Sie sind auf Seite 1von 10

Fire Safety Journal 46 (2011) 558–567

Contents lists available at SciVerse ScienceDirect

Fire Safety Journal


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

A thermal theory for estimating the flammability limits of a mixture


Tingguang Ma n
Boots & Coots Center for Fire Safety & Pressure Control, Department of Fire Protection and Safety, Oklahoma State University, 492 Cordell South, Stillwater, OK 74078, USA

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

Article history: Because it is difficult to treat the contributions of diluents explicitly using Le Chatelier’s rule, a
Received 10 September 2010 methodology based on thermal balance is proposed for estimating the flammability limits of a mixture.
Received in revised form This method converts the flammability information of a mixture into a binary domain of heating/
5 September 2011
quenching potentials and, after some simple manipulations, converts them back into the flammability
Accepted 7 September 2011
Available online 29 September 2011
domain. The advantage of this conversion is the separation of the heating and quenching potential
sums. The dual contribution (heating and quenching) of each species is stressed, while the simplicity of
Keywords: hand calculation is preserved. This method is equivalent to Le Chatelier’s rule but has increased
Flammability limits flexibility in dealing with various fuel/oxygen/diluents combinations. It will help safety engineers gain
Flammable mixture
more confidence in the hazard analysis of flammable mixtures involving diluents.
Le Chatelier’s rule
& 2011 Elsevier Ltd. All rights reserved.
Thermal balance

1. Introduction limits, is widely applied to predict approximate upper limits.


Mashuga and Crowl [17] provided a derivation of this rule using
The role of flammability in characterizing fire and explosion necessary assumptions.
hazards has long been recognized in both academia and industry, For the third category, LCR cannot be directly used. To apply LCR,
as noted by Britton [6,7], who reviewed 200 years of research the flammability of a pseudo-fuel is established to incorporate the
efforts regarding its measurement. He also stressed the difficulty in contribution of an inert. Zabetakis [27] and Haessler [11] address the
measuring flammability, showing the need to understand the contributions of a diluent by pairing one fuel with one diluent to
process using various equipment for hazard analysis. Recent explo- produce a new pseudo-fuel. The flammability of such a pseudo-fuel
sion disasters, such as the natural gas explosion in Connecticut is referenced from a diluent-fuel-ratio flammability diagram (Fig. 1b).
(February 7, 2010) and the BP explosion in the Mexican Gulf (April This diagram is converted from the experimental flammability
24, 2010), have brought a new wave of attention to understanding diagram (Fig. 1a). Though primitive and imprecise, this diagram
the flammability of mixtures. helps determine the flammability of a fuel-diluent combination,
According to Hansel et al. [12], common problems with making LCR more universally applicable.
flammability limits are: Kondo et al. [29] tried to use an extended LCR to incorporate
the contribution of an inert. However, they simply provide an
(1) Single fuel and inert in air; experimentally determined coefficient to the concentration of an
(2) Multiple fuels and an inert (here inert and diluent are inert, and no theory is referenced for doing so. Thus, their
interchangeable for a species contributing no thermal energy extension is not applicable to cases that have not been tested.
in a combustion); Recognizing the problems with LCR, Hansel et al. [12] tried to
(3) Single fuel or multiple fuels with multiple inerts. estimate the possibility of ignition using the critical adiabatic
flame temperature (CAFT). Their method requires another numer-
ical tool to determine the CAFT, which does not lend itself to hand
Regarding the first category, Coward et al. [28], Zabetakis [27] calculation. To deal with the ignition potential of hot gases in a
and Kuchta [14] compiled much original flammability informa- ceiling layer, Beyler [3] used a similar method for the burning
tion for chemicals of different classes, which has served as the potential of a hot layer. Though universal and general in theory,
major source of information for industrial applications. his approach is a little tedious (employing both the concentration
Regarding the second category, Le Chatelier’s rule (referenced and the temperature information of combustion products) and
as LCR hereafter), although developed for lower flammability can only be applied to lower limits.
Shebeko et al. [22] studied the analytical solution to the
n
Tel.: þ405 744 8772 (o). flammability limits of gaseous mixtures of combustible-oxidizer-
E-mail address: ting.ma@okstate.edu diluents. Too many efforts were focused on the details of reactions

0379-7112/$ - see front matter & 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.firesaf.2011.09.002
T. Ma / Fire Safety Journal 46 (2011) 558–567 559

Nomenclature xL lower flammability limit (volume ratio)


(% or dimensionless)
MSDS material safety data sheet xst stoichiometric fuel/air volume concentration
CAFT critical adiabatic flame temperature (K) (% or dimensionless)
CO the oxygen coefficient in a reaction (dimensionless) xU upper flammability limit (volume ratio)
Cst the stoichiometric number for a reaction (% or dimensionless)
(dimensionless) y variable volumetric concentration (%)
HQR heating/quenching ratio (dimensionless)
HO the heating potential of oxygen based on air Subscripts
(dimensionless)
HF the heating potential of fuel based on air a ambient
(dimensionless) d diluent
LFL lower flammability limit (volume ratio) i component of a mixture
(% or dimensionless) m sum of a mixture
LCR Le Chatelier’s rule st stoichiometric
QD the quenching potential of diluent based on air AF Adiabatic Flame
(dimensionless) L lower flammable limit
QF the quenching potential of fuel based on air U upper flammable limit
(dimensionless) D diluent-based potential to air potential
Ta ambient temperature (K) F fuel-based potential to air potential
UFL upper flammability limit (volume ratio) O oxygen-based potential to air potential
(% or dimensionless)

