Return to BIOS Table of Contents

B.I.O.S. Final Report No.1612
Item No.19

Fundamental Work On Combustion in Germany

This report is issued with the warning that, if the subject matter should be protected by British Patents of Patent applications, this publication cannot be held to give any protection against action for infringement.

British Intelligence Objectives Sub-Committee

London-H.M. Stationery Office

Price: 17s.6d.net.

Fundamental Work On Combustion In Germany

by D.S. Petty
E.P.Wright
F.H. Garner

Department of Chemical Engineering, The University. 

Birmingham

Technical Information & Documents Unit

40, Cadogan Square, London S.W.1

Note: Mr. H.A. Cheetham and Mr.T.L. Raine assisted in translating and collating documents used in this report. Considerable assistance was given by both Dr. J. Horne and Dr. A.H. Nissan in connection with translation of the documents and arrangements for the presentation of the report.

Fundamental Work On Combustion In Germany During The War Years

Contents of Report

  1. Physics and Chemistry of Oxidation and Flames.

    1. Reaction Kinetics of the Slow Oxidation of Hydrocarbons.

    2. Ignition Delay

      1. Factors influencing ignition delay:-in particular temperature and pressure effect.

      2. The delay in ignition of fuel sprays.

      3. The reaction mechanism in the ignition process.

    3. The Mechanism of Explosive Gas Reactions.

    4. Flames.

  2. Combustion in Spark Ignition Engines.

    1. Cause of Knock

    2. Prevention of Knock. Phenomena Occurring with High Valve Overlap.

    3. Knock Determination

      1. Methods

      2. Accuracy of knock determination

      3. Reference fuels.

      4. Knock measuring instruments.

      5. Viewpoints on possible future methods of fuel testing.

    4. Power Boost.

    5. Fuels and their Properties.

  3. Combustion in Diesel Engines.

    1. General Theory and Practice.

    2. Modifications in Engine Design and Operation

    3. Evaluation of Diesel Fuels. 

    4. Ignition Accelerators.

    5. Cold Starting and the Use of Additives

    6. Fuels and their Properties.

  4. Table of Symbols.

  5. References.

  6. Subject Bibliography.

  7. Appendix.

Physics and Chemistry Of Oxidation And Flames

A. Reaction Kinetics of the Slow Oxidation of Hydrocarbons

Laboratory experiments made in Germany on hydrocarbon oxidation had, as their main purpose, the investigation of ignition  behaviour, and were usually made in some form of adiabatic  compression apparatus. A certain amount of work, however, was done in tubes and static apparatus on the reaction kinetics of slow oxidation of preflame combustion. 

The slow oxidation of normal and isoparaffins at pressure less than atmospheric was studied by Muffling in a static system. With the n-paraffins, hexane and heptane, only a slight pressure effect was observed, although at certain intermediate temperatures the reaction presented a region of discontinuity- explosion occurring spontaneously on raising the pressure slightly. This reaction was compared with that of some peroxides which dissociate spontaneously above a critical pressure, but only slowly below it; thus a dissociation process is shown in Muffling's experiment with the n-paraffins by the appearance of almost equal concentrations of carbon monoxide and hydrogen. 

In contrast to the n-paraffins the iso-paraffin, iso-octane, showed a strong pressure effect. Changing the pressure in the system by inert gas addition had a marked effect on the iso-paraffin oxidation, the mechanism of which was therefore considered to consist of primary reaction and chain branching in the gaseous phase with breaking on the wall. This, together with the fact that oxidation was only observed at temperatures at which thermal dissociation occurs suggested that the oxidation resulted from this dissociation. In contrast, inert gas addition had little effect on the reaction of the n-paraffins, this and other observations indicating that with the n-paraffins both chain induction and breaking occurred on the walls. The appearance of oxidation at temperatures far below those of thermal dissociation of the molecule supported this view. 

The oxidation of two types of highly knock resistant compounds-arematics and ketones- at pressure less than atmospheric was also been investigated using a static apparatus.

Ketone oxidation was studied with methyl- ethyl ketone and diethyl ketone. With methly- ethyl ketone, reaction  only occurred at temperatures greater than 430º C. Within certain time limits the reaction speed followed the Semenoff equation. w = Ae Øt...1 and the reaction was therefore assumed to be of a chain character and to involve chain branching. The effect of changes in the wall surface of the reaction vessel indicated that it played an essential part for chain induction or breaking. The apparent activation energy found for the reaction was approximately 60 k. cals., which appeared concordant with the high kneck resistance of the compound. Carbon monoxide and carbon dioxide formed in the main reaction  were found to account for at least 85-90% of the carbon burnt. With increasing temperature, at a constant reaction time, the carbon monoxide formed reached a maximum at about 495º C. Here the oxidation of methyl ethyl ketone was essentially completed and higher temperatures resulted in oxidation of the carbon monoxide. At temperatures greater than 520ºC, explosion occurred, probably by a chain-thermal mechanism. The dependence of the oxidation reaction on pressure suggested a formula involving a minimum starting pressure.

In the pressure range used, cool flames were not observed. The effect of increasing the mixture strength was to displace markedly the curves to lower temperatures, and there was a considerable divergence in the calculated activation energies for the different mixture strengths. Diethyl ketone behaved similarly.

Using the same apparatus, the aromatics, benzene and teluene were studied. With benzene the temperature at which oxidation first became noticeable was 560º C, the reaction appearing after a definite induction period which decreased exponentially as the temperature was raised. Again the reaction velocity followed the Samonoff equation, but only within a very limited time interval; the reaction was considered to be of a chain character involving chain-branching. For the slow oxidation, an apparent activation energy of 62±6 k.cals. was obtained. The slow oxidation up to temperatures of 619ºC gave a pressure time curve of asymmetrical sigmoid shape and only the first development of reaction occurred exponentially. The asymmetrical course of the curve was attributed to the initial stages of the reaction resulting in the formation of carbon monoxide as a stable intermediate product which only underwent oxidation at the end of the reaction. At temperatures greater than 619ºC, this carbon monoxide oxidation resulted in a pure carbon monoxide explosion following the sigmoid shaped pressure-time course. The activation energy for this explosive reaction was calculated to approximately 89 k.cals. At 629ºC a "direct" explosion occurred without any preliminary sigmoid shaped pressure-time curve, the reaction prior to the explosion following the exponential law. The activation energy of this "direct" explosion was found to be the same as that of the pure carbon monoxide explosion. The oxidation of benzene was found to be proportional to a power of the total pressure. As with the ketones, increasing the partial pressure displaced to the oxidation curves to lower temperatures. Also, the ration maximum pressure change/initial pressure was found to be a characteristic of the hydrocarbon partial pressure and to increase with increasing stoichiemetric proportions. Inert foreign gases were found to aid the reaction by preventing the diffusion of active particles to the vessel walls. Helium had  a less marked effect than nitrogen. Both the addition of hydrogen and of acetaldehyde caused a lowering of the activation energy of the benzene oxidation. Acetaldehyde in particular caused a sharp decrease. There was a marked change in the course of the reaction because of active interference of the radicals resulting from the acetaldehyde. Oxidation of toluene was observed to follow in a similar manner to that of benzene, with the reaction occurring at somewhat lower temperatures. In all the experiments with benzene and toluene cool flames were never observed. The previously described investigations were all made by the static method. A flow type of apparatus was, however, developed for the kinetic investigation of very rapid homogenous gas reactions. The apparatus is shown in Fig. 1, throttle valves in the inlet and exit lines permitting the absolute pressure in the reaction tube to be changed. In the development of the apparatus the following requirements had to be met: 

(i) Concentration and ease of collection of the reaction products. 

(iii) Simple and definite standardisation of times of reaction of 10-1 to 10-4 seconds. 

(iv) Production of comparable concentration and temperature relationships in the reaction zone with the different reaction times. 

The first requirement was met by cooling after the predetermined reaction time. Wall reactions were excluded by separate heating and correct choice of reaction chamber dimensions, so that active material concentration at the wall could be kept negligible until the cooling zone was reached. For equal linear velocities of the two concentric gas streams, suitable dimensions were calculated to ensure that the active material concentration at the wall did not exceed a predetermined value. Reaction time was standardised by the choice of gas velocity and was altered by changing the absolute pressure so that the mass velocity remained constant. Assuming negligible chemical heat evolution, this ensured that the relative temperature and concentration fields would remain unaltered. Both these fields depend upon three dimensionless numbers which remain constant when the mass velocity is constant even though the pressure varies, providing the free path of the gas molecules is small in comparison with the vessel dimensions. 

The complete apparatus was never used because of the inadequacy of existing methods of analysis, but a model apparatus, for experimental work, was constructed according to the calculated dimensions. In this model apparatus provision was made both for sampling and determining the temperature at any point in the reaction zone. It was found that with different stream velocities the concentration profile remained unchanged, particularly in the region of greatest chemical change. Also, although widely different temperatures prevailed over the whole reaction zone, the temperature altered but little in the region of greatest chemical change; thus, investigations under isothermal conditions were possible, provided the heat liberated locally was slight. Experiments were made in the model apparatus on the effect of increasing the velocity of the inner gas stream so that the two gas streams entered at speeds considerably different from each other. In this particular apparatus, concentration of material at the wall of reaction zone remained negligible until the inner gas stream so that the two gas streams entered at speeds considerably different from each other. In this particular apparatus, concentration of material at the wall of the reaction zone remained negligible until the inner stream velocity reached 1500 cms/sec. Under atmospheric pressure this meant that the heating time could be reduced to 2 x 10-2 seconds, and even shorter times were possible under vacuum. So the apparatus could be used to investigate systems initially containing thermally labile materials, like tetraethyl lead of peroxides. Methods were given by Damköhler and Sander whereby for a homogenous gas reaction the reaction order and the apparent heat of activation could be directly determined in the complete apparatus (Fig. 1). In order to determine the absolute value of the velocity constant of the reaction, however, measurements of relative concentrations have to be made in the model apparatus for a particular type of reaction chamber to allow the calculation of an integral for the entire reaction zone. 

The method of sound dispersion has been suggested, as an alternative to spectroscope, for the determination of reaction kinetic data. It was assumed that it would be applicable in the high temperature region to study dissociation reactions of between 2 x 10-6 and 1 x 10 -3 secs., such as those occurring in flame and detonation fronts. The procedure would be to determine experimentally at constant temperature and pressure, the frequency function of the sound velocity of of the amplitude damping constant. Then from a general expression derived for those differential adiabatic exponent in dissociating gases, which is type and number of dissociation the real adiabatic be calculated by using assumed for the forward and reverse reaction velocities. Theoretical curves can then be calculated for velocity or amplitude damping constant as a function of frequency. The curves comparing best with the experimental curves are those obtained using correct values for the reaction velocities. The method is without ambiguity if only one dissociation equilibrium moves in the sound wave. Otherwise, a method employing successive approximations must be used. Moreover, only the forward and reverse velocities of the overall reactions are given, additional hypotheses being required to deduce the elementary reaction velocities.

