Combustion, Ezplosion, and Shock Waves, Vol. 33, No.

2, 1997

FORMATION AND COMBUSTION OF GAS CLOUDS

IN A C C I D E N T A L D I S C H A R G E T O T H E A T M O S P H E R E

G. M. Makhviladze, J. P. ttoberts, x and S. E. Yakush UDC 532.525.2-t-536.46

The formation and combustion of a fire-hazardous cloud arising in accidental discharge of

a flammable gas is analyzed. A criterion is obtained which makes it possible to identify
instantaneous, continuous, and intermediate types of discharge from the known physicochemicaI
properties of the material, vessel and hole configurations, and storage conditions. Relations for
the critical hole dimensions corresponding to the boundaries between the different discharge
regimes are given. Ranges of parameters for which ignition of the outflowing gas can give rise
to a flame or a fireball are established. The most probable limits for the coefficient of fuel
participation in the burning fireball for various ignition delays are determined. The results are
compared with the available ezperimental data.

INTRODUCTION

Accidental discharge of fire- and explosion-hazardous substances to the atmosphere is among the main
hazards in modern chemical industry [1, 2]. The typical causes of such discharge are depressurization or
failure of industrial vessels, pipelines, and transportation and storage facilities. Ignition of the resulting fuel-
air mixture can proceed in both an explosion regime (a shock wave is the major hazard in this case) and a slow
combustion regime, in which the thermal radiation from the flame is the major damaging factor. Realization
of one or the other regime of cloud combustion depends on the ambient conditions (for example, on the degree
of blockage and limitedness of space) and on ignition conditions (ignition delay, energy input in initiation of
the reaction, etc.).
At present, a number of engineering procedures for estimating the chemical process risk have been
developed (see, e.g., [3-5]). These procedures are based on schematization of physical processes proceeding in
one or another course of accident, and on the description of each process by simple mathematical models
invoking empirical information for the choice of resulting constants. As applied to the consequences of
depressurization of vessels with compressed gas, the following two limiting cases are usually considered:
- jet (quasi-continuous) outflow from a small hole in the casing of a vessel,
-

- instantaneous release of the entire mass of the gas with complete damage to the vessel.
-

Ignition of the gas discharged from a source located near the site of accident gives rise a burning flame in the
first case and to a fireball in the second case.
At present, procedures for calculating the gas discharge rate, flame geometry, thermal fluxes, and critical
distances corresponding to various degrees of damage have been developed for each of the above-mentioned
cases. The intermediate case of fuel discharge has been much less studied, although it is not less important.
The criterion by which gas discharge can be identified as the instantaneous or the continuous type of discharge

Institute of Problems of Mechanics, Moscow 117526; the present site is the University of Central
Lancashire, Preston PR12HE, UK. 1University of Central Lancashire, Preston PR1 2HE, UK. Translated
from Fizika Goreniya i Vzryva, Vol. 33, No. 2, pp. 23-38, March-April, 1997. Original article submitted
October 10, 1995; revision submitted November 10, 1996.
144 0010-5082/97/3302-0144 $18.00 Q 1997 Plenum Publishing Corporation
 has been purely subjective so far. For discharge of finite duration, there are no estimates of how mixing of the
outflowing gas with air can decrease the coefficient of fuel participation in fireball combustion with variation
of the ignition delay. At the same time, it is clear that, to use calculation techniques correctly in evaluating
hazard, it is necessary to establish qualitative criteria for determining flow type and flame configuration.
The present paper proposes a criterion for classification of types of gas discharge of finite duration.
Based on the geometry of vessel and hole and also on the properties of materials and storage conditions,
the criterion makes it possible to determine the type of discharge to the atmosphere [a jet, a cloud, or an
intermediate type ("cloud-like')], and to assess ignition delays for which one can expect gas combustion in
the form of a flame or a fireball. In the latter case, the coefficient of fuel participation was determined as
a function of discharge parameters and ignition delay. Final formulas are obtained for subcritical (at a low
internal pressure) gas discharge, and also for critical discharge at a constant high internal pressure (for example
in discharge from pipelines) and at an internal pressure decreasing with time (for example, for discharge of a
finite-volume vessel).