(multi-step reaction), which significantly limited their method’s 2. A thermal look at flammability
utility. Special attention was paid to the extinction condition, which
lost the generality of a thermal system when stressing the chem- 2.1. A binary view of a tertiary (fuel/oxidizer/diluent) system
istry involved.
Without a direct means of computing the flammability of a When dealing with flammability limits, it is common practice
mixture, one must resort to complex algorithms for thermody- to present data in a tertiary axis system [27]. Sometimes, a two-
namic computations (for example, see [18]), which is not con- axis plot is used with an implicit assumption that the third axis is
venient for field workers, who carry calculators instead of oxygen, which can be derived from the two known axes [2,3]. This
computers for field inspection. Though computers are more treatment is useful in presenting the data and showing the trend
powerful than calculators, more training is required for their regarding safety. However, the real contribution of each species to
use, not to mention the black-box treatment of proprietary soft- the flammability is not clear in this diagram. Specifically, the
ware. There is still a need to find a simple method that is quenching capabilities of a fuel and oxygen are overlooked, so
comparable to LCR and suitable for field estimation but that has there is little theoretical consistency in making general flamm-
more universal applications (coverage) for the purpose of hazard ability predictions.
analysis. This work provides such an estimation methodology. In a fuel/oxidizer/diluent system, each species contributes to
This paper is organized as follows. A literature review is the flammability either by heating (releasing energy) or by
provided for insight into the current status of manipulating quenching (absorbing energy). All agents have a mass, so they
flammable mixtures. A thermal theory is proposed to determine absorb energy during the ignition process (to overcome the
the flammability of a gaseous fuel, with special emphasis on the barrier of an Arrhenius reaction to reach the flame temperature).
contribution of all reactants before the reaction is initiated. The Some agents will release energy to maintain the flame tempera-
flammability diagram is analyzed as an application of the pro- ture, which can prevent the reaction from extinction. When fuel is
posed thermal theory. The limitations of the proposed method abundant, which is the case near the upper limits, the availability
and comparisons with other methods are provided in Section 6. of oxygen determines the total amount of energy released.
The equivalence of this methodology to LCR is established in the Fortunately, oxygen calorimetry theory tells us that the energy
Appendix A. released based on oxygen consumption is relatively constant

Fig. 1. Classical method for computing the flammability of a fuel mixture with atleast one diluent.
560 T. Ma / Fire Safety Journal 46 (2011) 558–567

by mass for most fuels [13]. This empirical rule is helpful in which is reasonably small for estimation purposes. In addition,
generating information at the upper limits. with the exception of helium, the quenching potential is close to
The reason for this choice is that most previous work does not the molecular ratio between the species. This is generally true and
recognize the quenching role of reactants (fuel or oxygen) or the quite intuitive. The larger a molecule is, the more energy it will
oxygen-limited energy release at upper flammability limits. absorb to raise its temperature. Thermal capacity is a function
Therefore, the contribution of an inert cannot be incorporated closely related to molecular weight [5].
or manipulated easily. By isolating the binary contribution of each One exception is noted for CO2. From the thermal properties of
component, the flammability limits are easier to determine from CO2, we have a quenching potential of 1.603 (measured against
the competing roles of heating and quenching. that of air). However, when predicting the inerting concentration
Thus, a binary system is constructed in this work. A fuel (at of carbon dioxide, Senecal [21] found that the quenching potential
lower limits) and oxidizer (at upper limits) determine the total of carbon dioxide is more than 1.603, which is probably due to the
energy release, whereas all agents absorb energy in the process of radiative loss term at the CAFT or the decomposition energy term
maintaining the flame temperature. The balance or competition of CO2 [32]. Using Senecal’s data, the difference is found to be 9%
between the heating and the quenching capability determines of the quenching potential; thus, a value of 1.75 is used in this
the flammability limits. To perform a hand calculation, a simple work. If the radiative property is the major reason, it should be
method is needed to scale all agents to comparable values. noted that only two diluents have this feature; the other is water
Ambient air is one such omnipresent agent that can be used for vapor. However, little attention was paid to water vapor as a
scaling purposes. suppression agent.
Without recognizing the quenching potential of oxygen,
Schroeder and Molnarne [19,20] used a dimensionless value K, a
2.2. Scaling thermal properties coefficient of nitrogen equivalency, to scale the inerting effect of
each diluent. Unfortunately, the relationship between this K value
The major difficulty in applying the flame temperature theory and the energy-absorbing potential of this species was not
to calculate flammability is dealing with the non-linearity of explored further, so there are various K values that exist for
thermal properties. NIST maintains a complete and up-to-date different standards. This significantly increased the uncertainty in
database with polynomial correlations for the thermal properties hazard analysis. Here, oxygen also contributes to the quenching
of species (see NIST chemistry WebBook) [1]. However, such potential, and almost all flammability tests are performed in air;
information is not easy to manipulate through hand calculations, thus, air is a natural choice for scaling.
nor is it readily available. Usually, an averaged or a middle value The purpose of this scaling is to avoid the non-linearity of
is used for hand calculations [3,10]. The approximation due to the specific heat. Because most species share the similar non-linearity
assumption of the reaction temperature is the major source in the target temperature range (from ambient to the CAFT), the
of error. fluctuation of their ratio is small in this range (1300 K–1850 K).
In this work, the ignition is assumed to occur at a certain Thus, the integration of the average specific heat over a certain
temperature around 1600 K [3,4,30]. Hereafter, this value is range [3] or the numerical determination of the CAFT [12] is
referred to as the CAFT (critical adiabatic flame temperature). avoided. This makes hand calculation feasible. Some researchers
Zabetakis [27] determined the relatively constant CAFT of most (such as Hansel et al. [12], Schroeder and Molnarne [19,20]) use
fuels, though the original finding has been attributed to Burgess nitrogen equivalency to scale the inerts without recognizing the
and Wheeler [31]. quenching potential of a fuel or oxygen. The latter is the major
The enthalpy difference observed for air between ambient contribution of this work.
temperature and the CAFT can be the baseline unit for scaling all
other agents in the combustion process. The humidity in air may 2.3. Major assumptions/derivatives
change this enthalpy value significantly; however, humidity is not
considered in all flammability measurements and will not be To scale other energy terms to those of air, the implicit
included here. Thus, the non-linearity is removed if all agents can assumption is that the CAFT will be constant for all fuels at
be compared to each other via a non-dimensionalized potential ignition. With limited exceptions, this rule generally applies and
term. This scaling process can simplify the estimation work has already been used by Sheinson et al. (1989) for suppression
significantly. and Beyler [3,4] for flammability. In fact, this is the basis for all
Here, the quenching potential of a diluent (or oxidizer/fuel) is estimation methods for mixture properties, including LCR [17]. It
the enthalpy change at the CAFT with respect to that of air. From is known that the critical condition at the UFL will be dominated
the NIST online database (2010), we can compare the quenching more by incomplete combustion [3]. However, the incomplete-
potentials for the following diluents at various levels of tempera- ness of a reaction will be scaled up or down in the estimation
ture (Table 1). Here, 1300 K is the CAFT of CO, 1600 K is the CAFT process accordingly.
for most hydrocarbons at ignition, and 1850 K is close to the CAFT Oxygen calorimetry (constant energy release per oxygen con-
at complete inertion [21]. The maximum fluctuation in quenching sumption by mass) is another assumption; that is, the oxygen
potentials is about 3.6% (helium, in the range of 1300 K–1850 K), consumption is closely related to the total energy release by the
system. Because oxygen-based energy release is roughly constant
Table 1 for most fuels [13], a relationship between fuel-based energy and
Quenching potentials of some gaseous agents (diluents). oxygen-based energy can be established. This is especially useful
in the information at upper flammability limits (UFL), when
Temp (K) AFT (K) He Argon N2 O2 CO2 Air
oxygen availability dominates a reaction. It is known that the
QDA ¼ ECAFT 1300 0.655 0.655 0.991 1.046 1.578 1.000 reaction at the UFL is incomplete [3], but it is difficult to
Ea

