EXPLOSIONS

There is a critical initial temperature for an explosive mixture which gives the maximum rate of pressure rise to the explosion which occurs upon igni...
0 downloads 0 Views 898KB Size
There is a critical initial temperature for an explosive m i x t u r e which gives the maximum rate of pressure rise to the explosion which occurs upon ignition. This critical initial temperature depends in a large measure upon the composition of the products of co m bu s t io n which, in turn, depend upon the composition of the explosive mixture. Other conditions, such as whether the charge is ignited at constant density or at constant pressure, are also important. These factors aid in explaining the observed differences in depreciation of fuels of different types in supercharged engines as compared with ordinary internal combustion engines. In reporting the effect of temperature upon explosions, the existence and value of the critical initial temperature must be considered if significant results are to be obtained.

GASEOUS EXPLOSIONS CRITICAL INITIAL TEMPERATURE FOR MAXIMUM RATE OF PRESSURE RISE' WILLIAM A. PEARL* AND GEORGE GRANGER BROWN University of Michigan, Ann Arbor, Mich.

I

K AN EA4RLIERpaper (2) the existence of a critical initial temperature giving maximum rate of rise of pressure to the subse-

quent explosion was demonstrated by experimental and theoretical data, and was used to explain otherwise contradictory results. Lafitte (11) has since extended and confirmed Dixon's earlier results (6) that an increase in initial temperature from 20" C. in theoretical mixtures with oxygen retards the formation and velocity of the "explosive" or "detonation" wave. But Lichty (12) using hydrogen-air mixtures of varying proportions, and assuming that rate of flame propagation and rate of rise of pressure are the same, was unable to find a definite relationship between rate of flame propagation and initial temperature or to verify the existence of a critical initial temperature. Aubert and Duchene (1) studied the combustion of hydrocarbon mixtures in a gas engine through a window in the combustion chamber, and noted a marked effect of cylinder temperature on rate of flame travel. Fenning (IO)found that a rise in initial temperature increased the tendency to detonate in hexane-air and benzene-air rich mixtures. Egerton and Gates (9) observed that in acetylene or pentane mixtures with oxygen and nitrogen a t constant initial pressure, an increase in initial temperature from room temperature to 230' C. retarded detonation, and that mixtures giving carbon monoxide burn faster than those giving 1 The previous eight papers in this series appeared in 1925, 1927, 1928, and 1829. Preient address, Armour Institute of Technology, Chicago, Ill.

carbon dioxide, and that carbon dioxide has a greater delayhg effect than nitrogen, which is probably due t o dissociation of the carbon dioxide. Lichty (12) reports that the theoretically correct mixture of hydrogen burns the fastest, and Campbell, Littler, and Whitworth (4) report that rate of rise of pressure is a maximum a t approximately a theoretical mixture with oxygen. Dumanois (6) found that engine knock in a C. F. R. engine a t 600 r. p. m. and jacket temperature of 150" C. was eliminated a t a high compression ratio a t a mean apparent temperature in the combustion chamber of 340" C. The effect of initial temperature on gaseous explosions is important and not well understood. This paper reports new data particularly concerned with the critical initial temperature which gives the maximum rate of rise of pressure. The apparatus used in this work consisted of four bombs of the type shown in Figures 1 and 2, three pressure elements, and a two-cylinder engine.

Description of Bombs Each bomb was machined from a solid block of steel. All joints in connection with the jacket of bomb 2 (Figure 1) were welded. The tops or covers were threaded, screwed in place, and welded. The arrangement shown in Figure 1 provides a rapid and easy method of controlling the temperature. With valve e wide open and valves c and d closed, if valve N' is opened, the cudden drop in pressure of the carbon dioxide as it enters the jacket forms solid carbon dioxide, which in 3 or 4 minutes fills the jacket. By this means, temperatures as low as -60' c. are easily obtained when the bomb is well insulated with asbestos. By the introduction of high-pressure steam into the jacket, any point from room temperature up to the maximum temperature of the steam can be obtained. The standard temperature for charging is attained quickly by admitting water through valve d to the jacket. 1058

