THE INFLAMMATION OF MIXTURES OF ETHANE AND AIR. 81 VIIL-The Inflammation qf Mixtwes of Ethane and Air in a Closed Vessel The Efects of Turbulence. By RICHARD VERNON WHEELER. WHEN describing the inflammation of mixtures of methane and air i t was noted that the speed a t which flame spreads through the mixture in a closed vessel is demonstrably dependent on the degree of mechanical agitation imparted to the mixture as indeed is the speed of flame in all combustible mixtures and under all conditions other than those existing during the propagation of the explosion wave. This important fact appears first to have been observed or a t all events first commented on by Schloesing and de Mondesir about the year 1864. Their experiments which involved an extended study of the mode of propagation of flame were carried out mainly with mixtures of carbon monoxide and air and were undertaken in connexim with a research on the working of gas engines.Mallard and Le Chatelier to whom the results of the experiments were communicated verbally have thus described the’m (Ann. des Mines, 1883 [VIII] 4 298): “Ces recherches ont inis en 6vidence un fait d’une grande im-portance l’influence de l’agitation du melange gazeux sur la vitesse de propagation de la flamme. Des melanges trhs lents (et par cette expression nous entendrona ceux dam lesquels la vitesse dei propaga-tion est faible) peuvent donner lieu B des propagations pour ainsi dire instantan& c’est-&-dire B de veritables explosions quand on provoque au moment de l’inflammation une agitation interieure tr6 82 WHEELER THE INFLAMMATION OF MIXTURES OF ETHAN.E AND vive telle que celle que I ’ m obtient en faisant d6boucher au milieu d’une mass8 gazeuse en r e p s un jet de gaz anim6 d’une grande vitesse.” These observations appear to hav’e been overlooked or forgotten until the subject of the agitation or turbulence of gaseous mixtures became of manifest importance during the investigation of gaseous explosions instituhed by the British Association f o r the Advance-ment of Science. New experiments on the subject by Dugald Clerk m d Hopkinson are recorded in the Fifth Report of the Committee on Gaseous Explosions (Rep. Brit. L 4 ~ ~ ~ ~ . 1912 201). To quote from hid Gustave Canet lecture (Junior Institution of Engineers 1913) Dugald Clerk “had long ago observed that gas engines would ’have been impracticable had the rates of explosion been the same in actual engine cylinders as in closed-vessel experi-ments.” During his experiments in 1912 he “found that the rate of explosion rise in the same engine varied with the rate of revo-lution increasing with increased number of rotations per minute, and was due to the turbulence or eddying caused by the rush of gases into the cylinder during the suction stroke which persisted during the compression stroke.” By drawing in a charge of mixture into the gas-engine cylinder in the ordinary way and then tripping the valves and compressing and expanding the charge for one or two revolutions before igniting it the turbulence was given time to die away.It was found that the effect of thus damping down turbulence was to retard the rate of inflammation of the mixture1 to a remarkable extent.For example with a mixture of coal-gas and air containing about 9.7 per cent. of gas ignition in a gas-engine cylinder under normal conditions a t the end of the first compression stroke (the engine being run a t 180 revolutions per minute) resulted in the maximum pressure being attained after 0.037 sec. ; whilst when ignition was a t the end of the third cornpression stroke after the charge had been expanded twice and turbulence had subsided the time taken foir the attainment of maximum pressure was 0.092 sec. Hopkinson experimented on the effects of turbulence a t the same time as Dugald Clerk using a cylindrical vessel 30.5 cm.in diameter and 30.5 cm. long. A small fan was mounted a t the centre of the vessel and comparison was made of the results of igniting similar mixtures with the fan at rest ahd in motion. With mixtures of coal-gas and air containing 10 per cent. of gas the times that elapsed between ignition and the attainment of maximum pressure were (1) with the fan a t rest 0.13 sec.; (2) with the fan running at 2,000 revolutions per min. 0.03 sec.; and (3) with the fan rucning a t 4,500 revolut’ions per min. 0-02 sec AIR IN A CLOSED VESSEL THE EFFECTS OF TURBULENCE. 