QDA ¼ ECAFT 1600 0.642 0.642 0.992 1.046 1.603 1.000 characterize. Therefore, it is assumed that a reaction at UFL is
Ea
ECAFT 1850 0.632 0.632 0.992 1.046 1.615 1.000 complete and similar to a reaction at LFL.
QDA ¼ Ea
MW 2.000 18.000 28.000 32.000 44.000 28.840
From the major assumption of constant flame temperature,
MW 0.069 0.624 0.971 1.110 1.526 1.000 several derivatives can be proposed. First, a binary thermal
MWair
system is additive with respect to its components. Catalytic or
T. Ma / Fire Safety Journal 46 (2011) 558–567 561

synergistic effects are precluded in the estimations. Some gases, has a quenching potential alone. Here, quenching potential and
such as H2, will change the flame temperature due to their heating potential are abstract (or dimensionless) properties of a
reactivity; therefore, it is difficult to apply this method directly. certain species. The difference between the heating and quench-
Some diluents, such as Halon 1301, will raise the flame tempera- ing potentials is measured as the heat of combustion of a fuel.
ture by scavenging the reacting radicals (Sheinson et al., 1989). Under a critical condition, such as at the lower flammable limit,
Their role is synergistic because they increase the quenching we can set up the energy balance shown in Eq. (3). For LFL, the left-
potential of all species involved (due to the raised temperature hand side of Eq. (3) is the total quenching (energy absorbing)
threshold for a reaction to take place, see Williams [26]). With a potential, which is obtained from a fuel and air, whereas the right-
constant CAFT, the quenching potential and the heating potential hand side is the total energy released by the fuel only. The low
are interchangeable between different fuels. This limitation flammability limit is the point at which the heat released is barely
makes the theory universally valid among thermal agents. enough to raise the system temperature to the CAFT or the minimal
A further derivative of the constant CAFT is the concept of a energy input required for the system to reach the critical condition
baseline enthalpy. Flammability limits are volume percentage or a at the CAFT.
dimensionless unit with respect to the background (air in this case).
xL QF þð1xL Þ ¼ xL HF ð3Þ
Thus, the enthalpy difference observed for air between ambient
temperature and the CAFT is used as the baseline enthalpy and used where x stands for volumetric ratio or molar concentration, and xL is
everywhere in this paper to non-dimensionalize the equations. the volumetric ratio at the lower flammable limit or LFL. All
Using this baseline enthalpy, the exact value of the CAFT is not parameters are scaled by one unit volume of air, so this equation
important (see Table 1). Everything is simplified to a form compar- is already non-dimensionalized.
able to the flammability limits. Similarly, we can set up the energy balance for the upper
In summary, we have made two major assumptions: flammable limit using the oxygen-based energy release for a fuel-
rich flame. Again, the left-hand side is the total heat sink, whereas
 Constant CAFT; and the right-hand side is the total energy release due to oxygen
 Oxygen calorimetry applies at both lower and upper limits consumption. Here, xU is the volumetric ratio at the upper
(or the reaction at UFL is complete and similar to the reaction flammable limit.
at LFL). 1xU
xU QF þ ð1xU Þ ¼ HO ð4Þ
4:773
From the first assumption, two derivatives are generated and
used here: here we have two equations with three unknowns, which are
difficult to solve directly. However, we have the constitutive
 A thermal system is binary and additive; relationship (for oxygen calorimetry) shown in Eq. (5).
 The properties of air can be used to scale other agents. HF ¼ C O HO ð5Þ
By combining the above three equations, the energy terms can
3. Development of the theory
be computed as follows:

To preserve generality, the following simple and universal 1 CO ðxU xL Þ


QF ¼ 1 þ ð6Þ
combustion reaction is considered. xL CO xU xL ððð1xU ÞxL Þ=4:773Þ
Ca Hb Oc þ CO ðO2 þ 3:773N2 Þ þCd D-aCO2 þ 0:5bH2 O xU xL
þ 3:773aN2 þCd D ð1Þ HO ¼ ð7Þ
CO xU xL ððð1xU ÞxL Þ=4:773Þ

