Sie sind auf Seite 1von 10

Combustion and Flame 159 (2012) 1016–1025

Contents lists available at SciVerse ScienceDirect

Combustion and Flame


j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / c o m b u s t fl a m e

Stirred reactor calculations to understand unwanted combustion


enhancement by potential halon replacements q
Gregory T. Linteris a,⇑, Donald R. Burgess b, Fumiaki Takahashi c, Viswanath R. Katta d,
Harsha K. Chelliah e, Oliver Meier f
a
Fire Research Division, National Institute of Standards and Technology, Gaithersburg, MD 20899-8665, USA
b
Chemical and Biochemical Reference Data Division, National Institute of Standards and Technology, Gaithersburg, MD 20899-8320, USA
c
Case Western Reserve University, Cleveland, OH 44106, USA
d
Innovative Scientific Solutions, Inc., Dayton, OH 45440, USA
e
University of Virginia, Charlottesville, VA 22904, USA
f
The Boeing Company, Seattle, WA 98124, USA

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

Article history: Several agents are under consideration to replace CF3Br for use in suppressing fires in aircraft cargo bays.
Received 4 March 2011 In a Federal Aviation Administration performance test simulating the explosion of an aerosol can, how-
Received in revised form 20 September ever, the replacements, when added at sub-inerting concentrations, have all been found to create higher
2011
pressure rise than with no agent, hence failing the test. Thermodynamic equilibrium calculations as well
Accepted 22 September 2011
Available online 7 November 2011
as perfectly-stirred reactor simulations with detailed reaction kinetics, are performed to understand the
reasons for the unexpected enhanced combustion rather than suppression. The high pressure rise with
added C2HF5 or C3H2F3Br is shown to be dependent upon the amount of added agent, and can only occur
Keywords:
Fire suppression
if a large fraction of the available oxidizer in the chamber is consumed, corresponding to stoichiometric
Flame inhibition proportions of fuel, oxygen, and agent. Conversely, due to the unique stoichiometry of CF3Br, this agent is
CF3Br predicted to cause no increase in pressure, even in the absence of chemical inhibition. The stirred-reactor
C2HF5 simulations predict that the inhibition effectiveness of CF3Br is highly dependent upon the mixing con-
Halon replacements ditions of the reactants (which affects the local stoichiometry and hence the overall reaction rate). For
Cargo-bay fire suppression C2HF5, however, the overall reaction rate was only weakly dependent upon stoichiometry, so the fuel–
oxidizer mixing state has less effect on the suppression effectiveness.
Published by Elsevier Inc. on behalf of The Combustion Institute.

1. Introduction chemical formulas are listed in Table 1.) The agent CF3Br, at sub-
suppressing concentrations, does not cause the overpressure.
Because of its destruction of stratospheric ozone, production of The Aerosol Can Test [1] simulates the situation in which a fire
the effective fire suppressant CF3Br has been banned by the Mon- in an aircraft cargo bay container heats an aerosol can (e.g., hair
treal Protocol. Although a critical-use exemption has been granted spray) causing it to burst, creating an explosion. In the test, a
to the aviation industry for use of recycled halon in cargo bay fire heated container at about 16 bar, releases its contents (propane,
suppression, the European Union requires replacement of halon in ethanol, and water) as a two-phase impulsive spray via a fast-
new design aircraft by 2018, and in existing aircraft by 2040. Sev- acting valve. A continuous DC arc across electrodes located about
eral replacements have been proposed, but they have all been 1 m downstream of the valve ignites the mixture. The fireball
found to produce enhanced burning in the FAA Simulated Aerosol expands into the chamber atmosphere of ambient air and water
Can test [1], and hence they fail FAA’s Minimum Performance Stan- vapor and premixed suppressant, and the temperature and
dard [2]. In particular, C2HF5, C3H2F3Br, and C6F12O all produce pressure in the chamber increases (over a time on the order of a
higher peak pressures in a simulated cargo bay as compared to second). During each test, instruments record the pressure, tem-
no added agent, when they are added at concentrations less than perature, visual images, and concentrations of agent and oxygen.
that required to completely suppress the explosion. (Names and Unfortunately, when added at sub-inerting concentrations, the
final pressure rise in the chamber with any of the halon replace-
ment agents is higher than in the absence of agent. The agent
q
C2HF5, added at a volume fraction of 13.5%, suppressed the explo-
Official contribution of NIST, not subject to copyright in the United States.
⇑ Corresponding author. Fax: +1 301 975 4052. sion; however, when added at volume fractions of 6.2%, 8.9%, and
E-mail address: linteris@nist.gov (G.T. Linteris). 11.0%, the peak pressure rise was about 3.6 bar, or about twice that

0010-2180/$ - see front matter Published by Elsevier Inc. on behalf of The Combustion Institute.
doi:10.1016/j.combustflame.2011.09.011
G.T. Linteris et al. / Combustion and Flame 159 (2012) 1016–1025 1017

Table 1
Chemical names of compounds of interest.

Formula Trade name Common name Systematic name Structural formula


CF3Br Halon 1301 Bromotrifluoromethane Bromotrifluoromethane BrCF3
C2HF5 HFC-125 Pentafluoroethane 1,1,1,2,2-Pentafluoroethane CF3  CHF2
C3H2F3Br 2-BTP Bromotrifluoropropene 2-Bromo-3,3,3-trifluoro-1-propene CH2 = CBrCF3
C6F12O Novec 1230, FK-5-1- Heptafluoroisopropyl pentafluoroethyl 1,1,1,2,2,4,5,5,5-nonafluoro-4-(trifluoromethyl)-3- CF3CF2C(=O)CF(CF3)2
12 ketone pentanone

