340 Messrs. Bairstow and Alexander .
Explosions of [ Feb. 7 from 10'-9 to \#151 ; 188 ' , thus giving the latent heat of air as 53'63 .
These last values come very near those found by Fenner and Richtmyer ; but the results are so varied that it is clear the question of the latent heat of air of very high oxygen-concentration is one requiring further investigation .
Explosions of Mixtures of Coal-Gas and Air in a Closed Vessel .
By Leonard Bairstow , A.B.C.Sc .
, Wh .
Sch .
, and A. D. Alexander , AR .
C.Sc .
, Wh .
Sch .
( Communicated by Professor Perry , F.R.S. Received February 7 , \#151 ; Read March 9 , 1905 .
) ( Abstract .
) The experiments were proposed by Professor Perry , and with his sanction and encouragement extended to more than two years ' continuous work .
The work had for its original object the determination of the explosive properties of compressed mixtures of coal-gas and air .
The main apparatus was designed by Messrs. McDiarmid and Mann , students of the Royal College of Science , South Kensington , and was made before the authors of this paper became connected with the work .
INDICATOR Bourdon to MERCURY A GAUGE to GAS-HOLDER Induction COIL AIR-RESERVOIR and AIR-PUMP IETER to CAS " I SUCTION-PUMP INDICATOR TUBE TUBL EXPLOSION CHAMBER MIXING \\\N !
k\\\\\\\\\\\\\ : DRAIN 1905 .
] Mixtures of Coal-Gas and Air in a Closed Vessel .
341 In the preliminary experiments it was found necessary to mix the gases independently of diffusion , and most of the experiments have consequently been made at an initial pressure of 35 lbs. per square inch absolute , as this allowed a considerable volume of air to be pumped into the cylinder after the coal-gas had been admitted .
Eventually it was found necessary to take still greater precautions in special cases , and mechanical mixing was resorted to .
The figure shows the explosion cylinder , which was 18 inches long and 10 inches in diameter internally .
The records were taken with an ordinary indicator , the spring being usually 150 lbs. per square inch per inch .
The recording drum was made to revolve continuously at a speed giving 42-5 inches of diagram per second .
By varying the firing arrangements the mixtures could be ignited at different points , and a further deviation from previous work was made by altering the length of the tube which led to the indicator .
The firing tube shown in the figure was closed at the top where the spark was produced , and communication with the cylinder took place through a pin-hole near the bottom .
The proportion of coal-gas to air was kept constant , and using the above tube the mixture was fired at different points .
In general , changes resulted both in the time of explosion and in the maximum pressure .
For the most explosive mixture the change in the maximum pressure was small ( fig. 8 of original paper ) , being only of the order of 1 per cent. , but in weak mixtures the variations sometimes exceeded 30 per cent. Such differences can only be observed in a vessel which absorbs heat , but it does not thereby follow that the differences are entirely due to the heat lost by the gases before the highest pressure is reached , since the maximum pressure does not coincide with the end of combustion when the rate of cooling approaches the rate of reception of heat by explosion .
This and the cooling loss can be differentiated in some of the diagrams obtained , and the experiments were used as a guide to correction .
Convection currents were also indicated by the method of firing , by explosions starting at the bottom of the cylinder being quicker than an otherwise similar one beginning at the top .
The currents are sometimes very considerable .
In order to reduce the cooling effects , the explosion was often commenced at four points in the axis of the cylinder .
The four spark-gaps were in one continuous circuit , and the sparks , therefore , occurred simultaneously .
Keeping the initial pressure constant the mixture was varied .
The highest pressure and the most rapid explosion was produced when the oxygen of the air was just completely burnt .
The pressure fell continuously as the pro2 b 2 342 Messrs. Bairstow and Alexander .
Explosions of [ Feb 7 portion of coal-gas to air was diminished , until the fractional volume of the mixture occupied by the coal-gas was about 1 in 12 .
A sudden change then Fig. 8 .
End of Indicator Tube at top of Cylinder Rich Mixture 0.142 Initial Pressure 3SlbS^-ns* depth of pinhole ( inches ) ,200- ioo- ioo- ir ioo200- Time Seconds 200- IOO- 200- IOO- THE dotted diacrams are DUE TO A SECOND INDICATOR RECORDING ON THE SAME DRUM O A .
Real Beginning , of ignition XB .
First movement of pencil FOR full CURVE 1 c. Fl RST movement of pencil FOR DOTTED CURVE occurred , and combustion remained incomplete after explosion .
Mixtures between 1 in 12 and 1 in 17 were still explosive , but the amount of gas burnt 1905 .
