International Journal of Thermal Sciences 75 (2014) 87e94

Contents lists available at ScienceDirect

International Journal of Thermal Sciences

journal homepage: www.elsevier.com/locate/ijts

Ventilation effects in confined and mechanically ventilated fires

J. Lassus a, L. Courty a, *, J.P. Garo a, E. Studer b, P. Jourda c, P. Aine d
a
Institut P’, CNRS, ENSMA, Université de Poitiers, 1 Av. Clément Ader, Téléport 2, BP 40109, 86961 Futuroscope Chasseneuil, France
b
Commissariat à l’Energie Atomique, DEN/DANS/DM2S/SFME/LTMF, CEA de Saclay, 91191 Gif-sur-Yvette Cedex, France
c
Commissariat à l’Energie Atomique, CEA/PMR/DPSN, CEA de Fontenay-aux-Roses, BP 6, 92265 Fontenay-aux-Roses Cedex, France
d
AREVA NC, 33 Rue la Fayette, 75442 Paris Cedex 09, France

a r t i c l e i n f o a b s t r a c t

Article history: The main objective of this work is to study underventilated and highly underventilated fires in a
Received 12 February 2013 compartment equipped with a mechanical ventilation network. Heptane and dodecane pool fire ex-
Received in revised form periments are performed in a reduced-scale room. A changing-scale analysis is performed in order to
19 June 2013
obtain species concentrations and temperature levels at a larger scale. Carbon dioxide concentration
Accepted 19 July 2013
Available online 31 August 2013
increases linearly with the decreasing of oxygen concentration. Heat release rate depends on ventilation
condition in the compartment and we can distinguish three cases. The first one corresponds to suffi-
ciently ventilated fires, where heat release rate is higher or equal to the one of an open fire and where the
Keywords:
Underventilated fires
reaction is almost complete. The second one includes underventilated fires for oxygen concentration
Froude scaling law ranging between the Minimum Oxygen Concentration of the given fuel and around 7% (this value cor-
Ventilation effects responds to the Minimum Oxygen Concentration of carbon monoxide). At this percentage of oxygen, the
Heat release rate Heat Release Rate is lower than the one of an open fire and the carbon monoxide production is pro-
Heptane portional to oxygen concentration. The third one corresponds to very underventilated fires for oxygen
Dodecane concentration lower than 7%, and where the fire can stop due to a lack of oxygen. In this case, carbon
monoxide concentration increases strongly with the decrease of oxygen concentration.
Ó 2013 Elsevier Masson SAS. All rights reserved.

1. Introduction experiments requires the confirmation of the validity of the simi-

Many investigations were conducted on unburnt gas production
during compartment fires. This production depends on the amount
of oxygen supplied. Carbon monoxide and oxygen concentrations
produced during fires have been measured by several authors
(Beyler [1], Tewarson [2], Morehart et al. [3], Gottuk [4] and Utiskul
[5] among others). Pitts [6] has summarized a large number of in-
vestigations conducted for carbon monoxide production. The
transport of exhaust gases to another compartment has also been
investigated (Lattimer et al. [7], Wieczorek [8], Lassus [9]). In many
of these investigations, available oxygen is considered using global
equivalence ratio. Beyler has shown that the carbon monoxide level
generated during a fire is correlated to fuel type [1]. The main
objective of the present work is to study the ventilation conditions
during a fire in a reduced scale room, and especially the production
of some species, such as carbon monoxide and carbon dioxide. As
pointed by Viskanta [10], such experimental data are important to
increase the efficiency of Computational Fluid Dynamics
(CFD)-based fire dynamics models. The use of laboratory-scale
larity between the original full-scale compartment and the scaled
model. In this paper, scaling laws are proposed for pool fires in a
compartment. Fire investigations have been previously conducted
in full scale compartments, such as the International Collaborative
Fire Model Project (ICFMP) conducted by Rowekamp et al. [11], Full-
Scale Enclosure (FSE) fires performed by Pitts et al. [12] in order to
extend the earlier reduced-scale enclosure (RSE) study carried out
by Bryner et al. [13] and FLIP test presented by Melis and Audouin
[14] for the study of vitiation effect on Heat Release Rate (HRR). A
comparison between these full-scale fires and the reduced-scale
room fire performed in this work is also given. The aim of the pa-
per is to study the validity of scaling laws and to check the validity
of our enclosure to reproduce the behavior of bigger enclosures. A
description of the different combustion regimes is also given. Sec-
tion 2 is devoted to the derivation of the scaling relations. Material
and methods are presented in Section 3 and the results and dis-
cussions in Section 4.
2. Scaling relations

* Corresponding author. Tel.: þ33 549 366 087. Investigations on fire compartment require experiments. It
E-mail address: leo.courty@ensma.fr (L. Courty). is often necessary to quantify temperature, pressure, species

