DILUTION LIMITS OF INFLAMMABILITY OF GASEOUS MIXTURES. 27 1.-The Dilution Limits of In$ummability of Gaseous Mixtures. Part III. The Lower Limits of some Mixed Injarrimable Gases with Air. Part IV. The Upper Limits of some Gases Singly and Mixed in Air. By HUBERT FRANK COWARD CHARLES W~LLIAM CARPENTER and WILLIAM PAYMAN. PART 111. IN Part I of this series of papers (Coward and Brinsley T. 1914, 105 lS59) the wide variation in the values assigned by different observers to the limits of inflammability of hydrogen and other gases in air was shown to be due to the very different criteria of inflammability used. The meaning of the term " inflammability " was therefore discussed and it was concluded that inflammability could and should be regarded as a characteristic property of a gas mixture apart from the precise nieans used f o r ignition and from the form of the vessel that might happen to bel chosen for experimelnt.It was argued that a gaseous mixture should be termed inflamniabls per se a t a stated temperature and pressure if and only i f it were capable of indefinih self-propagation of flame while1 the unburnt portion of the mixture was maintained a t the stated temperature and pressure.;+ I n Part I1 (T. 1914 105 1865) the1 lower limits of inflamma-bility of hydrogen methane and carbon monoxide individually in air were examined experimentally by means of apparatus speci-ally designed to enable1 the progress of flame to be observed in much wider and longer vessels than had been hitherto employd.Jr The presemtl paper records the results of experiments carried out to determine the lower limits in air of various mixtures of hydrogen methane and carbon monoxide taken two a t a time or * This definition has been discussed by Burgess and Wheeler (T.1914, 105 2591). Several other papers on the subject of " dilution limits of in-flammability ¶ ' have appeared since Coward and Brinsley's paper was published but as they have not been concerned with the question of indefinite propagation of flamc but mercly with tho inflammation of very liinitod ninount,~ of gasoous mist;lircs they 1ia.w no direct bearing on tho present inquiry. f Burell and Oberfell (U.S.A. Bureau of Mines Y'echrhicwl Paper No. 119 1915) havo adopted a oudiomster of the same Eiize 8s that used by Con.nrc1 and Brinsloy 28 COWARD CARPENTER AND PAYMAN THE DILUTION LIMITS all three together and finally of the complex mixture a “town’s gas.” A simple formula of purely additive character has been put forward by Le Chatelier to connect the lower limita of single gases with the lower l’ta of mixtarea of them.This formula origin-ally limited to binary mixtures is generalised thus: = 1. where N, N2 Ar3 . . . are the lower limits in percentages of the whole air mixture for each combustible gas separately nl n2, m3 . . . are the proportions in percentages of the whole air mix-ture of each combustible gas a t the dilution limit. The percentage of total combustible gas present in the limit mixture is thus This formula enables the lower limit L of a combustible mix-ture to be calculated from the dilution limits of its several con-stituents.I f the proportions of each of the combustible con-stituents are pl p2 p3 . . . so that pl+p2+p3+ . . . =loo a simple transformation gives its dilutdon limit in air as L = T + m 2 + % + . . . The physical meaning of the formula may be best appreciated by the consideration of a particular case. A mixture of air carbon monoxide and hydrogen which contains onequarter of the amount of carbon monoxide necessary to form a lower limit mixture, together with three-quarters of the amount of hydrogen necwsary, will be a lower limit mixture. In other words the lower limits of inflammability form a series of inflammability equivalents for the individual gases of a mixture. It may also be deduced from the formula that lower limit air mixtures if mixed in any proportions give rise to mixtures which are also a t their lower limits.Experimental support for the formula rests on observations by Le Chatelier ( A n m . des Mimes 1891 [viii] 19 388) with three mixtures of methane and coal gas by Ls Chablier and Boudouard (Compt. rend. 1898 126 1344) with three mixtures of hydrogen and carbon monoxide and with one mixture of aaetylene and carbon monoxide and by Eit’ner (Habilitationsschrift Miinchen, 1902) with hydrogen and carbon monoxide in equal volumes and with coal gas. The difference between the calculated and the observed dilution limits rarely reached one twenty-fifth part o OF INFLAMMABILITY OF GASEOUS MIXTURES. PAET 111. 