NATIONAL ADVISORY COMMITTEE FOR AERONAUTICS

TECHNICAL M]_40RANDUM 1261

O
ON THE THEORY OF THE PROPAGATION OF DETONATION IN

GASEOUS SYST_4S*

By Y. B. Zeldovlch

SUMMARY

The existing theory of detonation is critically examined. It

is shown that the considerations with which the steady value of the
velocity of detonation is chosen are not convincing. In connect-
ion with the problem of the process of the chemical reaction in a
detonation wave, the objections raised against the conceptions of
Le Chatelier and Vieille of the 19th century with regard to the
ignition of the gas by the shock wave are refuted. On the basis
of this concept, it is possible to give a rigorous foundation
for the existing method of computing the detonation velocity.
The distributions of the temperature, the pressure_ and the velocity
in the detonation wave front as the chemical reaction proceeds,
are considered. On the assumption of the absence of losses, the
pure compression of the gas in the shock wave at the start of
the chemical reaction develops a temperature that is near the
temperature of combustion of the given mixture at constant
pressure. The specific volume and the pressure are connected
by a straight line that passes through the point corresponding
to the initial state of the gas (straight line of Todes), which
is given by the following equation:

D2
P - P0 _ _u (v_v0)
v02

The effect of the hydrodynamic resistance and the heat losses

is considered. The losses during the chemical reaction decrease
the velocity of propagation of the detonation, thus leading to a
lowering of the temperature in the shock wave, the extinguishing

*"K Teorii Rasprostranenia Detonatsii v Gasoobraznykh Sistemakh."

Zhurnal Experimentalnoi i Teoreticheskoi Fiziki, T. I0, 1940,
pp. 542-568.
2 NACATM1261

of the gas, a drop in the velocity of the chemical reaction, and a

further increase in the losses. On the basis of these consider-
tO
ations, the existence of a limit of propagation of the detonation 0

is established: the maximumlowering of the detonation velocity

comparedwith the theoretically computedvalue is not large for
reactions having a rate that increases with temperature. At the
limit, the chem!calreactlon is extinguished at a distance equal
to several tube diameters. The theory developed permits, for a
chemical reaction with known kinetics, computing beforehand the
velocity of propagation of the detonation under real conditions
with account taken of the losses, the limits of steady propa-
gation of the detonation, and the distribution of pressure,
temperature, mass velocity, density, and concentration in a
steadily propagated detonation wave. The results of the theory
referring to the structure of the deto_t_on wave front and the
process of cooling and slowing downof the products are compared
with experimental data.

The practical application of the developed one-dimensional

theory to the computation of the effect of losses on the velo-
city of the detonation is limited by the consideration that even
at the limit the time of the reaction is small and the heat trans-
fer and resistance do not occupy the entire cross section of the
pipe. Moreover, in a very large number of cases, even long before
the attainment of the limit, there is observed the so-called spin,
a spiral or periodic propagation of the detonation not described
by the present theory. Someconsiderations are advancedwith
regard to nondimenslonal criterions on which the spin depends.
1. CLASSICAL
THEORY
OFVELOCITYOFPROPAGATION
OFDETONATION

According to the classical theory of propagation of a deto-

nation wave of Chapman(reference 1), Schuster (reference 2),
Jouguet (reference 3), and Crussard (reference 4) constructed by
analogy with the theory of shock waves of Riemann(reference 5),
Hugoniot (reference 6), Raylelgh (reference 7), and Ranklne (refer-
ence 8) on the assumption of the absence of any dissipative forces
(heat transfer or momentum transfer to the outside, effect of vis-
cosity and heat conductivity in the direction of propagation), the
conservation equations may be written in the following form:

o(D-u) = constant = AI (la)

p + 0(D-u)2 = constant = A2 (lb)

o(D-u)[E + _I+ p(D-u) = constant =A 3 (Ic)

P

NACA TM 1261 5

The above equations are the laws of the conservation of mass,

momentum, and energy, respectively, for a steadily propagated wave
with velocity D, that is, for the case where all the magnitudes p,
p, u, and E depend on the time and the coordinates only in the
O combination x - Dt; for example,
o)

p = p(x,t) _ p(x-Dt) (2)

where p is the pressure ((g cm/sec2)_cm2)_ 0 is the density

(g/cm3), E is the unit energy ((g cmZ/secZ)/g), and u is the
velocity of motion of the gas (cm/sec); E may include the chemi-
cal energy. All these magnitudes may vary, as the detonation wave
progresses, from point to point, whereas the magnitude D (the
velocity of propagation of the detonation wave in cm/sec) for a
given regime remains constant. The constants AI, A2, and A3
in equations (la), (ib), and (lc) have the simple meanings of the
flow of mass, the sum of the flow of the momenta, and the sum of
the flows of the thermal, chemical, and kinetic energies and the
work of the pressure forces, respectively, through a surface moving
with velocity D together with the detonation wave so that the
magnitude x - Dt is constant over the surface. The expressions
for the kinetic energy and the other data are in the system of
coordinates of the observer moving with velocity D together with
the wave; for example, the kinetic energy of unit mass is l(D-u)2
1
and not _ u 2. Using the more suitable unit of specific volume
v (cm3/g) instead of the density and writing all expressions for
the initial state of the explosive gas, the fundamental system of
equations is obtained:

(D-u)Iv
: A1 = Dlv0 (3a)

p + (D-u)Zlv
= A2 = P0 + D21v0 (3b)

I + ½(D-u)2= A31AI= 10 + D2/r (3c)

where the enthalpy I = E + pv (cal/g or (g cm2/sec2)/g) is intr(_

duced.

All magnitudes for the initial state are indicated by the sub-
script O; the velocity of motion of the gas in the initial state is
assumed to equal zero (so that in this system all velocities are at
rest with respect to the initial mixture.
4 NACATM1261

After eliminating the magnitudes D and u from equation (2), _o

O

_4
O-P0
D 2 = v02 (4a)
vo-v

(D_u)2 = _v2 D2 = v 2 P'P0 (4b)

v02 v0-v

_ Vo'V
u v0 D =4(v0-v)(p-p0) (4c)

the fundamental equation, the so-called dynamic adiabatic of

Hugoniot (reference 6), is obtained:

I(p,v)- Io (po,v0)- 1 (v0+v)(p_p0)

= o (s)

In the case wherein the state of the substance under consideration

the substance has the same chemical composition as in the initial
state so that the functions I and In are equal (a shock wave
without change of chemical state), the curve satlsfyirg equation (5)

vin =the p,v-plane has

is antheevident
form3BAMsolution.
(fig. i);
v0 (point A) If, in
on particular,
the contrary, P -
th _0,

transition from state P0 and v0 to the state p and v is

accompanied by the process of exothermic chemical reaction so that
for given p and v

I(p,v)_I0(P,V)

because I0 still contained chemical energy that is no longer in I,

the relation between p and v, according to equation (S), is repre-
sented by a curve of the form GECZIKD (fig. i).

In correspondence with equation (4a), which connects the velo-

city of propagation with the change of state, the curve GECZIKD
(fig. i) breaks up into three parts. The part KD corresponds to
the propagation of the flame with a velocity less than the velocity
of sound in the initial mixture - the so-called deflagration. The
usual velocities of propagation of the flame correspond to the points
very near the point K at which P = P0" According to equation (4b),
the velocity of the products of combustion is negative, that is, they
move in the direction opposite to the motion of the flame.

The part IK of the curve corresponds to the imaginary values

of the velocity of propagation and therefore does not correspond to
 NACA TM 1261 5

any real process. Finally the part GEC?I of the curve corresponds
0 to the propagation of the flame with velocity greater than the velo-
city of sound in the initial gas, that is, of the detonation, the
velocity of the gases having the same sign as the velocity of deto-
nation; the products of combustion, compressed to high pressure and
density higher than the initial, move in the direction of the
initial substance.

On the curves of figure i, the detonation velocity ma_ also

assume various values from a certain minimum D to infinity,
whereas the velocity of deflagration (quiet combustion) may vary
from zero to a certain maximum D 1.

Experiment, however, shows a sharp difference between deto-

nation and deflagration in this respect.

The velocity of deflagration is considerably smaller than the

characteristic value D1 computed from these gas-dynamical con-
siderations. The velocity of deflagratlon is several tens or
hundreds times less than the magnitude D1 and can vary greatly
for a small change of state of the mixture (for example, on add-
ing traces of hydrogen to mixtures of carbon dioxide) without
changing D 1. Accordingly, in all theories of deflagration the
velocity of the latter is associated with the temperature conducti-
vity and the velocity of chemical reaction in the mixture. For
detonation, on the other hand, the large stability of the value of
the propagation velocity, which depends very little on the external
effects, is very characteristic.

In a large number of cases, with all the accuracy that may be

reasonably demanded, the measured velocity of detonation agrees
with the minimum velocity D, which, in general, is possible on the
branch of the curve GECZI of figure 1 (Chapman reference I).

The corresponding regime differs also in this notable proper-

ty (Jouguet) that the velocity of sound in the products of com-
bustion is accurately equal to the velocity of the detonation with
respect to the products of combustion:

C = D - u (8)

where C is the velocity of sound

C 2 = v2(3p/_v)s (7)
 6 _ACA _ 1261

The equivalence of the two conditions (Chapman, reference i,

and Jouguet, reference 3) in the classical theory of detonation
without losses was strictly demonstrated by Crussard (reference 4).

For accurate numerical computations of the velocity of deto- O

nation in the absence of losses but with account taken of dissoci-
ation, the dependence of the specific heat on the temperature, and
so forth, the condition of Chapman appears more suitable (see the
computations of Ratner and Zeldovich, reference 9). However, the
physical meaning of the condition of Jouguet (reference 6) lEd. note:
Correction, reference 3_ is very much clearer and the considerations
that follow on the mechanism of the process of chemical reaction in a
detonation wave refers to this condition of Jouguet's.

2. CHOICE OF DEFINITION OF THE VELOCITY OF DETONATION

IN EXISTING THEORIES

Although the preceding considerations, based on incontestable

laws of mechanics, permit any velocity of detonation, greater than
or equal to D, the choice of a single defined value of the velo-
city and correspondingly of a single defined state of the products
of combustion at a definite point on the segment GECZI (fig. l)
requires the introduction of additional considerations.

