94 MORGAN THE IGNITION OF EXPLOSIVE IX.-Fhe Ignition of Explosive Gases by Electric Sparks." By JOHN DAVID MORGAN. SOME time ago an investigation was carried out by Dr. R. V. Wheeler and Prof. W. M. Thornton on the ignition of explosive gases by sparks produced in signal bell circuits (Home Office Report on Electric Signalling with Bare Wires R. V. Wheeler and W. M. Thornton June 1916). They used iron-core coils in con-junction with mechanical means for breaking the circuit Corn-menting on the results obtained they state : "It may be said that ignition by a rapid break flash a t a low circuit voltage depends on the inductance voltage a t which the flash is formed and the igniting power of the flash is proportional to the product Li (where L is the inductance of the circuit and i the current prior to breaking the circuit).When the break of the circuit is made slowly the igniting power of the flash has been found to depend upon its energy +Li2. There are thus two limit-ing conditions for the igniting power of the flash; a t the one the inductance voltage is of importance a t the other the energy. For any given gaseous mixture there is a range of rapidity of break over which the two types of ignition blend so that under certain conditions the igniting power of the flash may be proportional * Published with the permission of the Advisory Council for Scientific and Industrial Research GASES BY ELECTRIC SPARKS. 95 neither directly to i nor to 9 but to some intermediate value of it.” Then referring to a previous report (Home Office Report on Batt.ery Bell Signalling Systems R.V. Wheeler January 1915), khey cite a case i n which it was found that the igniting power of the break flash could be expressed by the relationship J,il-4= constant. The figures by Wheeler and Thornton in support of the conclusion that Li is constant are given in table I. TABLE I. Inductance (L). Henries. 0-27 0-47 0.70 0.90 1.04 1.18 1-27 1-31 1-60 2.00 Igniting current (i) at 25 volts. Ampere. 0-82 0.45 0.26 0.20 0.17 0-156 0.145 0.13 0.11 0.09 Li. 0.220 0.212 0.182 0- 180 0.177 0.183 0.184 0.170 0.176 0.180 I n the same report they give the number of layers of wire on the magnets used by them together with the igniting currents, These are given in table 11.I have added a third column giving the product N V where N=number of layers. As the layers each have the same number od convolutions N is proportional to the turns. It will be noticed that it can also be said that N2i2 is constant a quantity which has not the same pnysical significance as Li. The expressions are only comparable when each contains either of the terms i or 9. TABLE 11. Layers of wire on magnet ( N ) . 4 8 12 16 18 20 22 24 28 32 Igniting current (i). 0.82 0.45 0.26 0.20 0.17 0.155 0.145 0.13 0.11 0.09 N V . 10.8 12.8 9.7 10.0 9.3 9.6 10.0 9-7 9.6 8.4 I n the earlier report by Dr. Wheeler a table is given from which the number of layers on the magnet coil can be deduced an 96 MORGAN THE IGNITION OF EXPLOSNE the igniting current is added.lated N%2 and the figures are given in table 111. Using these figures I ha.ve calcu-TABLE 111. Number of layers ( N ) . Igniting current (i). Pi2. 20 0.17 11.6 16 0.23 13.6 12 0.33 16.6 8 0.66 19.2 6 0.96 32-6 The figures in the third column show that in this case the pro-duct N2i2 is by no means constant but progressively increases. When the flux produced by the current is linked with the whole of the windings the product N2i2 is proportional to the electro-kinetic energy of the system so long as the permeability is constant. When the linkage is imperfect or the permeability varies the energy is not proportional to N2i2. From the resulh above referred to it is found that when a low tension igniting spark is defined in terms of the coil constants ( N or L ) and the current (i) prior to the spark the energy required to produce a spark that will ignite a gas mixture of given composition appears to be constant in some cases and not constant in others.Experiments with low tension sparks have led me to suspect that such results as those above mentioned though apparently diverse have some constant factor in common and that the dis-parities are due to the manner in which the resiilts are expressed. There is not implied by this remark any suspicion of the accuracy of the work done by Wheeler and Thornton. They were concerned mainly with determining what circuit conditions could give rise €0 dangerous sparks and from that point of view the results were expressed in terms of practical utility.The question raised is as t o whether the results as expressed can be employed to determine the property of the spark on which ignition depends. I therefore decided to make a new investigation with low tension sparksj and arrange the experiments to cover a wide range of different magnetic conditions. Six short air-coae coils were made according to the particulars given in table IV. TABLE IV. 1 100 2 2 200 4 3 300 6 4 400 8 5 600 10 6 700 14 No. Number of turns. Number of layers GASES BY ELECTRIC SPARKS. 