1290-0729/$ e see front matter Ó 2013 Elsevier Masson SAS. All rights reserved.
http://dx.doi.org/10.1016/j.ijthermalsci.2013.07.015
88 J. Lassus et al. / International Journal of Thermal Sciences 75 (2014) 87e94

Nomenclature Q_ heat release rate (kJ s1)

Re Reynolds number
ACPH air change per hour S fuel pan surface area (m2)
cp specific heat (J kg1 K1) t time (s)
D fire diameter (m) T temperature (K)
Fr Froude number u gas velocity (m s1)
g acceleration of gravity (m s2) l thermal conductivity (J m1 K1 s1)
h fuel height in the pan (m) d thermal penetration distance within wall (m)
hconv heat transfer coefficient (J m2 K1 s1) DHc heat of combustion (J kg1)
L characteristic length (m) 3 emissivity
m _ 00 burning rate per unit area (kg m2 s1) r gas density (kg m3)
MOC Minimum Oxygen Concentration of fuel (% vol.) s StefaneBoltzmann constant (J m2 K4 s1)
q_ cond conductive heat flux (J s1) f global equivalence ratio
q_ conv convective heat flux (J s1) s flow time scale
q_ rad radiative heat flux (J s1) m dynamic viscosity (N s m2)

concentration and to consider parameter influences such as hconv

ventilation flow or HRR. Two cases of fire tests can be considered: Pconv ¼ (3)
r0 $cp $g1=2 $l1=2
r
fire tests in full-scale room and in reduced-scale room. Experiments
in full scale are expensive, that limit the numbers of tests. It may be In order to preserve this number, heat transfer coefficient must
1=2
difficult to analyze results in these conditions. With reduced-scale ensure hconv flr , the convective heat flux scaling is therefore
5=2
room, more tests may be carried out but the scale reduction must obtained if q_ conv flr . This condition is automatically obtained
respect several scaling laws adapted to the studied problem. In with HRR scaling.
many works, Reynolds number analogy is used to insure similarity The conduction term can be scaled if a condition is applied on
between the prototype and the full-scale compartment. However, the wall thickness. Indeed, from Fourier equation and Zukoski
fire requires the preservation of too many groups to ensure com- number, another dimensionless term appears:
plete similarity. Partial scaling is made by establishing dimen-
sionless variables from the conservation equation. Scaling relations lwall
for compartment fires have previously been proposed by Heskestad Pcond ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
  (4)
1=2
[15], Emori and Saito [16], Quintiere [17] and others. This work is s$ r$cl p $r0 $cp $g 1=2 $lr
wall
based on the one of Quintiere [10] and a particular attention is pffiffiffiffiffiffiffiffiffi
devoted to heat loss. Combined with the scaling laws for flow time ðsf lr =g Þ, this
3=2
With constant temperature and pressure from small scale to full relation can be preserved if ðl$r$cp Þwall flr . Except specific heat
scale, Reynolds and Froude numbers cannot be preserved simul- of steel, specific heat of solids are nearly constant. Thus, the
taneously. In this study, fire compartment tests have been con- proportionality cpwall fl0r appears. Assuming that thermal con-
ducted at a smaller geometric scale, maintaining the Froude ductivity and gas density of a material have similar trends, we
3=4
number, Fr ¼ u2/gL, constant. Reynolds number is large enough to obtain: lwall frwall flr . Conductive heat flux scaling requires the
ensure the turbulence of the flow. The Froude number is then made preservation of wall thickness by the following dimensionless
1=2
constant by ur ¼ lr that comes from dimensionless variable of term:
momentum conservation. Flow time scale is obtained from mass
1=2  1=2  1=4
conservation bysflr =ur flr . Since the ventilation flow rates dwall lwall g
(ACPH) is the inverse of s, it scales with l1/2. Pwall ¼ ¼ $ $dwall (5)
d rwall $cpwall lr
Pool fire is a phenomenon with a low Mach number that has a
nearly constant pressure. By neglecting pressure variation, the en- where d is the thermal penetration distance within wall.
ergy equation (1) in mono-dimensional form can be expressed by: In order to preserve this number, the following proportionality
relation on wall thickness must be obtained:
" # 000
l vT~ vT~ m l v2 T~ Q_ $lr
r$ r $ ~ þ u$
~ ¼ $ $ þ (1)
ur $s vt v~x r0 $ur $lr m$cp v~x2 r0 $ur $cp $T0 l1=2 $l1=4
r
dwall f  (6)
The last term, which corresponds to the source term, contains rwall $cpwall 1=2
the Zukoski number PZu ¼ Q_ =r0 $Fr $lr $g 1=2 $cp $T0 [18] that can
5=2
Using the proportionality relation of thermal conductivity, the
lead to the heat release rate scaling by preservation of the termQ2/ 1=4
gas density and Eq. (6), we get the requirement: dwall flr . If this
L5. Hottel [19] with the results of Blinov and Khudiakov [20], 5=2
term is preserved, conductive flux scaling is scaled with: q_ cond flr
Thomas et al. [21] and Yumoto [22] concluded that flame radiation
and this condition is obtained with HRR scaling.
plays the most important role in fuel combustion.
Assuming that walls are black bodies, radiative heat flux can be
Heat can be exchanged by radiation and convection at the
expressed with the StefaneBoltzmann law:
bounding surfaces. Convective heat flux can be expressed by:
 
