Академический Документы
Профессиональный Документы
Культура Документы
a r t i c l e i n f o a b s t r a c t
Article history: The earlier reported simplified model for multi-component droplet heating and evaporation is gener-
Received 22 September 2010 alised to take into account the coupling between droplets and the ambient gas. The effects of interaction
Received in revised form between droplets are also considered. The size of the gas volume, where the interaction between
21 February 2011
droplets and gas needs to be taken into account, is estimated based on the characteristic thermal and
Accepted 23 February 2011
mass diffusion scales. The model is applied to the analysis of the experimentally observed heating and
Available online 1 April 2011
evaporation of monodispersed n-decane/3-pentanone mixture droplets at atmospheric pressure. It is
pointed out that the effect of coupling leads to noticeably better agreement between the predictions of
Keywords:
Droplets
the model and the experimentally observed average droplet temperatures. In most cases, the observed
Multi-component fuel droplet temperatures lie between the average and central temperatures, predicted by the coupled
Heating solution. For the cases reported in this study, the observed time evolution of droplet radii cannot be used
Evaporation for the validation of the model. It is pointed out that the number of terms in the series in the expressions
Diffusion equation for droplet temperature and species mass fraction can be reduced to just three, with possible errors less
Heat conduction than about 0.5%. In this case, the model can be recommended for the implementation into computational
fluid dynamics (CFD) codes and used for various engineering applications, including those in internal
combustion engines.
Crown Copyright 2011 Published by Elsevier Masson SAS. All rights reserved.
1. Introduction but ignored the effects of droplets on gas. Furthermore, the issue of
optimisation of the code was not addressed in [9]. The choice of the
The problem of modelling multi-component droplet heating number of terms in the analytical solution was based exclusively on
and evaporation is a long-standing one, and has been widely dis- the numerical values of the ignored terms, without taking into
cussed in the literature (e.g. [1e8]). A new simplified model for account the computational cost. These two issues will be addressed
bi-component droplet heating and evaporation was suggested in in this paper. Also, the model suggested in [9] will be generalised to
[9]. This model is based on the analytical solutions for heat arbitrary number of species and validated against experimental
conduction and species diffusion equations inside droplets, which data different from those used in [9]. The new analytical solution
are incorporated into a numerical code. The results of calculations for species was given in [9], without showing the details of the
were validated against experimental data obtained at the Univer- derivation. Since these details are not trivial, they are reproduced in
sity of Nancy (France). this paper.
Although the model, suggested in [9], was recommended for As in [9], we will assume that the surface of the droplet remains
incorporation into computational fluid dynamics (CFD) codes, there spherical (see [10e12] for modelling the time evolution of the
are still a number of issues which need to be addressed. The original droplet surface using the Volume of Fluid methodology), and its
model of [9], took into account the effect of ambient gas on droplets radius remains constant during the time step (this assumption was
relaxed in the most recent papers [13,14]).
Basic equations and approximations are briefly summarised in
* Corresponding author. Tel.: þ44 1273 642 677; fax: þ44 1273 642 309. Section 2. Our approach to the coupled solution and the numerical
E-mail address: s.sazhin@brighton.ac.uk (S.S. Sazhin). algorithm are described in Sections 3 and 4. Section 5 is focused on
1290-0729/$ e see front matter Crown Copyright 2011 Published by Elsevier Masson SAS. All rights reserved.
