel ratio 11 1 11 9 13.8 15.2 15.7 Brake m.e.p., Ib. per sq. in. ... Fuel, pounds per brake horse power per hour 122 622 122 589 119 515 116 480 114 0.470 Exhaust valve temperature, de- grees Centigrade 706 717 747 752 747 An increase in mixture strength beyond that necessary for maximum power reduces the temperature of the cylinder sur- faces, as shown below for an engine operating at full throttle and constant speed. Air-fuel ratio . 10 5 11 4 12 9 13.5 15.4 Cylinder head temperature, degrees Centigrade . . 200 215 237 229 215 Tests show that the maximum temperature of the head of an air-cooled cylinder must not exceed 270C. for satisfactory work- ing. If the temperature exceeds 280 there is usually trouble THE COOLING SYSTEM 349 from preignition. Higher working temperatures are permissible with larger cylinders. If the temperature is kept at 200 to 220C., the economy and capacity obtained are quite as good as for water-cooled cylinders of similar design and size. The temperature of the exhaust valve at its hottest point should not exceed 720C.; with valves not exceeding 1.5 in. in diameter it is possible to reduce this temperature to 650C. In a well designed aluminum cylinder of the overhead-valve type, operating in a 60-mile-per-hour wind, a provision of 0.28 to 0.35 sq. ft. of cooling surface per brake horse power is sufficient to give satisfactory operation, the larger area applying to cylinders of about 4-in. bore, and the smaller to cylinders of about 6-in. bore. For steel or cast-iron cylinders with overhead valves this area must be increased about 50 per cent and for L-head cast-iron cylinders by 100 per cent. At reduced wind speeds the cylinder temperature increases; the mean temperature difference between the fins and the air varies inversely as T 70 "*. Thus in a given series of tests a reduc- tion of wind speed from 80 to 40 miles per hour increased the cylinder temperature from 229 to 296C. There are practical difficulties in the way of providing sufficient cooling surface for operation at full throttle below certain limiting wind speeds. The wind speeds can be less with smaller cylinders. The mini- mum air velocity for good performance of air-cooled cylinders of good design and material under full throttle is given below. At lower air speeds partial throttle only should be used. Diameter, inches 23468 Minimum air velocity, miles per hour ... 30 40 50 70 90 Cylinder Materials. With cylinders of normal design the middle portion of the head is the hottest point. Free air flow to this point is impeded by the valve ports and gears so that it is almost impossible to provide adequate cooling surface there. The heat has to travel outward and is dissipated mainly from the cooling surface surrounding the combustion head. It is therefore important to use a material of maximum thermal conductivity. The three practical materials for cylinder con- struction, steel, cast-iron and aluminum alloy, have conductivities (in C.G.S. units) of 0.12, 0.10 and 0.38 respectively. Aluminum is consequently the most desirable material. The alloys most suitable for cylinders are copper-aluminum alloys with about 350 THE AIRPLANE ENGINE 90 per cent of aluminum. The high-zinc alloys are unsuitable because their tensile strength is low at 200C. All the alloys show rapid decrease of strength as the temperature increases beyond 250C. The following table gives data on this point. Composition per cent Tensile strength, Ib. per sq. in. Cu Sn Mg Al Ni Mn At 250C. At 350C. 7.0 1.0 92.0 12,300 6,700 12.0 88.0 , . . 23,500 13,400 14.0 85.0 . . 1 21,300 14,500 4.0 1.5 92.5 2 . . 24,600 11 ; 200 9.0 89.0 2 19,000 10,100 Water Cooling. By water-cooling the cylinder and the exhaust ports it is possible to run with higher speeds and com- pression ratios than are practicable with air-cooled cylinders. The possible increase in speed and ratio of compression are rel- atively unimportant when compared with the performance of the best recent constructions in air-cooled cylinders; they are considerable as compared with the average air-cooled cylinder. With water cooling it is possible to maintain almost any desired cylinder temperature. If the temperature is low the volumetric efficiency and the capacity of the engine will be improved (see p. 37) but the engine friction increases and its efficiency falls off. The temperature of the jacket water after leaving the radiator must be below the boiling point of water at the pressure existing on the suction side of the pump, otherwise the pump will not function well but will suck in water vapor. As fuel economy is ordinarily more important than capacity, the jacket water is usually kept at as high a temperature as the boiling point will permit. The mean jacket temperature at the ground is usually 160 to 180F. Water is not the ideal cooling agent. A less volatile fluid would permit higher cylinder temperature; higher efficiencies might be obtained without running into such temperatures as would cause preignition. The same result might be obtained by operating a closed water-cooling system under pressures greater than atmospheric, but this would necessitate heavier material for the THE COOLING SYSTEM 351 radiator core and consequent increase in weight and decrease in airplane efficiency. The heat which is removed by the jacket water is practically equal to the b.h.p. or is 42.4 B.t.u. per brake horse power per minute. In a closely cowled engine this same amount of heat would have to be removed from the radiator. With the usual cowling there is considerable removal of heat by the air stream from the engine and water-j acket surfaces, so that only about 31 B.t.u. per brake horse power per minute has to be removed from the radiator: with an uncowled engine this quality falls to 23 or 25 B.t.u. The principal parts of a water-cooling system are the jackets, the pump and the radiator. The last of these will be considered first. Radiators. Airplane radiators have developed from auto- mobile practice but certain types of automobile radiators are entirely unsuited to air- plane practice. The suc- cessful commercial types have cores made of thin brass, or copper ribbons or tubes from 0.004 to 0.006 in. thick. Common types are shown in Fig. 269, which illustrates: a and 6, rectangular air pas- sages; Cj rhombic passages; and d, circular passages with hexagonal ends. Other com- mon types have hexagonal or elliptical air passages The water passages are nar- row, varying from 0.03 to 0.08 in. The air tubes are commonly not more than Y in. in maximum cross-section dimension and are from 3 to 5 in. long (depth of core). The metal sheets are stamped or rolled to the desired form with the front and rear ends of the pair of sheets forming each water passage in contact with one another. These ends are soldered by dipping them into a shallow pool of molten solder. Great care must be exercised to keep down the weight of solder as much as possible; it often amounts to 25 per cent of the total weight of the radiator core. The top and FIG. 269. Types of radiator core. 352 THE AIRPLANE ENGINE bottom ends of the water passages are inserted through slots in the top and bottom headers respectively; the two sheets of each water passage are spread apart and soldered to the header. In the case of type d, Fig. 269, the ends of the circular tubes are expanded into hexagonal forms which are soldered together; the expansion is made enough to give the desired width of water passage between the tubes. Type b differs from type a not only in the method of assembly but may also be made of a cor- rugated surface which is intended to give greater strength and larger radiating surface. The types a, 6, c and d in Fig. 269 have water in contact with all the radiating surface and are said to have only "direct" radiating surface. Many automobile radiators have extensions of this direct radiating surface in the form of fins on flat or circular tubes (e, Fig. 269), metal spirals, and so forth. Such " indirect" radiating surface is found to have too high a ratio of head resistance to heat-removing capacity tO;be satisfactory for airplane use. The dimensions or external shape of a radiator can be adapted to suit its location and desired performance. The location may be such that air may pass through or around it without obstruction, in which case it is said to be in an " unobstructed " position. On the other hand, the radiator may be located in the nose of the fuselage, or in the plane of the wing, in which case the air flow is materially affected by other parts of the plane and the radiator is said to be " obstructed." The performance of such a radiator will depend not only on the size and type of the core but on its position or surroundings. Examples of typical unobstructed locations are shown in Fig. 270 (at the sides of fuselage) and in Fig. 271 (over the engine); common obstructed positions are in the nose of the fuselage, and in the wing. A comprehensive study of the properties of various types and dimensions of radiator cores has been made at the Bureau of Standards and published in the Fifth Annual Report of the National Advisory Committee on Aeronautics. The following discussion is mainly from that source. Two quantities are of importance in determining the heat transfer of a core. They are the temperature difference between the entering air and the mean water temperature; and the mass flow of air. The temperature difference should ordinarily be taken as the difference between the mean summer air tempera- ture and the mean water temperature. The mass flow of air, THE COOLING SYSTEM 353 Outlet Pipe. FIG. 270. Side radiators. FIG. 271. Overhead radiators. 23 354 THE AIRPLANE ENGINE M, is the weight of air flowing per second per square foot of frontal area of the core. Its amount (at constant air density) is found to be proportional to the free air speed or the velocity with which the core moves through the air when the core is unobstructed; the mass flow is always less for obstructed positions than for unobstructed. The energy dissipated or heat transfer is expressed in horse power per square foot of frontal area, and, for purposes of com- parison of the properties of various cores, a temperature difference of 100F. is assumed between the air entering the radiator and the mean water temperature; the heat transfer is proportional to this temperature difference. One horse power is equivalent to 42.54 B.t.u. per minute. The head resistance of the core is the force required to push it through the air and is expressed in pounds per square foot of frontal area. This head resistance is found to vary approxi- mately as the square of the free air speed; in most cases the exponent is slightly less than 2. If R is the head resistance, and V the free air speed in miles per hour, then R = cV 2 and c is called the head resistance constant. The horse power absorbed by a radiator is the engine power required to overcome the head resistance and support the weight of the radiator. The work done in supporting the weight can be calculated if the lift-drag ratio of the plane as a whole is known. An average value of 5.4 may be assumed for this ratio. If W is the weight of the core and contairied water in pounds per square foot of frontal area, the propeller thrust required to sup- port the weight is TF/5.4. The horse power absorbed is V X 5 ' 280 ' 60X~3pOO It should be noted that this method of calculation neglects the effect on the lift-drag ratio of the addition of the radiator. The lift-drag ratio varies between different planes and varies even more widely between climbing and level flight. The selec- tion of a core for a given plane cannot be made satisfactorily without a knowledge of the relative importance of climbing speed and top speed. A lift-drag value of 5.4 is a good average THE COOLING SYSTEM 355 and gives about equal value to climbing and level speed. If the rate of climb is of prime importance the value may be as low as 3, while if speed on the level is the most important, a value as high as 10 may be used. A small additional power charge against the radiator is that required to overcome the resistance to water circulation in the radiator. It is usually so small as to be negligible. The definition just given of horse power absorbed applies only to the case of an unobstructed radiator. If the addition of a radiator necessitates alterations in structure (such as the sub- stitution of a flat nose for a stream-line fuselage or the enlarge- ment of the fuselage to accommodate the radiator required) the consequent increase in resistance of the structure should be charged to the radiator. A comparison of the performance of various cores can be obtained when the heat transfer per unit of power absorbed is known. This quantity is called the Figure of Merit and is a pure number. The comparison must be for the same tempera- ture difference and free air speed. It applies only to unob- structed radiators. The general conclusions derived from the tests at the Bureau of Standards are as follows: Heat transfer is a function of mass flow of air and is inde- pendent of the air density. Heat transfer is roughly proportional to mass flow for a core having only direct cooling surface. When there is a considerable amount of indirect cooling surface the heat transfer increases less rapidly than mass flow at high air speeds. Heat transfer is proportional to the temperature difference. '[Heat transfer is not greatly affected by the rate of water flow pro- vided the rate is above 2 gal. per minute per inch of core depth per foot width of core. It should be noted, however, that this is true only when the mean water temperature is regarded as constant. Heat transfer from direct cooling surface is not appreciably affected by the composition of the metal. When fins and other indirect cooling surface are used the thermal conductivity of the metal is important. Heat transfer is somewhat increased, but at the expense of a large increase in head resistance, by spirals or other forms of passages which increase the turbulence of the air stream. Heat transfer is greater for smooth than for rough tube walls, for, if 356 THE AIRPLANE ENGINE the surface is rough, it will be covered with a layer of more or less stagnant fluid. Head resistance for any particular core varies approximately as the square of the free air speed. The head resistance of a core appears to be closely related to its mass flow so that, in general, anything which tends to cut down the flow of air through the core will cause a considerable increase in head resistance. Head resistance varies directly as the air density for a given free air speed, and inversely as the density for a given mass flow. Head resistance is considerably increased by projections, indentations, or holes in the air tube walls. Head resistance per square foot is not appreciably affected by the size of the core within the limits used, viz., 8 by 8 in. to 16 by 16 in. and 12 by 24 in. Special conclusions with reference to types of cores are as follows : For a high figure of merit the core should have smooth, straight air passages, easy entrances and exits for the air and a large per- centage of free area. Under these conditions the figure of merit increases as the depth increases up to at least 20 times the di- ameter of the air tubes, which is as far as experiment has gone. Even greater depths may be of advantage. By far the most satisfactory radiator for use in unobstructed positions seems to be one of thin flat plates with water space not over KG in. wide and spaced J^ in. on centers. The plate should be at least 12 in. deep. As the figure of merit changes but slightly with increase of depth beyond 12 in. the depth may be made 20 in. or more if it is desirable to reduce frontal area. The chief defect of the type is mechanical weakness. Of the commer- cial radiators tested, those have given highest figure of merit at high air speeds which have only direct cooling surface in the form of tubes about J^ m - square and about 5 in. deep. The figure of merit of this type at 120 miles per hour free air speed varies from about 8 to 8.4, whereas flat plates 9% in. deep and J^ in. on centers have a value of 10.7. The energy dissipated per square foot of frontal area is less in the above flat plate radiator than in the best square tube radiators so that a larger frontal area will be required with flat plates but the power absorbed will be less. The British Air Ministry has adopted as standard a circular THE COOLING SYSTEM 357 tube 10 mm. in diameter expanded at the ends to a hexagonal section 11 mm. across the flats (Fig. 269 d). The standard length of the tubes is 120 mm. The material is 70 30 brass with wall thickness of 0.005 in. The actual power absorbed by the radiator in being lifted and pushed through the air (see Table 18) varies from about 3 h.p. to 6 h.p. per square foot of frontal area at 100 miles per hour. This amounts to from 5 to 20 per cent of the total engine power. A small gain in radiator performance may have an appreciable effect at high speed. The selection of a radiator core for an obstructed position is more difficult. An obstructed position involves a large absorp- tion of power. The resistance of a fuselage fitted with a nose radiator is two or three times the resistance of the same fuselage with a stream-line nose. The increase in resistance due to the substitution of a radiator for a stream-line nose is greater than the increase that would be caused by using a radiator of the same core construction and the same cooling capacity in an unobstructed position. At any given free-air speed the total resistance of a fuselage with a flat nose radiator is increased by increasing the air flow through the radiator, either by opening exit vents for the air or by decreasing the resistance of the radiator to the passage of air. This indicates that a nose radiator should be of compact con- struction with high heat transfer, for low air flows through the core, requiring a core of high resistance. This fact is of special importance since the space available for a nose radiator is so limited that the highest possible mass flows are used in practice. A nose radiator with air exit vents equal in area to the free air passage through the radiator is found to cut down the heat transfer about 35 per cent as compared with the same radiator in an unobstructed position. Indirect cooling surface may be of advantage if it is made of copper, crimped from the water tube walls and well soldered to them at all possible places. Several types of core show good heat transfer at low speeds, but here again the square tube, with direct radiating surface only, gives best result of all commercial types, and flat plates spaced J4 m - on centers show excellent performance. The fin and tube type with its small amount of direct surface has no use in airplanes except possibly in a wing position where a high head resistance is no disadvantage. 