As is clear from the preceding section by the conditions of

Chapman-Jouguet, corresponding to test results, the point of tan-
gency C of the straight llne ACB drawn through this point
from the point representing the initial state to the dynamic
adiabatic is selected. The higher lying points can be eliml-
nated more or less convincingly by noting that for the state GEC

c >D - u (s)
so that a disturbance (wave of rarefaction) may overtake the front
of the detonation wave and weaken it.1

1Wendlandt (reference i0) emphasizes the analogy between the com-

pression detonation wave on the branch CEG and the simple com-
pression shock wave without chemical reaction, which is likewise
overtaken and weakened by the rarefaction waves behind. The deto-
nation wave at the point of contact, however, for which Chapman-
Jouguet condition is satisfied, is similar to a sound wave and is
converted to a sound wave as the thermal effect of the reaction
approaches zero.
 NACATM1261 7

The occurrence of waves of rarefaction is connected with the

fact that in the detonation wave front there is an increasing of
_o the density (above the initial), whence it is seen that on the
O
igniting of the mixture in a closed pipe there should be a region
of lowered density because the total amount of substance and there-
fore its mean density must remain constant. In considering a steady
propagation, the impossibility of maintaining a constant higher
density (corresponding to a constant positive and different from
zero mass velocity according, to equation (la) or (3a)) and a
constant rise in pressure and temperature are obtained as a
result of friction of the gas at the wall and the heat transfer
to the side walls of the pipe.

By anticipating the detailed investigation of the heat trans-

fer and friction in the detonation wave, it is not difficult to
establish that the only possible final state of the products of
combustion a sufficient time after the passing of the detonation
wave is characterized by a temperature equal to the temperature of
the walls, due to the heat interchange, and velocity of motion of
the gases equal to zero, due to the slowing action of the walls.
In agreement with equation (la) (only this equation of the conser-
vation of matter remains valid notwithstanding the introduction of
resistance and heat interchange which change the form of equations
(la), (ib), (Ic), (3b), and (3ci), for a velocity of the gases
equal to zero the density does not differ from the initial density.

The rarefaction, the drop in density and pressure as compared

with the conditions in the wave front, is present but at the same
time as a result of the cooling there is also a drop in the velo-
city of sound to a value considerably less than the velocity of
the detonation.

Hence if it were possible to construct a regime in which for

the state represented by the point E on the segment GEC, where
expression (8) is satisfied (at the instant of ending of the chemi-
cal reaction) the substance would be subjected to a resistance and
loss of heat and there would at that point arise a layer of the
substance with lowered (on account of the lower temperature) velo-
city of sound, which would shield the detonation front from an$ r
additional rarefaction waves.

Thus the question of the exclusion of the segment GEC assumes

in the strict theory of the steady propagation a distinct aspect.
Actually even in this more strict theory the setting up of the
over-all regime (with succeeding retardation and cooling), in
which at the end of the chemical reaction relation (8) would be
8 NACATM1261

satisfied, appears impossible. The usual concepts on a wave of

rarefaction overtaking the detonation wave is found to be very (o
O
close to the actual conditions. _0

Entirely unsatisfactory, however, are the considerations by

which the lower branch CZD(fig. I) of the Hugoniot adiabatic is
excluded. Jouguet points out that the points described on the
segment CZD of regimes for which

C <D - u (9)

are unstable. Jouguet correctly remarks that as a result of the

velocity of the wave (relative to the products of combustion)
greater than the velocity of sound, any small disturbance of state
of the products of combustion cannot overtake the wave front. On
the contrary, the distance between the distrubed region and the front
will increase. However, it is impossible to agree with the con-
clusion that this means an "instability" of the wave, for the
distance between the disturbance and the wave increases but the
disturbance itself does not increase (in the presence of dissi-
pative forces it even decreases), that is, does not disturb the
propagation of the wave.

Becker (reference Ii) points out that for a given velocity

of detonation (determined by the slope of the straight line from
point A, for example AZE3, fig. i or 2) the entropy on the lower
branch (point Z) is less than at the point of intersection on the
upper branch E (at which point the inequality (8) holds). Further
on Becker writes as follows "It seems as though for a given velo-
city of detonation the products of combustion had the choice of
passing over either _nto the lower point (Z) or the upper point
(z)". And further, "If we imagine that the products of combustion
at the instant of their formation assume a state to which in the
sense of statistical mechanics there corresponds a greater proba-
bility, then it may be concluded that the products of combustion
choose the point C (on the upper branch in the present notation)
so that the lower part of the detonation wave branch will not
correspond to any real process."

The unconvincing nature of these considerations is evident.

A very simple example, contradicting such a priori assertions, is
furnished by the propagation of compression shock waves in a gas.
In the shock wave the entropy increases, the probability decreases,
but in order that the transition should occur to a more probable
state it is not sufficient merely that the gas should undergo an
increase in entropy - there is also required the motion of the
piston compressing the gas.
2 NACA TM 1261 9

Simulating even more closely the case of detonation is the

following experiment, which may be conceived as carried out: The
O gas that is enclosed in a pipe with transparent wall is by light
o_
reflection imparted a certain small quantity of energy. By rota-
ting the mirror, any velocity of displacement along the pipe of
the point giving out the energy (even a velocity greater than the
velocity of light) can be brought about. It is evident that for
a velocity of displacement of the reflector greater than that of
sound the state subject to the action of the light of the gas will
describe the point Z (fig. 2) corresponding to the given velocity
for which the increase in pressure and temperature is proportional
to the energy of the light ray bringing about the change in state.

For the upper point E corresponding to the given velocity

of the reflector, on the contrary there is obtained a great increase
in the pressure, temperature, and entropy depending only on the
velocity but practically independent of the intensity of the light
ray itself.

It is evident that the regime E (in the absence of a

piston compressing the gas in addition to the action of the light
reflector) is impossible. The impossibility of realization, in
detonating combustion of gas mixtures, of states corresponding to
the lower branch of the curve may be due (and actually is due as
shall be seen later) only to the concrete mechanism of the liber-
ation of the chemical energy in detonation. In reference return
to the example given previously of the propagation of disturbances
in the motion of the reflector, the illumination of the gas may
not only heat the gas as a result of the absorption of the light
but bring about a photo chemical reaction with the giving off of
heat. 2

What constitutes the difference between true (usual) deto-

nation and that of pseudopropagation with an externally given
velocity (by the motion of the mirror)?
J

In the present example, the heating of only a small volume

of the gas followed after (but not as a result of) the heating of
the preceding volume following the motion of the light reflector.
But in an actual propagation of detonation "post hoc = propter hoc,"
the giving off of chemical energy in a certain volume is connected
in a casual relation with the process of the chemical reaction in
the preceding (along the path of the wave) volumes of the gas. In

" 2See section 3.

i0 NACA TM 1261

this sense the truth is more closely approached by the qualitative

considerations of Jost (reference 12): namely, on the lower branch
(particularly on approaching the point I corresponding to constant _o
0
volume), the velocity of the propagation of the detonation is
greater than the velocity of sound and the chemical energy liber-
ated over a certain distance cannot be imparted to the wave front.
Only by excluding by such unconvincing considerations both branches
above and below the point C does modern theory approach the only
indisputable point at which accurately

C =D- u

namely, the point of tangency C, which also gives, as was previously

shown, a single well-deflned value of the velocity of detonation
confirmed by experiment. It hardly serves to make mention here of
the attempts to obtain the point of tangency as the only possibility
from other unproved conditions (mlnimumvelocity of detonation or
mlnimumentropy, and so forth) taken neither as supplementing nor
as substituting for the equations of hydrodynamics and chemical
kinetics (reference 43).

Finally, entirely inadmissible at the present time are the

attempts to identify the velocity of detonation with the velo-
city of motion of any particular molecules, atoms, or radicals in
the products of combustion, the corresponding particles being
assumed active centers of a chemical reaction chain (reference i_).

However good the numerical agreement, such an attempt is no

more than a make-shift and a clear backward step with respect to
the thermodynamic theory as is evident from the fact alone that it
is entirely unclear what mean or mean square velocity, or other
velocity of the molecules, should enter the computation.

The author of this theory points out the arbitrariness

of the choice of the carrier of the reaction. In the case of the
detonation of a mixture of acetylene with nitrogen oxide, there is
computed the velocity of the atomic oxygen, the atomic carbon, and the
molecular nitrogen. The arithmetic mean of these three values gives
to an accuracy of 0.8 percent the measured detonation velocity.

Lewis points out the desirability of a combination of his

"chain theory of the detonation velocity" with the classical theory
of Chapman-Jouguet. This last theory would hardly, however, gain
from such a combination.
 NACA TM 1281 Ii

3. PRESERTVIEWS 0NTHE PROCESS OF THE CH]_ICALREACTION

IN A DETONATION WAVE
O
o_
Thus even in the classical theory of the dynamic adiabatic
and the detonation velocity, though providing a number of indispu-
table successes, the unquestionable equations (1) to (5) and the
practically entirely satisfactory method of computing the velocity
of detonation, there is no complete clarity as regards its logical
basis. Matters are even less favorable as regards the problem of
the conditions under which the chemical reaction proceeds in a
detonation wave.

The majority of the investigators are inclined to the view,

which is analyzed and developed further on, that the start of the
chemical reaction, the ignition, is connected with the heating of
the gas above its self-ignltion temperature by adiabatic com-
pression in the shock wave. It is for this reason that other
points of view and the objections encountered in the literature
against such assumption were considered first.

0nly for the sake of curiosity there may be mentioned the

"quantum mechanical resonance" between the components of the gas
before and after the passage of a detonation wave, giving rise in
some mystical manner to the process of the chemical reaction
(reference 14).

The impossibility of the completion of the chemical reaction

over the distance of one length of free path, in particular for
somewhat complicated reactions requiring several encounters of an
entirely definite kind, is sufficiently well demonstrated, as for
example, in the book by Jest (reference 12). Thus between two
ternary collisions, the molecule for usual densities is subjected
to about 1000 ordinary collisions. The heat of activation of the
order of 40,000 calories per mole even at a temperature of 3000 ° K
decreases the probability of the reaction to 0.001 so that there
are again required about 1000 collisions for an elementary reaction.

As soon as it has been shown that the chemical reaction cannot

proceed over a length of the order of a mean free path, all theo-
ries fall away in which the direct impact of the molecules of the
products of the reaction with the molecules of the initial substances
play a leading part. In fact, between the fresh nonreacting gas
and the products of reaction there is a more or less wide zone in
which the reaction proceeds; there is a change in concentration,
temperature, density, pressure, and mass velocity of motion.
12 NACATM 1261

Because the width of this zone is considerably greater than

the length of the free path of the molecule, the transfer of energy
by direct impact is not considered but the heat conductivity _D
(D
and other dissipative processes in the gas - diffusion and viscosity -
connected with the gradients of the t_perature, concentration, and
velocity on the normal to the wave front are considered.
It is to the heat conductivity _ along the wave (in the dir-
ection of propagation of the detonation wave) that Becker (refer-
ence ll) ascribes fundamental importance in the propagation of
detonation. Becker, without investigatin_ the problem in more
detail, gives an example of the propagation of a shock wave in
which the increase of the entropy is due, in the final analysis, to
the dissipative forces themselves, primarily to the heat conduc-
tivity in the wave front.
An analysis of the conditions of the propagation of the normal
flame permits evaluating the order of magnitude of the amount of
heat transferred by the heat conductivity.