83 Simultaneously with and independently of the experiments thus made on behalf of the Gaseous Explosions Committee of the British Association a problem under investigation for the Explosions in Mines Committee of the Home Office was found to involve a study of the effgcts of turbulence on the inflammation of gaseous mix-tufes.The prolblem was to determine the effect if any of the presence of incombustible dusts in suspension on the limits of inflammability of mixtures of firedamp and air. A series of experi-ments on the ignition of mixtures near the lower limit of inflam-mability was made with a spherical vessel of about 4 litres capacity (described in T. 1918 '113 855) provided with a fan which could be rotated a t a high speed so as to agitate the mixture and maintain dust in suspension. Naturally the fan was rotated whether dust was present or absent 301 as t o ensure that the comparative experi-ments required should be made under as far as possible identical conditions.The pronounced effect of turbulence or agitation of a gaseous mixture on the speed a t which flame travels through it thus became manifest for 'many experiments had previously been made with similar mixtures in the same sphere without the fan. The fan had four blades and was attached to a horizontal shaft passinq through an air-tight gland near the bottom of the sphere. Each blade extended for 7.5 cm. alonq the' shaft and had a maxi-mum width of 2.5 cm. the edge having a radius of curvature of 9.5 cm. The shaft was so fitted that there was a clearance of 1 cm. between the side of the sphere and the edges of the fan-blades. A slight helical twist was given to each blade. Several experiments were made with mixtures of ethane and air near the lower-limit of inflammability which with ignition a t the centre of a closed spherical vessel of glass of 2.5 litres capacity is 3.10 per cent.ethane. With 3.0 wr cent. of ethane flame travels slowly throughout nearly the whole of the (nm-turbulent) mixture in such a vessel; and with 2.9 and 2.95 per cent. of ethane flame spreads through about one-third of the mixture (T. 1911 99, 2026). It will therefore be realised that even thouqh a mixture may not contain sufficient ethane to ensure continued self-propaga-tion of flame part of the mixture may be burnt with a consequent development of pressure in a closed vessel. The earlier experiments with turbulent mixtures were made with the fan running a t 100 revolutions per second. The means of ignition was a secondary dischame (from a ' I 10-inch " X-ray coil) across a spark-gap of 12 mm.a t the centre of the sphere produced by breaking a current of 10 amperes in the primary circuit of the coil the trembler being locked. Such a dischayge is more thah adequate to ignite any inflammable mixture of ethane and air whe 84 WHEELER THE INFLAMMATION OF MIXTURES OF ETHANE AND the mixture is still yet it was found that no ignition or rather, no propagation of flame took place with a mixture of ethane and air containing as much as 3.2 per cent. of ethane when that mix-ture was agitated by the fan a t 100 revolutions per s p n d . On stopping the fan and allowing the turbulence to subside ignitim took place readily with complete inflammation of the mixture and the development of a pressure of 3.4 atmospheres.Similarly with mixtures containing 3-15 and 3.05 per cenb. of ethane no ignition could be obtained whilst the fan was running (at 100 revolutions per second) however frequently the discharge was passed although when the mixtures were free from turbulence ignition occurred on the first passage of the discharge. Details of these and similar experiments are as follow: Ethane in mixture. 3-20 Per cent. Result. No ignition when the fan was running a t 100 revolutions per sec. With the fan a t 40 revolutionsper sec. ignition took place a pressure of 4.5 atm. being recorded 0.25 sec. after ignition. Without the fan running a pressure of 3.4 atm. was developed. No ignition could be obtained when the fan was run-ning at 100 revolutions per sec.Without the fan ignition occurred at once a pressure of 3.2 atm. being recorded. With the fan at 40 revolutions per sec. ignition occurred on the fourth passage of the discharge. With the fan at 20 revolutions per sec. ignition occurred a t once. A pressure of 4.4 atm. was developed on both occasions, 0.177 sec. after ignition in the first experiment and 0.287 sec. after ignition in the second. No ignition could be obtained when the fan was run-ning at 100 revolutions per sec. Without the fan ignition occurred at once and a pressure of 2.8 atm. was recorded. No ignition with the fan a t 100 revolutions per sec. With 20 revolutions per sec. ignition occurred a t once and a pressure of 4.3 atm. was recorded 0.30 sec. after ignition.With the fan running a t 20 revolutions per sec. ignition occurred when the discharge was maintained (the trembler of the coil being in action). A pressure of 4.2 atm. was recorded. 