358 THE AIRPLANE ENGINE The properties of cores selected as typical of common con- struction are given in Tables 17 and 18. The constructions vary from' the flat tubes (E-Q and E-8) with 100 per cent direct cooling surface and a minimum of head resistance to finned circular tubes (F-5) with only 12.3 per cent of direct surface and three times as much head resistance per square foot of frontal area as the best flat tubes. The dimensions of the core are given in Table 17; the performance at various wind speeds in unobstructed posi- tions in Table 18. The " figure of merit " necessarily diminishes \ 0.020 0.040 0.060 0.050 0.100 Mass Flow Factor (k) FIG. 272. Head resistance constant and mass flow factor. with increase of wind speed; the order of merit of the different cores is different at different speeds. At 120 miles per hour the flat tubes E-8 and E-Q are seen to be best and the finned circular tubes F-5 the poorest. Table 19 gives the constants in the empirical equations for "Head Resistance," "Mass Flow" and "Energy Dissipated" for the cores listed in Table 17; these quantities are obtained from the data of Table 18. The Head Resistance Constant and the Mass Flow Constant appear to be connected by a simple relation; plotting these quantities for all the cores tested gives the curve of Fig. 272. THE COOLING SYSTEM 359 i .-i O5 00 >0 CO cS2 1C CO c 1 00 CO o w 1 o o CO 05 CO tet (N t^ 00 O t- CM O 05 58 2 rt VJK, n > $5 g OH 35 o o C IN t^ on t^ 00 Tt* t^ O O ?> $00 CO CO o o o o 00 c 8 o M OS O5 CO T-t CO CO CN 1-1 co (N (N t^ CO 10 CO Iv. 00 O CO CM CO * $8 s d d (N d Ki <* P v* rf> p ; UJ 8 O5 IN i-H CO 1-H CO 5 or 05 s ll 1 (N ; o. CO Tj< 00 CO r-H t* CO O 1C CO CO ^ O O CO t^ 8 10 1 ! i c CO 1C 00 1 CM : o t-j t^ Tt< Tj to 43 1 1 1 u fl : : 3 . . O" > 'A Number Depth, inches 1! ..|a H^fc I Thickness of metal, Weight, pounds per Empty Water content. ^i s 1 1 4 P. Cooling surface, sq Per cent direc Water tubes: Length, inches WiHtVi inphps 1 2 X c c X J ^_ c c c C i i i 1 Sz Area normal to Hydraulic radii Air tubes: Hydraulic radi M "M a 3 .2 1 360 THE AIRPLANE ENGINE TABLE 18. RADIATOR PERFORMANCE IN TERMS OF FREE AIR SPEED (FOR UNOBSTRUCTED POSITIONS ONLY) Grade A represents very good performance; grade E, very poor Radiator Speed, miles per hour Air flow, Ib. per sq. ft. per sec. Energy in h.p. dissipated per square foot of Head resistance, Ib. per sq. ft. frontal area H.p. absorbed per sq. ft. of frontal area Figure of meri Front Surface A-7....... J" Grade 30 60 90 120 2.20 4.40 6.60 8.80 27.2 45.9 61.7 76.5 A 20.3 37.2 52.9 68.0 B 20.2 31.0 39.8 47.9 D 14.8 26.4 37.8 48.5 D 13.8 24.7 32.3 38.7 D 29.3 51.1 71.3 90.5 A 17.2 29.8 41.0 51.7 C 13.8 21.7 27.8 33.0 E 23.1 39.5 53.6 67.0 B 0.43 0.73 0.99 1.22 0.37 0.68 0.97 1.25 0.36 0.55 0.71 0.85 0.46 0.83 1.18 1.52 0.58 1.04 1.32 1.63 0.37 0.65 0.91 1.15 0.44 0.76 1.05 1.32 0.40 0.63 0.81 0.96 0.52 0.88 1.20 1.50 1.72 6.70 14.90 26.3 D 1.40 5.61 12.10 21.3 C 1.72 6.88 15.48 27.5 D 1.37 5.47 12.30 21.9 C 1.25 4.75 10.48 18.4 B 1.57 6.27 14.10 25.1 D 0.78 3.12 7.03 12.5 A 2.52 9.65 21.6 38.3 E 2.40 8.65 19.1 33.4 E .50 1.79 4.66 9.87 D 0.44 1.54 3.88 8.11 C 0.41 1.64 4.52 9.89 D 0.30 1.26 3.53 7.79 C 0.26 1.08 2.99 6.53 B 0.51 1.78 4.54 9.59 D 0.27 0.92 2.31 4.84 A 0.33 1.81 5.57 12.8 E 0.40 1.80 5.21 11.5 E 54.6 25.7 13.2 7.8 . B 46.6 24.1 13.6 8.4 B 49.8 18.9 8.8 4.8 D 48.7 20.9 10.7 6.2 C 53.5 23.0 10.8 5.9 C 57.2 28.8 15.7 9.5 B 63.3 32.4 17.8 10.7 A 41.3 12.0 5.0 2.6 E 58.0 22.0 10.3 5.8 C A-23 30 60 90 120 2.29 4.58 6.87 9.16 9$ Grade B-8 30 60 90 120 2.12 4.24 6.36 8.48 Grade C-4 30 60 90 120 2.40 4.80 7.20 9.60 Grade D-l 30 60 90 120 30 60 90 120 2.41 4.82 7.22 9.63 Grade E-6 2.12 4.24 6.36 8.48 -Hh&i~ Grade E-8 30 60 90 120 2.74 5.48 8.23 10.97 1 " l^L Grade F-5 30 60 90 120 1.82 3.64 5.46 7.28 - -kl'd? t|mX HI" ' %'MJ: \ O i i i 1 IK** . H ~~*^ 100 Grade G-3 30 60 90 120 1.88 3.75 5.62 7.50 Grade THE COOLING SYSTEM 361 A plotting of the data in Table 18 for core E-S is given in Fig. 273. Size of Radiator. If a type of core has been selected and its properties are known and plotted as in Fig. 273, the necessary size of an unobstructed core is directly obtainable. The dimen- sions must be calculated for the most unfavorable condition, which, ordinarily, will be maximum climbing speed near the ground with summer temperatures. The mean water temper- ature will be fixed by the maximum temperature allowable in the jackets and by the quantity of water circulated. The Free Air Speed, Mi.perHr. 50 IQO 39.2 Sq.H: Surf ace !00% Direct Surface 975% free Anta 32 Ib.Fmpfy 0246 8 10 Mass, Flow of Air, Lb. perSq.FtperSec. FIG. 273. Properties of a flat tube radiator core. maximum allowable water temperature is ordinarily 20 to 30F. below the boiling point and varies with the altitude of the plane; its value will be determined by (1) its influence on the volumetric efficiency of the engine (see p. 37) and (2) the water resistance of the -radiator (see p. 364). If the water system is closed, with no vent to the atmosphere, the last-mentioned factor disappears. The heat to be dissipated should be determined if possible but may be assumed equal to the brake work of the engine if more exact knowledge is not obtainable. The effect of propeller slip should be estimated and allowed for. Allowance should also be made for the cooling effect of the radiator headers and of the exposure of the engine to the wind. 362 THE AIRPLANE ENGINE Occasionally, instead of designing for maximum climb some other condition may impose maximum service on the radiator as, for example, in flying boats and seaplanes intended for training, where much taxi-ing is done at low plane speed and maximum engine power. If the radiator is in the nose of the fuselage some assumption must be made as to the relation of mass flow to the speed of the plane. The mass flow will usually vary from 0.04 to 0.07 time the speed of the plane (in miles per hour), depending on the type of radiator, the amount of cooling and the masking effect of the propeller. The power absorbed is seldom calculable because of the uncertain effect of the radiator on the resistance of the fuselage. TABLE 19. CONSTANTS IN THE EQUATIONS R = cV 2 ;M =/cF/ANDQ = Gm n R = Head resistance in pounds per square foot. V = Free-air speed in miles per hour. M = Mass flow of air in pounds per second per square foot. Q = Energy dissipated in horsepower per square foot per 100F. tem- perature difference. m = "mass flow constant," which is the ratio of the mass of air passing through 1 square foot of radiator to the mass of air passing through 1 square foot of free area in front of the radiator. Radiator c X 10 3 k X 10 2 m G n A-7 1 86 7 34 667 15 1 75 A -23 1.56 7 63 694 10 3 0.85 B-8 1 91 7 07 643 13 1 60 C-4 1 52 8.00 727 7 1 0.85 0-1, 1 32 8 03 730 8 1 70 E-Q 1.74 7.07 643 16.1 0.80 E-8. 867 9 13 830 7 6 80 F-5 2.68 6.07 0.552 10.0 0.60 G-3. 2 40 6 25 568 14 8 75 The mass flow for a wing radiator depends on the angle of incidence but is probably not over 0.01 time the plane speed even at the best climbing angle. The relative efficiencies of radiators in various positions are given by Liptrot 1 as follows : 1 Aeronautics, Apr. 29, 1920. THE COOLING SYSTEM 363 Relative Position of radiator ~, . efficiency Unobstructed Underslung, side or overhead, but close to fuselage Twin nose radiator Nose radiator with core entirely above or below propeller shaft . . Nose radiator with propeller shaft in center Behind engine 1.000 0.973 0.716 0.656 0.585 0.423 The use of a small projecting lip or stream line entrance around the core may reduce the necessary core size slightly but at the cost of a considerable increase of head resistance. Rate of Water Flow. One gallon (231 cu. in.) of water at 200F. weighs -^-= 8 Ib. approximately. With a temperature difference of 10F., 1 gal. of water per minute SO will give up 80 B.t.u. or , A . = 2 h.p. approximately. With a 4Z.4O temperature difference of 5F. the flow of water in gallons per minute should equal the engine horse power. The entering temperature of the water is fixed by the necessity of keeping at a certain point below boiling. With fixed entering temperature, if the amount of water circulated is increased the mean temperature of the water is raised and consequently the temperature difference between air and water is increased. The influence of water velocity on the heat transfer is found by experiment to be very small so long as the velocity is above 2 gal. per minute per foot width per inch depth of core, which is much below usual rates. With a circulation of J^ gal. per minute per horse power the temperature fall of the water is 20F. ; increasing this to % gal. reduces the temperature fall of the water to 10F. and increases the temperature difference between air and water by 5F. With an infinite amount of water circulated this tem- perature difference could be increased only another 5F. The increase in pump work with increased water flow makes it undesirable to circulate more than about % gal. per minute per horse power, and with radiators that are relatively long and narrow a flow of ^ gal. per minute per horse power should be used. 364 THE AIRPLANE ENGINE The pressures required to maintain water flow through the cores of radiators vary greatly with the dimensions and type of construction. Those types having the widest and straightest water spaces offer least resistance whereas those with many right angle bends will offer much resistance. The range for 12 com- mercial radiators tested at the Bureau of Standards, all of them 8 in. square in frontal section and of depths varying from 2% to 4 in. with a total water flow of 20 gal. per minute, was from 0.27 to 10.2 ft. of water pressure drop. These pressure drops may be assumed to vary directly as the height of the core, but the rate of change with change of water velocity follows an exponential law in ah 1 cases, though with a widely varying exponent in the different types. The resistance seems to depend largely on the care used in manufacture and on the form of the water tube entrances and exits. It would seem well to include a test for pressure necessary to produce water flow in acceptance specifi- cations for complete radiators. The water enters the top header of the radiator, at which place atmospheric pressure is usually maintained through the overflow pipe. The suction pressure at the pump cannot be less than the vapor pressure of the Tank on Top Plane I ' Carburetors ' ,., ., Radiators on each side of Machine FIG. 274. Cooling system of Benz engine. water leaving the rad- iator if the pump and radiator are at the same level. If the water leaves the radiator at 190F. the corresponding vapor pressure is 9.2 Ib. or about 5 Ib. below at- mospheric pressure. The maximum pressure avail- able for overcoming the resistance of the radiator in this case will be 5 Ib. per square inch or 11.5 ft. of water. With a reserve tank in the upper plane, as in Fig. 274, the head available in overcoming radiator friction is increased by the height of the tank above the suction. If the resistance of a proposed radiator is in excess of the available pressure, its height must be decreased and its width correspond- ingly increased in order to give the necessary radiating surface. Occasionally radiation or expansion tanks instead of being vented to the atmosphere are provided with safety valves opening THE COOLING SYSTEM 365 at 2 or 3 Ib. per square inch. This diminishes the loss of water from evaporation and may permit a higher water temperature. Effect of Altitude on Radiator Performance. The investiga- tions at the Bureau of Standards have yielded the following general conclusions: The effect of the lower air temperature is to increase the heat transfer in proportion to the increase in the mean temperature difference between the entering air and the water. The decrease in air density reduces the mass flow of air and decreases the heat transfer at any given plane speed in proportion to the air density. Head resistance is proportional to air density and is therefore reduced with increased altitude. The combined effect of temper- ature and density changes is to decrease the heat transfer but not as rapidly as the engine power diminishes; consequently the cooling capacity of the radiator becomes excessive at high altitudes and may be more than double the required capacity. As the head resistance falls off more rapidly than the heat transfer the " figure of merit " of the radiator increases with altitude. From the above con- clusions the performance of a radiator at any alti- tude can be calculated when its ground perform- ance is known. For ex- ample, take the flat plate core (E-8) for which ground data are given in Tables 17 and 18. It is desired to calculate its performance in summer at 10,000 ft. altitude and 120 miles per hour. The ground data are: Mass flow of air at 120 miles per hour = 10.97 Ib. per square foot per second. Head resistance at 120 miles per hour = 12.5 Ib. per square foot. Weight of core and contained water = 14.15 Ib. per square foot. The mean temperature of the water in the radiator may be V ^ ov \ \ ^ ^ e- 4U ft. V ^^ &, ^ r^ \ _ Radiatormaskingat altitudes, flow) as altitude increases. The curve of Fig. 277 shows how much masking is possible for the flat plate radiator E-S at 120 miles per hour, but the curve is practically the same for other cores and speeds. THE COOLING SYSTEM 367 It should be remembered that climbing should be considered as well as level flight in any discussion of radiators and of mask- ing. The speed for maximum climb may be only one-half that of level flight at certain altitudes, and the cooling must be adequate for the climbing condition. This consideration alone would require a masking of 50 per cent for such planes in level flight. If the relation between maximum climbing speed and level speed is known, and also the change in engine revolutions and power, the mass flow of air can be determined under both conditions and the desirable degree of masking can be found. Valve Pump .Pump To connect systems in series, turn A" through 30 Radiator Engine Radiator FIG. 278. Radiator interconnections for dual engines on lighter-than-air machines. The twin-engined dirigible offers a special case of importance. Such a craft may operate for long periods with one engine only, which therefore operates at low speed but full power. If the radiator is designed for maximum speed each radiator will be too small for its engine at this reduced speed. To obviate the use of a larger radiator the installation may be arranged as in Fig. 278 in case the water pumps are of such construction as to permit the water to pass through when they are idle. Turning the valve A through 90 deg. will circulate the water through both radiators and through the jackets of both engines, and will thereby prevent the idle engine from freezing up and will give more than adequate radiating surface. Some provision for masking the radiator is especially desirable in this case. 368 THE AIRPLANE ENGINE Masking can be partially accomplished by varying the water flow, as by by-passing some of the water from radiator inlet to outlet. The effect of reducing the quantity of water is to reduce the mean temperature of the water and thereby to reduce the mean temperature difference between air and water. The possible range of control by this means is small. Shutters across the radiator front answer the purpose more fully, although they add to the head resistance. They may be operated by the pilot, or as in some German planes, may be under the automatic control of an electrical resistance thermometer. Closed shutters on a nose radiator decrease the head resistance: on a free air radiator, they increase it. A retractable side or bottom radia- tor, which may be drawn within the body to decrease the cooling effect, is occasionally used. It may be arranged most conven- iently as an auxiliary radiator in series with a fixed main radiator which has no masking device and is adequate for high-altitude evel flight. The auxiliary radiator is retracted as altitude is gained. The increased water resistance from two radiators in series is objectionable. Yawing is another possibility. Effects of Yawing Airplane Radiators. The air stream does not always approach the radiator at right angles to its face. The most common causes of this are : 1. Radiator mounted in the propeller slip stream where the air strikes the radiator at angles other than normal to its face. 2. Radiator mounted in the wing (or other position) where the axes of its passages for the air are not parallel to the direction of motion of the plane. 