29 the combustible mixture.None of these experiments however, was carried out in apparatus large enough to indicate whether the mixtures used were capable of continued propagation of flame. Thus Le Chatelier and Boudouard used for the lower limit of hydrogen ifi air the figure 10 per cent. whereas the recent experi-ments have shown that mixtures containing upwards of 4.1 per cent. of hydrogen are capable of propagating flame apparently indefinitely in an upward direction. Le Chatelier had in fact, found that the 10 per cent. hydrogen mixtare was the weakest which would propagate flame downwards through a somewhat narrow tubel. It is only necessary to ignite from below to produoe a self-propagating flame in mixtures considerably weaker in hydrogen. The experiments now t o be described were therefore carried out in the eudiometer previously described (Coward and Brinsley Zoc.cit.) 1.8 met’rres (6 feet) in length and 30 cm. (1 foot) square in section with a capacity of 170 litres. In each case the mixtures under experiment were saturated with moisture a t 18-19O and were maintained a t approximately atmospheric pressure during inflammation. The source of ignition was a spark gap of variable length between small platinum spheres. A 6-inch Apps coil with two four or six storage oells was used t o produce single sparks. The various gases were pre-pared in a state of purity and each mixture with air was made to the desired composition which was checked by the analysis of samples taken just before firing. E X P E R I M E N T A L . The lower limits of a number of mixtures of hydrogen carbon monoxide and methane taken two or three together and also of a “itown’s gas,” are recorded in table I (p.30) together with the lower limits calculated by means of the Le Chatelier formula from the limits of the individual gases. Several of the experimental results differ from those calculated by amounts exceeding the errors of observation and experiment; nevertheless the formula givw a useful approximation over the whole range of mixtures examined and may be applied therefore, to water gas and to coal gas as well as to town’s gas. The most striking anomaly was shown by the mixture contain-ing 10 per cent. of hydrogen and 90 per cent. of carbon monoxide, where the large difference was in the opposite direction to that usually noted.This anomaly was more pronounced in experi-ments with the same mixture in a narrower tube (5 cm. diameter) 30 COWARD CARPENTER AND PAYMAN THE DILUTION LIMITS TABLE T . Composition of gas (before admisture with air). TAower limit, of iiiflainmability in air. Carbon Hydrogen. monoxide. 100 -75 25 50 50 26 75 1 0 90 I 0 0 100 -33.3 33.3 55 15 ” Town’s gas ” Methane. ----1 0 25 50 60 75 100 76 50 25 10 -33.3 30 Observed. Per cent. 4.1 4.7 6.03 8-2 10.3 12-5 11.0 9.6 7 . 7 7.2 6.4 5.6* 4.7 4.6 4.1 4.1 4.1 5-7 4.7 5.35 Calculateti. Per cent. I 4.9 6.2 8.3 10.4 __ 1 1 . 1 !).ci 7.7 7-1 6.5 I 5.0 4.7 4.4 4.2 -6.0 5.0 5-36; DiHerenco.Per cent. -- 0.2 -0.15 -0.1 + 0.4 - .--0.1 -0.1 0.0 +o-1 - 0.1 -- 0.3 -0.1 - 0.9 -0.1 -- 0.3 - 0.3 0.0 * This figure is choeen rather than the lower value given by Coward and Brinsley (Zoc. cit. p. 1885) for the reason stated on page 1877 of that paper : “ . . . the flames of mixtures containing 5-3 to 5.6 per cent. of methane are very sensitive t o extinction by shock . . . a 5-6 per cent. mixture will invariably propagate flame when the shocks are no greater than those occa-sioned by the somewhat violent bubbling of gas through water . . . (but) when the circumstances are such that a tranquil passage is rtssured 5-3 per cent. is the lower limit of inflammability of methane in air.” In none of the present experiments with methme mixtures did we observe the curious tranquil passage of flame noted with the 5.3 per cent.methane mixture so that 6.6 per cent. seems the correct figure to employ for calculations in con-nexion with these experiements. C0,=2.6. 02=0-5. C,H etc.