C L A S S I F I C A T I O N OF F I N I T E - D U R A T I O N D I S C H A R G E

We consider a flammable gas discharging to the atmosphere through a round hole with diameter d at
velocity Ub for time tr. If the discharge duration exceeds considerably the time of turbulent mixing of the gas
with ambient air tmix, a quasi-stationary jet forms in the atmosphere, and ignition of this jet leads to the
formation of a burning flame. Conversely, when the discharge proceeds almost instantaneously (i.e., over time
tr << tmix), a fuel-enriched cloud arises in the atmosphere, which burns in the diffusion regime as a fireball.
A number of intermediate cases which correspond to different ratios of the characteristic times tr and tmix are
observed between these two limiting cases.
To obtain a quantitative criterion for classifying finite-duration discharges, we consider two typical
types of discharge upon depressurization of gas vessels: a cloud of finite mass which has a an initial directed
pulse and a propagating nonstationary jet which is generated upon sudden switching of a mass and momentum
source. It is assumed that the buoyancy forces are negligibly small (purely inertial flow). In addition, the effect
of wind is not taken into account, and the factors limiting discharge are not considered. In each case, the
time during which the volume concentration of the discharged gas decreases to the upper concentration limit
of combustion is used as the characteristic time of mixing. This gives a temporal scale of mixing processes
which make the gas mixture combustible. The motion of the gas is considered axisymmetric; in addition, the
time of attainment of self-similar velocity and concentration distributions is considered much smaller than the
characteristic time of dilution of the gas to the upper concentration limit. This is due the fact that in jet flow,
self-similar distributions are established at a distance equal to several diameters of the discharge hole [6, 7],
whereas the gas dilution to the upper concentration limit, as will be shown below, proceeds at significantly
larger distances. Consequently, the self-similar stage takes most of the time of dilution.
The evolution of an instantaneous-discharge cloud can be described by the integral model of [6, 8-
9 10]. The cloud is approximated by a cone whose dimensions increase with motion because of mixing with
ambient air. The dynamics of the cloud and the variation in its volume are determined by integrating the
laws of conservation of mass, energy, and momentum. The values of the entrainment coefficient fl, which is
the average slope of cloud expansion, are, according to [6], in the range of 0.16-0.58, the large spread of the
data being due to the intrinsic irregularity of the turbulent cloud. Below, we use the most typical value of the
entrainment coefficient adopted in [6, 10]: fl ~ 0.25. The increase in the cloud volume due to mixing with air
is satisfactorily described by the formula

--= 8 ~ 2p__~b t
Vo pa
where V and V0 are the initial and current volumes of the cloud, t is time, U is the gas velocity, d is the hole
diameter, and p is the gas density; the subscripts a and b refer to ambient air and the gas at the outlet level,
respectively. Since the total mass of the flammable gas in the cloud is a constant quantity, its volume average

145
gas concentration decreases with an increase in the volume as C = pbYO#a/paV#g, where #a and #g are the
molecular masses of air and the gas. Hence, the time required for gas dilution to the upper concehtration limit
is estimated as
d ( , ~ ) 1 / 3 ( ~ : ) 4 / ~ _4/3
t c = 2ub vurL' (1)

where CUFL is the volume concentration corresponding to the upper concentration limit.
The mixing of the gas with air in an unsteady (propagating) jet with sudden switching of the source can
be described as follows. As is shown by experiments, a propagating jet consists of a leading part followed by
a conical part which is similar to a steady buoyant jet [6, 11]. We approximate the concentration distribution
in the propagating jet by a model by replacing the leading part by a sharply defined front located at distance
h from the virtual source. The concentration ahead of the front is assumed to be zero. Behind the front, we
shall use the concentration field in a stationary axisymmetric jet with a hyperbolic law of decay of the axial
concentration and a Gaussian concentration distribution in the radial direction:

x \pay /.tg
Here Cax is the axial volume concentration and B is an empirical constant whose value, according to various
experimental data, is in the range of 4.0-5.9 [7, 12-16] (below, we use the value B = 4.75 given in [7]). The
slope of jet flow spread, according to the data of [15, 16], is a = 0.10-0.16 (below we shall use the value of
a = 0.132 recommended in [16]).
The dependence of the front coordinate on time can be found from the law of conservation of mass,
by equating the mass of the gas discharged by the source up to time t to the mass of the gas that is present
in the atmosphere and is determined by integration of concentration (2) multiplied by the gas density under
normal conditions Pg,a.
Integration from 0 to cx~ over the radial coordinate and from 0 to h over the axial coordinate gives the
dependence

' (3)

where the value of the coefficient is equal to 2.45 for a = 0.132 and B = 4.75.
It is interesting that the same form of the functional dependence of h(t) can be obtained by using
the integral model of [6, 8, 9], in which the jet is approximated by a cone with height h and the base radius
r = alh. Equating the current m o m e n t u m of the cone 3/2paVU (where V = ~rr2h/3 is the volume of the
cone and U = dh/dt is the front velocity) to the total momentum (Trd2/4)pbU~t produced by the source up to
time t and integrating the resulting differential equation, we obtain relation (3) in which the numerical factor
should be replaced by a~-1/2. The slope of the jet external boundary given in [6, 10, 11] is al = 0.I66-0.25.
This gives constants of 2.0 to 2.4, which are close to the value obtained above.
The characteristic time of mixing in the propagating jet is defined as the time during which the jet
leading edge reaches the point at which the axial concentration is CUFL- From (2) and (3) we obtain

t j = 2 2B 3 (4)

Dividing (1) by (4) we find that the ratio of the characteristic times of mixing
to 1 ( p~~11~( #g ~213o:la

is smaller than unity, since the concentration CUFL is small, as a rule (--~ 0.05-0.2 for typical hydrocarbons).
This means that gas dilution by air in the instantaneous-discharge cloud proceeds more actively and terminates
in a smaller time than in jet flow.

146
 Fig. 1. Classification of finite-duration discharges based on
comparison of discharge time and hole diameter with critical
values: (a) instantaneous discharge with cloud formation;
(b) intermediate (cloud-like) discharge; (c) jet discharge.

