1436 PAYMAN THE PROPAGATION OF FLAME IN cx XXV.-The Propagation of Flame in Complex Gaseous Mixtures. Part 1. Limit &fixtures and the Uniform Movement of Flame in such Mixtures. By WILLIAM PAYMAN. IN order that flame may propagate through a mixture of an in-flammable gas with air or oxygen the heat developed by a given “layer” on burning must be sufficient to raise the contiguous layer of unburnt gas t o its ignition temperature. In a “limit mix-ture ” there is just sufficient and only just sufficient heat developed to accomplish this. If limit mixtures -of two or more inflammable gases be mixed together this heat balance should remain un-altered provided that all of the limit mixtures are of the same kind that is t o say all lower-limit or all upper-limit mixtures.It follows that all mixtures in any proportions of limit mixtures should remain limit mixtures the limiting percentage being that of the mixed inflammable gas. Conversely any limit mixture of a complex inflammable gas will consist of a number of limit mixtures of the individual gases it contains. Assuming this reasoning to be correct imagine a limit mixture with air of a complex inflammable gas. . . . be the simple constituents of this inflammable gas and their limits of inflammability N, ,VB N, . . . respectively. Suppose also that the complex limit mixture contains a per cent. of A b per cent. of B c per cent. of C . . . Then this limit mixture will contain Let A B C, aA + bB + cC + . . . + [lo0 - (a + b +c + . . .) J air. ‘This limit mixture ex hypothesi comprises a series of limit mix-tures of the simple inflammable gases.I n such a simple limit mixture of A for example every N parts of A are associated with 100-Nd parts of air so that every a parts of A are associated with (T) loo - N~ a parts of air. Similarly for B C . . . The complex limit mixture will therefore contain Since these terms equal 100 so that b+(E!-$)b G + c . . . (‘oO.zi“rJ,N”) are expressed as percentageg they will together c + . . * =loo. 100 - N,‘ 100 -&‘ b + c + a + ( N )a+b+(T-) (7 COMPLEX GASEOUS MIXTURES. PART I. 1437 This expression on simplification becomes This is the formula of Le Chatelier which has been shown to apply accurately for both the upper and lower limits of inflamma-bility of a number of complex gaseous mixtures with air (Coward, Carpenter and Payman this vol.p. 27). This “formula,” but not the generalisation from which it has been deduced applies only to mixtures of inflammable gases with an atmosphere of constant composition such as air. For the numerical quantities involved in the formula relate to the com-bustible gases only and admit of no allowance being made for variations in the proportions of inert gas present. The general-isation however should hold good for all limit mixtures; for mixtures in which t.he proportion of inert gas is greater or less than in air or even in which its proportion is not constant.* The effect of an inert gas (nitrogen) on the limits of inflamma-bility of methane has been investigated by Burgess and Wheeler (T. 1914 105 2596) who determined the limits for this gas in several artificial atmospheres of oxygen and nitrogen containing less oxygen than air.During the course of the present inquiry into the mode of combustion of mixed gases it became necessary to extend their work to include atmospheres containing more oxygen than air and pure oxygen. The method of experiment used by Burgess and Wheeler which involved the central ignition of the mixtures in a large globe was not employed in the present research. This investigation is mainly concerned with the uniform movement of flame and the deter-minations of the limits were made in the same apparatus as was used for measuring the speed of propagation of flame. This con-sisted of a horizontal glass tube 2.5 cm. in diameter open a t one end and closed a t the other the mixtures being ignited close to the open end of the tube by means of an electric spark.The criterion of inflammability was theref ore the horizontal propaga-tion of flame throughout the length of the tube. ‘The determinations were carried out by the method of trial and error using mixtures which differed in composition by about 0.10 per cent. of methane. Throughout this paper the term ‘‘ limit mixture,” whether upper or lower implies that mixture in which flame was just able to propagate. The limits were always sharply defined. On sparking a mixture con-The results of the determinations are given in table I. * This will be described in future as the *‘ limits generalisation.” 