Following Combustion in the Gasoline Engine by Chemical Means

Following Combustion in the Gasoline Engine by Chemical Means. Llyod Withrow ... Engine Studies of Preknock Reactions ... Flame Temperatures in Engine...
1 downloads 0 Views 963KB Size
September, 1930

I S D U S T R I A L AND ESGIAVEERIKGCHE-PIISTRY

&The alpha-cellulose contents of the residual woods of the sodium carbonate cooks were generally lower than those of the corresponding water series, and decreased with increasing concentration of the salt. 9-Sodium carbonate in 20 or 40 per cent concentration (based on the weight of' oven-dry wood) cannot be regarded as an inert ingredient in a cooking liquor.

94 5

Literature Cited (1) Aronovsky and Gortner, IND. ENG.CHEM.,92, 264 (1930). (2) Bray and Andrews, Poper Trade J . , 76, NO 19, 49 (1923). (3) Fromherz, 2. physiol. Chem., 50, 209 (1906). (4) Griffin, J. Am. Chem. Soc., 2 4 , 235 (1902). ( 5 ) Schorger, IND. ENG, CKBM., 16, 141 (1924). (6) Schorger, "Chemistry of Cellulose and Wood," p. 405, McGraw-Hill, 1926. (7) Wells, Grabow, Staid], and Bray, Paper Trade J . , 76, No. 24, 49 (1923).

Following Combustion in the Gasoline Engine by Chemical Means' Lloyd Withrow, W. G. Lovell, and T. A. Boyd GENERAL .MOTORSRESEARCH LABORATORIES, DETROIT,-MICK.

Measurements have been made of the oxygen concentration in gases withdrawn from the cylinder of a gasoline engine with a new and improved sampling valve which was located a t different places in the combustion chamber and opened a t different times during the combustion of the charge. Under the conditions defined in this investigation the combustion process In the gasoline engine consists of a narrow combustion wave which proceeds from the spark plug through the combustion chamber a t a finite rate. The combustion zone travels a t a greater rate and follows a different type of acceleration curve through the middle portion of the combustion chamber than along the side walls of the combustion chamber. Over the range investigated the average speed of the . . . . . a

I

N THIS investigation an auxiliary sampling valve which

could be opened at any desired point in the cycle was used to obtain information concerning the chemical reactions between gasoline and air in the combustion chamber of a gasoline engine. This method of experimentation appears to have been developed first by Brooks (1 ), who used it in connection with an oil injection engine. A sampling valve of a design similar to that employed by Brooks was developed independently in this laboratory by Lovell, Coleman, and Boyd @),who used it for studying the rate of burning of the charge in a gasoline engine. Note-In the Bulletin of the American Petroleum Institutefcr 1927, No. 20, R. A. Millikan and W. M. Zaikowsky announced their intention of using a sampling valve in connection with Project 11, "Analysis nf the Gradual Oxidation Prior t o Ignition of Fuels in Internal Combustion Engines and To date, however, the the Relation of Such Oxidation t o Detonation." authors of this paper have not seen a detailed description of Ihis apparatus. Ricardo and Thornycroft, World Power Conference (London), Vol. 111, p. 662 (1928),report the discovery of partial oxidation of fuel in a petrol engine prior to ignition by means of a sampling valve, but. these writers do not describe their sampling apparatus.

I n this latter study the rate of burning of the charge in the combustion chamber was measured by making a chemical analysis of samples which were emitted from the sampling valve when it was adjusted to be open for a short interval a t various times during which the actual combustion process mas going on. The results of these analyses were plotted against the various angles of revolution at which the valve was adjusted to open and curves were drawn. The slopes of these curves apparently gave some information about the rate of 1 Received June 7, 1930. To be presented before the Division of Petroleum Chemistry a t the 80th Meeting of the American Chemical Society, Cincinnati, Ohio, September 8 t o 12, 1930.

combustion zone increases with the engine speed. The progress of the combustion zone is unaffected by a change in spark timing or by the addition of sufficient lead tetraethyl to the fuel to stop detonation until after it has traveled the greater portion of the distance across the combustion chamber. The disturbance which is known as the knock or detonation is confined to that part of the charge which burns last. With respect to events in the latter portion of the combustion space where detonation occurs when it is present, definite conclusions about the differences between the characteristics of normal and detonating combustion must await the results of further experiments which are under way.