CO ¼ a þ b=4c=2 For a certain fuel/diluent system, the fuel properties (CO, xL and
Cst ¼ 1þ 4:773CO ð2Þ xU) can be found via any MSDS (Material Safety Data Sheet). Thus,
QF, HO and HF can be calculated from the material properties at the
where D stands for diluent, CO is the oxygen coefficient represent- two critical conditions. Derivatives of the flammability limits,
ing chemistry or a stoichiometric oxygen number, and Cst is the these terms are unique to each fuel, so they can be treated as the
stoichiometric number. Caution is taken here such that Cst is thermal signature of each fuel. Referring back to the critical
commonly used (such as [27,3]) for the slope of the stoichiometric concentration, we can determine the following relationships:
line. In the work by Zabetakis and Beyler, Cst is, in fact, an inverse
term of the stoichiometric number. In this work, xst ¼ 1=Cst is 1
xL ¼ ð8Þ
used for the stoichiometric concentration of a fuel/air mixture. 1 þHF QF
A more complex version of Eq. (1) can be found elsewhere (for
HO =4:7731
example, [25]). xU ¼ ð9Þ
HO =4:7731þ QF
Generally, CO is the single most important parameter repre-
senting the chemical difference between fuels. For example, Eq. (8) is used for the lower flammability limit and can be
Catoire and Naudet [9] use the same CO in producing correlations understood as the minimum excess heating potential of fuel
for flammability estimation because this parameter reflects the (HF QF) that heats one unit of air (QA ¼1) to the CAFT. Eq. (9) is
consumption of oxygen calorimetry or final energy release. This used for the upper flammability limits and can be understood as
variable is widely adopted in making flammability correlations. the minimum excess heating potential of air (HO/4.76 1 due to
With the baseline enthalpy, all energy terms are introduced in oxygen consumption because the fuel is in excess) that heats one
a non-dimensional form scaled by one unit volume of air. Here, Q unit of fuel (QF) to the CAFT. Thus, the physical meanings of
stands for the quenching potential (the required energy to raise flammability limits are rather clear.
the mixture temperature from the ambient value to CAFT) and H It seems circular to use HFQFHO for xLxU while using xLxU for
stands for the heating potential (the energy released during the HFQFHO. During the transformation process, the physical meaning
flammability tests to raise the reference mixture from the of xLxU is clear, and the energy terms HFQFHO are additive and easy
ambient temperature to the CAFT). Fuel and oxygen both con- to manipulate with increased flexibility. The measurable quantity
tribute to quenching and heating, respectively, whereas an inert for the excess heating potential HF–QF is the heat of combustion,
562 T. Ma / Fire Safety Journal 46 (2011) 558–567

DHC , so a correlation of DHC can be derived for xL, which will be 3


discussed later. Here, we only focus on the manipulation of the
HQR1 (Methane)
energy terms in a direct way (compared to using correlations).
HQR2 (Methane)
For these energy potential terms in a mixture, several additive
HQR = 1
rules apply. The following are some additive rules used in the
mixing process. 2
X

HQR
yi ¼ 1
i
X
CO,m ¼ CO,i yi
i 1
X
QF,m ¼ QF,i yi
i
X
HF,m ¼ HF,i yi
X
i 0
HO,m ¼ HF,i CO,i yi ð10Þ 0 0.05 0.1 0.15 0.2
i Volume Yield
Therefore, the flammability limits of the mixture are predicted Fig. 2. HQR curves for methane explaining the physical meaning of xL and xU.
using Eqs. (11) and (12), which are similar to Eqs. (8) and (9), with
the energy potentials being replaced by mixture values.
1
xL,m ¼ ð11Þ
1þ HF,m QF,m

HO,m =4:7731
xU,m ¼ ð12Þ
HO,m =4:7731þ QF,m
The above methodology converts the concentrations into their
counterparts in the energy domain. This is the first step in
obtaining the thermal signature of each fuel. For mixtures, the
manipulations are performed in the energy domain to include the
contribution of various reactants (step 2). Finally, the lumped
energy terms will be converted back to the flammability (or
concentration) domain (step 3). With the foregoing assumptions
and through the conservation of energy, this conversion is
physically meaningful and easy to manipulate.

4. Thermal analysis on flammability limits


Fig. 3. Modification of the flammability diagram by adding a diluent to methane.
4.1. Flammability limits

Table 2
The critical energy balance at the flammability limits can be
Spreadsheet calculation on the flammability limits of a mixture (methane 15.2%,
illustrated graphically using a new variable, the heating-quench- nitrogen 79.8%, and carbon dioxide 0.5%).
ing ratio (HQR), which is defined as the heating potential over the
quenching potential (as shown in Eqs. (13) and (14), which are Fuel CO yi xL xU QF HO
variations of Eqs. (3) and (4)).
Methane 2.0 0.152 0.053 0.150 12.276 15.072
HF xF Nitrogen 0.798 0.992 0.0
HQR1 ¼ ð13Þ Carbon dioxide 0.005 1.750 0.0
QF xF þ 1xF
Mixture 0.304 1.000 0.353 0.441 2.745 15.072
HO ð1xF Þ=4:773
HQR2 ¼ ð14Þ The underlined numbers are known inputs. The numbers that have been italicized
QFA xF þ 1xF are intermediate results in the energy domain of the quenching/heating poten-
tials. The resulting mixture properties are in bold.
here, Eqs. (13) and (14) are different because the condition at the
LFL is an oxygen surplus, whereas that at the UFL is a fuel surplus.
Thus, we can have two HQR curves and two cross points with 1xL xD
HQR¼1. One cross point is xL, and the other is xU, as shown in xU QF þ ð1xU xD Þ þ xD QD ¼ HO ð16Þ
4:773
Fig. 2. The flammability limits are those conditions that are met
when quenching and heating are balanced (HQR¼ 1). Similarly, we can have two HQR curves with variable diluent
concentrations. The cross-point of these two curves with HQR ¼1
4.2. Role of additional diluent(s) are the modified flammability limits of the new diluent. It
illustrates the change in the flammability diagram caused by
The mixture of multiple diluents can be treated as a new adding/removing diluents. To visualize the contribution of mod-
diluent with a cumulative quenching potential (QD), which is the ifying the flammability, Eqs. (17) and (18) are used in Fig. 3, with
only parameter considered for the quenching capability. Eqs. (15) data listed in Table 2.
and (16) manipulate the added contribution of a diluent.
HF x
HQR1 ¼ ð17Þ
xL QF þxD QD þð1xL xD Þ ¼ xL HF ð15Þ QF x þQD xD þ1xxD
T. Ma / Fire Safety Journal 46 (2011) 558–567 563