3.15 3.10 3.05 3.00 2.95 Strong agitation of a mixture poor in combustible gas renders it difficult to ignite or t o be precise renders it difficult for the flame that no doubt occurs during the passage of the discharge to spread away therefrom and travel throughout the mixture. This difficulty increases as the degree of agitation is increased and as the prcenb age of combustible gas is decreased. When however the flame in such an agitated mixture does manage t o spread away from the source of ignition it travels rapidly. From the high pressure developed when a mixture was ignited that contained 2-95 per cent.of ethane and t o which turbulenc AIR IN A CLOSED VESSEL THE EFFECTS OF TURBULENCE. 85 had been imparted by a fan running a t 20 revolutions per second, it seemed that flame must have travelled through a greater propor-tion of the mixture than the one-third observed when the mixture was quiescent. An apparatus was therefore devised to enable the appearance of tho flames in turbulent mixtures to be examined. The apparatus which consisted essentially of a globe of glaea of about 4 litres capacity is shown in Fig. 1 and needs no descrip-tion. Preliminary experiments were made to determine the direc-tion of the air-currents induced by the fan which had two helical blades and revolved on a vertical axis. From the behaviour of coloured powders introduced into the globe while the fan was spinning it appeared that air was drawn from the centre of the globe towards the axis of the fan and was discharged a t the periphery of the latter as a spiral current directed obliquely* around the walls of the globe.Mixtures of methane and air were used for the experiments. Normally the lower-limit for central ignition of methaneair mix-tures in a closed sphere is 5.6 per cent. methane; the flame travels upward from the spark a t the centre until it occupies onethird of the vessel when it travels downwards as a horizontal disk to the bottom. The appearance of the flames in mixtures containing less than 5.6 per cent. of methane is shown in Big. 3 T. 1911 99, 2025. When a 5.6 per cent.mixture of methane and air was agitated by spinning the fan a t about 50 revolutions per second a succes-sion of discharges from an induction coil the trembler of which was in operation in the usual manner apparently failed to cause ignition. On close observation however it was seen that a pointed tongue of flame appeared a t each passage of the discharge directed dowrzwurds towards the axis of the fan apparently drawn thither by the current. The flame was ab40ut 2 cm. long and formed a sharp-pointed cone having the spark-gap (12 mm. in length) as its base. Occasionally if the discharge were maintained a fine fila-ment of dame darted rapidly over a distance of a few an. towards the fan. The speed of the fan was now reduced to about 30 revolu-tions per second and a discharge passed across the gap.The sequence of events was too rapid to be followed by the eye. It was observed that a downward-pointing tongue of Aame was produced as before and that this tongue after some hesitation shot towards the axis of the fan; the whole vessel then seemed to fill with flame and the glass was shattered into' powder. Further experiments were made with mixtures containing less methane. On two occasions the globe was shattered owing to the * No doubt owing to an unequal setting of the blades of the fan 86 WHEELER TEE INPLAMMATION OP MIXTURES 0.E gTHANE Am rapidity with which the mixture contained in it was inflamed but in a number of experiments notably in several with a mixture containing 5.0 per cent. of methane (see T. 1914 106 2595) the movement of the flame could be followed; or a t all events owing to the persistence of retinal impressions the course taken by the flame was apparent.An attempt has been made to indicate tho appearance of the flame to the eye a t a given instant by the shaded additions to Fig. 1. The impression produced can be described as that of a spiral whirlwind of flame the axis of the