with no added suppressant (1.8 bar). With C3H2F3Br, volume frac- Using a constant volume combustion device, Shebeko et al. [14]
tions of 3% or 4% both gave a pressure rise of 4.3 bar, and volume studied the effect of various halogenated hydrocarbons on the
fractions of 5% or 6% gave a pressure rise of about 6.7 bar. With combustion of methane– and hydrogen–air mixtures. The flamma-
C6H12O, a volume fraction of 4.2% gave a pressure rise of 4.5 bar, bility limits generally narrowed with addition of the agents; how-
and above that volume fraction (or sometimes at 4.2% itself) the ever, for C4F8 addition to CH4–air flames, the lean limit widened,
over pressure was suppressed. In contrast, addition of CF3Br at a and for C2HF5 or C2F5Cl addition to CH4–air flames, the lean limit
volume fraction of about 3.8% suppressed the explosion, and 2.5% was nearly unchanged. That is, for a CH4–air flame at the lean limit,
CF3Br, the pressure rise was only 0.28 bar. (For reference, uncon- a C2HF5 mass fraction of 30% had essentially no effect on the
fined tests without suppressants create a 3.4 m diameter fire ball flammability.
[3]). The goal of the present work is to understand the reasons In addition to the flammability limits, the pressure rise, as well
for the higher pressure rise in the FAA aerosol can test with as the rate of pressure rise were also used by Shebeko et al. [14] to
C2HF5 or C3H2F3Br (at sub-inerting concentrations), and the lack illustrate the promoting effect of the halogenated fire suppressant
of this effect with CF3Br. for some conditions. For example, when added to lean H2–air
flames, C2HF5 increased the maximum pressure rise at any concen-
2. Background tration (up to the extinction point). Further, C4F8 (perfluorocyclob-
utane) addition increased the maximum pressure rise for both lean
Enhanced combustion in the presence of fire suppressants has and rich H2–air flames. This lack of kinetic inhibition was even fur-
been observed in previous work. A promotion effect has been ther substantiated for the agent C4F10 when added to CH4–air
shown for halogenated hydrocarbons in shock-tube ignition stud- flames, for which even the rate of pressure rise was higher with
ies for some fuels and conditions. For example, it is known that agent added up to a volume fraction of 4%.
the H2–O2 mixture can be sensitized by HBr [4–6] and HCl [4], with In experiments with high-speed turbulent flames in a detona-
promotion under some conditions and inhibition under others. tion/deflagration tube, Grosshandler and Gmurczyk [15–18] ob-
Moen et al. [7] performed experiments and modeling on the influ- served more vigorous combustion with CF3I or CF3Br, or various
ence of CF3Br, CF4, and CO2 on the detonability of ethane–air and hydrofluorocarbon inhibitors, while using either propane or ethyl-
ethane– and hydrogen–oxygen mixtures, and found that the inhi- ene as fuels. For some conditions, the premixed addition of the hal-
bition ability of CF3Br was greatly reduced as compared to flame ogenated agent to the air stream increased both the deflagration/
inhibition, and that CF3Br was a sensitizer for some situations. Sim- shock propagation rate as well as the pressure ratio across the shock.
ilarly, Babushok et al. [8] computationally studied the ignition de- The results varied with fuel type, stoichiometry, agent type, and the
lay of hydrocarbons (H2, CH4, CH3OH, and C2H6) and air with presence or absence of turbulence-inducing spirals.
various halogenated fire suppressants (Br2, HBr, CH3Br, CF3Br, In tests with co-flow diffusion flames, halogenated hydrocar-
and CF3H), and also found both promotion or inhibition of the igni- bons added to either the fuel or air stream have been shown to
tion. The varying behavior was complex, likely depending upon increase total heat release. Holmstedt et al. [19] reported that
subtle changes to the kinetic pathways as the pressure, tempera- HFC-227ea (C3HF7) or HFC-134a (C2H2F4) added to the fuel
ture, or mixture composition change. (propane) stream of a turbulent jet burner increased the total heat
Wider flammability limits in the presence of halogenated com- release, by a factor of 2 and 3.8, respectively, for concentrations
pounds have also been reported. Saito et al. [9] conducted flamma- just below that required for extinguishment of the flame. Similarly,
bility limit experiments in a tubular flame of premixed Katta et al. [20] found that CF3H added to the oxidizer stream in a
hydrocarbon (methane or propane) and air, with added CF3Br, methane–air cup-burner experiment increased the total heat
CF3H, C3HF7, or C4F10. The traditional fuel–air concentration limits release.
[10] were determined as a function of agent concentration. With Increased pressure can also make HFC–air mixtures more flam-
addition of CF3Br, both the rich and lean limits narrowed steeply mable. In a survey paper, Ural [21] noted that while halogenated
(the mixture became less flammable); whereas with addition of hydrocarbons are often considered to have no or low flammability
various hydrofluorocarbons (HFCs), the rich limit narrowed less potential, some have been shown to become flammable at elevated
steeply, and the lean limit sometimes widened (the mixture be- pressure. In a recent study, Kondo et al. [22] measured the rich and
came more flammable). These results confirmed the ignition tests lean flammability limits of HFC-32 (CH2F2) and HFO-1234yf
of Moore et al. [11] performed in a spherical combustion chamber, (C3H2F4) at pressures of 101–2000 kPa. They found that in general,
also with premixed reactants. Recently, Kondo et al. [12] measured at increased pressure, the lean flammability limit was relatively
the flammability limits of C2HF5 for spark-ignited premixed gases unchanged, while the rich limit widened significantly.
in a glass flask [13]; fuels were methane, propane, propylene, In large-scale tests, Mawhinney et al. [23] found that applica-
methyl formate, and HFC-152a (1,1-difluoroethane, CF2H–CH3), tion of water mist to a fire caused unwanted accelerated burning,
with dry air. The addition of the agent C2HF5 narrowed the rich which they believed was due to fluid–dynamic enhancement of
limit, roughly linearly, up to the inertion point. Conversely, the the burning. Hamins et al. [24] reviewed previous work on en-
lean limit was widened for all fuels, with the order of ranking: hanced burning with application of fires suppressants, and also
C3H6 < C3H8 < HCOOCH3 < C2H4F2  CH4 (i.e., largest widening of concluded that the enhanced combustion was due to more rapid
the lean limit occurred for the fuels with higher H/C ratio). Hence, mixing of fuel vapor with air, from the combined effects of en-
these experiments also showed an enhance burning of the fuels hanced turbulent mixing and more vigorous liquid fuel atomiza-
caused by C2HF5 under fuel-lean conditions. tion from agent jet impingement.
1018 G.T. Linteris et al. / Combustion and Flame 159 (2012) 1016–1025