B). Ignition Delay

Extensive laboratory experiments have been made on the self ignition of hydrocarbons because of its importance in relation to knock in the spark ignition engine and the ignition reactions in the diesel engine. A number of types of adiabatic compression apparatus, giving rapid compression of the gas mixtures, have been used to obtain the ignition delay with the elimination, as far as possible, of its thermal predecessors. The end of the ignition delay period has been determined either by the first appearance of luminosity or by the onset of rapid pressure rise. F.A.F. Schmidt has pointed out that some of the diversity in the result probably arises from this use of different methods of measurement, particularly where pressure is relatively low with long delay in ignition and slow burning in the initial stages without any violent pressure rise. It has been sought, by Zeise, to make a distinction in the ignition delay between the slow initial reaction process and the subsequent spontaneous ignition.

(a) Factors influencing ignition delay; in particular temperature and pressure effect.

From measurements made in an adiabatic compression apparatus, the variation of ignition delay with changes in mixture strength and pressure was thought by Jost, Rögener and Weber to be solely an indirect effect caused by variations in temperature produced by these changes. Results obtained by F.A.F. Schmidt, both in a bomb and in an adiabatic compression apparatus, indicated, in agreement with deductions made from engine tests, that in the temperature and pressure range occurring in the engine there is a marked direct pressure effect. Evaluation of the ignition delay results was obtained, for a limited temperature and pressure range, by the equation, this equation being an altered form of the expression.

Much experimental evidence was cited by Zeise to show that the reaction mechanism in ignition processes depends both on the temperature and pressure range involved. Zeise concluded that it is impossible to cover the whole temperature and pressure range for a given fuel with the same values of E and n. The reaction mechanism will vary from range to range with E and n, this variation having a general character of which two limiting cases are. From theoretical considerations Zeise showed that a pressure effect would be expected in certain regions. Thus if the activation energy at an assumed constant pressure of a reversible chemical process in ignition, is replaced by the variation of the free energy, equation leads to. The free energy change, however, is dependent on the pressure since, for example, with an ideal monatomic gas as reactant. Often, also, in ignition processes the energy is distributed on more of the quadratic terms i.e. So to determine the real activation energy the modified Arrhenius equation must be used. The apparent activation energy, which would be obtained from ignition delay measurements if the simple Arrhenius equation were valid, is related to the real activation energy as in the equation. Thus various values of E can be obtained from on value of Es, according to the temperature and number of degrees of freedom involved. The latter will depend on temperature, pressure, and prevailing reaction mechanism.

Ignition delay tests made in a bomb by Lonn showed that with nitrogen-oxygen mixtures it was primarily the oxygen partial pressure that determined the delay. He also showed that the effect of the addition of lead tetraethyl was to increase the delay, the effect being relatively greater at low pressures. F.A.F. Schmidt using a compression apparatus showed that ignition delay is increased by the addition of tetraethyl lead, although the main properties of a fuel as regards the pressure and temperature effect on the process remain unchanged. 

(b) The delay in ignition of fuel sprays.

The commencement of combustion of fuel sprays was theoretically considered by Dreyhaupt, who based his considerations on a typical individual fuel droplet. 

Initially, the fuel droplet being brought into the hot air in a short finite time will be surrounded by very little vapour. Thus, as shown in curve (1) of Fig. 2, there will be a steep rise in temperature with increasing distance from the droplet surface.

Later the temperature course flattens out because of vapourisation from the surface of the droplet, see curves (II) and (III). 

Curve (Ia) represents the temperatures necessary for ignition of the fuel air mixture under the initial conditions shown in curve (I). Similarly, curves (IIa) and (IIIa) correspond to conditions in the later stages, as shown in curves (II) and (III). It is not until the third stage, that the fuel vapour temperature curve (III) touches the ignition temperature curve (IIIa) and ignition ensues. Thus a slightly higher temperature is needed for ignition of the fuel droplet than that required for ignition of the most ignitable mixture.

Progressive combustion of the fuel droplet depends on diffusion processes for the supply of oxygen. Heating, evaporation and diffusion of the residual droplet is accelerated  by the release of reaction heat from the ignition zone. Dreyhaupt gives mathematical and graphical ways of obtaining the heat required by any fuel droplet for any interval of time up to the actual ignition of the droplet. These considerations allow for the variation in mixture strength at different distances from the droplet surface, and also for the gradual reduction in droplet size as evaporation occurs. Lack of data on fuel properties and engine conditions, however, makes the results only qualitatively accurate.

An interesting conclusion from these considerations of the initiation of combustion of a fuel droplet is that if the individual fuel droplets in a spray are too small, ignition will not occur. The droplets will have completely evaporated before reaching the ignition temperature. Referring to Fig.2, vapourisation will be complete and the vapour concentration at the former position of the residual fuel droplet surface will start to decrease before the fuel vapour temperature curve touches the ignition temperature curve. The decrease in vapour concentration will result in a reversal of the movement of the ignition temperature curve, so making ignition impossible.

The author shows that very large drops in the centre of a fuel spray may have insufficient air to support combustion; and, in the centre of a spray, the mass of oil in relation to the mass of air may be such that the drops are cooled so as to be non-ignitable. 

The fuel spray envelope must contain some droplets of molecular dimensions, nevertheless the ignition delay-time is of much greater order of magnitude than of the impact time for the fuel molecules. It therefore appears that a certain droplet size is necessary for ignition. 

The ignition and combustion of fuels injected into a bomb were studied by Blume by means of high speed photography and pressure measurements. Using gas oil, injected through commercial nozzles, ignition delays of approximately 2 x 10-3 seconds were obtained at temperatures of 500ºC and pressures of 35 atms. Photographs showed that only a small proportion of the fuel spray had evaporated at the moment of ignition. Using a flat-seated nozzle, ignition occurred nearer the nozzle, and the delay period was shorter because of the better mixture formation. The combustion speed was calculated from the observed pressure rise in the bomb, and the duration of the ignition delay was found markedly to affect the course of combustion. As ignition delay decreased, the maximum pressure attained also decreased and the combustion speed became  more even over the entire combustion period. An increase of the air temperature reduced the ignition delay but increased the total contribution time. This was attributed to the poor mixture formation in the later stages of  combustion because of insufficient turbulence in the bomb.

The effect of fuel sprays striking the walls was also studied. Gasoline was reflected from the wall surface. This was attributed to the gasoline being in the form of a liquid rod or stream in consequence of the low air densities employed. Diesel fuel differed, since atomisation gave a spray  behaving more nearly like a saturated vapour. This spray maintained its form after collision with the wall and traveled parallel to the wall surface instead of being reflected. Diesel fuel thus only reached the wall in special circumstances, and could be completely prevented from doing so by maintaining the wall temperature above 550ºC. 

A jet striking the wall had a smaller ignition delay than a non-striking jet. The ignition delay for a striking jet decreased with increase of wall temperature, until at about 700ºC the delay was approximately half that of a non-striking jet. The total combustion time for a striking jet increased with increasing wall temperature and was about 30% greater than that of a non-striking jet. This resulted from the striking jet being unable to expand on the wall side.

Mention should here be made of another series of experiments made in bombs shaped like combustion chambers. The purpose of the experiments was to clarify the factors affecting heat losses in engines. The fact that the heat loss up to the pressure maximum occurring after ignition  decreased with increasing initial bomb pressure emphasised the thermodynamic advantage of supercharging internal combustion engines. Consideration of the relationships of pressure and temperature after completion of reaction resulted, for the heat transfer coefficient, in the empirical formula. This differed considerably from an earlier relationship derived by Nusselt. The heat losses in small bombs were found to be very great, and it was concluded that it is advantageous to build an engine with a few large cylinders rather than with many small ones. 

Experiments have been reported which were made to investigate the theory that the energy losses in a knocking engine are due to the greatly increased heat transfer coefficient of the fuel/air mixture and exhaust gases under these conditions. In these experiments propane-air mixtures in a cylinder were rapidly compressed by means of a piston. The gas mixtures were then suddenly released into a combustion chamber and ignited while in a turbulent state. Flame photographs and pressure measurements were taken. The calculated heat transfer coefficient for combustion with violent oscillation was sometimes as much as 20 times greater than that for combustion with damped oscillation. Thus abnormal heat losses do occur during knock.

(c) The reaction mechanism in the ignition process. The spontaneous combustion of pure hydrocarbons by adiabatic compression was studied by Jost and reported in BIOS Final Report No.532. For such reactions he obtained pressure-time traces if the type shown in Fig.3. In his considerations of such a curve, Jost assumed that very little chemical change occurred, and so the increase in the number of molecules in the system was slight. Thus the value of the absolute temperature could be calculated from the pressure, and dp/dt taken as a measure of the rate of temperature rise dt/dt. The first delay period could be approximately represented by an exponential temperature function. Assuming the action during this delay period to be purely thermal, the rate of heating would be given by the equation. 

Jost calculated dt/dt for different values of t from an experimental curve, and so obtained values for tan = tan dt/dt. Consecutive values of tan did not however, bear the same relation to each other as the corresponding values of eE/RT, when a reasonable value of E was used. Thus he concluded that in the first induction period a chain, and not a thermal reaction was proceeding. He also concluded that the rate of change of reaction velocity with temperature in the region c-d was far too great to be accounted for by thermal mechanism. For the evaluation of the second delay period, development of the true value of the rate of the primary reaction and use of the hypothetical final level for this reaction enabled the dotted curve for the second part of the primary reaction to be drawn in Fig.4. Thus, two possibilities arise:-

(I) Accepting the analysis which allowed the second part of the primary reaction to be drawn in Fig. 4 resulted, as suggested by the asymmetric S-form of the primary reaction curve, in the conclusion that chain breaking occurred at the end of the primary chain reaction. Partly overlapping this was a secondary reaction which could have been of a thermal nature, although Jost favoured the concept of a chain mechanism. (II) Otherwise, a reaction velocity with negative temperature coefficient, becoming zero at the point of inflexion, was indicated by the course of the pressure rise. This reaction, which would follow on the primary reaction immediately after the velocity maximum (first point of inflexion), was considered by Jost to be unlikely. 

BIOS Final Report No. 532, referring to an interrogation of F.A.F. Schmidt, gives a diagram of a pressure-time plot obtained by him in an adiabatic compression apparatus. In this diagram the appearance of two separate delay periods is not evident, but in the delay period that does appear, there occurs a slight pressure rise corresponding to a temperature increase. This is suggested as the result of exothermic reactions of both the thermal and chain types proceeding during the delay period. The fact that when evaluating the delay periods by an equation of type (3) the value of "n" generally lay between 1 and 2 was thought to indicate that several reactions were proceeding simultaneously. 

Further evidence for occurrence of chain reactions in the ignition process was furnished by Zeise who evaluated some results obtained in an adiabatic compression apparatus on the basis of his previously mentioned subdivision of the ignition delay into the slow prereaction period followed by the true ignition reaction.

The slow prereaction period gave a linear log vs 1/T graph, the gradient of which gave activation energies of the order of those found in the kinetics of reactions with free radicals.

C.) The Mechanism of Explosive Gas Reactions

In the reaction theory of N. Semenoff, the criterion of an infinite reaction speed for explosive conditions in a chain branching reaction leads to the conclusion that both branching and breaking must be first order reactions. The assumptions, however, of a stationary condition and constant branching probability used as a basis for Semenoff's calculations do not correspond to conditions in practice.

