Fluid Phase Equilibria 312 (2011) 101–105

Contents lists available at SciVerse ScienceDirect

Fluid Phase Equilibria

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

Measurement and calculation of flash point of binary aqueous–organic and

organic–organic solutions
Tooraj Khalili, A. Zarringhalam Moghaddam ∗
Department of Chemical Engineering, Tarbiat Modares University, P.O. Box 14115/138, Tehran, Iran

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

Article history: Flash point for binary systems, pentan-1-ol + water, cyclohexanone + water, pentan-1-ol + cyclohexanone,
Received 28 February 2011 pentan-1-ol + 1,2,3,4-tetrahydronaphthalene, were measured using the tag close cup apparatus by equi-
Received in revised form 5 September 2011 librium test method. First two systems were partially miscible and the other two were completely
Accepted 5 September 2011
miscible.
Available online 10 September 2011
In this study attempt has been made to extend Liaw et al. algorithm for calculation of the flash point
of both miscible and partial miscible mixtures. The proposed algorithm was evaluated with experimen-
Keywords:
tal data for both miscible and partial miscible mixtures. The results show good compatibility for both
Flash point
Binary interaction parameter
aqueous–organic and organic–organic solutions flash point calculation.
Vapor–liquid equilibrium © 2011 Published by Elsevier B.V.
Activity model

1. Introduction assumption, White [4] simplified the Affens model by ignoring the
temperature dependence of LFL, Gmehling and Rasmussen [5] pro-
Flash point is the major property of the flammable liquids to posed the model according to the non-ideal liquid phase solution
analyze the safety in chemical processes. Flash point is the lowest and LFL dependence on temperature assumption and Liaw et al. [6]
temperature at which the concentration of vapor of the substance simplified the Gmehling and Rasmussen [5] model by ignoring the
in the air reaches the LFL (lower flammability limit). Flash point can LFL temperature dependence and developed their model for total
be determined in two ways; measuring by a test analyzer apparatus miscible flammable mixtures, afterwards Liaw et al. [7] extended
and calculating by predictive models. Close cup and open cup appa- their model for partially miscible and aqueous–organic mixtures.
ratus are two major test analyzers for experimental determination All the models which are derived from Le Chatelier rule often
of the flash point. require try and error calculation to determine the flash point
Since the flash point data for all the mixtures is not available of the mixture. In the present study by utilizing Liaw et al. [7]
and its accurate prediction is desirable for safety analysis. There- and Rachford-Rice [8] algorithm an extended method of calcu-
fore, several authors attempted to develop the predictive model lation is proposed to predict the flash point of the mixtures
over their researches. Wickey et al. [1] developed a model to cal- which could be used for both miscible and partial miscible
culate the closed cup flash point of the petroleum blends. This was mixtures. The ability of the proposed algorithm was evaluated
based on the use of flash point indices for the stocks in the blends. by using the experimental flash points of two partial misci-
This method does not work well for mixtures containing compo- ble mixtures of; pentan-1-ol + water and cyclohexanone + water
nents exhibiting non ideal solution behavior. Catoire [2] developed and two miscible mixtures of; pentan-1-ol + cyclohexanone and
the model assuming the flash point as a function of normal boiling pentan-1-ol + 1,2,3,4-tetrahydronaphthalene and were compared
point, number of carbon atoms in the vapor phase and standard with Wickey et al. [1] and Catoire [2] models.
enthalpy of vaporization. This model is applicable for miscible com-
bustible solvent mixtures. Based on the Le Chatelier rule several
authors have developed models to calculate the flash point of the 2. Materials and methods
mixtures. Affens and McLaren [3] proposed the model according
to the ideal liquid phase solution and LFL temperature dependence 2.1. Materials

All components were purchased from Merck Company

∗ Corresponding author. (Germany). Pentan-1-ol, CAS Number 71-4-0 was synthesis grade
E-mail addresses: [email protected] (T. Khalili), [email protected] with a minimum purity of 98%, cyclohexanone, CAS number 108-
(A. Zarringhalam Moghaddam). 94-1 was in extra pure grade with a minimum purity of 99% and