The proposed classification of discharge types based on a comparison of the discharge time with the
characteristic time of turbulent mixing can be specified as follows. If tr <~ re, the discharge belongs to the
instantaneous type and leads to the formation of a cloud that is poorly mixed with air. Conversely, for G/> ta,
quasi-continuous (jet) gas flow is realized. Finally, for t c ~< G ~< t j, the intermediate type of discharge occurs,
in which the gas is discharged fairly rapidly, on the one hand, so that a steady jet has no time to form
during the discharge, but, on the other hand, the discharge proceeds rather slowly, so that part of the gas is
mixed with air to concentrations lower than the upper concentration limit. We shall call this type of discharge
"cloud-like" to emphasize the difference from jet flow.
The formulated criterion can be given in a more convenient form for practical application, if we
introduce the total mass of the discharged gas M and express the discharge time as G = 4M/~rd2CclpbUb,
where Cd = 0.6-0.9 is the discharge coefficient that takes into account the features of the discharge hole [3,
4]. Below, we use the value Cd = 0.85 recommended in [17, 18] for gas flows. Equating the time tr to each
of the characteristic times t c and t j and solving the resulting equations for the hole diameter, we find two
critical diameters that correspond to the boundaries between the different types of discharge:
d3c _ 8M ( flg,a "~4/3 ttg g-~4i3
cd p ,o urL. (5)
d3 = 2M ( Pg,a'~ 3/2 ["#g "~3/2~2
t-77{) (6)
Finally, the criterion is written as
tr ~ t c (d >1 de) - - instantaneous discharge with cloud formation,
t c <~ tr <~ t j (dc >>-d >>.d j) - - intermediate (cloud-like) discharge, (7)
tr >~ t j ( d <~ d j) - - jet discharge.
The relationship between the discharge time t~ and the critical times t c and t j, and also the
corresponding relations between the diameter d and the critical diameters dc and dj entering into (7) are
shown schematically in Fig. 1. Note that for noncircular holes, this classification is possible if the effective
diameter d = ,/4/4/4~/r
y , calculated from the surface area S is introduced as a geometrical characteristic.

DISCHARGE FROM LOW- AND HIGH-PRESSURE VESSELS

In formulas (5) and (6), the gas density at the level of the discharge hole (radius rb) depends on the gas
parameters in the vessel, and, hence, requires additional definition. We shall consider two cases: discharge at
low (lower than the critical pressure) vessel pressure and discharge at high pressure. Furthermore, approximate

147
formulas are given which make it possible to analyze discharge of a vessel of finite volume with allowance for
the variability of gas-flow rate.
If the excess pressure in the vessel is small, the flow is essentially isothermic, and the gas density in
the vessel p0 and the density at the outlet differ only slightly from the gas density at atmospheric pressure
Pb. In this case, the critical parameters pg,a and dc take the form
d~ - 8M gg ~4/3 8 ~tg r ,T
CdTrPg,a ga VUFL -- CdTr~a'JUFL v0' (8)

2M /'#g'~a/2 ~ 2 2 (g~)3/2 2
d~r= Cd~rPg,a0~2B3 ~,-'~a) t.~UFL -- CdTrO~2B3 _ _ ~ CUFL~ , (9)
where Vo = M/po ~ M/pg,a is the volume of the gas discharged to the atmosphere under storage conditions.
If the storage pressure exceeds the critical value p. (about twice as high as atmospheric pressure), the
flow proceeds at sonic velocity, and the gas pressure and density at the outlet are
f 2 "~k/(k-1 ) /, 2 ,~l/(k-U (k ~ l)k/(k-l)
Pb = P0t, V ; - 7 ) , Pb = P 0 t y ; - 7 ) for P0 > P, = P~ - - , (10)
where k is the adiabatic exponent for the gas. The discharged-gas pressure becomes equal to atmospheric
pressure in the immediate vicinity of the outlet, after which the flow is essentially isobaric. To take into
account the additional expansion of the gas, we introduce an equivalent hole diameter deq and an equivalent
gas density peq by defining t h e m from the condition of continuity of the mass-flow rate through the true and
equivalent holes, as is done in describing underexpanded jets [15, 18, 19]:

4 PbUb= ~ PeqUb, P rmeq "-~ Pb

Substituting the equivalent diameter deq = d(pb/pa)1/2 and the density pcq for the true quantities with
allowance for relation (10), we obtain the following relations for the critical diameters dc and d j:

d ~ = cd---V___-- FLk 0) Y0, (11)

d3 = 2 ( +___.__.!l
k ) 3/2(k-1) 3/2 2 (p=~1/2

Here V0 = M/po = M/pg,a(pa/po) is the volume of the gas discharged to the atmosphere under storage
conditions.
We now consider how the present criterion can be used for analysis of discharge of vessels of finite
volume. In the case of storage in a vessel with low pressure, formulas (8) and (9) can be used directly, but,
in this case, one should take into account that the quantity V0 may not coincide with the total volume of the
vessel V. Thus, for example, if a vessel has a rigid casing, and its volume is constant, discharge will proceed
only until the internal pressure becomes equal to the external pressure. In this case, only part of the gas
[having volume Vo ~ V(po - P a ) / P a ] will be discharged at pressure to the atmosphere. The exchange of the
remaining gas with the atmosphere will proceed under the action of diffusion processes and buoyancy forces;
this stage is not described by the present model. If a vessel has a variable volume (for example, it consists of
movable sections or has an elastic casing), the discharge volume is determined by the design features of the
vessel and should be specified in each particular case.
For discharge of high-pressure vessels up to pressure equalization, one can assume that the gas is almost
entirely discharged to the atmosphere and the part of the gas remaining in the vessel is a negligibly stoat1
portion of the stored mass. Thus, as the quantity V0, one can use the volume of the vessel itself. In this case,
however, one should take into account that, with a decrease in the internal pressure, the equivalent diameter
and the equivalent gas density change. We introduce the average gas pressure and density over the vessel, so
that the corresponding gas-flow rate is equivalent to the average flow rate during the time of discharge. As is
shown in [20], in adiabatic flow, the average flow rate is G~v = r#G0, where Go is the initial flow rate calculated