SEPTEhIBER, 1936

IKDUSTRIAL AND ENGINEERING CHEMISTRY

Bomb 2 wa-: provided with three threaded openings for spark plug?: one in the center of the bottom, one in the side near the bottom, and the third, P , in the side about halfway up. One side of the bomb was equipped with two windows, f , one near t'he top and the other near the bottom. These windows were made of 0.25-inch plate glass and were held in place by a hollo~vnut and sealed with litharge and glycerol. A lens, g, was placed in front of each of these windows, f, to focus the light from the burning gas in the bomb on a rotating drum, S,in order to measure the time required for the flame to fill the bomb and duration of the flame. An opening, into which a t'hermocouple, X , was fastened, was located in the side of the bomb about halfway between the top and the bottom so that the thermocouple junction mas in the center of the bomb. Bomb 3 (Figure 2 ) was not provided with window and was heated by gas burners on the sides and cooled by running water over it. It was provided with a removable jacket t h a t could be filled with solid carbon dioxide to cool the bomb for the lower temperatures. This bomb was bet'ter suited to the extreme temperatures because there were no windows to be damaged by great changes in temperature. Bomb 1 was similar to bomb 3 except for size. For extremely low temperatures bomb 4 (Figure 2 ) was used as it could be cooled with liquid air. It was of such a size that it' could be submerged in a wide-mouthed Thermos

1059

bottle of liquid air. A small thermocouple, charging valve, and special pressure element were located in the top of the bomb in such a way as not to interfere with the submersion of the bomb in the Thermos bottle. The spark plug was located in the center of the bottom. All other connections were made in the top of the bomb. The bomb was charged at room temperature, then placed in the Thermos bottle, and cooled to the required temperature. It was then removed, placed in an asbestos jacket, and mounted in the frame (Figure 2) in which position the charge was ignited. The frame was a means of holding this small bomb to one of the larger bombs so t h a t the same pressure recording derice could be used. TABLE I. DIMEXSIONS AND MEASURED VOLUMES OF BOMBS Bomb NO.

1 2 3 4

Diameter Cm. 5.4 7.8 7.3 4.75

Length Cm. 17.75 12.7 10.2 11 4

Measured Vol. cc. 402 604 415 224

The charging valve, G (Figure l), can be used either for a liquid or for a gaqeous fuel. Valve K has a twofold purpose, for charging liquid fuels and for purging the header, L , the piping, and valve G. \Then the bomb was to be charged with

INDUSTRIAL AND EYGIXEERISG CHEMISTRY

1060

~

PRESS&RE EL EMEN 7 ODE/$ YG -CHARbI/rG

VAL/€

-TnFaMAL

B p B 3

GP:&l&G

coUDLEG.W,~L

1-

cog- vc- ,Ac/Er DEWO/fD

VOL. 28, NO. 9

tive timing with reference to breaker points d and B which were driven on the same shaft. When breaker points B opened, breaking the primary circuit, a spark occurred a t the spark plug. A primary voltage of 10.5 with the 6-8 volt, high-tension coil, R, in combination with this timing device gave a strong spark of uniform intensity. The same source of current was used for the 50 c. p., 6-8 volt headlight bulb in order to obtain the maximum brilliance. Detail 2 (Figure 3) shows the method of determining the timing of the spark. The high-tension current from coil R jumped a fixed gap between two black plates. The plates were 0.05 mm. apart and mounted in a vertical position perpendicular to the rotating drum. The light from the spark across the gap produced vertical black lines on the photographic paper attached t o rotating drum C. Since a spark produced by this method does not jump in one direction and cease, but oscillates back and forth decreasing in intensity for a period of time, many black lines, parallel to each other and decreasing in intensity, appeared on the bromide paper. The first black line is the ignition spark. The distance of this line from a reference point on the bromide paper gives the exact timing of the spark. The reference line was produced by the spring wire that held the bromide paper on the drum. By measuring the distance from this reference point to the point a t which the igniting spark occurred, the delay in the igniting spark due to increased pressure or temperature was determined. Similarly, the delay in apparent rise of pressure after the igniting spark was determined. '