Muffling, therefore, considered the general differential equation for the time change of concentration of chain carriers in which the speed of chain branching is proportional to the concentration of the starting material and that of the chain carriers. Using as a basis equation (12) and the equation he investigated the time factors of theoretical reactions having chain breaking of the first and second and simultaneous first and second order.

Calculated transformation-time curves and reaction velocity-time curves for first order breaking are shown  in Figs. 5 and 6, respectively. 

The differently shaped curves were obtained from different values of k a, the velocity constant of the chain breaking reaction. Similarly calculated curves for second order breaking are shown in Figs. 7 and 8, and are of a type that had been frequently observed in oxidation  reactions, (e.g. Fahlbusch loc. cit.), although an explanation was previously lacking. Thus, from the experimental curves, it appears possible to make a conclusion as to the order of the chain breaking reaction. From the calculations with first and second order chain breaking processes, an explosion limit curve was calculated with the upper and lower limits typical of a chain explosion.

Muffling considered that it was difficult to determine if explosion occurred with strictly or almost isothermal transformations, and it was therefore best to use the term "explosion" only when transition into heat explosion occurred. Explosion is usually taken to mean the complete consumption of the reactants in a very short time, but in first order breaking reactions 100% transformation only occurs if there is no chain breaking, since, as seen in Fig.5, the transformation depends on the ratio k1c10:k. This difficulty does not occur in reactions involving second order chain breaking since these reactions never quite come to a standstill. The velocities reached with first order breaking, and are often sufficiently high for the processes to be considered as explosions, particularly as where second order chain breaking occurs conditions favour a thermal explosion.

D.) Flames.

Hubner and Klaukens derived an indirect method for determining the true thickness of the luminous zone of the inner cone of a bunsen burner flame, from the measurement of the intensity distribution of a photograph of the flame. The zone width observed depends on the radial distribution of luminous intensity in the flame, and although the distribution of luminous intensity in a photograph is not the same as this true radial distribution it was shown to depend on it. 

Stationary flames and explosions at low pressures were studied using acetylene-oxygen, acetylene-argon-oxygen, acetylene-air, propane-air, pentane-air and benzene-air mixtures. Both forms of combustion could be obtained down to the pressure limits imposed by the experimental equipment (3-4 mm Hg for acetylene mixtures.) Tube diameter and gas flow had to be increased, however, as the pressure was lowered. The limiting pressure for stationary flames varied inversely with the tube diameter.

At very low pressures the thickness and volume of the luminous cone of the flame were reported to be considerably increased. Sander, however, considers that the volume and thickness only appear to increase because of oscillations of the flame. 

Spectroscopic investigations have been made on the flames which originate from the reaction of atomic oxygen with acetylene or methyl alcohol. With methanol it was found that the hydroxyl groups already present in the alcohol played no part whatsoever in the hydroxyl emission spectrum. With the acetylene flame, the weak bands occurring between 3000 and 4000 A, which von Vaidya had attributed to the HCO radical, were investigated. The emission spectrum of C2D2 was also studied, whereby it could be ascertained that the band carrier contained hydrogen.

These "hydrocarbon flame bands" were also reported to have been quite clearly observed in the emission spectrum of the flame in a diesel engine. In a spark ignition engine Zeise observed that in the decrease of intensity of the OH bands at 3064 A when knock occurred, only part of the fine structure (rotational lines) near the band head disappeared. The intensity of the lines further away from the band head was not decreased so much. This phenomenon was attributed to an induced predissociation of hydroxl radicals by collision, a process which would be of importance in furnishing chain carriers for preflame reactions.

Further information on the reactions occurring in flames is given in BIOS Final Report No.532. In diffusion flames of rich mixtures of hydrocarbons burning in air, cracking occurs with the formation of carbon particles, and the location of the cracking zone depends on the stability of the hydrocarbon. Thus benzene breaks down on approaching the flame front and the carbon aggregates do not penetrate the front but are carried upwards and discharged from the tip of the flame. Acetylene, being more thermally stable, is partially decomposed in the burning zone, and the carbon, having a relatively slow burning rate is, for the most part, discharged from the outer boundary of the flame. 

Formation of hydrogen in the cracking process results in flame disturbance. At very high flame temperatures, of the order of 8000ºC, dissociation of fuels is reported to cause the formation of NO and CO instead of water and carbon dioxide, as in combustion at lower temperatures.

The effect of turbulence on the ignition velocity of a coal gas-air mixture has been investigated in vertical and horizontal tubes. Using spark ignition, the flame velocity in the direction opposite to that of the gas flow was measured by ionization gaps. Both with increasing fresh gas velocity and tube diameter the ignition velocity increased, but not as an explicit function of the Reynolds number. The ignition velocity, however, was affected by the flow conditions inside the tube. From the experimental results two equations were derived for the ignition velocity, one for laminar and the other for turbulent flow.

Turbulence cannot be characterised by one quantity; and Damköhler attributes the differing influence on flame speed to the size of the turbulence eddy (equivalent to the Prandtl mixture path) relative to the laminar flame thickness. Eddies are classified as coarse or fine, depending on whether they are greater or less than the flame thickness. 

Flame velocity was shown to increase considerably with the onset of fine spherical turbulence, as to result in distribution of the flame front over a wider zone, so increasing the burning rate. Using propane-air mixtures E. Schmidt et. al. studied the influence of turbulent movement of gases on their combustion rate. Flame and Schlieren photographs were taken of the combustion inside a long steel tube fitted with glass windows. With the combustion tube closed at one end, the flame velocity when the gas mixture was ignited at either the closed or open and was found to be altered by vibrations and and turbulent motion of the gas inside the tube. 

When the combustion tube was closed at both ends, however, ignition of the propane-air mixture at one end of the tube gave a flame, the velocity of which was mainly influenced by pressure and temperature changes. 

2. Combustion in Spark Ignition Engines

Most of the research made on spark initiated combustions has been concerned with engines. Experiments have, however, been made to determine the effect of grinding sparks on ignitable mixtures of a number of gases. From these experiments conclusions were made as to whether or not spark free instruments are necessary when working in the presence of these mixtures. It was also shown that the upper limit for ignitability with grinding sparks may like below the most explosive mixture. The ignitability of mixtures by grinding sparks was found to correspond with results using the criterion for safety from explosion with electrical traction - i.e. whether or not ignition occurs through a 0.8 mm mesh. 

The most important and widely studied phenomenon of combustion in spark ignition engines is that of knock . One report describes it as occurring in a finite time range of approximately 10-4 seconds. The knock oscillations run back and forth in the combustion changer as pressure waves of great amplitude, the steepness of which depends on the form of the combustion chamber. Although the primary wave of the knock oscillation is usually equal to that of the self oscillation of the chamber, oscillations of a higher order can be stimulate. The apparent pressure propagation speed is about 950 m/sec., and as strong knock the extent of the knocking zone may amount to several cms. The amplitude spectra of the knock shocks, even at frequencies of several 10,000 Hertz, still show amplitudes of some  atmospheres, so that the knock disturbance may be explained as cavitation phenomena.

A.) Cause of Knock

Knocking results from the combustion of the residual charge, and the cause of this combustion has generally been attributed to spontaneous ignition. Dreyhaupt considered the possibility of other causes. He concluded that the formation of knock oscillations without pressure increase, by resonance of the self oscillations, was highly improbable as was also the postulation of detonation to explain knock. He did not exclude, however, the preparation of the uninflamed charge by shock waves.

The prereactions which determine the residual charge have been studied in an externally driven engine, without ignition. The temperature increase of the mixture during passage through the engine was taken as a measure of the prereactions. These were shown to depend markedly on temperature and induction time. A strict relationship between octane number and the prereactions was only found in chemically similar fuels. Raising the octane number by lead addition generally, but not invariably, caused a marked decrease in prereactions. It was observed that in some indicator diagrams a marked hump near T.D.C occurred at every alternate cycle. This was assumed to be due to some residual gas being mixed with the fresh charge. Thus gas containing end products of prereactions would probably inhibit such reactions in the fresh charge, whilst gas containing products of prematurely interrupted prereactions would probably promote further reaction in the new charge.

The action of residual gases will vary with the mixture strength, and lean mixtures will probably have a mainly thermal effect. Analysing some of the products of these prereactions Damköhler and Eggersgluss found the ratio of higher aldehydes to formaldehyde. This was such that, assuming the higher aldehydes were formed by progressive degradation the formaldehyde must have been produced by a different process, probably by radical chains.

The methods available for the quantitative determination of aldehydes in gasoline have been compared by Widmaier. He concluded that of the usual recognised methods the hydroxylamine hydrochloride method is the most suitable as the others give results that are too low. From experiments in which butylaldehyde and benzaldehyde were added to leaded and unleaded gasolines he concluded that the effect on the antiknock value of the gasolines by the aldehyde group in these compounds was very small. The main effect could be attributed to their molecular structure.

Similar experiments made with peroxides showed that both their active oxygen content and their molecular structure greatly influenced the octane rating of the gasoline. A comparison of the stannous chloride method and the thiocyanate method for determining peroxides in gasoline, favoured the latter.

The absence of a method of determination which is specific for any one peroxide or group of peroxide led to an attempt by Eggersgluss to effect group separation by chromatographic adsorption. Six groups of peroxides were studied in experiments made using various solvents and adsorbents. 

The various peroxide groups each showed a characteristic behaviour, and the adsorption was primarily dependent on the kind of polar group present in the peroxide. It was claimed that it was practical possibility to devise a method, involving suitable combinations of adsorbents and solvents, to separate an unknown mixture of peroxides into the various peroxide groups.

B.) Prevention of Knock. Phenomena Occurring with High Valve Overlap

In his previously cited paper Dreyhaupt suggests the astatic fading out of the pressure rise resulting from the spontaneous ignition of the residual gas as a means of counteracting the obnoxious effects of knock. Shaped piston crowns may function as knock oscillation dampers as well as hindering any shock waves which originate from the primary flame. He also draws attention to the large compression ratios permissible in sleeve valve engines, presumably because of the absence of hot spots likely to cause glow ignition knock.

Tetraethyl lead was almost exclusively the only anti-knock material used in practice. It was usually limited to a maximum of 0.12% vol. At one time the question of raising the anti-knock value of 87 octane number gasoline to 100 octane number, by the addition of methyl aniline and by increasing the content of tetraethyl lead to 0.16% vol., was considered. It was not found practicable, however, because of valve corrosion etc., caused by the increased lead concentration. 

Some experiments have been reported in which a fluid of the composition  : - Gasoline 50%, Iron Carbonyl 40%, Methyl Aniline 10%, was added to fuel to give a concentration of 0.05% of iron carbonyl. 

Other methods, besides the addition of anti-knocks were used in Germany to limit knocking. It is reported that coating the piston with fine colloidal graphite gave quite good results in preventing knock. With highly supercharged engines the D.V.L. distributed injection method was used, in which the bulk of the fuel was injected during the compression stroke and the combustion period. This resulted in a flattening and considerable raising of the knock limit curve, particularly with lean mixtures. Increases of mean effective pressure of 2-4 kg/cm2 have been reported.