....... decrease of the concentration of oxygen and the rates of increase of the concentrations of hydrogen, carbon monoxide, carbon dioxide, and water while combustion was taking place in the combustion chamber. From the shapes of these curves it was concluded that the combustion reaction was of considerable duration and that after some sort of a flame spread the entire charge to the rear of the flame front burned more or less homogeneously. These results appeared very promising and indicated that the sampling valve could be a powerful tool for this type of investigation. Since the publication of this earlier work, however, it has been discovered that the sampling apparatus described therein was not well suited to such an investigation. By means of an electrical valve-interval-deterinator, which was afterwards incorporated as an essential part of the sampling apparatus and which is described below, it was found, first, that the sampling valve opened from 5 to 10 degrees of revolution of the crankshaft after the angle of revolution at which it was set to open; and second, that the sampling valve remained open for 20 or more degrees of revolution of the crankshaft instead of 2 degrees. Since the performance of the sampling valve must be considered in interpreting the results of analyses of samples taken from the engine, considerable time and effort have been expended to develop a sampling mechanism which would permit proper correction for these errors. After a careful analysis of the types of mechanism which would open and close a sampling valve in as short an interval as possible and still remove a suitable portion of the gas from the combustion chamber during each compression cycle of the engine, a new sampling valve and a new mechanical device for actuating this valve were designed.

946

INDUSTRIAL AXD ENGYNEERIXG CHEMISTRY

With this improved sampling apparatus and a new engine, samples have been removed at several different positions in the combustion chamber and a t various times after ignition while the engine was running with gasoline as fuel. The results of the analyses of these samples for oxygen were plotted against the respective angles of revolution at which the samples were removed, and the conclusions about the nature of the combustion reaction in the gasoline engine which have been derived therefrom, RS well as in what respects they differ from the previous conclusions, are presmted in this paper.

Pieuro 1-Ensine and Sampiing A P I I B ~ ~Used ~ Y I In Study of Cornbusrion

Apparatus

These experiments were made with the single-cylinder engine and auxiliary apparatus shown in Figure 1. This engine, which was built in thB laboratory, was of the watercooled ell-head type, with 2'/&ch bore and 43/4-in~hstroke. The design of the cylinder heads used in all these experiments &-as such that the compression ratio was fixed a t 5.0:l. Figure 2 shows the general details of construction of the engine and the sampling apparatus. The engine was connected to a dynamometer, which absorbed the load and thereby controlled the engine speed. By means of an evaporative cooling system which provided for ample water circulation about the cylinder block and cylinder head, the water temperature was kept at 100' C. The spark timing was observed by the flash of a neon tube, attached to the flywheel, as it passed a stationary spark protractor which was insulated from the engine and connected to the high-tension wire leading to the spark plug. A Stromherg carburetor of the horizontal type with a 1-inch throat TZW wed with the throttle wide open in these experiments. The carburetor was attached to the engine by means of an adapter consisting of a piece of steel tubing about 3 inches long with a flange of the proper size at either end. No heat \vas applied t o the mixture before it entered the combustion chamber. The exhaust gases passed into a special ventilating system which is provided in this laboratory to carry the fumes out of the building. SAMPLIXGA P P m T u t i T h e power for operating the sampling valve was transmitted from the camshaft to the t.op of the engine by means of the vertical shaft, A (Figure 2), which was rotated by the spiral gear, B, on one end of the camshaft. Near the top of A was another spiral gear, C, for rotating the cross shaft, F. F was held in place by the cross-