q_ conv ¼ hconv $S$ðT  Twall Þ (2) q_ rad ¼ 3 $s$S$ T 4  T04 (7)

From Zukoski number and the preservation of the Froude number, a Using Zukoski number and the preservation of Froude number, a
dimensionless term appears: dimensionless term appears:
 J. Lassus et al. / International Journal of Thermal Sciences 75 (2014) 87e94
3$s$T03
Prad ¼
r0 $cp $g1=2 $l1=2
r release rate.
Radiative heat flux scaling requires therefore the preservation of
1=6
this term, that can lead to T0 flr . However, this term cannot be
preserved. It results that radiative heat flux does not scale. In order
to obtain good agreement between the prototype and the full scale
compartment, radiative heat flux from wall must not have an in-
fluence on HRR. Consequently, the term Prad (Eq. (8)) must be
negligible with regard to Zukoski number for HRR obtained from
energy equation (Eq. (1)).
3. Materials and methods
In order to reduce the cost of fire tests and to be able to carry out
fire tests more reproducible than fire in a compartment of high size,
a reduced-scale room has been built. This experimental device is a
reduced-scale room based on a 100-cubic-metre compartment,
which is cubic and whose wall thickness is 0.3 m. The overall di-
mensions of this reduced scale room are scaled geometrically from
those of the full scale compartment, resulting in a 8-cubic-metre
room test. A picture and a scheme of the experimental setup are
presented in Fig. 1. In order to preserve conductive heat flux, walls
have a thickness of 0.2 m. Indeed, in order to preserve the
conductive flux, we have to take the wall thicknesses into account.
The term (lwall/rwall$cp,wall)1/2$(g/lr)1/4$dwall must be preserved (cf.
[24]). This term depends on the nature of the material used to build
the enclosure and on the thickness of this material. Walls of the
full-scale enclosure are made of concrete and are 0.3 m thick. Ac-
cording to Quintiere, concrete material has a thermal conductivity
between 0.8 and 1.4 W m1 K1, a density between 1900 and
2300 kg m3and a specific heat of 0.88 kJ kg1 K1. If we take the
maximum of these values, the term written above takes the value
17. For a reduced-scale enclosure in cellular-concrete material, this
term is preserved for walls of a 0.2 m thickness.
Walls are made of reinforced concrete. Scaling HRR imposes the
2
preservation of the quantity Q_ =L5. The powers tested are lower
than 130 kW, which corresponds to a maximum HRR of 1 MW at
full scale. The radiative heat flux from wall must not have an in-
fluence on heat release rate. To ensure this condition, a comparison
is made for different heat release rates presented on Table 1 and the
value Prad ¼ 0.00017 is obtained.
For a fire in the reduced scale compartment of 75 kW, the effect
of radiative heat flux on HRR can be considered insignificant.
A mechanical ventilation network is used. It delivers a flow of 24
or 40 m3 h1 (Air Changes Per Hour (ACPH): 3 or 5). Other venti-
lation flows are tested to study ventilation influence on HRR,
temperatures and species. Ducts have a square section of 0.2 square
Fig. 1. Photograph and scheme of the reduced-scale fire compartment with instrumentation.
89
Table 1
(8) Dimensionless term PZu as a function of heat
Q_ (kW) PZu
5 0.0001
10 0.0002
20 0.0004
30 0.0006
50 0.001
75 0.0015
100 0.0019
meters. Inlet duct is located at the bottom of the room test at a
0.3 m height and extraction network is located near the ceiling, at a
1.7 m height. Acquisition time (temperature, gas velocities, gas
concentrations) is set at 6 s.
Experiments are performed with heptane and dodecane pool
fires. The stainless-steel pans are located in the center of the room
test, at a height of 0.5 m, and placed on a load cell to measure fuel
consumption as a function of time. Heptane used contains 71% of n-
heptane and has a density of 710 kg m3, a flash point of 4  C and
a boiling point of 98  C. Dodecane used contains 99% of n-dodecane
and has a density of 749 kg m3, a flash point of 74  C and a boiling
point of 216  C. In order to avoid freeboard effects, fuel quantity is
determined as a function of pan diameter with the relation h/
D ¼ 11.5%, where h represents the fuel height and D the pan
diameter. Indeed, before each test, water is first poured on the pan
and then on the fuel until it reached 1 mm below the pan lip.
During the combustion, the location of the fuel/water interface
remains fixed. Therefore, since the height of the burning fuel de-
creases as combustion progresses, the freeboard length increases
during the experiment. However, this freeboard length increase
appeared to have only a minor effect on the measured combustion
rate. Fuel is ignited with a butane torch.
Different sizes of pan diameter (0.10, 0.15, 0.23, 0.30 and 0.40 m)
are used. HRR is measured by means of a load cell using Babrauskas’
law [23] (Eq. (9)) or calculated on the basis of the oxygen con-
sumption taking into account the incomplete combustion of carbon
monoxide to carbon dioxide.
Q_ ðt Þ ¼ S$m
_ 00 ðt Þ$DHC (9)
As oxygen consumption is difficult to use during under-
ventilated fires because of the oxygen provided by the dilution flow,
HRR is measured by mass loss in this paper. Consequently, for
underventilated fires, HRR is overestimated. The flow rates of gases
in the inlet duct and dilution duct is measured with hot wire system
placed at the entrance of these ducts. Their sensitivities are lower
90 J. Lassus et al. / International Journal of Thermal Sciences 75 (2014) 87e94