Considerable research was made on the knock behaviour of fuels in engines with large valve overlap. The general effect of valve overlap in raising and flattening the knock limit curve was fully appreciated by the Germans. Valve overlaps as great as 120º were used in some aero engines and enabled very good weak mixture performance even with highly aromatic fuels.

The advantageous  effects of valve overlap resulted from scavenging of the residual gases and cooling of the cylinder walls, valves, etc. Use of high valve overlaps, however, introduced the problem of how far it was possible to apply the knock limit curves obtained by the D.V.L. supercharge testing method to main engines with large valve overlaps. Often, instead of the conventional knock limit curve with its minimum near stoichiometric mixture strength, a knock limit curve was obtained with its minimum far in the rich region. Although most experiments, in which this type of curve was obtained, were made by the DB 601 engine, it was recognised that such a curve is not necessarily a characteristic of or peculiar to liquid cooled engines. Indeed, Penzig, in his experiment using the DB 601 engine with a valve overlap of 113º, seldom observed such curves.

The knock limit curves having their minimum in the rich region were also found to have a reverse temperature sensitivity; i.e. knock free performance was actually improved by raising the boost air temperature. With highly aromatic or alcoholic fuels the phenomenon of a knock limit curve minimum in the rich region was not observed. 

Franke has explained the formation of the knock limit curve with its minimum in the rich region and allied phenomena by invoking Callendar's theory for the formation of peroxides on droplets. In the DB 601 engine the mixture formation had been reported to be poor. This poor mixture formation would result in the formation of unvapourised fuel droplets increasing in proportion with increase in the mixture strength, and these droplets in most instances would be conducive to peroxide formation. Due to this peroxide formation the knock limit curve would be expected to decline continuously as the mixture is enriched. The conventional type of knock limit curve, would be expected with aromatic and alcoholic fuels, since aromatics are little affected by peroxidation in the presence of droplets; and alcohols; although easily peroxidised, require very little energy of activation for further oxidation. Their resulting products of oxidation do not lead to rapid reactions which can produce knocking because of their low energy of dissociation. The reversed temperature effect is readily explained; the increase of the temperature will hinder droplet formation, and thus the knock limit curves will be raised and their minima displaces further into the rich region. Indeed, above a certain temperature no droplets will appear for a certain range of mixture strength, and so a minimum will appear around the stoichiometric mixture strength. A second minimum, however, will appear in the extreme rich region. Fig. 9 is a three dimensional representation showing how the knock limit curve is influenced by this effect of temperature on the mixture formation. 

It was shown by Franke that with regard  to the characteristics of the knock limit curve, neglecting the position of the curve, a relationship existed between boost air temperature and valve overlap. Thus, as far as the shape of the curves is concerned, the effect of valve overlap can be compensated by a corresponding alteration of the boost air temperature. As would be expected, the temperature change required to compensate for additional increase in valve overlap is smaller than that required for the first increase, since once scavenging is complete increase of valve overlap has a smaller effect in reducing internal thermal stress. 

In the same paper were reported a series of tests made to ascertain the influence of valve overlap on the anti-knock value of fuels; i.e. on the position of the knock limit curve. These tests were made at high boost air temperatures, so that in all cases a minimum occurred near stoichiometric mixture strength. This facilitated comparison of results. Fig. 10 shows the change of the knock limit at the minimum with varying valve overlap. Due mainly to the decrease in internal thermal stress, it is seen that the antiknock value of all fuels tested increased up to 80º valve overlap. The effects of further increase in valve overlap were irregular, in spite of the fact that thermal internal stress must have been further decreased. The anti-knock value of the fuels with high thermal sensitivity, i.e. benzol and alcohol blends, actually decreased. This outweighing of the effect of decrease of internal thermal stress was attributed to the fact that with these fuels, residual gases favourably affect the kinetics of reaction influencing the knock behaviour. Thus increased scavenging reduced the residual gas quantities until they were no longer of any importance, and so their favourable influence on the knock limit was lost. Whether or not decrease of internal thermal stress outweighed the loss of residual gas was dependent on the type of fuel.

In a later paper Franke described a series of tests to ascertain if correlation with rating by the D.V.L supercharge method could be obtained from knock limit curves having their minima in the rich region. These tests were made on numerous fuels of differing chemical constitution, in the BMW 132 N, DB 601 A and DB 601 E engines.

Use of two valve overlaps (40º and 120º) with the DB engine gave the two types of knock limit curves, from which it could be seen whether any variation in rating was due to differences in engines or solely to the different shapes of the knock limit curves. Knock limit curves obtained from values of the mean effective pressure were completely temperature insensitive in the rich region when low aromatic content fuels were used. Thus the knock limit curves were plotted using absolute boost air pressure values.

Three graphs were plotted:- 

(I) Minima of the knock limit curves for the various fuels at 1.05 for the DB 601 A vs. those for the BMW 132 N. 

(II) Anti-knock values at 1.05 for the DB 601 E vs. the minima at 1.05 for the new BMW 132 N.

With a few exceptions the ratings seemed superficially to be in good arrangement. For the results obtained from the two engines to be in complete agreement, however, it would be necessary for the graph curves to be straight lines passing through the origin at 45º. Although this was approximately true in one graph, others differed in slope. Each difference in gradient results in the requirement of a different constant for the conversion to the BMW 132 N values. Thus it was evident that fuel rating by determining the anti-knock values according to the D.V.L supercharge method in the BMW 132 N engine could not be used for engines with knock limit curves of different characteristics. Two engine types were recognised- the one where the anti-knock value of fuels rose with increasing boost air temperature, and the other where it fell. The fact that it might be possible to choose a temperature where test and theoretical curves agreed would not permit the fuel rating in one engine type to be applied to the other engine type. It was found that, quite accidentally, the value of 130ºC for the boost air temperature of the D.V.L supercharge test gave a good agreement between the anti-knock values in the DB 601 E and BMW 132 N, provided only the value at 1.05 was considered.

It was concluded that where only a general evaluation of the anti-knock value of any fuel was required, the normal D.V.L. supercharge method with the BMW 132 N cylinder could be used. Even for this purpose, determination of the order of rating for fuels in engines giving knock limit curves differing from those obtained with the BMW 132 N cylinder is only permissible for fuels with a low aromatic content. Compared with the DB 601 E cylinder, the BMW 132 N cylinder underrated gasoline-benzol blends but, surprisingly enough, rated mixtures with a high isopropylbenzene content too favourably. For the temperature characteristics and the extent of the knock region, however, it is essential to test the fuels in engines which give knock limit curves of a similar kind over the whole operating range.

C. Knock Determination

(a) Methods.

(I) Octane Number Determination. The small mono-cylinder engines used in Germany for this work were the standard C.F.R knock rating engine and the German I.G. engine. Automobile fuels were originally rated by the C.F.R. Research Method, but from 1st April, 1943, the Army adopted the C.F.R. Motor Method as corresponding better with operating conditions in military vehicles. Day to day control of aviation fuel production was made using both the I.G. and C.F.R. engines operating according to the C.F.R. Motor Method. For the determination of the practical knock limit, however, rating of aviation fuels under Motor Method conditions was known to be too severe for all but highly paraffinic fuels. Thus the I.G. Oppau Method was developed in Germany having milder test conditions corresponding to the C.F.R. Research Method. The operating conditions in the Oppau Method are 

(I) Speed   600 RPM (as in Research Method)

(II) Ignition  22º BTDC

(III) Coolant Temperature 100º C ( as in Research Method)

(IV) Mixture Temperature 125ºC

(V) Inlet Pressure  1000 mm Hg. 

(VI) Carburettor setting variable between 0.7 and 1.2 

The Oppau Method is a multipoint one but is simpler than the supercharging method in the BMW 132 N cylinder with which, however, it is in reasonable agreement. It was reported that, in general, rich mixture running has a greater influence in the BMW than in the I.G. engine. With routine tests by the Oppau Method, the knock behaviour is read off the octane number dial has been calibrated previously with reference blends at mixture strengths always corresponding to maximum knock.

The method of test is to vary the compression ratio until incipient knock occurs (knockmeter reading 50), with the carburettor adjusted to the strongest knocking mixture. The knock behaviour is then read off the octane number dial. The mixture strength is changed and the knock decreases. To restore it to its original value (knockmeter reading 50) the compression ration is raised. Thus another octane number is obtained. From six to eight such octane numbers an octane number vs graph is plotted.

Under the conditions described, this method was found unsuitable for testing unleaded fuels, or fuels with particularly high benzene or alcohol contents. 

(2) Supercharge Testing. The failings of single point methods, such as the C.F.R. methods of octane number determination, for evaluating the behaviour of a fuel in an aero engine are well known. Seeber has outlined these. Correlation between small engines, like the C.F.R., and full size engines was also found difficult.  A method was therefore adopted by the D.V.L of examining aviation fuels in a single cylinder aero engine under conditions approximating as closely as possible to the actual stresses on the fuel in the full sized engine. The test conditions finally employed in the D.V.L Supercharge Method are given in BIOS Final Report No.119 as (I) Speed 1600 RPM, (II) Intake air temperature 130ºC supercharged (III) Valve overlap 40º (IV) Cooling air pressure 300 mm Hg. (V) Oil temperature 90ºC (VI) Cylinder Head Temperature 240-250º C maximum (VIII) Ignition 35º BTDC.

The variable in this method is in the boost pressure, which is provided by an externally driven compressor. As the mixture strength is altered, the boost pressure is varied to give the same knock. (6-10 knocks/minute by ear). A knock limit curve is thus obtained by plotting these values of boost air pressure as a function of air/fuel ratio. The knock limit curve may also be drawn in the form of mean effective pressure as a function of air/fuel ratio. 

An attempt was made to develop a supercharge method using a small motor, which would give results agreeing with those obtained by the D.V.L Supercharge Method using the BMW 132 cylinder. The N.S.U. 501 O.S.L. motor was used for this purpose. The order of rating fuels in the BMW and NSU engines was the same, but the knock limit curves obtained with the NSU engine were flatter and closer together. Also, the highest mean effective pressures obtained were below those obtained with the BMW engine because the NSU engine was not designed or the test conditions used.

Other small supercharged engines, also, were developed on the principle of the D.V.L. Supercharge Method in the BMW 132 cylinder, and tests were made on the I.G. engine, in which the compression ratio was kept constant and the boost pressure was varied.

In 1944, Witschakowski reported supercharging tests made in a D.B. cylinder and a BMW 132 cylinder. Instead of the usual conditions of the D.V.L. Supercharge Test, the conditions used correspond more with main engine operation. From these tests it was concluded that a number of aviation fuels used at that time were not being exploited to the full in  the existing aero engines.

(3) Main Engine Tests. In BIOS Final Report No. 532, fuel rating in full scale aero engines was reported to be little advanced in Germany as the results with the D.V.L. Supercharge Method in the BMW 132 N cylinder had proved sufficiently reliable.

(b) Accuracy of knock determination

(I) Octane Number Determination. From 1936 to 1943 a series of investigations were made by the various German test laboratories to ascertain the accuracy of octane number determination.

