Cetegen, B. M., Zukoski, E. E., & Kubota, T. Entrainment in The Near and Far Field of Fire Plumes

This article was downloaded by: [Cornell University Library]
On: 15 November 2014, At: 05:54

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK
Combustion Science and Technology

Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/gcst20
Entrainment in the Near and Far Field of Fire Plumes

a a a
B. M. Cetegen , E. E. Zukoski & T. Kubota
a
Karman Laboratoryof Fluid Mechanics and Jet Propulsion, California Institute of
Technology , Pasadena, CA , 91125
Published online: 17 Apr 2007.
To cite this article: B. M. Cetegen , E. E. Zukoski & T. Kubota (1984) Entrainment in the Near and Far Field of Fire Plumes,
Combustion Science and Technology, 39:1-6, 305-331, DOI: 10.1080/00102208408923794
To link to this article: http://dx.doi.org/10.1080/00102208408923794
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
Combusrion Scicncr and Technology, 1984, Vol. 39, pp. 305-331 0 1984 Gordon and Breach Science Publishers, Inc.
0010-2202/84/3906-0305$18.50/0 Printed in Great Britain
Entrainment in the Near and Far Field of Fire Plumes
B. M. CETEGEN. E. E. ZUKOSKI and T. KUBOTA Karman Laboratoryof

Fluid Mechanics and Jet Propulsion, California Institute of Technology,
Pasadena, CA 91 125
Presenred at the Fire Research Conference in honor of Howard Emmons, Washington D.C., 1983
Abstract-This paper describes entrainment measurements made in fire plumes with a new tech-
nique. Measurements were in plumes rising from natural gas diffusion flames stabilized on 0.10,
Downloaded by [Cornell University Library] at 05:54 15 November 2014
0.19 and 0.50 m diameter burners and the heat release rates ranged from 10 to 200 kW. The
heights examined ranged from elevations starting very close to the burner surface to distances
about five times the average flame heights. Experiments indicate the presence of three regions: a
region close to the burner surface where plume entrainment rates are independent of the fuel flow
(or heat release) rates; a far field region above the flame top, where a simple point source model
correlates the data reasonably well; and an intermediate region where entrainment appears to be
similar to that of a turbulent plume.
1 INTRODUCTION
One of the artifices used in current models of fire spread through complicated structures
is that of dividing each space within the structure into two stratified layers, an upper
layer or volume which contains the hotter gas and a lower volume which contains the
cooler gas. The assumption is made that the gas in each volume is homogeneous and
the processes which occur in each are analyzed by making use of conservation equa-
tions for mass, species and energy. In these models, a description must also be given
of the various transport processes for mass, species and energy which are allowed to
occur between the gas in one layer and all other layers with which it communicates
and all the surfaces which bound it.
In the early stages of a fire within a room, the primary transport mechanism between
the layers is the buoyant plume which rises above the fire. This plume entrains cool
air from the lower layer, heats it in the fire and transports it to the upper layer. The
amount of air entrained in the lower layer has a strong influence on the temperature
and concentration of combustion products in the upper or ceiling layer and the volu-
metric flot rote uf smoke from the fire room to other rooms within the structure.
Thus it is clear that the entrainment process of the plume is a key element in the
development of a fire in a structure.
Although considerable information is presently available concerning entrainment in
buoyant plumes, it has principally been obtained in plumes which have risen far above
their origin and which have densities little different from that in the surrounding gas.
In contrast, in building fires, we are interested in regions of the plume located very
close to the origin where the finite size of the heat release region and large density
differences may influence the entrainment process. When flames extend into the hot
ceiling layer gas, the problem may be further complicated by greatly reduced oxygen
concentrations in the ceiling layer and the possibility that unburnt fuel can build up
in that layer and eventually result in phenomena such as flashover.
The previous work o n fire geometry and plume entrainment is extensive and was
305
306 B. M. CETEGEN, E. E. ZUKOSKI AND T. KUBOTA
recently reviewed (Zukoski ef al., 1980; Cetegen et al., 1982). The experimental work
on entrainment can primarily be divided into two types. In the first, point measure-
ments of time averaged fluid properties are made in a turbulent, highly fluctuating
environment and then used to estimate the plume mass flow rates by neglecting the
cross correlation between the gas density fluctuations and the velocity fluctuations.
A typical example of the difficulties encountered with point measurement technique is
illustrated .by the gas velocity measurements of Cox and Chitty (1980) who used a
technique based on the cross correlation of random thermal fluctuations in a turbulent
diffusion flame. In this technique, the time elapsed during the passage of a thermal
disturbance between the two thermocouple probes and the separation distance between
the probes are used to estimate the gas velocity. This technique will be valid when the
disturbances travel with the local gas velocity and the probes are aligned with the
velocity vector. Although these conditions may be satisfied most of the time along the
centerline of the plume, they are not a t the edges of the plume where the area change
due to an increment of radius is largest. Hence, this method is inappropriate for use
in fire plumes.
The alternate approach to making entrainment measurements concerns the use of a
direct measurement of mass flow in the plume. This technique, described earlier
(Zukoski et al., 1980; Cetegen ef al., 1982) and later in this paper, involves capturing
the entire plume mass flow in a hood. In this technique, the plume plunges through
an interface formed between the hot gas captured in the hood and the cooler ambient
air. Thus, the technique measures the total flow into the ceiling layer which may be
slightly larger than the plume mass flux. Two methods were used to determine the
mass flow rates. First, mass balance argument was employed when ceiling gas is
relatively cool and second was based on the chemical analysis of the species in the
ceiling gas when the gas was very hot.
, The theoretical work on fire plumes can also be divided into several groups. The
earliest work delt with adiabatic plumes rising from point or distributed sources of
heat, mass and momentum, and the problem was solved by use of an integral tech-
nique. More recently, the much more elaborate k-E-g method for analysis of turbulent
flows has been extended to include the study of buoyant turbulent jets. The inclusion
of the heat addition in either approach requires the use of ad hoc assumptions or
procedures which, in our opinion, are not satisfactory. In particular, large scale,
structures are present in the buoyant diffusion flames of interest to us (see Kubota,
1977) and we expect that they influence the combustion and entrainment process in a
manner not now included explicitly in any of the fire plume models.
This paper is essentially an extension of the work published earlier by the authors
(Zukoski et al., 1980). It contains a brief study of flame geometry, reported in detail
elsewhere (Zukoski et al., 1984). A review is then given of integral models for plumes
with large density differences which rise from finite sources with finite values of mass,
momentum and heat fluxes a t the source. These models are then used to introduce the
idea that the simple point source plume model, used to correlate the far field plume
data, requires an offset for the origin of the plume. The effective origin lies near the
burner surface a t an appreciable distance below the top of the flame.
This material is followed by a discussion of the far field data and the location of the
virtual origin of the model plumes. Finally the experimental data concerning entrain-
ment in the near field of fire plumes is presented and discussed with the help of several
ad hoc models.
The material discussed here is available in somewhat greater detail elsewhere
(Cetegen et al., 1982) and the experimental data are tabulated there.
ENTRAINMENT I N FIRE PLUMES
2 EXPERIMENTAL TECHNIQUE AND APPARATUS
The basic idea for the technique used to make plume mass flux measurements is best
illustrated by the use of the sketch shown in Figure 1. In the early stages of a fire, the
hot products ofcombustion produced by a flame are usually segregated in a well stirred
ceiling layer whose properties are roughly homogeneous. The flame and the buoyant
plume it produces entrain cool air from the room and heat it in the fire by combustion
processes and in the plume by mixing with the plume gas. If the fuel consumption
rate is tivand the air entrainment rate is m E , this heated gas flows into the ceiling layer
at a rate ( m ~ + t i l ~ ) .Other gas may flow into the ceiling layer as a result of the disturb-
ance produced by the plume as it plunges into the ceiling layer, &, or as a result of
other mixing processes which may occur at the interface separating the hot ceiling
layer from the cool air layer, rill. Thus the total mass flux to the ceiling layer, t A c
will be the sum of these contributions,
The terms ( t ; ~ ~ + t A ~ + mare