HO ð1xxD Þ=4:773 Compared to LCR, this method is more consistent and theore-
HQR2 ¼ ð18Þ
xQF þ ð1xxD Þ þ xD QD tical in dealing with contributions from various terms (fuel/
oxidizer/diluent). By stressing the thermal processes governing
ignition, the physical meaning of each term is clear. The results
are close to those produced using LCR, as evidenced by the
4.3. Role of additional fuel(s) equivalence of these two methods established in the Appendix A.
The spreadsheet operations in Table 3 can be presented in the
The above analysis of flammability can be extended to mix- form of HQR (Heating/Quenching Ratio) curves as shown in Fig. 4.
tures of multiple fuels. The three fundamental inputs of a fuel are Without a diluent, Eqs. (13) and (14) are used to draw the sum of
CO, xL and xU. Together with compositional information, three the HQR curve for the mixture. Because ethyl acetate provides the
intermediate parameters (or the thermal signature of each reac- largest contribution and the other two fuels compensate each
tant, QF, HO and HF) are derived (using Eqs. (8) and (9)). It should other, the HQR curves of the mixture follow those of ethyl acetate.
be noted that HO and HF are related to each other through CO. The The physical meaning of these curves is clearly shown.
thermal terms are summed (Eq. (10)), and the lump-sum thermal
parameters are used to recalculate the flammability limits of a
mixture (Eqs. (11) and (12)). A spreadsheet used to compute the 5. More applications of this thermal theory
mixture properties is listed in Table 3. This problem is directly
taken from [17]. 5.1. Flammability diagrams

One advantage of this method is the incorporation of inerting


Table 3
Spreadsheet for the determination of the flammability of a mixture. agents (diluents) directly in computing flammability. Here, a
diluent is introduced into the system with a concentration of
Fuel CO yi xL xU QF HO yd and a quenching potential of QD. Eqs. (15) and (16) are the
thermal balance equations for this case.
Ethyl acetate 5.0000 0.6350 0.0220 0.1150 9.7881 10.8437
Ethanol 3.0000 0.2080 0.0430 0.1900 6.3580 11.8914
Thus, we have
Toluene 0.1570 0.0140 0.0680 12.9788 17.3465 11.1007 1ð1QD ÞxD
Mixture 5.2120 1.0000 0.0206 0.1144 10.2613 11.1020 xL ¼ ¼ xL,0 ½1ð1QD ÞxD  ð19Þ
1QF þCO HO
Le Chatelier’s rule 0.0207 0.1133
Relative error (%)  0.679 1.021  
QD
xU ¼ xU,0  xU,0 þ xD ð20Þ
QF 1 þ HO =4:773
where the subscript 0 is used to denote the initial flammability
limits without any diluent. These two curves produce a flamm-
ability diagram (Fig. 5). The inerting concentration is found to be
the cross-point of the xL and xU curves. An additional curve, the
stoichiometric line, is also presented. This is governed by the
reaction stoichiometry (Eq. (21)).
1xD
xst ¼ ¼ xst,0 ð1xD Þ ð21Þ
1 þ4:773CO
It should be noted that the prediction curve is based on an
isothermal process (assuming constant flame temperature at all
critical conditions), so the concentration of total inertion (some-
times called inertion (or inertization) point, for example, see [22])
is larger than experimentally determined values. Macek [16] found
that the flame temperature is raised at extinction (due to the
change in flame structure, a higher temperature is needed to
sustain the flame), so less agent is required (due to the introduction
of more inerting potentials along with the rise in temperature).
Fig. 4. Determination of the flammability for a mixture. Shebeko et al. [22] tried to approximate this inertion point more

15 xL (Ar) xL (N2)
xU (Ar) 12 xU (N2)
Stoichiometric (CO2)
Flammability (%)

Stoichiometry(CO2)
Flammability (%)

12 xL (CO2) xL (CO2)
xU (CO2) 10 xU (CO2)
Published Ar Published N2
published CO2 8 Published CO2
9
6
6 4

2
0 10 20 30 40 50 60 0 20 40 60
Inerting Concentration (%) Inerting Concentration (%)

Fig. 5. Flammability diagrams for (a) methane and (b) ethane.