While the possibility of enhanced flammability in the presence determined based on the value of g and the specified volume frac-
of HFC fire suppressants has been described in the work presented tion of each component.
above, no analyses were performed in those investigations to The initial inhibitor volume fraction in the oxidizer gases was
understand why promotion occurred. More typically, HFC agents varied from zero to 13.5% for C2HF5, 6%, for C3H2F3Br, and 5% for
act as fire suppressants, and based on previous work [25–27], CF3Br, corresponding to approximately the maximum amount
C2HF5 would be expected to extinguish—or at least weaken—the added in the FAA tests [1,33]. The fraction of the chamber volume
FAA aerosol can test explosions. In previous work [28], an analysis involved in the combustion, g, was varied from about 0.23–1.00.
based on equilibrium thermodynamics was used to gain insight
into the behavior of the agents C2HF5, C3H2F3Br, and CF3Br in the 3.2. Equilibrium thermodynamics
FAA tests. The analyses were able to predict the experimental pres-
sure rise for addition of C2HF5 or C3H2F3Br at sub-extinguishing The equilibrium conditions of the aerosol can test were calcu-
volume fractions (0–11.2% and 0–6%, respectively). The enhanced lated using both the STANJAN-III program of Reynolds [34], and
combustion with addition of agent to the oxidizer was shown to CEA2 of Gordon and McBride [35]; the two codes gave results very
be due to the higher heat release, caused by involvement of close to each other. The calculations were performed over the wide
increasing fractions of the chamber volume of oxidizer. Nonethe- range of initial conditions described above. Constant enthalpy, con-
less, for CF3Br at all concentrations, and C2HF5 at its suppression stant pressure solutions were obtained, as described in Ref. [28].
volume fraction (13.5%), the pressure rise was less than predicted
by equilibrium thermodynamics, possibly due to kinetic limita- 3.3. Kinetic mechanism
tions in the rate of energy release caused by addition of these
chemically acting agents [29–32]. Hence, it is of interest to exam- A kinetic mechanism to describe the chemical inhibition of the
ine the kinetic behavior of the aerosol can test fuel with air and aerosol can test fuel (propane, ethanol, water) with the HFC and
added agent (C2HF5 or CF3Br). Because of the distributed nature bromine-containing species was assembled from sub-mechanisms
of the explosively dispersed fuel-agent–air mixtures in the FAA available in the literature. For the hydrocarbon mechanism, an
tests [1], perfectly-stirred reactor (PSR) calculations were em- optimized model for ethylene oxidation proposed by Wang and
ployed, as described below. While the conditions in the FAA test co-workers was employed [36,37], that included 111 species and
are likely to include more non-premixed or partially premixed 784 elementary reactions. This model has been optimized by con-
structure (due to the fuel-rich core), premixed stirred reactor cal- sidering experimental ignition delay and species profiles data from
culations nonetheless seem to be a reasonable next step in com- shock tubes, laminar flame speeds, species profile data from flow
plexity (after the equilibrium calculations), that will provide reactors, and species profile data from flat flames. To this mecha-
some insight into the relevant kinetic limitations. nism, more detailed reactions of ethanol were added (5 species
and 36 reactions), as proposed by Dryer and co-workers [38–40].
3. Numerical methods For the reactions of the hydrofluorocarbons (HFCs) in hydrocarbon
flames, the NIST HFC mechanism was used [41,42]. Subsequent up-
3.1. Initial conditions dates to that mechanism were made by NIST workers, as noted
L’esperance et al. [43]. Other changes to the NIST HFC mechanism
The volume fraction of agent in the oxidizer Xinh, is a variable. were made in the present work based on recent experimental mea-
The turbulent mixing during the aerosol release makes specifica- surements and theoretical calculations [44–50] as listed in Table 2.
tion of the oxidizer–fuel ratio near the flame difficult, so calcula-
tions are performed for a range of fuel–oxidizer ratio. However,
because the oxidizer consists of both air and agent (which is also Table 2
Reaction rate modifications as suggested by Saso et al. [44], Takahashi and co-workers
a reactant), the oxidizer is partially premixed. Presentation of the [45,46], Vetters (for T Ü 330 K) [47], Fernandez and Fontijn [48], and Tsai and
results in terms of a fuel–oxidizer stoichiometric ratio is not prac- McFadden [49,50], that were incorporated into the NIST HFC mechanism.
tical: as the agent concentration in the oxidizer is increased, u for
Reaction A (cm/mol/s) b Ea (J/mol) References
stoichiometric conditions changes due to the agent’s oxygen de-
mand, as well as due to changes in the equilibrium products as CHF3 + H ) CF3 + H2 3.76  1013 0 54,810 [44]
CO + F + M ) CFO + M 3.09  1019 1.4 2040 [44]
the hydrogen–halogen ratio in the flame changes. Hence, the stoi-
CFO + H ) CO + HF 2.5  1013 0 0 [44]
chiometry effect is presented in terms of the fraction (g) of the CF3 + O ) CF2O + F 1.54  1013 0 0 [45]
available oxidizer (the chamber volume) involved in the reaction CF3 + H ) CF2 + HF 5.33x 1013 0 0 [45]
with the fixed quantity of fuel species from the aerosol can CF2 + O ) CFO + F 2.45  1013 0 0 [46]
CF2 + H ) CF + HF 3.98  1013 0 19,000 [46]
contents.
CF + O2 ) CFO + O 6.62  1012 0 7070 [45]
The fuel quantity is taken to be the contents of the aerosol can CHF3 + O ) CF3 + OH 3.07  1014 0 79,290 [48]
simulator (270 g ethanol, 90 g propane, and 90 g water). The oxi- CHF + H ) CH + HF 0.64  1014 0 0 [50]
dizer consists of ambient air (oxygen, nitrogen, and water vapor) CHF + H ) CF + H2 2.30  1014 0 0 [50]
and the added suppressant. Relative humidity of the test air was
not reported in the FAA tests; however, analysis of local weather
data for the days of the tests indicates that water vapor volume
fractions were always in the range of 0.0016–0.01 (corresponding Table 3
to 7–40% R.H. at 21 °C). Hence, the calculations were performed Additional reactions added to the NIST HFC mechanism to account for C2HF5 reaction
for a water vapor volume fraction Xwv of 0, 0.0125, and 0.025 in with early – forming radicals from propane and ethanol decomposition.

the O2/N2/H2O oxidizer mix, corresponding to 0%, 50%, and 100% Reaction A (cm/mol/s) b Ea (J/mol)
R.H at 21 °C. CHF2  CF3 + C2H5 = CF3  CF2 + C2H6 5.7  1010 0.0 49,400
To summarize: the fuel components and their quantities were CHF2  CF3 + nC3H7 = CF3  CF2 + C3H8 5.7  1010 0.0 43,100
fixed; the ratio of fuel to total oxidizer (N2, O2, H2O, and agent) CHF2  CF3 + iC3H7 = CF3  CF2 + C3H8 5.7  1010 0.0 56,500
was variable, expressed as g, the fraction of chamber volume CHF2  CF3 + C2H4OH = CF3  CF2 + C2H5OH 5.7  1010 0.0 44,400
CHF2  CF3 + CH3CHOH = CF3  CF2 + C2H5OH 5.7  1010 0.0 66,500
(11,400 L) involved in the combustion, and the quantity of each
CHF2  CF3 + CH3CH2O = CF3  CF2 + C2H5OH 5.7  1010 0.0 37,200
oxidizer component in the initial mixture for the calculation was
G.T. Linteris et al. / Combustion and Flame 159 (2012) 1016–1025 1019

A list of potentially important reactions of C2HF5 with the radicals glected in the present analyses. The governing equations of conser-
from initial fuel (propane or ethanol) decomposition was devel- vation of mass, species, and energy form a system of coupled non-
oped, and the rates were estimated as given in Table 3. The barriers linear algebraic equations, which can be solved numerically. In the
for the reactions were estimated in Evans-Polanyi fashion by anal- present work, we employ the SANDIA PSR code [59].
ogy to that for the reference reaction CHF2  CF3 + CH3 = CF3 – To obtain the characteristic chemical time at extinction using a
CF2 + CH4 contained in the NIST HFC mechanism by increasing stirred-reactor model, one must determine the blow-out condition.
the barriers in proportion (0.3) to the decrease in the heat of reac- The process is illustrated in Fig. 1, which shows the reactor tem-
tions relative to the reference reaction. The HFC sub-mechanism fi- perature as a function of residence time, for three values of the vol-
nally adopted contained 51 species and 600 reactions. To describe ume fraction of C2HF5 in the oxidizer. At a very low reactor mass
flame inhibition by CF3Br, the bromine parts of the mechanism of flow rate, the residence time in the reactor is long, essentially
Babushok et al. [8,51] were adopted, adding 10 species and 74 yielding the equilibrium conditions. As the mass flow in the reactor
reactions. That sub-mechanism has been validated in premixed is increased, the temperature decreases slightly due to incomplete
flame speed and ignition delay studies [8,29,52]. The final mecha- reaction, and there eventually becomes a point at which there is
nism used for the simulations of the aerosol can test fuel, with insufficient time to achieve substantial reaction in the vessel;
C2HF5, and CF3Br, has 177 species and 1494 reactions. because of the exponential dependence of reaction rate on temper-
The aerosol can test fuel is predominantly ethanol (about 2/3 of ature, this point is a very abrupt change, where the mixture
the energy release), and other models are available that describe ‘‘blows-out,’’ without reacting, yielding a blow-out time schem.
ethanol combustion [53–55] (and even more for propane). None- Near blow-out, a criterion of <0.5% change in the mass flow rate
theless, we selected the C1–C4 model of Wang as the base-case was used to determine schem.
(and added the ethanol reactions of Dryer and co-workers) since Figure 1 shows the blow-out condition for a mixture of the aer-
extensive validation studies have been performed by the develop- osol can test fuel, with 65% of the oxidizer in the test chamber
ers of these mechanisms. (g = 0.65) at a relative humidity of 50%, for C2HF5 added to the
chamber air at 0%, 7.7%, and 14.4% volume fraction. As indicated,
3.4. Perfectly-stirred reactor calculations the characteristic chemical times, schem, obtained for these inhibi-
tor concentrations are 0.43 ms, 2.4 ms, and 32 ms, corresponding
Flame extinction caused by suppressants is controlled by the to overall chemical rates in the stirred-reactor (xpsr = 1/schem) of
characteristic times for chemical reaction and transport, as de- 2330 s1, 416 s1, and 31 s1. As the concentration of added
scribed by the Damköhler number Da = sr/sc, in which sr is the flow C2HF5 increases in this range, the characteristic chemical time in-
residence time, and sc is the chemical time [56]. Hence, an impor- creases (i.e., the overall characteristic reaction rate decreases).
tant step for understanding flame suppression is to obtain some
measure of the overall reaction rate. Given the explosive, two-
4. Results
phase, turbulent mixing process occurring during release of the
aerosol can test simulator fuel [3], the reaction zone there may
4.1. Equilibrium calculations
be simulated reasonably well by a stirred reactor. Further, the stir-
red-reactor blow-out residence time has been correlated with both
The adiabatic flame temperature (of the involved reactants) Tad
the laminar flame speed [57] and with extinction of laminar diffu-
was calculated for the FAA aerosol can test in the presence of each
sion flames with added inert suppressants [58], indicating its util-
of the three suppressants (C2HF5, C3H2F3Br, CF3Br) premixed in the
ity as a measure of overall reaction rate. The residence time in the
chamber air. Calculations were performed over a range of values of
reactor s is defined as s ¼ qV=m,_ in which q is the mixture density,
V is the reactor volume, and m _ is the mass flow. Heat losses from
the reactor to the surroundings can also be considered, but are ne-
C2HF5
X inh
2500
η ≈ 0.65
2000 0
50 % r.h.