148
 Fig. 2. Criterion for determining the type of discharge in
dimensionless coordinates (~, ~): boundary curves 8j(~) and
8c(~) separate the regions of instantaneous (a), intermediate
(b), and quasi-continuous (c) types of discharge.

from the gas-storage parameters, and the proportionality factor 77 depends on the ratio of the pressure in the
storage vessel to the external pressure po/p~. The quantity 7? decreases monotonically with an increase in
po/pa. Thus, under nearly critical conditions, 7/~ 1, whereas at high initial pressures (po/pa > 100), the value
of 77 decreases to 0.25. The formula for the dependence ~(Po/P,,) obtained in [20] is very complex, but, for
po/pa > 10, it can be approximated with acceptable accuracy by 7/= s(pJpo) 1/6, where s ~ 0.6.
Since the gas-flow rate with flow choking is proportional to the square root of the product of the vessel
pressure and density, the average pressure and density can be written as pay = ~p0 and Pay = r/p0. Using pay
and p~v instead of P0 and p0 in (10) in calculating the gas outlet parameters, we finally obtain the following
expressions for the critical diameters:
8 (k-Fl"~(k+8)16(k-1)#gC4/3 Ipa~ 114
d~-cd~rS312~--'~) ~a UFL ~ 0 ) TV~, (13)

2 (-~a) CUFL\-~OI Vo. (14)

DIMENSIONLESS FORM OF CRITERION

The criterion proposed in the present work for classification of finite-duration discharges can be
written in dimensionless form as follows. We define the dimensionless hole diameter ~ and the parameter
characterizing the physicochemical properties of the flammable gas as
d ('--~ ~ ~ = (~ 1/~'~2/~ (15)
= V1/-'--~\pa/ ' k-~a] ~UFL'

where e = 0 for low-pressure discharge, 1/6 for high-pressure discharge, and 1/12 for discharge of a high-
pressure vessel. In the new coordinate, critical conditions (9), (12), and (14), which distinguish jet discharges
from cloud-like (intermediate) discharges, are described by a linear dependence, and the boundary between
the intermediate and the instantaneous types of discharge (8), (11), and (13) is described by the exponential
dependence
#: = ~:~, ,~c = "~c~213. (16)
The values of the proportionality factor -),j and 7c for different cases of discharge are listed in Table 1, where
the values of the constants are calculated for a = 0.132, B = 4.75, k = 1.4, Cd = 0.85 and, s = 0.6. The
boundary curves (16) are shown in Fig. 2: the lower curve 6j({) separates the jet and the cloud-like types of

149
 I
I I
v .at

-44

O0

I I
.at I

D-
c/

~o

0
--a

~a ~o
P~ b

150
 TABLE 2
Low pressure High pressure Discharge
Substance
~J ~c ~J ~c ~J 6c
Methane 0.210 0.155 0.509 0.195 0.647 0.252 0.834
Hydrogen 0.219 0.162 0.522 0.204 0.664 0.263 0.857
Propane 0.257 0.190 0.582 0.239 0.739 0.308 0.953
Butane 0.271 0.201 0.603 0.252 0.767 0.325 0.989
1-Butene 0.286 0.212 0.625 0.266 0.794 0.343 1.02
Propylene 0.289 0.214 0.629 0.269 0.800 0.347 1.03
Ammonia 0.327 0.242 0.684 0.304 0.869 0.392 1.12
1, 3-Butadiene 0.342 0.253 0.704 0.318 0.895 0.410 1.15
Ethylene 0.487 0.360 0.880 0.453 1.12 0.585 1.44

discharge, and the upper curve 6c(~c) corresponds to the transition from the intermediate type of discharge
to discharge with cloud formation. The coordinate axis shown on the left refers to low-pressure discharge,
and the axes shown on the right refer to high-pressure discharge at constant pressure (left) and at a pressure
decreasing with time (right).
The values of the parameter ~ for a series of widely used gases - - hydrogen, volatile hydrocarbons, and
ammonia - - are listed in Table 2, where the corresponding critical diameters 6j and 6c are also indicated.
Evidently, for the entire set of gases, the parameter ~ varies from 0.2 to 0.5; this range seems to be the most
important for further theoretical and experimental investigation.