Vol. 22, No. 9

shaft bracket, R, which carried a hearing for the cross shaft F and was fastened to the top of the engine. In the latest design the cross shaft F carried the rotor shown in section A-A in Figure 2, but while most of these experiments were being performed it carried a cam for actuating the sampling valve by means of a rocker a m aa shown in Figure 3. When the rotor is used, the valve stem is moved downward by the impact of the hammer, G, upon the rocker arm, H , and when the cam is used the valve stem is moved downward by the rocker arm, H , in Figure 3. The valve is closed by the spring, 0, which is made of heavy spring wire with one active coil. The purpose of this design waa t.o give this spring as high a natural period as was convenient for this device. The electrical valve-intervaldeterminator consisted essentially of a simple electrical circuit from a Gvolt battery with one terminal attached to a spark coil and the other terminal attached to the insulated micrometer screw, I. (Figure 2) This screw projected through a hole in the rocker a m directly above and in the same vertical plane as the valve stem and made contact with the top of the valve stem without contacting the rocker arm at any point. From the insulated micrometer screw the current passed to the engine by way of the valve stem and then to the rotating contact, J , at the top of the vertical shaft, A, which made a11electrical connection with the copper strip, K , during each revolution of the camshaft. From K the current passed around to the binding post, L,and from there back to the Gvolt battery, passing on the way through a spark coil: the secondary of which led to a small neon tube. To operate the electrical valve-interval-deterinator tlir insulated contact strip K was rotated about the vertical shaft, A , by means of the handle, N , to an angle a t which it was known that the valve was not opening. The insulated microineter screw, I , was then screwed down until it just contacted the top of the valve stem, as was indicated by the appearance of regular flashes from the neon tube for each revolution of the vertical shaft, A . The position of K was then varied until the angle was found at which the circuit was broken by the impact of the rocker arm upon the valve stem, thereby indicating the opening of the valve. The readings were made quickly and easily aa the scale opposite K was calibrated in terms of degrees of revolution of the camshaft. With this device it was possible to determine accurately both the t i e and the duration of valve opening, It was found that these factors were constant for a given engine speed when either the cam or the rotor was used to actuate the valve. At engine speeds from 500 to 1200 r. p. m. the duration of the valve opening ranged from 6 to 12 degrees of revolution of the crankshaft when the cam was used to actuate the sampling valve and from 2 to 4 degrees of revolution when the rotor was used. The time of opening of the valve was varied by loosening the binding nut, l', and rotating the cani or rotor ahout the cross shaft, P, to the desired angle. By releasing the clutch shown a t S the timing of the valve could be changed without stopping the engine and disturbing the conditions of engine operation. This clutch was of the simple dog-type which would engage in one position only. The male portion was the sliding member and the female portion was attached directly to the spiral gear, B. An important feature of the sampling apparatus was its flexibility with respect to the point of location of the sampling valve in the combustion chamber. When the cross-shaft bracket, R, and the vertical-shaft bracket, P, were loosened, the cross shaft F waa free to slide back and forth in the horizontal plane, while at the same time it could be rotated about the vertical shaft, A, in the horizontal plane. As a result, the mechanism for actuating the sampling valve could be

September, 1930

947

I X D U S T R I S L A S D ENGISEERING CHEMISTRY

G

i 4

w

P

Figure 2-Sectional

View of Engine and Sampling Apparatus

adjusted to any position desired and there fastened to the cylinder head. The sampling valve itself was designed to perniit efficient cooling of the sample as soon as it had passed out of the combustion chamber. This was accomplished by allowing the gas sample to expand, immediately after passing by the valve seat, into six small water-cooled copper tubes which led to the valve port. The entire valve was cooled by a stream of cold water which entered the valve a t X and left at Y . (Figure 2) I n order to avoid the hot spots arising from obstructions in the water jacket and so to insure normal cooling of the combustion chamber, a separate cylinder head was provided for each position of the sampling valve. With one exception there were in each cylinder head only two 7/s-inch holes, one to accommodate the spark plug and the other the sampling valve. I n one case a 12-mm. metric spark plug was substituted for the 7/s-inch (22-mm.) spark plug in order not to decrease the capacity of the water jacket. Procedure

I n making determinations of the type which are described herein the accuracy of the control of the conditions of engine operation determines in a large measure the value and the consistency of the results. Consequently, before any samples were removed from the combustion chamber, the engine was permitted to run at least hour to allow the engine conditions to become constant. The mixture ratio was then adjusted to maximum power, which in this engine corresponded to 80 per cent of the air required for complete combustion.

Then after the timing of the spark had been adjusted for maximum power the engine was ready for the sampling operations. Since the sampling apparatus was adjusted to remove approximately 1 cc. of gas per explosion, a short time was required for the removal of each sample, and as a result each sample represented an average of several explosions. The gas was collected directly in a modified O r s a t apparatus and I analyzed immediately for carbon dioxide and oxygen. When samples had been removed f r o m t h e combustion chamber at several diferent times after ignition, the mixture ratio was rechecked by an analysis of the gas in the regular exhaust for d and Rocker Arm Used to' carbon dioxide and car- Figure 3-Cam Actuate Sampling Valve bon monoxide. If it was found that the mixture ratio had changed during the course of the run, the experiment was repeated. ~

Results

The results of these determinations are shown graphically in Figures 4 and 5 and in Figures 7 to 11,inclusive, and tabulated in Tables I to IV, inclusive. I n Figures 4 and 5 the per-