=2.8. CO= 14.1. H,=46.6. CH,=19.4. C,€&=4-0. N,=9.2 per cent .The benzene etc. vapours were estimated by the deter-mination of their partial pressures by Burrell and Robertson’s method (J. Ind. Eng. Chem. 1915 7 669). For calculating this figure the following lower limits of the individual gases were used Hydrogen carbon monoxide and methane &s in table I, ethylene 3.4 per cent. (Eitner) ethane 3-1 per cent. (Burgess and Wheeler), benzene 1.4 per cent. (Kubierschky). The last three values represent results obtained in small vessels but may probably be safely used in view of the comparatively small amounts of the three gases present in the “ town’s gas.” The non-inflammable constituents of the town’s gw amounted to 12.3 t Composition of the “ town’sgas ” CdHI etc.vapours=O*8 OF INFLAMMABILITY OF GASEOUS MIXTURES. PART 111. 31 A brief reference to the general character of the flames is necessary. Inflammable mixtures rich in hydrogen including the town's gas gave thin vortex rings of flame increasing in diameter as they rose through the first 30 or 40 cm. and then breaking into luminous segments which subdivided into balls of flame. The latter rose increasing in number to the top of the vessel. The flames in mixtures somewhat below the lower limit were extinguished a t s m 0 stage of their journey.The flames of these mixtures showed in fact similar behaviour to t.hose described by Coward and Brinsley for pure hydrogen but the luminosity was much enhanced. Inflammable mixtures containing no hydrogen gave thick rings which as they progressed developed into flames of strongly convex front spreading from side to side of the box; similar mixtures just below the limit of inflammability gave rings of flame breaking into striae which were extinguished in the next 50 or 100 cm. of their journey. The appearance of all the flames of mixtures is apparently compounded additively of those of the individual components. There was no difficulty in deciding upon the figure for the limit within about 0.1 per cent.Conchions (Part 111.). The lower limits of inflammability in air of mixtures of hydrogen carbon monoxide and methane taken two a t a time or all together and also the lower limits of wat,er gas coal gas and town's gas may be calculated with approximate accuracy from t.he lower limits of the individual gases by means of L e Chatelier's formula. PAXT 1v. The Upper Limits of some Gases Siiigly and Mixed in Air. The upper limits of inflammability of hydrogen methane and carbcon monoxide severally in air have been investigated by a number of observers; their results are quoted in T. 1914 105, 1859. The hydrogen figures show t.he greatest range of variation, namely from about 55 t o 80 per cent. of hydroge'n. The methane figurw mostly lie bet.ween 12 and 17 per cent'.of methane and per cent. which on the lower limit mixture represents only 0.66 per cent. of the whole. In view of the known slight influence on the lower limit of methane of the substitution of small amounts of carbon dioxide or nitrogen for equal amounts of air it is safe to assume for the purposes of the calculation that the non-inflammable constituents of coal gas (and likewise of water gas) can bo treated as air 32 COWARD CARPENTER AND PAYMAN THE bfLUTfON LIMITS the carbon monoxide figures are in the neighbourhd of 75 per cent. of carbon monoxide. The Upper Limit of Hydrogen .-Some preliminary experiments were conducted with the object of discovering whether the flames in mixtures just below the upper limit resembled the flames in mixtures just above the lower limit.If so the apparent dis-crepancies between the resulk3 of earlier workers might be ex-plained on the same lines as the discrepaiicies noted in lower limit figures. It was soon evident however that comparatively weak or short electric sparks which were quite strong enough to ignite lower limit mixtures were unable to inflame upper limit mixtures. Stronger sparks in the lataker mixtures started flames which travelled throughout the whole mixture. This promised a clue to the main cause of discrepancy of the results of others and by the use of igniting sparks of such variable strength as might well have been employed in ordinary laboratory practice a range of results was obtained nearly as wide as those of the previous un-correlated list.The experiments were carried out. in a half-litre globe with a spark of variable length in the centre. A 6-inch Apps induction coil was used with a constant break and the current in