] Mixtures of Coal-Gas and Air in a Closed Vessel .
343 decreased , at first suddenly and then regularly until the limit of inflammability was reached .
If then , the gases remaining in the cylinder , after an explosion of a mixture less rich in coal-gas than 1 in 12 were used instead of air in a subsequent explosion , higher pressures should result .
This was observed to be the case , and with a necessary modification furnishes the explanation of the experimental results obtained by Mr. Grover in 1895.* In the diagrams of fig. 8 ripples are shown between the beginning of explosion and the time at which the maximum is reached .
It is obvious from their position that they are not due to the inertia of the indicator .
With four sparks or a single spark at the top of the cylinder these occur early in the diagram , and are not usually very noticeable .
When , however , the point of firing is altered , the change is accompanied by a difference in the position of the ripples .
A measurement of them , when sufficiently uniform to be measured , showed that the period was independent of the pressure at which they occurred , and eventually they were traced to sound waves in the indicator tube produced by the arrival of the flame at its open end .
All the observed conditions are then satisfied .
Supposing then that the end of the indicator tube is in the upper surface of the cylinder ( as in the cases of fig. 8 ) , and that the mixture is fired below the middle of the cylinder , the flame will reach the bottom first and explosion will be completed when the ripples occur , that is they will occur on the top of the diagram , as in the figure .
The conclusion as to their cause was confirmed by altering the length of the indicator tube and noting the change produced in the position and period of the ripples .
When the ripples occurred at the maximum pressure the amplitude was sometimes enormous , the extra pressure due to the ripples being often as great as the pressure due to combustion in the cylinder .
This phenomenon is probably connected with the detonation wave investigated by MM .
Mallard and Le Chatelierf and by Professor H. B. Dixon4 There is no evidence of the deconation wave in the cylinder itself in any of our experiments .
The big movements are never found with weak mixtures , or even in rich mixtures fired with short tubes .
With weak mixtures , even when combustion is complete after explosion , the ripples indicate that the flame travels through the more inflammable portions of the mixture first , and therefore that the constituents burn * " Modern Gas and Oil Engines , " Grover .
+ ' Ann. des Mines , ' 1883 .
I ' Phil. Trans. , ' A , 1903 .
344 Messrs. Bairstow and Alexander .
Explosions of [ Feb. 7 , successively .
This type of combustion can be detected until the coal-gas occupies more than one-eighth of the volume of the mixture .
As explosion begins very slowly , a small amount of friction in the indicator introduced an appreciable error in the time of explosion .
This was estimated by making an external gap in the spark circuit which included the indicator paper .
By this means a hole was pierced in the paper at the time of sparking and ( fig. 8 ) shows the amount of error in a particular case .
( The necessary corrections are given in the full paper , which is preserved in the Archives of the Royal Society .
) Experiments on Compressed Mixtures .
These experiments form an extension of the work of Mr. D. Clerk in 1886* The arrangement of four sparks was found to produce the most consistent results , and it was then noticed that slight changes in the composition of the coal-gas affected the experiments , particularly the rate of cooling .
To make comparative experiments , therefore , it is necessary to do them at the same time and on the same gas .
Three series are given , each the result of one day 's experiments .
The gas used in the first two is the same , but different to that used in the third series .
Series 1 ( fig. 11 ) .
Mixture constant ( volume of coal-gas divided by the volume of air = 0T68 ) .
Initial pressure varied from \#163 ; to 3 atmospheres .
Series 2 ( fig. 12 ) .
Mixture constant ( 0105 ) .
Initial pressure as in Series 1 .
Series 3 ( fig. 13 ) .
Initial pressure constant .
Mixture varied .
( Tables accompanying these figures are given in the full paper .
) From fig. 11 it will be seen that for a mixture of 0168 the time of explosion is almost independent of the initial pressure between one and three atmospheres , but considerable increase is noticed for the two lower pressures .
Similarly , between the same limits the ratio of the maximum to the initial pressure is constant .
With the weaker mixture the time of explosion increased as the initial pressure decreased , without any decided effect on the ratio of the maximum and initial pressures ( fig. 12 ) .
, The mixtures for Series 3 are such that the volumes of coal-gas burnt * 'Proc .
Inst. Civil Eng. , ' 1886 .
1905 .
] Mixtures of Coal-Gas and Air in a Closed Vessel .
345 Fig. 11 .
Pressure MAX ?
-4eu\gt ; 3r.-.AT O.IOSEC ao2 0.06 0.03 0.04 Time Seconds .
0.05 increase in arithmetical progression .