~ ) directly associated with the fire plume and are called the
plume mass flux, i f pBecause
. the interface between the hot and cold gas is stabilized
by the gravitational field, we expect that the mass entrainment rate at the interface
far from the fire, m l will often be negligible in the absence of some other mixing pro-
cess such as drafts through a door or window. Observations of gas motion (made
visible by adding smoke to the ambient gas) have been made in our apparatus and they
suggest that very little entrainment if any occurs at the interface and that most occurs
a t the edges of the plume itself.
Since we need to be able to describe entrainment as a part of any model which is
used to predict the behavior of the early stages of a room fire, the primary purpose of
our experiments was to measure the plume related mass flux, mp as a function of height
above the fire source, Z t ; heat release rate, of and the initial fire geometry expressed
as burner diameter, D. When Zd is above the top of the flame, measurement is accom-
plished by use of the hood shown schematically in Figure 2. The interface level of
FIGURE 1 Entrainment contributions into the ceiling layer.

B. M. CETEGEN. E. E. ZUKOSKI AND T. KUBOTA
FIGURE 2 plume mass flux measurement technique.

the ceiling layer is maintained at a constant height above the fire source by with-
drawing a suitable flow of hot ceiling layer gas through an exhaust pipe. The hood
was made large enough to ensure that the gas in the hood is kept relatively quiescent
so that it will not entrain air a t the interface (rill of Figure 1 ) . When this entrainment
can be neglected, a simple conservation of mass argument shows that the mass with-
drawal rate required to maintain a fixed interface height is the plume mass flux, mp.
When the interface lies well below the top of the flame, the gas temperature in the
ceiling layer becomes too high for the apparatus described above, and a modification
of the first technique was required. The method used to measure the plume mass flow
rates in the near field of the fire plume is based upon measuring the fuel flow rate and
the mole fractions of combustion products in the well-stirred hot gas captured by a
small hood. If the complete oxidation of methane, CH4, into COzand Hz0 is achieved,
then the measurement of the concentration of either COz, or excess 0 2 allows the
calculation of the overall air-fuel ratio of the gas which enters the hot gas layer. Since
the amount of fuel supplied to the burner is measured, we can then determine the
entrainment rate of ambient air in the lower regions of the flame which are exposed to
ambient air. The complete combustion of methane in the presence of x moles of
excess air can be written as,
then for dried products,
8.52 Yo, 1- 8.52 Yco,

X = or X = (3)
1 - 4.76 Yo, 4.76 Ycon
Here Yo, and Yco, are measured mole fractions of Oz and COz respectively. When
we are only interested in the air-fuel ratio of the exhaust gases, we can ensure that all
carbon is reduced to COz and hydrogen to Hz0 by passing the gas through a furnace
in the presence of excess oxygen. The equivalence ratio (which is the stoichiometric
air-fuel ratio divided by the measured air-fuel ratio) in these experiments was never
greater than 0.70; and hence excess oxygen was always present in the hood gas.
Either of the two independent estimates of I, obtained from Eq. ( 3 ) can be used to
estimate the entrained flow rate of air as,
ENTRAINMENT IN FIRE PLUMES 309
The large hood and the associated equipment used to make our far field entrain-
ment measurements (interface height, Zr greater than the average flame height,Z,J
are described in detail elsewhere (Zukoski et al., 1980; Cetegen et al., 1982). The
hood dimensions are 2 . 4 x 2 . 4 ~1.56 m deep; it was made out of steel and insulated
with lOcm thick glasswool. The burner and the floor can be moved vertically to
change the elevation of the interface height between 2.3 m and a lower bound of
roughly 1.0 m which is set by our desire to keep the ratio of interface thickness to the
interface height below 1/10.
The near-field entrainment measurements were made in the small hood, a 1.2 m
cube, open on the bottom side, which was designed to be used a t gas temperatures up
to 1C0O0C. The hot gas is allowed to spill out under the edges of the bottom side of
the hood. The plume mass flux is deduced (as explained above) from the measure-
ments made of the carbon dioxide and oxygen partial pressures in the gas samples
withdrawn from the hood and from the measurements of the fuel flow rates.
The experimental set-up is shown in schematic diagram of Figure 3. The small hood
(I .2 x 1.2 x 1.2 m), made out of steel, is lowered upon the fire by a pulley mechanism
and the fires are stabilized by the 0.10,0.19 and 0.50 m diameter axisymmetric burners
described below. A gas sample is withdrawn from the small hood through a 6.4 mm
dia. stainless-steel tube inserted to various depths in the small hood. An aspirated
chromel-alumel thermocouple is placed at the entrance of the probe to monitor the
gas temperature. The sample was first passed through a furnace which insures com-
plete oxidation of incomplete combustion products (mainly CO), and is then dried
and filtered to remove particulates from the gas stream. The furnace is an 84cm-long,
U-shaped, 0.64-cm O D stainless-steel tube (0.025 cm wall thickness) kept at about
1000°C by electrical heating. The COz and C O contents of the sample was measured
with Beckman Model 864 Nondispersive Infrared analyzers. The Oz concentration
FIGURE 3 Apparatus for chemical sampling.