2000
Reactor Temperature / K

7.7

X C2HF5 = 0 % 0.072
Tad / K

1500 14.4

1500
0.135
1000

500

0.43 2.4 τ = 32 ms 1000


0 0.0 0.2 0.4 0.6 0.8 1.0
0.0001 0.001 0.01 0.1 1 10
η
Residence Time / s
Fig. 2. Calculated adiabatic equilibrium temperature (Tad) as a function of chamber
Fig. 1. Stirred reactor temperature as a function of residence time for the FAA volume fraction (g) involved in combustion with the contents of the aerosol can
aerosol can test fuel, g  0.65, 50% R.H., and C2HF5 added at 0%, 7.7%, and 14.4% simulator. Different curves refer to different initial volume fractions of the
volume fraction in the oxidizer stream. suppressant C2HF5 in the chamber air.
1020 G.T. Linteris et al. / Combustion and Flame 159 (2012) 1016–1025

g and Xi, as described in more detail in Refs. [28,60], which provide at Xinh = 6%, the explosion from the aerosol can fire ball was not
more details on specification of each of the mole fractions (pro- suppressed (despite all the oxidizer estimated to be consumed).
pane, ethanol, and water in the fuel mix, and N2, O2, ambient water As with C2HF5, as Xinh increases with added C3H2F3Br, the curves
vapor and agent in the oxidizer stream) for the initial conditions. for Tad are very flat near the peak—that is, Tad is not very sensitive
The pressure was 1.01 bar. to the value of g for high Xinh.
The results for C2HF5 are shown in Fig. 2. With no agent, the The results for CF3Br are different. As shown in Fig. 4, addition of
shape of the curve mimics the variation in Tad with fuel–air equiv- CF3Br up to suppression concentration (around 4%) again decreases
alence ratio [61], and the peak adiabatic flame temperature Tad,peak the peak Tad only slightly (100 K), but the shape of the curves at
is reached when about one third of the chamber volume of oxidizer high Xinh are essentially the same as at low Xinh. That is, for peak Tad,
reacts with the aerosol can contents (g = 0.33). As Xinh increases, the fraction of chamber volume required is constant (g = 0.33), not
Tad,peak decreases slightly, but still remains near 2200 K, and the increasing as Xinh increases, as occurs with addition of the other
peak value is reached at larger values of g. For inhibitor loadings two agents.
greater than Xinh = 0.072, Tad,peak decreases abruptly, and thereafter The effects of the inhibitor volume fraction Xinh on the peak adi-
decreases more steeply as Xinh increases. This occurs because above abatic flame temperature are summarized for the three agents
Xi = 0.072, the fluorine content of the system is greater than the (C2HF5, CF3Br, and C3H2F3Br) in Fig. 5. The left axis shows the peak
hydrogen content, so HF cannot be formed as the most stable final Tad, while the right axis shows the value of g required to achieve
product; rather, COF2 is also formed, with the attendant lower total the peak Tad (gmax). As indicated, all the agents have a minor (low-
chemical enthalpy release. The most notable feature of these ering) effect on the peak Tad for Xinh < 7%, with CF3Br lowering the
flames is that as Xinh increases to the value which suppressed the temperature the most. At higher Xinh with the agent C2HF5, the
aerosol can test explosion, 0.135, the peak Tad is still around peak Tad is lower, due to the inability of the system to form HF
1900 K, and all of the chamber volume (of oxidizer–inhibitor (when there are more F atoms in the system than H atoms). The ef-
mix) is required to achieve the peak temperature. Further, at high fect of the different agents on the fraction of chamber volume re-
values of Xinh, the behavior near the peak Tad is more of a plateau quired to achieve peak Tad is dramatic. As agent is added, gmax
than a peak, such that a wide range of values of g will produce increases rapidly for C2HF5, and even more rapidly for C3H2F3Br—
nearly the same Tad. This is different from the behavior at low Xinh, but not at all for CF3Br. For the halon alternatives, this is a result
for which Tad drops off on either side of the peak, similar to the of their fuel-like nature, (i.e., carbon and hydrogen content), so that
uninhibited case. Since reaction rates are exponential in tempera- the larger molecules have a larger effect on gmax. That is, adding the
ture, this behavior is likely to affect the overall chemical rates at agent is like adding fuel to the air stream, so that to oxidize the
low and high Xinh. aerosol can test fuel, more oxidizer volume is required. As Xinh
The results for C3H2F3Br are shown in Fig. 3. The result is similar goes up, the value of the oxidizer for this purpose is decreased
to that of C2HF5; however, the system reaches unity hydrogen to more, so a higher volume fraction of the chamber is required.
fluorine ratio at Xi = 0.06, so values above that condition are not The reason that CF3Br does not show this effect is due to the unique
displayed in the figure. As a result, the Tad,peak is maintained nearly stoichiometry of this agent in hydrocarbon systems. Since [F]/[H] is
constant at about 2200 K for all values of Xinh up to 0.06. As with always <1 for CF3Br in the FAA aerosol can test, and since there is
C2HF5, increasing amounts of C3H2F3Br require increasing amount sufficient water in the products, there are always sufficient H and O
of chamber volume of oxidizer to achieve Tad,peak, such that with molecules left over from the hydrocarbon oxidation (ordinarily in
Xi = 0.06, nearly all of the chamber volume is involved at the the form of H2O), to supply the H and O necessary to oxidize
fuel–oxidizer ratio corresponding to Tad,peak. Nonetheless, note that the CF3Br. A global reaction (exothermic) representing this is:
in the FAA aerosol can test experiment, even with C3H2F3Br added CF3Br + 2H2O ) CO2 + 3HF + HBr (note that HF is a more stable
product for H atoms than H2O).
C3H2F3Br

CF3Br
X inh
2000
0
2000
0.046
Tad / K

Xinh
Tad / K

0
1500
1500

0.03

0.06

1000 1000
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

η Fraction of Chamber Volume Involved in Combustion, η

Fig. 3. Calculated adiabatic equilibrium temperature (Tad) as a function of chamber Fig. 4. Calculated adiabatic equilibrium temperature (Tad) as a function of chamber
volume fraction (g) involved in combustion with the contents of the aerosol can volume fraction involved in combustion with the contents of the aerosol can
simulator. Different curves refer to different initial volume fractions of the simulator. Different curves refer to different initial volume fractions of the
suppressant C3H2F3Br in the chamber air. suppressant CF3Br in the chamber air.
G.T. Linteris et al. / Combustion and Flame 159 (2012) 1016–1025 1021

C3H2F3Br

CF3Br

2000 1.0

Tad,max / K

C2HF5
C3H2F3Br

max
0.5

CF3Br

1500 0.0
0 2 4 6 8 10 12 14

Xinh (%)

Fig. 5. Calculated maximum adiabatic equilibrium temperature (left scale) as a function of initial volume fractions of the suppressant in the chamber air, for C2HF5 (4), CF3Br
(h), C3H2F3Br (s). The fraction of chamber volume (g) involved in the combustion which gives the peak temperature is shown on the right scale for each suppressant (same
symbols); error bars: effects of water vapor.