than 1%. In the extraction duct, a laser and a fast digital camera are Table 3
used in order to determine gas flow rate with Particle Image Comparison between oxygen concentrations and maximum temperature during full
and reduced scale heptane fires of equivalent heat release rate.
Velocimetry (PIV) technique. Water-cooled MEDTHERM and CAP-
THERM captors are used to measure the radiative heat flux received Parameters Reduced scale Compartment Compartment
by walls. They are placed on the wall at a 0.45 m height and at 0.3 m compartment of 120 m3 of 400 m3

of the center axis, respectively on the inlet side and on the Q_ 20 kW 190 kW 520 kW
extraction side. Their range of sensitivity is [0e0.1] mV (W/m2). [O2] (% vol.) 16 14.9 16.6
Tmax ( C) 130 162 126
Pressure is also measured continuously by a pressure captor that
has a sensitivity of 2 Pa.
Continuous measurements of oxygen, carbon dioxide and car-
described in Prétrel [25]. Hydrogenated Tetra-Propylene (HTP) and
bon monoxide are performed at the entrance of the extraction duct
30% vol. Tri-Butyl-Phosphate (TBP). HTP fire tests were also per-
by on-line analysis. Local measurements of oxygen, carbon dioxide,
formed [15] in a compartment of 120 m3. A detailed description of
carbon monoxide, hydrogen and unburnt hydrocarbons are per-
this compartment is given in Nasr et al. [26]. HTP has a similar
formed in the exhaust duct by probe sampling and analyzed by gas
chemical formula than dodecane. Melis and Audouin [14] only
chromatography. All gas concentration measurements presented in
show fuel mass loss during heat release rate plateau. Maximal heat
this paper are performed at the entrance of the extraction duct in
release rate is estimated using the method of mass loss (Eq. (9)).
order to avoid dilution flow. The temperature measurements in the
Results are presented in Table 3. Good agreements with oxygen
compartment are performed with chromel-alumel thermocouples
concentrations and temperature are found between the results in
(type K) of a 0.5 mm wire diameter; they give values with an un-
the reduced scale room and those in the 400 cubic-meter
certainty of 1.5  C. Other details of experimental device are
compartment. Consequently, the similarity between full-scale
described by Lassus et al. [9] and Lassus [24] and photographs of the
compartments and the scale model is satisfactory. With the 120
experimental setup are presented in Fig. 1. Three fire tests for each
cubic-meter compartment, oxygen concentration and temperature
condition are made to check the reproducibility of the main pa-
are higher and that may be due to inlet duct position in this
rameters. Good reproducibility of temperatures, species concen-
compartment, near the ceiling.
trations and heat release rate are found [24].