m 9

OF BOMBS3 FIGURE 2. DIAGRAM

AND

4

oxygen, for example, valve K was opened and valve G closed, and the pipes were purged with oxygen. Valve K was then closed and the bomb was ready to be charged with oxygen. Manometer t was used for low pressures, and gage H for higher pressures. Pressure elements of the diaphragm and of the Midgley piston type were used. In the latter it was necessary to seal the piston t o prevent leakage by placing a thin rubber diaphragm between the bottom of the pressure element and a small flange on the bomb. This method was very satisfactory except for exceedingly low and high temperatures. At the low temperatures the piston would freeze to the wall and fail to move. At the high temperatures the rubber would soften and leak. A long pressure element was necessary to prevent the mirror from becoming covered with frost when working a t low temperatures. A little water was placed on the diaphragm to keep it cool when tests were being made a t high temperatures. A complete wiring diagram is shown in Figure 3. Variations in the lag of apparent rise in pressure after the spark o c c u r r e d might be due to variations in the intensity of the spark caused by variations in the period of buildup in the coil, Since it was impossible to o p e r a t e the hand switch so as to make a contact a t exactly the same position of the cam every time a charge was ignited, a second cam and contact, A , was used. The arm on switch T was moved to the right as indicated. A contact on copper strip k was made before the contact on copper strip T . This gave time for the intensity of light D to become a m a xi m u m before ignition. When contact r was made, current flowed through solenoid S when breaker A made a contact. Breaker points Y on the solenoid acted as an ignition switch with a posi-

Description of Engine The engine was constructed by modifying a small two-cylinder, opposed, four-cycle motor. One cylinder was used as a pump cylinder, the other was operated as a four-cycle gas engine. Coiled copper pipes led from the exhaust of tbe pump cylinder to the intake port in the engine cylinder. The carburetor wa8 attached to the intake port in the pump cylinder. The engine cylinder was equipped with a Midgley pressure indicator. The rotating drum, on which the beam of light was reflected, was gear-driven and ran at the same speed as the crankshaft. The pump cylinder was so modified as to have a minimum clearance and the camshaft modified so that the cylinder functioned as a four-stroke cycle pump which delivered a constant charge by weight to the engine cylinder at all times. The tem erature of the charge was changed as desired in the copper tuiing between the two cylinders. A thermocouple was placed in the copper tubing next to the engine cylinder for indicating the

FIGURE3.

W I R I N Q DIAGR.4M

SEPTEMBER, 1936

IXDUSTRL4L AND ENGINEERING CHEMISTRY

intake temperature, and a manometer was connected t o an opening in the copper tubing to indicate intake pressure. The small engine was direct-connected to a large synchronous motor that ran at 1200 r. p. ni. and maintained constant speed of the engine regardless of its power output.

Calibration of Apparatus The pressure elements were calibrated in place by filling the bomb with oil and uqing a dead-weight loading device previously described ( 2 ) . The deflection was directly proportional to the pressure for all three pressure elements. The ChronielAlumel thermocouples were calibrated by comparison with a standard thermometer and a t several melting points a t low temperatures. The speed of the drum that carried the bromide paper was maintained constant by using a 1/3 h. p. synchronous motor as driring means.

Procedure The I J C J(Figure I ~ ~ ~ 1) was brought to a temperature above 100' C and purged with nitrogen by charging six times to a pressure of 100 pounds per square inch (7 kg. per sq. cm.) and exhausting through port &. Test runs made after purging the bomb three times checked similar tests made after purging eight times as already explained. After purging, the bomb was brought t o room temperature and charged with oxygen and fuel. With all valves on the bomb closed tightly, the mixture was heated t o 150" C. to mix the gases, which were then taken to the desired initial temperature. T o eliminate any question as to the effect of this preheating, numerous tests were run heating the mixture to ZOO", cooling i t to 30",then bringing it to the desired initial temperature before igniting, and the rate of rise of pressure checked tests in which the gases were not heated and cooled before they were brought to the initial temperature a t which the test was to be made. The work of Dumanois and Mondain-Monval(7) indicates t h a t preheating up to within 10" C. of the ignition temperature does not cause