The relative humidity (R.H.) of the ambient air is known to have


a significant effect on the behavior of HFCs in hydrocarbon flames
8
(because of the sensitivity of the system to the fluorine–hydrogen
atomic ratio [62–64]). Hence, the thermodynamic equilibrium cal- C3H2F3Br
culations (as well as the kinetic calculations described below) were
performed for water vapor in the air corresponding to 0%, 50% and
/ bar

100% R.H., at 294 K; i.e., a water vapor volume fraction in the ambi- 6
ent air of 0, 0.0125, and 0.025. For the aerosol can test chamber
max

with no inhibitor, saturating the air with water vapor (100% R.H.)
Pressure Rise at

as opposed to dry air (0% R.H.) lowered Tad by about 6% for stoichi-
ometric and lean conditions (0.34 6 g 6 1.0), while for rich condi-
4
tions (0.25 6 g 6 0.20), Tad was raised by 1–2%. With addition of
C2HF5
C2HF5, and g > 0.30, Tad was 1–5% higher with saturated air. For
CF3Br addition, Tad was at most 1% lower with saturated air. Note
than in the FAA tests, the variation of ambient water vapor content
was much smaller (6–40% R.H.) than the range examined here, 2
(0–100%). CF3Br
Based on flame equilibrium calculations, and using the value of
g which produces the peak adiabatic flame temperature Tad (gmax),
it is possible to estimate the pressure rise in the chamber [28,66].
0
In the calculation, the value of g is selected, and determines the
0 5 10 15
fraction of chamber volume oxidizer (O2–N2–agent) allowed to re- Xinh (%)
act with the fuel mix in the aerosol can simulator. An equilibrium
calculation (constant P, constant H) gives the volume and temper- Fig. 6. Calculated pressure rise based on equilibrium thermodynamics evaluated at
ature of the products, and these are allowed to mix adiabatically gmax, as a function of inhibitor volume fraction in the chamber air, for addition of
with the fraction of chamber volume (1  g) which is treated as in- C2HF5, C3H2F3Br, or CF3Br (small symbols and lines: calculation with polynomial
ert. The ideal gas law gives the pressure which this volume of prod- curve fits; large circles: FAA experimental data; error bars: effects of water vapor).
ucts would have attained. (Note that equilibrium calculations
assuming constant U, and constant V for the reacting fuel and
chamber fraction, essentially the Adiabatic Mixed Explosion Model to the case of C2HF5 at 13.5%, the pressure rise is not predicted well
[65], gave equivalent results.) with added CF3Br added at around 4%, for which the explosion is
The results of the calculations of pressure rise are shown in again completely suppressed. Apparently, kinetic effects limit the
Fig. 6, together with the experimental data from the FAA tests extent of reaction with CF3Br, or C2HF5 at the suppression point
[1,33]. The error bars (in Figs. 5 and 6) show the effects of water (or with CF3Br at lower values of Xinh). Below, we employ stirred-
vapor (the error limits showing the results for 0% R.H., and 100% reactor simulations to estimate the overall rate of reaction of the
R.H.). As indicated, for C2HF5 and C3H2F3Br, the calculations predict aerosol can test fuel with air and suppressant; as in the equilibrium
the pressure rise reasonably well, except for the case of calculations above, PSR simulations were performed for a range of
XC2HF5 = 0.135, for which the explosion was suppressed. Similar values of g and Xinh.
1022 G.T. Linteris et al. / Combustion and Flame 159 (2012) 1016–1025

4.2. Perfectly stirred reactor calculations The effect of addition of the agents on the overall chemical rate,
over a range of g and Xinh, is shown in Figs. 9 and 10. For C2HF5, Fig-
The initial conditions for the stirred-reactor calculations are ure 9 shows that the effect of C2HF5 on the overall rate depends
identical to those of the equilibrium calculations, and the results upon the value of g. At low g (rich conditions), adding agent re-
can be presented in a similar format. The stirred-reactor tempera- duces xpsr drastically. At intermediate values of g (around 0.55),
ture Tpsr is evaluated at the reactor residence time just above the addition of C2HF5 has no effect on xpsr for Xinh up to about 0.06;
blow-out condition. It is shown as a function of g and Xinh for while at high g (lean conditions), adding C2HF5 first increases the
C2HF5 and CF3Br in Figs. 7 and 8. For each of these suppressants, chemical rate, then decreases it. That is, for large fractions of the
the peak Tpsr is lower than the peak Tad, by 150–400 K, varying with chamber air involved, adding suppressant enhances the reactivity
g, Xinh, and inhibitor type. For C2HF5, Figure 7 shows that as Xinh in- of the system. Conversely, for a given amount of agent, the effect
creases, Tpsr decreases for rich conditions (low g), but increases of g is different at low Xinh versus high Xinh. For example, with no
then decreases for lean conditions (large g), illustrating the fuel- agent, xpsr is highly dependent upon g, whereas for Xinh = 0.135,
like nature of C2HF5 in the stirred reactor. Also, as with the adia- the overall chemical rate is more mildly dependent upon g (i.e.,
batic flame temperature, for a given value of Xinh, Tpsr is sensitive the curve in Fig. 9 is relatively flat). Clearly the amount of oxidizer
to g for low Xinh, but relatively insensitive to g for high g; i.e., the (air–agent–water vapor) involved in the reaction with the aerosol
curves in Fig. 7 flatten out for high Xinh and g. For CF3Br, Figure 8 can contents has a large effect on the behavior of the system.
shows that for rich flames (g < 0.33), Tpsr is unaffected by addition
of agent, while for lean flames (g > 0.33), Tpsr increases somewhat.
C2HF5
10000

X inh
C 2HF5 0

X inh 1000

1700 0
-1
ω psr / s

100
0.072
Tpsr / K

1500 0.072

10

0.133
1300
0.133
1
0.0 0.2 0.4 0.6 0.8 1.0
η
1100
0.0 0.2 0.4 0.6 0.8 1.0 Fig. 9. Overall chemical rate xpsr, for C2HF5 as a function of chamber volume
η fraction (g) involved in combustion with the contents of the aerosol can simulator.
Different curves refer to different initial volume fractions of C2HF5 in the chamber
Fig. 7. Calculated Tpsr as a function of chamber volume fraction (g) involved in air.
combustion with the contents of the aerosol can simulator. Different curves refer to
different initial volume fractions of the suppressant, C2HF5, in the chamber air.