3 ~ 1438 PAYMAN THE PROPAGATION OF FLAME IN taining a little less than the lower-limit percentage of inflammable gas there usually arose a ball of flame which travelled some 5 or 6 cm.along the tube. A mixture containing slightly more in-flammable gas than the higher-limit percentage usually produced a flame which travelled the short distance from the spark to the open end of the tube owing to the dilution of the mixture there by diffusion of the outside air. In the limit mixtures flame travelled steadily and a t an approximately uniform speed through-out the length of the tube. I n no instance did the flame of the burning limit mixture fill the whole cross-section of the tube but it was usually similar to the trailing flames described by Burgess and Wheeler (T. 1914, 105 2593). This was most marked with the upper-limit mixtures and with the lower-limit mixture of methane in pure oxygen.These flames were about 10 mm. in diameter and about 15 mm. long and had a short tail resembling a “Prince Rupert’s drop” in shape. In three instances the deposition of carbon was noticed during the passage of flame through a limit mixture namely in the upper-limit mixtures of methane with atmospheres containing 50 66, and 100 per cent. of oxygen.* The flames resembled that of a tallow candle and the odour of the residual gases was similar to that caused by the smouldering wick of such a candle. I n general, the upper-limit flames were olive-green in colour. The colour of the lower-limit flames was pale blue. TABLE I. The Limits of Inflammability of Methane in Mixtures of Oxygen and Nitrogen.Percentage of oxygen in atmosphere. 13.7 17.0 21.0 (air) 33-0 50.0 66.0 100.0 c CH,. 6-4 6-1 5.8 5.8 5.8 5.8 5.7 Percentage composition of limit mixtures. , A Lower limit. Upper. P - 0 2 - N2. CH,. 0 2 . 12.8 80.8 6-9 12.7 16.0 77.9 8.9 15.5 19-8 74.4 13-3 18.2 31-4 62.8 25.1 25.0 47.1 47.1 38.8 30.6 62.8 31.4 47.5 35.0 94.2 - 59-2 40.8 7 N 2’ 80.4 75-6 68-5 49.9 30.6 17-5 -* According to Bone (Phil. Trans. 1915 215 275) when methane and oxygen mixtures are exploded under pressure “ there is EL total cessation of any separation of carbon (which is very marked with mixtures 2CH + 0,) after the proportion of oxygen in the original mixture exceeds the limit 3CH + 20,.” No carbon was deposited when a mixture containing 59.3 pe COMPLEX GASEOUS MIXTURES.PART I. 1439 0 0 30 The results are plotted in the diagram the ordinates representn-ing percentages of methane and the abscisse percentages of oxygen in the limit mixtures. If the “limits of generalisation ’’ given earlier in this paper applies to these mixtures the values for each of the two sets of limik should lie on a straight line. It will be seen that this holds accurately over a large range of mixtures namely over those con-taining more than about 17 per cent. of oxygen.” Mixtures con-taining less than this amount of oxygen require rather more methane than the theoretical quantities to attain both the upper and lower limits. 0 40 I Oxygen in limit mixture per cent.The dotted lines in the diagram represent the values obtained by Burgess and Wheeler (Zoc. cit.). The shapes of both curves are cent. of methane and 40.7 per cent. of oxygen was exploded under a pressure of 12.7 atmospheres. In the present series of experiments a t atmospheric pressure a mixture of approximately the same composition as that used by Bone (59.2 per cent. of methane and 40.8 per cent. of oxygen) deposited carbon as did also mixtures containing less oxygen in proportion to the methane present, namely those intermediate in composition between 3CH4+20 and CH,+ 0,. It would therefore appear that the limiting composition at which the deposi-tion of carbon ceases is not fixed but varies with the initial pressure of the mixture.* According to Burgess and Wheeler (Zoc. cit.) no mixture of methane, oxygen and nitrogen is capable of propagating flame when there is less than about 13 per cent. of oxygen present 1440 PAYMAN THE PROPAQATION OF FLAME IN similar the difference in magnitude being due to the difference in experimental conditions. Little change was observed in the lowe; limit until the mixtures contained a large excess of nitrogen; whilst the value with pure oxygen was only slightly lower than that with air. The latter observation is not in agreement with the results recorded by Parker (T. 1914 105 lOOZ) who found the lower limit of inflammability of methane to be slightly higher with oxygen than with air (6.0 per cent. and 5-8 per cent. respectively). The apparatus used by Parker was similar to that previously used by Burgess