was monitored with a Beckman Model 741 Amperometric oxygen detector and the
amount ofany remaining unburnt hydrocarbons was measured with a Beckman Model
400 Flame Ionization Hydrocarbon analyzer. The location of the hot-cold gas inter-
face was determined by a shadowgraph system and the screen surrounding the large
hood helped to suppress the disturbances in the ambient air. The large hood was used
to remove continuously the hot gases spilled around the edges of the small hood.
The gaseous fuel used in these experiments was city gas taken without processing
from the Southern California Gas Company mains and is a mixture of hydrocarbons
and its principal constituents, with their mole fractions, are methane 0.924, ethane
0.042, nitrogen 0.015 and propane 0.01. The lower heating value is about 47.5 MJ/kg
and density is about 0.72 kg/m3 a t 20°C and one atmosphere. These properties are
average values which d o not vary more than a few percent from week to week. A
Meriam laminar flowmeter was used to measure the flow rate of the fuel.
The fuel was fed into the flame after passing through a porous bed of spherical glass
beads (0.2 to 0.5 cm in diameter) which was 2 to 5 cm deep. The surface of the bed
was made flat and flush with the metal burner. Three burner designs with diameters of
0.10, 0.19 and 0.50 m were used and they are described in detail elsewhere (Zukoski
et a/., 1980, 1984). Experiments were performed with and without a 2.4-m square
floor mounted flush with the upper surface of the burners. In the experiments without
the floor, the burners resembled right circular cylinders which extended a t least 2 to 3
diameters above the floor of the laboratory. This arrangement prevented strong shear
interactions developing between the entrained air and the hot-cold gas interface when
the interface height was less than 25 cm above the top of the burner.
The velocity of the fuel a t the surface of the burner is dependent on its temperature
at that point. Because of the intense thermal radiation to the glass beads at the surface
of the burner and subsequent heating of the fuel, the gas temperature a t the surface is
about 600°K. Values of Richardson, Rij and Reynolds, Ref numbers based on burner
diameter and fuel velocity a t the top of the burner are shown in Table I. Minimum
TABLE 1
Burner initial parameters
values of the Richardson number are large enough to ensure that the buoyancy of the
jet dominates the initial momentum even a t small heights equal to the burner diameter
in all our experiments. The maximum value of the Reynolds numbers shown in the
table (for the 10-cm burner at 126 kW)was about 2500. Although this value is large,
the increase in kinematic viscosity which occurs a t the flame surface (due to temperature
rise caused by combustion) will keep the fuel "jet" in a laminar state for a t least a
number of diameters. Hence, we expect that the momentum flux of the fuel is never
very important in our flames and that close to the burner lip, the flow will be laminar
though it may be unsteady. Here, we also note that for large diameter fires (0.50-m
ENTRAINMENT I N FIRE PLUMES 311
burner a t 126 kW) Richardson numbers are very large and Reynolds numbers are
small. Thus, for these large diameter fires, which are typical of many real fires,
buoyancy is clearly dominant and laminar conditions are expected.
3 FLAME STRUCTURE REVIEW
The authors have presented elsewhere (Zukoski el a/., 1980, 1984) the results of a
photographic study of the flames which are used in this entrainment study. The prim-
ary results of interest are summarized here. First, we found that large vortex-like
structures were produced in the flame sheet close to the burner with a frequency of
about 1/3d(g/D). These structures have an initial diameter close to the burner dia-
meter and grow as they rise to the top of the flame. Near the burner surface, the
smallest disturbances in the flame sheet are 2-4cm wrinkles o r folds and the smaller
scales required to modify transport phenomena are not observed. The extinction of
combustion in the structure at the top of the flame causes the visible flame height to
drop to the next heighest structure and this process causes the pulsations in the flame
height. The scale of this intermittent flame region A decreases from 80 to 40 percent
of the flame height Zf1 as ZfdD increases from 1 to 10. Finally, the flame height
(defined as the 50 percent intermittency height) is approximately given in terms of a
parameter, QD*=&/~, CpT,d(gD)D2, as:
Typical values are shown in Table 11.
TABLE I1
Flame characteristics
Taken from Kubota (1977).
4 INTEGRAL METHODS OF FIRE PLUMES
During the present work, we reviewed schemes for calculating plume mass flow rates
and made several analytical studies to obtain a feeling for the influence of the burner-
flame configuration on the properties of the buoyant plume in the region near the top
of the flame and the flame region itself. Some of these studies are presented in detail
elsewhere (Cetegen el al., 1982) and a very brief description of the results is given here.
312 B. M. CETEGEN, E. E. ZUKOSKI A N D T. KUBOTA
Consider first the idealized buoyant plume solution described by Morton et at.
(1956). They examined the plume rising above a point source of heat and found a
simple algebraic solution when they made the assumptions: (1) that density differences
across the plume are so small that they need only be considered in the buoyancy term
of the momentum equation (the Boussinesq approximation); (2) that entrainment rate
per unit height is proportional to p, Wmb where p, is the ambient density, Wm is the
velocity on the centerline of the plume, and b is a measure of the plume radius; and
(3) that velocity and density profiles are self-similar.
This weakly buoyant, point source plume produces a mass flux, ma which can be
expressed as:
where
All of these quantities are defined in the list of symbols. Cm is a constant and we have
found that a value of 0.21 brings the predictions of Eq. (6) intogoodagreementwith
the data of a number of experiments including those of Yokoi (1961) and ours.
Later, Morton (1965) relaxed the approximations listed above in a n effort to
account for the effect of large density differences on plume development. He assumed
that the entrainment rate should be modified by multiplying the value for the weakly
buoyant plume by the density ratio function, .\/(pmlp,), i.e., that the rate is propor-
tional to l/(pm/pm)PmWmb. Here, pm is the density on the plume centerline and it
is a function of elevation. In addition, he made what amounted to the Howarth-
Dorodnitsyn transformation in the radial direction, and assumed that the temperature
and velocity profiles could be expressed as self-similar functions in the transformed
radial scale rather than the real radial scale, b. Morton showed that these modifica-
tions have the mathematically useful result that the transformed equations for the
strongly buoyant problem have exactly the same form as the equations for the weakly
buoyant plume.
Given this result, we have shown that the mass flow in the strongly buoyant point
source plumes is also given precisely by Eq. (6). That is, the algebraic form for the
plume mass flux is the same for weakly and strongly buoyant plumes.
Finally consider the strongly buoyant plume with a finite source. Morton (1959)
showed that a solution could be found when the source of the plume was a horizontal
disk of unspecified diameter a t which heat, momentum and mass are added. The
source diameter is fixed by mass and momentum fluxes. Later, Kubota (1977) showed
that this finite source solution could conveniently be put in the form of hypergeometric
functions. This solution has the property that far above the source, it is asymptotic to
a solution for a point source plume with the same heat addition but with an origin
which is vertically displaced from that of the finite source plume. Thus, the far field
mass flux for the strongly buoyant, finite source plume can be expressed as ms[Z+Zvo]
where ma is the function given in Eq. (6), Zvois the offset of the virtual origin below
the elevation of the finite source and Z has its origin a t the elevation of the source.
The magnitude of the offset for the virtual source term Zvocan be found given the
choice of the mass and momentum fluxes a t the source of the plume.
Calculations have been carried out for a number of strongly buoyant plumes with
a finite source. The initial mass and momentum fluxes were selected to crudely model
our fire plumes with the source located a t the time-averaged top of the flame. We also
found the corresponding point source solutions which were asymptotic to these finite
source solutions. Comparison of these solutions showed that near the elevation of the
finite source (i.e. near the top of the flame), the mass fluxes in the two plumes differed
by less than 15 percent.
Thus, these idealized plume calculations suggest that the predictions of Eq. (6) for
the simple point source plume can be used to describe with little error the mass fluxes
in buoyant fire plumes for elevations above the top of the visible flame as long as a
suitable origin can be selected for the point source. In addition, given the mass and
momentum fluxes a t the flame top, we can calculate the value of the offset, Z,,,,.
We have also applied this technique for the strongly buoyant plumes to study the
effects of heat addition on the entrainment process. The problem examined was that
of a finite source, strongly buoyant plume with the additional complication that, in the
energy equation, heat release due to combustion is added over the visible flame region.
The heat release is distributed in an ad hoe way over the visible flame. A number of
arbitrarily selected heat addition distributions were examined for problems in which
experimental values of the flame height and initial values of the fuel mass and momen-
tum flux were used to crudely simulate some of the real fires. Results of several of
these exercises are presented elsewhere (Cetegen er al., 1982) and they showed us that
a weakly buoyant plume with its source located at the burner surface would have
plume mass fluxes a t the elevation of the flame top which were within &30 percent of
the values we obtained for the finite source strongly buoyant plume with heat addition.
In summary, these calculations suggest that the mass flux calculated from Eq. (6)
will give a useful estimate of the flow in a fire plume in the regions above the top of
the flame and also that the heat addition region, which acts as a distributed source,
will cause the effective origin for the point source plume to be displaced only a small
fraction of the flame height from the burner surface. Numerical values of these results
are presented in the following section and previously by Cetegen er al. (1982).
5 PLUME MASS FLUX MEASUREMENTS I N T H E FAR FIELD
Far field experiments were primarily carried out with the mass balance technique
described in the previous section and in detail elsewhere (Zukoski et al.,1980; Cetegen
et al., 1982). The values of plume mass flux were obtained for three positions of the
burner corresponding to approximate interface heights of 1.0, 1.5 and 2.3 m and with
three burners (0.10, 0.19 and 0.50 m in diameter). Heat input rates were varied from
10 to 200 kW and we used two floor configurations (a 2.4-m-square floor and no floor).
The data are presented in Figures 4, 5 and 6 as the ratio of measured mass flow rate,
m, to the ideal value, m3 [Zi] obtained from Eq. (6) versus the interface height, Zr
divided by the average flame height, ZJ~. This form was chosen for the presentation
of the data because the ratio of mass fluxes emphasizes the similarity of measured to
ideal mass flow rates and removes most of the dependence on heat addition and height
above the burner. In addition, normalizing the elevation by the flame height draws
attention to the separation of the region below and above the flame top.
In calculating the ideal value, ms, we used the heat release rate based on the fuel
flow rate and lower heating value of fuel,Qr. This value was used to be consistent with
1 I I
0 - 0.5om.
Y
Y A
a flool
Y no floor
FIGURE 6 Far field entrainment data for 0.50 m diameter burner with and without the 2.4 m
square floor surround.
our near field measurements since we did not measure the heat loss due to radiation
in the near field experiments.
If we consider the data obtained for 0.10-m dia. burner with 2.4-m-square floor
surrounding it, we find that the ratio mp/mslies around a constant value of about 0.85
for Zt/Zfl > 2. This means that the actual entrainment is about 15 percent less than
the value predicted from the point source model without offsets. However, if the heat
release rate used in calculating, ms is reduced by the radiation losses discussed by
Zukoski era/. (1980), the ratio mp/m3would increase by about 13 percent and lie about
a value of 0.96. In the region 1I; Zr/Zfl< 2, entrainment decreases below the point
source buoyant plume correlation indicating that the dependence of entrainment on
interface height and heat release rate changes. The scatter in this representation is less
than 20 percent. The effects of changing the floor surround were too small to be
observed given the accuracy of the experiments.
The same general pattern is observed for the 0.19-m dia. burner. In this case, experi-
ments were performed both with and without the floor surround. It appears that the
effect of the floor surround is to slightly lower the plume mass flux in the upper region
(Zr/Zfi > 2) and the opposite effect is observed in the lower region. The value of
ti~~/m is sabout 1.0 above Zt/Zrl = 2. These values would be around 1.13 if the heat
release after the radiation losses were used to evaluate m3.
Similar effects are observed for the 0.50-m dia. burner. However, for this geometry,
the mass fluxes without the floor are approximately 30 percent higher than those with
the floor. The data with floor surround cluster about a value of i , / t i ~ s= 1.1 which
would be about 1.24 when the radiation losses were taken into account. These values
of Ijzp/m3are considerably larger than the values for the other two burners.
The presentation of data in these figures show that the point source plume with its
316 B. M. CETEGEN, E. E. ZUKOSKI A N D T. KUBOTA
origin at the burner surface does give a reasonable estimate for the far field data.
However, the diameter dependence of the entrainment ratio mp/mg appears to be re-
lated to the height of the heat release zone above the burner. This behavior is not
surprising given the foregoing discussion concerning the effective origin of the buoyant
plume. The effects of the floor surround can also be tied to the proximity of the heat
release region to the floor (low Zft/D). The entrainment for these short fires is much
more affected by the floor geometry as we would expect.
The theoretical analysis discussed above leads us to expect that the heat release
region will produce a shift in the origin of the point source plume which could be an
appreciable distance above or below the burner surface. Several analyses were made
to determine the offsets of the equivalent point source plumes from the far field entrain-
ment data. The first one involved the determination of the constant Cm in Eq. (6) by
a least mean square analysis for the far field data segregated by burner diameter. The
results of this analysis, which are available in detail (Cetegen et a/., 1982), showed that
the entrainment constant lies between 0.21 and 0.23. Since, in the far field of these
fire plumes, it is expected that the entrainment constant Cm would be the same for
different burners regardless of scale or geometry, we decided to use Cm=0.21 as a
reasonable estimate in the following analyses.
As a next step, we investigated the heat release and burner diameter dependences of
offsets, ZO. This was done by plotting the data in a form which allows us to easily
perceive the dependence of entrainment rates on the structure .of the heat addition
region and the elevation from an effective origin of the far field plume. The method
used is based on the observation that we can rewrite Eq. (6) in a form such that a mass
flux parameter depends linearly on elevation,
Thus, if our far field plume does follow the predictions of the simple model [which
leads to Eq. (6)], we expect that
Here, Z is measured from the burner surface and Zo is the offset from the burner sur-
face (not the flame top) required by the effects of the heat addition region. Typical
data plotted in this manner are shown in Figure 7 and are presented in more detail by
Cetegen el a/. (1982). The parameters actually plotted are [mP/@Pl3/5 versus Z.
The slopes of the lines on these plots were calculated from Eq. (8) with Cmy0.21. The
best fits through the data were'made by eye. Representative data are shown for the
0.19-m dia. burners for data obtained without floor and with the mass balance
technique.
The slope of the data do agree remarkably well with that predicted from Eq. (8)
and there is a definite offset for most of the experiments. Values of the offsets for a
large number of experiments were determined in the manner described above, and the
results are presented in Figure 8 where the values of Zo/D are plotted as functions of
Zft/D. The offsets for each burner-floor configuration tend to collect near a particular
value and vary slowly with the Zfl/D ratio. However, the data for all three burner
diameters can be used to define a straight line of the form,
ENTRAINMENT I N FIRE PLUMES 317
DATA WITHOUT FLOOR