For normal (quiet) combustion the propagation of which can be

due only to the heat conductivity 3, the flow of heat is a magnitude
of the sameorder as the heat of combustion given out in unit time.
The width of the front should be of the sameorder as the product
of the time of the chemical process by the velocity of propagation
of the flame.

The conditions in a quiet flame shall nowbe comparedwith the

conditions in a detonation wave. The order of magnitude of the
temperatures and the temperature differences In both cases is the
same. For the samechemical reaction time, the width of the zone
in the case of the detonation wave, is greater than in the case of
the quiet combustion in regards to the ratio of the velocities
D/D"qulet; in this case, the gradients of the temperature and the
concentration drop in the reverse ratio D"/D and in the same
ratio there is a drop in the flew of heat transferred by conduc-
tivity (or diffusive flew). On the other hand for the sameor
nearly the samecalory content of detonating and deflagrating
mixtures, the amountof heat given out per unit time per unit area
of flame front is greater in the case of the detormtion than in the
case of the deflagration in the ratio D/D". Finally, assuming t!mt

50r the other transfer process entirely analogous in its molecular

mechanism, of the diffusion of active centers. The substitution of
heat conductivity by diffusion in no way changes the conclusions.
 NACA TM 1261 13

in the deflagration (normal combustion) the flow of heat trans-

ferred by heat conductivity is of the same order as the heat of
combustion, there is obtained for the detonation the ratio of heat
(._
O
conducted to heat given off.
Or)

QD " _ 10-9 ÷ 10-5 (i0)

where k is the heat conductivity in (cal)(deg'l)(cm-l)(sec'l)

or (erg)(deg'l)(cm-l)(sec "I) = (g)(cm)(deg'l)(sec-3); T is the
temperature: x is the coordinate normal to the flame front (cm);
Q is the calory content of the detonating substance in (cal/cm3)
or (erg/cm3) = (g/cm sec2): D the velocity of detonation (cm/sec);
and D" is the velocity of deflagration (cm/sec).

The obtained ratio (equation (I0)) is so small that there is

no need of attempting to obtain a more accurate ratio of the heat
transfer and heat of reaction in any theories of normal combustion
(references 3, 4, an& 15 to 18) or more accurate differences in
temperatures in the detonating wave, and so forth, by operations
that in no way can alter the fundamental factors; namely, the small-
ness of the flow of heat in the direction of propagation of the deto-
nation, which to a very great accuracy is adiabatic (provided the
heat losses on the side walls of the pipe are neglected), the process
of the chemical reaction in the detonating wave, and the impossi-
bility of assuming any important effect of the heat transfer by the
heated products of combustion on the ignition of the fresh gas.

It remains to clarify the underlying reasons for the error of

Becker, the great difference between the conditions in the shock
wave and the detonation wave.

In considering the steady propagation of a compression shock

wave, ther_ is no previously given characteristic magnitude of time
or length. Such magnitudes - the width of front of the sh_ck wave,
the time of compression in the shock wave - appear only as soon as
dissipative phenomena are considered: heat conductivity and viscosity
in the wave front. The width of front of the shock wave is chosen
(or computed) in such a way as to give a sufficient large temperature
gradient in order that the heat flow be sufficiently large and to
assure the required increase in entropy independent of the value of
heat conductivity and is determined by the difference between the
dynamic adiabatic of Eugoniot and the (static) adiabatic, the isen-
tropic of Poisson.
14 HACATM1261

The smaller the heat conductivity the greater should be the

temperature gradient, the smaller the width of the front and the (o
0
time of compression in the shock wave. As shownby Prandtl (refer-
ence 19) and Becker (reference ll), for someconsiderable value of
the compression, the width of the front in a shock wave in a gas is
approximately equal to the length of the free path.
The matter stands entirely differently in the case of the deto-
nation wave. In this case, there is an entirely definite character-
istic of the time - the time of the process of the chemical reaction.
In combination with a definite linear velocity of propagation of the
detonation, the width of the zone of the chemical reaction is obtained,
which cannot change (as was the case with the shock wave) in follow-
ing the changes of the heat conductivity. The chemical reaction can-
not proceed for the time of a single collision, manymolecular col-
lisions being required_ and the width of the zone extends over many
free path lengths.

If, according to the computations of Becker, the heat conduc-

tivity becomesnoticeable only for a width of the order of the length
of free path, then naturally for a zone extending over several lengths
of free path the flow of heat will becomenegligibly small in corr_
spondencewith result of equation (10). It may again be emphasized
that in the approximation of equation (lO) the introduction of the
velocity of deflagration was only as a meansof estimating the order
of magnitude of the time of the chemical reaction at a hlgh tempera-
ture by known experimental data.

Of great interest is the attempt to construct a theory of deto-

nation of Ismailov and Todes (reference 20), which was never published
azmlwas known to the author from the lecture at the Institute of
Chemical Physics in 1934.

Combining the first two fundamental equations of the steady

propagation, namely the equation of conservation of matter and the
equation of conservation of momentum so as to obtain the equation

D 2 = v02 P-P____O
vO.v (4a)

Ismailov and Todes give this equation a new interpretation pointing

out that in a steadily propagating medium with a certain definite
velocity, the manner in which the chemical reaction proceeds, the
specific volume and the pressure must be linearly connected accord-
ing to equation (4a).
 NACA TM 1261 15

The motion in the p-v plane as the chemical reaction proceeds

should be along the straight line ACB (fig. l) passing through the
initial point A, the straight line, which shall be called the
o "constant velocity line of Todes."
o]

The question arises in what direction the motion represented by

the straight line of Todes proceeds, lemailov and Todes assumed that
the point rises directly from the initial state A lying evidently
on the dynamic adiabatic for the fresh unchanged chemical gas to the
point C lying on the adiabatic of Hugoniot and correspondingto the
ended chemical reaction. A number of intermediate adlabatics can now
be drawn, which would correspond to the process of the chemical
reaction at 25-_0-75 percent and the relation between the chemical
reaction and the motion of the point along the straight line of Todes
from A to A25 percent, A50 percent, and so forth, can be easily
determined for the points of intersection with the corresponding
adiabatics (fig. 3).

It is not difficult to see, however, the physical inadmissi-

bility of the chosen direction of the motion. In the considerations
of similarity, there now necessarily enters the velocity of the chemi-
cal reaction at low temperatures, the initial temperature, and temper-
atures near it.

At room temperature, the mixture of hydrogen with oxygen reacts

very slowly, whereas the detonating wave travels 2 to 3 kilometers
in a second.

The heat conductivity, as has been seen in the case of the deto-
nation, cannot be responsible for the initial rise in temperature
where the temperature is low and the velocity of the chemical reaction
is clearly insufficient.

Finally, the entire motion of the point considered over the

distance AC occurs in regions in which the velocity of sound is
less than the velocity of propagation of the wave.

c <D - u (9)

Exactly as'in the case of the pseudopropagation in the example with

the light ray, the relation (9) contradicts the possibility of a
casual relation between the different phases of the combustion. The
reaction starts somehow of itself at the initial temperature inde-
pendently of the approach of the detonating wave because the trans-
mission of the ignition by a material agent in a dense medium is
impossible through a layer of substance fo_ which the relation (9) holds.
 16 NACATM1261

Although the considerations of Ismailov and Todes regarding

the path of motion of the system, the start of the reaction at the
initial temperature, are rejected, in the following discussion use (D

is madeof the clear concept of the straight line of Todes

The objections raised by Becker (reference ll) against the

ignition of the explosive mixture by a shock wave, with references
to Nernst (reference 21) and van't Hoff (reference 22), are entirely
incorrect and are based on evident miscomprehension. Basing his
objections on these authors, Becker states that for an adiabatic
ignition there are required extremely higher pressures, up to 100 or
2E0 atmospheres, than the pressures that are available in the deto-
nation wave.

Actually the 250 atmospheres of van't Hoff mentioned by Nernst

(reference 21) are the same pressures that may be developed in adia-
batic combustion by self-igniting explosive mixtures compressed to
the ignition temperature.

On the contrary, the pressures of the adiabatic compressions

(isentropic of Poisson), which are required to bring the mixture
to self-ignition, are 19.5 to 23.9 atmospheres according to the
computation of van't Hoff and 25 to 40 atmospheres according to the
data of Falk (references 23 and 24) presented by Nernst.

Moreover, in the shock wave (adiabatic of Hugoniot) for a

strong compression the raised temperatures considerably exceed the
heating in an isentropic compression (adiabatic of Poisson) to the
same pressure.

Thus, the compression of the gas by the shock wave is necessary,

the only possible method of igniting the gas, brought about by the
chemical reaction in the propagating detonation wave.

What are the characteristics of the shock wave propagated ahead

of the detonation wave, which ignite the gas? There is often

encountered the statement (Jouguet_ reference 3; Sokolik, refer-

ence 25, and Crussard, reference 4) that the shock wave has the same
pressure as the detonation wave (or less, Crussard (reference 4)).
The pressure in the shock wave propagated in the fresh mixture is
equal to the pressure corresponding to the point of tangency of the
dynamic adiabatic, the pressure of the gas at the state of the gas
at the moment of ending of the chemical reaction.

There is presented here a comparison (borrowed from Jouguet,

reference 5) of the initial state of the gas, the state of the gas
at the instant of ending of the reaction, and the state in the shock
 NACA TM 1261 17

wave, developing precisely the ignition temperature (555 ° C, table I).

_D The excellent numerical agreement is seen: For a mixture with

O
a very small addition of nitrogen (a greater addition makes the
,-4
mixture unsuitable for detonating), the pressure of the detonation
is Just sufficient for self-ignltion to occur in the shock wave of
the same pressure. However, the smaller velocity of propagation of
the shock wave (in comparison with the detonation wave) for equal
pressure renders impossible the propagation of a steady regime in
which there is an igniting shock wave ahead of the products of com-
bustion. In the example given it is not clear in what manner the
shock wave, the velocity of propagation of which is only 1210 meters
per second, assures the propagation of the chemical reaction of deto-
nation with the velocity of 1660 meters per second.

From the consideration (inaccurate, but it shall not be dis-

cussed here) of the conditions that give rise to the detonation wave,
in particular, basing the results on the experimentally established
fact of the appearance at the point of occurrence of the detonation
of a reverse compression, the so-called "retonation wave," propagated
over the products of combustion, Crussard (reference 4) concludes
that the ignition shock wave should satisfy the conditions

P < Pdet
(ll)
u >Udet

Satisfying these two conditions assures the "correct" regular

occurrence of the detonation.

The practical applicability of conditions (ii) is complicated

by the fact that the detonation wave has often a velocity and a
pressure considerably greater than in the stationary regime (for
example, the data of Bone, Fraser, and Wheeler, reference 25)
according to which in a mixture of CO and 02 the velocity of
detonation at the instant of its occurrence reaches 3000 meters per
second, whereas the steady velocity is equal to 1760 meters per
second.