and Wheeler namely a 2-litre globe in which the mixtures were ignited a t the centre.This lack of agreement is undoubtedly due to the difference in the position of the point of ignition in the two sets of experiments. It has frequently been noted that the limits of inflammability vary with the position of the point of ignition according as the flame has to pass upwards or downwards through the gas mixture. The fact that a flame will pass more readily upwards than downwards is well illustrated when a lower-limit mixture of methane in air, for example is ignited by a spark a t the centre of a globe. As soon as the spark passes a flame shoots to the top of the vessel, bends over and then moves slowly downwards to the bottom.I n order to investigate this point further a series of experiments was carried out to determine quantitatively the effect of varying the point of ignition on the limits of inflammability of methane in air and in oxygen. A glass tube 2.5 cm. in diameter was used, closed a t one end and fitted with firing points a t the other (open) end. TABLE 11. Limits of ZnfEarnmability of Methane with Bifferent Positions of the Point of Ignition. Percentage of methane in lower-limit mixture. - Mode of propagation. Air. Oxygen. Upward ........................ 5-5 5.4 Horizontal ..................... 5.9 5.8 Downward .................. 6.1 6.3 Central ignition (Parker). . 5-8 6-0 Both for upward and horizontal propagation the lower limit of inflammability of methane is less in oxygen than in air.Bor downward propagation however the order is reversed. The differences observed are not very great although too large to be accounted for by experimental error. Of the factors which deter-mine the value of the limiting percentage of inflammable gas th COMPLEX GASEOUS MIXTURES. PART I. 1441 transference of heat by convection and the absorption of heat by the mixture may be mutually opposed. During the downward propagation of flame convection does not materially affect the transference of heat to unburnt layers of the mixture; the influence of the slightly higher specific heat of oxygen as compared with that of nitrogen therefore becomes apparent. With horizontal and upward propagation of flame however the influence of con-vection currents masks the effect of the higher specific heat of oxygen.The change of order of the results dependent on the direction of travel of the flame is more marked when the results for methane in air are compared with those for hydrogen. Such a comparison is made in table 111. TABLE 111. Lower Limits of Inflammability in A.ir of Methane and of Hydrogen. Percentage of inflam-mable gas - Mode of propagation. Methane. Hydrogen. Upward ........................ 5-5 4.2 Horizontal ..................... 5.9 6-2 Downward .................. 6.1 9.7 Central ignition ............ 5.8 9.2 Attempts have been made to calculate the limits of inflamma-bility of a gas from its thermal constants. It will be clear from a consideration of the results recorded in table I11 that any such calculation is doomed to failure unless allowance can be made for the influence of convection currents.Since the lower limit of inflammability of methane (downward propagation of flame) is less with air than with oxygen it might be expected to be less still with an atmosphere containing less oxygen than air. The lower limit of inflammability of methane in an atmosphere containing 17 per cent. of oxygen was found to be 6 . 3 per cent. for downward pro-pagation of flame. This limit is thus affected in the same sense as both ICmits for horizontal propagation in mixtures containing only a small percentage of oxygen; that is t o say more methane is required to form the limit mixture than would be expected from results with mixtures richer in oxygen.This displacement of the range of inflammability corresponds with the displacement of the range for maximum speed of uniform movement of flame in mixtures of methane and air. It has been generally assumed that the latter displacement is due to the higher thermal conductivity of methane as compared with that of air. A This however is not so 1442 PAYMAN THE PROPAGATION OF FLAME IN similar displacement is found however when the inflammable gas has a thermal conductivity less than that of air as will be shown in a subsequent communication. The displacement under con-sideration in the present paper and other similar displacements, have one feature in common namely that the mixtures contain a large proportion of inert gas (nitrogen) together with only a slight excess of one or