1061

partial combustion, since no carbon dioxide or aldehyde was formed until ignition occurred with a rapid rise in pressure. After the bomb had been charged and brought to the desired temperature, the room was darkened and the bromide papers were mounted on drums C and N . The motor was started and, when the drums had come to constant speed, the bomb was ignited by operating the lever on switch T (Figure 3). The bromide papers were then developed. The products of combustion were taken through the sampling tube from the bomb for analysis. The bomb was then ready to be purged and charged for the next test. The initial density or weight of charge was kept constant for each series of tests, so that each series indicates the effect of initial temperature at a constant initial density except series 38, 39, and 40, in which the densities of charges were varied so as to maintain the initial pressure constant in each series.

Results In all, some two thousand tests were made. The most significant dat'a obtained from these tests are given in Figures 4 to 7 , and summarized in Table 11. In plotting these curves, from which the critical initial t'emperatures given in Table I1 were determined, the average rate of rise of pressure ( A P ) / ( I t ) was used. Where A P = rise of pressure, lb. per sq. in. At = increment of time, after initial rise in pressure, required for pressure to reach a maximum, sec.

B o m b Tests The results of this work verify the existence of a temperature a t which the rate of rise of pressure is a maximum, and in addition show that it is not a fixed temperature but can be controlled by changing initial conditions. The most extensive tests were made using methane as a fuel, and the effect of initial conditions on it's critical initial temperature will be dis-

INDUSTRIAL AND ENGINEERING CHEMISTRY

1062

TABLE11. SUMMARY OF CRITICAL INITIAL TEMPERATURES AS DETERMINED IN BOMBS Test Series No,

1

2

Abs. Bomb Charging No. Pressure Ns Cm. H a ,---Per Acetylene 1 221.1 45.3 Fuel

1

3

1

4

1

1

0 2

Fuel

Vol. Ratio, Os/Fuel in hZixture

Critical Initial Temp.

c.

cent bu uolume--

2.72:l 1.54:l

.... ....