These experiments show clearly that the heating value of unit volume of coal-gas increases as the mixture gets weaker ( fig. 13 ) .
Fig. 14 gives the cooling curves for the three series .
By analysing the curves it will be seen that the rate of cooling increases with increase of initial pressure , but is independent of the relative proportions of coal-gas and air .
Hypotheses Introduced in the Calculation of Explosive Phenomena .
The outstanding differences between calculated and experimental results can now be explained by dissociation or an increasing specific heat .
The hypothesis of a specific heat increasing with temperature is based entirely , so 346 Messrs. Bairstow and Alexander .
Explosions of [ Feb. 7 Fig. 12 .
Pressure 020 0.30 Time Seconds .
Fig. 13 .
200 PR 100 C A r es : J.RE / LSS INS* J 1/ / ) o. 1 o. 2 O. 3 O. Time 4 O. Seconds 5 O. 6 O. T O. , 3 0.9 far as experimental evidence is concerned , on the work of MM .
Mallard and Le Chatelier .
They examined their cooling curves carefully in the search for a discontinuity at the lower limit of dissociation .
Fig. 15 shows an analysis 1905 .
] Mixtures of Coal-Gas and Air in a Closed Vessel .
347 Fig. 14 .
LlI ( D iO Hi ^ ,1 l.l \#151 ; ui u in of two of their curves.* The lower curve shows the discontinuity in question at C. In both explosions concerned in diagram 15 , the products are the same * ' Ann. des Mines , ' 1883 .
348 Messrs. Bairstow and Alexander .
Explosions of [ Feb. 7 , Fig. 15 .
^Mixture BEFORE EXPLOSION 2H2+Oz+ 2COz 49.1 i d a Mixture BEFORE EXPLOSION 2\\ + 2.CO + 2.0* ( water VAPOUR NOT condensed ) UNIT OF TIME 0.305 SECS .
po =s PRESSURE AFTER CONDENSATION both in composition and mass .
In comparing results during cooling therefore , equal pressures mean equal temperatures .
The rate of loss of pressure divided by the pressure is shown as the ordinate of the figure and the ratio of the two ordinates for any given abscissa , therefore gives the ratio of the rates of cooling in the two cases .
Where is 300 , the rate of cooling in one case is 50 per cent , greater than in the other .
The only apparent difference between two such cases lies in the maxima reached , and the only other experiments which afford any evidence on this point* agree in showing that the higher the maximum pressure and temperature , the lower the rate of cooling at any given subsequent temperature and pressure .
The rise in the temperature of the cylinder walls would produce just such a change .
A mathematical and experimental investigation showed that the metal itself does not increase in temperature greatly , and a film must therefore exist on the surface which sometimes attains a temperature of several hundred degrees Centigrade .
Such a hypothesis satisfies all the observed experimental conditions , whilst it would be exceedingly difficult , if not impossible , to explain either the upper steep part of the curve in fig. 15 , or the difference in the rates of cooling for the two cases , on the idea that C is the limit of dissociation .
Experiments at ordinary temperatures have not shown any such increase in specific heat as is necessary for the above hypothesis .
In order to determine the fraction of heat developed , the composition of * 'Ann .
des Mines , ' 1883 , p. 427 .
1905 .
] Mixtures of Coal-Gas and Air a Closed Vessel .
349 the coal-gas and its calorific value were obtained .
For the richest mixture ( 0-184 of coal-gas to 1 of air ) the pressure obtained was 657 per cent , of that calculated .
This fraction increased as the mixture was weakened , and was about 80 per cent , when combustion became incomplete .
The highest temperature for the richest mixture is 2430 ' C. absolute , and the heat developed 63*2 per cent. On the hypothesis of increasing specific heat , the temperature would be 6 per cent , greater .
Summary .
Mixtures of coal-gas and air are not inflammable until the volume of coal-gas is greater than one-seventeenth of the combined volumes .
Only a very small fraction of the gas then burns , the amount burnt rapidly increasing with increased richness of the mixture until the coal-gas is one-twelfth of the total volume .
The least inflammable of the constituents then burns , and combustion becomes and remains complete so long as air is in excess .
In these latter cases it is still probable that the constituents burn successively and not simultaneously .
The hypothesis of a specific heat increasing with temperature is not supported by direct experiment , and cannot be proved by any work on the pressures produced by explosion , the authors believing that a proof would require the measurement of temperature .
Direct experiments by Deville at temperatures below 1400 ' C. have shown that both steam and carbon dioxide are partially decomposed , and this dissociation is therefore taken by us as the sole explanation of the difference between the pressures calculated for explosions in a closed vessel and those actually obtained .