918

I-VDVSTRIAL A S D EXGILVEERISG CHEMISTRY

T’ol. 22, s o . 9

centages of oxygen by volume which were found in the samples are plotted against the respective angles of revolution of the crankshaft a t which the samples were removed. One of the curves shows the time when the sampling valve opens and the other the time when it closes. I n this study the time when the piston was at top dead center has been taken as a reference point and is marked zero on the figures. The data in Figure 4 were obtained with the sampling valve 2 inches (5.1 cm.) from the spark plug and equidistant from the side

that the main combustion reaction completes itself within a comparatively narrow zone. Since the sampling valve thus offers means of following the combustion zone across the combustion chamber, an attempt was made to do this by locating the sampling valve at various points in the roof of the combustion chamber and determining the angle at which the oxygen concentration began to decrease rapidly. This angle of revolution was assumed to be the time of arrival of the combustion process at any given point, even though from 4 to 10 degrees of revolution elapsed before the oxygen had completely disappeared. This assumption is p a r t i a l l y justified when it is noted that in each case the sampling valve was mounted in the cylinder head in such a manner that the valve itself was flush with the roof of the combustion chamber, and it is probable that the gases midway between the roof and the floor of the combusFigure 4--Sampling Valve 2 Inches (5.1 cm.) Figure 5-SamP1ing Valve 3 Inches (7.6 Cm.1 tion chamber burn before the gases from Spark Plug from Spark Plug Percentage of Oxygen in Samples Withdrawn from Engine at Various T i m e s after Ignition near the relatively cold sampling Engine speed.. . ,600 r. p. m. valve. Spark.. . . . . . . . .I6 degrees before top dead center Figure 6 shows a horizontal cross Mixture ratio.. .80 per cent of theoretical air Full throttle section of the combustion chamber o Fuel =gasoline (engine knocking) r Fuel =gasoline + 5.4 cc. lead tetraethyl per gallon (engine not knocking) and the various Dositions at which the time of the diiappearance of the walls of the combustion chamber, and the data in Figure 5 oxygen was determined. The results of these determinations were obtained with the sampling valve 3 inches (7.6 cm.) from are shown in Tables I to IV and in Figures 7, 8, and 9. The data in Figure 7 show the effect of engine speed upon the spark plug and equidistant from the side walls of the combustion chamber. I n each case the spark plug was located a t the progress of the combustion zone through the middle the end of the combustion chamber over the piston. portion of the combustion chamber. These data were obI n interpreting these data the duration of the valve opening tained from the analysis of samples of gases taken from the must be considered. At the angle of revolution where the combustion chamber at positions 1, 2, 3, and 4 as indicated in samples from the combustion chamber just begin to show a Figure 6. The time intervals (in seconds) which were redecrease in oxygen concentration, the portion of the sample quired for the combustion zone to progress from the spark which is emitted during the latter part of each opening of the plug to the various posisampling valve may be completely burned while the portion tions of the sampling valve of the sample which is emitted during the early part of each are plotted as a b s c i s s a s opening may be entirely unburned. Hence, for determining and the distances from the the time interval between ignition and the arrival of the spark plug to the various combustion wave at the sampling valve, the factor that must positions of the sampling be measured is the time in degrees from ignition to the point valve are plotted as ordiwhere the curve marked “valve closes” shows a change in nates. The curves show slope. Similar considerations must be applied to the data the results which were obwhen determining the time interval from ignition until the tained at engine speeds of oxygen has completely disappeared at the sampling valve, 1200, 800, and 500 r. p. m. with the result that in this case the data are obtained from with the spark adjusted for maximum power a t the curve marked “valve opens.” Figure 4, then, shows that the oxygen begins to disappear each speed. These data rapidly a t 2 degrees before top dead center and Figure 5 i n d i c a t e that, over the shows that the oxygen begins to disappear rapidly at 7 or 8 narrow range of engine degrees after top dead center. Since the data in Figure 4 speeds which were studied, were obtained with the sampling valve 2 inches (5.1 cm.) from the average speed of the the spark plug and equidistant from the side walls of the com- combustion wave v a r i e s Figure 6-A Horizontal Cross bustion chamber, while the data in Figure 5 were obtained approximately w i t h t h e Section of Combustion Chamber, Showing Various Positions of with the sampling valve similarly located but a t a distance of engine speed. This effect Sampling Valve is mobablv due to the in3 inches (7.6 cm.) from the spark plug, these results show that a combustion wave passed between these two points in creased tuibulence of the charge in the combustion chamber the combustion chamber in about 10 degrees of revolution or with an increase in the engine speed. The form of the inat the rate of about 25 feet (7.5 meters) per second. It is also dividual curves in Figure 7 indicates that during normal significant to note that when the valve is 2 inches (5.1 cm.) combustion a combustion zone passes through the charge from the spark plug (Figure 4) the oxygen has completely with an accelerating speed until it is approximately threedisappeared a t 3 degrees of revolution after top dead center fourths of the way across the combustion chamber, and then or 4 degrees of revolution of the crankshaft before the oxygen travels the remaining distance a t a decreasing rate of speed. Figure 8 shows the effect of engine speed upon the progress begins to disappear when the sampling valve is 3 inches from the spark plug, (Figure 5 ) These data, then, indicate of the combustion zone along the wall of the combustion