spiral being inclined a t an angle; in effect the flame seemed to follow the course of the current induced by the fan. It appeared alw that the flame passed several times through the mixture before i t finally died away a t the centre of the sphere. Analysis of the products of combustion of the 5.0 per cent. mixtures of methane and air showed that all the methane had been burnt.There can be little question as a result of these observations that the action of the form of turbulence studied in causing an enhanced speed of combustion of a weak inflammable mixture of methane or ethane and air within a closed vessel is purely mechanical. The flame which normally would be propagated mainly by conduction of heat from a burning t o an unburnt “layer” of mixture is forcibly dragged in the wake of the rapid current induced by the fan bcrning the mixture in its path. The difficulty experienced by the flame in sucli weak mixtures in travelling away from the source of ignition if the speed of the fan is very great is no doubt due to the fact that mixtures of the paraffins with air exhibit a considerable “ time-lag ” when the temperature of the source of heat that causes ignition is but little above the ignition-tempera-ture a condition obtaining with the flames of limit mixtures.With richer mixtures in which flame normally spreads a t an equal speed in all directions from the source of ignition the action of turbulence is mechanical also. To quote Mallard and Le Chate-lier (Zoc. cit. p.. 350) : “Lorsque le gaz dans lequel progresse la flamme est B 1’Btiit d’agitation la vitesse de propagation augmente parceque la chaleur se transmet non seulement en vertu de la conductibilit6 du melange gazeux mais encore en vertu des differences de vitesse des diverses parties de la masse. La surface de la flamme au lieu de garder une forme constante et rBguliSre se deforme ii chaque instant augmente de largeur en multipliant les points d’inflammation et par suite en rendant plus rapide la progression de la combustion.” If this explanation is correct it follows that (1) the greater the turbulence the more rapid should be the combustion; and (2) II'o face p .S AIR IN A CLOSED VESSEL THE EFPECTS OF !KJRBULENCE. 87 mixture in which the speed of flame normally is slow should be more susceptible t o the effects ,of turbulence than one in which the speed of flame normally is rapid. The first deduction has received experimental verification by Hopkinson whose results have already been quoted. His results are confirmed by a series of experiments in the 4-litre sphere with mixtures of ethane and air containing 3.85 per cent. of ethane, the timeipressure curves f o r which are reproduced in Fig.2. The time-intervals between ignition and the attainment of maximum pressure were mixture a t rest 0.146 sec.; fan running a t ( a ) 20 revs. per sec. 0.091; ( b ) 40 revs. per sec. 0.070 sec.; ( c ) 100 revs. per sec. 0.045 sec. Additional points that should be FIG. 2. 6 Time seconds. Time O=tiine of ignition. noted as regards these curves are (1) the slight increase of pres-sure obtained with the turbulent mixtures (a) and ( b ) and the marked increase with the turbulent mixture ( c ) as compared with that produced by the quiescent mixture; and (2) the disappearance from the curve fo,r turbulent mixture ( c ) of 'the1 horizontal portion a t maximum pressure noticeable in the other three curves. An explanation of these effects is offered later.I n order t o test the second deduction that should follow if the explanation suggested for the action of turbulence is correct two series of experiments were made with mixtures of ethane and air ranging between the lower-limit mixture and that giving the maxi-mum pressure on combustion. I n the one series the fan was run a t a constant speed of 100 revolutions per second; in the other th 88 WHEELER THE INFLAMMATION OF MIXTURES OE’ ETHANE AND Time seconds. Time O=time of ignition. Time seconds. Time O=time of igniton. * It should be noted that the unit of time employed in plotting the curves This contraction of t8he time- in Fig. 3 (and Fig. 2) is double that in Fig. 4. scale is rendered necessary from considerations of space AIR IN A CLOSED VESSEL THE EFFECTS OF TURBULENCE.89 turbdent mixtures occupying the left-hand portion of each diagram. From these curves the time that elapsed between ignition and the attainment of maximum pressure for each mixture can be deter-mined. These times together with the times for mixtures not included in Figs. 3 and 4 are recorded in the table that follows: Ethane in mix-ture. Per cent. 3.30 3.45 3.60 3-80 3-85 4-05 4.30 4.35 4-60 4-65 4.70 4.80 5.00 5.25 5.35 5-60 5.95 6.00 6.40 6-45 6-75 7-06 7.15 Time between ignition and the attainment of maximum pressure. Seconds. 