0378-3812/$ – see front matter © 2011 Published by Elsevier B.V.

doi:10.1016/j.fluid.2011.09.003
102 T. Khalili, A. Zarringhalam Moghaddam / Fluid Phase Equilibria 312 (2011) 101–105

Table 1 Tfp is the flash point of the mixture and Imix is the index of the
Activity coefficient calculation by UNIFAC [8] model.
mixture that can be calculated by Eq. (2):
ln i = ln iC + ln iR 
Which Imix = ˙i Ii (2)
˚i  ˚

m
i
z
ln ic = ln xi
+ q ln ˚i
2 i
+ li − i
xi
xj lj (Combinatorial part)
i
i is the volume fraction of the component i in the mixture and Ii
j=1
is the flash point index that is calculated by Eq. (3):
10 (i)
lj = 2
(rj − qj ) − (rj − 1) and qi = v k
Qk
2414
k log10 (I) = −6.1188 + (3)

m
 
m TF /k + 503.71
(i)
˚i = xi ri / rj xj and ri = v Rk and i = qi xi / qj xj
k TF is the flash point of the pure component or the blends.
j=1 k j=1
And
 3.1.2. Catoire model [2]
ln iR = v(i)
k
(i)
(ln k − ln k ) (Residual part) Catoire proposed the empirical model to calculate the flash point
k  of the mixtures as Eq. (4):
m = Qm Xm / Qn Xn and  mn = exp(− amn /RT)
Tfp
 T 0.79686
eb ◦
n = 1.477 × 0.16845
× Hvap × n−0.05948 (4)
K K
Teb is the normal boiling point temperature of the mixture in the
◦
Kelvin unit, Hvap is the standard enthalpy of vaporization at
1,2,3,4-tetrahydronaphthalene, CAS number 119-64-2 was in syn- 298.15 K in kJ/mol unit and n represents the number of carbon
thesis grade with a minimum purity of 96%. All these chemicals atoms in the fictitious compound representative of the fuel mixture.
were used without further purification. Water was purified by the Teb is the temperature in which the vapor pressure of the mixture
water purification apparatus (Barnsteed, model D4742, USA). equal to 1 atm vapor pressure can be calculated by Eq. (5):

Patm = xi i Pisat (5)
i
2.2. Methods
Patm is the standard pressure (101.3 kPa), xi is the mole fraction
of the substance i in liquid phase.  i is the activity coefficient at
A tag close cup flash point analyzer (SDM 930, Italy) was used
the temperature of Teb which can be can be calculated from the
to measure the flash point of the mixtures. The mole fraction of
equations such as Non-Random Two Liquid (NRTL) [10] and the
each component is determined by measuring the mass using the
Universal Quasi Chemical (UNIQUAC) [10]. These models require
Sartorius digital balance (TE-214S: sensitivity 0.0001 g, maximum
two or more binary interaction parameters that are extracted from
load 210 g). The sample was prepared and transferred to the cup of
experimental equilibrium data. In the cases that no such data are
the apparatus at least in 10 ◦ C below the expected flash point. The
available, UNIFAC [11] model can be used which the calculation
apparatus was set to heat the sample in 0.5 ◦ C not less than 1.5 min.
method is reported in Table 1. The required group–group interac-
The ignition source was used from 5 ◦ C below the expected flash
tion parameters in UNIFAC model are from parameters reported by
point temperature and was repeated in 0.5 ◦ C interval by inserting
Poling et al. [11]. Pisat is the vapor pressure of the substance i at the
and removing in the cup within 2.5 ± 0.5 s. This procedure is that
Teb which can be calculated by the correlations given in Table 2.
followed in standards including ASTM D3971 [9]. ◦
Hvap is calculated by Clausius–Clapeyron law. Vapor pressure
of mixture in three temperatures in vicinity of 298.15 K (20 ◦ C, 30 ◦ C
and 35 ◦ C) must be considered which can be calculated by Eq. (6):
3. Theory 
P= xi i Pisat (6)
3.1. Calculation models i