97 Two iron cores of relatively large cross-section were also made, one a laminated bar and the other a closed laminated frame for use with the same coils. The experiments were divided into! three groups which were distinguished only by the differences in the magnetic conditions of the cores.Diagrammatic representations of the coils are shown in Fig. 1. Current was obtained from a 12-volt accumulator. The circuit was completed by a variable FIG. 1. Air-core coil 0,pen iron,-core coil. Closed ivon-core coil. resistance of negligible inductance an ammeter and a ‘flick’ con-tact breaker the latter being enclosed in the explosion chamber. The contact breaker consisted of a flexible steel prong capable of being rotated into contact with a fixed steel stem and then flicked over the stem. A coal gas and air mixture of constant composi. tion was used throughout the investigation. The least currents required to produce ignition are given in table V.VOL. cxv. 98 MORGAN THE IGNITION OF EXPLOSIVE TABLE V. Air-core Coils. No. of layers ( N ) . Current (i) amperes. 2 4.6 4 2.05 6 1.2 8 0.83 10 0.575 14 0.35 Open Iron-core Coils. No. of layers ( N ) . Current (i.) amperes. 2 1-05 4 0.51 6 0-35 8 0.26 10 0.2 1 14 0.15 Closed Zrm-core Coils. No. of layers ( N ) . Current (i) ampere. 2 0.63 4 0.32 6 0.2 8 0.16 10 0.13 14 0.09 N2i2. 81 67.4 54 44 33 24 N V . 4.4 4.3 4.4 4.3 4-4 4.4 N V . 1.6 1.64 1.44 1.64 1.69 1.6 Figs. 2 and 2a give the results graphically. It will be noticed that the product N2i2 is not constant for the air-core coils although it tends t o a'constant value a t the upper value of N and is constant with the open iron and clwed iron-core coils although the value of i V 2 9 is different in the latter two cases.I n other words the results may be said t o be similar in kind to those obtained in Wheeler and Thornton's investigations. The present investigation differs from those as recorded in the cited reports of Wheeler and Thorntm in that I have carried out measurements on the circuits after interruption. The first step consisted in the use of an arrangement as shown in Fig. 3. This is a Wheatstone bridge in conjunction with a ballistic galyancl 'meter. The inductance coil a non-inductive balance resistance b , ammeter c flick contact breaker d variable non-inductive resist-ance e and battery f are all (exceptling 6 ) as used i n the explwion experiments.u and b are shunted by non-inductive resistances T of sufficiently low resistance to eliminate sparking a t d when the circuit is broken. Using with each coil the current values require GASES BY ELECTRIC SPARKS. 99 to give the igniting sparks the observed ‘(kicks” were plotted against N . The straight lines indicate that for each group the energy associated with the system prior to interruption was constant but they give no information as t o These art3 given in Fig. 4. a b 6 1.2 5 1.0 4 0.8 3 0.6 2 0-4 1 0.2 8 10 12 14 N 0 2 4 6 a = air-core coil. b = open iron-core coil. c= closed iron-core coil. 2 4 6 8 10 12 1 4 N a = air-core coil. b=open iron-me odl. c = Closed iron-core mil. E 100 MORGAN THE IGNITION OF EXPLOSIVE whether the energy was the same for each group.The induct-ances were therefore measured and found to give +Liz=constant for each group but different for different groups. As nothing is gained by quoting all the values of all the induct.ances only the largest for each group is recorded in table VI. TABLE VI. Air-core coil 14 layer ..................... 0.01 0-0006 joule. Open iron-core coil 14 layer ............ 0.07 0.0008 .. Closed iron-core coil 14 layer ............ 0.56 0.0023 .. L. ;Liz. It is clearly not permissible to conclude that the energy pro-jected into the sparks in the explosion experiments is constant for a constant magnetic condition but different when that condition FIG. 3. is changed until it is proved that; the differences found are not accounted for by cosre or other losses.A further step involving direct spark measurements was therefore necessary. After try-ing various schemes the apparatus shown diagrammatically in Fig. 5 was adopted. a is the flick contact breaker used in the explosion experiments. This is enclosed in an ebonite chamber, b to which a capillary tube is sealed. Two things appeared a t first to render this device useless. The heat developed by the current passing through the contact breaker when closed was sufficient tol interfere with proper measurement of the heating effect of the spark produced on opening the contact breaker. Further it was difficult t o maint,ain a perfectly gas-tight joint around the rotatable stem carrying the prong of the contact breaker. These troubles were avoided by permitting a slight leak in the chamber and observing (through a microscope) only the “kick” given to the liquid thread in the capillary tube.Usin GASES BY ELECTBIC SPARKS. 