4. Experimental results 4.2. Ventilated fires

4.1. Scaling effects: comparison with full scale compartment fire
In order to validate results obtained in the reduced scale room,
scaling effect is studied by comparing species concentrations and
maximal temperature measured in the experimental device with
those obtained during heptane and dodecane fire tests in full-scale
rooms. As stated above, HRR is scaled by the preservation of the
term Q2/L5, where L is the characteristic length equal to the cubic
root of room volume. A heptane fire of 1 MW performed in ICFMP’s
compartment [11] is compared to heptane fire of 28 kW performed
in the reduced scale room with a pan diameter of 0.23 m, at 3 ACPH
(Air Change Per Hour). Results are presented in Table 2. ICFMP’s
compartment is similar to a cubic room. Temperature at the
exhaust duct, maximal carbon dioxide and Minimum Oxygen
Concentrations are in good agreement for these two scales, despite
the fact that thermal radiation from the walls is not considered in
this comparison. The difference between CO concentrations ob-
tained during the reduced-scale experiments and ICFMP experi-
ments can be due to the differences in the sampling places in the
extraction duct. Indeed, we are comparing different enclosures that
have not been built to scale one as a function of the others; sam-
pling places can therefore be different.
Results obtained for dodecane fire tests performed in the
reduced scale room are also compared to fire tests performed at a
larger scale. Indeed, Melis and Audouin [14] performed fire tests in
a compartment of 400 m3. The experimental enclosure is also
Ventilation influence is considered in this paragraph for heptane
fires performed with a pan diameter of 0.15 m. The ventilation flow
is ranging between 1 and 7.6 ACPH. This pan diameter allows
obtaining fires sufficiently ventilated. HRR for different ventilation
flows are presented in Fig. 2. HRR obtained for an open fire with the
same pan diameter (0.15 m) is also presented in this figure. With
this pan diameter and for all the ventilation flows, oxygen quantity
initially present in the reduced scale room and the one that is
supplied by mechanical ventilation allow obtaining a sufficiently
ventilated fire.
For low ventilation flows (1e3 ACPH), HRR reaches a plateau
that is almost the same as the one obtained in open conditions:
fires have a behavior similar to the one in an open fire and there is
no confinement effect. Flame shows the classical structure of
diffusion flame (Fig. 3, photograph a).
When ventilation flow increases, flames are sloping (cf. Fig. 3,
photographs b and c). This effect is highlighted by the evolution of
radiant heat flux in the compartment and is illustrated in Fig. 4. For
low ventilation flow, radiant heat fluxes measured on each side of

Table 2
Comparison between species concentrations and maximum temperature during full
and reduced scale heptane fires of equivalent heat release rate.