3. Radiator pivoted about an axis perpendicular to the direction of motion of the airplane for the purpose of changing its inclination for the regulation of cooling capacity (masking). The effects of yawing a radiator through angles from to 45 deg. are (1) to decrease slightly the mass flow; (2) increase the head resistance by as much as 50 per cent in the case of cores of low head resistance but much, less in the case of high-resistance cores; and (3) in some cases, for angles up to 20 or 25 deg. to increase slightly the heat transfer. These effects vary largely with different types. The complete radiator consists not only of the core but of top and bottom headers. The top header may serve merely as a distributor or it may have sufficient capacity to serve as reserve and expansion tank also. The latter practice reduces complica- tions and is therefore used on small machines intended for short THE COOLING SYSTEM 369 flights. For large machines used for long flights, an adequate water capacity would entail a large frontal surface and excessive head resistance of the header. The desirable reserve capacity in British practice is given by the formula: Gallons h.p. X (endurance in hours) 1,600 f-fbfn Baffle Plate Instrument Flange ^ x*3" 9aOkfMt Supporting Brackets ^"Shutter Bracket FIG. 279. Details of typical nose radiator. The reserve water tank is often located in the upper wing but there is danger of freezing unless, as in Fig. 274, the water circula- tion is through the tank; the objection to including it in the circulation is the increased length of pipe through which the water has to be forced. The lower header is a collector only and should be as small as practicable. Both headers should be stream-lined. The headers and their contents will usually add 50 per cent to the weight of the core and its contents. Occasionally (as in the Maybach 24 370 THE AIRPLANE ENGINE plant) the headers are divided into halves by vertical baffles on the fore and aft line. Water enters the left-hand side of the lower header, passes to the left-hand side of the upper header, then over the baffle to the right-hand side and down to its exit at the right-hand side of the lower baffle; this arrangement causes greatly increased water resistance if the same weight of water is circulated; if the weight of water is halved so as to maintain the same velocity in the radiator passage, the mean temperature difference between air and water will be diminished, necessitating the use of a large radiator. A complete nose radiator is shown in Fig. 279. Among the details to be noted are the filler, inlet and outlet pipes; the perforated baffle plate between the inlet and the upper tank; the overflow pipe; the upper and lower supporting brackets; and the shutter brackets. The filler cap is of hard rubber with a safety chain ; a better construction, avoiding loss from the snap- ping of the chain, is with a hinged cap held closed by a snap wire. Pumps. As previously pointed out (p. 363) the cooling water required is not more than J gal. per brake horse power per minute. Ordinarily it is J4 gal. per minute or less. The re- sistance to the circulation of the water is chiefly in the radiator, but is considerable in other parts of the system; its magnitude is variable, but may be assumed to be from 4 to 8 Ib. per square inch in good installations. The water horse power of the pump of a 100-h.p. engine using ^ gal. (2 Ib.) of water per horse power per minute against 8 Ib. 2 X 100 1 per square inch pressure is 8 X 144 X X QQ QQQ = 0.116 h.p. If the efficiency of the pump and its drive is 20 per cent, the horse power used to drive the pump will be 0.166 -r- 0.2 = 0.58 h.p., which is a very small fraction of 100 h.p. Consequently, the water pump efficiency is comparatively unimportant and the type selected should be one of maximum simplicity and minimum weight. The single impeller volute centrifugal pump meets these conditions best and is univer- sally used. In a volute pump, water enters axially, is caught by the im- peller blades and is given a high velocity of rotation before it is discharged into the volute casing, from which it escapes through one or more outlets. The number of outlets is usually the same as the number of banks of cylinders. In order to keep down the THE COOLING SYSTEM 371 size and weight of the pump the impeller rotates at a speed greater than that of the engine; one and one-half engine speed is common. If the impeller blades are radial (Fig. 280) the theoretical dis- charge pressure in feet of water is given by V 2 /2g where V is the tip speed of the impeller. Taking an impeller diameter of 4 in. / 4 2,400\ 2 and a speed of 2,400 r.p.m. this becomes ( ir X TO X QQ J -f- 2g = 32 ft. = 13.8 Ib. per square inch. The water velocity leaving the impeller cannot be converted completely into pressure head and there are various impeller and casing losses, so that the Outlet Inlet FIG. 280. Water pump of Liberty engine. actual discharge pressure will be much less than that calculated above; it would probably be less than one-half the theoretical value. The resistance to be overcome by the pump is entirely fric- tional, and varies as the square of the amount of water circulated. The amount of water circulated is proportional to the speed of the pump. The work done by the pump is proportional to the volume of water circulated multiplied by the resistance, or is proportional to the cube of the pump speed. The pump of the Liberty engine, Fig. 280, has a 2-in. inlet and two outlets. It runs at 1^ times engine speed, and has a capacity of 86 gal. per minute at 2,000 r.p.m. of the engine. The impellers are radial and are partly shrouded. The packing of the 372 THE AIRPLANE ENGINE impeller shaft against waiter leakage is kept compressed by a coiled spring. The King-Bugatti (410 h.p.) pump (Fig. 281) has impellers completely shrouded on one side. As there is only one outlet the casing is of complete volute form. The impeller is 5% in. diameter with eight vanes, the web being drilled to equalize the water pressure. The shaft is packed with graphited asbestos rope packing, held under compression by a coiled spring. The inlet is 2}^ in. in diameter; the outlet is 2% 6 m - It is coupled direct to the engine shaft. FIG. 281. Water pump of Bugatti engine. The Austro-Daimler (200 h.p.) pump weighs 7.6 lb., has an impeller 4.4. in. in diameter, inlet and outlet diameters 1.42 in., and a ratio of pump to engine speed of 1.89. It is driven from the rear end of the crankshaft by a bevel ear which is integral with a sleeve forming an extension shaft (EX 282). The pump bevel gear floats on the end of the pump spindle, and is fitted with a large-diameter thrust ball-race and retaining spring, which, being at the bottom end of the spindle, are as far away as possible from the impeller. Both the pump spindle bearings are lubricated through two holes drilled in the pump body and oil grooves cut in the spindle bearings. The impeller is formed with six vanes THE COOLING SYSTEM 373 and is completely shrouded ; it is keyed to the spindle and secured by a gun-metal nut and washer. A conically-faced shoulder is fn/ef. FIG. 282. Water pump of Austro-Daimler engine. machined on the pump directly beneath the impeller. This shoulder beds into the bevelled face of the bronze bearing, form- 50 J- ? 20 1 ,0 1000 1200 1400 1800 2000 1600 R.D.m FIG. 283. Performance curves of Austro-Daimler water pump. > ing a water-tight joint. The performance curves of this pump are given in Fig. 283. 374 THE AIRPLANE ENGINE The Maybach (300 h.p.) pump (Fig. 284) has an impeller 4.46 in. in diameter, inlet 2.13 in. in diameter, outlet 1.97 in. in diameter, and a ratio of pump to engine speed of 2. The pump spindle is driven through a dog clutch at its lower end by a short vertical spindle running in a bronze bush- ing; this spindle is driven by a bevel gear meshing with the main bevel fixed on the rear end of the crankshaft. The top por- tion of the pump spindle bearing is cupped to form the housing for a thrust ball-race, above which is fixed the impeller. The impeller is a gun-metal casting, having six helical vanes. The lower half of the pump body FIG. 284. Water pump of Maybach engine. is an aluminum casting, to the inlet passage of which the diagonal water pipe from the radiator is coupled by a rubber con- nection. The top half of the water pump body, which is a gun- metal casting, is formed with six helical passages leading in 10 90 100 110 120 J30 SO 60 70 Gallons per Minute FIG. 285. Performance curves of Maybach water pump. a reverse helical direction to the impeller. These passages connect with the common vertical outlet passage in the top of the body casting. The center portion of the top body casting, inside THE COOLING SYSTEM 375 the helical passages above the impeller, is domed and fitted with a screwed plug. This plug is drilled with a small hole, to prevent an air-lock. Two other holes are also drilled in the bottom of the impeller between the vanes for the same purpose. The steel ball thrust race is exposed to the flow of water, a disadvan- tageous feature. Performance and efficiency curves for this pump at 2800 r.p.m. are given in Fig. 285. It will be seen that the maximum pump efficiency of 28 per cent is obtained with a discharge head of about 22 ft. of water and a capacity of 100 gal. per minute. Piping. Water velocities in pipes vary from about 8 ft. per second in small engines to 16 ft. per second in large engines. Actual pipes sizes are from 1}^ in. diameter for 90 h.p. to 2 in. diameter for 400 h.p. The frictional resistance to flow of water through straight pipes is given by h = 4f (l/d) (V 2 /2g) where h is the loss of head in feet, I and d are the length and diameter of the pipe respec- tively in feet, V the velocity in feet per second, and / is a coeffi- cient whose value is likely to vary from 0.004 to 0.010, depending on the roughness of the pipe. Inlets, outlets and bends will each offer a resistance equivalent to a length of 10 to 20 diameters. Large pipe sizes diminish the resistance and work of the pump but they weigh more and hold more water. The pump suction should be of ample diameter and as short and direct as possible. The connecting rubber hose should be firm and non- collapsible. Pipe lines should be of light tubing, bent to easy radii, with a minimum of bends and fittings. Hose connections at junction points should be very short, and should fit over cor- rugations. The fastenings should be by smoothly-bearing steel clamps which do not cut the rubber. Tape should be applied over hose and clamp and the whole shellacked. The pipes should be arranged to avoid air pockets if possible; if such occur, vent cocks must be applied. Particular care must be given to the vent cock on the pump casing. Water. The water used should be free from lime. Filling the system with boiling water makes starting easy in cold weather. Anti-freezing solutions are all more or less objectionable, and it is best to drain the system when the plane is not in use. Fig. 286 shows the properties of some anti-freezing mixtures. Alcohol lowers the boiling point and makes close control of temperature essential; the strength decreases and the freezing point is elevated 376 THE AIRPLANE ENGINE sv 45 \ \ Af) \ o> "*> \ \^^ \ > oi> X 'E xo ^ S^ S^ \ V t O\J -c < * or ^ fe \ ^ 1" ?o ^ \ \ S dg & JC ^ X. % v \ ^ \ ID X \ \ ^ Jv t X, x s -10 -5 5 10 15 ZO 25 30 Freezing Poitrf, Decj.Fahr. FIG. 286. Properties of anti-freezing mixtures. 'Expansion \ Tank '" Center Section To Expansion Tank j Honey tomb / from Engine "Airlnfake Shufors WirestoShutferi \ " Control Lever i ^"- Drain Cock Confrpf in Cock . Fron-1- Vi ew of Drain Cock Radio for Side View FIG. 287. Cooling plant of S E-5 airplane. THE COOLING SYSTEM 377 as the alcohol evaporates. Periodical tests by hydrometer are advisable. Glycerine does not evaporate, but impairs circula- tion and is detrimental to rubber. The glycerine should be stirred slowly into the water. Typical complete cooling systems are shown in Figs. 287 and 288. Figure 287 is for a 180 h.p. Hispano-Suiza engine in a SE-5 plane with two tubular nose radiators at the sides of the engine shaft, each 30 by 7K by S 1 ^ in., with a total frontal area of 450 sq. in. and a radiating surface of 12,700 sq. in. The water capacity is 83^ Ib- an( * the flow rate 30 gal. per minute. The system is provided with an .expansion tank which occupies the Filler Cap To Motor 'ToMoh>r Top View of Rad/crfor FIG. 288. Cooling plant of Le Pere airplane. leading section of the middle panel of the upper wing. A small portion of the water leaving the cylinders passes around the intake manifold and is then returned to the pump; the rest of it goes through the radiator. The radiator is masked by shutters. Figure 288 shows a 360-h.p. Liberty engine in a Le Pere two- seater plane with a wing radiator in the center section of the middle panel of the upper wing. The radiator is 31 in. long, 27 in. wide and 7 in. deep; has a frontal area of 783 sq. in. and^a radiating surface of 35,520 sq. in. The water capacity is 41.6 Ib. and the flow 80 gal. per minute Free water area 61.6 sq. in., free air area 1,247 sq. in., weight 127 Ib. The water pumped around the manifold goes to the radiator before returning to the pump. CHAPTER XV GEARED PROPELLER DRIVES A well designed airplane engine develops its maximum power at a speed (r.p.m.) considerably in excess of the most efficient propeller speed. In order to combine maximum power develop- ment with most efficient utilization of that power it is necessary to resort to a geared drive. Geared drives have been employed in a number of successful installations. A German analy- sis of these 1 is the basis for the discussion which follows. The simplest type is a single -reduc- tion with spur gears as in Figs. 289 and 290. In the Renault engine (Fig. 289), the gear ratio is two to one and consequently can be used for driving both camshaft and propeller shaft; in the Hispano-Suiza engine (Fig. 290) the gear ratio is four to three. Gears of this type show heavy wear. A design for a single reduction with internal gear is shown in Fig. 291; the internal gear housing is attached to the crankcase by an eccentric centering flange which permits accurate adjustment of the gears. This type permits great simplicity but there is difficulty in arranging satisfactory bear- ings on both sides of the gear wheels. With single-reduction gears the propeller shaft cannot be in the same axial line with the crankshaft; when this arrange- ment is desired double-reduction gears must be used. There are many possible arrangements; both pairs of wheels may be fitted with internal or external gears and in addition any one of the three shafts may be fixed while the other two drive and are driven respectively. Some of these arrangements are shown ^UTZBACH: Technische Berichte, Vol. Ill, Sec. 3. 378 FIG. -- Crank Shaft 20T.P4.57r(fnmm) 289. Renault single-reduction- gear. GEARED PROPELLER DRIVES 379 FIG. 290. Hispano-Suiza single-reduction-gear. FIG. 291. Single-reduction internal gear. Iran k Shaft anci Propeller Turning in the Same Crank Shaft and Propeller Turning! _P_ir_ect^n I Opposite Directions In termed ' ' rection 7 re bnarr turning in I he Same Direct/on as Crank Shaff Intermec/icrh Shaft Turning '- Opposite Or reef ion A Irrkrmediote Shafting Fixed in Housing B Intermediate Sfrafr Revo/vi'ng wH-h ^ Propeller Intermediate Shaft Revolving with Crank Shaft FIG. 292. Possible arrangements of double-reduction gears. 