In 1940, Wilke compared U.S. and German results and attributed the poorer accuracy obtained in Germany to lack of experience the use of both the C.F.R. and I.G. engines, lack of standard sub-reference fuels and the wide range of synthetic fuels tested. He gave the value of ± 1 Octane Number as the mean error for the C.F.R. engine, under both Research and Motor conditions. Using the I.G. engine and its octane number dial, the mean error, as determined with the use of reference fuels, in the same engine, was ± 1.27 Octane Numbers.

From 1940 to 1941 the main purpose of the research group as the unification of the sub reference fuels and transfer graphs for octane number evaluation. Singer and Wilke have reported on this. For secondary reference fuels I.G. standard gasoline (replacing heptane) and the purely pafaffinic fuel "Z" (replacing iso-octane) were introduced. All testing stations used one calculation graph instead of individual curves for standard gasoline and standard reference fuel "Z", and the limit of accuracy for a storable petrol was found, over a large number of engines, to be ± 1.5 Octane Numbers. It was thought that by using heat sensitive and heat insensitive fuels some idea of the heat standard of the engine would be gained. However it was found that the spread of points hardly changed when a heat sensitive or heat insensitive fuel was used. The accuracy of blending values (i.e. blending octane numbers) was reported as ±4 Octane Numbers. Neumann reported that the I.G. test motor gave values of 0.8 and 0.4 Octane Numbers less than the C.F.R Research and Motor Methods respectively. He also commented on the considerable spreads obtained in the knock measurement of synthetic fuels, and mentioned that better results are obtained when a reference fuel similar to the fuel under test is used.

The position of German knock measurement after the completion of the 10th series of investigations was discussed at the conference of the Knock Panel at Oppau in 1943. Sixty laboratories with one hundred and eight test engines took part in the investigations, in which four automobile petrols and four aviation petrols were rated according to the Research and Motor Methods respectively. In both methods the average accuracy was ± 0.6 octane numbers, 82% of all the results falling within the previously accepted limit of ±1 octane number. For the Research Method the measured value in both the I.G. and C.F.R. engines was on the average equal. In the Motor Method the value for the I.G. engine was on average approximately 0.3 octane numbers lower than with the C.F.R. engine. However, this difference lay within the known limits of accuracy. These limits of accuracy did not apply to synthetic fuels. For example, Fischer-Tropsch gasoline, although giving better reproducibility when the secondary reference fuel "Z" was admixed, still gave greater variations than normal gasoline. Such variations had been suggested as due to (I) Effect of some physical differences in the composition of the petrol (e.g. due to vapourisation losses in storage and handling). (II) Effect of some chemical influence (e.g. peroxide formation). (III) Effect of peculiarities of individual test engines, and in method of measuring to which synthetic petrols react in some special way. 

Even when precautions were taken to eliminate the effect of (I) and (II), ratings still gave average variations of ±1.2 and ±1.4 octane numbers. 

It had been previously thought that knock measurement was more reliable the more the reference fuel resembled the fuel under the test. Three benzol blends, however, were rated by different test laboratories using (I) Reference Fuels i-octane and n-heptane. (II) Secondary reference fuels benzene and I.G. standard gasoline. (III) Secondary reference fuels "Z" and I.G. standard gasoline.

The assumption that the reference fuel must conform to the fuel under test was not established. It appeared that secondary reference fuel "Z" would be suitable for all types of samples.

(2) D.V.L Supercharge Method-Wenzel considered that the limits of error laid down in the regulation of 1940, namely Pme ± 4% and charge pressure ± 1.5% were too stringent.

Making tests on different days seemed markedly to affect results. He concluded that, using the simplified D.V.L procedure (constant spark advance), even with the greatest precautions, limits of error of less than ± 5% could not be expected. 

A detailed consideration of the spread limits of supercharge curves in a single test engine was made by Seeber , in which fuel side influence and the influence of experimental technique were differentiated. He also presented data obtained on the reproducibility of knock limit curves using identical fuels in several test engines, and concluded that, even where several motors are used, the following error limits should not be exceeded: Pme ± 2% for paraffinic and isoparaffinic fuels, Pme ± 4% for aromatic fuels (those containing more than 35% of aromatics or unsaturated hydrocarbons).

If greater errors occurred in spite of all control and experimental precautions, he recommended that the cause of errors should be investigated by use of the D.V.L pressure acceleration procedure.

It appears that even these error limits may have been too strict, and BIOS Report No. 119 gives the reproducibility of performance curves as to within 4-6%.

Penzig showed that better reproducibility could be obtained by using liquid cooled cylinders. Ease of repeatability was also greater the flatter the knock limit curves.

Witschakowski suggested that for ordinary fuel testing the simplified procedure was adequate since any advantages of variable ignition are annulled by poor repeatability. He also stressed the necessity for identical engine details if good comparability between several test engines were to be obtained.

The importance of using the lubricating oil specified for the supercharge test was apparent from investigations made by Franke. He found that, under D.V.L. supercharge conditions and also main engine operational conditions, the lubricating oil had an effect on the antiknock value of the fuel. The magnitude of this effect was dependent on the aromatic content of  the fuel, and only became appreciable for fuels with a high aromatic content . It appeared, however, that the lubricating oil had no effect on the lead susceptibility of a fuel. 

(c) Reference Fuels.

(I) Octane Number Determination. The C.F.R. and I.G. engines were used for testing fuels of less than 100 octane. As reference fuels for such tests iso-octane and n-heptane were used. Reference has already been made to the use of I.G. standard gasoline and "Z" gasoline as secondary reference fuels. The secondary reference fuel "Z" has the advantage over benzene that its standard curve is a straight line, giving simpler reading and easier extrapolation (if required) for Octane Numbers greater than 100. It also has better anti-freeze properties and lead sensitivity.

(2) Supercharge Testing. Reference fuels used in the D.V.L. Supercharge Test depended on the type of fuel under test.

(I) For examining B4 fuel, reference fuel "Eich B 4" was used, the latter being a Leuna hydrogenation gasoline containing 0.12 % vol. of T.E.L.

(II) For examining C 3 fuel, reference fuel "Eich C 3" was used, the latter consisting of 80% DHD gasoline from I.G. Ludwigshafen, 20% iso-octane and 0.12% vol. of T.E.L.

(d) Knock measuring instruments.

In the I.G. and C.F.R test engines the contact bouncing pin was in general use despite the known inaccuracies in this method of knock determination Schutz discussed the sources of  error arising in the use of the contact bouncing pin both from the viewpoint of the theory of the method of measurement and of the mechanical errors arising during measurement. He assumed that knock is a spontaneous ignition arising from a number of points in the residual charge. On the basis that the size of this charge governs the knock intensity, and the energy distribution of the charge the point form which the knock wave starts he suggested the following methods for determining the knock resistance of fuels: (I) Determination of the amplitude of  the knock vibrations. (II) Determination of the locality of the knock centre from the time interval by which quartz indicators differ as to the start of knock. (III) Determination of the combustion time by the use of ionisation gaps, from the assumption that the combustion time ends at the onset of knock.

Lindner, has also discussed other methods of measuring knock resistance, and an electrical indicator, in which a capacitance is altered by change in cylinder pressure, has been described by Meurer. 

To eliminate contact burning, adjustment difficulties, etc., which occur with contact bouncing pins, an electrodynamic bouncing pin was developed. In this the pin carried a coil which, with knocking combustion, moved in a magnetic field. The alternating voltage induced, after being rectified, gave a measure of the knock intensity.

Experiments were made to compare electrodynamic bouncing pin, contact bouncing pin and a pressure acceleration knockmeter. An I.G. engine was used, and the tests were made according to both the Motor and Oppau methods. It was considered that in the Motor method the use of the electrodynamic bouncing pin was unobjectionable. In the Oppau method, however, when the electrodynamic bouncing pin was compared with the pressure acceleration indicator, the latter gave a much flatter knock limit curve. This was particularly evident at high lead and benzol contents, where presumably machine vibrations due to the high compression ratios affected the bouncing pin. It was concluded that the acceleration meter has possibilities as an objective knock measuring instrument.

The D.V.L. Zeiss Ikon acceleration meter has been described by Wende. A quartz indicator of frequency between 35,000 and 50,000 Hz., screwed into the cylinder, was used to pick up oscillations in the cylinder, and the course of the second differential of the pressure was observed on a cathode ray oscillograph. The mean value of the amplitude of this d2p/dt2 impulse was indicated on an ammeter. This second differential of pressure is very sensitive to knock and gives an objective measure of it. 

It was reported that developments were in progress to enable use of the indicator as an external attachment to the cylinder. 

For the purpose of defining the knock limit, the onset of knock was considered of more importance than the knock intensity, and was obtained. 

(i) From the d2p/dt2 vs. motor operational value curve. The kink in the curve indicated the onset of the knock.

(ii) In main engines, where variable charge pressure could not be used, from the d2p/dt2 vs. fuel consumption graph. Here the two bends in the graph indicated the knock region. The theoretical basis of the kink which characterises the beginning of knock was discussed by Lichtenberger. He also reported the kind always occurred at approximately the point where the incidence of knocking, as determined aurally, was 8 to 10 sharp knocks per minute. The accuracy of determination of the onset of knock was within ±20 mm Hg. boost pressure even in the most unfavourable cases, while in most of the measurements it was within ± 5 mm Hg.

Utilizing the D.V.L. quartz indicator, a mechanical counter was developed which recorded the number of seperate knocks per minute. 

In the later piece-electric instruments, accurate measurements could be obtained up to 100,000 cycles/second. Another advance in knock measuring was the development of an electric-acoustic method for knock determination. This was described by Kneule. The effect of the gas vibrations on a pick up, placed in any suitable position on the cylinder block, is measured. The basic engine sound can be damped out and the original signals are taken through amplifiers, sensitive to the knock frequency, and are observed either on a cathode ray oscillograph or a sensitive galvanometer. The main advantages claimed for this system are: (i) Ease of attachment of pick up. (ii) Recording of all intensities from light to strong knock. (iii) Applicability to mulitcylinder engines. (iv) Applicability to engines in actual operation. 

It has the disadvantage, however, that the onset of knock may be detected at various compressions on degrees of supercharge, and not at any absolute fixed point, according to the sensitivity of the amplifier. 

Agreement between electro-acoustic measurements and road tests was reported to be fairly satisfactory. 

It should be noted that in spite of the development of the  pressure acceleration and electro-acoustic methods of knock determination, the observation of the onset of the knock in the D.V.L. Supercharge Test continued to be made aurally.

(e) Viewpoints on possible future methods of fuel testing. 

Some thought was given to the possibility of characterising fuels from their ignition properties. On this basis it would appear possible to use the same characterisation for both spark ignition and diesel engines. Considering the applicability of ignition delay results for the fundamental characterisation of the self ignition process, F.A.F. Schmidt concluded that the regularity of the self ignition process could be reduced to three values: (i) The velocity of the reaction process under a fixed standard state, depending on the nature of the fuel. (ii) The temperature dependence of the reaction process. (iii) The pressure dependence of the reaction process.