220.8 50.1 213.1 51.9 200.3 55.5

40.0 14.7 30.3 19.6 24.0 24.1 39.2 5.3

128.5 78.2 122.6 60.6

21.8 39.4

0.181g. Theoretical

78.0 84.0

40.3 55.8

52.2 39.5

7.5 5.0

-2.0 62.0

1:l

7.4:l

15.0 8.0

5 6

Benzene

7

Butane

1 1

184.3 199.0

9 10 11

csz

1 1 1

162.5 45.7 162.5 45.5 161.4 45.5

64.3 54.5 54.5

0.5Og. Veryrich 0.39g Rich

12 13 14

Ethylene

3 3 3

332.0 76.8 462.0 82.7 332.1 75.9

15.1 13.0 18.1

9.1 4.3 6.0

1.66:l 3:l 3:l

200.0 120.0 20.0

15 16 17

Hydrogen

1 1

50.3 41.0 41.2 49.0 55.9 70.8C01

20.0

29.7 19.4 39.6 14.4 44.4 17.0 34.0 14.7 29.7 14.6 14.6

0.67:l 0.492:l 0.325:l 0.5:l 0.5:l 1:l

35.0 21.0 32.0

3

250.3 250.1 250.0 176.3 332.3 332.3

1 1 1 2 2

119.1 122.2 126.0 146.0 146.3

62.2 60.7 58.7 50.7 50.7

25.2 24.6 23.8 37.7 32.8

12.6 14.7 17.5 11.6 16.4

2:l Below 0 1.67:l 30.0 62.0 1.36:l 3.25:l 2.0 2:l Below -45.0

26 27 28 29 30

3

3 3 3 3

332.3 231.1 332.3 332.3 332.3

64.7 64.7 64.7 64.8 64.8

23.4 23.4 21.7 28.1 20.0

11.7 11.7 13.5 7.1 15.2

2:l 2:l 1.61:l 3.96:l 1.31:l

31 32 33 34 35

3 3 3 3 4

462.3 74.6 462.3 74.5 0 117.3 0 114.1 191.3 38.8

16.9 20.2 66.6 79.6 40.8

8.5 5.3 33.3 20.5 20.4

92.0 2:l 3.81:l 138.0 2:l Below -120.0 3.88:l -100.0 2:l -110.0

36 37 370 38 39

4 4 3 3 3 3 2

332.3 462.3 592.3 280.0* 418.7* 177.1* 151.0

64.7 74.8 81.2 64.9 74.8 38.9 49.1

23.4 16.8 13.2 23.4 40.7

11.7 8.4 6.6 11.7 8.4 20.4

44.3

6.6

8

1

1 2

18

19 20 21 22 23 24 25

3 >!ethane

40

41

Propane

* Constant

initial pressure teat series.

16.8

0.088g. Lean

0.3Og.

6.97:l 7.9:l Lean

2:l 2:l 2:l 2:l 2:1 2:l

6.72:l

68.0 90.0 60,130

6.0

100.0 115.0

-16.0 -18.0 210.0 25.0 248.0

-22.0 98.0 186.0 -50.0 Above 160.0 Below -40.0 2.0

VOL. 28, NO. 9

cussed first. At constant initial density, tests indicate that the bomb size does not materially affect the critical initial temperature as shown by comparing test series 31 and 37, and also comparing series 26 and 36. One bomb had approximately twice the volume of the other and about half the diameter-length ratio. Similarly the initial density does not affect the critical initial temperature as shown by comparing series 26 and 27. The concentration of the diluent was found to be an important factor in determining the critical initial temperature. Excess oxygen above that required for a theoretical complete combustion, excess fuel, or a so-called inert gas such as nitrogen or carbon dioxide raises the critical initial temperature. Nitrogen in a theoretical mixture of methane and oxygen raises the critical initial temperature as shown by a comparison of series 33, 35, 36, and 37 (Figure 8). With no diluent present, the rate of rise in pressure increased uniformly as the initial temperature was lowered, indicating that, if a critical initial temperature exists for a theoretical mixture, it is a t a very low temperature, as shown by series 33 (Figure 6). Methane and oxygen in a mixture containing 38.6 per cent by volume excess oxygen (series 34, Figure 6) shows a critical initial temperature of about - 100O C., similar to that found with a like percentage of nitrogen (series 35, Figure 6). A correlation of critical initial temperatures when the diluent was all nitrogen, and when it was part nitrogen and part excess oxygen, was not established. A comparison of series 26 and 29 (Figure 4) indicates that in the presence of 64.7 per cent nitrogen, 13.9 per cent by volume excess oxygen raised the critical initial temperature 43' C. If the 13.9 per cent excess oxygen had been additional nitrogen, the critical initial temperature would have been

raised 166" C. A similar condition is shown by a comparison of series 31 and 32 (Figure 4). However, a comparison of series 24 and 27 (Figure 4) indicates an inverse relation. When the diluent was 50.7 per cent by volume nitrogen and 14.5 per cent excess oxygen, the critical initial temperature was 15" C. higher than when the total diluent was nitrogen. The effect of the preieiice of excess fuel in a methaneoxygen-nitrogen mixture is shuwn by comparison of seriec 26, 28, and 30 (Figure 4). Excess methane to the extent of 24.5 per cent raised the critical initial temperature 226" C., and 52.0 per cent excess methane raised the critical initial temperature 264" C. above that of a theoretical mixture containing the same nitrogen concentration. Similar tests were made on theoretical mixtures of methane a t constant initial pressure (series 38, 39, and 40, Figure 5). When 74.6 per cent nitrogen was present (series 39) a t constant initial pressure, the critical initial temperature was above 160" C. For the same composition of gas ignited a t constant initial density, the critical initial temperature was 92' C. With 64.9 per cent nitrogen present (series 38) a t constant initial pressure, the critical initial temperature was -50" C. And for the same composition ignited a t constant initial density, the critical initial temperature was - 16" C. Tests on different fuels under similar conditions indicate considerable difference in the critical initial temperature. A comparison of series 16 with 35 (Figure 6) indicates that the critical initial temperature of this hydrogen mixture is 131O C. h her than methane in a similar mixture under similar conditie s. Comparison of series 14 and 31 (Figure 4) indicates that the critical initial temperature of a methane mixture is about 72" higher than a similar ethylene mixture under similar conditions. A correlation between the effect of nitrogen on the critical initial temperature of ethylene and methane is noted by comparing series 13 (Figure 5) and series 14 (Figure 4) with series 31 and 37a (Figure 4). Increasing the nitrogen 6.8 per cent raised the critical initial temperature of ethylene