WITH MASS BALANCE
0 0 00 05 1.0 1.5 2.0
FIGURE 7 OKset measurement examples for the 0.19 m dia. burner for far field data obtained
without floor. Lines have a slope corresponding to Cm=0.21.
HESKESTAO +=0.8-033
(1W)
Yrn 0
A
V
0:O.lOm
00.19
.50
0.50
MASS BAL.
CHEM.
1
i
A !
f SOLID POINTS WITHOUT FLOOR
/ TT
I
I
f
FIGURE 8 Plume origin offsets as a function of flame height, both normalized by the burner
diameter.
B. M. CETEGEN, E. E. ZUKOSKI A N D T. KUBOTA
with slightly different values of c and d for the burner-floor and burner-no floor con-
figurations. c=O.S, d= -0.33 and c= -0.8, d= -0.33 are reasonable values for these
two configurations. Both of these lines lie close to the offset proposed by Heskestad
(1983)who plotted Zo/D versus Q2/5/Dwhich has the same form in the region QD* > 1 .O.
The offsets decrease sharply as flame height to burner diameter ratio decreases and
they are quite small for QD* < 1.0. However, this does not imply that the offsets for
small QD*can be neglected, since for these short flames the offsets may be appreciable
fractions of the flame height. For example, the plume mass flux a t the flame top
depends on Zll+Zo and thus the ratio Zo/Zfr is the important parameter. For
Q ~ ~ c 0 .2.02 ,and 20.0, the values of Zfl/D are about 1, 4 and 11 and the values of
Zo/Zf,estimated from Figure 8 are about -0.13, $0.13 and f0.24. Hence, for each
of these examples, there will be an appreciable effect of the offsets on the entrainment
a t the flame top. Here, we note that the sign convention used is such that a positive
offset places the effective origin below the burner. The scatter of the data shown in
Figure 8 and the analysis discussed in the previous section suggest to us that the offset,
Zo does not depend simply on Z/JD and that the process is more complex than
suggested here.
We have also calculated values of ZOby using the theoretical approach described
above. Remember that if we know the mass and momentum flux in the plume a t the
flame height, we can compute the location of the virtual plume origin with respect to
the flame top, Zooand hence with respect to the burner location, Zo. We carried out
this exercise for Of=60 kW and D=0.10, 0.19 and 0.50 m using the values of mass
flux and flame height interpreted from our measurements. Since we did not measure
the momentum flux, we estimated this quantity from the mass flux, assumed gaussian
velocity and temperature profiles, and the velocity along the fire plume centerline
measured and correlated by McCaffrey (1979). The results of this computation agreed
well with those presented in Figure 8 but d o depend strongly o n the assumed centerline
velocity.
Close agreement between the calculated and measured sets of virtual origins indicates
that the mean properties of fire plume above the mean flame height may be estimated
with reasonable accuracy by the approximate solution incorporating the assumptions
made in the strongly buoyant plume analysis. It also suggests that the variations with
diameter seen in Figures 4, 5 and 6 are real effects and are not artifacts of the experi-
mental method.
For the sake of developing a general representation of the mass flux computations,
we tentatively suggest that the offset from the burner surface can be expressed as a
function of Zft/D as indicated in Figure 8 and that the following formulae be used to
estimate the plume mass fluxes for ZT Zlr.
zo
Burner withoutjfoor: - - 0.80-0.33-
D D
In summary, despite the variations in the levels of the plume mass flux ratios for differ-
ent burners, the results show that a simple point source model can be used to make
good estimates of plume mass fluxes when the top of the visible flame lies below the
interface between the ceiling layer and cooler layer near the floor. Agreement of
experimental and calculated values within & 15 percent is achieved without any offset
for the origin of the plume or other device used to account for the small values of the
ratio of fire height to burner diameter, the finite vertical extent of the heat addition
region and the large density differences present in the flame zone. A more precise
estimate can be made when the offset defined by Eqs. (10) are used. A value of
Cm=0.21 gives a good representation of our data and is close to that obtained by
Yokoi (1961). The plume mass fluxes measured in the far field in our experiments are
larger by about 25 percent than the data reported by Ricou and Spalding (1961) and
are considerably smaller than the results of several experiments reported by Thomas
er a / . (1965).
In regard to our measurements, several qualifications are necessary and these are
described in more detail by Zukoski el al. (1980). First, we used the heat release rate
of the fuel, Qf (fuel mass flow rate multiplied by the lower heating value of the fuel)
to scale the plume mass fluxes. The effects of radiation loses from the fire are not
negligible and in our experiments we believe that 25 to 30 percent of
radiation. Second, although natural gas was used in most of our experiments, we
of
is lost by
carried out several experiments in which we examined the effects of the fuel heating
value on entrainment process. We could see no systematic dependence on this para-
meter, which was varied by adding nitrogen to the fuel, even when it was changed
from about 10 to 48 MJ/kg.
Third, these plumes were produced in as quiet an atmosphere as we could maintain
in the laboratory and we have shown (Zukoski eta/., 1980) that small ambient disturb-
ances can produce 20 to 50 percent increases in plume mass flows. Similar and more
extensive observations were subsequently made by Quintiere et al. (1981) who obtained
2- to 3-fold increases in entrainment over the undisturbed values for a variety of
door-window opening configurations.
6 PLUME MASS FLUX MEASUREMENTS I N T H E NEAR FIELD
Entrainment measurements in the visible flame region were made by chemical analysis
of combustion products in the ceiling layer as described in Section 2. The plume mass
fluxes have been measured for three burners (0.10, 0.19 and 0.50 m in diameter) and
heat input rates of 10 to 100 kW without the floor surround. In these experiments, no
direct measurement of the flame height has been made, but rather the estimates made
by observers looking up into the small hood a t oblique angles suggested that the flame
lengths in the hood were not greatly different from the free flame heights reported by
Zukoski e t a / . (1984). In some of these experiments (large heat inputs and small bur-
ners, large Q D * ) the flames were observed to impinge on the hood intermittently.
Certainly, the hot vitiated air did not produce increases in flame height by more than
20 percent.
Data obtained in a typical experiment are presented in Table I1 where we list values
of the fuel-air ratio of the gas which entered the hood given as a fraction of the
stoichiometric value, d h , with the heat release rates, o f ; the plume mass flux, r n , a t a n
interface height of Zg z 0.33 m and the ratio of interface height to free flame height,
Zi/Z1l, for the 0.19 m burner. The values of +h were determined from the gas sample
B. M. CETEGEN, E. E. ZUKOSKI AND T. KUBOTA
TABLE I11
Entrainment and Equivalence ratios for the
0.19 rn burner
data and they increase from about 10 percent a t the lowest heat input rate to almost
70 percent a t 81 kW. If the residence time for the gas in the hood, tra, is defined to
be the mass of gas in the hood divided by the plume mass flux into the hood, we find
that the residence time varies from about 20 sec at 10 kW to about 10 sec a t 81 kW.
This residence time is certainly long enough to insure that the composition of the gas
in the hood will have reached a steady state even if it is not in equilibrium. The most
important observation from Table Ill is that the plume mass flux is a very weak func-
tion of the heat release rate of the fire when the interface height, Z; is small. This
important result will be discussed later in detail and will form the basis of an ad hoc
model for entrainment.
In all the near field experiments, the temperature of the walls of the hood and gas
temperature were not controlled and the walls of the small hood were not insulated.
Gas temperatures in the hood ranged from 500 K to 750 K and combustion efficiencies
were high (see Cetegen et al., 1982). We kept the equivalence ratio of the hood gas, +r,
below 0.70 to prevent soot formation.
The plume mass fluxes, mp normalized by the theoretical value for the far field,
mi)g[Zi]as a function of interface height, Zt normalized by Z,l are shown in Figures 9,
10 and 11 for three burners with a range of heat release rates. (The lines on these
figures are sketched roughly through the data for a given heat release rate.) As
expected, the entrainment rates deviate substantially from the far field correlation a t
low elevations. Examination of data for all three burners reveals the fact that the
plume mass flux is relatively insensitive to the heat release rate of the fire, Qf in the
initial region. Even though the heat inputs change by a factor of 3 to 4 in this region,
the change in entrainment is often within the accuracy with which we can make our
plume mass flux measurements. Another observation is that the comparison of the
entrainment data for these three burners suggests that plume mass fluxes vary almost
linearly with the burner diameter. These observations together with the appearance
of these diffusion flames in the region close to the burner surface led us to consider
the simple ad hoc model for the flame which is discussed in the following section.
6.1 Simple Model for Initial Region