the primary was varied by using a battery of 2 to 12 volts. We have to acknowledge our indebtedness to Mr. F. Brinsley for conducting this series of experiments the results of which are recorded in table 11. TABLE 11. Percemtage of Hydrogen in Apparent Upper Limit Air Mixture. Spark gap. 2 4 1 mm. 57.5 -2 4 8 70.2 70.7 16 20 32 46 56 - -- -- -I -- - - -- -8 67.0 70.2 7 1-2 7 1-2 72.5 73.5 73.5 --Voltage of accumulators. __7 --/--_ 12 ---72.2 74.5 75-5 -I -These figures suggest an upper limit of ydrogen-a,r mixtures in the neighbourhd of 75.5 per cent.of hydrogen but the volume of gas used was much tnoo small to indicate whether the flames observed were capable of indefinite self-propagation. Further-more the gases were confined and so were not maintained under constant pressure during inflammation FIG. 1. Hydrogen. FIG. 3. FIG. 2. S4etlzune. FIG. 4 Methane and carboil. inonoxide mixture. Upper linzit Jlames in tube of 5 em. diameter. [Yo J h c e page 33 OF INFLAMMABILITY OF QASEOUS MIXTURES. PART IV. 33 A series of experiments was therefore carried out in a 15-litre bell-jar just dipping under the surface of water with a spark gap near t'he botltom of thel mixture. With a suitable single spark, ignition was obtained with mixtures containing 73.1 73.8 and 74.0 per cent.of hydrogen but failed with 74.4 and 75.0 per cent. of hydrogen. The flames travelled rapidly throughout the whole mixture. The limit indicated was thus approximately 74.2 per cent. of hydrogen. The next step in determining the true limit fosr continued pro-pagation of flame was the use of long vessels. A tube 1.5 metres long and 5 cm. wide was used. Flame travelled rapidly through this tube with mixtures containing 71.2 and 71.4 per cent. of hydrogen. The appearance of this flame is indicated in Fig. 1. Mixtures containing 71-6 and 73.0 per cent. could not be ignited, or if ignited the flame was extinguished before it had travelled more than a felw cm. from the spark. I n order to fix the upper limit precisely it would be necessary to use vessels of dimensions comparable with those of the box previously described.This would involve the construction of a much stronger vessel than the one available but a t the time this was contemplated the experimenh had to be abandoned and oppo'rtunities for continuing them will not be available in the near f uture. It is however certain that the upper limit of hydrogen is some-what higher than 71.5 per cent.; it is probably near to 74.2 per cent. The Upper Limit of Methane.-In the 15-litre bell-jar mixtures containing 15.1 and 15.3 per cent. of methane propagated flame, of a reddish-brown colour edged wihh blue upwards throughout the mixture. A 15.5 per cent. mixture could not be ignited but when a rapid succession of sparks was passed a blue flame-cap was observed above them.I n the 1-5-metre tube mixtures containing 14.4 14.7 15.0 and 15.1 per cent. of methane propagated flame throughout the tube; with a 15.2 per cent. mixturel a flame was initiated but was extinguished after pas3i.n-g some pn. up the tuba. I n each case, the flame seemed to consist of two distinct portions the upper-most blue with a convex front followed by a reddish-brown conical tail which suggested a secondary reaction of combustion (see Fig. .2). The limit for indefinite propagation is therefore more than 15.1 per cent. and probably approaches 15.4. This conclusion is supported by the experiments of Burrell and Oberfell (Zoc. cit.) who used for upper limit experiments on methane an iron pipe 30 cm. (12 inches) in diameter 2.1 metres VOL.cxv. 34 COWARD CARPENTER AND PAYMAN THE DILUTION LIMITS (7 feet) high with a series of glass windows. Their experimelli:; showed the upper limit to lie between 15.0 and 15.4 per cent. for upward propagation of flame. The Upper Limit of Carbom Monoxide.-In the 15-litre bell-jar flames travelling rapidly upwards through the whole of the mixture were obtained when 73.7 and 74-0 per cent. of carbon monoxide was present. Flameis were initiated in 74.5 and 75.0 per cent. mixtures but were extinguished after travelling a short distance. A 75.2 per cent. mixture gave only a blue halo round the spark. I n the 1.5-metre tube a flame travelled up through a 72.9 per cent. mixture but no more than a tongue of flame1 was obtained with a 73.1 per cent.mixture. The walls of the tube evidently exerted a notable cooling influence. The1 self -propagating flame had a strong convex front was blue with a bright whitish-blue edging but had no “tail,” as was the case with methane flames (see Fig. 3). The limit for indefinite propagation is therefore more than 73.0 per cent. and probably approache? 