These characteristics do not remain constant in different pressure and temperature ranges. 
Jost pointed out that to form a simple classification of fuels in the test engine, the engine properties have to be eliminated, since knock is a combined quality of fuel and engine. Thus a physical apparatus might just as well be used for fuel testing. He considered that in the future it might be possible to state fuel properties by definition with physical characteristics, and to gauge engines by using in them two fuels of differing characteristics. A similar method was envisaged by Rögener, for determination of the decisive factors regarding knocking, such as final temperature, density and time for reaction. In his view, however, instead of a  physical test, all fuels would be raised, for all engine types, from one engine test. Assuming that knock depends on temperature, T, density and time ,then the theoretically true expression for the knock limit of fuel 1 would be. Calculating from engine results the expression would be. i.e. different values in different engines corresponding to the same true variables.

If three, suitably chosen fuels had true knock limit surfaces (on a T-P-Y diagram) intersecting at point P, the knock limit surfaces obtained from engine results would intersect at P1 for engine 1 and P2 for engine 2. Repeating in different engines the equivalent points (i.e. intersection points) could be plotted as co-ordinates in a three dimensional system. Hence, with a new fuel, by plotting the knock limit surface in one engine, those of other engines could be calculated; since the positions of the surfaces are the same relative to the equivalence points of the other engines as the surface in engine 1 to the equivalent point of 1. 

Phillippovich was doubtful if a simple laboratory method would be able to replace knock measurement in engines, since engines experiments had shown that knock in engines is not simply the result of a heat explosion but also of cold reactions with a negative coefficient of temperature in certain regions. 

D.) Power Boost

Amongst the fuels used for temporary increase in output for such special purposes as starting and air combat were methanol and ethanol, leaded or unleaded. A number of experiments were made which confirmed the advantages claimed for secondary injection of methanol water mixtures and pure water. In particular, water-methanol mixtures with more than 50% methanol were found suitable for a double fuel operation. 

Over a period of four years nitrous oxide was extensively used as a means of temporarily boosting engine power at high altitudes. At first it was used liquefied under pressure at atmospheric temperature. This method gave somewhat erratic running since pressure altered with the external air temperature and with vapourisation. Later super-cooled, pressureless nitrous oxide was used, which allowed larger quantities to be carried and effected an improvement in the specific performance due to increased internal cooling. In spite of the wide use of nitrous oxide for boosting power, C.I.O.S. Report No. XXXII-44 mentions the tendency to look upon it solely as a makeshift for a good supercharger and internal cooling.

The basic problems involved in the use of oxygen carriers such as nitrous oxide, were considered by Lutz. His calculations were based on the fact that whilst boosting the power, the alteration of indicated power results from the change in the charge weight due to the injection of the carrier, and the energy content of the latter, according to the equation.

The "air value" is a complex quantity depending, amongst other things, upon the extend of vapourisation of the oxygen carrier.  The other factor is equation, namely the charge weight change, depends on : (a) The influence of the blower due to the change in boost pressure. (b) The change in boost air temperature because of the mixture effect. 

It was shown that as regards these influences both the latent heat of vapourisation and the extent of vapourisation of the oxygen carrier are very important. Thus the absolute oxygen content of the carrier is not the only factor determining its suitability. 

From its effect on the weight of air charge by reason of its great heat of vapourisation and its "air value", hydrogen peroxide would, if completely vapourised, even outstrip oxygen in the increase given in indicated power. Complete vapourisation, however, would be impossible since cooling of the boost air would be too great for the process to be controllable, and at the low temperatures involved hydrogen peroxide would become solid. The use of hydrogen peroxide as a carrier is thus excluded, and knocking and corrosion factors etc., limit the only practicable oxygen carriers to nitrous oxide and oxygen. Both liquid nitrous oxide and liquid oxygen are readily vapourised. Liquid nitrous oxide has the advantage that because of its high latent heat of vapourisation there is a higher absolute rise of output for the same thermal stress on the engine than is possible with oxygen. The specific output gain of liquid nitrous oxide, however, is only 60-75% of that which is available from liquid oxygen.

Lutz, therefore, defined two ranges of application of these oxygen carriers. He suggested that for large gains in output over short periods nitrous oxide should be used, whilst for smaller gains in performance over longer periods, oxygen is best.

The high thermal loads encountered when operating with nitrous oxide, as compared with normal operation, could be countered by increasing the amount of fuel injected. The same effect could be obtained by using a nitrous oxide alcohol combination, with or without water spraying. This also allowed a greater increase in output. Alcohol or water-alcohol mixtures require an ignition advance, whereas the quick pressure rise, when nitrous oxide is used, makes a retarded ignition desirable. It was expected, however, that with simultaneous injection of nitrous oxide and alcohol no special ignition timing would be necessary.

The importance of place of injection of nitrous oxide on the possible increase in output was investigated by F.K.F.S. As was expected, injection in front of the blower resulted in a higher output yield than injection behind the blower.

Discussing the conclusions of Luts, F.A.F. Schmidt referred to experiments showing that the actual temperature difference when working with liquid oxygen as compared with liquid nitrous oxide was not so great as theoretically calculated. Further, he envisaged the admixture of methanol as a coolent in order to achieve extremely high gains in performance with liquid oxygen.

Difficulties in the use of liquid oxygen because of its tendency to give rise to vapour look and also to become surrounded with a non conducting layer of gas which prevents evaporation, were said to have been overcome by the insertion of a magnetically controlled valve in the supply line. This valve automatically vented any gas to the air.

The use of methyl alcohol as a fuel or as an internal coolant was sometimes complicated by the appearance of glow ignition phenomena.

Experiments made did not result in an additive being found which would increase the ignition temperature and stop the catalytic reaction. An increase in the ignition temperature was attained, however, by a high percentage admixture of the hydroformed, hydrogenation naptha, DHD-gasoline.

E.) Fuels and their Properties.

The use of hydrogen as a special fuel was considered both from theoretical aspects and from the results obtained using a single cylinder, external ignition side valve, stationary engine. The engine was operated under pressure charge because of the high output attainable by this method. Even with high compression end pressures the combustion in the engine tended to be irregular, especially with very lean mixtures. Except for greater combustion speeds, however, the behaviour of the hydrogen motor did not deviate from that of other gas motors. In contrast to results obtained in bomb tests, flame speeds in the hydrogen motor were not greatly in excess of those liquid fuels. Ignition limits were found to be similar to those observed in bomb tests. Self ignition was never observed, and the combustion, over a wide range of mixture strength was found was found insensitive to ignition adjustment. Output was limited by knock, and the highest output was obtained with mixtures 20% richer than stoichiometric. Comparing the results with other engine methods, the output of the hydrogen motor was seen to be in the range of the diesel engine. Where container weight is not importance, as in locomotive traction or stationary engines, it was considered that hydrogen would be a suitable fuel. In respect to thermal properties and economy, hydrogen could be rated equally with ordinary fuels. 

The problems involved in knock testing of gaseous motor fuels were discussed by Ruess. He gave values for the octane numbers of various gases as shown in Table 1 of the appendix.

As regards liquid fuels, the knock behaviour of a large number of alkyl benzols was reported by the I.G. Farben, Ludwigshafen. In Tables 2A and 2B of the appendix are presented the motor octane numbers of these compounds (diluted with Eichbenzin I.G. 10 in the proportion 50: 50 Vol % and with the addition of 0.15% Vol % T.E.L.) and their superchargeability, as determined in the BMW 132 cylinder by the D.V.L simplified procedure. These results were summarised as follows: 

1)Of the alkyl benzols treated with a sulphuric acid catalyst the iso compounds show a better knock behaviour than the normal compounds, but the contrary is true in treatment with an aluminum chloride. catalyst.

2) In the series mono, di, tri, and tetraethyl benzol the diethyl benzol shows the most favourable knock behaviour. In the propyl substitution the highest knock resistance is shown by tripropyl benzol i.e. superchargeability increases from mono over di to tripropyl benzol, but is much smaller with tetrapropyl benzol.

3) With the exception of ethyl and propyl benzol, the knock behaviour in alkyl benzols deteriorates with increasing carbon atom number and boiling temperature.

It should be emphasised that these results do not refer to pure compounds, and should, therefore, be considered with caution. The above results do not appear to be in accord with the statement by Phillippovich that in the cases examined, supercharging could be increased as the number and length of side-chains of the benzene nucleus increased; and that of two highly aromatic fuels, the one containing aromatic compounds of higher boiling point could be used for greater supercharging. 

Data on hydrocarbons and other compounds used as fuels and fuel components were given in another report from I.G. This data was presented in the form of tables according to substance clases and not chemical structure. From these tables and other sources were obtained the octane numbers of the large number of compounds given in Table 3 of the appendix.

Three grades of aviation gasoline were used by the Germans; for training purposes, for normal operational use, and for aircraft having high duty engines. The specifications for these fuels, as given in BIOS Report No.119, are reproduced in Tables 5 and 6 of the appendix. As with all the other fuel specifications given, these show a relaxation in some of the requirements compared with the specifications laid down earlier in the war. It should be noted, however, that the octane number requirements were never relaxed.

For completeness, the specification for gas turbine fuel J-2 is also reproduced-Table 10. It has been reported that kerosine complying with this specification was only accepted as a gas turbine fuel because of the shortage of gasoline. 

In Table 7 is shown the specification for motor gasoline used in military vehicles. 

3. Combustion in Diesel Engines

The diesel engine was used for general automotive and marine (mainly submarine) propulsion, and only to a small extent for aviation purposes. Work on diesel fuels was directed mainly to an improvement of their starting qualities, and to the development of alternative fuels as supply of the normal fuels diminished.

A) General Theory and Practice

The efficiency of an engine in relation to the ideal engine was discussed by List. Idealised curves for the combustion laws needed for the different types of diesel engine cycle were given, taking into account variation in compression ration, limiting maximum pressure and limiting maximum pressure rise. For a constant maximum pressure, the mean pressure of the cycle was shown to depend strongly on the combustion law. The considerations showed that for greater efficiency a better control of combustion must be attained, so as to give diagrams with a small pressure rise in relation to the compression and pressure. Bearings must also be improved to permit the use of higher maximum cylinder pressures.

Tests were made in a single cylinder engine using direct injection, prechamber, turbulence chamber and air storage chamber. The combustion law for each process was determined by accurate engine indication and careful fuel and air consumption measurements. It was shown that high initial combustion speeds which decreased during the course of combustion were not desirable; also, further knowledge of the processes occurring between injection and ignition were required to allow for adequate control during this part of the cycle-combustion after ignition already being well controllable.

Combustion chamber design, in contrast to that in Britain, tended towards the use of air storage chambers. Several engine and bomb investigations of the combustion process have been made using this type of chamber. 