CF3 Br
10000
CF3Br

X inh = 0.046
1000
1700
-1

X inh
ω psr / s
Tpsr / K

100 0
1500

0.024

0.046
10
1300 0

1
1100 0.0 0.2 0.4 0.6 0.8 1.0
0.0 0.2 0.4 0.6 0.8 1.0 η
η
Fig. 10. Overall chemical rate xpsr, for CF3Br as a function of chamber volume
Fig. 8. Calculated Tpsr as a function of chamber volume fraction (g) involved in fraction (g) involved in combustion with the contents of the aerosol can simulator.
combustion with the contents of the aerosol can simulator. Different curves refer to Different curves refer to different initial volume fractions of CF3Br in the chamber
different initial volume fractions of the suppressant, CF3Br, in the chamber air. air.
G.T. Linteris et al. / Combustion and Flame 159 (2012) 1016–1025 1023

Variation in the water vapor content of the air (from Xwv = 0 to contents has a large effect on the expected suppression efficiency
Xwv = 0.0125) has a varying effect on the overall chemical rate in of the two agents. As discussed above, for the agent C2HF5 it is pos-
the PSR. When small amounts C2HF5 are added, the effects are sible to estimate the values of g from the pressure rise observed in
large. For example, with Xinh = 0.02, added water vapor increases the experiments (and shown in Fig. 6). That is, if fairly large values
xpsr by 12% at g = 0.2, and decrease it by 50% at g = 0.94. With of g are not employed, it is impossible to have enough energy re-
Xinh = 0.06, the effect of water vapor is smaller, decreasing xpsr by lease to get the pressure rise observed in the FAA tests. Using Fig-
0–13% for the same range of g. With Xinh = 0.14, the decrease in xpsr ure 5, we obtain the value of gmax (i.e., g necessary to give Tad,
with water vapor addition is 26% at g = 0.4, and 8% at g = 1.0. While which was required to yield the observed pressure rise for C2HF5
these effects are significant, the variations in xpsr in Fig. 9 with g or and C3H2F3Br). These values of g are shown as the solid circles in
Xinh are much larger. Fig. 11. As shown, for C2HF5 at Xinh = 0.075, g must be about 0.47
For CF3Br addition, Figure 10 shows that the overall chemical to get the observed pressure rise, while for Xinh = 0.135, g  1.05
rate is always reduced, and the effect is much stronger for lean is required. As indicated, at the higher values of Xinh, values of g
conditions (large g). This is consistent with the known high tem- close to unity are required to give the observed pressure rise. For
perature sensitivity of the effectiveness of the HBr catalytic radical CF3Br, the specification of g is more difficult, since, as discussed
recombination cycle [66]. For example, at g = 0.6, CF3Br addition at above, the predicted pressure rise with CF3Br is insensitive to the
Xinh = 0.045 lowers the overall reaction rate by three orders of mag- value of g used in the calculation, so the pressure rise cannot be
nitude (as compared to the uninhibited case), in contrast to C2HF5 used to estimate g. (Of course, the actual pressure rise for CF3Br
(with g = 0.6), for which the inhibitor was relatively ineffective is even less than that predicted based on thermodynamics, due
when added at concentrations up to Xinh = 0.08. For CF3Br, water to kinetic influences.) If the actual g in the FAA tests were 0.33
vapor has a significant effect on xpsr at g for stoichiometric com- for CF3Br, then the PSR calculations suggest that it should not
bustion (0.33), lowering xpsr 12% at Xinh = 0% and 90% at Xinh = 0.05; extinguish the aerosol can test as effectively as C2HF5 (which is
whereas at low (0.16) or high (0.64) g, the effect is smaller, about contrary to the experimental result). Hence, it is more likely that
12–40%. the actual g for the aerosol can test with CF3Br addition is higher
The results for C2HF5 and CF3Br discussed above are plotted to- than 0.33. If g were about 0.47 for either agent, then the explana-
gether in Fig. 11. Curves for no agent, and for C2HF5 and CF3Br at tion shown by Fig. 11 is consistent with the FAA experimental re-
their suppressing concentration (Xinh,sup = 0.135 and 0.038) and sults for the agents, since CF3Br is more effective than C2HF5 at
roughly one half that value ½ Xinh,sup (Xinh = 0.075 and 0.022) are g = 0.47, as shown in Table 4. The sensitivity of the inhibitor effec-
shown. The importance of the actual value of g in the aerosol can tiveness to the value of g is consistent with the findings in some of
tests is illustrated by evaluating the effect of agent addition on xpsr the FAA tests [67] that a delay in the energy application to the igni-
at g = 0.33 and 0.47 (as denoted by the vertical dotted lines in the ter had a large effect on the explosion properties of the system.
figure and listed in Table 4). For the case of g = 0.33, xpsr = 7600 s1 That is, delaying the spark increases the mixing time prior to igni-
for the uninhibited flame, and 625 s1 and 2700 s1 for CF3Br and tion, changing the effective value of g in the system (as well as
C2HF5 at ½ Xinh,sup, respectively. That is, C2HF5 reduces the reaction changing the local gas composition near the ignitor). Similarly,
rate more than CF3Br with each added at half its suppression con- tests with CF3Br with added N2 in the oxidizer stream found the
centration. Similarly, when added at Xinh,sup, xpsr = 40 s1 and combination to work better than either agent alone. This is also
980 s1 for CF3Br and C2HF5, so that again, C2HF5 is expected to likely due to the enhanced effectiveness of CF3Br at lower temper-
slow the reaction rate more. On the other hand, referring to the ature, as illustrated in the simulations above for which higher g
dotted line in Fig. 1 for g = 0.47, xpsr = 2900 s1 for uninhibited leads to lower overall reaction rate (i.e., higher dilution with the
flames, and 1100 s1 and 90 s1 for C2HF5 and CF3Br at ½ Xinh, ext, relatively inert air stream lowers the peak temperature, increasing
and 55 s1 and 14 s1 at Xinh,sup, and so that for g = 0.47, the inhibi- the effectiveness of the HBr catalytic cycle).
tion by CF3Br is stronger. At higher values of g, the superior inhibi- To highlight the importance of the value of g on the effective-
tion by CF3Br is even greater. ness of the agents in reducing xpsr, Figure 12 shows, xpsr, as a func-
It is clear from the above discussion that the actual fraction of tion of the fraction of the suppressing concentration. The different
the chamber gases involved in the reaction with the aerosol can curves in the figure are for different assumed values of g (0.33,
0.47, and gmax). The following conclusions can be drawn from the
10000 X inh = 0 %
figure:

1. More agent generally reduces xpsr, for the assumed values of g.


X CF3Br = 2.4 %
2. For the case g = 0.47, there is little change in xpsr for the curve
1000 for C2HF5 up to 30% of the suppression value.
X CF3Br = 4.6 % X C2HF5 = 7.2 %
3. For C2HF5 (blue curves), the reduction in xpsr with addition of
agent is similar regardless of the value of g; i.e., for g = 0.33,
-1
ω psr / s

g = 0.47, or g(Tad,peak).
100

X C2HF5 = 13.3 %

Table 4
10 Overall chemical rate predicted by the stirred reactor calculation for C2HF5, CF3Br, and
2-BTP simulant at values of constant g = 0.33 and g = 0.47.

Overall chemical rate, xpsr (s1)


g: 0.33 0.47
1
0.0 0.2 0.4 0.6 0.8 1.0 Agent: C2HF5 CF3Br C2HF5 CF3Br
η Fraction of suppression volume fraction (%)
0 7600 7600 2900 2900
Fig. 11. Overall chemical rate xpsr, for C2HF5 and CF3Br as function of g; curves are 50 625 2700 1100 55
given for each agent at the extinction volume fraction, and approximately one half 100 40 980 90 14
of that value.
1024 G.T. Linteris et al. / Combustion and Flame 159 (2012) 1016–1025

10000 CF3Br, the reduction in the overall reaction rate with addition of
CF3Br is highly dependent upon g. Hence, the relative effectiveness
of these agents for reducing the overall reaction rate in the stirred
reactor simulations is highly dependent upon the premixed state of
the reactants—the assumed composition of which depends upon
CF3 Br, η = 0.33
1000 the mixing state in the FAA test. That is, the relative effectiveness
of these agents for inerting the explosion of the FAA test may be
ω psr / s -1

dependent upon the particular test conditions employed.