All the parameters in Eq. (6) are as the same in Eq. (5). The slop of
3.1.1. Wickey et al. model [1] ◦
ln(P) vs. 1/T with P in kPa and T in Kelvin is equal to (−Hvap /R),
Wickey et al. [1] have proposed a model to calculate the close ◦
that R is 8.314 in J/mol K and −Hvap is the standard enthalpy of
cup flash point of petroleum blends and mixtures. In their model
vaporization in J/mol.
flash point is calculated by Eq. (1):
n can be calculated by Eq. (7):

Tfp 2414 n= yi ni (7)
= + 42.59 (1)
k 6.1188 + log10 (Imix ) i

Table 2
Vapor pressure equations.

Component Vapor pressure equation Merck [16]

VPa @ 20 ◦ C (hPa) TNBP b (◦ C)

Pentan-1-ol [7] Log (Pisat /Pa) = (7.39824 − 1435.570/[−93.202 + (T/K)]) · 131.58 3 138
−4 −7 ·(T/K)2 }
1,2,3,4-Tetrahydronaphthalene [14] Log (Pisat /Pa) = ([1 − 480.364/(T/K)] · 10{0.85916−5.75417·10 ·(T/K)+4.41971·10 ) · 131.58 0.24 207
Cyclohexanone [15] Log (Pisat /Pa) = (7.4705 − 1832.2/[244 + (T/C)]) · 131.58 4.5 155
Water [15] Log (Pisat /Pa) = (8.07131 − 1730.63/[233.426 + (T/C)]) · 131.58 23 100
a
Vapor pressure.
b
Normal boiling point temperature.
Tfp
Pure
, Pure component flash point.
 T. Khalili, A. Zarringhalam Moghaddam / Fluid Phase Equilibria 312 (2011) 101–105 103

Compute all xiα

Start
zi
xiα =
1 + R.(C i − 1)
Set model for
β
γ iα and γ i Compute new C i
γ iα
Ci =
Set all z i and Compute all xiβ γ iβ
e1 , e 2 , h xiβ = xiα .Ci

Compute all Loop

1
Modifey all x iα , β α β
Initial guessT γ i and γ i
α ,β x ia, β
xi =
∑ x iα , β
ComputeNewR
Initialguesses F
zi .(Ci − 1) R (k +1) = R k −
forallC i and R F =∑ ∂F
1 + R.(Ci − 1) ( )
∂R

No
∂F z .(C − 1)2
F < e1 ( ) =− ∑ i i
∂R [1+ R.(Ci −1)]2

Yes

No
Mutual miscible
0 < R <1
Set x i = z i
Yes

Partial miscible
Set xi = xiα
Set newT Loop
T (k +1) = T (k) − h 2
Compute OB
(eq 10)
No

Yes No
Set new T
OB < 1 OB − 1 < e 2
T (k + 1) = T (k) + h
Yes

Report flash point as

T fp
=T

Fig. 1. Algorithm for calculating the flash point of mixtures.

ni is the number of carbon atoms in the component i and yi is the Because of the low pressure (1 atm), the vapor phase of the mix-
mole fraction of component i in the vapor at the flash point tem- ture assumes to be ideal gas thus the yi can be calculated from Eq.
perature. Because the flash point temperature of the mixture is (9):
unknown, yi is defined here as the mole fraction of component i
in the vapor at the Teb . xi i Pisat
yi = (9)
Patm
3.1.3. Extended algorithm
According to the Le Chatelier rule [12], the flash point of a xi is the mole fraction of the flammable substance,  i is the activity
flammable mixture is the temperature which satisfies Eq. (8): coefficient and Pisat is the vapor pressure of the flammable sub-
 y stance at the temperature of the mixture flash point.
i
1= (8) LFLi of each component can be calculated by Sly [13] relationship
LFLi
i as is shown in Eq. (10):

yi is the vapor mole fraction of flammable component i, in equilib- sat

Pi,FP
rium with liquid phase, and LFLi is the lower flammability limit of LFLi = (10)
the pure component i. Patm
104 T. Khalili, A. Zarringhalam Moghaddam / Fluid Phase Equilibria 312 (2011) 101–105

Table 3
Pure component flash point.