564 T. Ma / Fire Safety Journal 46 (2011) 558–567

10 published CO2
published N2
Stoich. (CO2)

Flammability (%)
8 Stoich. (CO)
xL (N2)
xU (N2)
6 xL (CO2)
xU (CO2)

2
0 10 20 30 40 50
Inerting Concentration (%)

Fig. 6. Flammability diagrams for (a) propane and (b) hexane.

precisely, though with some additional complexity. Because it


0.16 He
is better to make conservative estimations (for the inerting con-
centrations), no refinement was made in this work. Further rescaled N2
refinement should remove the assumption of similar flame struc- N2
ture because the flame temperature is higher at the inertion point. rescaled CO2
As Beyler [3]summarizes the flame temperature is higher at 0.12 CO2
extinction (premixed flame) than at ignition (diffusion flame) due
xF
to a distorted flame structure.
When applying this treatment to hexane, the prediction shows
significant deviations. The inertion point is on the stoichiometric
0.08
line of CO (Eq. (22)) instead of that of CO2 (Eq. (2)). This shift is
attributed to incomplete combustion and to the preferential
diffusion of reactants [3]. Such a shift is clearly shown in the
flammability diagrams for propane and hexane in Fig. 6. The
inertion points appear to be on the stoichiometric line for the 0.04
0 0.2 0.4
reaction involving CO.
xD
CO ¼ a=2þ b=4c=2
Fig. 7. Scaling of the flammability diagram.
Cst ¼ 1 þ4:773CO ð22Þ

Macek [16] has discussed this bias in greater depth. The devia- contributions of quenching potentials.
tion at the inertion point arises due to several reasons: incomplete
reaction, decreased critical flame temperature, increased radiative xD1 QD1 þ xF1 QF þ 1xD1 xF1
HO ¼ ð23Þ
heat loss terms, etc. Caution should be taken if the composition of a 2xF1
mixture is close to the mixture’s inertion point. Usually, the process
of determining the flammability range takes place far away from xD1 QD1 þ xF QF þ 1xD1 xF1
HO ¼ 4:773 ð24Þ
this point, so there is no need to discuss this matter further. 1xD1 xF1

QD1
xD2 ¼ xD1 ð25Þ
5.2. Incomplete reaction near inertion point QD2

For most estimation work, this incompleteness of a reaction is xF1


xL2 ¼ ð26Þ
generally ignored. Oxygen calorimetry (second assumption) is 1xD1 þ xD2
affected most by the incompleteness of a reaction. We can reduce
the dependence on oxygen calorimetry by converting one flamm- ð1xD2 ÞððHO =4:76Þ1ÞxD2 QD2
ability diagram to another. xU2 ¼ ð27Þ
QF 1 þ HO =4:76
The above predicted diagrams are based on a constant flame
temperature under critical conditions, which lead to sharp iner- Eq. (23) obtains oxygen information from the LFL (Eq. (3)),
tion points in estimates. The real inertion point is low in whereas Eq. (24) obtains oxygen information from the UFL
temperature and incomplete in reaction, and radiation dominates (Eq. (4)); both sets of information are obtained from an experi-
the loss terms. To consider an incomplete reaction near the mentally determined flammability diagram. Eq. (25) rescales the
inertion point, we can project the flammability diagram from contribution of a new diluent. Eqs. (26) and (27) produce the LFL
one thermal agent to another thermal agent. Of the three inter- and UFL informations with a rescaled contribution from this
mediate parameters, we will keep HO (or HF, since they are diluents, respectively.
coupled by CO) floating (because it represents the incomplete The rescaled flammability diagram is shown in Fig. 7. The data
chemistry near the upper limits) while keeping QD and QF points from one curve (helium) are projected onto other curves
constant. Following the procedure illustrated by Eqs. (23)–(27), (CO2 and N2) by following the above conversion steps. Such a
we can rescale a flammability diagram from that of one thermal conversion process retains some features of an incomplete reac-
agent to that of another thermal agent by only rescaling the tion by rescaling the contribution of a diluent.
T. Ma / Fire Safety Journal 46 (2011) 558–567 565

6. Discussions combustion products only because the specific heat of a fuel cannot
be retrieved from any reference. The advantage of this treatment is
6.1. Comparison to Le Chatelier’s rule (LCR) the possibility of incorporating the equivalence ratio or final-state
temperature into the estimation scheme, thus allowing the method
Both methods are based on a constant CAFT. Both take the to address vitiated combustion in a ceiling layer.
variation in the heating potentials of each into consideration. X n
ci DHC,i =100
However, LCR does not differentiate between the variations in the R Tf Z1 ð28Þ
quenching potentials, so it is poor in dealing with new diluents. i ¼ 1 T0 np cp dT