380 THE AIRPLANE ENGINE schematically in Fig. 292. In the top row the intermediate shaft is fixed; in the second and bottom rows it revolves forming the so-called planetary gears. The shaded gears are fixed and do not revolve. Some of these arrangements offer considerable diffi- culties for actual construction, notably in the matter of pro- viding suitable bearings on both sides of the gears; in others the space occupied may be great and the revolutions of the intermediate shaft very high. A simpler arrangement is one in which both pairs of gears have one gear in common. Schematic outlines of such reductions are shown in Figs. 293, 294 and 295. Figure 293 is developed from Ai. Fig. 292; Fig. 294 from A 4 ; and Fig. 295 from B 2 . The Rolls-Royce planetary gear, Fig. 296, is an actual construc- tion of Fig. 295. The three revolving intermediate shafts are JT FIG. 293. FIG. 294. FIG. 295. Double-reduction gears with a common gear. carried in a spider, C. The internal gear, a, on the crankshaft drives the three gears, 6, on the intermediate shafts, and the three gears, c, on the same shaft mesh with the gear, d, which is held against revolving in the housing. The spider, C, revolves and carries the propeller shaft. The advantage of the double-reduction gear over the much simpler single-reduction gear lies in the perfectly axial trans- mission of the power, from which the best condition of loading of the housing (pure torsion) is obtained. When the power is transmitted through two, three or four intermediate gears at equal angles, springing of the gear shafts from unequal peripheral forces or inaccurate tooth forms is avoided. Certain arrangements also make it possible to use heavy revolving masses (for instance, those of the intermediate shafts or the larger internal gears), thereby improving the uniformity of transmission and avoiding reversals of tooth pressure. The principal advantage, however, consists in the fact that on account of the load being divided GEARED PROPELLER DRIVES 381 o 382 THE AIRPLANE ENGINE between two to four intermediate gears the tooth pressures per unit of tooth face are low. Consequently small pitches and small gears can be used which in turn have smaller construction defects since the defects of manufacture resulting from the use of inaccurate dividing wheels increase with increasing radius. The disadvantage of the double reduction gear is its weight and cost and the need for exact adjustment of the intermediate shafts if all the gears are to work equally. Furthermore, a complicated construction is necessary to ensure a solid and secure assembly of the gear. To obtain and keep proper adjustment of the reduction gearing as wear occurs, it is necessary to fit a joint between the crank case and the gear case, or, in the transmission, between crankshaft and FIG. 297. Rolls-Royce single-reduction gear. gear, which will adjust itself automatically while running or can be adjusted in assembly. In the Rolls-Royce planetary gear, Fig. 296, a sliding cross linkage is used in a fixed housing. The link, (B) and e, lies between the outer engine housing and the intermediate gear wheel, d, which is held in the housing. Con- sequently the gear wheel, d, can adjust itself and always remains concentric with the crankshaft. The whole set of planetary gears also remains concentric with the crankshaft which may shift in the casing but not with the casing. The forward bearing, g and h, must be adjusted on each overhaul of the engine by the screws, /. In the Rolls-Royce spur gear, Fig. 297, the upper gear can be adjusted by eccentrically-set ball-bearing cages, c and d, and the lower gear can be adjusted on the engine shaft by screws. A universal joint is used between the crankshaft and the gear, a. GEARED PROPELLER DRIVES 383 Many difficulties have been encountered in the actual oper- ation of reduction gears principally, fracture, wear and heating of the gears. The bending stress in a gear tooth of the common involute form may be taken as W f 14 X r- approximately, where W is the load in pounds on the gear tooth, b is its width and p is the circular pitch in inches. The mean value of the loading on the tooth can be determined from the known engine power, P, and the speed, V, of the pitch circle _ 550 XP V The maximum loading on the teeth may be considerably greater than the mean loading either because of acceleration pressures, resulting from incorrect pitch or form of teeth, or because of irregular delivery of power from the engine, or on account of reinforced vibration near a resonance period of the shaft. The values of / calculated for a number of successful engines run from 30,000 to about 40,000 Ib. per square inch; the material used is generally case-hardened chrome-nickel steel. These high stresses are calculated on the assumption that all the load is carried on one tooth. With accurate pitching the deformations of the loaded tooth will transfer load to the next tooth. With oblique teeth, such as herring-bone gears, the tooth pressure is distributed on an oblique line running from the root to the tip and the bending stress is thereby reduced. The stresses are worse if the teeth bear unevenly as a result of warping in hardening, untrue keying or poor forming. The surface pressure of the opposing curved tooth faces must not be sufficient to squeeze out the oil film. The relative sliding speed of straight-toothed gears is zero at the pitch circle, and the oil is more easily squeezed out under this condition than when there is relative motion. The bearing pressure is given by the W expression -JT-J, where d is the diameter of the relative curvature of the teeth at the rolling circle. With involute teeth having radii of curvature of e\ and 62 at the rolling circle, 2 1 1 -; = H ) d a ~ e 2 384 THE AIRPLANE ENGINE the + sign applying to external gears, the sign to internal gears. Calculations from successful engines indicate that with hardened gears the bearing pressure may go up to 1,400 Ib. per square inch; if the gears are not hardened it should not exceed 450 Ib. per square inch. With internal gears the bearing pres- sures become low and hardening is, as a rule, unnecessary. Experience with roller bearings, where hardened rolls run between hardened rings, indicates a permissible bearing pressure of 2,800 Ib. per square inch or more at low peripheral speeds; if the rolls bear directly on the unhardened shaft the value falls to from 150 to 300 Ib. per square inch. With oblique toothed gears the contact shifts with great speed from side to side, as a result of which there is less tendency to squeeze out the lubricating film. Heating of the gears results from the sliding contact at the teeth and may be of such magnitude as to lead to trouble. The heat is best carried off by thermal conduction from the gears to the outer casing, but if this is not sufficient it must be assisted by oil cooling. The lubrication should not be so heavy that the oil heats up through churning; this may occur through the use of wide gears which catch the oil and force it out sideways with great force, or through locating the gears very close to the housing. For smooth running it is necessary that there should be no reversals of pressure in the gears. Four-cylinder engines give such reversals of pressure and so do six-cylinder engines at a low torque, or, with very heavy reciprocating parts, at high speeds. With a larger number of cylinders with crank angles equally spaced reversals will not occur. Central power plants have been used on several planes. The principal advantage which they offer is the possibility of con- centrating power plants in a central engine room (where they can be under constant supervision) and the resulting reduction of drag of the complete machine. There is also the possibility of reducing the number of mechanics required in a multi-engined plane. The disadvantages are the loss of power (possibly 5 per cent) in the transmission shaft and gears, and the increase in weight. Chain-driven propellers were used successfully by the Wright Brothers in 1903 and by others later. In recent years the chain drive has not been used but shaft and bevel gears have been GEARED PROPELLER DRIVES 385 employed with some degree of success. Siemens-Shuckert multi-engine planes of several sizes have used bevel gear drives. The largest of these with six engines and four propellers, is ar- ranged with the four rear engines driving the two rear propellers at half engine speed and the two front engines driving the two front propellers with a reduction ratio of 14 to 9. The couplings between the engines on the main transmission gear are a combina- tion of friction and independent couplings. The latter enable the engine to be disengaged and stopped if damaged. The articu- lated transmission shafts are connected at both ends through laminated spring couplings. The main difficulty in the operation of shaft drives has been in the setting up of "torsional resonance/' which has caused break- age of shafts and universal joints. This has been overcome by the use of a flywheel on the engine and a special clutch which combines a dog clutch and a friction clutch. The shaft should rotate at engine speed and the gear reduction should be near the propeller. Some trouble has resulted from " whirling" of long shafts but probably because bearings have been placed at nodal points; this can be avoided. With these difficulties overcome it is doubtful whether the expense, weight and complication of the flywheels, clutches, shafts and gears will not more than counterbalance the advantages of a central engine room. CHAPTER XVI SUPERCHARGING Change of Engine Power with Altitude. The indicated work in the cylinder of a gasoline engine is the product of the heat of combustion of the fuel by the thermodynamic efficiency of the engine. The thermodynamic efficiency is unaffected by the air density and depends only on the ratio of compression. The heat of combustion is determined by the weight of fuel which can be burned and this depends on the weight of air admitted and con- sequently on the density of the air. All other conditions remain- ing constant, the indicated power of an engine would vary directly as the density of the air. The brake horse power of the engine is the difference between its indicated power and the power required to overcome engine friction. At 'constant engine speed the friction will not change greatly with the air density; it increases with lowered temperature of the lubricant and decreases with lowered pressures at rubbing surfaces. If the frictional resistance remained constant the brake horse power would fall off much more rapidly than the indicated power at high altitudes. For example, an engine developing 100 i.h.p. at the ground will give 85 b.h.p. with 15 friction h.p. Operating in air at one-half ground density the theoretical indicated power is 50 h.p. and with 15-h.p. friction there would be 35 b.h.p. The brake power would be diminished in the ratio 35/85 = 0.412, while the indicated power is halved. The actual diminution in brake power is not as great as the preceding calculation would indicate; the conditions are complex and not susceptible of exact calculation. The friction horse power is partly rubbing friction and water and oil pump work and partly work done in overcoming throttling losses at the intake and exhaust of the gases. The former losses may be assumed constant with varying air density; the latter may be assumed to vary directly as the air density. If the total fricton loss is 15 per cent of the full indicated power at the ground, and the throttling losses are assumed to be one-third of the total friction loss, and if the indicated horse power is propor- 386 SUPERCHARGING 387 tional to the relative air density, d, then the brake horse power, B, at any air density is given by d - 0.10 - 0.05d 0.95d - 0.10 0.85 0.85 where B is the brake horse power at the ground. The following table gives horse powers calculated for different altitudes. It will be seen that the brake horse power is nearly proportional to Altitude, feet Relative air density, d Relative air pressure Relative b.h.p., B/B. 1.0 1.0 1.0 6,000 0.829 0.801 0.808 12,000 0.694 0.645 0.660 18,000 0.581 0.518 0.532 24,000 0.485 0.414 0.424 30,000 0.411 0.333 0.342 the air pressure. Actual tests support these calculated quantities for altitudes up to 10,000 ft.; for high altitudes the brake horse power does not decrease as rapidly as the air pressure nor so slowly as the air density but according to some intermediate law. Tests made at the Bureau of Standards 1 show a variation of brake horse power with barometric pressure as in column 7 of Table 20. The ratio of brake power to air density is given in the eighth column. It is seen that the brake power falls off more rapidly than the air density and that at one-half ground density the brake power is about 0.43 time the ground power. The variations with air density of the mechanical, volumetric and thermal efficiencies of the Liberty 12 engine and the Hispano- Suiza 300 at a speed of 1,600 r.p.m. are given in Fig. 298. The relative horse powers of Table 20 are based on constant engine speed. They may more properly be regarded as relative engine torques. Engine speed falls off with increasing altitude so that the actual horse power developed falls off more rapidly than is indicated in Table 20. With constant revolutions per minute the resisting torque at the propeller diminishes in direct proportion to the air density and consequently falls off less I 4th Annual Report, National Advisory Committee for Aeronautics, 1918, p. 502, Fig. 6. 388 THE AIRPLANE ENGINE rapidly than the engine torque. Since the engine torque is practically independent of the revolutions per minute the engine speed will diminish as altitude is gained until that speed is reached at which propeller torque equals engine torque. The actual engine power at any altitude is given by P = PG X K X No where P G is power developed at the ground, No is revolutions per minute at the ground, N is revolutions per minute at altitude, K is the quantity in the seventh column of Table 20. Barometric Pressure in Cm. of Ha. Approximate 74.4 63.7 523 43.6 36.0 29.9 74.51 63.29 52.51 43.43 35.81 29.74 0.030 0.070 0.060 0.050 0.040 0.030 O.OoO 0.070 0.060 0.050 0.040 0.030 Air Densi-hj in Lb.per Cu. Ff. Liberty 12. Hispano-Suiza 300. FIG. 298. Variation of engine efficiencies with air density. Table 20 shows that the engine power at constant speed is almost exactly proportional to the barometric pressure. On this basis the engine power at an elevation where the barometer is B cm. is Supercharging. The diminution in power of a gasoline engine with increasing altitude results in a moderate reduction of speed in horizontal flight. If greater power were available the ground speed could be maintained at all elevations or exceeded, if desired. Much effort has been expended in attempts to prevent or reduce SUPERCHARGING 389 5 PQ fc O 8 a 11 O V ^3 T3 ' a> t, II 3 J J H! IP U3 iO U3 O '-i>OOt^ OOOOOOOOOOO O O 05 O Oi OOOOCiOiCiOiaiOiOOOOOOOOOO r-I 1H rH O O O O O O O O O O O t-o ^OOC^OOe5O5>Oi-HOO^J<'-lOOOCO^H i-Idddddddo'ddddo'dd I I I I O5 (N O5 CD CO -< OS t "5 O d U5 1C IO id Oit N -*OCOC^OO500CO l C^COC^^OOi CIC^C^OlC^Wi-li-li-lrHi-tT-lr-tT-ll-l i -t^-OOO5 390 THE AIRPLANE ENGINE this falling off in engine power. Such falling off would be entirely avoided if steam power could be substituted for gasoline power, since the boiler and condenser pressures would be independent of the barometer pressure. Attempts to design a light-weight steam plant have not been successful; there is no difficulty with engine or condenser (which takes the place of radiator), but it has not been found practicable to design a boiler to withstand high steam pressures and of sufficiently extended heating surface without arriving at weights which are prohibitive for airplane use. Furthermore, the lower fuel economy of a steam plant would necessitate the carrying of a greater weight of fuel. Two general methods present themselves for increasing gas engine power at high altitude: 1. To select an engine so large that it will give the desired power when running with wide-open throttle at the high altitude at which the airplane is intended to fly, and to operate it at partial throttle at all lower altitudes. 2. To select an engine which gives the desired power at the ground and add some device for supplying the cylinder with air at a pressure greater than the barometric pressure when desired. This process is known as pre- compression or supercharging. As an illustration, suppose it is desired to fly at 20,000 ft. developing 400 h.p. This can be accomplished either by install- ing an engine which would develop 800 h.p. with wide-open throttle at the ground; or by installing a 400-h.p. engine provided with a supercharging device which is able to maintain that horse power at all altitudes up to 20,000 ft. If the large engine is used the engine weight will be increased. An estimate made of the increase of weight which would result from doubling the power of a Liberty motor by doubling the piston area per cylinder, while keeping the stroke constant, indicates this increase would be about 40 per cent. If the power were doubled by doubling the number of cylinders the weight would be nearly doubled. It should be noted that if the engine is not permitted to develop more than 400 h.p. at any elevation, the radiator, water pump and general cooling system will not be larger than for a 400-h.p. engine. If the smaller engine is used the engine weight will also be increased by the addition of the supercharging apparatus and the engine becomes more complicated. Oversized Engine. In this system, the greater weight of the engine is offset not only by greater simplicity (as compared with a supercharging engine) but also by greater economy. Such SUPERCHARGING 391 engines should be provided with an automatically controlled throttle valve, actuated by some device (generally similar to an aneroid barometer) which responds to changes in atmospheric pressure. An example of such a device is given in Fig. 299 in which an airtight flexible chamber filled with air at low pressure actuates a balanced double-seated throttle valve. If the throttle is placed before the carburetor, the top of the float chamber must be kept in communication with the low-pressure side of the throttle. The control may be so adjusted as to give constant horse power at all altitudes up to that at which the throttle is wide open; the power cannot be maintained beyond that point. With an engine so operated it is possible to use a higher ratio of compression, without danger of preignition, than with an engine which has wide-open throttle at the ground. Air Tight Chamber _T Outside Afr u TJ and Tnlef from Compressor FIG. 299. Automatic throttle control for oversized engine. With constant power output the weight of the charge admitted per cycle will be approximately constant and the pressure in the cylinder at the beginning of compression is also approximately constant. The latter quantity is actually