It is entirely inadmissible to apply conditions (Ii) to the

shock wave continuously igniting the gas in the steadily propagating
detonation. In this case, there are no retonation waves. The con-
ditions required more or less for the occurrence of detonation have
no relation at all to the steady propagation.
18 NACATM 1261

The conditions of the equality of the pressures in a shock

wave or wave of detonation (or P< Pdet or U>Udet, relations
(iI)), which from Jouguet (reference 3) "are natural", actually,
in the field of gas dynamics and velocities comparable with the (D

velocity of sound has no basis in fact. The more detailed experi-

mental work on the detonation limit refutes the agreement found by
Jouguet. (See the following discussion.)

If a strictly steady regime of propagation of the detonation

is considered in which the entire wave moveswith a single well-
defined and constant velocity, it is evident that the shock wave
igniting the gas must satisfy the single condition, that is, be
propagated in the gas with a velocity equal to the velocity of
detonation.
In the earlier literature mention is madeof shock waves of
velocity of propagation D. Thus Vieille (reference 27) in a note
emphasizing the part of the discontinuities (shock waves) in the
propagation of explosions for the explosive gas 2H2 + 02, the
velocity of sound being 510 meters per secondj the velocity of
detonation 2800 meters per second, finds the pressure in the shock
wave of the samevelocity equal to 40 atmospheres and further
Judges the possibility of attaining such pressure for a combustion
without change in velocity of the gas initially compressedto
several atmospheres.

Crussard (reference 4) for the mixture 2C0 + 02 in which

the measuredvelocity of detonation was varied and was approxi-
mately 1210 meters per second showsthat in a shock wave, propa-
gated with the velocity of 1210 meters per second, the tempera-
ture of the compression is 720° C, so that at this temperature
there is still possible an appreciable lag of the self-ignition.
In 1924-25, Wendlandt (reference lO), a student of Nernst, ener-
getically defended the point of view according to which the gas
is ignited by a shock wave of velocity equal to the velocity of
the detonation wave. Wendlandt investigated in detail the concen-
tration limits of the detonation propagated in explosive mixtures
and measuredthe velocity of the detonation near the limit on
the ignition of the investigated mixture by detonation of the
explosive gas in a special part of the pipe. The steadiness of
propagation of the detonation is established by the comparison of
the velocity over two sections of the path. Near the limit the
velocity drops sharply, differing considerably from the computed
value. The agreements obtained by Jouguet are found to be com-
pletely illusory. On the contrary, the temperature in the shock
wave of equal velocity (but not equal pressure) at the limit is
 NACA TM 1261 19

found to be of the order of the temperature of self-igniting with

minimum ignition lags, the temperatures of the start of the rapid
o_ reaction. Figure 4, (taken from reference i0) for hydrogen-air
0
o_ mixtures, gives the results of all Wendlan_t's tests and compu-
tations. On the abscissa are plotted the hydrogen content in per-
cent and on the ordinate, the velocities in meters per second.
Curve i gives the velocities of detonation computed by the classi-
cal theory. There are also plotted the results (curve 2), of the
computations of S. B. Ratner and Zeldovich (reference 9j 1940).

The small crosses denote the velocities of propagation on the

first part of the path nearest to the point of ignition, and the
clrcles on the second part further removed. Above the limit in
the region of steady propagation, both velocities coincide. Below
the limit the detonation wave is extinguished; the velocity in the
second part is less than in the first.

Finally, the dashed curves give the velocities corresponding

to the shock waves in which are developed the temperatures denoted
on the curves (i000 ° K, 900 ° K, and so forth). It is seen that at
the limit in the shock wave of equal velocity (but not equal
pressure as is assumed by Jouguet), there is attained the tempera-
ture of self-ignition with minimum ignition lags.

In conclusion of the review of the existing views on the

propagation of detonation, the phenomenon of spin in the detonation
should be briefly considered and several attempts made to explain it.

Campbell and Finck (reference 28) revealed the known periodicity

on the photographs of the detonation in certain mixtures.4 A number
of succeeding tests showed that such periodicity may be connected
with the propagation of the detonation in spirals.

The only theoretical work (reference 50) in which an attempt

is made to describe such propagation in a cylindrical pipe in three
dimensions is not convincing because the assumptions of equations
of spiral flow upon which it is based is without foundation. Of
importance, however, is the relation computed between the period
and the diameter of the pipe, a relation excellently confirmed by
tests.

4The periodicity of the distribution of the illuminated particles

in the cooling products of detonation is so accurately observed
that it must not be considered as accidental (Dixon, reference 29).
20 NACA_ 1261

On the other hand, in someof the work of very recent times

this relation is ignored as also the direct demonstration of spiral
propagation. The spin is described as a one-dimensional but not
steady phenomenon. (D

r-_

Becker in his work of 1936 (reference 31) considers a mixture

with insufficient velocity of chemical reaction for which the steady
detonation satisfying the condition of Chapman-Jouguetis impossible
because the temperature is not sufficiently high. The compressed
detonation wave corresponding to the upper branch of the Hugoniot
adiabatic with a temperature higher than that corresponding to the
point of tangency is propagated, igniting the mixture, but the wave
is gradually weakenedto normal. The detonation wave is then dis-
continued but the shock wave due to the still sufficiently lively
reaction again causes detonation of the gas over a certain distance
ahead, which then again breaks off, and so forth.

Jost (references 12 and 32) considers a reaction beginning only

after the passage of a certain time T (period of induction) after
the heating of the mixture. The shock wave entering the gas during
the time T is propagated along the gas up to the start of the
reaction and weakens gradually. After the period of induction T
the ignition of the compressed mixture occurs and shock waves are
propagated from the point of inflammation. The forward traveling
wave overtakes the first weakened wave, strengthens it, and the
process starts anew.

Quite similar views with regard to the periodic changes of the

process in which the shock wave first travels ahead then is over-
taken by the flame are developed by Avanesov and Rukin (reference 33)
who consider especially the chain character of the process, the part
played by the active centers, and so forth.

These considerations still, however, are far from constituting

a theory of periodic propagation and their unconvincing character
is evident. In order to prove the existence of a periodic state,
it is first of all necessary to consider possible steady states
(complete absence of reaction, deflagration, and detonation) to
explain the regularity and stability, to construct a quantitatively
assumed periodic regime, in particular to find its mean velocity
of propagation, and to show that this regime does not transform
asymptotically into any steady regime.

0nly by knowing the conditions of ignition in the steady deto-

nation wave will it be possible to decide whether it is possible for
an explosive mixture unsuitable for detonation with normal velocity
to ignite by shock waves due to the "still lively reaction" to quote
Becker (reference 31).
 NACA TM 1261 21

Under the same assumptions as those made by Jost (refer-

ence 32), It is entirely possible to set up a strictly steady
state - a shock wave with successive additional compression to
compensate for the losses and a front of rapid chemical reaction
0
O_ at a constant distance (product of the velocity of the detonation
by the period of induction) from the shock wave front. It Is
not clear whether the propagation in steps described by Jost will
lead to such a regime.

In the present work, the conditions of one-dlmenslonal

steady detonation, effect of heat transfer and resistance, and
distribution of temperature, pressure, and so forth, in a
strictly steady regime shall be studled.

Such computations are required as a startlng point of the

whole more complicated theory of the propagation of detonation
in t_hree dimensions, the theory of periodic states, and so forth.

4. IGNITION BY COMPRESSION IN THE SHOCK WAVE AND THE CHOICE OF

A DEFINITE VALUE FOR THE VELOCITY

In the following section, a strict theory of the steady prop-

agatlon of detonation s.hall be constructed with account taken of
the losses that are required in all cases for determining the llm-
itlng conditions. First, an elementary proof of the fact that the
mechanism of ignition by a shock wave actually excludes the possi-
bility of the realization of the lower branch CZI shall be given.

To the ideas of Le Chatelier (reference 15), Berthelot (ref-

erence 34), and Vlellle (reference 27) on ignition by shock com-
pression, the categorical proof of Wendlant (reference 10) on the
ignition by a shock wave of velocity equal to D, and the consid-
erations of Todes and Ismallov (reference 20) that there must be
a linear relation between the specific volume and the pressure
from the laws of conservation, there corresponds the following
plcture of the process in the p-v plane (flg. 1): There occurs
an instantaneous Jump from point A to point B - the sudden
compzesslon without chemical reaction. Whereas In the initial
state A the velocity of the chemical reaction was negligibly
small, the state B corresponds to a high temperature at which
the chemical reaction proceeds with considerable velocity. To
the process of chemical reaction corresponds the motion of the
point along the straight line of Todes ACB In the direction from
B to C.
22 NACATM 1261

At greater velocity of detonation after the Jump A3 the

motion occurs over the segment 3E and the impossibility in the
detonation to arrive at the point Z of the lower branch CZI thus
follows directly from the mechanismof the chemical reaction pro- UD
O

cess requiring for its start a shock compressionwith following W9

smooth motion along the straight line of Todes. The Jumplike

motion along the straight line of Todes (corresponding to shock
waves) is possible only in the direction from bottom to top because
the increase of entropy corresponds only to this direction.

It should be expecially emphasizedthat from the present con-

siderations there follows the possibility of the propagation of t_
detonation with a velocity exceeding that computedfrom the condi-
tion of Jouguet with the existence in the products of reaction of
the state Z on the lower branch of the Hugoniot adiabatic. This
assumption is true in that case in which the igniting agent, for
example, the radiation of the products of the reaction, is propa-
gated more rapidly than the shock wave and produces a chemical
reaction in the substance in state A or very near it. In this
case corresponding to the initial assumptions of Todes and Ismailov
on the motion over the segment AC, there is also possible a motion
with another velocity, for example, along the segment AZ with
the realization of the point Z on the lower branch notwithstanding
the smaller entropy. This possibility may be of importance for th_
theory of detonation of porous condensedexplosive substances.

5. PROPAGATION OF DETONATION IN A PIPE WITH ACCOUNT TAKEN

OF RESISTANCE AND HEAT TRANSFER

The propagation of a detonation wave in a pipe shall be con-

sidered with account taken of the heat exchange and resistance at
the side walls. The discussion is restricted to one-dimensional
theory in which the heat exchange and resistance are uniformly
distributed over the entire section of the pipe. The coordinate
reckoned from the front of the detonation wave in the direction
toward the fresh gas is denoted by x on which alone depend all
the magnitudes in the steady, one-dimensional theory:

d D-U
=0
dx v
(12)

D-U D
v v0
NACA _ 1261 23

d-_ + =-F (13)

_- + --- + O - DF (14)
dx II _I D-U
V -

where D is the velocity of the detonation, u the velocity of

motion of the substance (all measured with respect to the undis-
turbed fresh gas), p the pressure, v the volume per unit mass
(cm 3 /g), I the enthalpy of unit mass, M the mass flow with
the velocity of detonation over unit area (g/cm3) (sec)6, equal to
the mass velocity of combustion, F the.force of resistance per
unit area of the pipe referred to unit section

=- (15)
d 2v

where according to the definition used in hydrodynamics of the

nondlmensional coefficient of friction

fj = tj o

G is the amount of heat, referred to unit section, given off by

the gas to the walls over unit length of the pipe in unit time.
In the case where the velocity of heat exchange is determined
from the heat transfer from the gas to the walls,

2d v
+_2 - % (16)

where ICT is the enthalpy of gas at the temperature of the wall,

the coefficient m = I, if the analogy of Reynolds between the
heat exchange and friction holds, m < 1 in rough pipes (see
any course in gas dynamics and heat transfer). In equation (14),
the term DF is the work of the friction forces in a system of
coordinates moving with the wave.