7- Without With turbulence. turbulence. - 0.176 - 0.096 0-332 -0.152 -0-146 0.045 0-124 0.036 - 0-033 0.094 -- 0.026 0.073 -- 0.029 0.070 -0-063 0-024 - 0.021 0,064 0.020 0.052 - 0.019 0.0465 -- 0.019 0,046 -0.0466 0.019 0.050 0.020 0.082 -It has been shown (T.1918 113 852) that these time-inhrvals can be used to' calculate for each mixture the mean speed of propagation of flame between the' centre and the top of the sphere, a distance of 9.75 cm. The speeds thus calculated are shown plotted against percentages of ethane in Fig. 5. Allowing for the irregularities which are naturally more noticeable with the tur-bulent than with the quiescent mixtures the speeds for equivalent percentages of ethane in the two sets of experiments as deduced from the smoothed curves are given in'the table on p. 90. The conclusion that a mixture1 in which normally the speed of flame is slow should be affected by turbulence to a greater extent than one in which normally the s p e d of flame is rapid is thus proved experimeiitally by the gradual diminution in the value of the ratio B / A .Tk e DeveZoi~.rne~~t of Pressure.-On referring t o the time-pressure curves for mixtures without turbulence given in Figs. 3 and 4 an 90 WHEELER "HE INFLAMMATION OF MIXTURES OF ETHANE AND FIG. 5. comparing them with the curves for mixtures of methane and air previously published (Zoc. cit. Fig. 2 p. 847) i t will be seen that Mean Speed of Propgation of Flame f r o m C'etztre t o Top of Sphere. Cm. per see. Ethane in Without With mixture. turbulence. turbulence. Pep cent. ( A ) . (B). Ratio RIA. 3.6 35 142 4.06 3.8 55 195 3.54 4-0 75 237 3-17 4.2 95 2 84 2.99 4.4 112 320 2.85 4-6 129 360 2.79 4.8 144 400 2.77 5.0 158 430 2-72 5.2 172 462 2-68 5.4 185 485 2-62 5.6 195 500 2.56 5.8 202 510 2.52 6.0 210 518 2.47 6.5 212 518 2.44 6.7 200 495 2.47 both sets of curves are of the same type.All the mixtures o AIR IN A CLOSED VESSEL THE EFFECTS OF TURBULENCE. 91 ethane and air up to and including that containing 5.6 per cent. of ethane have time-pressure curves which exhibit the three stages of development noticeable with the mixtures of methane and air. The explanation of these stages offered when describing the methane curves can be applied also in the present instance, Support is given to the assumption then made that the second stage of development during which the recorded pressure remains constant represents a balance between a gradual decrease of pres-sure that begins as soon as inflammation 'sf the mixture is complete and is due to cooling by the walls of the vessel and an increase of pressure incident a t the same moment and due to the gradual attainment of thermal equilibrium.For it will be found that a graphical " correction " applied in conformity with this assump-tion in the manner described (Zoc. cit. p. 849) yields results for the maximum pressures in close agreement with the maxima recorded by equivalent mixtures when turbulent over the whole range from 3.80 per cent. ethane (at and above which percentage the flame travels from the centre in all directions a t the same speed) upwards . This is best shown in Fig.6 where the observed maximum pres-sures for all the mixtures experimented with both turbulent and quiescent and the " corrected '' maxima for the latter are shown plotted against percentages of ethane. It should be observed that the magnitude of the correction as is to be expected diminishes in proportion as the speed of inflammation of the mixture increases. Similarly the magnitude of the difference between the maximum pressures recorded with like mixtures when turbulent and quiescent also decreases as the speed of inflammation of the latter increases, until with mixtures containing mom than 5.6 per cent. of ethane no difference is observable between the two sets of pressures. Further the crests of the time-pressure curves for the quiescent mixtures that contain more! than 5.6 per cent.of ethane no longer remain hori-zontal over a measurabl'e length of time but the cooling curves begin as soon as the maxima are attained. Pier (Zeitsch. Elektrochem. 