COMPARISON WITH EXPERIMENT

Estimation of the accuracy of the criterion proposed in this paper requires conducting experiments
in which both geometrical (the vessel volume and the hole diameter) and physicochemical parameters of
the substance and storage conditions are varied. We are unaware of such detailed studies, but the available
literature data can be used to verify at least the consistency of the results obtained.
Chaineaux et al. [21, 22] studied methane and hydrogen discharges from a vessel with a volume of
0.12 m 3 at p0 = 10 MPa. Nozzles with diameters of 6- 10-3, 1 . 2 . 1 0 -2, and 2.4- 10 -2 m were used. In
the experiments, turbulent-jet propagation was observed for the first few seconds after depressurization. After
that, the jet existed for a few tens of seconds, its length decreasing slowly with time due to the pressure drop in
the vessel as the gas discharged. Thus, the jet type of flow was observed in the experiments. For the indicated
holes, the dimensionless diameter 6 = dVol/3(po/p~)Un is equal to 1.78.10 -2, 3 . 5 7 . 1 0 -2, and 7.15- 10-2,
respectively, which is much smaller than the critical diameter, which has close values 6j = 1.2 ~ ~ 0.26 for
methane (~ = 0.21) and for hydrogen (~ -- 0.219) (see Table 2). In the plane (~, 6), the corresponding points
are in the region of jet flow (points 1 in Fig. 2), which is in complete agreement with the observation results.
A wider range of pressures was used in the experiments of [23], where hydrogen and methane discharges
proceeded from a vessel with a volume of 2.5 9 10 -4 m 3 with outlet diameter of 5 9 10-4-8 9 10 -3 m and
p0 = 5-20 MPa. Substitution of these values into (15) gives the range of values of the dimensionless diameter
= 0.01-0.20 in the region of jet flow (as above, 6 < 6j ~ 0.26). This agrees with the photographs given in
[23], in which the jet arising during discharge is clearly seen. For illustration, the values of 6 = 0.01, 0.1, and
0.2 belonging to the indicated experimental range are shown by points 2 in Fig. 2.
Another type of discharge (instantaneous discharge) was studied by Landis et al. [10]. The vessel was
a tube 0.102 m in diameter and 0.305 m long. It has a hemispherical flange at one end, and the other end was
closed by a discontinuous membrane with a rupture pressure of 3.22, 7.17, 20.52, and 70.69 bar. The setup was
used to model the emergency separation of the end part of a typical high-pressure vessel. In the experiments,

151
 the vessel was filled with a nitrogen mixture with additives facilitating visualization of the resulting gas cloud.
Immediately after rupture of the membrane, a conical cloud was discharged to the atmosphere." Landis et al.
[10] identified the observed type of discharge as instantaneous. Since the test gas was not flammable and the
upper ignition limit CUFL was not determined for it, the present criterion cannot be used directly. However,
for estimation, we assume that the values of CUFL for the gas are in the range of 0.05-0.2, which is typical
of many flammable gases; this gives a range for ~ of 0.125 to 0.35. For the above-mentioned initial pressures,
the dimensionless parameter 5 is 0.8, 0.86, 0.93, and '1. In Fig. 2, the range of 0.125 ~ ~ ~< 0.35, 0.8 ~< 5 ~< 1
is dashed. Evidently, for part of the parameters, the points fall in the region of instantaneous discharge, and
for the other part, the points fall in the region of intermediate [according to classification (7)] discharge This
agrees with experimental data, especially if one takes into account that the transition from one regime to
the other is smooth. Note that the values of 5 near the boundary between the regions of instantaneous and
intermediate discharge were obtained when the outlet dimension was comparable with the vessel dimensions.
This means that truly instantaneous discharge with the formation of a gas cloud which is poorly mixed with
air can be expected only with very strong or complete failure of the vessel casing. This conclusion also agrees
with the results of observations.
Finally, one more series of experiments that can be used to verify the proposed criterion is described in
[24]. In this experiments, of 1.5 to 13 g of methane and propane were discharged to the atmosphere to study
the time of combustion of fireballs upon their ignition. Different nozzles were used: circular, perforated, slit,
and radial, the effective diameter of the discharge hole being 2 to 5 cm. Rubber balloons were used as working
vessels. They produced an excess pressure of 0.7 to 14 kPa by extension of the casing, and the linear discharge
rate varied from 9 to 88 m/sec. Thus, in contrast to the experiments of [10, 21-23] described above, Roper
et al. [24] studied subsonic discharge with a small pressure drop. Visualization of the resulting flow without
gas combustion showed that the discharge has the shape of an elongated cloud and was close in structure
to a propagating jet. This means that, according to the classification (7) proposed in this work, this type
of discharge can be regarded as intermediate (cloud-like) or close to jet. We compare this conclusion with
the results obtained according to criterion (7). Roper et al. [24] do not give a description of the experiments
performed, but only indicate ranges of the gas masses used (1.5-13 g), which corresponds to the values
V0 = (2.3-19.6) 9 10-3 and (0.9-7.6) 9 10-3 m 3 for methane and propane. Calculations for d = 2 9 10 -2 and
5-10 -2 gave the following results for methane. With the least value of the volume, 6 = 0.152 and 0.38 for small
and large holes, respectively, whereas for the largest volume, 5 = 0.08 and 0.185, respectively. In the case of
low-pressure methane discharge (~ = 0.210), 6c = 1.44~ 2/3 = 0.509 and 5j = 0.74~ = 0.155. Evidently, for
the smallest hole and the largest volume, the values of 5 are in the region of jet discharge, whereas for the
largest outlet and smallest volume, the perimeter 5 falls in the region of intermediate (cloud-like) discharge.
In the two other cases, the values of 5 are near the boundary 5j. Similar calculations for propane (~ = 0.257,
5c = 0.582, and 5j = 0.19) give 5 = 0.21 and 0.52 for the smallest volume and 5 = 0.1 and for the largest
volume, for small and large holes, respectively. As earlier, only one point (for the largest volume and the
smallest holes falls in the region of jet discharge, and the remaining points are near the boundary 6j or in
the region of intermediate (cloud-like) discharge. The calculation results are shown in Fig. 2 by points 3 for
methane and points 4 for propane.
Thus, the results of the comparison show that the criterion (7) proposed in the present paper allows
one to determine the type of discharge using only the geometrical parameters d and ~ , the physicochemical
properties of the substance CUFL and ttg, and the initial pressure p0. The criterion is given as a relationship
between the dimensionless parameters 5 and ~, and this makes it very general. Of course, more detailed
experimental studies using various substances and discharge conditions are required to obtain a comprehensive
estimate of the accuracy and range of application of the given criterion, but even the above comparison shows
that the criterion gives fairly reasonable results.