%

100" C.; increasing the nitrogen 6.6 per cent raised the critical initial temperature of methane 93" c. A large difference is noted in the effect of excess fuel on critical initial temperatures. A comparison of series 26 and 30 (Figure 4) indicates an increase of 264" C. caused by 52 per cent excess methane. In the case of hydrogen, 54 per cent excess fuel raised the critical initial temperature only 11" C. as shown by a comparison of series 16 and 17 (Figure 6). In the case of ethylene, 80 per cent excess fuel raised the critical initial temperature 180" C. as shown by a comparison of series 12 (Figure 5) and series 14 (Figure 4). In each of the bomb tests an analysis was made of the products of combustion, The products of combustion were not materially affected by the initial temperature, if the fuel remained in the vapor state. The percentage of carbon dioxide increased a t high temperatures by an amount not exceeding 2 per cent. When excess oxygen was present, it appeared as free oxygen in the products of combustion. The same was true when excess hydrogen was present. However, when the fuel was a hydrocarbon, the percentage of carbon dioxide decreased and the percentage of carbon monoxide and hydrogen increased rapidly as the excess fuel increased up to a certain richness, a t which point free carbon appeared. It is evident that the initial temperature is an important factor in determining the rate of pressure rise. Failure to recognize the existence of a critical initial temperature giving a maximum rate of pressure rise has led to contradictory results when considering the effects of the initial temperature. It should also be recognized that some fuels will not remain in the vapor state a t the critical initial temperature under given conditions.

Engine Tests The bomb tests indicate that a rich mixture containing a concentration of nitrogen equivalent to that of air has a critical initial temperature similar to the initial temperature in an

INDUSTRIAL AKD EXGINEERING CHEMISTRY

1064

automobile engine cylinder. Furthermore, such mixtures have a higher critical initial temperature under conditions of constant initial pressure as existing in the usual engine than a t constant initial density. A supercharged engine apprordmates a condition of constant initial density, while an engine that is not supercharged operates a t approximately constant initial pressure, all conditions being the Same except the initial temperature. In this case the supercharged engine might be working above the critical initial t e m p e r a t u r e and the non-supercharged engine below. The results of the engine tests shown by series 42 and 43 (Figure 7) verify the existence of this condition. Over the range of temperatures at which the tests were made a t constant initial c3 pressure, the engine was operating below the critical initial temperaINTAKE TEMPERATZREture, and the rate of rise of pressure ON R~~~~OF pRESSmEincreased with increased initial temRISE perature. In the case of constant initial density tests such as in the supercharged, the engine was operating above the critical initial temperature, and, as the temperature of the entering air was raised, the rate of pressure rise decreased. This difference in the critical initial temperature of supercharged and conventional gasoline engines may be the cause of the observed differences in depreciation of fuels of high octane number. If the knocking tendency of a fuel is a function of autoignition temperature and of rate of rise of pressure as has been suggested (S), the knocking tendency would be increased more noticeably by an increase in temperature below the critical initial temperature (which would increase the rate of pressure rise on ignition as well as approach the autoignition temperature) than by an increase in temperature above the critical initial temperature which would decrease the rate of rise of pressure and thereby tend t o compensate for the other effects of increase of temperature. The observations of Dumanois (6) that knock could be eliminated by high operating temperatures may be explained on the basis that he was operating above the critical initial temperature.