other of the reacting gases above the quantity required for complete combustion.A possible explanation of the results is that the mode of com-bustion in such mixtures differs from that in mixtures containing a large excess of either of the reacting gases. Such an explanation is supported by the analyses of the “flame gases” recorded by Burgess and Wheeler in the paper to which reference has already been made. The samples of gas were rapidly snat,ched from the flames in such a manner as to cool the primary products of com-bustion before secondary reactions could come into play. It will be seen on examining the table of analyses (p. 2604) that all mix-tures containing less than 15 per cent. of oxygen appear to be influenced by the deficiency of reacting gas (whether methane or oxygen) and it is in these mixtures that the generalisation regard-ing limiting percentages no longer holds.With the upper-limit mixtures of low oxygen content the primary products of combus-tion contain smaller quantities of hydrogen than of carbon mon-oxide whereas with the higher-limit mixtures containing a greater proportion of oxygen the quantities of hydrogen and carbon mon-oxide produced are equal. Similarly with the lower-limit mixtures of low oxygen content, the primary products of combustion contain more carbon monoxide than hydrogen whilst with lower-limit mixtures containing more than 15 per cent. of oxygen these gases are absent altogether from the products of combustion. Further consideration of these results is reserved for a future ,communication as is also the consideration of the displacement of the range for maximum speed of uniform movement of flame in mixtures of air with inflammable gases.The Unifoym Movement of Flanae in Linzit Mixtures. !l!he speed of horizontal propagation of flame in the limit mix-tures in a tube 2.5 cm. in diameter was determined by the method described by Wheeler (T. 1914 105 2606). The results are given in table IV COMPLEX GASEOUS MIXTURES. PART I. 1443 TABLE IV. Speed of Propagation of Flame in Limit Mixtures of Methane, Oxygen and Nitrogen in a Tube 2.5 cm. in Diameter. Percentage of oxygen in atmos-phere. 13.7 17.0 21.0 (air) 33.0 50.0 66-0 100.0 Speed in cm. per sec. - Lower limit.Upper limit. 21.9 19.1 22.4 19.0 23-3 19.1 23-0 18.9 22.8 18.9 21-3 19.4 19.9 18.9 'The upper-limit speeds are identical within the range of experi-mental error. The speeds in the lower-limit mixtures are through-out slightly higher than the corresponding upper-limit speeds, although with pure oxygen the difference is very small. A notice-able feature of these flames common to them all was their small size in comparison with the diameter of the tube. This was more marked with the flames in the upper-limit than in the lower-limit mixtures a fact which no doubt accounts for the slower speed of the former flames. For the smaller the flame the greater is its surface in proportion t o its volume and the greater in proportion is the transference of heat from the flame to the walls of the tube.I f this explanation be correct it follows that the speeds of flames in limit mixtures should increase with increased diameter of the tube in which they travel. This was found to be so by Mason and Wheeler (T. 1917 111 1052). With tubes of very small diameter on the other hand the speed of flame a t the limits is comparatively high (Payman and Wheeler, T. 1918 113 656) but for another reason. With such tubes the cooling effect of the walls is so great as to have a marked effect on the value of the limits the range of inflammability of the mixtures rapidly narrowing as the diameter of the tube is diminished. Moreover convection currents have no appreciable influence in tubes of such small diameter. It seemed probable that! the speed of flame in a limit mixture, determined under standard conditions should approach a constant value irrespective of the nature of the inflammable gas.To test this the speeds of flame have been determined in limit mixtures of air with several of the paraffin hydrocarbons. The results are given in table V which is of value also in recording the limits of inflammability (horizontal propagation of flame). The limits differ slightly from those found (central ignition in 3 H 1444 PAYMAN THE PROPAGATION OF FLAME IN a large globe) by Burgess and Wheeler whose results are inserted in the table in brackets. I n the upper-limit mixtures the flames vibrated rapidly about half-way along the tube and were sometimes extinguished there. The difference between a limit mixture and one which could only propagate flame for a short distance if a t all was however well marked.TABLE V. Limits of Inflammability and Limiting Speeds of Flame in Mixtures of Air with the Parafin Hydrocarbons in a TuBe 2.5 cm. in Diameter. Lower limit. Upper limit. 