-

ISDL'STRIAL .AND ENGINEERING CHEMISTRI'

September, 1930

chamber. These results were obtained by analyzing samples of gases which were taken from the combustion chamber a t positions 5, 6, 7, and 8 as indicated in Figure 6. Here again the time intervals (in seconds) between ignition and rapid disappearance of oxygen a t the sampling valve are plotted against distances from the spark plug to the sampling valve and the curves correspond to engine speeds of 1200, 800, and 500 r. p. m. These data indicate that the average speed of the combustion zone along the side wall of the combustion

F i g u r e 7-Effect of E n g i n e Speed u p o n Progress of C o m b u s t i o n Wave t h r o u g h M i d d l e P o r t i o n of C o m b u s t i o n C h a m b e r

Table I-Time f r o m I g n i t i o n u n t i l t D i s a p p e a r a n c e of Oxygen Begins a t Various P o i n t s in t h e C o m b u s t i o n C h a m b e r a t 500 r. p. m. Spark advance.. . . . . . . . . . . . 1 6 degrees before top dead center Mixture ratio.. . . . . . . . . . . . ,SOper cent of theoretical air Fuel. ..................... Gasoline (engine knocking) Compression ratio. . . . . . . . ..5.0: 1 Full throttle

1 2 3 4 5 6 7 8

Cm. 2.5 5.1 7.6 10.2 3.8 5.1 7.6 10.2

Of

TIMEAFTER IONITION Degrees revolution Seccnds ( x 10-3) 15 4.9 14 6.0 26 8.6 33 10.9 15 4.9 20 6.6 30 9.9 37 12.0

T a b l e 11-Time from I g n i t i o n until D i s a p p e a r a n c e of Oxygen Begins a t Various P o i n t s i n t h e C o m b u s t i o n C h a m b e r a t 800 r. p. m. Spark advance.. . . . . . . . . . .. 2 2 degrees before top dead center Mixture ratio.. . . . . . . . . . . ., 8 0 per cent of theoretical air Fuel.. . . . . . . . . . . . . . . . . . ..Gasoline (engine knocking) Compression ratio.. ....... . 5 . 0 : 1 Full throttle

POSITION OF DISTANCE FROM SAMPLINQ VALVE S P A R K P L U G TO (Fig. 6) SAMPLING VALVE 1 2 3 4 5

6 7 8

Inches 1 2 3 4 1.5 2.0

3.0 4.0

Cm. 2.5 5.1 7.6 10.2 3.8 5.1 7.6 10.2

F i g u r e 9-Spark Advance u p o n Progress of t h e C o m b u s t i o n Wave A-Through middle of combustion chamber B-Along wall of combustion chamber o-Spark 22 degrees before top dead center (maximum power) s-Spark 30 degrees before top dead center

increase of 8 degrees of spark advance slightly decreased the time required for the combustion zone to travel from the spark plug to the sampling valve. This apparent decrease in time is not large enough of itself to warrant the conclusion that detonation is accompanied by an increase in the rate of progress of the combustion zone. A number of determinations were made to find the effect of lead tetraethyl upon the speed of the combustion zone as it progresses across the combustion chamber. Since the data in Figures 4 and 5 indicated that lead tetraethyl had no effect upon the progress of the combustion zone, at least until after it had passed position 3 (Figure 6), runs were made with the T a b l e 111-Time f r o m I g n i t i o n until D i s a p p e a r a n c e of Oxygen Begins a t Various P o i n t s in t h e C o m b u s t i o n C h a m b e r a t 1200 r. p. m. Spark advance.. . . . . . . . . . .. 2 6 degrees before top dead center Mixture ratio.. . . . . . . . . . . .. 8 0 per cent of theoretical air Fuel. .................... .Gasoline (engine knocking) Compression ratio. . . . . . . . . ,5,0:1 Full throttle

POSITION OF S A X P L I N G VALVE

(Fig. 6)

P O S I T I O N OF DISTANC!E FRO31 SAMPLING VALVE S P A R K €'LUG TO (Fig. 6) SAMPLIW VALVE

Inches 1 2 3 4 1.5 2 3 4

graphically with the results obtained with the spark adjusted for maximum power in Figure 9. Curve A of Figure 9 shows the schedule of the combustion zone in passing through the middle portion and curve B that passing along the side wall of the combustion chamber. The data indicate that this variation of the spark advance and the resulting increase in the intensity of the knock had no effect whatever upon the oxygen curves, except when the valve was located at position 4 of Figure 6, in which case, an