1909 15 536) who used the pres-sures developed by the inflammation of different mixtures in a closed veasel to determine the specific heats of various gases has made observations which have a bearing on the question of the effecta of turbulence. Using a manometer of similar construction t o the Petavel gauge (Phil. Mag. 1902 [vi] 3,461) Pier found exact agreement between the observed and the calculated pressures produced by mixtures the combustion-temperatures Qf which exceeded 1 6 0 0 O . For this reason he combatted Nagel's opinion ((' Versuche uber Zundgeschwindig 92 WHEELER THE INFLAMMATION OF MIXTURES OF ETHANE AND heit explosibler Gasgemische,” Mit teilungen iiber Forschungs-arbeiten des Zngelzieurwesem Vo,l.54 1908) that with central igni-tion in a spherical vessel the mixture near the walls must be raised in temperature by adiabatic compression before flame reaches i t FIG. 6. I ‘7 / ‘ I c, P‘c 0% 0 o x 0 1 I / I -2-5 3 4 5 6 7 7 Ethane per cent. Quiescent observed x , corrected Turbulent observed @ (an opinion that had already received experimental verification by Hopkinson) and suggested that the interchange of heat between different portions of the mixkxre within blie vessel must be practi-cally in&antaneous AJB IN A CLOSED VESSEL THE EBFECTS 02’ TURBULENCE. 83 This result Pier supposed would be effected by a rapid whirling and mixing of the contents of a spherical vessel owing to a sudden increase of pressure on ignition a t the centre.It is clear if only by reason of the difference observable in the character of the time-pressure curves for ethane-air mixtures with and without arti-ficially-produced turbulence that Pier’s contention cannot be cor-rect; and Hopkinson’s measurements of the temperatures within a closed cylindrical vessel a t the moment of maximum pressure pro-duced by the inflammation of a mixture of coal-gas and air (Proc. Roy. SOC. 1906 [A] 77 387) should have convinced Pier of its falsity. . I n the absence of knowledge regarding the composition of the products of combustion a t the moment of attainment of maximum pressure when the ethane-air mixtures contain excess of ethane it is not possible to calculate the theoretical pressures that should be given by such mixtures on ignition in a closed sphere were there no loss of heat during the propagation of flame.Calculation can, however be made for those mixtures in which the combustion of ethane can be presumed to be complete. The mixture of ethane with air in which ethane and oxygen are in the theoretical propor-tions for complete combustion to form carbon dioxide and steam contains 5-63 per cent. of ethane. The dotted line in Fig. 6 repre-sents the calculated maximum pressurm mer the range 3-8-5.5 per cent. ethane.* It will be seen that a loss of heat of between 9 and 12 per cent. presumably due to radiation during the propagation of flame is indicated.A matteq for further study is the fact that the mixtures of ethane and air which produce the highest pressuretj are not those within close range of the mixture containing ethane and oxygen in theo-retical proportions for complete combustion (5.63 per cent. of ethane) but lie over a considerably higher range namely 6.5-7-0 per cent. The time taken for the attainment of maximum pressure reaches a minimum over the same range or in other words the speed of propagation of flame under the conditions of the experi-ments is fastest in mixtures containing between 6.5 and 7.0 per cent. of ethane. I n this respect the results obtained with mix-tures of ethane and air differ markedly from those with methane and air. Further comparison of these results with those obtained with mixtures of methane and air is reserved for a future communica-tion which will include the results of similar experiments with other n embers of the paraffin series of hydrocarbons. * The cdculations were made in the manner described in T. 1918,113, 868 using Langen’s values for the specific heats of the gases 94 MORGAN THE IGNITION OF EXPLOSIVE E x P E R I M E N TAL. The apparatus used (Clitre sphere) and general method of pro-cedure f o r the experiments has already been described (Zoc. cit., p. 854). The ehane was prepared by the action of water on zinc ethyl and was purified by liquefaction by liquid air; the ratio C I A on explcsion analysis was 1-25 showing that it contained no impurity. The majority of the experiments described in this paper were carried out during the year 1912 with the assistance of Mr. M. J. Burgess. [Received November lBth 1918.