a little more at the ground than at higher altitudes because the engine is exhausting against a higher barometric pressure and consequently there is a greater weight and pressure of burned gases remaining in the cylinder to be mixed with the new incoming charge. Further- more,' the efficiency at the ground would be lower than at high levels on account of the higher back pressure. With constant power output the pressure in the cylinder at the beginning of com- pression would be less than 7 Ib. per square inch at 20,000 ft. elevation, and probably less than 8 Ib. per square inch at the ground. That is, the maximum pressure to be expected at the beginning of compression is 8 Ib. per square inch as compared with 14 Ib. in the usual engine. This results in lower compression and 392 AIRPLANE ENGINE explosion pressures. Furthermore, the cylinder temperatures are lower throughout the cycle mainly in consequence of the smaller amount of heat developed by the explosion and the greater cooling effect of the water jacket. Under these conditions it is possible to employ higher compresssion without danger of Meters 3048 46Z7 10,000 14,000 18000 Altitude, Feet. 22POO FIG. 300. Effect of altitude on variation of engine power with compression ratio. preignition. Engines have been operated in this manner with a ratio of compression as high as 7. The employment of a high ratio of compression will increase the available power, particularly at high altitude. This is shown clearly in Fig. 300, which gives the results obtained at the Bureau of Standards with an engine supplied with three different sets of pistons to give different ratios of compression. The curve B is for a compression ratio of 5.3, which is here regarded as standard. The curves A and C are for compression ratios for 4.7 and 6.2 respectively. It will be seen that, calling the horse power with Per Cent no-ease of M.E.P 3 in 5 u |V'/5W ^'^ x ^ r"*--'^ V^%17 X 100 so -t- c SL 6 B 40 33 o n D :E 20 40 00 80 100 120 Air Intake Tern perature,Dec} Fahr. FIG. 324. Temperature change in intake manifold. With air initially cold it will not be possible to vaporize the fuel completely by heat absorbed from the air because at low temperatures the air will become saturated with the fuel vapor before all the fuel is vaporized. Table 14, page 229, shows, for example, that with a theoretically correct mixture, the air cannot hold all the pentane in the vapor form at a temperature below about 38F. In such case the only chance for complete vapor- ization is by supplying heat from outside. This is accomplished by utilizing some of the heat either of the jacket water or of the exhaust gases. The possibilities are (1) to preheat the air before it enters the carburetor, (2) to heat the mixture in the manifold, and (3) to heat the manifold locally at some place on which the 1 SPARROW, Technical Note, No. 55, Nat. Adv. Comm. Aeronautics. MANIFOLDS AND MUFFLERS 419 liquid drops impinge so as to supply heat to the liquid only (hot-spot method). All preheating is objectionable in that it diminishes the density of the charge and thereby decreases the capacity of the engine. That method of preheating is best which causes vaporization with the minimum resulting temperature of the mixture. All three methods of preheating are employed in airplane practice. In some of the German engines preheating the air is accomplished by taking the air through pipes in the crankcase (Fig. 78), which has the advantage of cooling the lubricating oil. The more common procedure is heating the manifold by jacket water. FIG. 325. Liberty engine intake manifold. There is no general consensus of opinion as to the best form of manifold. It is desirable that sharp turns should be avoided as far as possible, that the various branches should have approxi- mately the same length, that sudden enlargements should be avoided and that the velocities should be high enough to prevent deposition of liquid drops but not so high as to cause a large frictional resistance. Mean velocities of 120 to 200 ft. per minute are common but values up to 250 ft. per minute have been used successfully. Manifolds usually divide themselves into two classes, the short-branch type and the long-branch type. The standard Liberty manifold (Fig. 325) is a good example of the short-branch 420 THE AIRPLANE ENGINE type; it is water-jacketed and has a baffle plate opposite to the inlet to equalize the lengths of the three branches. The Benz manifold (Fig. 326) is an example of an unjacketed long-branch manifold with all three branches of the same length and with long- turn elbows. Another design for accomplishing the same purpose is shown in the Hall-Scott engine (Fig. 63). Manifolds for vertical engines can usually be arranged in any way the designer likes. In the Maybach engine (Fig. 80) the carburetors are at the ends and the manifolds run along the side of the engine with no attempt to equalize the lengths of the branches. Ordinarily such arrangements as those of Figs. 325 and 326 are used. In 90-deg. Vee engines there is plenty of room in the Vee to accommodate the carburetors and the intake is usually inside the FIG. 326. Benz intake manifold. Vee. With this location a short-branch manifold must be used. The Hispano-Suiza engine (Fig. 50) shows a typical arrange- ment with the transverse pipe water-jacketed. If long-branch manifolds are to be used they must either be placed in the rear of the engine, as in the Curtiss engines, Figs. 59 and 62, or the intake valves must be on the outside of the Vee. In 60-deg. and 45-deg. Vee engines the space inside the Vee is small and a favorable design of manifold is difficult if the car- buretors are placed inside the Vee. If the inlet valves are placed outside the Vee, the exhaust pipes will be crowded inside the Vee and may give rise to troubles caused by their proximity to the valve springs, etc. An alternative arrangement is to provide a space between the two central cylinders of each block (as in the Bugatti (Fig. 67) and Fiat engines (Fig. 76)) and to lead the induction pipes from carburetors mounted outside the Vee through these spaces to manifolds inside the Vee. MANIFOLDS AND MUFFLERS 421 In radial and rotary engines the distribution problem is com- paratively simple, especially where an induction chamber is provided in the crankcase. The special distributing chamber of the Bristol " Jupiter" engine (Fig. 148) is noteworthy. Exhaust Manifolds. The function of the exhaust manifold is, primarily, to conduct the exhaust gases away from the airplane with the minimum back pressure at the engine and without fire risk to the airplane or annoyance from the discharged gases to the pilot. An additional function may be to muffle the sound of the exhaust, although this has usually been considered unimpor- tant in military machines. The manifold is usually required to have a clearance of 2^ in. from wooden parts and of 3H m - from fabric parts of the airplane. FIG. 327. Hall-Scott exhaust manifold. The simplest exhaust piping is either its complete absence as in rotary engines and certain stationary engines or the use of short tubes discharging outwards or upwards. In some engines these stub tubes are cut off on the outer end at an angle of 45 deg. so as to discharge backwards as well as outwards. The absence of exhaust pipes or the use of short straight pipes is advantageous not only in reducing back pressure but also in permitting better cooling of the exhaust valves by radiation and avoiding heating of the exhaust valve springs. This arrangement does not conduct the gases away from the crew of the airplane. A method of discharging the gases overhead and to the rear with small back pressure is shown in the Hall-Scott manifold of Fig. 327. This arrangement obstructs the view of the pilot in a tractor machine but diminishes the noise heard from below. Long radius curves are very essential for all the branches if back 422 THE AIRPLANE ENGINE pressure is to be kept down; the radius should be about 2K times the diameter of the pipe. The arrangement of Fig. 328 with discharge to the rear and with an exhaust main of increasing diameter not only carries the gases away from the crew but slows down the gases before exit and thereby tends to diminish noise. As the exhaust period lasts more than two-thirds of a revolution there is overlapping of exhausts in a manifold connecting three or more cylinders. The velocity of the gases immediately after the opening of the exhaust valve is extremely high and its effect on the exhaust from any other cylinder whose exhaust valve is open at that time should be carefully considered. By the use of two concentric exhaust mains to which the cylinders are connected so FIG. 328. Hall-Scott exhaust manifold. that no two consecutive exhausts go into the same main, an ejector effect can be obtained from the action of a newly opened exhaust on the exhaust which is closing, which may cause substan- tial scavenging in the closing cylinder. The cross-section area of manifold branches is governed by the size of the ports in the cylinders; an average value is about 0.15 sq. in. per brake horse power of the cylinder. Mufflers, to be efficient, must slow down the exhaust gases to velocities below that of sound (1,100 ft. per second); actual gas velocities probably exceed 2,000 ft. per second. For airplane use the muffler must be of light weight and yet durable. The con- structions employing reversal of direction of the gases and dis- charge through small holes are likely to give excessive back pressure. Tests by Diederichs and Upton show that the volume of the muffler should be about three times that of a single cylinder MANIFOLDS AND MUFFLERS 423 of the engine, and that the inlet to the muffler should be tangen- tial so as to give the gas a whirling motion. For durability the muffler should be attached to the end of a tail pipe 6 or 8 ft. long which will cool the gases sufficiently to prevent excessive oxida- tion of the muffler. One of the simplest and most successful mufflers is shown in Fig. 329. The tangential inlet pipe starts of circular cross-section and flattens out in a fan shape so as to give admission for almost the whole length of the muffler. An inner Inlef Inlet FIG. 329. Diederichs and Upton exhaust muffler. shell, AB, is provided with a large number of holes except in the small arc between A and B; these holes are smallest near B and increase in diameter from B to A (contra-clockwise). The gas passes through these holes and then through more holes in the innermost shell and finally escapes at the open end of the inner- most shell. With this muffler the exhaust noise can be reduced 80 per cent with about two-thirds of 1 per cent reduction in engine power. Mufflers of this type are found to give less back pressure than those with axial admission. CHAPTER XVIII STARTING The starting of an airplane engine depends on three things, (1) obtaining an explosive mixture in the cylinder, (2) a device for igniting it, and (3) a device for turning the engine over. Hand Starting. The idling device on the carburetor will fur- nish a rich mixture to the cylinders if the throttle valve is closed and the engine is turned at a sufficient speed. If the revolutions per minute of the engine is too low (below 20 to 30 r.p.m.) the velocity of the air in the intake manifold will be insufficient to carry the fuel into the cylinders; this will be the usual condition FIG. 330. Priming system for a 12-cylinder Vee engine. with hand starting. To overcome this difficulty the cylinders may be primed through individual priming cocks or through such a priming system as is shown in Fig. 330. With the engine cold only a small part of the gasoline will be vaporized and the rest will go as liquid into the cylinder and will dilute the lubricant. In very cold weather there will be difficulty in vaporizing enough of the gasoline to form an explosive mixture unless the jacket is filled with hot water or some other device has been used to heat the engine. In such a case a number of attempts may have to be 424 STARTING 425 made before the engine starts, and as an excess of gasoline is put in the cylinder before each attempt the walls may be washed clear of lubricant and scoring of the cylinder may occur when the engine finally starts. To overcome these difficulties a more volatile fuel such as ether may be used for priming. A more satisfactory practice, for cold weather, is to use hydrogen as the starting fuel. This has the great advantage that it does not have to be vaporized and that it forms an explosive mixture throughout a great range of strengths. A hydrogen-air mixture will explode if the hydrogen forms from 10 to 66 per cent of the mixture by volume; with gasoline vapor the range is only from 1.5 to 4.8 per cent. The hydrogen may be admitted through the priming system as shown in Fig. 330 ; it is best to take the gas from a fabric balloon in which it is at atmospheric pressure and not from a high-pres- sure bottle which would be likely to give an excess of gas. The hydrogen must be shut off shortly after the starting as it gives more violent explosions than gasoline. The regular magneto will not turn fast enough with hand starting to give a spark. If a battery system of ignition is used there is no difficulty from this source. If a magneto system is used it is customary to supply a hand-operated starting magneto which is geared to run at high speed and is turned independently of the engine. With hand starting the engine may be pulled over several times to fill the cylinders with explosive mixture, after which a shower of sparks is sent from the starting magneto through the distributor to the fully retarded cylinder which is ready to fire, or the ignition may be left on, fully retarded, while the propeller is pulled over either by hand or by rope. The directions for starting the Liberty engine by hand are as follows: Inject jj^> oz. of lubricating oil through each priming cock. Turn switch "off." Turn the engine over five times. Open throttle slightly. Retard spark fully. Prime each cylinder twice. Turn engine over twice. Turn on one switch. Pull down and forward on propeller blade. After the engine starts: Advance spark half way. Turn on both switches. Leave throttle undisturbed for 5 min. The lubricating oil should be warm by the time the jacket outlet water has reached 150. If, in cold weather, it is not, stop the engine for 5 min. ; then start all over again. Accelerate and slow down the engine a few times. Note that the oil gage registers pressure, 5 Ib. at 600 426 THE AIRPLANE ENGINE r.p.m. At this speed, the ammeter should show " discharge. 7 ' At 1,000 r.p.m. it should indicate " charge." After 5 min. more, open the throttle wide. The speed should rise to about 1,600 r.p.m. The necessity of turning the engine over preliminary to starting can be avoided by the use of a mechanism which will lift the valves and permit the pumping of an explosive charge into the cylinders. In the Maybach engine this is accomplished (Fig. 331) by lifting all the tappets (inlet and exhaust) off their cams by depressing the hand lever, A, which rotates the two lay shafts, BB, and lifts the tappets through slots in the lay shafts engaging with small lugs projecting from the tap- pets. At the same time the shutter, C, in the exhaust main is closed. The hand pump, E, is then operated and air is sucked through the carburetor and the intake manifold and through the cylinder to the exhaust man- ifold and to the pump. When the cylinders are thus charged the lever, A, is brought ver- tical, which restores the engine to its operating position. The starting magneto is then used to start the engine. The whole operation can be carried out from the pilot's seat. The starting torque in high-power airplane engines is yery considerable. The load consists of two parts, (1) the compression load, and (2) the friction load. The compression load increases with the cylinder diameter and the compression ratio. The maximum torque does not change with the number of cylinders because high-compression pressure will exist at any moment in one cylinder only. As the compressed gas re-expands the mean torque for two revolutions does not increase with increase in number of cylinders. Friction is the principal resistance in starting the engine, especially in cold weather when the lubricating FIG. 331. Starting mechanism of May- bach engine. STARTING 427 oil may be near the solidifying point. Tests at McCook Field 1 to determine the average starting torque for various engines have yielded the following results. The average air temperature was AVERAGE STARTING TORQUE, POUNDS-FEET Engine Number cylinders Rated horse power Normal speed t | g" 1 Stroke, inches Number of engines tested Number of trials Starting torque, pounds-feet Throttle open Throttle closed Maxi- mum l H J Liberty 12 8 6 8 6 400 300 210 180 185 ,700 ,800 ,700 ,600 ,400 5.00 5.51 5.00 4.75 5.90 7.00 5.91 7.00 5.25 7.08 2 2 2 1 1 .6 6 6 3 3 130 106 133 110 150 124 102 105 82 139 156 101 135 87 153 143 96 110 77 145 Hispano-Suiza Liberty Packard BMW 75F. With freezing temperatures the results would have been much higher probably doubled. It should be noted that the starting torque does not increase nearly as rapidly as the engine horse power. FIG. 332. Compression release of Benz engine. From the preceding table it may be presumed that with low temperature the mean starting torque for a 400-h.p. engine will be from 300 to 400 Ib.