6 The magnitude AI of equation (la).

24 NACA TM 1261

The enthalpy I in addition to the pair of variables defining

the physical state of the substance (for example p and s or p
and v) depends also on the chemical variable n - the intensity _0
0
of the process of an irreversible chemical reaction and which for
definiteness is identified with the concentration (nondimensional
g/g) of the final products of the reaction. It should be borne in
mind that the reaction proceeds, particularly at the start,
irreversibly.

I = l(p, v, n)

(17)

dl = (3_pD dp + (_ dv + 8(_nD dn
v,n p,n p,v

where (31/_n)p,v is the heat of chemical reaction, taken with

opposite sign, referred to unit mass of the reaction product, posi-
tive in an exothermal chemical reaction. It may be shown that the
heat of reaction of the detonation wave entering the equation is

Cp Qv-Cv Qp
(18)
Cp-C v
p,v

where Cp and cv are the specific heats and Qp and Qv the

heats of reaction at constant pressure and volume, respectively.
In the case of a reaction at which the number of molecules does
not vary

g --Qp - Qv (18a)

The derivatives taken at constant value of the chemical para-

meters, that is, in the absense of an irreversible chemical reac-
tion may be transformed with the aid of the thermodynamical rela-
tlons. By comparing equations (17) with I = I(p,S,n) and
S = S(p,v,n)

an
dp + 3(_) dS + _(_n) dn = vdp + TdS + 3(_)
n p,n p,S p,S

+ dn
v, p,n p,v P,
 NACATM1281 25

it is found that

v,n v,n

O
(z9)
o)

( Ip,n= T I- Ip,n
= " T ( IS,n ,n __

In the following discussion

S,n (20)

T _S • =v 7

v_n

The constant H has the same dimensions as M, the magnitude C is

the velocity of sound.

After simple algebraic transformations, equations (12) to (14)

are reduced to the form

dp + M2dv = - F dx (21)

Vo-V vo Q
dp + H 2 dv = - _--c7- F dx +-V_ Gdx - -- dn = - _ Fdx (22)
V V'

.....v0-v v0 G +____
Q dn = v0-v v0 G + Q dn (22a)
v' v'D F v'F dx v' v'D F v'F(D-u) dt

During the energetic reaction, _ > I; after the end of the

chemical reaction in the process of slowing down and cooling of
the reaction products, _ < I.

For concrete computations, the equation of chemical kinetics

must be added

dn/dt = f(n, T, p) = $(n, p9 V) (23)

For the following discussion, however, it is sufficient to use

the most general considerations with regard to the kinetics of the

7For an ideal gas, v' = _v

26 NACATM1261

reaction. The velocity of the reaction is large in the region of

small and mediumcombustion and becomeszero in passing to the
equilibrium state.
(D

The boundary conditions under which the equations are inte- C)

_0

grated are the following: For x > 0 the mixture is in the ini-
tial state (A, fig. i or fig. 5). For x = 0 there is a discon-
tinuity (shock wave) bringing the substance into state B
completely determined by the velocity of propagation D of the
wave. At a large distance after the passage of the wave x--_ - _,
a state must be established in which as a result of the slowing-
downaction and the heat transfer

F=0
G=0
u= 0
(24)
V = V0

T = TCT

P = PeT = QTcT/v0

Here use had already been made of equation (2) to obtain v = v0:

TCT is the temperature of the walls, PCT, the corresponding

pressure at the initial density and initial specific volume. The
integrated curves p, v, and n as functions of x are completely
determined by equations (12) to (14) or (21) to (22) and the ini-
tial conditions. The imposing of boundary conditions permits, in
addition, the determination of the special value of the detonation
veloclty D entering the equations as a parameter indirectly
through M and through the coordinates of the point B in the
p-v plan e.

Thus the equations do not require, in order to find the defi-

nite value of the detonation velocity, the introduction of any out-
side supplementary conditions, such as introduced by Chapman
(reference I) or Jouguet (reference 5), which is entirely natural
because in the equations and boundary conditions (21) to (24) are
included not only the conditions in the wave front but also the
succeeding slowing down and cooling of the products.

Solving equations (15) and (16) ,

(M2-H 2) dv

(M2-H 2) dp
= (_-i)

= (_M2-H 2) F
F d_x

dx
} (25)
 _ACA TM 1261 27

The line M 2 = H2 is a special line of the equation (dashed,

fig. 5). When the detonation is defined as a state in which the
propagation of the flame proceeds with a velocity greater than the
velocity of sound in the initial gas, at point A C < D,
C2v0 2< D2/v0 2, and H2 < M 2. After the shock compression, at
I-., point B, as is known, C >D and H 2> M 2. The shock compres-
0 sion is accompanied by a Jump through the line M = H.
o'_

The final state of the products of combustion (point C)

according to the boundary conditions differs very little from
the inltial - only as a result of the change in the number of
molecules during the reaction. At the point C, again H 2 < M 2.

At the same time, it is recalled that the magnitude _ of

equations (21) and (22a) considerably exceeds unity in the zone
of energetic chemical reaction_ in the process of cooling and
slowing down of the gas in the absence of chemical reaction or
very slow reaction, _ < 1.

The transition from B to K requiring the intersection of

the integral curve and the line H = M is possible only for
simultaneous (curve I, fig. 5):

H = M
=l ? (26)
•
It is not difficult to show, considering equations (25), that
when x decreases from 0 to - _ if the value _< 1 is attained
at H _ M in the upper region, the integral curve, not intersec-
ting the line M = H, bends back and the boundary conditions can-
not be satisfied (curve 2, fig. 5). This result is obtained if in
a mixture with less heat content than for curve 1 there is attempted
to construct a regime with the same velocity of detonation.

If, however, at the instant of attaining M : H, _ > i (curve 5,

fig. $), the integral curve, reaching the line M = H, is not pro-
longed, If 7 denotes the distance of the point from the line M = H
(positive in the upper region M < H) then for small values of 7
the differential equation holds

7d7 ~ (_-i) Fdx

(z7)
72 ~ (G-l) Fx + constant

If 7 : 0 for _ > I, a further decrease In x passing through,

In the system of coordinates chosen, the values from 0 to - _ leads
28 NACATM1261

to imaginary values of 7. Case 3 is obtained if, without changing

the velocity, the calory content of the mixture is increased as
compared with curve i.

tC
After the intersecting of the line M = H at _ = I (curve i) C

and satisfying of the boundary conditions (24), reaching point C

by the integral curve is assured. By substituting the values F
and G into the equations, it is not difficult to show that at the
end of the chemical reaction point C is a singular point of the
node type through which pass all the integral curves of the lower
region. In the lower region

D-___u
> C_ (28)
V V

H2 < M2 t
D>C+u

Naturally, in whatever manner the cooling and slowing of the reac-

tion products proceeds, they always lead to the final state (24),
any transmission of a disturbance ahead and any back reaction on
the regime being impossible.

The consideration of equations (21) and (22) thus led to the

condition of Jouguet (reference 3):

H 2 =M 2

that is, (29)

C D-u
V

C=D-u

At the point at which this condition or condition (6) is satis-

fied, the chemicai reaction has not yet ended: _ = i corresponds
to a definite velocity of the chemical reaction balancing the action
of the heat transfer and friction:

dn vG (v' + v-v0)(D-u )
d-_=-_ + Q F (30)

that is, assumes the presence of a certain quantity of fuel the

reaction of which Is continued in the cooling zone, M > H and
has no effect on the velocity of detonation.
 NACA TM 1261 29

In the literature, the statements are found that for a slowed

chemical reaction not all the heat of the reaction is given off in
the detonation wave front, which is explained by the lowered deto-
nation velocity as compared with that computed by the classical
O
theory (Wendlandt reference 10; Lewis and Friauf (reference 35),
Rivin and Sokolik (reference 36), and Jost (references 12 and 32).

Equation (30) shows that the incomplete combustion losses in

the wave front are connected with the velocity of heat transfer and
hydrodynamic resistance. The less the heat transfer and resistance,
the smaller should be the velocity of chemical reaction on the line
M = H at _ = 1 (at the rear boundary of the wave front); a
smaller concentration of the incompletely burned substance is neces-
sary for maintaining a smaller velocity of reaction. Together with
the losses in incomplete combustion, decreasing the heat of reaction
in the wave front, there must be considered a second form of the
losses of friction and heat transfer in the wave front during the
energetic chemical reaction. In the zeroth approximation, considering
the velocity of the chemical reaction very large, _--*_ in the wave
front_ _ = 1 corresponds to an extremely small incompleteness of com-
bustion approaching zero, the losses during the reaction also approach-
ing zero. These equations give in the limit the classical equations of
the detonation wave together with the condition of Jouguet, which is
obtained as a mathematical consequence of the equations and boundary
conditions (21) to (24) and requires no extraneous considerations.

Integrating the equations up to the point xI (x I < 0) at which

M = H, _ = l, there is obtained

D D-u 1
v0 vI

(D-ul)2 D3 _Xl
Pl + Vl = P0 + -- - Fdx
v° 3o

(D.Ul)2 D2 v0
II + _ = I 0 +_- + __0 xl gdx - v 0_0 xl Fdx (31)

- (_Pl/_I)s, n = D2/v02

Io = I0' + QO (no)

II = Ii' + ql
30 NACATMIZSI

In these equations the enthalpy was broken up from the chemical

energy Q0(n0) - the heat producing capacity of the initial sub-
stance - and ql the heat producing capacity of the incomplete combus-
tion at point Xl, the amount of unreacted substance at this point
being determined from condition (30), so that it should give a velocity
of reaction balancing the losses _ = 1. C

r-

For briefness the notation (the sign of D follows from equa-

tions (9) and (10) and xI < 0 in the chosen system) is introduced:

- D--JO G dx = g >0

(32)

-v Fdx=f>O
0_0 xl

When it is assumed that the heat capacity of the fresh mixture is

greater both as compared to the physical heat of the mixture at the
initial temperature and as compared with the losses,

f' g, ql, I0 _ QO (35)

It is foun_ that on the assumption of a constant heat capacity,

for the deviation of the velocity from that computed from the
classical theory in the absence of losses

AD i f i g I ql
(34)
D 2(kl2-1) Q0 2 Q0 2 Q0

where kI is the exponent of the Poisson adiabatic at point x I.