From this perspective, this method is an extension of LCR because


Table 4 redoes the estimation in Table 3. The current method is
it incorporates the quenching potentials into the estimation.
comparable to Beyler’s method for LFL, though Beyler’s method
As shown in the Appendix A, this method is only equivalent to
fails when excess fuel appears in combustion products (for
LCR for flammable components. For additional diluents in the
conditions near UFL).
mixture, the modified flammability diagram (Fig. 1b) implicitly
Hansel et al. [12] tried to use a similar method for reactants;
assumes that the diluents and the fuel share the same quenching
however, an external numerical tool is needed to compute the
potential, which is a simplification. The current method can consider
CAFT of a mixture, which can no longer be performed by hand.
the variations in the thermal properties of a diluent, so it is better
The current method focuses on the energy balance from a
than the modified LCR method (shown in Fig. 1). The thermal theory
reacting fuel/oxidizer/diluent perspective. This method is basi-
behind this method establishes the flammability diagrams auto-
cally a conversion scheme. It converts the flammability data into
matically and implicitly, removing the need for an experimental
the domain of heating/quenching potentials. After some addi-
diagram. This is the major advantage of this method over LCR.
tional manipulations in that domain, the method then converts
Compared to the major assumptions in deriving LCR [17], the
the flammability data back into the domain of flammability limits.
first assumption of constant heat capacities for combustion
The accuracy of the estimation results depends mainly on the
products is dropped. Using the enthalpy of air to scale all energy
inputs. As the flammability diagram (Fig. 7) shows, the measure-
terms, the dependence of heat capacities on temperature is no
ment error from a pure substance will be passed on to its mixture.
longer important. This simplifies the derivation (see Appendix A)
The assumptions will also introduce some errors; generally,
and estimation process. Only the dimensionless thermal quanti-
however, these errors are small compared to the uncertainty in
ties are compared, avoiding the non-linear specific heat terms at
the measurement of the flammability limits [6,7]. In addition, this
various temperature ranges.
method has the potential to deal with combustion in a hot
Another advantage of the theory is the use of the oxygen
environment, as Beyler’s method does. More research is needed
calorimetry at the UFL. This increases the error if the reaction at
to predict the flammability under non-standard conditions.
the UFL is incomplete. However, for most common fuels, this error
is only significant close to a fuel’s inertion point. For dilution-
6.3. Limitation of this model
estimation points close to a fuel’s measuring points (without
additional dilution), the estimation error will be propagated and
Because the estimated results are highly dependent on the
rescaled from the initial measurement error. The conversion
accuracy of the inputs, any errors introduced by initial measure-
process introduces some new errors, except for those due to the
ments will be passed on to the estimated results. For some
initial assumptions. In essence, the thermal theory for flammability
flammability data, the published LFL is small in value. Caution
diagrams (in Section 5.1) calls for the replacement of the flamm-
should be taken in using these data for prediction.
ability diagram in Fig. 1(b), thus avoiding the look-up error
The flame structure at the UFL is assumed to be similar to that
associated with LCR. Keeping this in mind, we can find more
at the LFL. This is not generally accepted because incomplete
applications of this method.
combustion is involved at the UFL. However, this methodology
requires both conditions to derive the thermal properties of a fuel.
6.2. Comparison to Beyler’s method The small sacrifice in precision will lead to a wider application
range. In addition, all flammability data are still dependent on the
Another comparable method reported in the literature is Beyler’s flame temperature or flame structure. Hydrogen changes the
energy-balance theory [3], which is similar to this work in that an flame thickness through its diffusivity, thus decreasing the flame
energy balance is performed at the CAFT. Though the integration is temperature. Halon 1301 changes the flame temperature through
used in Eq. (28), the averaged specific heat of each product is used to its scavenging effect on reacting ions. Both will be difficult to
avoid the numerical integration process. The original idea was to predict using this theory. However, in reality, LCR is also used for
determine the flammable mixture of unknown origin with a thermal mixtures involving hydrogen without verifying the problem of
balance to determine if a reaction would run away (combust). lowered flame temperature [3].
Therefore, it can be used to test the flammability of a mixture. This limitation can also be applied to other estimation methods,
However, Beyler’s method determines the heat balance of including LCR and Beyler’s method, because all estimation methods

Table 4
Comparison of four estimation methods for predicting flammability limits of a mixture.

Fuel Yield LFL UFL QF HO DHC CO2 H2O N2

Ethyl acetate 0.6350 0.0220 0.1150 9.8700 10.8649 2060.8 4.0 4.0 18.9
Ethanol 0.2080 0.0330 0.1900 6.3938 11.8989 1235.5 3.0 3.0 11.3
Toluene 0.1560 0.0120 0.0710 17.3666 11.0778 3733.9 9.0 4.0 34.0
Mixture 0.0206 0.1145 10.3066 11.1023 2148.1 4.6 3.8 19.6

Le Chatelier’s rule Current method Beyler’s method


LFL 0.0208 0.0206 0.0205
UFL 0.1135 0.1145 n/a
566 T. Ma / Fire Safety Journal 46 (2011) 558–567