-ft. It is difficult for a mechanic to exert this torque on the propeller even in a land plane without danger to himself from overbalancing; in a sea plane it is even more difficult to accomplish. The starting torque can be diminished by reduc- ing or eliminating the compression. For this purpose compres- sion release cams may be provided on the camshaft to keep the exhaust valves open during the compression stroke. In the Benz engine (Fig. 332) the compression release cams are brought 1 Air Service Information Circular, No. 126. 428 THE AIRPLANE ENGINE FIG. 333. Compression release of Basse- Selve engine. into action by the axial movement of the camshaft effected by a square-thread screw operated by a small lever at the rear of the crankcase. The camshaft is returned to its normal running position by a spring inside the front end of the shaft. The relief cams open the exhaust valves 35 deg. early and close them at 22 deg. late. In the Basse-Selve engine (Fig. 333) the compression release is operated by means of a rod which lies horizontally along the outside of the camshaft casing directly underneath the exhaust-valve rocker arms. The rod is slotted in such a way (see separate detail, Fig. 333) as to form cams which lift the exhaust valve rockers when the rod is partially rotated by means of the hand lever at the end of the rod. With this device the exhaust valves remain open so long as the rod is in the rotated position. In addition to the difficulty which is experienced in starting large engines by the propeller there is a considerable element of danger, which has proved fatal in many cases. To obviate this, recourse may be had to a portable engine cranker, which is usually available only at airdromes, or to a starting mechanism integral with the engine. Whatever the nature of the starting mechanism it should be thrown automatically out of action as soon as the engine fires and it should also go out of action if the engine fires before the dead center and starts to turn backward. The con- nection from the starter to the engine is through a dog or clutch, as in Fig. 334, which is pushed out of mesh when the engine starts forward but will remain in mesh if the engine starts backward. FlG - 334> ~ Dog clutch ' The acceleration of the engine shaft as a result of firing one of the cylinders is very great and if the engine starts backward this acceleration will be transmitted to the starting mechan- ism. As the starting mechanism is always geared, with a ratio which may be 100:1 or more, the motivating element will be given an enormous acceleration which will produce high teeth pressures STARTING 429 in the gears and will probably strip the teeth of the gears or the dog projections of the clutch. If the gears are hand-operated through a crank the gear ratio may be 10 or 20 to 1, and, although the teeth may hold on a back fire, the mechanic will be endan- gered by the high speed of the handle. To safeguard the starting gear some kind of friction clutch or other safety drive should be placed close to the dog. Multiple-disc clutches are suitable on account of their compactness and low weight, but all such friction devices are uncertain in their action. A different type of safety device with an oscillatinggmember is shown in Fig. 335. \ The oscillating member, A, takes no part in the trans- mission of the load. The engine drive is to the right of the figure. In normal rotation, A is driven by the gear wheel, C, and is free to oscillate as determined by the tapered sides of the stationary projections on D. If a reverse rotation FlG - 335. Safety clutch, occurs D prevents the rotation of A, which causes the sleeve on which C is mounted to move to the right, and thereby throws C out of mesh. As the action cannot take place until some back- ward movement has occurred this gear should be placed on a shaft geared to the engine shaft, so that release may occur quickly. In a hand-operated system it might be on the handle shaft. Integral cranking mechanisms are either operated by hand, by compressed air, or by electric motor. It is usually necessary to have the engine rotating at a speed of from 10 to 20 r.p.m. before regular operation will take place. Hand mechanism must have the operating handle at the side or rear of the engine and consequently must employ worm, bevel, or helical gears, usually in conjunction with spur gears. The efficiency of the gear train is poor and the attainable cranking speed very low. The hand mechanism for the Hispano-Suiza engine is shown in Fig. 336. It includes double-reduction spur gears, a dog clutch, a releasing spring and a gear-driven starting magneto. Compressed air may be utilized (1) in a motor which cranks the engine, (2) it may be carbureted and sent through a dis- tributor and exploded in the cylinders, or (3) it may go direct to the cylinders at a high pressure through a distributor and SHERMAN, The Automobile Engineer, December, 1919. 430 THE AIRPLANE ENGINE special non-return valves. If a compressed air motor is used, it can be run from the engine as an air compressor to store up the -'Starting Magneto FIG. 336. Hand-starting mechanism for Hispano-Suiza engine. - me rear flange, the -Engine Face FIG. 337. Radial air compressor. air necessary for starting. In the Motor-Compressor Company's starter (Fig. 337) a multi-cylinder radial compressor is lined up STARTING 431 with the crankshaft at the rear of the engine. When operated as a compressor it is directly connected through a positive clutch to the engine and runs at engine speed, compressing air up to 230 Ib. pressure in a wire-wound tank. When used as a starter it drives the engine through a train of spur gears and rotates at seven times engine speed. Automatic arrangements are provided for throwing out the compressor when the air pressure reaches 230 Ib. and for throwing out the cranker when the engine starts. The apparatus weighs about 50 Ib. for a 200-h.p. engine. FIG. 338. Compressed air distributor. In the Christensen system compressed air is sent through a special carburetor, a distributor and non-return valves to the cylinders. The air pressure is sufficient to start the engine (say 100 Ib.) and a retarded spark gives a late explosion and initiates regular operation. This system uses a minimum of compressed air for the starting. An air compressor driven by a hand-operated clutch from the crankshaft and an air tank are necessary parts of the system. The whole weighs about 40 Ib. for a six-cylinder engine. 432 THE AIRPLANE ENGINE Compressed air can be used inside the cylinders without carbureting. If no air compressor is provided a steel air tank must be carried of sufficient capacity for several starts. This arrangement uses a distributor and non-return valves; it will weigh less than the Christensen system, but is likely to leave the aviator stranded on occasion. The details of an air distributor are shown in Fig, 338. When the starting lever, H, is thrown in, the central main air valve, Bj is opened by the boss at the end of the sliding sleeve and the individual valves, D, E, admitting air to the individual cylinders are opened in turn by the rotation of the face cam, G. Another method which has been used but is now abandoned is to insert a black powder cartridge in a special fitting in the cylinder head. On detonation this will give a considerable pressure and should start the engine. It causes a deposit of carbon in the cylinder and is not adapted to remote control. Electric cranking has been used considerably. It requires a 12-volt battery which will normally have to supply 100 amperes but may be called upon to give to 200 amperes for a half minute or more; an electric motor which will operate at 3,000 to 4,000 r.p.m., carrying a small spur gear on the armature shaft; and a double reduc- tion gear with a speed reduction ratio of 100 to 150. The last gear is pref- erably faster 1 f o Xl propeller hub motor reduction gears being mounted on the crankcase between the FIG. 339. Electric starter. front cylinders and the propeller hub. The gears can be brought into mesh by the use of a solenoid wired in series with the motor. The Bijur starter used on Liberty-12 engines has a six-cell battery weighing 35 Ib. with a capacity of starting the engine 150 times under normal conditions. The starting motor and gear may weigh 24 Ib. and give an engine speed of 40 to 50 r.p.m. The maximum torque available at the engine crankshaft is 1,300 Ib.-ft. An arrangement in which the gears are kept inside the nose of the engine is shown in Fig. 339 ; x this arrangement sup- 1 SHERMAN, loc. cit. STARTING 433 poses the starter incorporated in the engine design and not adapted, as usual, as an afterthought. If this starter is mounted at the rear of the engine it can be made a combined hand and electric starter by putting a dog clutch and cranking handle on the intermediate shaft. Various portable crankers have been devised for airdrome use. The U. S. Air Service has used an electric cranker driven by an automobile-starting motor and storage battery through double- reduction gearing, and mounted on a motor truck. As the weight of the gearing does not have to be considered in this case it has been found unnecessary to provide an automatic release in case of engine back fire. The cranker is mounted in a spherical bowl permitting universal adjustment; it is brought up to the end of the propeller shaft and adjusted so that its shaft is in line with the engine crankshaft. An engagement lever then pushes the perforated face plate at the end of the cranker shaft against the front propeller hub flange when some of the nuts enter the perforations. The engagement lever is withdrawn as soon as the engine starts. The Odier portable starter, which has had considerable use, employs a long single-acting steel cylinder and piston operated by carbon dioxide from a steel bottle of liquefied carbon dioxide. The piston carries a pulley at its free end, over which a cable is passed with one end fastened to the cylinder and the other wrapped around a drum and then fastened to an elastic cord. The drum has four bolts placed symmetrically around the periphery at one end parallel to the drum axis projecting sufficiently to engage a kind of dog clutch on the front propeller hub flange. The cylinder is carried on an inclined wooden arm and a vertical leg of adjustable height, so that the bolts can be brought up to the propeller hub level. The high pressure of saturated vapor of carbon dioxide (308 Ib. per square inch at 0F.) provides a large starting force in a cylinder of small diam- eter. The piston stroke is such as to give two revolutions of the engine with high speed. The cranker is thrown forward and out of mesh when the engine starts by the action of the dog teeth on the propeller flange. In case of a back fire the piston is pulled back, the gas is recompressed, and the elastic cord becomes slack and permits the drum to revolve freely. The weight of the whole apparatus is 44 Ib. so that it can be carried in the plane if desired. 28 CHAPTER XIX POTENTIAL DEVELOPMENTS Increased Compression. The best airplane engines give a notably better performance, both as regards fuel consumption and horse power per unit of piston displacement, than other gasoline engines. The possiblity of this improved performance results from the use of a higher compression ratio, which in turn is only possible through the use of the volatile aviation gasoline. Ordinary commercial gasoline containing larger fractions of the heavier paraffines (nonane and decane) would detonate at the compression pressures reached in airplane engines. The de- pendence of fuel consumption on the compression ratio is shown in the following table, 1 which gives the theoretical consumption of gasoline (lower heat value 18,600 B.t.u. per pound) per brake horse-power hour with an assumed mechanical efficiency of 90 per cent and with variable specific heats. Ratio of compression 4 4 5 5.0 5.5 6.0 Gasoline per brake horse-power hour 416 398 375 0.361 0.350 The best modern engines use 0.45 to 0.50 Ib. per brake horse- power hour with a compression ratio around 5.0; that is, the attained efficiency is 0.375 -f- 0.45 = 83 per cent of the theore- tical efficiency. With higher compressions this relative efficiency tends to increase. At maximum load, which is obtained only with richer mixtures (see p. 33), the relative efficiency averages 75 per cent for water-cooled and 76.5 per cent for radial air-cooled engines with aluminum cylinders. The use of still higher compression pressures, and consequently higher efficiencies, is possible by the use of gasoline mixed with toluol, alcohol or other substance which will prevent detonation (see p. 236). This field of improvement is now under active investigation and promises considerable improvement. 1 GIBSON, Trans. Royal Aeronautical Society, 1920. 434 POTENTIAL DEVELOPMENTS 435 Use of Inert Gases. A high compression pressure can be used without danger of detonations, and consequent preignitions, by taking in cooled exhaust gases with the charge. The influence of such admixture is shown in Fig. 340, which is taken from Ricardo's tests. 1 With a high-grade fuel which, when operating at full throttle and with an economical setting, detonates at a compression ratio of 4.85:1, the full power can be maintained by admitting inert gases in sufficient quantity to prevent detonation up to a ratio of compression of 6:1. That is, the decrease in weight of fresh charge taken in is fully compensated by the in- crease in engine efficiency up to that ratio of compression. If still higher compression is used (for an oversized engine for high- 1 150 140 130 g 0.6 0.4 0.3 40 4.5 5.0 5.5 60 6.5 7.0 7.5 8.0 Compression Ratio. FIG. 340. Effect of addition of inert gases on engine performance. altitude flight, see p. 390) the power will fall off. The increase in economy by the use of the inert gases, with increase of com- pression ratio from 4.85 to 6, is seen to be about 6 per cent. The dotted lines show the performance obtained using a fuel doped to prevent detonation and without the admission of inert gases. With a compression ratio of 7:1 the exhaust-controlled engine develops 84 per cent of the power which would be developed by a pure non-detonating mixture. Any increase in efficiency from increase of compression ratio will also increase the power output in practically the same ratio and is therefore doubly valuable. 1 Trans. Royal Aeronautical Society, 1920. 436 THE AIRPLANE ENGINE The maximum brake m.e.p. possible for an engine with inlet valve closure of 50 deg. late, with volumetric efficiency of 88 per cent and mechanical efficiency of 90 per cent, is as follows: 1 Nominal compression ratio 4 5 5. 5 5 6 Maximum brake m.e. P 130 5 138. 143 5 148 The best recorded results for both air- and water-cooled engines with compression ratios of 4.5 to 5.0 are very close to these figures. Tests of single engines have shown consistently better results than those of multi-cylinder engines. The best air-cooled single- cylinder engines have shown a relative thermal efficiency of 91 per cent. The difference between the best performance of a single-cylinder water-cooled engine and the performance of a 12-cylinder Vee engine is from 8 to 10 per cent. The difference depends on the efficiency of the induction system and represents the possible saving by better distribution in the induction system. As the efficiencies of the best single-cylinder engines using weak mixtures are within 10 to 15 per cent of the theoretical maximum, it is evident that little further progress is possible in improving the thermal efficiency of engines using the present cycle of operations. Increase in capacity (b.h.p. per unit of piston displacement) can be obtained by supercharging or by im- proving the volumetric efficiency. This latter method offers some chance for improvement as the measured values vary from 70 to 85 per cent with exceptional values up to 90 per cent. The fire risk in airplanes could be practically eliminated by the use of kerosene as fuel. Kerosene may be used either (1) by vaporizing it outside the engine, or (2) injecting it into the cylin- der as a liquid either during the suction stroke or at the end of compression. To vaporize a reasonable proportion the initial temperature must be not less than 140F., which results in a reduction in the weight of the charge of about 20 per cent as compared with gasoline and a corresponding decrease in engine power. Furthermore, it is chemically much less stable than gaso- line and detonates at a lower temperature so that a lower com- pression ratio must be used, which further diminishes the power and decreases the efficiency. The heavier fractions condense on the cylinder wall and, passing into the crankcase, thin the 1 GIBSON, loc. tit. POTENTIAL DEVELOPMENTS 437 lubricating oil. Injection of the fuel during the suction stroke intensifies this last trouble but reduces the loss due to preheating. Injection at the end of compression presents many difficulties common to Diesel engines, which are not yet satisfactorily solved. Modifications of the Otto cycle would seem to offer considerable possibilities for increased efficiency. The pressure at the opening of the exhaust valve is usually from 60 to 70 Ib. per square inch. If more complete expansion could be obtained a considerable increase in efficiency might be effected. Attempts have been made to realize this potential increase in work along two lines, (1) by more complete expansion in the cylinder and (2) by expanding the gases after leaving the cylinder. 