Theoretical Discussion The equation for rate of rise of pressure a t constant volume previously suggested (3) is dP at

h

P

--B

a

(l)

= c x ~ ~ x X( e~ T )

where C, is a function of the absolute temperature and may be replaced by the following equation for heat capacity in the terms of absolute temperature (8): C. = b dT + f T 2 gT3 (2) Substituting Equation 2 for C, in Equation 1:

+

+

Differentiating Equation 3, keeping P / T (density) constant, and equating equal to zero t o obtain the value for the absolute temperature a t which the rate of rise of pressure is a maximum, 3,TC4 - (gB 2f)Tc* (fB- d)TC2 dBT, bB = 0 (4) where T,is the critical initial temperature a t constant initial density. Similarly, differentiating Equation 3, keeping P constant,

+

+

+

+

(a9

VOL. 28, NO. 9

+ %)To4- (aj + gB + 2f)Tc3(ad -.fB

+ d)Tc2 - (ab - dB)Tc + bB = 0

(5)

where TCis the critical initial temperature a t constant initial pressure. In Equation 4 some positive value of T, will satisfy the equation. The same is true of Equation 5 . That is, for some positive finite initial temperature the rate of rise of pressure will be a maximum. If the numerical values (8) for b, d, j , and g are inserted in Equation 4, the value of T, may be found in terms of B. If the products of combustion are carbon dioxide, the critical initial temperature (T,) is low; when the products are carbon monoxide, nitrogen, and oxygen, T , is higher; it is still higher for water and highest for the case of hydrogen as a product of 25'-combustion. Clearly the critical initial temperature is a function of the products of combustion. The experimental results correlate, in all cases, with the theoretical analyqic. The richer the mix$ ture, the larger is the percentage of carbon monoxide 5 and hydrogen, and the less carbon dioxide in the products of combustion of a hy2 ' drocarbon fuel. indicating (as reported) a higher critical initial temperature as the richness was increased. Similarly, excess hydrogen increases the critical initial - gc temperature, but relatively less than the same excess of 40 60 BO a hydrocarbon fuel because PEP c i l -VOG~N there is less difference beFIGcRE IxITIaL T E Y P E R . ~ T ~ RP EER C E ~ T A G E tn een water and hydrogen NITROGEXFOR THEORETIC~Lthan between carbon diox~ I I X T U R EOFS METHANE AND ide and carbon monoxide OXYGEN plus hydrogen. The products of combustion for fuels with different carbon-hydrogen ratios would be different and therefore exhibit different critical initial temperatures.

Conclusions The critical initial temperature is a function of the products of combustion and the diluents. The explosive mixtures giving products of combustion whose heat capacities have low temperature coefficients are indicated as having high critical initial temperatures. Listed in order of decreasing temperature coefficient are carbon dioxide, carbon monoxide, nitrogen, oxygen, water, and hydrogen, so that as the composition of the products of combustion shifts from carbon dioxide toward hyd;ogen, the critical initial temperature of the explosive mixture is raised. The results obtained indicate that the greater the carbon-hydrogen ratio of a fuel, the lower is the critical initial temperature. Increasing the diluent, or excess oxygen or fuel, raises the critical initial temperature in all cases. Increasing the excess hydrocarbon fuel raises the critical initial temperature rapidly up to a certain richness, after which it has relatively less effect. Increasing the excess hydrogen fuel raises the critical initial temperature relatively much less than increasing the excess hydrocarbon fuel. The critical initial temperature under conditions of constant initial pressure is lower than a t constant initial density for

SE‘PTEMBER, 1936

INDUSTRIAL A4NDENGINEERING CHEMISTRY

theoretical and near-theoretical mixtures containing small concentrations of diluent nitrogen. For mixtures containing concentrations of diluent nitrogen equal to or greater than that of air, the critical initial temperature a t constant initial pressure is higher than a t constant initial density. Similsr relationships are found in gasoline engines and bombs, and may help explain the effects of supercharging upon fuel combixtion characteristics.