I n this simple model, the initial region is viewed as a highly wrinkled, thin, laminar
diffusion flame surrounding the burner perimeter. The diffusion flame separates air
on one side from fuel on the other and is in a very unsteady flow field. This picture
described above may seem to be unreasonable especially for large fires. However, the
Reynolds number based on fuel flow properties is proportional to heat release rate, 01
divided by the burner diameter, D. Thus a 126-kW fire on 0.50 m diameter burner
ENTRAINMENT IN FIRE PLUMES
FIGURE 9 Plume mass flux normalized by the far field correlation equation (6) as a function
of interface height divided by the flame height for a range o f heat release rates on 0.10 m diameter
burner.
0
a
N x I I I 6,.w
z 10.0 0 61.0
0 19.0 A 73.5
0
131.5 +80.5 -
0 41.0
a t # -
0
yl
-
a
0
a I I I
0.0 0. SO 1.00 1.50 2.00
=i / Z j l
FIGURE 10 Plume mass flux normalized by the far field correlation equation (6) as a function
of interface height divided by the flame height for a range of heat release rates on 0.19 m diameter
burner.
B. M. CETEGEN, E. E. ZUKOSKI AND T. KUBOTA
FIGURE I 1 Plume mass flux normalized by the far field correlation equation (6) as a function
of interface height divided by the Rame height for a range of heat release rates on 0.50 m diameter
burner.
FIGURE 12 Definition of scales in the initial and turbulent flame regions.