Parameters ICFMP fire test Reduced scale compartment

Q_ 1 MW 28 kW
[O2] (% vol.) 12 12.7
[CO2] (% vol.) 6.5 6.1
[CO] (% vol.) 0.01 0.07
Fig. 2. Heat release rates as a function of time obtained with heptane fires performed
Tmax ( C) 255 235
with a pan diameter of 0.15 m at different ACPH.
 J. Lassus et al. / International Journal of Thermal Sciences 75 (2014) 87e94
Fig. 3. Pictures of heptane flames at 3, 5 and 7.6 Air Changes Per Hour (influence of the
ventilation flow on the flame slope).
central axis are similar: flame presents a symmetrical evolution. At
7.6 ACPH, the flame incline results in different radiant heat fluxes
on extraction and inlet sides. Consequently, the peak of HRR is
much higher than the one of an open fire.
4.3. Underventilated fires
With a larger pan diameter, oxygen quantity initially present
and supplied by mechanical ventilation can become insufficient to
obtain a complete reaction. The evolution of HRR obtained for
heptane fires of a 0.23 m diameter pan is presented in Fig. 5, for
different ventilation flows. The HRR of an open fire for this diameter
is also presented. In the frame of potential use of these experiments
for simulation validation, Fig. 6 presents the temperature profile at
the entrance duct (i.e. at the exit of the compartment) for a heptane
fire with a pan diameter of 23 cm at 3 ACPH. Oxygen concentration
as a function of time is also presented in this figure.
Similarly to fires performed with a pan diameter of 0.15 m and at
high ventilation flows, fire is sufficiently ventilated and a blowing
out effect coupled with a confinement effect is observed. HRR
reaches a high value but seems to stay at a steady state level after 5
ACPH. We may suppose that, for an upper ventilation flow, air flow
will become too strong: an opposite effect of this blowing effect,
described by the Damköhler number, will result in a lower HRR. For
low ventilation flows (ACPH  3), HRR reach plateaus and are lower
than the one reached by HRR in open geometries: it is due to a lack
of oxygen. With these ventilation flows, oxygen supply is insuffi-
cient to burn fuel at the same HRR as the one of an open fire. These
underventilated fires stop due to a lack of fuel.
Fig. 4. Non-dimensional radiant heat flux (ratio Frad/radiant heat flux at 100 s) as a
function of time of heptane fires (pan diameter of 0.1 m and ventilation flow of 3 and
7.6 ACPH).
91
Fig. 5. Heat release rates as functions of time of heptane fires performed with a pan
diameter of 0.23 m and at different ACPH.
4.4. Very underventilated fires
With a heptane pan of a diameter of 0.3 m, fire can become very
underventilated, resulting in a large amount of carbon monoxide.
HRR as a function of time is presented in Fig. 7 for heptane fires of a
0.3-m diameter pan, at 3 and 5 Air Changes Per Hour. This figure
also shows HRR obtained during open fire. The noisy evolution can
be explained by the fact that the fire in this case is very under-
ventilated and close to the extinction. This is the reason why the
flame is very perturbed. In order to avoid damages in the room test
with too stronger fires, only these two ventilation flows are tested.
Fig. 8 shows the evolutions of oxygen concentration for heptane
fires of a 0.3-m diameter pan, at 3 and 5 Air Changes Per Hour.
At 3 ACPH, HRR reaches a plateau similar to the one of an open
fire. But oxygen amount initially present in the compartment and
supplied by ventilation flow is not sufficient to maintain this py-
rolysis rate. Oxygen is quickly consumed and fire stops by lack of
oxygen. We can see in Fig. 8 that oxygen concentration increases
after the flame extinction. After this extinction, there is an accu-
mulation of oxygen in the compartment supplied by mechanical
ventilation. Then, auto-ignitions near heptane pan can take place.
Auto-ignitions were observed during the experiments. As auto-
ignition is a random phenomenon, the number of these auto-
ignitions varies for each fire tests.
Fires performed at 5 ACPH present a higher HRR than those
obtained for open fire due to confinement effect coupled to a slight
25 250
O2
temperature
20 200
Temperature (°C)
15 150
[O2] %
10 100
5 50
0 0
0 500 1000 1500 2000
Time (s)
Fig. 6. Temperature at the duct entrance and oxygen concentration as functions of
time (heptane fire with a pan diameter of 23 cm at 3 ACPH).
92 J. Lassus et al. / International Journal of Thermal Sciences 75 (2014) 87e94
Fig. 7. Heat release rates of heptane fires performed with a pan diameter of 0.3 m and
at different ACPH.
Fig. 8. Oxygen concentration as a function of time of heptane fires (pan diameter of
0.3 m and different ACPH).
blowing out effect. The bigger amount of oxygen that enters the
compartment at this ventilation flow can also contribute to this
high value of HRR. Similarly to fires performed with a ventilation
flow of 3 ACPH, these fires become very underventilated and stop
by lack of oxygen.
4.5. Evolution of species concentrations
The normalized yield of oxygen is plotted as a function of global
equivalence ratio (Fig. 9) in order to study the evolution of com-
bustion completeness. It is defined as the oxygen mass yield
divided by the theoretical maximum mass yield of oxygen for the
given fuel. The normalized yield of oxygen is approximately one for
Fig. 9. Normalized oxygen yield as a function of equivalence ratio for different heptane
fires (present work and Ref. [1]) and hexane fire [4].
Fig. 10. Carbon dioxide concentration as a function of oxygen concentration for hep-
tane fires: present work with pan diameters of 0.23 and 0.3 m at different ACPH and
ICFMP fire tests [11].
sufficiently ventilated conditions, where the reactions are nearly
complete. Results are similar whatever the ventilation flow applied
is and whatever the pan diameter used is. For each condition, dots
form two curves that respectively tallies with the oxygen depletion
during fire and the oxygen increase at the extinction. Good agree-
ments are found with Gottuk’s hexane results [4]. Beyler’s model
[1] for heptane fires is quite good for sufficiently ventilated con-
ditions and gives an upper limit of this yield during underventilated
conditions.
The carbon dioxide concentration is plotted against oxygen
concentration for heptane fires performed in different conditions
along with the results of ICFMP heptane fire [10] in Fig. 10. The
carbon dioxide yield is on the same order as ICFMP results. There is
no influence of ventilation flow and compartment geometry on the
carbon dioxide yield. The type of ventilation used (natural or me-
chanical) has nearly no influence on the evolution of carbon dioxide
concentration. The evolution of carbon dioxide concentration as a
function of oxygen concentration is linear during sufficiently
ventilated fires and underventilated fires. There are no measure-
ments of these parameters at very underventilated conditions for
heptane fires. During very underventilated conditions obtained
with dodecane fires, a plateau is reached for oxygen concentration
below 7%. This result is presented in Fig. 11 for a dodecane fire
performed at 5 ACPH with a pan diameter of 0.4 m.
The carbon monoxide concentration obtained for different
conditions is plotted against oxygen concentration in Figs. 12 and
13, respectively for heptane and dodecane fires. Similar behaviors
are found for these two fuels. The evolution of carbon monoxide
concentration as a function of oxygen concentration presents three
12
40 cm, 3 ACPH
10 40 cm, 5 ACPH
30 cm, 3 ACPH
8 30 cm, 5 ACPH
CO2 (%)
0
0 5 10 15 20 25
O2 (%)
Fig. 11. Carbon dioxide concentration as a function of oxygen concentration measured
during a dodecane fire performed at 3 and 5 ACPH with pan diameters of 0.3 and
0.4 m.
 J. Lassus et al. / International Journal of Thermal Sciences 75 (2014) 87e94 93