Experimental (◦ C) Supplier (◦ C)

Pentan-1-ol 49 49 [16]
1,2,3,4-Tetrahydronaphthalene 72.5 71–77 [16]
Cyclohexanone 43 43 [16]

sat is the vapor pressure of component i at its flash point and the
Pi,FP
Patm is the ambient pressure which is assumed 1 atm. The vapor
pressure of the studied components can be calculated by the cor-
relations given in Table 2.
Substituting Eqs. (9) and (10) in (8) leads to Eq. (11) from which
the flash point of mixtures can be calculated:
 xi i P sat
i
sat
=1 (11)
Pi,fp
i

In Eq. (11) index i indicate the properties of the, ith flammable

component.
As is shown in Eqs. (8) and (9), the flash point calculation is
based on the liquid–vapor equilibrium state of the gas phase and
Fig. 2. Comparison of the calculated flash point with the experimental data for:
the liquid phase which has the same surface interface. Liaw et al. pentan-1-ol (1) + water (2) system.
[7] has shown in their work that for partial miscible mixtures the
liquid–liquid equilibrium state must be considered and introduced
an algorithm for calculating the flash point of such mixtures. In In Table 4 mean standard deviation of experimental data for each
present work attempt has been made to facilitate and to speed up mixture and absolute average deviation between the experimental
the utilization of this algorithm by introducing the relevant equa- flash point data and calculated are reported.
tions at each step in the algorithm. In Figs. 2 and 3 experimental flash point data of pentan-1-
In Fig. 1 the algorithm is presented which shows LLE and flash ol + water and cyclohexanone + water is compared with calculated
point can be calculated in loop 1 and loop 2 respectively. In the values. The calculation was carried out using extended algorithm.
algorithm subscript ˛ indicate aqueous zone and ˇ indicate organic Results show that for these two partial miscible mixtures in the
zone, e1 is the error for LLE calculation and the e2 is the error to cal- two liquid phase region the flash points remain approximately con-
culate the flash point, h is the temperature increasing or decreasing
step, zi indicate the overall mole fraction of the components, R is
the molar ratio and indicate the moles of components in ˇ phase
correspond to the total moles in the mixture.

4. Results

The flash point for pentan-1-ol + water, cyclohexanone + water,

pentan-1-ol + cyclohexanone and pentan-1-ol + 1,2,3,4-
tetrahydronaphthalene mixtures was measured by tag close
cup apparatus by equilibrium test method. Binary mixture of
pentan-1-ol + water and cyclohexanone + water are partial misci-
ble and in contrast binary mixtures of pentan-1-ol + cyclohexanone
and pentan-1-ol + 1,2,3,4-tetrahydronaphthalene are total misci-
ble. An algorithm was proposed to calculate the flash point of the
mixtures. The Wickey model [1] and Catoire models [2] which are
not applicable for partial miscible mixtures however could be used
for total miscible mixtures. The activity coefficient in liquid phase
is calculated using UNIFAC model. In Table 3 the flash point of
pure flammable substances from experimental measurement and
supplied by the supplier is reported. Comparing these data shows
that experimental data are in close agreement with those reported
by the supplier. In this study for the calculations which the pure Fig. 3. Comparison of the calculated flash point with the experimental data for:
component flash point is required, the experimental data are used. cyclohexanone (1) + water (2) system.

Table 4
Mean standard deviation of experimental data and absolute average deviation of calculated flash point from experimental data.