will have a n initial Reynolds number of less than 500 and so will a 1 MW fire on a
4-m burner. Jets at this Reynolds number are laminar for a t least a number of initial
diameters downstream of the jet source. Also, the photographs and shadowgraph
images of the flames close to the burner surface, show that the small scale turbulence
(small compared to diffusive thickness of laminar flame sheet) are not present. See
more complete discussion of these points in Zukoski e l al. (1984).
In a laminar diffusion flame, air and fuel enter the region adjacent to the flame by
molecular diffusion processes and are heated by the reaction in a thin flame front and
rise due to natural convection as shown in the sketch of Figure 12a. All transport
processes are assumed to be molecular. Because the temperature of the gas a t the
flame itself must be close to the adiabatic flame temperature, we can assume that the
region near the visible flame has a n average temperature which is a constant fraction
of the adiabatic flame temperature. Thus, the average temperature and density in the
flame zone are independent of height 2.
In this simple model, we need to estimate the surface area of the laminar diffusion
flame surrounding the buoyant fuel jet. The pictures and the video recordings of our
flames suggest that the surface area of the flame scales like the product of the burner
perimeter, n D and the height above the burner surface, Z. Flame width measurements
presented by Zukoski et al. (1980) and Cetegen et al. (1982) indicate that the widths
are of the order of the burner diameter and subsequently grow larger. Also, the
minimum diameter of the flame near the burner is never much smaller than the burner
diameter. Furthermore, large scale vortical structures are present in these flames and
these structures and the smaller scale wrinkling of the flame surface could easily
increase the surface area estimated above, n D Z by a large factor. These observations
are in contrast with what we expect for the shape of a steady, buoyant jet. In this
problem, the jet fluid accelerates due to its buoyancy and hence its diameter must
contract sharply with height. We believe that the differences between the observations
and the steady solution are due to periodic shedding of large scale vortical structures
described above and by Zukoski et al. (1980) and Cetegen et al. (1982).
In the initial region, the momentum balance for the thin flame sheet (shown in
Figure 12a) of width, S and horizontal length proportional to burner diameter, D is
d
-(6 Dpm Wm2)a Apg SD
dZ
where we characterize the whole flame region by top hat profiles for velocity, Wmand
density, pm as well as temperature and density differences (AT and Ap). Also, we
ignore the momentum contribution of the entrained air and fuel.
The energy balance is,
where ha is the heating value per massofair and 22 is the molecular diffusion coefficient.
We have taken 916 as a scale for the velocity at which oxidizer is entrained by the
diffusion flame. Note that only two equations are required here, because the maximum
temperature in the layer is fixed by the chemical reaction and is constant.
324 B. M. CETEGEN, E. E. ZUKOSKI AND T" KUBOTA
If we assume that horizontal length scale is constant and equal to the burner diam-
eter, D, and the average density and temperature in the flame region are independent
of Z, we can show that a similarity solution exists when,
Wm ex. Zl/2 and S ex. Zl/4 (13)
Given these results, the mass flux at height Z, rill, is proportional to
(14)
and in terms of the constants, it can be written as,

(15)
where g' = [(Pa) Pm) --1 ]g. Note that the plume mass flux in this region does no t
depend on the total heat release rate of the fire ( Qf or Zfl).
6.2 Turbulent Flame

The initial region described above later develops into an intermediate region which
contains most of the intermittent flame zone. This latter zone is important since it
comprises a vertical distance which in some cases may be a large fraction of the flame
height. However, the transtition between the two regions is not clearly defined by
photographic evidence and is only clearly defined in the entrainment data. The corre-
spondence between the three regions we use here and those defined by McCaffrey
(1979), on basis of temperature or velocity, needs to be clarified by further work.
In this region, we: assume that air consumed by combustion (or heat released) per
unit height in the turbulent flame is a constant fraction of the rate of ambient air
entrainment per unit height at that height. This hypothesis ensures that average tem-
peratures and densities in the buoyant flame plume remain constant. It is used as a
matter of convenience: and is in rough agreement with experimental evidence of
McCaffrey (1979) only for the lowest part of the flame. Following the treatment of the
:!fiT = W m1] (16)
where 1] is the diameter of the flame as shown in Figure 121:>. We assume that the mass
is entrained per unit height at a rate given by,
(17)
and that a constant fraction of this mass is burned in the volume,
(18)
ENTRAINMENT IN FIH..E PLUMES 325
Then, conservation of momentum and energy become,
(19)
(20)
where K is the fraction of the entrained mass consumed in combustion. Since we