6. CONDITIONS FOR THE PROCESS OF THE CH_ICAL REACTION

The start of the reaction in the detonation wave is connected

with the compression and the heatlng of the gas by the shock wave
(the Jump A-B, fig. 1 or 5). The conditions for the process of the
chemical reaction accompanied by a change in state more or less closely
following the equation of the straight line of Todes shall be
considered.
 NACATM 1261 31

The conditions in the detonation wave shall be compared with the

conditions in the quiet "deflagrating" combustion. In the second
case, the velocity of the reaction in the initial state is negligibly
(_g small; the start of the reaction is connected with the heating up of
C)
o] the mixture through heat conductivity by the products of combustion.
Practically the reaction as a whole proceeds in a zone of tempera-
tures very near the temperature of combustion. To the molecular mech-
anism of the processes of transfer of heat and diffusion in the gases
corresponds the composition of the reacting mixture approaching the
composition of the products of combustion (reference 18). In a deto-
nation wave, however, the substance by rapid compression reaches a
temperature at which energetic chemical reaction proceeds without
change in composition.

Preliminary computations on the assumption of constant heat

capacity, absence of dissociation, and so forth, and other simplifi-
cations show that in the shock wave propagating with a velocity equal
to the velocity of detonation (point B, fig. I or 5) the density of
the gas is six times as great as the initial, the pressure twice as
great as the pressure at the instant of the end of the reaction
(point C, fig. I or 5), and four times as large as the explosion
pressure in a closed vessel. The temperature is very near (for a
reaction without change in the number of molecules) to the combustion
temperature of a given mixture at constant pressure. For the numeri-
cal computations of the temperature see Wendtlandt (reference I0).

As has already been shown, the continued heat exchange and diffu-
sion in the detonation wave can be neglected so that the chemlcal read
tion proceeds almost adiabatically. In the case of an autocatalytic
reaction, the absence of diffusion of the catalyzing products may con-
slderably hinder the process of the reaction in the detonation wave.
When the reaction proceeds along the straight line of Todes, the giving
off of the heat is associated with a considerable rise in the
temperature.

It is of interest that the maximum temperatures on the straight

llne of Todes is somewhat displaced to the left (fig. 1 or 5) rela-
tive to the point of tangency C so that between the maximum and the
point C there is a "paradoxical" region in which the process of the
exothermal reaction and the liberation of heat are accompanied by the
increase in temperature due to the simultaneous expansion of the sub-
stance. The liberation of heat in this region is accompanied by an
increase in entropy. Preliminary computations give an increase in the
maximum temperature above the temperature at the point C of the
order of 50 ° to i00 o at the end of the reaction.

The integration of equations herein in the region of cooling and

slowing down of the products of reaction leads to the approach of the
32 NACA_ 1261

integral curve toward a final point on _he side of v > v0, to which
corresponds in the steady regime, according to the equation of the
conservation of matter, a change in sign of the mass velocity u < 0
in qualitative agreement with the test results of Dixson (reference 281

Further it may be shown that the point xI itself (point C,

fig. 5), at which there is intersection with the line M = H the con-
dition of Jouguet equation (29), is satisfied and at the same time
= 1. Equation (30) does not correspond to any special peculiarities
in the curves of pressure, densities, and other magnitudes. The space
distribution of the various magnitudes entering the equations is shown
in figure 6.

The conditions for the process of the reaction are very near the
conditions of adiabatic explosion.

As is known, the time to develop the explosion for all reactions,

the rate of which increases with the temperature, and all the more,
for autocatalytic reactions, is determined essentially by the minimum
rate of the chemical reaction at the lowest initial temperature
(references 37 and 38). 8

T - exp (35)

In the expression (34), the first two terms are determined by the
total time of the reaction in the interval 0 - Xl or in the
p,v-plane from point B to the intersection with the llne M = H.
The last term is determined by the velocity of the reaction at the
point xI on the intersection of the line M = H in the p,v-plane,
that is, at a temperature considerably higher than the temperature at
the instant of compression, x = 0, point B.

In the absence of special reasons, for a reaction the velocity

of which increases with the temperature, the first two terms of
expression (34) - the losses in friction and heat transfer during
the process of the reaction, determined essentially by the conditions
at the start of the reaction, at point B - are considerably greater
than the last term, which gives the incomplete combustion in the wave.
The reverse can be the case only in systems in which the chemical
reaction falls into two stages, the second of which is considerably
slower - even at a higher temperature than the first. Examples of

8 In a quiet flame propagation, the velocity of the flame is deter-

mined by the rate of the chemical reaction at a temperature near the
maximum temperature of combustion. The zone of low temperature and
small rate of reaction is lengthened by the action of the heat
conductivity.
 NACATM 1261 33

such kind are mixtures of diazine and oxygen where the formation of
carbon dioxide occurs considerably more rapidly than its oxidation
(Dixson (reference 29)), the decomposition of nitroethers where in
the first stage there is easily former nitrogen oxide (Berthelot
O (reference (34)), Appin and Belayer (reference 39). In this case,
O_
the second slow stage may preceed as a whole in the zone x < x1
and _ < 1 and show no effect on the velocity of the detonation.

In the general case of a single reaction the velocity of which

increases with increase in the temperature, this is not so and the
fundamental losses are determined by the composition of the substance,
the velocity of its motion, and the time of reaction near point B
(fig. 5). Using the approximate relation for the detonation and shock
wave for a mixture with large heat content, there is obtained

AD 2 + 2_k12 - 2_ DT E/RT B
--=- e (36)
D (kz + i)2

where TB is the temperature at point B, describing the shock com-

pression of the gas wave propagated with the velocity D, E is the
heat of activation of the reaction, K1 and K2 are the exponents
of the adiabatics of Poisson at points C and B respectively;
for the meaning of the other symbols see section 5.

When account Is taken of the dependence of the temperature TB

on the detonation velocity (in the limit for a heat of reaction con-
siderably exceeding the initial heat content of the mixture the last
condition (33)),

TB ,- D 2 [
(37)
]
there is easily obtained the transcendental equation describing the
limit of propagation of the detonation arising from the effect of
the losses.

DO _. exp

(the magnitudes with the subscript zero are computed with the
absence of losses)
54 NACATM1261

RTB°
ADIim = " D (39)
E

2
(_ +ak 2 - 2a), D0To 2E
_llm = -- - _ - (40) c
(k I + 1) 2 d ekTB °

wDere TO is the time of the adiabatic reaction (the period of

induction of the adiabatic explosion) for the initial temperature
TB °, and kI and k2 are the ratios of the specific heats (expo-
nents of the adlabatics) at points C and B (fig. i or 5); for
definitions a, I , and d see equations (15) and (16).

7. COMPARISON WITH TEST RESULTS

The question of the deviation of the observed velocity of the

reaction from the va]ue computed has often been raised in the last
I0 to 15 years. A sharp drop in the velocity near the limit of
Wendlandt (reference I0) (see in detail his data that follow and
fig. 4) is explained by the fact that the heat given off in the
wave is less than the total heat of reaction.

Lewis and Friauf (reference 35) compared the velocity of deto-

nation in an explosive gas to which was added argon and helium. For
the same addition, the computed temperatures of detonation, the
pressure, and all dissociations were the same. The computed deto-
nation velocity is in the ratio of the square root of the density
and therefore in mixtures with large amounts of helium is twice as
large as in the same mixtures with argon.

Experiment shows in mixtures with argon that the velocity of

propagation iS nearer to the value computed; the product of the
velocity by the square root of the density (the molecular weight)
is greater than in mixtures with helium, a fact that indicates
smiler losses in mixtures with smaller velocity of detonation.

The authors state that at a greater velocity of the wave, the

chemical reaction does not succeed in entering the wave front and
the incomplete combustion decreases the velocity of the detonation.
There is here clearly seen the unsatisfactoriness of all considera-
tions of such kind. The velocity of the chemical reaction is of an
entirely different dimension than the velocity of detonation. The
statement that the velocity of detonation Is greater than the velocity
 NACA TM 12_I 35

of the chemical reaction is without meaning. From these trivial

considerations, it is already seen that without introducing any new
time intervals (or lengths -- in the presence of the velocity of deto-
nation of the dimensions of a length time this is e_uivalent), it is
0 impossible to describe incomplete combustion, the deviation of the
0")
measured velocity from the computed.

In the theory herein such a time is introduced through the losses-

the time of retardation, the tLme of heat transfer.

It is the ratio of these times with the times of the chemical

reaction that determines the relative losses." For sufficiently
increased friction and heat transfer, it may be imagined that even
the slowest reactions, for example, the combustion of dust will
have time to enter the detonation wave and can lead to a detonation.
Greater losses In mixtures in which the detonation velocity is
greater are associated with the intensification of the turbulent heat
exchange and friction on increasing the velocity of motion propor-
tlonal to D.

In the literature there is a complete absence of mention of

another type of losses; namely, the friction and heat losses during
the chemical reaction (the magnitudes f and g of equations (32)
to (54) in which the incomplete combustion is denoted by ql)"
As has already been shown, it is these losses, depending on the
minimum rate of the chemical reaction to the total time of the
process, that are the most essential.

In table 2, are presented computations according to Wendlandt

(reference I0), of the times of reaction in the detonation wave
propagated in a hydrogen-air mixture. It is assumed in equation (36)
that _ = 0.02, _ = 0.25, kl = 1.3, k2 = 1.4 and the losses in
_ncomplete combustion are neglected so that

D/D = 0.0O8DT/d (41)

In the last column are given the heats of activation computed

from the relation of the time of reaction and temperature (for each
neighboring pair of points). (See equation (35).) If it is
assumed, along with Wendlandt, that the last mixture (18.5 percent)
]ies at the limit, then equation (59)

AD = kTD
2E

will give

E = 5000
36 NACA TM 1261

The values of the heats of activation of about 5000 to I0,000

due to the strong drop of the velocity of detonation to 75 percent
of the computed value as observed by Wendlandt are entirely
inadmissible.
UD
(D
_O
In a later work of Breton (reference 40), the velocity at the
limit behaves very differently in different mixtures. In certain
cases the velocity drops markedly and in other cases remains with-
out change (hydrogen - oxygen mixtures). Finally for amonia with
oxygen according to Breton the velocity at the limit increases
although actually the scatter of the test points Is so large that
it may equally well be'asserted that the velocity drops.

Thus the detonation picture is essentially unclear. Breton

associates any decrease in velocity as a function of the diameter
with spin and notes that in his tests with mixtures near the limit
that spin always occurs; as a result the conclusions herein are not
directly applicable.

The chronoelectrical method of Wendlandt (reference i0) does

not permit observation of the structure of the wave.

Livelng and Dewar (reference 41) observed the reversal of the

red llne of cadmium in the spectrum of the detonation propagated
parallel to the slit of the spectrograph.

The conclusion drawn by the authors as to the temperature

gradient in the wave front entirely agrees with the views pre-
sented herein. Here too, however, it is not clear whether the
effect observed by these authors is connected with the reflec-
tion of the wave at the window, which covers the pipe (reference 29).