F i g u r e 8-Effect of E n g i n e Speed u p o n Progress of C o m b u s t i o n Wave a l o n g Wall of C o m b u s t i o n C h a m b e r

chamber increases with the engine speed over the short range which was studied. The form of the individual curves, however, indicates that, the combustion zone travels the entire length of the combustion chamber at an accelerating rate. Since under the above conditions of engine operation the intensity of the detonation was such that 2.7 cc. of lead tetraethyl per E. s. gallon completely stopped the knock, it was thought desirable to repeat some of the above determinations with the engine knocking still more severely. To do this the spark was advanced from 22 degrees to 30 degrees while the other conditions of engine operation were kept as nearly constant as possible. With this degree of spark advance the engine knocked very violently and the output was 80 per cent of the maximum. Also 5.4 cc. of lead tetraethyl per U. S. gallon of gasoline failed to remove the knock completely. The results of this run are t,abulated in Table I V and are compared

1 2 3

T a b l e IV-Time

DISTANCE FROM SPARK

P L U G TO

SAMPLING VALVE Inches 1 2 3

Cm. 2.5 5.1 7.6

TIMEAFTER IONITION Degrees of revolution Seconds ( X 10-8) 26 3.6 32 4.6 34 4.8 6.0 3.9 4.8 6.2 7.4

f r o m I g n i t i o n until Disappearance of Oxygen Begins

a t Various P o i n t s in C o m b u s t i o n C h a m b e r a t 800 r. p. m., w i t h E n g i n e K n o c k i n g Violently Spark advance.. ........... 3 0 degrees before top dead center Mixture ratio.. . . . . . . . . . . ..80per cent of theoretical air Fuel.. ....................Gasoline (engine knocking violently) Compression ratio.. ........ 5 . 0 : 1 Full throttle

POSITION O F

TIMEAFTER Ion1nox Degrees of revolution Set:onds ( X 10-3) 20 4.0 26 5.0 30 6.0 42 8.4 27 5.4 30 6.0 37 7.4 45 9.0

949

DISTAKCE FROM

SAMPLING VALVE SPARKPLUGTO (Fig. 6 ) 1 2 3

SAMPLING VALVE

Inches 1 2 3

4

4

6

2.0

7

s

-1 . 5

2.0

4 .o

Cm.

2.5 5.1

7.6 ..

-1_0 . 2 3.8 6.1 7.6 10.2

TIMEAFTER IGNITION Degrees of revolution Seconds ( X 1 0 - 9 21 4.2 25 5.8 30 6.0 40 8.0 28 5.6 31 6.2 37 7.4 45 9.5

950

I N D U S T R I A L A N D ENGINEERING CHEMISTRY

Vol. 22, No. 9

sampling valve in positions 3 and 9 of Figure 6. While these samples were being removed, the rotor (Figure 2) was used to actuate the sampling valve in place of the cam. The results of the analyses of these samples for oxygen are shown in Figures 10 and 11,where the percentages of oxygen by volume are plotted against angles of revolution. In each of these figures the results of the analyses of samples taken from the engine at position 3 (Figure 6) are shown with broken lines, while the results of the analyses of samples taken from position 9 are shown with solid lines. The data in Figure 10 were obtained while operating the engine with 5.4 cc. of lead tetraethyl in each U. S. gallon of gasoline, and therefore without detonation; the data in Figure 11were obtained while operating with gasoline alone, and accompanied by detonation. The differences between the results obtained under