are based on the assumption of a constant CAFT. The method Rearranging Eq. (A3), we have
discussed in this work is better than other methods because it 1
considers the contributions from the quenching potentials of fuel/ ¼ 1 þHF QF ðA4Þ
xL
oxygen/diluents, which are not discussed in other works.
Performing the summation on both sides of Eq. (A4) and
inverting both sides, Eq. (A1) is proved.
7. Conclusions Similarly, we must prove the following equation (Eq. (A5))
with respect to Eq. (A2).
In essence, this work attempts to add the variation of each 1 QF
species into the estimation scheme for predicting flammability ¼ 1 ðA5Þ
xU 1ðHO =4:773Þ
limits. LCR treats all fuels in the same way, assuming that all fuels
contribute equally to flammability. The modified flammability However, substituting Eqs. (A3) and (A6) into (A5), we find
diagram (Fig. 1) also treats a diluent as something sharing the that Eq. (A5) is always true.
same quenching potential as a fuel, so it can be part of a new xU xL
HO ¼ ðA6Þ
pseudo-fuel. This method tries to differentiate the dual contribu- CO xU xL ððð1xU ÞxL Þ=4:773Þ
tions of each species in a reaction, incorporating the variations of In conclusion, Le Chatelier’s mixing rule is not only empirical but
fuel and diluents into a simple scheme comparable to LCR. can also be proven using the thermal theory introduced in this work.
Here, a binary perspective is constructed to isolate the contribu- The derivation process drops the assumption of constant specific
tions of the heating and quenching potentials of each species. Thus, heat [17] and creates some physically meaningful intermediate
the contribution of fuel/oxygen/diluents can be managed under the terms. These terms provide flexibility in applications, especially
same systematic framework. With isolated contributions from fuel/ when dealing with diluents. LCR must use an experimental flamm-
oxygen/diluents, the energy source and the energy sink terms are ability diagram to incorporate the contribution of a diluent. This
simply additive and easily manipulated by a thermal balance. Thus, flammability diagram is implicitly used in this work. Thus, this
the method adds flexibility to LCR. The derivation in the Appendix A methodology can be more general and useful than LCR, though they
shows that the contributions from fuel/oxygen/diluents are equiva- are equivalent when no additional diluent is involved.
lent in some cases (when only fuels are involved in a mixture). To
avoid the non-linearity of specific heat in combustion, all energy
References
terms are non-dimensionalized by the properties of air. This is only
possible if the critical flame temperature is constant for all fuels.
[1] Anon, 2010, NIST Chemistry WebBook, /http://webbook.nist.gov/chemistry/S.
This scaling avoids the non-linearity in Beyler’s method and the [2] V. Babrauskas, Ignition Handbook, Fire Science Publishers, 2003, pp. 110.
numerical integration problem in Hansel’s method. Another con- [3] C. Beyler, Flammability limits of premixed and diffusion flames, SFPE Hand-
tribution of this work is the oxygen calorimetry at the flammability book, 1st edition, 1988, pp. 1–297.
[4] C. Beyler, A short history of the prediction of flame extinction based upon
limits. It is not an accurate assumption (complete reaction at both flame temperature, Fire Mater. 29 (2005) 425–427.
LFL and UFL), but it works under the current estimation scheme. [5] R.B. Bird, W.E. Steward, E.N. Lightfoot, Transport Phenomena, 2nd edition,
Attempts are made to explore the flammability diagrams from a 2006.
[6] L.G. Britton, Two-hundred years of flammable limits, Prog. Safe. Prog. 21 (1)
thermal perspective. The physical meaning of the flammability (2002) 1–11.
limits are better presented in the diagrams. The breakdown of the [7] L.G. Britton, Using heat of oxidation to evaluate flammability hazards, Process
assumption at the CAFT is the major reason for the deviation near Safe. Prog. 21 (1) (2002) 31–54.
[9] L. Catoire, V. Naudet, Estimation of temperature-dependent lower flamm-
the inertion point.
ability limit of pure organic compounds in air at atmospheric pressure, Prog.
This method can predict the flammability range, stressing the Safe. Prog. 24 (2) (2005) 130–137.
contributions of all participants in a reaction and providing a full [10] D. Drysdale, Introduction to Fire Dynamics, 2nd edition, Wiley, 1996.
coverage for all possibilities. It is simple to use, presents clear [11] W.M. Haessler, Fire: Fundamentals and Control, Marcel Dekker, Inc., 1972.
[12] J.G. Hansel, J.W. Mitchell, H.C. Klotz, Predicting and controlling flammability of
physics and is suitable for spreadsheet calculations; therefore, it multiple fuel and multiple inert mixtures, Plant/Oper. Prog. 11 (4) (1992)
should find additional applications in the hazard analysis of 213–217.
mixtures of flammable gases. [13] B. Huggett, Estimation of rate of heat release by means of oxygen consump-
tion measurements, Fire Mater. 4 (2) (1980).
[14] J.M. Kuchta, Investigation of Fire and Explosion Accidents in the chemical,
mining and fuel-related industries—a manual, Bur. Mines Bull. 680 (1985).
Appendix A. Application of the thermal theory in deriving Le [15] H. Le Chatelier, Estimation of firedamp by flammability limits, Ann. Mines 19
(ser. 8) (1891) 388–395.
Chatelier’s mixing rule
[16] A. Macek, Flammability limits: a Re-examination, Combust. Sci. Technol. 21
(1979) 43–52.
Le Chatelier’s mixing rule can be derived using this theory; [17] C.V. Mashuga, D.A. Crowl, Derivation of Le Chatelier’s Mixing Rule for
Flammable limits, Process Safe. Prog. 19 (2) (2000) 112–117.
therefore, this method is equivalent to Le Chatelier’s mixing rule
[18] G.A. Melhem, A detailed method for estimating mixture flammability limits
[15] (Eqs. (A1) and (A2)). using chemical equilibrium, Prog. Safe. Prog. 16 (4) (1997) 203–217.
[19] V. Schroeder, M. Molnarne, Flammability of gas mixtures, part 1: fire
1 1 potential, J. Hazardous Mater., A 121 (2005) 37–44.
xL,m ¼ ¼ PN ðA1Þ
1þ HF,m QF,m i yi =xL,i [20] V. Schroeder, M. Molnarne, Flammability of gas mixtures, part 2: influence of
inert gases, J. Hazardous Mater., A 121 (2005) 45–49.
[21] J.A. Senecal, Flame extinguishing in the cup-burner by inert gases, Fire Safe.
1HO,m =4:773 1 J. 40 (2005) 579–591.
xU,m ¼ ¼ PN ðA2Þ
1HO,m =4:773QF,m i yi =xU,i [22] Yu.N. Shebeko, W. Fan, I.A. Bolodian, V.Yu. Navzenya, An analytical evalua-
tion of flammability limits of gaseous mixtures of combustible–oxidizer–
From the fundamental equations of QF and HO in the paper, we diluent, Fire Safe. J. 37 (2002) 549–568.
have [25] I. Wierzba, S.O. Shrestha, G.A. Karim, An approach for predicting the
flammability limits of fuel/diluent mixtures in air, J. Inst. Energy 69 (1996)
1 CO ðxU xL Þ 122–130.
QF ¼ 1 þ [26] F.A. Williams, A review of flame extinction, Fire Safe. J. 3 (1981) 163–175.
xL CO xU xL ððð1xU ÞxL Þ=4:773Þ [27] M.G. Zabetakis, Flammability Characteristics of Combustible Gases and
1 1 Vapors, Bulletin 627, Bureau of Mines, Washington, 1965.
¼ 1 þ Co HO ¼ 1 þHF ða3Þ [28] H.F. Coward, G.W.Jones, Limits of Flammability of Gases and Vapors, Bulletin
xL xL
No. 503, US Bureau of Mines, 1952.
T. Ma / Fire Safety Journal 46 (2011) 558–567 567

[29] S. Kondo, K. Takizaw, A. Takahashi, K. Tokuhashi, Extended Le Chatelier’s [31] M. Burgess, R. Wheeler, The lower limit of inflammation of mixtures of
formula for carbon dioxide dilution effect on flammability limits, Journal of paraffin hydrocarbons with air, Journal of the Chemical Society (London) 99
Hagardous Materials 138 (1) (2006) 1–8. (1911) 2013–2030.
[30] R.S. Sheinson, J.E. Penner-Hahn, D. Indritz, The physical and chemical action [32] W.M. Haessler, Fire Fundamentals and Control, Marcel Dekker, 1989.
of fire suppressants, Fire Safety Journal 15 (1989) 437–450.

Das könnte Ihnen auch gefallen