doi:10.1016/j.ijthermalsci.2011.02.020
S.S. Sazhin et al. / International Journal of Thermal Sciences 50 (2011) 1164e1180 1165
the experimental setup and input parameters. The results of vT
calculation are compared with experimental data in Section 6. In h Tg Ts ¼ kl ; (2)
vR R¼Rd 0
Section 7, the accuracy of the model predictions versus the CPU
efficiency of the code are investigated. The main results of the and the initial condition Tðt ¼ 0Þ ¼ Td0 ðRÞ, where Ts ¼ Ts ðtÞ is the
paper are summarised in Section 8. droplet’s surface temperature, Tg ¼ Tg ðtÞ is the ambient gas
temperature, subscripte0 indicates that the derivative is taken
inside the droplet. The analytical solution to this equation is given
2. Basic equations and approximations and discussed in [9]. This solution is applied at individual time-
steps. To take into account the effect of droplet evaporation, gas
Most basic equations and approximations used in our analysis temperature should be replaced by the so-called effective
are essentially the same as used in [9]. They are briefly summarised temperature defined as T ¼ Tg þ r LR_ =h, where the value of R_
eff l de de
below. New equations, not used in [9], are described in more detail. is taken from the previous time step and estimated based on the
Droplets are assumed to be perfect spheres, and the process of following equation:
their heating is described by the following equation [15,16]
_d
m 2pRd Dv rtotal BM Sh
2
! R_ de ¼ ¼ ; (3)
vT v T 2 vT 4pR2d rl 4pR2d rl
¼ k þ ; (1)
vt vR2 R vR
where Dv is the binary diffusion coefficient of vapour in air, BM is the
where k ¼ kl =ðcl rl Þ is the liquid thermal diffusivity. Eq. (1) is to Spalding mass transfer number defined as:
be solved for t > 0 and 0 R Rd, where Rd is the droplet radius,
rvs rvN Yvs YvN
with the following boundary condition (ignoring the effect of BM ¼ ¼ ; (4)
evaporation):
rgs 1 Yvs
1166 S.S. Sazhin et al. / International Journal of Thermal Sciences 50 (2011) 1164e1180
P
Yv ¼ i Yvi is the vapour mass fraction, Sh is the Sherwood introduced by Approximation (11) would be less than 15%. This
number. This number and the Nusselt number for an isolated error is comparable or even less than the errors due to different
droplet (Shiso and Nuiso) were estimated based on the model sug- dependencies of Sh* and Nu* on spacing between droplets (cf. Eqs.
gested in [17] (this model was extended to a moving multi- (24) and (25)). Remembering (11), Eq. (8) is simplified to:
component droplet in [18]): !
cpv 1
h i ! 4 ¼ : (12)
ð1 þ Red Scd Þ1=3 max 1; Re0:077 1 cp;air Le
lnð1 þ BM Þ d
Shiso ¼ 2 1þ ;
BM 2FðBM Þ The latter expression for 4 is used in our analysis. This approach
(5) allowed us to avoid the iteration procedure to take into account the
dependence of 4 on BM,T.
h i The thermal expansion or contraction of droplets was taken into
! account as in [9]. The effect of recirculation inside droplet was
lnð1 þ BT Þ ð1 þ Red Prd Þ1=3 max 1; Re0:077
d 1
Nuiso ¼ 2 1þ ; taken into account based on the effective thermal conductivity
BT 2FðBT Þ (ETC) model [17] in which the liquid thermal conductivity kl is
(6) replaced by the effective thermal conductivity
where
nair keff ¼ ckl ; (13)
Scd ¼
Dv where the coefficient c is defined as
is the Schmidt number, h i
c ¼ 1:86 þ 0:86tanh 2:225log10 PedðlÞ =30 ; (14)
cpv Tg Ts
BT ¼ PedðlÞ ¼ RedðlÞ Prl is the droplet Peclet number, in which the liquid
Leff
transport properties and the relative velocities between droplets
is the Spalding heat transfer number, and gas were used.
All transport coefficients for the mixtures and individual
0:7 ln 1 þ BM;T substances were estimated as discussed in Appendix 1. Note that
F BM;T ¼ 1 þ BM;T ;
BM;T our analysis is focused on the evaporation only; the analysis of
condensation is beyond the scope of this paper.
QL X Q Equations for mass fractions of liquid species Yli hYli ðt; RÞ inside
Leff ¼ L þ ¼ ei L i þ P L ; droplets were used in the following form:
_d
m m_i
i
i !