explosions. This fact, however, does not invalidate the use of the present results to determine the probable form of the flame front as it proceeds away from the spark plug. The results of this work show that the combustion zone travels faster through the middle portions than along the side walls of the combustion space. This phenomenon is probably due to the cooling effect of the combustion chamber walls. Consequently it is believed that a flame front which is convex toward the direction of travel proceeds outward from the spark plug through all parts of the charge. These results show also that the speed of the combustion zone increases with the engine speed both through the middle portions and along the side walls of the combustion chamber. This increased speed is believed to be due, in part a t least, to the increased turbulence of the charge. The exact s i g n i f i c a n c e of the steep parts of the oxygen curves is not yet apparent. Therefore, no attempt has been made to estimate accurately certain essential factors, such as the actual time required for the combustion process to complete itself after the combustion zone has arrived at the sampling valve. The forms of the steep parts of the oxygen curves are affected by several factors. First, Figure 10-Engine Not Knocking Figure 11-Engine Knocking Violently as was indicated earlier in this Percentages of Oxygen by Volume in Samples of T h a t Portion of t h e Charge Which Burns Last paper, the longer the valve reEngine speed. . . ,1000r. p. m. Spark. , . . . . . . . . .25 degrees before top dead center mains open the flatter will be Mixture ratio. . . .80 per cent of theoretical air the oxygen c u r v e s . ConseFull throttle Fuel-gasoline quently, in the interpretation of Broken line-sampling valve located 3 inches (7.6 cm.) from spark plug Continuous linesampling valve located 46/4 inches (12.1 cm.) from spark plug the meaning of the slopes of the oxygen curves the d u r a t i o n the two sets of conditions are coniined to the slopes of the of the valve opening must be considered. Second, the slopes oxygen curves obtained with the sampling valve at position 9. of the oxygen curves are affected by fluctuations of the rate of When detonation is present they-are steeper than during flame propagation between successive explosions. If the time of arrival of the combustion zone at the sampling valve normal combustion. fluctuates over an angle of 10 degrees of revolution, and if the Discussion of Results width of the burning zone approaches zero as a limit, at least The results of this study indicate that there is a definite 10 degrees of revolution will still be required for the oxygen combustion zone which passes through the charge of gasoline concentration to reach zero a t that point. A more complete and air in the engine a t a finite velocity. This conclusion is discussion of this phenomenon will be given in a later paper. based upon the fact that as the distance between the spark Third, the forms of the oxygen curves are also influenced plug and the sampling valve was increased there was an considerably by the change of the pressure in the combustion increase in the time interval between the instant that ignition chamber during the interval that the valve is open. This is occurred and the instant that the oxygen began to decrease due to the fact that the differential rate of gas flow through rapidly a t the sampling valve. During these intervals the the sampling valve is roughly proportional to the pressure in oxygen concentrations in samples emitted by the sampling the combustion chamber and to the area of the valve opening. valve remained fairly constant, so that these portions of the As the duration of the valve opening is increased the effect of this factor becomes more pronounced in determining the shape oxygen curves were flat. The chief source of error which should be considered in of the oxygen curves. Other factors which probably influence measuring this time interval is the performance of the sampling the shape of the oxygen curves to some extent are the time valve itself. Since a valve-interval-determinator was used required for the combustion zone to pass by the sampling to check the time and duration of the valve opening while valve, the rate of combustion of the gasoline in the burning each sample was being removed, it was possible to determine zone, and the quantity of oxidation which takes place in the within 1 or 2 degrees of revolution the crank angle between expansion chamber of the sampling valve itself. The combined effect of the various factors enumerated the instant that ignition occurred and the instant that the above is to decrease rather than to increase the slopes of the oxygen concentration began to decrease rapidly. I n case there was some fluctuation of the speed of the oxygen curves. And, since these curves are rather steep in combustion zone from one explosion to another while these spite of the factors-Figures 4 and 11, for e x a m p l e i t seems experiments were performed, the measurements recorded in probable that the zone of combustion is quite narrow, althis paper would represent the time required for the fastest though as yet no precise width can be assigned to it. But the results do show that during normal combustion in traveling combustion zones to proceed from the spark plug to the sampling valve. And, indeed, some experiments the engine a comparatively narrow combustion zone, within which will be the basis of a later paper show that there is some which the oxidation reaction completes itself as far as the air fluctuation in the rate of flame propagation between successive present permits, proceeds from the spark plug through all

.

September, 1930

INDUSTRIAL A N D ENGIXEERING CHEMISTRY

parts of the charge at a measurable rate of speed. Whether or not this combustion zone is the actual flame itself cannot be determined from the results which are presented herein, because by definition the term “flame” means oxidation accompanied by the emission of light. If it is assumed that this combustion zone is the flame itself (experiments to be described in a later paper indicate that this assumption is probably correct), the data presented here indicate that, in one important respect a t least, combustion in an engine is similar to that in a bomb (6). Both in the engine and in the bomb there is a burning zone which proceeds from the spark plug through all parts of the charge a t a speed which in part is a function of the degrees of turbulence of the gases in the charge. EFFECT OF DETONATION UPON OXYQENCmvss-Figures 10 and 11 indicate that the presence of sufficient lead tetraethyl in the fuel to stop the knock has no effect upon the oxygen curves until after about three-fourths of the charge has burned. When the last part of the charge burns, however, the oxygen curves are much steeper when the engine is knocking than when it is running normally. The increased steepness of the slopes of the oxygen curves may be due to a higher rate of oxygen consumption. On the other hand, it may be due to the sharp rise in pressure which accompanies detonation, or to a combination of these factors. At the time of detonation a pressure rise of 100 pounds per square inch above the pressure of a normal explosion has been observed. These observations were made while using the electrical pressure (3, 4) upon the engine which was employed in this research. The results do indicate definitely, however, that the disturbance which is known as detonation occurs in that part of the charge which burns last, and the data of this paper are not inconsistent with the idea that there is a more rapid spread of the combustion wave through the last one-fourth of the combustion space in detonating explosions than in nondetonating explosions. Comparison of Present Results w i t h Previous Work