2.5 23 cm, 3 ACPH obtained for different compartments (and therefore for different
scales) for the three combustion regimes: sufficiently ventilated
23 cm, 5 ACPH
fires, underventilated fires and very underventilated fires. Values
2 30 cm, 3 ACPH are obtained thanks to scaling laws presented in Section 2.
30 cm, 5 ACPH
1.5 MOC (CO)
CO (%)

5. Conclusion

1 A reduced-scale room has been built in order to perform several

tests respecting several scaling laws. A comparison of species
0.5 concentrations and temperatures of the upper layers of fires is
made between full scale and reduced scale. Good agreement is
0 found for these parameters during ventilated fires. However, there
0 5 10 15 20 is no very underventilated fire in full scale compartment allowing
O2 (%) the validation of the reduced-scale room in these conditions.
Fire tests are performed in the reduced-scale room for different
Fig. 12. Carbon monoxide concentration as a function of oxygen concentration of pan diameters and ventilation flows. The results of this investiga-
heptane fires at 3 and 5 ACPH, with pan diameters of 0.23 and 0.3 m. tion show that three conditions of ventilation can appear. During
sufficiently ventilated fires, combustion is almost complete which
results in a very low carbon monoxide production. When fires are
2.5 MOC (CO)
underventilated, i.e. when oxygen concentration in the compart-
40 cm, 3ACPH ment is ranging between the Minimum Oxygen Concentration of
2 40 cm, 5ACPH the given fuel and around 7%, HRR of fire is lower than the one
obtained during open fire with the same pan diameter. Carbon
30 cm, 5 ACPH
monoxide is produced and its quantity depends on oxygen con-
1.5
CO (%)

30 cm, 3 ACPH centration in the compartment. Carbon dioxide concentration in-

creases linearly with the decreasing of oxygen concentration. For
1 oxygen concentrations lower than around 7%, when fires become
very underventilated, heat release rate becomes quite lower than
0.5 the one in open fire conditions and extinction by lack of oxygen
occurs. Some auto-ignition can appear around fuel pan and in the
0 room. Carbon monoxide concentration in the compartment
strongly increases due to the fact that it is no longer oxidized in
0 5 10 15 20
carbon dioxide and reaches a quantity higher than 1%.
O2 (%)
Fig. 13. Carbon monoxide concentration as a function of oxygen concentration of Acknowledgments
dodecane fires at 3 and 5 ACPH, with pan diameters of 0.3 and 0.4 m.
The authors are highly grateful to Pr. Jean-Pierre Vantelon for his
help during this work.
parts. The first part corresponds to sufficiently ventilated condi-
tions and is ranging between the Minimum Oxygen Concentration
References
(MOC) of the given fuel and the initial concentration of oxygen:
with these oxygen quantities, a nearly complete reaction is reached [1] C.L. Beyler, Major species production by diffusion flames in a two-layer
and carbon monoxide production is very low. MOC of heptane and compartment fire environment, Fire Safety Journal 10 (1986) 47e56.
dodecane are respectively 11.55 and 11.14%. The second part tallies [2] A. Tewarson, Generation of Heat and Chemical Compounds in Fires? SFPE
Handbook of Fire Protection Engineering, third ed., National Fire Protection
with underventilated conditions and is ranging between the MOC Association, 2002, pp. 3.82e3.161.
of CO (6.5%) and the MOC of the given fuel: during this range, [3] J.H. Morehart, E.E. Zukoski, T. Kubota, Species Produced in Fires Burning in
oxidation reaction is incomplete and there is a carbon monoxide Two-layered and Homogeneous Vitiated Environments. Report NBS-GCR-90-
585, National Institute of Standards and Technology, 1990.
production all the more important since oxygen concentration is [4] D.T. Gottuk, Generation of Carbon Monoxide in Compartment Fires. PhD
low. The last part tallies with very underventilated conditions and thesis, Virginia Polytechnic Institute and State University, 1992.
includes oxygen concentration lower than around 7%. It is the range [5] Y. Utiskul, J.G. Quintiere, A.S. Rangwala, B.A. Ringwelski, K. Wakatsuki,
T. Naruse, Compartment fire phenomena under limited ventilation, Fire Safety
of oxygen concentration where carbon dioxide reaches a plateau.
Journal 40 (2005) 367e390.
Carbon monoxide production in this part is much higher than the [6] W.M. Pitts, The global equivalence ratio concept and the formation mecha-
one during the second part due to the fact that carbon monoxide is nisms of carbon monoxide in enclosures fires, Progress in Energy and Com-
no longer oxidized in carbon dioxide. Table 4 presents HRR bustion Science 21 (1995) 197e237.
[7] B.Y. Lattimer, D.T. Ewens, U. Vansburger, R.J. Roby, Transport and oxidation of
compartment fire exhaust gases in an adjacent corridor, Journal of Fire Pro-
tection Engineering 6 (1994) 163e181.
Table 4 [8] C.J. Wieczorek, Carbon Monoxide Generation and Transport from Compartment
Heat release rate (in kW) of different enclosures of different volumes for three Fires. PhD thesis, Virginia Polytechnic Institute and State University, 2003.
combustion regimes: sufficiently ventilated fires, underventilated fires and very [9] J. Lassus, E. Studer, J.P. Garo, J.P. Vantelon, P. Jourda, P. Aine, Influence of venti-
underventilated fires. lation on ignition risk of unburnt gases in the extraction duct of underventilated
compartment fire, Combustion Science and Technology 183 (2010) 517e528.
Experimental setup (8 m3) 10 15 20 30.5 61 91.5 122 [10] R. Viskanta, Overview of some radiative transfer issues in simulation of un-
Enclosure of 100 m3 82 123 165 250 500 750 1000 wanted fires, International Journal of Thermal Sciences 47 (2008) 1563e1570.
Enclosure of 120 m3 96 143 191 291 583 874 1165 [11] M. Rowekamp, J. Dreisbach, W. Kelin-Hessling, K.B. McGrattan, S. Miles, M. Plys,
Enclosure of 400 m3 261 391 521 795 1589 2384 3178 O. Riese, International Collaborative Fire Modeling Project (ICFMP), Summary of
Enclosure of 583 m3 357 535 713 1088 2175 3263 4350 Benchmark Exercises. Report 227, GRS, 2008.
(ICFMP fire tests) [12] W.M. Pitts, N.P. Bryner, E.L. Johnsson, Combustion product formation in under
and overventilated full scale enclosure fires, in: Proceedings Combustion
94 J. Lassus et al. / International Journal of Thermal Sciences 75 (2014) 87e94