74.3 per centl. Apjdiicability of the Mixture Law t o Upper Limits of Inflammability . The additive character of the lower limih of inflammability was expressed by Le Chatelier in a formula quoted above (p. 28). The validity of a similar formula for t h e upper limits of mixeld combustible gases-in air has been obscured by experiments with hydrogen-air mixtures in which the sparks were insufficiently strong and therefore the figures obtained represented not the limiting composition for the propagatioln of flame but the limiting cmposibion for the initiation of flame by the sparks in use.It is shown below that t;he following formula holds approximately for the upper limits of mixtures of hydrogen methanel and carbon monoxide two or three a t a time;: where nl % . . . represent the proportions in percentages of the whole air mixture of each combustible’ gas a t the upper limit, N, N . . . represent thel upper limits in percentages of the whole air mixtnre of each combustible gas separately. F o r reasons stated abolve the experimental apparatus available did not combine the desirable width with the desirable length, and the choice lay between a bell-jar and a longer but narrowe OF INFLAMMABILITY OB' GA!E%OUS MIXTURES.PART IV. 35 glass tube (1.5 metres long 5 em. iii diameter). The latter was chosen because it enabled a flame to be ubserved travelling far enough from the original source of ignition; the disadvantage of narrowness was not great for the limits observed in the tube were lower than those indicated by the wider bell-jar by a not very considerable amount. (For hydrogen 2.7 on 74.2 per cent.; for methane 0.25 on 15.4 per cent..; for carbon monoxide 1.2 on 74-2 per cent.) The limits observed in the tube are recorded in table 111. TABLE 111. Percentage composition of gas (before admixture Upper limit of inflammability in air. with air) ,- A \ I \ Calcu- Differ. - 7 1.5 CH,. CO. Observed. lated.ence. - I H,. -- I 100 Single gases ... - 100 - 15.1 - - - 100 73.0 I 48.5 51.5 - 22.6 24.4 -1-S Binary 50 - 50 71.8 72.5 -0-7 mixtures. - 50 50 22.8 25.0 -2.2 Ternary mixture ...... 33.3 33.3 33.3 29.9_ 31.9J -2.0 Coal gas ...... ... See footnote.* 30.9 28.8t $2.1 * Composition of the coal gas. C,H, etc. = 1.2 ; CO2=O.1 ; 0 ~0.1 ; C2H,=2.9 ; C0=7-3 ; H,=50-6 ; CH,=29*7 ; C,H,=3.2 ; N2=4*9 per cent. -t For the calculations the upper limits of hydrogen methane and carbon monoxide given in the table were used together with the values C,H,=4.7 (Roszkowski) ; C2H,=22.0 ; C,H,= 10.7 (Burgess and Wheeler ignition centrally in large globe. Private communication). Analyses of the residual gases showed that mixtures just below the upper limit propagated flames which consumed the whole of the oxygen and therefore passed through the ~ h d e of the mix-ture.This behaviour is in sharp cont'rast with t.hat of lower limit mixtures in which self -propagating Aames may leave unconsumed a considerable fraction of tlhe mixture. Figs. 1 2 and 3 show that bhe upper limit methane flames are characterised in contradistinction from the hydrogen and carbon monoxide flames by the possession of flame tails. Mixtures con-taining methane wit,h carbon monoxide or hydrogen and air also exhibit the m a r k a b l e tail which suggests a secondary reaction (see Fig. 4). Evidelncs as to the nature of this reaction should bs readily obtained by an examination of the intercoiial gases, c 36 PAYMAN AND WHEELER THE PROPAGATION OF but the opportunity of attempting this experiment has not pre-sented itself, C’onclzcsions (Partl IV.) . The upper limits of inflammability in air saturated with water a t 18-19O of hydrogen met>hane and carbon monoxide are in the neighbourhood of 74.2 15.4 and 74.2 per cent. respectively. The upper limits in air of mixtures of these gases taken two o’r three a t a time and also the upper limit of coal gas may be calculated with approximate accuracy by means of a simple formula of an additive character. FACULTY OF TECHNOLOGY, MANCHESTER UNIVERSITY. [Received October 12th 1918.