According to the present views, the pressure, density, and

velocity in the detonation wave drop as the chemical process pro-
ceeds. The slower the chemical reaction, the more slowly does this
drop take place and the wider the zone of increased pressure and
increased velocity. A wave in which the reaction proceeds more
slowly has a greater store of energy in correspondence with the
fact that it is more difficult to produce it.

These considerations are evidently confirmed by the results

of Rivin and Sokollk (reference 36) according to which the mix-
ture of carbon monoxide and oxygen (with a small addition of an
explosive gas) produces detonation no worse if not better than an
explosive gas and a mixture of ethane with oxygen.
 NACA TM 1261 37

8. PROBL}_3 IN THE DEVELOPMENT OF THE THEORY

In the theory developed, the pressure and the velocity of

C_
motion in an explosion with chemical reaction (shock wave, point B,
O fig. 1 or 5) forming the front of the detonation wave are higher
O_
than the values assumed for detonation, that is, at the instant when
(neglecting losses) the chemical reaction has ended (point C, fig. 1
or 5, conditions in equations (6), (29), (fig. S)).

Can the detonation wave (C) 'push' before it a shock wave (C)
of greater pressure? The fact that this is possible is confirmed
by the example of deflagratlon (quiet burning) where the products
of combustion in expanding push ahead of them the still unburned
mixture, the pressure of the products of combustion being lower than
the pressure of the unburned mixture°

It is noted that the relation between B and C is precisely

the same as between the initial point A and the products of deflag-
ration on the branch KD. Detonation is no other than deflagration
of a shock-compressed gas heated to a temperature higher than the
temperature of self-ignltion.

The impulse of the pressure together with the flow of momentum

are likewise conserved in a detonation wave of the structure com-
puted herein and there is also conserved the sum p + o(D-u) 2 in the
shock wave sustained by the motion of the piston where each component
separately is constant.

The stability of such a shock wave with respect to small dis-

turbances is beyond doubt. At the end of the chemical reaction the
detonation naturally dies down but the shock wave also changes
together with other properties.

The first factor favorable for the stability of the wave is the
fact that as a result of the drop in the velocity of sound below the
velocity of propagation there is attained an independence of the con-
ditions behind the wave. On the other hand, when small disturbances
are imposed the velocity of the chemical reaction is likewise sub-
Ject to variation. It is, at present, impossible to predict the
result of the computation of the stability with respect to small dis-
turbances, which may depend also on the special properties of the
chemical kinetics (autocatalysis, heat of activation).

In any case the entire chemical reaction, which determines

the velocity of the detonation, proceeds in a region where the
38 NACATM1261

velocity of sound is greater than the velocity of the detonation

(the velocity is equal only at the end of the reactlon). 9
G
In considering the effect of the losses on the velocity of
the propagation, the discussion herein is restricted to the first
approximation. In the zeroth approximation in the absence of
losses, the state of the system varies according tc the equation
of the straight line of Todes. In the following first approxima-
tion, the losses and their effect on the velocity in varying the
system from the zero approximation were found. For a system wlth
a knownkinetics of the chemical reaction, the effect of the
losses, the limit of the propagation of the detomation, and so
forth, should be found by the first integration of equations (21)
and (22): the value of D may be found (because the limit had
been exceededand the steady regime exists) by selection. The
equations, with account taken of the dependenceof the kinetics
on the temperature, assures below the limit the absence of solu-
tions satisfying the boundary conditions.

It is of significance that In the absence of a reaction,

even if the analogy of Reynolds (a = l, see equations (15) and
(16)) !s satisfied, the relation of the heat transfer and fric-
tion is such that in the steady regime of a definite velocity
after compression in the shock wave there is an additional heat_
ing of the gas. This effect _s still further increased in a
rough plpe in which the friction increases more rapidly than the
heat transfer.

The taking into account of this additional heating may

somewhatwiden the limits of detonation.

It might also be possible, for greater accuz'acy, to take

account of the increased heat transfer and friction at the
start of the motion as comparedwith the stabilized velocity
profile and temperature. But here the limits of application
of the theorj have alread_ been exceeded. As soon as the pro-
file (distributions over the radius) Is considered, the one-
limensional theory is no longer possible.

9 Jost (references 12 and 32) supposes that in a normal (that

is, the stationary wave considered herein) detonation the
velocity of sound is exactly equal to the velocity of the deto-
nation and from this follows the instability of the normal deto-
nation, its conversion into a periodic regime.
 NACA%]4i_61 39

A detailed analysis of the conditions of heat transfer and

friction leads to the following conclusions. The Reynolds num-
ber of the flow is very large so that in the stabilized flow
O the resistance coefficient and the character of the flow may be
OD
considered as practically independent of the value of the ReFn-
olds n_ber. The length of the stabilization distance, however,
still depends very essentially on the Reynolds number. The
rapid chemical reactions evidently proceed over a distance far
from sufficient for stabilizing the flow. At the limit of deto-
nation, the reaction proceeds over a length of the same order as
the length Z over which turbulence of the boundary layer occurs
corresponding to the limiting value of the Reynolds number, formed
with this length:

Re z _ Zu _ constant ~5. i05 (42)

Returning to the Reynolds number formed from the diameter of the

pipe,
5
z : 5"10 • d/Re (43)
In view of the fact that this boundary limits the applica-
tion of the developed theory, in particular for determining the
effect of the losses on the velocity of detonation, further
investigations are required.

9. THE QU_TION OF SPIN IN DETONATION

Experiment indicates the existence of peculiar three-

dimensional states of the propagation of the so-called spin
(rotation) of a detonation wave in which the instantaneous dis-
tribution depends also on the angle in the cylindrical system
of coordinates, coaxial with the pipe, notwithstanding the com-
plete symmetry of the initial conditions.

At the present time, the fundamental reasons for such rota-

tion are entirely unclear.

Moreover, there have not been set up with sufficient accu-

racy those experimental conditions under which spin arises, in
particular the relation between the spin and the limit of deto-
nation. On the basis of the analytical investigation of the
simplest one-dimenslonal theory given herein, it shall be
attempted, by the methods of the theory of dimensions, to
explain what magnitudes will enter any future accurate theory
40 NACATM1261

of spin and limit of detonation. First of all, from the conditions

of agreement with the classical theory as the limit case in the
absence of losses and rapid reaction, it is evident that the velo-
city of propagation of detonation without losses D and such non- _o
0
_D
dimensional parameters as the ratio of this velocity to the velocity
of sound in the initial state, the ratio of the pressures and volumes
before and after the reaction will evidently enter. All these nondi-
mensional parameters vary relatively little, at least if the initial
temperature of the mixture, which is approximately the room tempera-
ture, is not varied within wide limits.
The final velocity of the chemical reaction is introduced. It
may be supposedthat the type of the chemical kinetics (autocatalysis,
or the classical reaction of a certain order) is in a certain sense
unessential; autocatalysis changes the absolute magnitude of the period
of induction and makes it dependon small amountsof additions to the
initial mixture but the shape of the kinetic curve is almost unchanged
because even in the classical reaction with considerable heat of reac-
tion there is observed a considerable scatter in connection with the
increase in temperature.

The time of the chemical reaction which, together with the velo-
city of the detonation gives the characteristic length DT of the
process of the chemical reaction appeared as a characteristic. Direct
comparison of this length with the diameter of the pipe cannot be made
because all the motion occurs along the axis of the pipe. For this
reason, it may be expected that there enters the characteristic dis-
tance of friction and cooling the magnitude d_ where _ is the
nondimensional coefficient of resistance.lO

At those Reynolds numbers, about I0 S, which are usually attained

in a detonation wave, the coefficient of resistance may be considered
as practically constant so that it would appear that in a turbulent
region at Re " l0 S the dependenceof the phenomenonon the Reynolds
numbervanishes.

The results of the preliminary computations (equations (41) to

(43)) of the distances over which a stabilized regime is established
and over which the friction and the heat transfer are propagated over
the entire sections, showthe reverse: Whereasin the stabilized flow
the dependenceon the Reynolds numberdrops out, the distance over which
this stabilization occurs dependsvery much on the Reynolds number.

I0 Our coefficient _ characterizing the ratio between the heat trans-

fer and friction evidently does not enter here.
3 NACA TM 1261 41

At large Reynolds numbers, even long before stabilization, a tur-

bulent boundary layer occurs at the distance 5.105 d/Re.

O The Reynolds number is a very essential factor particularly

g)
for the two-and three-dimensional theory, which includes the consid-
eration of spin. In the one-dimensional theory, the stabilization
distance differs only in the somewhat increased coefficient of resis-
tance, whereas actually (both in the two-and three-dimensional theory),
it is over this distance, where the most energetic chemical reaction
occurs, that the very character (laminar or turbulent) of the flow
changes. Thus from the conditions at the point of stabilization
Re_ = 5.10 5 there enters the length Z not depending on the dia-
meter Z _ 5"10 5 v/u or 5"10 5 U/D.

Thus three magnitudes of the dimensions of length essential for

the process; namely, the "chemical distance " DT. stabilized fric-
tion d/_, and "turbulence formation" 5.10 5 u/D are concerned.

From these three lengths, two nondimensio_al criter_la can be

constructed, for example, Re = dD/_ and DT _/d ("the criterion
of Rivin and Sokollk").

The requirement of at least two criteria for the description

of the phenomena of spin and limit of detonation can evidently be
established from a consideration of even the relatively meager experi-
mental data presently avallable.

According to the views of Rlvln and Shelkln if a wider interval

of pressures and diameters (that is, values of the Reynolds number)
than the one dealt with by Breton (reference 40) is considered, the
connection between the spin and the limit ceases to be unique. Such
unique relation would necessary follow from the theory of Rivin-
Sokolik with only the single criterion DT _ /d in which only the
width of the wave front is fixed replacing it by the diameter of the
pipe d or the distance of resistance d/_ = dr(Re). It might be
supposed that on increasing the criterion of Rivin-Sokolik, the
classical picture with the narrow plane wave front and velocity, not
differing from the one computed is passed from to the appearance of
spln - for one value of the criterion and limit of detonation - at a
stlll larger value of DT _/d. Such simple behavior is actually
observed in a series of tests in pipes of constant diameter at a con-
stant pressure (Breton), that is, at practically constant value of
the Reynolds number.
42 NACATM1261

Here, the point is madethat the increase in the criterion

DT _/d brings about the appearance of spin.

If, however, starting from a certain mixture at atmospheric _D

0
pressure in a pipe of average diameter (18 to 25 mm)the pressure _0

is lowered without changing the diameter and the composition, it

will evidently be possible in a numberof cases to attain the limit
without observing2pinll; a drop in pressure meansan increase in
the criterion DT_/d and a drop In the Reynolds number. In the
case of a mixture, however, in which spin is observed under the
usual conditions, an increase in the diameter does not lead to the
disappearance of the spin notwithstanding the decrease in the cri-
terion DT _/d. Instead of the disappearance of the spin in wide
pipes, there is observed the appearance of certain spirals (refer-
ence 42). Thus, it may evidently be concluded that the appearance
of spin considerably favors an increase In the Reynolds number.