Peterson, using a double air storage chamber fitted to a bomb, studied flame movement (with air movement) by high speed photography. He found that slightly late injection of fuel, equivalent to half load, resulted in the greater part of the fuel remaining in the main chamber near the first throttle point, although a small proportion reached the first storage chamber where ignition occurred. The flame raced along the fuel layers before the first throttle point, throwing it in the direction of the injection nozzle. From there, combustion spread laterally and filled the main chamber. It was found that there were two main courses of combustion. When injection took place after the maximum inflowing air speed, the second storage chamber took little part in the combustion, since the bulk of the fuel remained in the main chamber. When the fuel was injected earlier, however, the strong air stream carried it into the second storage chamber, and then the main chamber played only a small part in the combustion since little fuel was brought back into it. In this latter case soot formation occurred in the second storage chamber. Between these two injection extremes, there was a favourable range, dependent upon the fuel quality. It was found that the injection nozzle needed to be no wider than that necessary to bring the fuel into contact with an adequate supply of air. Increase of the angle of spray caused a drop in combustion speed, because the effect of each burning fuel droplet on its surrounding droplets became less than greater the distance between them.

In engine experiments, the combustion course was followed by the use of ionisation methods to determine flame travel, and by pressure pick ups. The latter were usually of the piezo-electric type, which the Germans favoured rather than the capacity type.

Tests were made by Dreyhaupt on an engine with a Lanova air-cell combustion chamber. With this engine at full load, ignition occurred only after the complete fuel charge was injected, and a smooth pressure rise occurred in the main chamber, thus eliminating "diesel knock". The fuel spray angle was fairly critical on this engine, and alteration from 6º to either 4º or 8º was detrimental to the combustion.

Experiments were also made on a Deutz prechamber engine, gas samples being taken from the points at which pressure and flame ionisation measurements were made. Although there should have been ample air available in the prechamber, combustion was actually found to occur under a great air deficiency. This led to soot formation into the prechamber at the commencement of combustion. The flame spread very quickly into the main chamber because of the high pressure developed in the prechamber. This engine was very insensitive to the ignition quality of the fuel and gave an almost constant delay period under all conditions.

B) Modifications in Engine Design and Operation.

An important development was the introduction of the recycle diesel engine for underwater submarine operations. Its ultimate aim was to enable the engine to be operated either entirely on oxygen supplied from high pressure cylinders, or under "Schnorkel" conditions in which a limited supply of air was drawn through a pipe from the ocean surface. A twenty cylinder Krupp Germanieawerft engine was run on this principle. Its supercharges were removed to reduce excessive exhaust temperatures, and an inlet temperature of approximately 100º C was used. Exhaust back pressures reached 20 p.s.i., and intake suction 1.4 p.s.i.

In German aero engines, considerable difficulty was experienced by failure of the spark ignition system at high altitudes. Trouble with the ignition also arose because of excessive plug fouling with high octane, lead containing fuels. An attempt was made to overcome these difficulties by the use of a low tension ignition system operating at about four hundred volts. Later, however, investigations centred on the use of the "Ring" process. This process had the additional advantage that it made feasible the use of high boiling point safety fuels. The work on the "Ring" process had been summarised by O'Farrel in BIOS Report No.1609, which also contains an extensive bibliography on the subject.

In the "Ring" process, ignition was produced by spraying a liquid into the combustion chamber, at the appropriate moment in the compression stroke. This liquid spontaneously ignited at the cylinder temperature, thus igniting the main fuel charge. Diglycol diethyl ether and butandiol diethyl ether were found particularly suitable as ignition fuels. The former, which had a cetane number of 188, was manufactured in quantity and was also used as a diesel starting fuel. The mechanism of the self ignition of this fuel was thought to be a rapid disintegration of the molecule under the action of heat. This reaction, being exothermic, produced a rapid temperature rise causing the products of the decomposition to ignite.

Operation of the "Ring" process could be made at a compression ratio of 7 : 1, but in practice a ratio of 8:1 was normally used. For weak mixture operation the optimum ignition fuel (R-fuel) quantity increased slightly, whilst with overrich mixtures the performance deteriorated because of longer ignition delays. In practice, the timing of the R-fluid injection was not too critical, and was usually kept constant.

Cylinder head temperatures were considerably lower with R-fluid injection than with spark ignition, although no appreciable difference in exhaust temperature was observed. If the operating temperature was too low, ignition difficulties arose. In this respect a great improvement was obtained when the R-fluid was injected on to the hot exhaust valve. This increased fuel consumption, which, however, was still an improvement on that obtained using spark ignition.

Compared with the spark ignition process, the maximum power output, when using R-fuel, was approximately the same at rich mixtures, but was considerably improved at weak mixtures. Although at high compression rations knock was considerably reduced because of the multipoint ignition of the mixture by R-fluid, at the compression ratios normally used there was little improvement over the usual method of engine operation. 

With the "Ring" process, starting was difficult, and sometimes necessitated an auxiliary spark ignition system. The other disadvantages were the cost of the injection equipment, even when using "pumpless" injection of the R-fluid, and the necessity for careful coolant control. Preliminary experiments were made to investigate the possibility of self ignition operation of mixture compression engines. Such a process would reduce the outlay on accessories in high speed multi-cylinder engines of small capacity, and might make possible the utilization of low boiling, low anti-knock fuels. 

The experiments were made in a single cylinder, air cooled, spark ignition engine of 200 cc. capacity and in 700 cc. water-cooled, single cylinder diesel engine. A fuel blend of 60% of low boiling, primary gasoline of octane number 50, and 40% of RCH diesel fuel of octane number -80 was found to give the most favourable conditions for self ignition operation. At first, however, knock-free operation was not attained. 

The knock could not be suppressed by additives, without adversely affecting the ignitability. Attempts to suppress the knock by altering the combustion chamber design were also ineffective. It was found, however, that for very rich and very weak mixtures (air: fuel weight ratios of 8-10 and 25-28 respectively) knock free self ignition could be obtained. This phenomenon was attributed to a reduction in the combustion velocity, Indicator diagrams showed that the pressure rise occurred fairly smoothly, a few degrees after T.D.C. The flame in the knock-free region was a faint blue in contrast to the bright white flame appearing with knock.

Starting from cold required the use of additives, preheating or a glow plug. Power and consumption in the knock-free lean region corresponded approximately to those of the automotive diesel engine.

It was concluded, however, that for practical self ignition operation, the lean mixture knock free-region had first to be extended. Also the development of a precise instrument for control of the air : fuel ratio was essential.

Other modifications made have mainly been in details. Thus M.A.N., on their two-stroke double acting engine injected lubricating oil through a nozzle on to the cylinder walls. This was all consumed, and it was claimed that both wear and fuel consumption were greatly reduced. Junkers, in their opposed piston aero diesel engine, had pistons made in three sections, the one being a steel crown that ran at 1300ºF. Four through bolts, with heavy springs to take up expansion, held it together. Klöckner- Humboldt-Deuts were developing a 16 cylinder diesel aero engine. In this, open combustion chambers were used, with six or eight hole fuel injection nozzles. The fuel injection time at 2000 r.p.m. was shortened from 50º to 30º. To prevent ring sticking the top piston ring, (a composite type) had its upper half made in two separate segments.

C) Evaluation of Diesel Fuels

In Germany, the four usual methods of measuring ignition delay were :

(a) Pressure measurement, which could be made anywhere in the combustion chamber provided it was of uniform shape.

(b) Ionisation measurement, which only referred to one local spot.

(c) Heat emission from the combustion chamber.

(d) High speed photography.

The latter two depended on the shape and position of the window, and their sensitivity was affected by soot deposits. 

An instrument was developed for the measurement of ignition delay, which utilised an inertia indicator to determine the start of ignition. A contact fitted on the nozzle holder, and operated by an elongated nozzle pin, closed a circuit when fuel injection commenced. When ignition occurred the pressure rise was so rapid that the rate of change of deflection of a diaphragm in the wall of the combustion chamber caused the contact in the interior of a light metal tube to be opened. This broke the circuit. A photo-cell could be used in place of the inertia indicator if so desired.

An attempt was made to develop a universal engine for both cetane number and octane number determination. An F.K.F.S engine with fuel injection was used. The compression ratio was kept constant and the delay period before self ignition was determined using a photo-electric cell. It was claimed that ratings obtained by this method agreed very well for the entire cetane range, and even up to approximately 50 units on the octane number scale. Above this figure the method could not be used successfully. Using mixtures of reference benzene and gasoline, Ernst found that the ignition delay octane numbers to be approximately 1-2 units higher than those obtained in an I.G. engine under C.F.R Research Method Conditions.

In 1942 the special Committee of the D.V.M for the Standardisation of Engine tests on Diesel Fuels recommended:

(i) Ignition delay process should be standardised as the basis for cetane number determination.

(ii) Ignition delay should be constant throughout tests. 

(iii) Engines used should be the H.W.A. and I.G. Test Diesel.

(iv) The ignition delay should be kept constant either by throttling or altering the compression.

(v) The measuring instruments used should be the inertia indicator (Rhenania-Ossag), the piezo-quartz of the F.K.F.S. photo-cell indicator.

The most common test procedure used the I.G. Test engine. The method was similar to the C.F.R. diesel method, except that an ignition delay period of 18º was used. Another method in which the delay period was kept constant, was that of the H.W.A. In this method the inlet air was throttled back to the limiting ignition and the resulting suction in the inlet tube was taken as a measure of the ignitability. It was claimed that this throttling method corresponded better with cold starting behaviour than did the other test methods.

In contrast to these methods which employed a constant ignition delay, the D.V.L., with its test-bench diesel, measured the duration of the ignition delay whilst keeping the compression ratio constant.

A correlation between the cetane number of a fuel and its specific gravity, aniline point, bromine number and pour point was drawn up in graphical form by the H.W.A.

D) Ignition Accelerators

Heinze, Marder and Veidt determined the effect of the addition of a large number of compounds to various brown coal tar and antracite tar distillates. The compounds tested included: ethyl nitrate, amyl nitrate, amyl nitrite, ethylene chlorohydrin nitrate, tetralin peroxide, benzoyl peroxide, nitroso n-methyl urethane, diethyl tetrasulphide, diethyldioxime, copper stearate, butyl bromide, p-nitroso dimethyl aniline, chlorodinitrobenzene, trinitrotoluol, benzoyl acetone, cyclohexanone oxime, dipentene and terpin. The last seven compounds had no effect on the ignition behaviour, whilst of the others, the nitrates and nitrite were the most effective, the rise in cetane number of the fuels being approximately proportional to their concentration. With nitroso n-methyl urethane, ethylene chlorohydrin nitrate and diethyltertrasulphide, however, the improvement per cent of additive decreased as the additive concentration increased.

The effect of the additives on a number of other fuels was also determined. Except for tetralin peroxide, an addition of less than 0.5% of additive always brought the Conradson Carbon value of the fuels well in excess of the British and U.S.A specification of 0.2%. All the additives tested increased the corrosion effect on zino, but only with tetral peroxide and amyl nitrate was the corrosion considered excessive. All the additives were completely stable in storage even after one year.

E) Cold Starting and the Use of Additives

Whereas in Britain and the U.S.A. additives have mainly been used to raise the cetane number of a fuel for normal running, in Germany the emphasis in research has been centered on their use for improving cold starting behaviour.

The effect on the starting behaviour of diesel fuels by the injection of special starting fuels into the air intake manifold of the engine was studied by Leib. He used mainly ethers and similar compounds for these starting fuels.