Literature Cited (1) Aubert, M., and Duchene, R., Compt. rend., 191, 123-5 (1930). (2) Brown, G. G., Leslie, E. H., and Hunn. J. V., ISD. ESG.CHEV, 17, 397 (1925).

1063

(3) Brown, G. G., and Watkins. G. B., Ibid., 19, 280 (1927). (4) Campbell, C., Littler, W. B., and Whitworth, C., J. Chem. Sac., 135, 339-48 (1932). ( 5 ) Dixon, C., Trans. Roy. Sac. (London), 184, A97 (1893). (6) Dumanois, P., Ann. combustibles Ziquides, 9, 143 (1934). (7) Dumanois, P., and Mondain-Monl-al, P., Compt. rend., 189, 761-3 (1929). (8) Eastman, E. D., Bur. Mines, Tech. Paper 445 (1929). (9) Egerton, A, and Gates, S. F., Proc. Roy. Sac. (London), A114, 137-51 (1927). (10) Penning, R. W., Phil. Trans., 225, A331 (1926). (11) Lafitte, P., Compt. rend., 186, 951-3 (1928). (12) Lichty, L. C., Trans. Am. SOC.Mech. Engrs., 51, 37-44 (1929).

RECEIVEDMay 11, 1936.

LIQUID PROPANE Use in Dewaxing, Deasphalting, and Refining Heavy Oils

ROBERT E. WILSON Pan American Petroleum and Transport Company, New York, N. Y.

P. C. KEITH, JR. ROBABLP the most remarkable development in the history of refining heavy oils The M, W. Kellogg Company, for the production of lubricants has been New York, N . Y . the recent development of the use of liquid propane, as a solR. E. HAYLETT vent (or, more accurately, as antisolvent) for the removal of asphalt, wax, and other undesirable constituents from lubriUnion Oil Company, Los Angeles, Calif. cating fractions. The purpose of this paper is to discuss both the theoretical and practical aspects of these new processes and to describe briefly the commercial installations which have alrender the oil marketable. In this paper the term “naphready been made in the United States for carrying them out, thenic” is used inclusively to denote all nonparaffinic conwith ernphasis on certain novel engineering features. stituents other than the asphaltic materials. Thus it may inIn a previous paper (6) the authors pointed out that the clude compounds of olefinic or aromatic characteristics, and important properties of lubricating oils, particularly for use as in any case refers to those undesirable constituents present motor oils, are lorn- carbon-forming tendencies, low pour test, in all lubricating cuts which have a relatively low stability, high viscosity index (i. e., low rate of change of viscositywith low viscosity index, and low hydrogen-carbon ratio. temperature), and high resistance to oxidation and sludging. Reasons were given -why these properties were particularly important from the standpoint of the performance of motor This paper necessarily touches only the more important oils in modern internal c o m b u Y t i o n developments which have occurred in propane refining engines.

P

Undesirable Constituents in Lubricating Fractions To obtain these properties, five types of constituents present in ordinary lubricating oil fractions must be eliminated. These five undesirable constituents are: ( a ) paraffin wax, which must be removed t o obtain a low pour point; (b) asphalt, the removal of which is necessary for several reasons, including instability and excessive carbon-forming tendencies; ( c ) the heavy ends of the lubricating oils, which also have high carbon-forming tendencies; ( d ) the “naphthenic” compounds which are, in general, responsible for low viscosity indices and low resietance to oxidation; and ( e ) color bodies which must he remowd, primarily to

during the past few years. I t indicates that propane is a cheap, available, safe, and versatile solvent that can be used advantageously in every step of lubricating oil manufacture. A t low temperatures its properties are such that wax can be quickly and completely removed; at high temperatures, due to rapid changes in its physical properties, it may be used to precipitate various undesirable constituents, since it tends to eliminate all those compounds which the refiner wishes to remove from his raw lubricating stock. Propane refining makes readily available as by-products a whole series of high-melting-point waxes and petrolatums of extremely high quality and new types of asphalt of unusually desirable emulsification properties, as well as excellent ductility penetration relations. Unquestionably these products will make themselves felt commercially in the near future.