assume that !:J.T is fixed by the heat release process, only two equations are required.
A similarity solution of these equations exists if
1] ocZ and Wm OCZ1/ 2 (21)
The corresponding mass flux is,
7T1]2
m2 OC pm - - Wm OC Z5/2 (22)
4
and given the constants it: can be expressed,
(23)
The only difference between this region and the far field is that the combustion stops
at the flame top and the temperature decreases with height.
The solutions for the three regions must now be matched at a llower transition point
(t) and at the top of the flame (f). Matching conditions at (f) are simple and logical
since we must conserve mass, momentum and enthalpy and because the only difference
between the two regions (in this simple model) is that the combustion stops at (f) and
consequently !:J.T becomes: a function of height. The solution at (f) suggests that both
regions will have the same offset.
These matching conditions require that the mass flux in the turbulent flame region,
m2 and the far field, rn3 be given by:
(24)
(25)
Here, Z/l is the vertical distance to the time-averaged top of the flame in the coordinate
system used for regions (2) and (3), and em has a value near 0..21.
If we normalize the mass flux in turbulent flame, m2 by the far field correlation, we
get:
(26)
326 B. M. CETEGEN, E. E. ZU][(OSKI AND T. KUBOTA
and thus the ratio does not depend explicitly on heat release rate or burner diameter.
This result is shown in Figure 13 as the curve (A) for Zt/Zjz < ZjZjZ < 1, where Z, is
the height at which the transition to the turbulent flame occurs. A zero offset is used
here.
The transition at (t) between the initial region and the turbulent flame, discussed
earlier, is not clearly defined in our discussion. The plume mass flux measurements
suggest that there is a change in the behavior of entrainment which occurs at elevations
between 0.6 and 1.0 m. But, the scatter in our data prevents us from making any
strong statements concerning this transition. However, if we neglect, for the moment,
the offsets in the vertical axis of the solutions, we find that the ratio of mass flux in the
initial region, ml to that in the far field, m3 is
ml .
_.- IX DZ-ll/12 Q-l/3 (27)

m3
The lines corresponding to Eq. (27) are sketched on Figure 13. A family of curves is
obtained because of the dependence on both D and Q.
Also if we equate the mass fluxes for the initial region and the turbulent flame at the
transition height, Zi, we: get (ignoring offsets)
Zt D4/7
Z, IX D4/7 or - - IX - - - - when Qn* > 1.0 (28)
ZfZ Q2/5
Thus, the transition height becomes a larger fraction of the flame height as the diameter
increases and the heat release rate decreases. For small heat release rates and large
diameter, Zi may increase until it is of the same order as or larger than Zn. Under
,/0
(3)
o L ----L ~
o t 1.0 ZJZn
FIGURE 13 Dependence of plume mass flux in three regions. Regions (1), (2) and (3) are
present for example (A); and regions (1) and (3) for (B).
these conditions, the turbulent flame zone may not be present and the matching
should be done at the flame top between the initial region and the far field. This is
shown as curve (B) in Figure 13 where numbers attached to the segments of the curves
correspond to the three regions described above.
The simple model developed above leads to a qualitative representation for the
plume mass flux measurements and a rational matching for the upper two regions
(turbulent flame and far field). The comparison of the model with the experiments in
the next section will show that even though the general trends predicted by the model
agree with the experiments, the transitions among the three regions are not sharply
defined. In an attempt to test the quantative predictions of the model, we calculated
the mass fluxes for turbulent flame, ril'!, and far field, m3 by using Eqs. (24) and (25)
which were corrected by the offsets as suggested before in Eq. (10). The comparisons
of the model computations with the 0.19 m burner entrainment data showed that a
better fit was obtained when the matching was made at the flame top Zfl+~/2 rather
than at Zfl. Similarly, the constant in Eq. (15) was chosen to make this equation fit
the experimental data obtained with the 0.19-m dia. burner for which the initial region
is most apparent. Finally, the transition between initial region and the turbulent
flame or initial region and far field is allowed to occur whenever the calculated mass
flux for initial region or far field exceeded that for initial region at the same height.
These model computations are shown in Figures 14, 15 and 16 together with the
experimental data. Here" we note that no further adjustment of the constant in the
initial region has been made for different burners.
6.3 Model Comparison with Experiments
Experimental data for plume mass fluxes are presented in several forms in Figures 9 to
11 and 14 to 16. In the first three of these Figures 9, 10 and 11, the plume mass flux,
mp normalized by the far field correlation (6) is shown as a function of interface height
divided by the flame height, Zi/Zfl. Even though the O.IO-m dia. burner data do not
clearly show the initial and the turbulent flame: regions, the data of the 0.19-m dia.
burner shows these regions for 20,40, 60 and 80 kW data. The 20 kW data exhibit
the behavior where the initial region directly approaches the far field. This happens
at a value of Qo" of about unity. Here, it is also apparent that the transition from the
turbulent flame region to the far field spans large: distances. This picture is similar to
Figure 13 which shows the model predictions. Note that in the turbulent flame region
the data do roughly collapse onto a single curve on these plots. The data for the
0.50-m dia. burner show the: transition from the initial region to the far field for 20,
30, 40 and 60 kW fires. For all of these heat release rates, the values of Qo" are less
than unity. Even though the scatter in the data and the simplicity of the model pro-
posed do not allow us to make precise, quantitative estimates, the trends observed
from the experiments are in agreement with the simple model.
Figures 14, 15 and 16 show the plume mass flux, mp as a function of interface height,
Z, with the heat release rate of the fire, Ql as a parameter. The initial region and the
transition between it and the turbulent flame region are clearly defined in these plots.
Comparison of the three sets of data shows that, in the initial region, mp scales roughly
with the burner diameter and the plume mass fluxes are independent of the fuel flow
rate. The values of n"zp for the O.IO-m dia. burner in the initial region are higher than
the model predictions. This result may be connected with the much larger initial
Reynolds numbers and smaller Richardson numbers obtained 'with this burner for a
B. M. CETEGEN, E. E. ZUKOSKI A N D T. KUBOTA
FIGURE 14 Plume mass flux as a function of interface height for various heat release rates
shown together with model computations for 0.10 m diameter burner.
shown together with model computations for 0.19 m diameter burner.
shown together with model computations for 0.50 rn diameter burner.
given heat addition rate, as compared with the corresponding values obtained with the
0.19 and 0.50-m dia. burners.
7 CONCLUDING REMARKS
The mass flow rates measured in these experiments were the total mass flux into a
stratified layer of hot gas due to a fire plume and thus may be larger than the fire
plume mass flux itself.
Entrainment measurements have been made in buoyant fire plumes which model
conditions we expect to find in large accidental fires in which initial fuel momentum
can be neglected. We found that three regions of the fire plume must be considered :
An initial region near the burner, an intermediate zone, and the far field region which
is nearly adiabatic. In the initial region, molecular diffusion appears to be important,
and entrainment is almost proportional to the burner diameter and is independent of
the total heat release rate of the fire. In the turbulent flame region, the plume mass
flux depends on both heat release rate and elevation above the fire source and entrain-
ment appears to occur by the same mechanism as in a turbulent plume. A simple
model was developed to describe the entrainment process in these two regions. The
model fits our data with an accuracy which makes it useful in models of fire spread in
buildings and it may be useful for predicting entrainment rates in fires much larger
than those we examined experimentally.
Finally, in the adiabatic far held region, we showed that a point source plume model
gave an accurate prediction for entrainment rates when appropriate offset was used.
The magnitude of the offset can be related to the parameters of the heat release region
by use of an integral model for strongly buoyant plumes with distributed sources.
ACKNOWLEDGEMENTS
This work was supported through grant G8-9014 from The Center for Fire Research of The
Na~ionalBureau of Standards, U.S. Department of Commerce. The authors are thankful to
Drs. .I.
Rockett and J. G. Quintiere of the Fire Research Center who gave valuable support.
NOMENCLATURE
physical half-width
plume mass flux constant in Eq. (6)
specific heat at constant pressure