7- - Per cent. of Speed cm. Per cent. of Speed cm. Hydrocarbon. combustible. per sec. combustible. per sec. Methane CH ......... 5.8 (5.6) 23.3 13.3 (14.8) 19.1 Ethane,*C,H ......... 3.3 (3-4) 18.1 10.6 (10.7) 19-7 Propane C3H8 ......... 2-4 (2.3) 20.8 7.3 ( 7.3) 20.3 Butane C,H, ......... 1.9 (1-6) 20.1 6.5 ( 5.7) 20.3 (CH + C,H,,*) ...... 2.6 (2-5)t 22.3 7-7 ( 7*7)t 20-7 * Equimolecular mixture of methane and pentane. -f Calculated from values for methane and pentane.Pentane C,H, ......... 1.6 (1.4) 20-2 5-4 ( 4.5) 20.2 The “limit speed” is thus found t o approach a constant value, as foreshadowed by Burgess and Wheeler (T. 1914 105 2596), not only with each of the paraffin hydrocarbons singly but also with the mixture of methane and pentane. There is no reason to doubt but that the limit speed of flame would have the same value for any mixture of the paraffins. The speeds of flame in limit mixtures with air of carbon mon-oxide and hydrogen have also been determined. With carbon monoxide the speed a t both limits (in a tube 2.5 cm. in diameter) is 19-4 cm. per second which agrees well with the speeds for the paraffins. With hydrogen the speed at the lower limit is remark-ably slow namely 10 cm. per second.The flame is exceedingly small consisting of a tiny ball of flame which however travels the full length of the tube. For reasons given in a previous com-munication (this vol. p. 41); it was not found possible to determine accurately the speed of flame in the upper-limit mixture of hydrogen and air, The equimolecular mixture of methane and pentane corresponds with propane in percentage composition and calorific value and yields the same products on complete combustion. The marked difference between the limits of inflammability of the mixed gases and those of propane shows that these are not the only factors on which the limit,s of inflammability depend. Similarly a mixtur COMPLEX GASEOUS MIXTURES. PART I. 1445 of three volumes of pentane and two volumes of hydrogen corre-sponds with propane but this mixture of gases has limits 2-5 (lower) and 8.6 (upper) as compared with propane 2.4 and 7.3.These differences are perhaps due to the ability of the con-stituents of the mixed inflammable gases to burn independently. This subject will be dealt with more fully in succeeding papers of this series. EXPE~RIMENTA L. The speeds of propagation of flame in limit mixtures were deter-mined in glass tubes by the method described by Wheeler (T., 1914 106 2610). Two tubes were employed both 2.5 cm. in diameter; one 3 metres long used for the majority of the experi: ments; the other used for the mixtures with atmospheres rich in oxygen was only 1.5 metres long so as to avoid the setting up of the detonation wave with consequent shattering of the tube.The platinum firing points were about 2 cm. from the open ends of the tubes. At measured distances along each tube were fused ground-glass tubulures which carried glass plugs with stout platinum wires fused through them. Fine " screen wires " of copper wete stretched across these platinum supports inside the tube and electrical connexion was established with an automatic commutator and chronograph by means of platinum terminals on the outside of the plugs. In order to fill the tubes with the mixture required for experi-ment they were exhausted of air by means of an oil-pump half filled with the mixture and re-exhausted before being finally filled. A sample of the gas was then taken for analysis. The limits of inflammability for upward and downward propaga-tion of flame were determined in a similar tube 1.5 metres long, but without side-pieces. The gases were prepared in the usual manner the paraffin hydro-carbons being purified by repeated liquefaction and subsequent fractional distillation until on explosion of a sample with excess of air and oxygen the theoretical value for the ratio C / A was obtained. The methane used in the limit determinations in pure oxygen contained 99.8 per cent. of CH, and the ratio CIA was found to be 2-00. The oxygen was prepared by gently heating recrystal-lised potassium permanganate and contained 99.6 per cent. of 0,. The gases were stored over water rendered alkaline by potassium hydroxide and the mixtures were therefore saturated with water vapour. ESKMEALS, CUMBERLAND. [Received September 4th 1919.1 3 H"