101 the coils and current values employed in the explosion experi-menb it was found that the “kicks” were the same throughout. It follows that the sparks obtained in the three widely varying groups of experiments were identical as regards their impulsive thermal effects. I do not think it can be argued from the above that the FIG. 4. spark i ~ a = air-core coil. b =open iron-core coil. c = closed iron-core cc~il. energy was the same in all cases although this possibility is not excluded. The final step consisted in an attempt to determine definitely whether %he energies of the different sparks were the same or not.For this purpose a high tension winding of fine wire and many turns was placed on one of the limbs of the closed iron core a 102 MORGlAN THE IQNI?l!ION OF EXPLOSIVE shown in dotted lines Fig. 1 and connected to a small permanent gap in a spark plug a Fig. 6 . The sparks produced were very small and several had to be produced in succession to give a deflection definitely readable through the microscope. I n place of the flick contact breaker in the primary circuit a cam-operated interrupter such as is used in ignition apparatus for internal-combustion engines was employed. With this interrupter twelve FIG. 5. sparks were obtained from each complete rotmation of the cam. The chamber was made perfectly gas-tight; and the deflections were different in character from those of the previous experiment in that they were relatively slow.Taking each of the coils in turn and using the current values obtained in the explosion experi-ments the deflections obtained after one complete rotation of the cam were observed. I n all cases they were found to be the same. FIG. 6. It follows that the total heating effect of the same number of sparks from each coil was the same; consequently the sparks were of equal energy. Seeing that the sparks in all three groups of experiments gave the same impulsive thermal effects and the sparks in one group gave the same energy effects it is permissible to argue that the energies of all the sparks in the three groups were the same GASES BY ELECTRIC SPARKS.103 The conclusion of the investigation is therefore that over the wide range of different Conditions examined tbhe igniting sparks had this in common-that they all liberated the same amount of heat energy a result which is not evident from measurements on the spark circuit prior to the production of the sparks. Regarding single spark ignition of explosive gases initially a t atmospheric temperature and pressure the main resulta of in-vestigations which have been published in recent years and which can be regarded as well established appear to be as follows: (1) With a low tension spark the least spark energy required t o ignite a given gas mixture diminishes with increase of the voltage impressed on the spark circuit prior to the production of trhe spark (The least energy required to &.artt a gaseous explosion, W.M. Thornton Phil. Mag. 1914 [vi] 28 734). (2) When the circuit voltage is constant the spark energy required for ignition of a given gas mixture by a low tension spark is constant (see above). (3) With a high tension spark (which consists of a capacity component preceding an inductance component,) the incendivity of the spark (or ability to cause ignitipn) can be increased by increasing the proportion of energy in the initial part of the spark without increasing the total energy of the spark (" Spark Ignition," J. D. Morgan Engineering November 3rd 1916). (4) The incendivity of a condenser or capacity spark is greater than that of an inductance spark dissipating the same amount of energy (Thornton Zoc.cit.). (5) With a capacity spark the least energy required for ignition of a given gas mixture diminishes as the spark voltage increases (Thornton Zoc. c i t . ) . These results clearly establish Dhe fact that the incendivity of a spark does not depend on the total energy of the spark. It is generally supposed that the energy required to produce ignition of a given inflammable gas mixture is constant for similar physical conditions. If the assumption is correct then the fact that the total energy of the least igniting spark is found experi-mentally to vary with the conditions under which t3he spark is produced suggests that not all of the spark energy is utilised in the process of ignition but only a portion at the commencement of the spark.It is of course possible that the inflammability of a gas as determined by the least energy required to produce igni-tion is not' constant' for identical physical conditions of the gas but i t would appear to be useless to attempt an investigation of this point by spark measurements having regard t o the facts above mentioned. It is important to note that a spark is a varyin 104 JEPIXCOTT THE PHYSICAL CONSTANTS OF NICOTINE. PART I. source of heat which very rapidly reaches its maximum intensity and then less rapidly disappears. Experiments prove that increase of the initial intensity of a spark results in increased incendivity. As already stated t,his suggests that ignition is due only to the initial part of the spark and that in every spark there is a certain amount of unused energy which makes no contribution to the process of ignition. The proportion of unused energy must diminish as t,he initial intensity increases but a t present any measurements of the effective portion of the spark appear to be impmsible. It follows from this suggested theory of unused energy that any attempts to specify t,he inflammability of a gas in terms of the total energy of the least igniting spark must necessarily yield the diverse results which have hitherto been obtained. THE MARKS AND CLERK LABORATORY, 13 TEMPLE STREET BIRMINGHAM. [Received November 21st 191 8.