152
E F F E C T OF I G N I T I O N D E L A Y ON T H E C O M B U S T I O N R E G I M E OF O U T F L O W I N G GAS

Accidental emissions of flammable gases are frequently accompanied by gas ignition, which is a hazard
because of thermal radiation. If the gas discharges as a quasi-stationary jet, ignition gives rise to a flame whose
combustion intensity is determined by the rate of fuel supply, and the time of existence can be estimated as the
period from the moment of ignition to completion of the discharge. Formulas for the flame length, heat-release
intensity, and heat fluxes at various distances are.known and can be found, for example, in [3-5].
Rapid discharge of the total mass of gas, conversely, gives rise to a compact fuel-rich cloud. With
immediate ignition, this cloud burns in a diffusion regime as a fireball. The main parameter determining the
diameter DFB of the fireball, the duration of its combustion tFB, and the heat flux from its surface is the initial
fuel mass M. At present, there are a number of empirical formulas describing these relations as exponential
functions of mass:
DFB = A 1 M p, trB = A 2 M q. (17)
Here p ~ 1/3 [3, 25]. The data of various authors for the time of fireball combustion are less consistent:
the exponent q varies from 1/6 to 1/3. The cause of this discrepancy was studied by Roper et al. [24], who
showed that the exponent 1/6 corresponds to the predominant action of buoyancy forces on the turbulent
combustion, whereas the regime 1/3 corresponds to the dominance of inertial forces (for example, in high-speed
jet discharge).
In the intermediate [according to classification (7)] type of discharge, the gas-combustion characteristics
can depend on the ignitibn delay. We consider the case of immediate ignition by which is meant ignition with
a delay ti that does not exceed the time of gas discharge ~r. Note that if cloud ignition proceeds with a
longer delay, a considerable amount of gas can be mixed with air to nearly stoichiometric concentrations, so
that ignition can lead to gas-cloud explosions accompanied by the formation of shock waves. This case is not
considered here; the corresponding calculation procedures are given, for example, in [1-5].
We introduce the parameter 8 = t r / t F B , which characterizes the ratio of the discharge time to the
time of combustion of the total gas mass. In addition, we define the dimensionless moment of initiation of the
reaction r = ti/t~, so that 0 ~< ~- ~ 1. If 8 < 1, diffusion cloud combustion is the limiting (i.e., slower) process.
In this case, one might expect that, irrespective of the specific moment of ignition, the burning gas forms a
fireball. The case 8 > 1 is more complicated. The gas mass discharged to the atmosphere up to the moment
of ignition is equal to T M , and the time of gas combustion referred to the time of combustion of the entire
g a s t F B is T B = 7 q [see (17)]. We compare this time with the time r E ----- ( t r - - t i ) / t F B = 8(1 - r) required for
completion of ignition from the moment of ignition (also referred to tFB). Analysis is conveniently performed
using Fig. 3, which shows a curve of T B ( r ) , which passes through the points (0, 0) and (1, 1) and the straight
line rE(z), which passes through the point (1, 0) and cuts the value 8 on the ordinate.
The point of intersection r. gives the moment of initiation at which the time of combustion of the
gas that has already been discharged to the atmosphere is equal to the time during which flow will continue.
For 0 ~< r ~ r., the discharge will be the limiting (slower) process. Consequently, one might expect that the
burning gas forms a flame which will exist for the remainder of the time up to the end of discharge. Conversely,
for r. ~< r ~< 1, diffusion combustion is the limiting process, and the general flame configuration will be close
to a fireball. The values of the delay r. that distinguish these two combustion regimes are determined from the
equation 8(1 - r.) = r.q. A graph of the function r.(8) is given in Fig. 4 for q = 1/3 and 1/6. Here for 8 < 1,
the corresponding branches are shown by dashed curves, since in this range, the occurrence of a fireball should
be expected for any 0 ~ 8 ~< 1. The curves divide the range of parameters into two subregions: the lower
region corresponds to the combustion of the gas discharging as a flame, and the upper region corresponds to
the formation of a fireball. Using this figures, one can estimate the most probable flame configuration for the
given ignition delays from the known discharge parameters (total mass and discharge duration).
Naturally, the formation of a clearly defined flame or a fireball is possible only far away from the
boundary between the combustion regimes r.(8). Whereas in the near-boundary region, in which the time of
combustion of the gas that has already discharged is Comparable with the time of completion of discharge,

153
 t/tF 1.0
,#.