The minimum compression ratios required  for the ignition of these fuels, without injection of the main fuel, were first determined. An I.G. engine was used in a room which could be cooled to -25ºC. The lower the room temperature the higher was the compression ration necessary for ignition. In general, decrease in vapour pressure of these starting fuels resulted in a higher compression ratio being required for ignition. An exception was ethyl iso-propyl ether, which, despite its lower vapour pressure, (and also lower octane number), required considerably less compression than di-ethyl ether. No unexpected effect resulted from a mixture of starting fuels. With the two lowest boiling fuels tested-diethyl ether and ethyl isopropyl ether-the engine would only start within a certain compression ratio range. Above an upper compression limit, spontaneous combustion in the cylinder resulted in all the fuel being burnt before the end of the compression stroke. With the higher boiling starting fuels this phenomenon was only observed at fairly high temperatures and compression ratios of nineteen to twenty.

Tests in which diesel fuel was injected into the cylinder and starting fuel into the induction line, showed that for easy starting fuels with high cetane number and high vapour pressure are needed. As with ethyl iso-propyl ether, however, the inner structure of the fuel is also a factor. For practical use with induction line injection the following order of starting fuels was given; ethyl iso-propyl ether, diethyl ether, di-iso-propyl ether and di n-propyl ether.

A series of investigations concerned with the improvement of starting behaviour was made at the Technische Hochschule, Stuttgart. All the tests were made on an F.K.F.S engine. Additives were either admixed with the fuel, introduced into the intake air, or directly injected into the cylinder separately from the main fuel. 

Denmer judged starting behaviour by motoring the engine under fixed conditions and determining the minimum compression ratio needed to keep the engine running by itself. Direct addition of ethyl nitrate to a commercial diesel fuel gave an improvement of 1/2 compression ratio. The improvement with Fischer Tropsch RCH oil was much less, and at high additive concentration the performance even deteriorated. The cetane number for both fuels increased with the concentration of ethyl nitrate. Admixture of carbon disulphide, although slightly improving the starting behaviour of commercial oil, had an adverse effect on the RCH diesel fuel. Carbon disulphide vapour, however, effectively improved the starting behaviour. The cetane number of the commercial diesel fuel decreased with carbon disulphide addition, but that of the RCH fuel increased slightly. Acetaldehyde had little effect when added directly to the fuels, but vapour addition improved the critical compression ratios by 2 or 3. The cetane number dropped appreciably on addition of acetaldehyde to the commercial fuel, but the RCH fuel showed the reverse effect. It was concluded that the raising of the cetane number of a fuel by the use of additives is not indicative of an improvement in the starting behaviour of the fuel. This agreed with the experience that the cetane number measured in an engine, even by the H.W.A. method, is no criterion of the starting behaviour of a fuel, at least at temperatures below 0ºC. In this connection it was suggested that possibly a useful characteristic of starting behaviour might be obtained by considering cetane number, boiling range and viscosity.

Similar experiments were made by Schutze but he added fuel admixed with additive to the intake air as well as simultaneously injecting it in the normal way. When using the highly paraffinic RCH fuel the engine could run for a time under its own power without any flames appearing in the cylinder. This indicated strong preflame reactions. The gradual heating of the engine eventually resulted in flames appearing, accompanied by considerable knock and an acceleration of the engine. Napthenes and aromatics did not show this phenomenon. Experiments with varying inlet air temperatures showed that the critical compression ratio when using RCH fuel containing 5% ethyl nitrate rose sharply as the temperature fell below 50ºC.This was attributed to the ethyl nitrate being sucked into the cylinder in the liquid state, the drops of ethyl nitrate being surrounded by a cool zone.

With normal quality fuels the effect of the heat lost by evaporation of the droplets would be far outweighed by the increased ignitability caused by the ethyl nitrate. With the highly ignitable RCH oil, however, ignition would be hindered. From engine tests the ignition temperatures of cyclohexane and cyclohexane containing 5% of ethyl nitrate were calculated using the equation. These calculated values were compared with tests made in the Jentsch ignition tester. It was concluded that provided fuel does not enter the engine cylinder in the liquid state the ignition point gives a suitable guide to starting behaviour.

From engine experiments in which the main fuel was not injected, Staats found, that ether added to the intake air from a carburettor required a smaller compression ratio for ignition than did ethyl nitrate in the vapour phase. Ether, when added in the vapour phase, however, required an even lower compression ratio.

In actual engine operation, the starting behaviour, when ethyl nitrate was used in the vapour phase as an ignition accelerator, was not dependent on the ignitability of the main fuel. In contrast, the effect of the different instabilities of the individual fuels was slightly noticeable when using ether vapour. It was found that there was an optimum ether vapour quantity; introduction of too much ether vapour so cooled the combustion chamber that , presumably, the prereactions between the ether and the air were retarded.

Cold room tests at temperature down to -40ºC showed that at such low temperatures ether was the only substance that would assist starting. Ethyl nitrate was only effective for temperatures down to 0ºC.

A somewhat different attempt to improve the starting behaviour of an engine was made by Gerschler. During starting he passed the products from the exhaust port back into the intake port. A very slight improvement resulted, but constructional difficulties prevented the practical use of the system. The work, however, furthered the development of the recycle engine.

F) Fuels and their Properties

The diesel fuels used in Germany were of the normal petroleum type. The Ruhr-Chemie, A.G. also produced a Fischer-Tropsch fuel which was usually used for blending purposes. They also produced a reference diesel oil (RCH reference), with a cetane number greater than 70. Increasing shortage of supplies, however, brought about the use of some alternative fuels. 

At first, an attempt was made to operate diesel engines with liquefied gas, such as a propane-butane mixture. Bubble formation in the fuel lines was a problem, but it was found that this could be avoided by maintaining the entire fuel system at a pressure of twelve atmospheres. The chief difficulty inherent in the use of liquid gas in the engine was its great compressibility compared with normal gas oil. A much larger injection nozzle was required, and the pump timing had to be advanced about 20º. Under such conditions the ignition delay was about 14º-18º. A larger injection pump was necessary because, in spite of the pressurized fuel system, the pump sum never completely  filled with liquid. The size of the anti-dribble valve in the outlet port of the pump was also increased to prevent lengthening of the injection period. Other difficulties arose in the form of leakages in the fuel system, and lack of lubrication of the injection nozzle value and injection pump plunger.

Further developments in the use of liquid gas was stopped by the introduction of the more conventional, wood fired gas generators. It was claimed that when using the gas from such generators, petrol engines developed 70-75% of their normal power output, and diesel engines about 80%. Compression ratios of 9:1 to 11:1 were used.

As early as 1940 information was given on running automotive diesel engines on gasoline and other fuels, without alteration to the engine, in the event of a temporary failure in the supplies of normal diesel fuel. Benzene, alcohols and other high octane fuels were unsuitable, but all commercial petrols, even if they contained lead or alcohol, could be used directly in the engine. In an emergency, middle distillates, lamp oil, etc., could also be used. When using gasoline that was not blended with at least 25% of gas oil, the addition of 5-10% of engine lubricating oil was advised in order to lubricate the fuel pump and nozzle. The use of gasoline in warm weather also necessitated cooling of the injection pump to prevent vapour formation. The low density and, in general, poorer ignitability of the alternative light fuels caused the engine output to be lowered by 10-15%. Starting was also difficult, and higher idling speeds had to be employed. 

A comprehensive investigation was made by Penzig on the use of nitroparaffins as fuels. In his report he tabulated the properties, and presented other data in respect of nitromethane, nitro-ethane, nitropropane, nitrobutane, and nitrobenzene. Approximately 25% more energy was obtained from a given amount of air when using nitropropane than when using n-heptane. When burning heptane in air to which nitrous oxide had been added in such proportion as to give 3 mols N2O/mol C7H16, however, almost the same energy was obtained, as with nitropropane and normal air. Thus when ordinary air was brought up to the same oxygen content by the addition of free oxygen, the energy liberation, although better than with normal air, was not so great as with the nitrous oxide containing air. 

In a reference to the work on glow ignition, it was stressed that nitroparaffins are readily ignited at hot surfaces, even though they are difficult to spark ignite. Glow ignition temperatures were obtained by passing the fuel vapours over an electrically heated coil. It was suggested that the nitroparraffins might be used as fuels for high altitude flying, and thus eliminate the necessity for the conventional spark ignition system. 

The glow ignition properties of the nitro parraffins were also shown by their cetane ratings. With direct injection the technical product S 3 ( a mixture of 1 and 2 nitropropane) had a cetane number of only fifteen. In a prechamber engine where the fuel contacted the chamber walls to a greater extent, it had a cetane number of forty five. 

The admixture of nitropropane ( in the form of S 3), with a number of gasolines gave surprising results. The fuel S 3 had an octane number of 72, but when mixed with certain gasolines it gave rise to mixtures with octane numbers lower than those of either component. The minima occurred with mixture containing approximately 30% of S 3. With other gasolines no such phenomenon was observed. Nitro-ethane and nitrobutane behaved similarly to nitropropane. The octane number of nitrobenzene could not be determined because of glow ignition, but experiments were made in which it was mixed with gasoline VT 702. Up to 50% nitrobenzene addition the octane number of the fuel was only slightly changed, but further increase in the nitrobenzene content, however, resulted in a considerable improvement in the octane rating.

Experiments were made using a number of diesel fuels. Corresponding to their effect on the octane numbers of gasoline, the nitroparaffins gave mixtures with cetane numbers greater than those of either component. The maxima, however,  were not so pronounced as the minima in the octane number curves, although the improvement was greater the higher the cetane number of the original fuel.

On recycle diesel operation, use of nitropropane as a fuel was estimated to be able to give a saving on the total weight of fuel and storage equipment of 16%. Its use in submarine operation would, however, be limited because of the bubbles that would arise from the evolved nitrogen.

Table 4 of the appendix is a compilation from the literature of cetane ratings of various compounds.

Finally Tables 8 and 9 show the general German specifications for diesel fuels used in aero, automotive, and marine work.

Where necessary the symbols, as given in the original documents, have been changed to ensure uniformity.
a Constant
a' Constant
A Constant
b Constant
c Concentration of chain carriers
c10 Original concentration of initial material
C1 Concentration of initial material
C Constant
D Tube Diameter
e Base of natural logarithms
E Energy of activation
Es Apparent energy of activation
G Free energy
k Velocity constant
k1 Velocity constant for reaction of chain carrier with initial material
k Velocity Constant of chain breaking reaction
k Velocity constant of chain breaking reaction
K Constant
K2 Constant
m Adiabatic exponent
M Molecular weight
n Constant
N10 Initial indicated power
N1 Indicated Power
P Pressure
Po Minimum starting pressure
Pme Mean effective pressure
R Gas Constant
t Time
T Absolute temperature
Te Absolute ignition temperature
Ta Absolute inlet temperature
U Heat transfer coefficient
v Fresh gas velocity
V Volume
w Reaction velocity
wo velocity of primary chain formation
WLO Initial air charge weight
WL Air charge weight
Wx Charge weight of oxygen carrier
x Proportion of initial material transformed
z Ignition velocity
Zj Sum of the internal degrees of freedom of a monatomic gas