Fundamentals and Applications, Combustion Institute Central and Western [19] H.C. Hottel, Review: certain laws governing diffusive burning of liquids, Fire
States (USA) and Combustion Institute Mexican National Section and American Research Abstracts and Reviews 1 (1958) 41e44.
Flame Research Committee, 1995, pp. 565e570. [20] V.I. Blinov, G.N. Khudiakov, Certain laws governing diffusive burning of liquids,
[13] N.P. Bryner, E.L. Johnsson, W.M. Pitts, Carbon Monoxide Production in Doklady 113 (1957) 1094e1098.
Compartment Fires: Reduced-Scale Enclosure Test Facility, National Institute of [21] P.H. Thomas, R. Baldwin, A.J.M. Heslden, Buoyant diffusion flames: some
Standards and Technology, 1994. NISTIR-5568. measurements of air entrainment, heat transfer and flame merging, in: 10th
[14] S. Melis, L. Audouin, Effects of vitiation on the heat release rate in mechanically Symposium International on Combustion, 1965, pp. 983e996.
ventilated compartment fires, in: 9th International Symposium on Fire Safety [22] T. Yumoto, Heat transfer from flame to fuel surface in large pool fires, Com-
Science, 2009, pp. 931e942. bustion and Flame 17 (1971) 108e110.
[15] G. Heskestad, Modeling of enclosure fires, in: 14th International Symposium on [23] V. Babrauskas, Ignition Handbook, Fire Science Publishers, 2003, pp. 41e140
Combustion, 1973, pp. 1021e1030. (Chapter 4).
[16] R.I. Emori, K. Saito, A study of scaling laws in pool and crib fires, Combustion [24] J. Lassus, Risque d’inflammation de gaz imbrûlés au cours d’un incendie en
Science and Technology 31 (1983) 217e231. milieu confiné-ventilé. PhD thesis, University of Poitiers, 2009 (in French).
[17] J.G. Quintiere, Fundamentals of Fire Phenomena, Wiley and Sons, 2006 [25] H. Prétrel, Interprétation des essais FLIP. Report, IRSN, 2002.
(Chapter 12). [26] A. Nasr, S. Suard, H. El-Rabii, J.P. Garo, L. Gay, L. Rigollet, Heat feedback to the
[18] E.E. Zukoski, Fluid dynamic aspects of room fires, in: 1st International Sym- fuel surface of a pool fire in an enclosure, Fire Safety Journal (2013) in press,
posium on Fire Safety Science, 1984, pp. 1e30. http://dx.doi.org/10.1016/j.firesaf.2012.12.005.