1.0

rr 0.5 ,/r
0.5 Rame

IJY! = I =(~ I
9z'. 0.5 r 1.0 0 1 2 3 4 5
Fig. 3 Fig. 4

Fig. 3. Determination of fuel-combustion regime by comparison of the burning time (rB)

for the gas discharged to the atmosphere up to the moment of ignition and the time of
completion of discharge (rE).

Fig. 4. Boundary between gas-combustion regimes plotted for q = 1/3 (curve 1) and 1/6
(curve 2).

a more complicated configuration can be observed, for example, the simultaneous existence of a fireball (in
which the gas discharged before the moment of ignition is burning) and a flame (in which the gas remaining
in the vessel at the moment of ignition burns).

T H E C O E F F I C I E N T OF F U E L P A R T I C I P A T I O N IN F I t t E B A L L C O M B U S T I O N

Since the fuel mass is the main parameter that determines the fireball characteristics, it is necessary to
estimate how the mixing processes can decrease the fuel fraction that burns in a fireball (i.e., the coefficient
of fuel participation). For rapid (instantaneous) discharge, the fuel practically has no time to mix with air,
and the coefficient of participation is close to unity. In the case of intermediate [according to classification (7)]
discharge, part of the discharging gas is diluted by air to concentrations lower than the upper concentration
limit. This decreases the fuel fraction that is present in the fuel-rich zone and burns as a fireball. We adopt
the following scheme for finding the possible coefficient of fuel participation. Let ignition of the discharge
occurs at time ti ~< t r . The gas discharged before this moment is partially mixed with air, so that the mass
of the gas with a concentration exceeding the upper concentration limit is smaller than the discharge mass
by a value AM. We shall assume that only the gas whose concentration at the ignition moment is higher
than the upper concentration limit takes part in fireball combustion, since it is this gas that will burn in
the diffusion regime. In addition, we assume that the entire gas discharged to the atmosphere after initiation
of the reaction participates in the reaction. This is caused by the fact that for this part of gas, an ignition
source is already available. Thus, we assume that the fuel mass in the fireball is smaller than the total mass
of the fuel discharged to the atmosphere M by a value A M , which is due to gas mixing before the moment
of ignition. This assumption gives the lower bound for the possible coefficient of fuel participation:
X = (U - AM)/M = 1 - AM/M.

For cloud-like discharges, the mass of the gas with concentration C ~< CUFL can be obtained by integration
of the concentration distribution in a propagating jet (2). In this case, the domain of integration is extended
along the axial coordinate from 0 to the front coordinate h at the ignition moment ti [see (3)], and along
the radial coordinate, it is extended from the surface on which C = CUFL to infinity. Integration gives

1.54
A M = ~ra2pg,aCuFLh3/3, which after substitution into (3) and some transformations reduces to the form
AM_ 2X/~'CuFL (pg,a]3/aC#g'~3/4(Vo'~l/2(po'~'/2.3p." (18)
M 3~c~B 312 ', Pb / \'-~=) \-~) \'~=)
Substituting the values of Pb and P0 into (18) and using 6 and ~ for tow- and high- pressure discharges and for
discharge of a finite-volume vessel, we obtain the following final formula for the coefficient of fuel participation
in the fireball:
= I - 312. (19)
Here the coefficient g for each case of discharge is given in Table 1. It follows from (19) that the coefficient
of fuel participation decreases as 3/2 with an increase in r and a decrease in 6. Since the minimum diameter
for which the discharge belongs to the cloud-like type corresponds to the boundary 6 -- 6j [see (16)], the
boundary value of the participation coefficient (the same for all discharge pressures) is
XJ = 1 - ~rjr 312, (20)

where crj = crrj

^-3/2 (see Table 1). It follows from (19) and (20) that the participation coefficient reaches a
minimum value of Xmin = I - - O'j ~ 0.33 when ignition proceeds precisely at the moment of completion of
discharge (r = 1) and the hole size coincides with the critical dimension that corresponds to the boundary
between cloud-like and jet discharges. Thus, mixing of outflowing gas with air prior to the moment of ignition
can decrease considerably (by a factor of three) the mass of the gas that can participate in fireball combustion.
The range of X = 0.33-1.0 can be used for a rough estimate of the maximum and minimum possible
consequences of discharge combustion in the fireball regime, whereas in a more detailed analysis, relation
(19) makes it possible to estimate the fireball parameters more exactly, with allowance for specific features of
the course of the accident.