Acknowledgments
100 C2HF5, η= 0.47
The authors thank Wing Tsang of NIST, and John Reinhardt at
C2HF5, η= 0.33 the FAA Technical Center for essential input to the work. Ken Smith
C2HF5, η=ηmax and Med Colket at the United Technologies Research Center per-
formed ab initio calculations for C3H2F3Br and graciously provided
CF3 Br, η = 0.47
10
the thermodynamic data for the C3H2F3Br molecule, allowing us to
0 50 100 perform the thermodynamic calculations. Mike Zehe at NASA
Xinh / Xinh,supp (%) Glenn Research Center assisted with implementing the C3H2F3Br
data in to the NASA CEA program. Brad Williams at NRL kindly pro-
Fig. 12. Overall chemical rate xpsr, for all C2HF5 (D) and CF3Br (C) as function of the vided his changes to the NIST HFC mechanism. The work was sup-
fraction of suppression concentration, for different values of g. ported by The Boeing Company.

4. The effectiveness of CF3Br is very sensitive to the value of g. References


5. For CF3Br to be more effective than C2HF5, g must be greater
[1] J.W. Reinhardt, Behavior of Bromotrifluoropropene and Pentafluoroethane
than about 0.4. When Subjected to a Simulated Aerosol Can Explosion, DOT/FAA/AR-TN04/4,
Federal Aviation Administration, 2004.
[2] J.W. Reinhardt, Minimum Performance Standard for Aircraft Cargo
Compartment Halon Replacement Fire Suppression Systems (2nd Update),
5. Conclusions
DOT/FAA/AR-TN05/20, Federal Aviation Administration, 2005.
[3] T. Marker, Initial Development of an Exploding Aerosol Can Simulator, DOT/
Chemical equilibrium and perfectly-stirred reactor calculations FAA/AR-TN97/103, Federal Aviation Administration, 1998.
have been performed for the purpose of understanding the unex- [4] D.R. Blackmore, G. O’Donnell, R.F. Simmons, Proc Combust. Inst. 10 (1965)
303–310.
pected enhanced combustion in the FAA Aerosol Can Explosion [5] L.A. Lovachev, V.T. Gontkovskaya, N.I. Ozerkovskaya, Combust. Sci. Technol. 17
simulation test. The equilibrium calculations were used to predict (1977) 143–151.
the overpressure in the FAA tests with C2HF5 or C3H2F3Br added at [6] L.A. Lovachev, L.N. Lovachev, Combust. Sci. Technol. 18 (1978) 191–198.
[7] I.O. Moen, P.A. Thibault, J.H. Lee, R. Knystautas, T. Dean, C.K. Westbrook, Proc.
sub-inertion concentrations. With either of these agents, the ob- Combust. Inst. 20 (2008) 1717–1725.
served pressure rise is a strong function of both the fraction of [8] V.I. Babushok, D.R.F. Burgess, W. Tsang, A.W. Miziolek, in: Halon Replacements,
chamber volume involved in the combustion (g), and the amount 1995.
[9] N. Saito, Y. Saso, C.H. Liao, Y. Ogawa, Y. Inoue, in: Halon Replacements, 1995.
of added agent. At increasing agent volume fractions, the observed [10] H.F. Coward, G.W. Jones, Limits of Flammability of Gases and Vapors,
large pressure rise is predicted only if a large fraction of the cham- AD0701575, US bureau of Mines, 1952.
ber is involved, corresponding to near stoichiometric conditions [11] T.A. Moore, D.S. Diedorf, S.R. Skaggs, in: Proceedings of the 1993 CFC & Halon
Alternatives Conference, 1993.
(including oxygen demand from reaction of the agent), with reac- [12] S. Kondo, K. Takizawa, A. Takahashi, K. Tokuhashi, A. Sekiya, Fire Saf. J. 44
tion to equilibrium products. On the contrary, the pressure rise (2009) 192–197.
with added CF3Br is shown to be nearly independent of both the [13] Number Designation and Safety Classification of Refrigerants, ANSI/ASHRAE
Standard 34-2007, American Society of Heating, Refrigerating and Air-
fraction of chamber volume involved, and the amount of added
Conditioning Engineers, 2007.
CF3Br. This result is shown to be due to the atom balance in the [14] Y.N. Shebeko, V.V. Azatyan, I.A. Bolodian, V.Y. Navzenya, S.N. Kopyov, D.Y.
CF3Br–hydrocarbon–air system, for which added CF3Br is predicted Shebeko, E.D. Zamishevski, Combust. Flame 121 (2000) 542–547.
to never show an enhanced pressure rise (even if the typical strong [15] G.W. Gmurczyk, W.L. Grosshandler, Fire Safety Science – Proceedings of the
Fourth International Symposium, vol. 4, International Association for Fire
chemical inhibition with CF3Br were not to occur). Nevertheless, Safety Science, 1994, pp. 925–936.
the high reactivity for the C2HF5 and C3H2F3Br at the concentra- [16] G. Gmurczyk, W. Grosshandler, Proc. Combust. Inst. 25 (1994) 1497–1503.
tions of the FAA tests are somewhat unexpected. [17] G.W. Gmurczyk, W.L. Grosshandler, in: W. Grosshandler, R.G. Gann, W.M.
Pitts (Eds.), Evaluation of Alternative In-Flight Fire Suppressants for Full-Scale
To understand the role of kinetics, the inhibition effects of the Testing in Simulated Aircraft Engine Nacelles and Dry Bays, National Institute
agents on the overall reaction rate of the system were examined of Standards and Technology, Gaithersburg, MD, 1995.
through stirred reactor calculations. Calculations were performed [18] A. Hamins, G. Gmurczyk, W.L. Grosshandler, C. Presser, K. Seshadri, in: W.L.
Grosshandler, R.G. Gann, W.M. Pitts (Eds.), Evaluation of Alternative In-Flight
for C2HF5, and CF3Br, over a range of agent concentrations and frac- Fire Suppressants for Full-Scale Testing in Simulated Aircraft Engine Nacelles
tion of chamber oxidizer involved in the combustion. The amount and Dry Bays, NIST SP 861, National Institute of Standards and Technology,
of oxidizer (air + agent) in the test chamber which participates in Gaithersburg, MD, 1994.
[19] G. Holmstedt, P. Andersson, J. Andersson, in: Fire Safety Science – Proceedings
the high-temperature reaction with the fuel (i.e., the amount of of the Fourth International Symposium, International Association of Fire Safety
chamber test volume which mixes with and reacts with the fire Science, vol. 4, 1994, pp. 853–864.
ball of the fuel reaction), is found to have a major influence on [20] V.R. Katta, F. Takahashi, G.T. Linteris, Combust. Flame 144 (2006) 645–661.
[21] E.A. Ural, Process Saf. Prog. 22 (2003) 65–73.
the relative effectiveness of each agent.
[22] S. Kondo, A. Takahashi, K. Takizawa, K. Tokuhashi, Fire Saf. J. (2010).
For the premixed system modeled by the PSR, the role of excess [23] J.R. Mawhinney, B.Z. Dlugogorski, A.K. Kim, in: Fire Safety Science –
oxidizer (air plus agent) entrainment with the reaction zone varies Proceedings of the Fourth International Symposium, International
with the agent. For C2HF5, the reduction in the overall reaction rate Association for Fire Safety Science, vol. 4, 1994, pp. 47–60.
[24] A. Hamins, K. McGrattan, G.P. Forney, Unwanted Accelerated Burning After
with addition of agent is found to be relatively insensitive to the Suppressant Delivery, SP-1004, National Institute of Standards and
fraction of chamber volume involved (g). On the other hand, for Technology, 2003.
G.T. Linteris et al. / Combustion and Flame 159 (2012) 1016–1025 1025