The conclusions about the character of combustion in the gasoline engine that have been reached on the basis of data contained in the present paper differ in some important respects from those presented in the earlier papers from this laboratory. The reason for these differences lies in the various factors that have since been found to influence the shape of the oxygen curve, as outlined above. Outstanding among these factors is the length of time the sampling valve remains open. On account of imperfections of the early apparatus, as already mentioned, it remained open much longer than was then believed-namely, 20 or more degrees of revolution instead of about 2 degrees. A valve that remains open for a long interval of time would yield samples of gas showing a

95 1

slow transition in composition from entirely unburned to completely burned. This would be true even if there were a substantially instantaneous transition from unburned to burned gases in that portion of the combustion chamber in which the sampling valve was located. It happened that in most of the earlier work the sampling valve was located in that portion of the combustion chamber within which detonation, when present, is now known to occur, and within which there is a distinct difference between the form of the oxygen curve obtained when the engine is knocking and when it is not knocking. The difference consists essentially in a steeper slope during detonation than during normal combustion. (Figures 10 and 11) The specific differences between the conclusions of the early work and those reported in this paper are in respect to the following points: (1) It was previously concluded that after ignition there was a very rapid spread of flame throughout the combustion chamber and that, following this, the duration of the combustion reaction as measured in degrees of revolution during which it persisted, was so considerable that it was going on simultaneously practically throughout the whole combustion chamber. The more accurate data obtained with the new sampling valve, which not only has a much shorter duration of opening but which is also provided with means for indicating the point of opening as well as the time of remaining open, and thus yields samples which are a close approximation t o the optimum instantaneous sample, show that the combustion wave proceeds from the spark plug through the combustion chamber a t a definite rate. (Figure 7) The data indicate that the zone within which the combustion reaction completes itself is quite narrow. (2) It was previously concluded from the slopes of the oxygen curves that during detonation the combustion of the fuel (meaning the rate a t which the fuel was consumed during the actual combustion reaction, as distinguished from the velocity of flame spread through the mixture) proceeded a t a rate more rapid than normal. The results of this further study show that the forward movement of the combustion wave is not affected by detonating conditions until i t has reached the zone of detonation, which is the latter portion of the combustion space within which the sampling valve was located during most of the early work. Acknowledgment

The authors desire to acknowledge the large contribution to this work that was made by H. V. Almen and C. M. Doyle, who designed and developed the sampling apparatus which made the experiments possible. Literature Cited (1) Brooks, “Chemistry of Internal-Combustion Engines,” 121. S. Thesis, Ohio State University, 1922 (unpublished work). (2) Lovell and Coleman with Boyd, IND. ENO.CHEM.,19, 373, 376 (1927). (3) Martin and Caris, J . Soc. Automotive Eng., 48, 87 (1928). (4) Martin and Caris, E l e c . J., 47, 87 (1930). ( 5 ) Maxweil and Wheeler, J . Inst. Petroleum Tech., 14, 175 (1928).

Activity in Construction of Gas and Oil Pipe Lines Construction of pipe lines for natural gas and the manufacture of pipe for the systems promises to be one of the major industrial developments of the year, according to the Bureau of Mines. Expenditures for long distance lines to transport natural gas alone may reach $250,000,000 for the year. Manufacture of the pipe for such lines has been one of the chief activities of the iron and steel industry. Improvement in pipe so as t o enable it t o withstand high pressure and better methods of welding which make lines nearly leakproof have been partly responsible for the growth of pipe lines within the last few years. Natural gas lines planned or under contemplation include one from the Texas Panhandle t o Chicago, extensions to various points of the line from Monroe, La., to Atlanta, a second line from the Kettleman Hills field to San Francisco, possibly a line

from the Texas Panhandle to Indianapolis, perhaps extensions of existing lines to serve Washington and cities along the Atlantic coast, and a line to Butte and Anaconda, Mont., from fields in north-central Wyoming or possibly from fields in Canada. I n addition, a gasoline line is being built from Marcus Hook, near Philadelphia, to Toledo. Work is about to start on a gasoline line connecting refineries in Oklahoma and Kansas with consumers in the Chicago-Milwaukee area. The latter will be a cooperative line and will be used by Continental and Barnsdall and possibly two or three other refining companies. Perhaps half the freight cost will be saved through use of the line, the saving amounting to several cents on a gallon on some shipments. Furthermore, a crude oil line is being laid from Oklahoma to Wood River, Ill.