1. A lower terminal pressure in the cylinder can be obtained either (a) by throttling or cutting off the admission of the charge or (b) by making the expansion stroke longer than the com- pression stroke. The former method (a) is that of the oversized engine (p. 390) and is employed primarily for maintaining power at high altitudes. Its use requires a larger and therefore heavier engine for a given capacity, which is a serious detriment. The method (b) can be carried out by the use of a variable -stroke engine and has the incidental advantage of permitting a more complete scavenging of burned gases. The Zeitlin engine which is now under development is an example of a variable stroke engine, but this feature is utilized in this case only to give better scavenging and thereby to permit the admission of a greater weight of charge. It is a single-valve, nine-cylinder air-cooled rotary which follows the Gnome engine in taking in its air supply through the open exhaust valve and mixing with it an overrich mixture through ports uncovered by the piston near the end of the suction stroke. The variable stroke is obtained by mounting, on the crankpin, eccentrics, which are driven by gearing around the crankpin in the same direction as the engine but at one-half engine speed. The connecting rods are mounted on these eccentrics and consequently the piston travel will vary throughout two revolutions of the engine. In the engine with 107.75 mm. crank throw, the working stroke of the engine is 181 mm. ; the exhaust or scavenging stroke is 203.5 mm. ; the suction stroke is 226 mm. and the compression stroke 203.5 mm. As the admission ports are not covered until the piston has made part of the compression stroke the effective compression stroke is practically the same as the working stroke. The admis- 438 THE AIRPLANE ENGINE sion ports in the cylinder are uncovered near the end of the suc- tion stroke. The length of the scavenging stroke is such as to clear the cylinder almost completely of burned gases, so that the new charge is undiluted by them. This engine is arranged to give variable compression by open- ing the exhaust valve during the early portion of the compression stroke and permitting some air to escape before it has time to mix with the overrich mixture. It is designed for a maximum compression ratio of 7 and has controls for decreasing this to 4.5. The maximum compression is for altitudes of 10,000 ft. or more; the minimum compression is for ground operation. The strength of the mixture will obviously change with the ratio of compression; the mixture will be rich at the ground and will be leaned to maximum economy strength at normal flying level. 2. The further expansion of the gases after leaving the cylinder can be carried out either in a gas turbine or in a reciprocating engine. The use of the exhaust gas turbine for driving a super- charging blower has already been discussed (p. 408) ; in view of the high speed of the turbine shaft, which is necessary if the turbine is to have fair efficiency, it is doubtful whether such a turbine could be geared down so as to help drive the propeller without excessive loss of power. For driving high-speed auxil- iaries such as the supercharging blower it has an important field. Compounding. Expansion of the gases in a reciprocating engine can be accomplished by following the methods used in compound steam engines. The problem is simplified in some respects in the gasoline engine because one double-acting low- pressure cylinder can take the exhaust from four high-pressure cylinders and as the temperature of the gases is greatly reduced by the time they reach the low-pressure cylinder it might be possible to operate without trouble from excessive piston tem- perature. As the friction work in the high-pressure cylinders is equal to about 15 Ib. per square inch of piston area there should be a pressure drop of at least that amount at exhaust, or with 70 Ib. terminal pressure in the high-pressure cylinder the receiver pressure should be about 50 Ib. Furthermore, in order to get a full charge in the high-pressure cylinder it would be necessary to have an atmospheric exhaust from the high-pressure cylinder at the end of the exhaust stroke and immediately after closure of the exhaust to the receiver. The piston displacement of the POTENTIAL DEVELOPMENTS 439 low-pressure cylinder would probably have to be about three times that of the high-pressure cylinder, or with the same stroke its diameter would be 1.7 times that of the high-pressure cylin- der. Attempts to construct compound gas engines have been made in stationary types but without commercial success. The extra power obtainable has not justified the additional first cost and maintenance. It is possible, however, that more success may be met in the airplane engine where first cost is not of prime importance. The weight of the engine per horse power should not be increased by compounding and the weight of fuel used should be appreciably diminished. Two-cycle. A modification of the Otto cycle which offers possibilities of considerable reduction in weight per horse power is two-cycle operation. The normal Otto cycle requires four strokes for the completion of the cycle of which two are used solely for pushing out the burned charge and taking in the new charge. The essential parts of the cycle are unchanged if the exhaust and admission processes are speeded up and made, in part, simultaneous by using a slightly precompressed charge to sweep out the exhaust gases after the exhaust pressure has fallen substantially to atmospheric pressure. With this arrangement the cycle of operation may be completed in two strokes, the number of cycles per minute doubled and the horse power almost doubled. Considerable experience with two- cycle engines is available from stationary practice and marine practice and indicates the possibility of increase of the power output from a cylinder of given size by from 60 to 80 per cent but with a falling off in efficiency of 20 per cent or more. There are two general methods of obtaining a precompressed charge, (1) by crankcase compression and (2) by the use of a separate compressor. With crankcase compression in a multi- cylinder engine, the crankcase must be divided to form a gas-tight compartment for each cylinder so that each piston on its down stroke may compress a charge which has been taken in during the up stroke. This arrangement is only possible in an engine with a single row of cylinders; it cannot be carried out in a Vee, W, or radial engine. It is simpler than the separate compressor but it limits the amount of charge which can be taken in to the volume sucked in during the up stroke of the piston and it will carry lubricating oil from the crankcase into the cylinder. A separate compressor should have a displacement 440 THE AIRPLANE ENGINE volume greater than that of the engine cylinder in order to send some scavenging air into the engine cylinder for the more com- plete clearing out of the exhaust gases and also to give a pressure at the beginning of compression which is fully up to or slightly above atmospheric pressure. The compression pressure required is about 5 Ib. per square inch. One double-acting air compressor would be required for two engine cylinders with discharge from the compressor direct to the cylinders. By the use of a receiver a larger compressor can be made to serve a larger number of cylinders. A centrifugal compressor would eliminate the need for a receiver. The two-cycle engine is just beginning to be used in airplane service. The principal difficulties to overcome are low efficiency and heat trouble. If the additional weight of fuel that has to be Annular Exhaust & Air Inkt Ports- Exhaust Pork tool ing Water Inlet FIG. 341. Junkers two-cycle solid-injection engine. carried for a long flight is equal to the saving in engine weight there is little advantage in the lighter weight engine except for short flights. The doubled number of explosions in the engine per unit of time increases the cylinder temperature and leads to serious heat difficulties with the piston. Exhaust valve troubles are eliminated by the use of exhaust ports uncovered by the piston in place of exhaust valves. There has been considerable experimental work in this field but with no practical results so far except in the case of the Junkers engine. The Junkers engine (Fig. 341) obtains high efficiency and eliminates heat trouble by departing entirely from the usual construction. There are six horizontal cylinders with their axes at right angles to the center line of the fuselage. The engine has two opposed pistons per cylinder and two crankshafts. All the pistons on each side of the engine are connected to a POTENTIAL DEVELOPMENTS 441 common crankshaft. The two crankshafts are geared together so as to make the two pistons of any one cylinder move in or out simultaneously. The combustion chamber is the space enclosed between the two pistons when they are on their inner dead center. Near the outer dead centers the pistons uncover cylinder ports, the exhaust ports (on the left) being uncovered first and the air admission ports (on the right) shortly afterwards. The propeller shaft carries the central gear with which the gears on the two crankshafts mesh; a blower is operated from the pro- peller shaft. There are no valves and no carburetors. When the pistons are on their outer dead centers, air from the blower passes through the cylinder from right to left and clears out the exhaust gases. This air is compressed, while the pistons make their inward strokes, to a pressure of 210 Ib. per square inch or more. Fuel is then injected into the combustion space through the nozzle at the bottom of the cylinder, is ignited by a spark plug immediately above it, and expands, driving the two pistons outward. The pistons are equipped with a special cooling device. They are made with a cavity which is partly filled with a heavy oil and then sealed. The oil is violently dashed backwards and forwards by the motion of the piston, absorbs heat from the piston head and carries it to the cooled piston sides. The efficiency of this method of cooling is shown by tests with thermo-elements which indicated a maximum temperature of the piston head of 350F. at maximum speeds and loads. The advantages of this method of construction appear to be manifold. The high compression gives high efficiency, which is helped by the small heat loss during explosion resulting from the smallness of the cooling surface of the combustion chamber. The excellent scavenging permits higher volumetric efficiency, which in conjunction with higher efficiency gives a higher m.e.p. than is obtainable in other types. The i.h.p. is consequently more than twice that obtained for the same piston displacement in four-cycle engines and the weight is reduced to 1.5 Ib. per horse power. The balancing of the reciprocating parts is practically perfect because the two pistons of each pair are at all times moving with equal accelerations in opposite directions; this condition is favorable to high engine speeds. The large size of the gas inlet and exhaust ports combined with positive admission' of the air and fuel permits also high volumetric efficiency at 442 THE AIRPLANE ENGINE high speeds; consequently this engine can be operated at higher speeds than the usual type. As the propeller is geared it can be run at its most favorable speed. A further feature of the engine is the great reduction of fire risk resulting from the direct dis- charge of the fuel into the engine cylinder; there is no explosive mixture outside the engine cylinder and less liability to fuel leaks. Moreover, a less volatile fuel can be burned. The principal apparent objection to the Junkers engine is the difficulty of accommodating an engine of its width in the fuselage. In the design shown in Fig, 341, the over-all width is 8J- times the stroke, or with a 6-in. stroke the width is 4 ft. 3 in. This dimen- sion is exceeded in large radial engines (see p. 195). The mechan- ical efficiency is low on account of the blower work and of the gearing losses; other Junkers engines, not adapted to airplane use, have given mechanical efficiencies of about 73 per cent. 1 Among the possible developments for airplane engines are Diesel engines, gas turbines and steam plants either turbine or reciprocating. The Diesel engine would be most advantage- ous in view of its higher efficiency and the safety and low cost of the fuels which it could employ. The difficulties in the way of its employment in airplanes in its present stage of development are its excessive weight and the large size of individual cylinder below which it has not been found practicable to go. In the modified form of the Hvid and similar engines, smaller size cylinders become practicable but the weight is still excessive. It is by no means certain that it will be found practicable to operate this cycle successfully at the high speeds necessary for airplane use. Gas turbines have been under active development for over fifteen years but the difficulties inherent in them have not as yet been overcome without sacrificing their potential efficiencies. The principal troubles are those resulting from the high temper- atures to which the combustion chamber, nozzles, and buckets are subjected. When these temperatures are reduced, by inject- ing water or excess air, the efficiencies fall off. Moreover, the efficiencies are low unless the air is precompressed and as centrif- ugal compressors seldom have efficiencies above about 60 per cent, a large part of the power developed in the turbine is utilized in driving the compressor. Over-all thermal efficiencies are usually about 5 per cent, although an unsubstantiated value of 1 SCOTT, Jour. Soc. AuL Eng., 1917. POTENTIAL DEVELOPMENTS 443 20 per cent has been claimed for a 1,000-h.p. unit. The gas turbine, if applied to airplane propulsion, would have to be geared down, probably with double reduction gear. Its simplicity and light weight have attracted many inventors, but there are no indications that it is ever likely to become practically available. Steam-power plants have the great advantage over internal combustion engines of maintaining their power at all altitudes. Both turbines and reciprocating engines of light weight are fully developed and can be regarded as immediately available for airplane use. The difficulties arise in connection with the boiler and oil burner. For efficiency the steam must be generated at high pressure and with high superheat, but no construction of boiler is known which does not entail weights which would be excessive for aircraft. A satisfactory kerosene (or other fuel) burner for operation with high rates of combustion in small space would also have to be developed; the fire risk from such apparatus would probably be considerable. And finally, the efficiency of the best steam plants is not nearly so high as that of existing airplane engines. It seems very unlikely that steam plants will ever be employed in aircraft. INDEX A. B. C. engines, (table), 194 Acceleration, in carburetors, 271 force, (see Inertia Force} Acetylene, as fuel, properties of, 225 Aerofoil, (see Wings) Air compression, power absorbed in, 397 temperature rise in, 395, 400 cooling, 344 cycle efficiency, (def. and table), 17 densities, at altitudes, (curves), 366 flow, pulsating, 254 'flow of, orifice coefficients for, 243 through venturi tube, 246- 250 fuel ratio, determination of strength of, 242 influence of charge dilution on optimum value of, 262 influence of, on engine per- formance, 260 optimum values of, 261 variation with air density of, 259 intakes, 416 pumps, 292 starting, (see also Starting'), 429 temperature, at altitude, (curves), 365 influence of, on capacity, 35 on engine power, 33, 35 on thermal efficiency, 35 weight flow of, (chart), 249 Alcogas, 242 Alcohol, as fuel, properties of, 224 effect on detonating pressure, 239 mixtures with gasoline, 242 Alloys (see Steels, and Aluminum Alloys) Altitude, air density at, (curves), 366 control of carburetor, 268 effect of, on radiator, 365 influence of, on engine power, 386-389 temperatures at, (curves), 365 Aluminum alloys, 118 for pistons, 135 American engines, description of, 80-98 Austro -Daimler engine, cylinder of, 131 dimensions of, 73, 124 inertia forces and bearing loads, 59 water pumps of, 372 weights of parts, 78 B B. R. 2 engine, description of, 190 Balancing, devices for, 58 in radial engines, 206 in rotary engines, 206 of reciprocating parts, 55 of rotating parts, 54 Ball and ball carburetor, description of, 283 Battery ignition systems, (see Igni- tion Systems) Battery, of Liberty engine, 316 Basse-Selve carburetor, description of, 284 engine, compression release of, 428 cylinder dimensions of, 124 dimensions of, 73 inertia forces and bearing loads of, 59 lubricating system of, 341 oil pumps of, 341 valve gear of, 171 valves of, 157 . weights of parts, 78 445 446 INDEX Baume" scale, conversion table for, 219 Bayerische Motoren Werke carbu- retor, description of, 285 Bearings, 144 ball and roller, in radial engines, 207 crankshaft, dimensions of, 74 friction work of, 328 loads on, 59, 76, 327 oil grooving of, 337 Benz engine, compression release of, 427 connecting rods of, 130, 141 cylinder of, 124, 129 description of, 110 dimensions of, 73 fuel pump of. 294 intake manifold of, 420 lubrication system of, 335 oil pump of, 336 performance curves of, 111 piston of, 136 propeller hub of, 148 valve gear of, 172 weights of parts, 78 Benzene, (see Benzol) Benzol, as fuel, properties of, 223 Bijur starting system, 432 Bosch