[25] D. Trees, K. Seshadri, A. Hamins, in: A.W. Miziolek, W. Tsang (Eds.), Halon [43] D. L’esperance, B.A. Williams, J.W. Fleming, Combust. Flame 117 (1999) 709–
Replacements: Technology and Science, ACS Symposium Series 611, American 731.
Chemical Society, Washington, DC, 1995. [44] Y. Saso, D.L. Zhu, H. Wang, C.K. Law, N. Saito, Combust. Flame 114 (1998) 457–
[26] E.J.P. Zegers, B.A. Williams, E.M. Fisher, J.W. Fleming, R.S. Sheinson, Combust. 468.
Flame 121 (2000) 471–487. [45] K. Takahashi, Y. Sekiuji, Y. Yamamori, T. Inomata, K. Yokoyama, J. Phys. Chem.
[27] M.R. Nyden, G.T. Linteris, D.R.F. Burgess, P.R. Westmoreland, W. Tsang, M.R. A 102 (1998) 8339–8348.
Zachariah, Evaluation of Alternative In-Flight Fire Suppressants for Full-Scale [46] Y. Yamamori, K. Takahashi, T. Inomata, J. Phys. Chem. A 103 (1999) 8803–
Testing in Simulated Aircraft Engine Nacelles and Dry Bays, NIST SP 861, 8811.
National Institute of Standards and Technology, 1994. [47] B. Vetters, B. Dils, T.L. Nguyen, L. Vereecken, S.A. Carl, J. Peeters, Phys. Chem.
[28] G.T. Linteris, F. Takahashi, V.R. Katta, H.K. Chelliah, O. Meier, in: Fire Safety Chem. Phys. 11 (2009) 4319–4325.
Science – Proceedings of the Tenth International Symposium, International [48] A. Fernandez, A. Fontijn, J. Phys. Chem. A 105 (2001) 8196–8199.
Association for Fire Safety Science, 2012, in press. [49] C.P. Tsai, D.L. Mcfadden, J. Phys. Chem. 94 (1990) 3298–3300.
[29] T. Noto, V. Babushok, A. Hamins, W. Tsang, Combust. Flame 112 (1998) 147– [50] J.S. Francisco, J. Chem. Phys. 111 (1999) 3457–3463.
160. [51] V. Babushok, T. Noto, D.R.F. Burgess, A. Hamins, W. Tsang, Combust. Flame 107
[30] G.T. Linteris, D.R. Burgess, V. Babushok, M. Zachariah, W. Tsang, P. (1996) 351–367.
Westmoreland, Combust. Flame 113 (1998) 164–180. [52] G.T. Linteris, M.D. Rumminger, V.I. Babushok, W. Tsang, Proc. Combust. Inst. 28
[31] G.T. Linteris, F. Takahashi, V.R. Katta, Combust. Flame 149 (2007) 91–103. (2000) 2965–2972.
[32] F. Takahashi, G.T. Linteris, V.R. Katta, Proc. Combust. Inst. 31 (2006) 2721– [53] N.M. Marinov, Int. J. Chem. Kinet. 31 (1999) 183–220.
2729. [54] P. Saxena, F.A. Williams, Proc. Combust. Inst. 31 (2007) 1149–1156.
[33] J.W. Reinhardt, Prevention of a Simulated Aerosol Can Explosion With a [55] A. Frassoldati, A. Cuoci, T. Faravelli, E. Ranzi, Combust. Sci. Technol. 182 (2010)
Mixture of Halon 1301 and Nitrogen, DOT/FAA/AR-TN08/49, Federal Aviation 653–667.
Administration, 2008. [56] F.A. Williams, J. Fire Flamma. 5 (1974) 54–63.
[34] W.C. Reynolds, The Element Potential Method for Chemical Equilibrium [57] R.B. Barat, Chem. Eng. Sci. 56 (2001) 2761–2766.
Analysis: Implementation in the Interactive Program STANJAN, Version 3, [58] S. Liu, M.C. Soteriou, M.B. Colket, J.A. Senecal, Fire Saf. J. 43 (2008) 589–597.
Stanford University, 1986. [59] P. Glarborg, R.J. Kee, J.F. Grcar, J.A. Miller, PSR: A FORTRAN Program for
[35] S. Gordon, B.J. McBride, Computer Program for Calculation of Complex Modeling Well-Stirred Reactors, SAND86-8209, Sandia National Laboratories,
Chemical Equilibrium Compositions and Applications, NASA Reference 1986.
Publication 1311, NASA Glenn Research Center, 1996. [60] G.T. Linteris, O. Meier, Unwanted Combustion Enhancement by Potential
[36] D.A. Sheen, X.Q. You, H. Wang, T. Lovas, Proc. Combust. Inst. 32 (2009) 535– Halon Replacements: Thermodynamic Considerations, NIST Technical Note
542. 1693, National Institute of Standards and Technology, 2011.
[37] H. Wang, X. You, K.W. Jucks, S.G. Davis, A. Laskin, F. Egolfopoulos, C.K.Law, USC [61] I. Glassman, Combustion, Academic Press, San Diego, CA, 1996.
Mech Version II. High-Temperature Combustion Reaction Model of H2/CO/C1– [62] G.T. Linteris, in: A.W. Miziolek, W. Tsang (Eds.), Halon Replacements, ACS
C4 Compounds, University of Southern California, 2007. <http://ignis.usc.edu/ Symposium Series 611, American Chemical Society, Washington, DC, 1995.
USC_Mech_II.htm>. [63] K. Takizawa, A. Takahashi, K. Tokuhashi, S. Kondo, A. Sekiya, Combust. Flame
[38] J. Li, A. Kazakov, M. Chaos, F.L. Dryer, in: Proceedings of the Fifth Joint Meeting 141 (2005) 298–307.
of the US Sections of the Combustion Institute, Combustion Institute, 2007, p. [64] G.T. Linteris, G.W. Gmurczyk, in: HOTWC-1995, National Institute of Standards
Paper 26. and Technology, 2006, pp. 227–238.
[39] J. Li, A. Kazakov, F.L. Dryer, Int. J. Chem. Kinet. 33 (2001) 859–867. [65] R.A. Ogle, Process Saf. Prog. 18 (1999) 170–177.
[40] J. Li, A. Kazakov, F.L. Dryer, J. Phys. Chem. A 108 (2004) 7671–7680. [66] G. Dixon-Lewis, R.J. Simpson, Proc. Combust. Inst. 16 (1977) 1111–1119.
[41] D.R. Burgess, M.R. Zachariah, W. Tsang, P.R. Westmoreland, Prog. Energy [67] J.W. Reinhardt, in: Aircraft Cargo Compartment ‘‘Testing Update’’,
Combust. Sci. 21 (1995) 453–529. International Aircraft Systems Fire Protection Working Group Meeting,
[42] D. Burgess, M.R. Zachariah, W. Tsang, P.R. Westmoreland, Thermochemical and Federal Aviation Administration, 2007.
Chemical Kinetic Data for Fluorinated Hydrocarbons, NIST Technical Note
1412, National Institute of Standards and Technology, 1995.

Das könnte Ihnen auch gefallen