burner diameter
molecular diffusion coefficient
gravitational constant
modified gravitational constant, (Ap/pg)
heat of combustion per unit mass of air
heat of combustion per unit mass of fuel
mass flux into the ceiling layer
mass flux entrained by the plume, see Figure 1
mass flux contribution a t the interface, see Figure I
measured plume mass flux, see Figure I
mass flux due to plunging of the plume, see Figure 1
computed mass fluxes.in initial, turbulent flame and far field regions
respectively
pressure
heat release rate based on mfh,
Dimensionless heat addition parameter, ( Q j / p , c p ~ , d ( gD ) D 2 ]
as above replacing D with Z
as above replacing Z with Zfl
Reynolds number, pf Vf Dlpf
Richardson number, ( ~ - - ~ f D/pf
) g Vf2
temperature
temperature difference
fuel velocity a t the burner surface
velocity in z direction
mole fraction
height above the burner surface
interface height
average flame height
offset height below the burner surface, virtual plume origin
offset height below the average flame height, Zo=Zvo-Zf1
Greek symbols
S boundary layer thickness
K fraction of entrained mass consumed in combustion, see Eq. (20)
P dynamic viscosity
v kinematic viscosity
ENTRAINMENT IN FIRE PLUMES
P density
Ap density difference
4 equivalence ratio (fuel-air ratio divided by the stoichiornetric value)
Subscripts
c ceiling layer quantities
E entrained
f fuel
fl flame
h hot or hood values
m mean or maximum
P plume quantities
T turbulent
co values at infinity
REFERENCES
Cetegen. B. M., Zukoski, E. E., and Kubota, T. (1982). Entrainment and flame geometry of fire
plumes, Report to Center for Fire Research, NBS, Daniel and Florence Guggenheim Jet
Propulsion Center, California Institute of Technology, Pasadena, CA.
Cox, G. and Chitty, R. (1980). A study of the deterministic properties of unbounded fire plumes.
Combustion and Flame 39, 191.
Heskestad, G. (1983). Personal communication.
Kubota, T. (1977). Turbulent buoyant plume in stratified media, Report t o Center for Fire
Research, NBS, Daniel and Florence Guggenheim Jet Propulsion Center, California Institute
of Technology, Pasadena, CA.
McCaffrey, B. J. (1979). Purely buoyant diffusion flames: some experimental results, NBSIR79-
1910, National Bureau of Standards, Department of Commerce, Washington, DC.
Morton, B. R. (1959). Forced plumes. J. Fluid Mech. 5, 151.
Morton, B. R. (1965). Modelling fire plumes. Tenth Symposium (International) on Combustion,
Academic Press, New York, p. 973.
Morton, B. R., Taylor, G. I., and Turner, J. S. (1956). Turbulent gravitational convection from
maintained and instantaneous sources. Proc. Roy. Soc. A234, 1.
Quintiere, J. G., Rinkinen, W. J., and Jones, W. W. (1981). The effect of room openings on fire
plume entrainment. Comb. Sci. and Tech. 26, 193.
Ricou, F. P., and Spalding, D. B. (1961). Measurement of entrainment by axisymmetrical tur-
bulent jets. J. Fluid Mech. 11, 21.
Thomas, P. H., Baldwin, R., and Heselden, A. J. M. (1965). Buoyant dimusion flames: some
measurements of air entrainment, heat transfer and flame merging. Tenth Symposium (Inter-
national) on Combustion, Academic Press, New York, p. 983.
Yokoi, S. (1959). The use of models in fire research, publication Nr. 786, pp. 186-206, National
Academy of Sciences, National Research Council, Washington.- D C (1961). See also: R e ~ o r t
29, ~ a p a n e s eBldg. Res. Inst.
Zukoski, E. E., Kubota, T., and Cetegen, Baki (1980-81). Entrainment in fire plumes. Fire Safety
Journal 3. 107.
Zukoski, E. E., Cetegen, 8. M., and Kubota, T. Visible structure o f buoyant diffusion flames,
Submitted to Combustion and Flame.

Cetegen, B. M., Zukoski, E. E., & Kubota, T. Entrainment in The Near and Far Field of Fire Plumes

Uploaded by

Copyright:

Available Formats

You might also like

Cetegen, B. M., Zukoski, E. E., & Kubota, T. Entrainment in The Near and Far Field of Fire Plumes

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Cetegen, B. M., Zukoski, E. E., & Kubota, T. Entrainment in The Near and Far Field of Fire Plumes

Uploaded by

Copyright:

Available Formats

This article was downloaded by: [Cornell University Library]

On: 15 November 2014, At: 05:54

Combustion Science and Technology

Entrainment in the Near and Far Field of Fire Plumes

To link to this article: http://dx.doi.org/10.1080/00102208408923794

PLEASE SCROLL DOWN FOR ARTICLE

Entrainment in the Near and Far Field of Fire Plumes

B. M. CETEGEN. E. E. ZUKOSKI and T. KUBOTA Karman Laboratoryof

2 EXPERIMENTAL TECHNIQUE AND APPARATUS

The terms ( t ; ~ ~ + t A ~ + mare

FIGURE 1 Entrainment contributions into the ceiling layer.

FIGURE 2 plume mass flux measurement technique.

then for dried products,

8.52 Yo, 1- 8.52 Yco,

FIGURE 3 Apparatus for chemical sampling.

3 FLAME STRUCTURE REVIEW

Typical values are shown in Table 11.

Taken from Kubota (1977).

4 INTEGRAL METHODS OF FIRE PLUMES

5 PLUME MASS FLUX MEASUREMENTS I N T H E FAR FIELD

DATA WITHOUT FLOOR

0 0 00 05 1.0 1.5 2.0

6 PLUME MASS FLUX MEASUREMENTS I N T H E NEAR FIELD

6.1 Simple Model for Initial Region

FIGURE 12 Definition of scales in the initial and turbulent flame regions.

Wm ex. Zl/2 and S ex. Zl/4 (13)

Given these results, the mass flux at height Z, rill, is proportional to

and in terms of the constants, it can be written as,

6.2 Turbulent Flame

:!fiT = W m1] (16)

and that a constant fraction of this mass is burned in the volume,

Then, conservation of momentum and energy become,

where K is the fraction of the entrained mass consumed in combustion. Since we

1] ocZ and Wm OCZ1/ 2 (21)

The corresponding mass flux is,

and given the constants it: can be expressed,

_.- IX DZ-ll/12 Q-l/3 (27)

6.3 Model Comparison with Experiments

specific heat at constant pressure

You might also like