The data used herein for the comparison of the present theory
with experiment is incommensurablysmall in comparison with all the
material accumulated for the last 80 years since the discovery of
detonation. The fault is partly ascribed to the theory itself, which
is extremely simplified and which does not describe the special phe-
nomenonsuch as spin.

This characterizes the style of the experimental work. Up to

the most recent times, the hydrodynamic theory of the velocity of
detonation, excellently confirmed by test, did not arouse a need for
seeking the mechanismof the chemical reaction and an investigation
of the conditions at the detonation wave front. If the present work
gives rise to newexperimental investigations, which penetrate more
deeply into the nature of the phenomenon,the object will have been
attained.

i0. SUMMARY

i. The classical theory of the velocity of detonation Is presented.

The considerations of Jouguet and Becker leading to the exclusion of
a number of possible states of the products of combustion correspon-
ding to a greater velocity of detonationare not convincing, as can
be proven by an experiment carried out in imagination.

Z. The views of various authors on the mechanism of the process

of the chemical reaction in the detonation wave are considered. It

Ii
From a personal communication with Rivin and Shelkin.
 NACATM1261 43

is shown that the diffusion of active particles and the heating by

thermal conductivity cannot play a significant part in the propaga-
tion of detonation. The ignition, the start of chemical reaction,
O
produces an instantaneous compression of the substance with an
O_
associated rise in temperature.

S. The process of the change of the specific volume in the

detonation wave and pressure corresponding to these views is con-
sidered. The exclusion of the states mentioned in item 1 and the
choice of a definite value of the velocity are consequences of the
mechanism described in item 2 of the start of the chemical reac-
tion and the conservation equations leading (Todes and Ismailov)
to a linear relation between the pressure and the specific volume
in the absence of losses.

4. The equations of the process of the chemical reaction, the

friction, and the heat transfer in a steadily propagated detona-
tion wave are set up. For a rapid chemical reaction, the equations
give agreement with the classical value of the detonation velocity.
In the succeeding approximations of the equations, the effect of the
losses on the velocity and the limit of detonation are described.

5. The developed one-dimensional theory is compared with the

test data available and the special features of the theory and the
limits of its applicability are pointed out.

6. Considerations are presented with regard to the descrip-

tion of the experimentally observed phenomenon of spin (spiral
propagation) in nondimensional criteria of similitude.

Translated by S. Reiss

National Advisory Committee for Aeronautics

 44 NACA TM 1261

REFERENCES

I.
Chapman, D. L.: Phil. Mag. and Jour. Sci., vol 47, ser. 5,
1899, p. .90.
_D
O
_9
2. Schuster, A.: Phil. Trans. Roy. S.c. London, vol. 152, 1893.

3. Jouguet, E.: Mechanlque des explosifs. Paris, 1917.

4. Crussard, L.: Bull. de S.c. l'Industrie Miner. S. Etienee,

6, 1 - 109, 1907.

.
Riemann, Bernard: Ueber die Fortpflanzung ebener Luftwellen
yon endlicher Schwingungsweite. Ges. Math. Werke, Aufl. 2,
1892, S. 159-181.

.
Hugoniot, H.: Memoire sur la propagation du mouvement dans
les corps et sp6cialement dans les gas parfaits. Premiere
partie. Jour. de L'_cole Polytechnique, Cahier LVII, 1887,
P. 5-97; Deuxieme Partie, Cahier LVIII, 1888, P. 1-125.

7.
Rayleigh: Aerial Plane Waves of Finite Amplitude. Proc. Roy.
S.c. London, vol. I/XXIV, no. A84, ser. A, July 8, 1910, pp.
247-284.

o
Rankine, W. J. Macquorn: On the Thermodynamic Theory of Waves
of Finite Longitudinal Disturbance. Phil. Trans. Roy. S.c.
London, vol. 180, 1870, pp. 277-288.

9. Zeldovich, Y. B., and Ratner, S. B.: Jour. Exper. and Theor.

Phys.

I0. Wendlandt, Rudolf: Experimental Investigations Concerning the

Limits of Detonation in Gaseous Mixtures. Part I. NACA TM
553, 1950; Part II, NACA TM 554, 1930.

ii. Becker, R.: Impact Waves and Detonation. Part I. NACA TM

505, 1929; Part II, NACA TM 506, 1929.

12. Jost, W.: Explosions - und Verbrennungsvorg_nge in Gasen.

Berlin, 1939.

15. Lewis, Bernard: A Chain Reaction Theory of the Rate of Ex-

plosion in Detonating Gas Mixtures. Jour. Chem. S.c.,
vol. 52, no. 8, Aug. 1930, pp. 3120-3127.
 NACA TM 1261 45

14. Anon.: Collection of papers on gas explosions. Avioavtoisdat,

1932, p. 215. (In Russian.)

/ I

15. Mallard, et Le Chatelier: Recherches experimentales et theo-

/
O rlques sur la combustion des melanges gazeux explosifs.
C_
/
Ann. des Mines, T. IV, ser. 8, 1883, P. 274-568.

tf
16. Nusselt, Wilhelm: Die Zundgeschwindigkeit Brennbarer Gasgemische.
VDI Zeitschr., Bd. 59, Nr. 43, Okt. 23, 1915, S. 872-878.

17. Daniells, P. J.: The Theory of Flame Motion. Proc. Roy. Soc.
London, vol. 126, no. 802, ser. A, Feb. 3, 1930, pp. 593-405.

18. Zeldovich, Y. B., and Frank-Kamenetsky, D. A.: Jour. Phys.

Chem. (SSSR), vol. 12, 1938, p. 100. Acta Physiochlm., vol.
9. 1938, p. 341.

19. Prandtl, L.: Zur Theorle des Verdichtungsstosses. Zeitschr.

f. Ges. Turbinewesen, Jahrg. III, Heft 16, Juni 10, 1906,
S. 241-245.

20. Ismailov, S., and Todes, 0.: Paper presented at NTS LIKBF,
1934.

21. Nernst, W.: Theoret. Chem., 8 - I0 Aufl., 769, 1931.

/!

22. Van't Hoff, J. H. : Vorlesungen uber theor, u. phys. Chem.,

2 Aufl., 1901, 245.

23 Falk, K. G.: The Ignition Temperatures of Hydrogen-0xygen

Mixtures. Jour. Am. Chem. Soc., vol. XXVIII, no. ll, Nov.
1906, pp. 1517-1534.

24. Falk, K. George: The Ignition Temperatures of Gaseous

Mixtures. Jour. Am. Chem. Soc., vol. XXIX, no. ll, Nov.
1907, pp. 153611557.

25. Sokolik, A. S.: Jour. Phys. Chem. (SSSR), vol. 13, 1939.

26. Bone, WilllamA., Fraser, Reginald P., and Wheeler, William

H.: A Photographic Investigation of Flame Movements in
Gaseous Explosions. Part VII - The Phenomenon of Spln in
Detonation. Phll Trans. Roy. Soc. London, vol. 235, ser.
A, 1936, pp. 29-68.

27. Vieille, P.: Comptes Rendus, 131, 413, 1900.

46 NACATM1261

28. Campbell, C., and Finck, G.: Jour. Chem. Soc., 1927, p. 1572.

29. Dixon, H.: Phil. Trans. Roy. Soc. London, vol. 200, 1903,
p. 315. (o
G
_o
30. Shchelkin, K. I.: Jour. Theor. Phys. (SSSR), T. IV, 1934,
p. 730.

31. Becker, R.: Zeitschr. f. Elektrochem., Bd. 42, 1936, S. 457.

32. Jost, W.: Zeitschr. f. phys. Chem., Bd. 42, Abt. B, 136, 1939,
S. 136.

33. Avanesov, D., and Rukln: Works, Leningrad Chim.-Tekhn. Instr.,

No. 7, 1939.

34. Berthelot, M.: Sur la force dee mati_res explosives. Paris,

1883.

35. Lewis, Bernard, and Friauf, James B.: Explosions in Detonating

Gas Mixtures. I - Calculation of Rates of Explosions in
Mixtures of Hydrogen and Oxygen and the Influence of Rare
Gases. Jour. &m. Chem. Soc., vol. 52, no. 10, Oct. 1930, pp.
3905-3920.

36. Rivln, M., et Sokolik, A.: Acta Physiochim., vol. 7, 19:57,

p. 825.

17. Semenov, N. N.: Chain Reactions, 1935.

38. Todes, 0. M.: Jour. Phys. Chem. (SSSR).

39. Appin, A. Y., and Belayer, A. F.: Laboratory W IKHF, unpublished

work.

40. Breton, M.: Theses, Univ. de Nancy (Paris), 1936.

41. Dewar, J., and Liveing, G. D.: Proc. Roy. Soc. London, vol.36,
p. 471.

42. Bone, William D., and Fraser, Reginald P.: Photographic Inves-
tigation of Flame Movements in Gaseous Explosions. Parts IV,
V, and VI. Phil. Trans. Roy. Soc. London, vol. CCXXX, ser. A,
1932, pp. 363-384.

43. Scorah, R. L.: On the Thermcdynamic Theory of Detonation.

Jour. Chem. Phys., vol. 3, no. 7, July 1935, pp. 425-430.
 NACA TM 1261 47

TABLE I

State Composition v Velo-

O
0_ of city
gab of
propa-
gation

Initial i i0 ° 0 = 283 ° K ii
1 02 + 3 1 N2

Detonation
wave 12.87 0.575 2019°C = 2292°K 1659
H2 0+3_ 1 N2

Shock wave 12.13 0.241 555 ° C = 8280 K 1210

1 02 + 3 1 N2

TABLE II

TT E
B2 Dcomp Dmeas Tcomp from
(per- (m/seC) (mlsec) (,ec) DmeaB (ca i/mole )
cent) (OK)

19.6 1686 1620 0.55.10 -4 1290 20.000

18.8 1645 1475 1.75.10 .4 1120 9.000

18.5 1635 1300 3.94- i0 "4 930 5.000

18.3 1625 1207 5.30-10 -4 840

48 NACATM 1261

E
o_
O
o)

K
Po
AI D
I
I M
Vo V

Figurel.

p4

Figure 2.
NACA _M 1261 49

Figure S.

1900

1700

1500

1300 900
O

_D

ii00

9OO

70O

500

500

0 i0 i5 20 25

H 2, percent
Figure 4.
50 NACA_4 1261

/M =H

B_ 2 I //

M<H 3_/__ =l

///// M>E k

//
/
/

M=H
, Ae

_C
) 1

Figure 5.

Po

_T To

Po

X=X X =o X

Figure 8.

NACA-Langley - 11-3-50 - 1100