Mixture Mean standard deviation (◦ C) Absolute average deviation (◦ C)

Algorithm Wickey Catoire

Pentan-1-ol (1) + water (2) 0.52 0.26 NA NA

Cyclohexanone (1) + water (2) 0.56 0.09 NA NA
Pentan-1-ol (1) + cyclohexanone (2) 0.47 0.37 1.74 4.21
Pentan-1-ol (1) + 1,2,3,4-tetrahydronaphthalene (2) 0.46 0.83 6.53 2.66
 T. Khalili, A. Zarringhalam Moghaddam / Fluid Phase Equilibria 312 (2011) 101–105 105

In Figs. 4 and 5 experimental flash point data of

pentan-1-ol + cyclohexanone and pentan-1-ol + 1,2,3,4-
tetrahydronaphthalene mixtures compared with values calculated
from extended algorithm, Wickey model [1] and Catoire model [2]
is shown. For these two miscible mixtures the extended algorithm
results show better compatibility with the experimental data in
compare with Wickey model [1] and Catoire model [2].

5. Conclusion

The experimental flash point of binary mixtures of pentan-1-

ol + cyclohexanone, pentan-1-ol + 1,2,3,4-tetrahydronaphthalene,
pentan-1-ol + water and cyclohexanone + water systems were mea-
sured.
In present work attempt has been made to facilitate and expe-
dite the utilization of Liaw’s algorithm by introducing the relevant
equations at each step in the algorithm. The results showed that
the calculation of the flash point by the algorithm have a good
agreement with those measured experimentally.

References
Fig. 4. Comparison of the calculated flash point with the experimental data for:
[1] R.O. Wickey, D.H. Chittenden, Hydrocarb. Process. 42 (6) (1963) 157–158.
pentan-1-ol (1) + cyclohexanone (2) system.
[2] L. Catoire, S. Paulmier, J. Phys. Chem. Ref. Data 35 (2006) 9–14.
[3] W.A. Affens, G.W. McLaren, J. Chem. Eng. Data 17 (1972) 482–488.
[4] D. White, C.L. Beyler, C. Fulper, J. Leonard, Fire Saf. J. 28 (1997) 1–31.
[5] J. Gmehling, P. Rasmussen, Ind. Eng. Chem. Fundam. 2 (1982) 186–188.
[6] H.J. Liaw, Y.H. Lee, C.L. Tang, H.H. Hsu, J.H. Liu, J. Loss Prev. Process Ind. 15 (2002)
429–438.
[7] H.J. Liaw, C.T. Chena, V. Gerbaud, Chem. Eng. Sci. 63 (2008) 4543–4554.
[8] H.H. Rachford Jr., J.D. Rice, J. Petrol. Techno 4 (10) (1952) 1–19.
[9] ASTM D 3941, Standard Test Method for Flash Point by the Equilibrium Method
with a Closed-Cup Apparatus, American Society for Testing and Materials, West
Conshohocken, PA, 2001.
[10] J.M. Prausnitz, R.N. Lichtenthaler, E.G. Azevedo, Molecular Thermodynamics of
Fluid-Phase Equilibria, third ed., Prentice Hall PTR, US, 1999.
[11] B.E. Poling, J.M. Prausnitz, J.P. O’Connell, The Properties of Gases and Liquids,
fifth ed., McGRAW-HILL, US, 2001.
[12] H. Le Chatelier, Ann. Mines 19 (8) (1891) 388–395.
[13] N. McManus, Safety and Health in Confined Spaces, CRC Press, Vancouver,
Canada, 1998.
[14] D. Mackay, W.Y. Shiu, Ma. Kuo-Ching, S.C. Lee, Handbook of Physical–Chemical
Properties and Environmental Fate for Organic Chemicals, second ed., CRC
Press, Boca Raton, 2006.
[15] I.M. Smallwood, Handbook of Organic Solvent Properties, John Wiley, Great
Britain, 1996.
[16] Merck, The Merck Index, twelfth ed., Merck & Co., NJ, 1996.

Fig. 5. Comparison of the calculated flash point with the experimental data for:
pentan-1-ol (1) + 1,2,3,4-tetrahydronaphthalene (2) system.

stant however decreases by increasing the mole fraction of the

flammable component.