magneto, 302 Brake mean effective pressure, (def.), 25 Bugatti engine, description of, 93-98 performance curves of, 98 spark adjustment in, 306 water pumps of, 372 C Cams, 163 followers for, 164 Camshafts, dimensions of, 72 Capacity of engine, 26, 29 influence of fuels on, 240 variation with air temperature, 33, 35 compression ratio, 37, 392 engine speed, 37, 392 jacket-water temperature, 37 mixture strength, 33 Carbon dioxide, dissociation of, 20 Carburetors, 245-289 acceleration in, 271 air discharge coefficients of, 252 altimetric compensation of, 267 control of, 268 atomization in, 271 construction of, 272 dimensions of, 74 float arrangements in, 272 flooding of, 272 idling device in, 271 intakes for, 272 mixture characteristics of, 258 performance of, 264 pressure drop in, 253 strainers for, 290 viscous flow type, 267 weights of, 78 Castor oil, 333 Central power plants, 384 Choke, (see also Venturi Tube), 250 Christensen system of starting, 431 Claudel carburetor, description of, 276 Clerget engine, description of, 189 effect of compression ratio on horsepower of, 393 performance curves of, 190 Combustion, air required for, 228 effect of turbulence on rate of, 235 higher and lower heats of, (def.), 214 products of, (table), 216 velocity of propagation of, 232 Compound engines, 438 Compression, ratio of, (def.), 14 variable, 438 pressures, effect of alcohol on, 239 effect of toluene on, 239 maximum allowable, 236, 237 variation with compression ratio, 16 engine speed, 16 ratio, 66, 72 influence of, on capacity, 37 on efficiency, 434 engine power at high altrtudes, 392 INDEX 447 Compression, release, 427 Compressors, (see Air Compression} Connecting rod, rods, 138 assembly, articulated, 203 dimensions of, 74 for rotary and radial engines, 203 materials for, 116 slipper assembly, 204 stresses in articulated, 143 weights of, 76, 78 Cooling fins, 345 systems, 344 anti-freeze solutions for, 375 piping for, 375 pumps for, (see Water Pumps) typical examples of, 377 water for, 375 Cosmos engines, (table), 195 induction chamber of, 196 "Jupiter," balancing of, 207 performance curves of, 196 Counterweights, 146 Crankcases, 148 cooling of, 150 weights of, (table), 78 Crank pins, loads on, 76 Crankshafts, 143-146 balancing of, 143 dimensions of, 74 materials for, 115 of radial engines, 211 strength of, 146 weight of, 78 Curtiss engines, crankshaft balanc- ing of, 144 cylinders of, 124, 127 description of, 86-90 lubricating systems of, 334 performance curves of, 90, 92 type K, 86 Cylinders, air-cooled, 200, 345 materials for, 349 temperature of, 347] arrangements, size and propor- tions of, 61 attachment to crank case, 125 liners, material of, 115 lubrication of, (see Lubrication) Cylinders, offset of, 48 thickness of, 123 types of construction, 122 g weights of, 78 D Detonation, 236 Diesel engines, 442 Dilution of charge, influence of compression pressure on, 263 influence on engine perform- ance, 393, 434 Dimensions, engine, (table), 65 of American and German engines, (table), 72 overall, of engines, (table), 79 Dissociation, of carbon dioxide, 20 of water vapor, 20 Distillation curves, 241 Distributor, Bosch, 304 Dixie, 305 Dixie magneto, 302 Double-rotary engines, 176, 191 Drag, of wing, 2 Duralumin, 121 E Engine speed, influence of, on capacity, 37 English engines, descriptions of, 98-107 Ether, as fuel, properties of, 226 Exhaust gas, composition of, 244 turbines, (see also Supercharg- ing), 398, 438 manifolds, (see Manifolds) mufflers, (see Mufflers) Explosion phenomena, (see also Fuels, Combustion), 231 intervals, in multicylinder engines, 63 limits, effect of carbon dioxide dilution on, 262 Explosive mixtures, 212 properties of, (table), 216 wave, 236 448 INDEX Fiat engine, description of, 107 lubrication system of, 335 valves of, 174 Firing order, 75, 325 Flash point, of oils, 330 Flight, power available for, 8 power required for, 1 Friction, laws of, 327 loss, at piston, 24 in engine, 24 Fuel, fuels, 212-226 air ratio, (see Air-fuel Ratio) air required for combustion of, (table), 216 detonating compression pressures of, (table), 237 explosive limits of air-fuel mix- tures, (table), 232 flow through jets, 254-258 heats of combustion of, (table), 217 ignition temperatures of, 233 influence on capacity of, 240 influence of temperature on fluidity, (curves), 257 hydrocarbon, classification of, 213 minimum vaporization tempera- ture of, 231 properties of, 212, 215 pumps, 289, 292 weights of, 78 specific heats of, 229 specific volumes of saturated vapors of, 229 systems, 289-294 for supercharging engine, 412 tanks, 290 temperature drop due to vapori- zation of, 230 vapor pressures of, 229 viscosity of, 257 G Gas turbines, (see also Exhaust Gas Turbines), 442 velocity, through valves, 152 Gasoline, (see also Fuel), 217-223 blended casing-head, 218 calorific value of, 223, 241 cracked, (synthetic), 218 distillation curves of, 221 mixtures with alcohol, 242 specifications for, 219 "straight" refinery, 218 tests for, 222 volatility of, 220 Geared propeller drives, 378-384 Gears, heating of, 384 pressure between teeth of, 383 stresses in, 383 Gnome engine, description of, 185 torque of, 53 Gudgeon pin, (see Piston Pin) Hall-Scott engine, cylinder dimen- sions of, 124 description of, 90 exhaust manifold of, 420 lubrication system of, 335 performance curves of, 95 Heat balance, of airplane engine, 38 dissipation, from air-cooled cylin- ders, 345 transfer, in radiator core, 352 Hispano-Suiza engine, air pump of, 292 cylinder of, 126 description of, 82 dimensions of, 73 effect of compression ratio on horsepower of, 393 hand starting mechanism of, 430 influence of air density on performance, 388 lubrication system of, 334 performance curves of, 86 propeller gears of, 378 , propeller hub of, 147 starting torque of, 427 test results of, 31 valves of, 173 weights of parts of, 78 INDEX 449 Horse power (see also Capacity) required for flight, 4 Hydrogen, as fuel, properties of, 225 Idling device, in carburetors, 271 Ignition, 295-326 assemblies, weights of, 78 spark advance, 306 systems, battery, 297, 312 comparison of, 324 cycle of operations in, 307 self-sustaining battery, 313 temperature of, 233 Indicated thermal efficiency, maxi- mum obtainable, 237 Indicator cards, actual, 15 negative pumping loop, 23 theoretical, 13 Induction coil, 296 Inertia factors, (table), 42 forces, in Austro-Daimler, Basse"- Selve and Liberty engines, 59 in radial engines, 52 primary, 55 secondary, 55 of reciprocating parts, 41 Intake manifolds, (see Manifolds) Jacket-water temperature, influence of, on capacity, 37 Jets, discharge coefficients of, 255 flow through, 254 Junkers two-cycle fuel-injection engine, 440 K Kerosene, as fuel in airplane engines, 436 Lanchester balancer, 58 LeRhone carburetor, description of, 288 29 LeRhone engine, description of, 185 effect of compression ratio on horse power, 393 oil pumps of, 342 performance curves of, 187 Liberty engine, air intake for, 417 battery ignition system of, 313 breaker mechanism of, 318 centrifugal oil cleaner of, 333 connecting rods of, 141 cylinder of, 124, 130 description of, 80 dimensions of, (table), 72 distributors of, 318 electric starting system for, 432 generator, 315 heat balance, 39 indicator card, 40 inertia forces and bearing loads, 59 influence of air density on performance of, 388 intake manifolds of, 419 lubrication system of, 334 method of starting, 425 oil pumps of, 339 performance curves of, 31, 83, 239 piston of, 138 starting torque of, 427 torque of, 43 valve action of, (curves), 168 valve of, 172 water pumps of, 371 weights of parts, (table), 78 Lift, on a wing, 1 Lubricating oils, properties of, (table), 331 reclaiming of, 332 specifications for, 330 tests of, 330 viscosity of, 329 Lubrication, 327-343 methods of, 333 of cylinders, 328 of radial engines, 211 oil consumption for, 338 450 INDEX M Magneto, 297-307 armature flux, 308 typical constants for, 307 for unequal firing intervals, 300 inductor type, 300 secondary voltage in, 309 speed of, 306 starting, 425 Manifolds, exhaust, 421 ejector effect in, 422 intake, 417-421 influence of, on engine perform- ance, 436 preheating of, 419 pressure drop in, 28 temperature change in, 418 Master carburetor, description of, 283 Materials, for special engine parts (table), 118 properties of, (table), 114 Maybach carburetor, description of, 286 engine, cylinder of, 129 dimensions of, 124 description of, 113 dimensions of, (table), 73 . fuel pump of, 293 lubrication system of, 336 performance curves of, 113 pistons of, 136 starting mechanism of, 426 valve action of, (curves), 167 valve gear of, 166 water pumps of, 374 weights of parts, (table) 78 Mechanical efficiency, 23 Mean effective pressure, (table) 67 brake, 25 maximum obtainable, 237, 238 Mercedes engine, air pumps of, 292 dimensions of (table), 73 weights of parts (table), 78 Miller carburetor, description of, 282 Mixture strength (see also Air-fuel Ratio) influence of, on engine perform- ance, 22, 32 Moss turbo-supercharger, 411 Mufflers,, exhaust gas, 422 X Napier "Lion, " connecting rods of, 142 description of, 103 lubrication system of, 335 performance curves of, 106 Odier portable starter, 433 Offset cylinders, 48 Oil pumps, 336, 338 performance curves of, 340 weights of (table), 78 Oil sumps, 150 Oil tanks, 343 Oversized engines, 390-394 Orifice, discharge coefficients for sharp-edged, 243 Otto cycle, 13 efficiency with variable specific heat, 19, 21 Oxygen, use in engine at altitudes, 413 Packard engine, cylinder head of, 131 description of, 80 dimensions of (table), 72 performance curves of, 84 weights of parts of (table), 78 Parasite resistance, 3 Performance curves, Benz engine, 111 Bugatti engine, 98 Clerget engine, 190 Cosmos "Jupiter" engine, 196 Curtiss engine, 90, 92 Hall-Scott engine, 98 INDEX 451 Performance curves, Hispano-Suiza engine, 86 LeRhone engine, 187 Liberty engine, 83 Maybach engine, 113 Napier "Lion" engine, 106 Packard engine, 84 Salmson engine, 199 Siddeley "Puma" engine, 106 Rolls-Royce engine, 102 Pipes, for fuel system, 290 Piston, Pistons, dimensions of, (table), 72 displacement, per horsepower, (table), 67 divided-skirt, 135 friction of, 133 material of, 132 pin, 138 dimensions of, (table), 74 loads on, (table), 76 rings, 137 dimension of, 74 slap of, 134. slipper, 133 speed, (table), 66 weights of, (table), 76, 78, 166 working temperatures of, 132 Pitch, of propeller, (see Propeller Pitch of) Power, (see Horsepower and Capacity) available, for flight, 8 required, for flight, 1 Priming, of engine, 423 Pumps, fuel, 289, 292 Pressure, Pressures, in ideal (air) cycle, 14 drop, in intake manifold, 28 past valves, 152 mean effective, 25 on bearings, 144 on crankpin, 43 on piston, 40 Propeller, 5 coefficients, 7 efficiency of, 6 geared, (see Geared Propeller Drives) hubs, 147 Propeller hubs, weights of, (table), 78 pitch, 6 pitch ratio, 6 slip, 6 speed, (table), 66 thrust, 6 thrust horse power of, 6 torque, 6 torque horse power of, 6 R Radial engines, 176, 193 air-cooled, dimensions of, (table), 66 ball and roller crankpin bearings in, 207 crankshaft of, 211 details of, 200 firing order of, 178 lubrication of, 211 number of cylinders of, 178 overall dimensions of, (table), 79 unbalanced forces in, 56 valve operation in, 211 water-cooled, dimensions of, (table), 68 Radiators, 351-370 constants of, (table), 362 construction of complete, 368 cores of, 351 dimensions of, 359 effect of position on performance of, 363 figure of merit of, 355, (table), 360 head resistance of, 354, 356, (table), 360 heat transfer in, 352, 354, (table), 360 horse power absorbed by, 354, (table), 360 limiting temperatures in, 363 masking of, 366, 368 mass flow of air in, (def.), 352 obstructed, 352 on lighter-than-air machines, 367 performance of, 358, 360 at altitudes, 365 resistance to water flow in, 364 selection of, 356 452 INDEX Radiators, size of, 361 water flow through, 363 yawing of, 368 Rateau supercharger, 410 Reduction gearing, (see Geared Pro- peller Drives) Renault engine, connecting rods of, 142 cylinder dimensions of, 124 propeller gears for, 378 Resistance, parasite, 3 of plane, 4 of wing, 2 Revolutions, of typical engines, (table), 66 Ricardo system, of supercharging, 401 Rocker arms, 171 Rolls-Royce engine, description of, 98 dimensions of, (table), 101 performance curves of, 102 propeller gears for, 380, 382 Rotary engines, 176, 179-193 balance of, 57 details of, 200 dimensions of, (table), 66 firing order of, 178 number of cylinders of, 178 overall dimensions of (table), 79 torque of, 50 S Salmson engine, description of, 198 performance curves of, 199 Saturation temperature of air-fuel mixtures, (table), 231 Side thrust, 48 Siddeley "Puma" engine, 11 connecting rod of, 140 description of, 105 performance curves of, 106 valves of, 173 Siemens-Halske double-rotary en- gine, 192 Slip, of propeller, 6 Spark advance, (see Ignition) gaps, 310 plugs, 319-324 causes of failure of, 320 Spark plugs, construction of, 322 cracking of insulators of, 321 dimensions of, 76 location and number of, 324 Sparking voltages, 311 Specific heats, at constant volume, (table), 20 Springs, safe loads and deflections of, 170 Starting, 424-433 air compressors for, 430 by hand, 424 Christensen system, 431 compressed air for, 429 compression release for, 427 electric systems for, 432 magneto, 425 mechanisms, dog clutch for, 428 integral, 429 safety devices for, 429 Motor-Compressor Company sys- tem, 430 portable crankers for, 433 torque, 426 use of hydrogen for, 425 Steam-power plants, 443 Steels, 116-117 . for exhaust valves, 160 Stewart- Warner carburetor, air dis- charge coefficients of, 252 mixture characteristics of, 258 performance curves of, 265 Storage cells, (see also Battery), 312 Strainers, for fuel systems, 290 Stroke-bore ratios, 66, 72 Stromberg carburetor, description of, 278 mixture characteristics of, (curves), 258 performance curves of, 266 Sturtevant engine, cylinder of, 128 supercharger for, 404 Supercharged engines, 394-413 explosion relief valve for, 408 fuel supply system for, 412 power developed by, 394 relief valve of, 413 throttle valve for, 408 INDEX 453 Superchargers, (see also Air Com- pression) t 397 Brown-Boveri, 407 coupling for, 407 exhaust gas turbine, 398, 408 Moss turbo-, 411 Rateau, 410 Schwade, 406 Sturtevant, 404 Supercharging, 368-415 centrifugal compressors for, 403 efficiency of, compressor, 400 exhaust gas turbine, 400 gearing of compressors for, 403 influence on plane performance, 415 methods, 401 multi-stage compressors for, 404 reciprocating compressors for, 403 Ricardo system of, 401 Roots positive blower for, 403 Tangential factors, (table), 44 Tanks, fuel, 290 oil, 343 Tappet clearance, adjustment of, 172 Temperatures, in ideal (air) cycle, 14 of jacket-water, influence on en- gine capacity, 37J Test results, correction to standard conditions, 32 Thermal efficiency, 25 effect of mixture strength on, 22 maximum observed, 21 variation with air temperature, 35 mixture strength, 33 throttling, 36 Throttling, influence of, on thermal efficiency, 36 Thrust, of propeller, 1, 6 Timing (see Valve Timing} Toluene, effect on detonating tem- perature, (table), 239 value (def.), 238 Torque, at engine crankshaft, 43 Torque, in rotary engines, 50 of engine, 26, 67 at starting, 426 of propeller, 6 ratio of maximum to mean, 46 variation with number of cylin- ders (table), 47 Torsional vibration, of crank-shaft, 146 Turbines, exhaust gas, (see Exhaust Gas Turbines, Super- charging) Turbo-superchargers, (see Super- chargers) Turbulence, 235 Turning moment, (see also Torque), 40 Two-cycle engines, 439 U Unbalanced forces, magnitude of, 56 neutralizing of, 58 periodic, effects of, 57 Valve, Valves, automatic throttle, 391 causes of failure of, 160 effect of lift and diameter on engine capacity, 158 exhaust, 156 dimensions of, 66, 72 temperature of, in air-cooled cylinder, 349 gears, 162 weights of, (table), 78 inlet, dimensions of, 66, 72 number per cylinder, 66, 72 lift of, 151 materials of, 159, 161 operation, in radial engines, 211 ports, 125 pressure drop past, 152 seats, 125 springs, 168 retainers for, 170 tension of, (table), 72 stem guides, 125 temperatures of, 159 timing of, 172, 173, 175 454 INDEX Valves, weights of, (table), 76 Variable-compression engine, 438 stroke engine, 437 Vee engines, angle of, 64 connecting rods of, 139 dimensions, (table), 70 overall dimensions of, (table), 79 unbalanced forces in, 56 Venturi tube, 251 discharge coefficients of, 252 Vertical engines, dimensions of, (table), 68 overall dimensions of (table), 79 unbalanced forces in, 56 Vibration, (see Torsional Vibration) Viscosimeter, 329 Viscosity, measurement of, 329 of fuels, 257 of oils, 329 Volatility, (see also Distillation), 220 Volumetric efficiency, 27, 28 W W engines, arrangements of, 65 dimensions of, (table), 70 Water cooling, 350 pumps, 370 horse power required for, 370 performance curves of, 373, 374 weights of, (table), 78 vapor, dissociation of, 20 Weights, of engines, 60, 67, 78 of engine parts, (table), 78 of reciprocating parts, (table), 76 of rotating parts, (table), 76 of water in engine, (table), 78 Wings, characteristics of, 2 Wiring systems, 325 Wright engine, description of, 84 Wrist pin, (see Piston pin) Zeitlen engine, description of, 437 Zenith carburetor, air discharge coefficients of, 252 description of, 272 mixture characteristics of, 258 performance curves of, 264 UNIVERSITY OF CALIFORNIA LIBRARY BERKELEY Return to desk from which borrowed. on the last date stamped below, LD 21-100m-7,'52(A2528sl6)476 THE UNIVERSITY OF CALIFORNIA LIBRARY