CONCLUSION

In this paper, we propose a quantitative classification of flammable-gas discharge from low- and high-
pressure vessels. The criterion determining the type of discharge of finite duration is given as a relationship of
two dimensionless quantities - - the hole diameter and the parameter including the physicochemical properties
of the substance. Formulas are obtained for dimensionless critical parameters that distinguish quasi-continuous
(jet), instantaneous (leading to cloud formation), and intermediate discharges. The criterion enables one to
describe in a unified manner discharge of low- and high-pressure vessels, and it is also applicable to the analysis
of discharge of finite-volume vessels. A comparison with the experimental data available in the literature
indicates that the proposed theoretical model gives reasonable results. Certainly, a comparison with detailed
experimental data and results of numerical simulation of unsteady discharges will play an important role in
determining the accuracy and range of application of the model.
Comparison of the characteristic times of discharge and combustion showed that, depending on the
ignition delay, combustion of the discharged gas can proceed in both the flame and the fireball regimes. In
the latter case, the effect of discharge conditions and ignition delays on the fuel mass that can participate in
fireball combustion are estimated. Thus, the results obtained makes it possible to analyze the type of discharge
during depressurization of vessels pressurized by a flammable gas and to determine the most probable flame
configuration in discharge combustion.
The proposed approach to the analysis of the formation and combustion of fire-hazardous clouds in
finite-duration discharge can be used in solving problems of chemical process risk evaluation, construction risk
maps, expert evaluation of the conseque=ces of acddents in pipelines, in pressurized gas storages, etc. The
dimensionless relations obtained in the paper can be of significance in the statistical analysis of accidents. The
paper deals only with one-phase (gas) flow, but the proposed approach can be extended to two-phase flow.
The authors are grateful to the Committee of Scientific Research of EPSRC (Grant G R / K 13486) for
support of this work.

155
REFERENCES
. V. Marshall, The Main Hazards of Chemical Processes [Russian translation], Mir, Moscow (1989).
2. M. V. Beschastnov, Commercial Explosions. Estimation and Prevention [in Russian], Khimia, Moscow
(1991).
3. Guidelines for Chemical Process Quantitative Risk Analysis, Center for Chemical Process Safety
(AIChE/CCPS) (1989).
4. Methods for the Calculation of Physical Effects Resulting from Releases of Hazardous Materials
(Liquids and Gases) - - TNO "Yellow Book," 2nd ed., Voorburg (1992).
5. SFPE Handbook of Fire Protection Engineering, Nat. Fire Protection Association, Quincy, MA (1988).
. R. S. Scorer, Environmental Aerohydrodynamics, John Wiley and Sons, New York (1978).
7. G. N. Abramovich, T. A. Girshovich, S. Yu. Krasheninnikov, et al., Theory of Turbulent Jets [in
Russian], Nauka, Moscow (1984).
8. H. C. Hardee and D. O. Lee, "Expansion of clouds from pressurized liquids," Accid. Anal. Prevent.,
7, 91-102 (1975).
9. A. F. Roberts, "Thermal radiation hazards from releases of LPG from pressurized storage," Fire
Safety J., 4, 197-212 (1981/82).
10. J. G. Landis, R. E. Linney, and B. F. Hanley, "Dispersion of instantaneous jets," Proc. Saf. Prog.,
13, No. 1, 35-40 (1994).
11. H. Phillips, "Ignition in a transient turbulent jet of hot inert gas," Combust. Flame 19, 187-195
(1972).
12. A. D. Birch, D. R. Brown, M. D. Dodson, and J. R. Thomas, "The turbulent concentration field of a
methane jet," J. Fluid Mech., 88, 431-449 (1978).
13. P. Sforza, "Mass, momentum and energy transport in turbulent free jets," Int. J. Heat Mass Transfer,
21,271-284 (1978).
14. D. R. Dowling and P. E. Dimotakis, "Similarity of the concentration field of gas-phase turbulent jets,"
J. Fluid Mech., 218, 109-141 (1990).
15. K. Moodie and B. C. R. Ewan, "Jets discharging to atmosphere," J. Loss Prevent. Proc. Ind., 3,
No. 1, 68-76 (1990).
16. C. J. Chen and W. Rodi, Vertical Turbulent Buoyant Jets. A Review of Experimental Data, Pergamon
Press, New York (1980), p. 31.
17. R. H.Perry and D. Green (eds.), Perry's Chemical Engineering Handbook, 6th ed., McGraw-Hill, New
York (1984).
18. A. D. Birch, D. R. Brown, M. D. Dodson, and F. Swaffield, "The structure and concentration decay
of high-pressure jets of natural gas," Combust. Sci. Technol., 36, Nos. 5 and 6, 249-261 (1984).
19. B. C. R. Ewan and K. Moodie, "Structure and velocity-measurements in underexpanded jets,"
Combust. Sci. Technol., 45, Nos. 5 and 6, 275-288 (1986).
20. J. L. Woodward and K. S. Mudan "Liquid and gas discharge rates through holes in process vessels,"
J. Loss Prevent. Proc. Ind., 4, No. 3, 161-165 (1991).
21. J. Chaineaux and G. Mavrothalassitis, "Discharge in air of a vessel pressurized by a flammable gas,
and the volume of the resulting mixture generated," J. Loss Prevent. Proc. Ind., 3, No. 1, 88-90
(1990).
22. J. Chaineaux, G. Mavrothalassitis, and J. Pinean, "Modelization and validation tests of the discharge
in air of a vessel pressurized by a flammable gas," Prog. Astronaut. Aeronaut., 134, 104-137 (1991).
23. V. K. Baev, V. V. Shumskii, and M. I. Yaroslavtsev, "Self-ignition of a fuel gas escaping into an
oxidizing medium," Fiz. Goreniya Vzryva, 19, No. 5, 73-80 (1983).
24. F. Roper, J. Arno, and H. C. Jaggers, "The effect of release velocity and geometry on burning times
for non-premixed fuel gas clouds," Combust. Sci. Technol., 78, 315-338 (1991).
25. R. W. Prugh "Quantitative evaluation of fireball hazards," Proc. Saf. Prog., 13, No 2, 83-91 (1994).

156