vYli v2 Yli 2 vYli
QL is the power spent on droplet heating, cpv is the specific heat ¼ Dl þ ; (15)
vt vR2 R vR
capacity of fuel vapour, ei ¼ ei ðtÞ are the evaporation rates of
P
_ i ¼ ei m
species i, m _d ¼
_ d ðm _
i mi Þ. BT and BM are linked by the where i > 1, Dl is the liquid mass diffusivity. Eq. (15) was solved with
following equation [17]: the following boundary condition [1]:
BT ¼ ð1 þ BM Þ4 1; (7) vYli
aðei Ylis Þ ¼ Dl ; (16)
vR R¼Rd 0
where
! ! and the initial condition Yli ðt ¼ 0Þ ¼ Yli0 ðRÞ, where Ylis ¼ Ylis ðtÞ
cpv Sh* 1 are liquid components’ mass fractions at the droplet’s surface,
4 ¼ ; (8)
cp;air Nu* Le
_ dj
jm
a ¼ ¼ R_ de : (17)
Le ¼ kg =ðcp;air rtotal Dv Þ ¼ Sc=Pr is the Lewis number, 4prl R2d
h i ! Assuming that species concentrations in the ambient gas are equal
ð1 þ Red Scd Þ1=3 max 1; Re0:077
d 1
*
Sh ¼ 2 1 þ ; (9) to zero ðYvN ¼ 0Þ, the values of ei can be found from the following
2FðBM Þ relation [9]:
h i ! Y
ð1 þ Red Prd Þ1=3 max 1; Re0:077 1 ei ¼ P vis ; (18)
* d i Yvis
Nu ¼ 2 1 þ : (10)
2FðBT Þ
where the subscript v indicates the vapour phase. We assume that ei
When deriving (9) and (10) we assumed that the corrections to Sh* are still defined by Eq. (18) even in the case when YvN s0, which is
and Nu* due to finite distance between droplets are the same. expected in the case of a coupled solution. The analytical solution of
Remembering that FðBM;T Þ is a slowly varying function in the range Eq. (15) is given in Appendix 2. This solution is generalised to the
from 1 to 1.285 [17] and assuming that Le is close to 1, we can case of moving droplets with the help of the effective diffusivity
expect that model [1,9] in which Dl in Eq. (15) is replaced with the so-called
effective diffusivity Deff [37]:
!
Sh* Deff ¼ cY Dl ; (19)
z1: (11)
Nu*
where
Note that for small Re, condition (11) is always satisfied. For large h i
cY ¼ 1:86 þ 0:86tanh 2:225log10 RedðlÞ Scl =30 : (20)
Re, ðSh* =Nu* ÞzLe1=3 . In this case, even if Le ¼ 1.5, the error
S.S. Sazhin et al. / International Journal of Thermal Sciences 50 (2011) 1164e1180 1167
Table 1
The numerical values of parameters in Eqns. (24) and (25).
Parameter AT GT CT AM GM CM
Value 0.155335 1.416767 0.535026 0.155335 6.503406 107 54302.36
1168 S.S. Sazhin et al. / International Journal of Thermal Sciences 50 (2011) 1164e1180
coefficient of the fuel vapour in the gas inside the region of influ-
ence. Equations (28) and (29) allow us to assume that this region of
influence is large enough to enable us to ignore heat exchange
between it and the ambient gas, and escape of fuel vapour from this
volume to the ambient gas.
Let us first focus on the mass transfer process.
patm Vg
Ntotal ¼ ; (30)
Ru T
where Ru is the universal gas constant. Initially, there is no vapour
in Vg and Ntotal ¼ Nair at T ¼ T0 . Once the evaporation process has
PN
started then a certain number of moles of vapour Nv ¼ i¼1 Nvi,
where N is the total number of species, penetrates into volume Vg.
Simultaneously the temperature changes from the initial temper-
ature T0 to T1. The number of moles of air in this case reduces (or
increases if T1 is sufficiently lower than T0) to:
Fig. 1. Schematic diagram for the spherical region of influence or the plane perpen-
dicular to the cylinder axis for the cylindrical region of influence. patm Vg X N
Nair ¼ Nvi : (31)
Ru T1 i¼1
differential equations describing mass, momentum and energy This allows us to calculate the mass fraction of vapour inside the
conservation between cells. Although this is true in the case of the cell as
Eulerian approach, the results usually start depending on the grid
PN
size for sufficiently small cells in the case of the Lagrangian/Euler- i¼1 Mvi Nvi
YvðcellÞ ¼ PN ; (32)
ian approach widely used for spray simulation. This problem was Mvi Nvi þ Mair Nair
i¼1
extensively discussed in [24], where the concept of the region of
influence was first introduced. The main idea of this concept is to where Mair is the molar mass of air, Mvi are molar masses of vapour
allow droplets to exchange mass, momentum and energy not only species. We assume that YvN ¼ YvðcellÞ . In the one-way solution
with gas in a cell, in which a droplet is located, but with gas in described in [9] it was assumed that YvN ¼ 0. When deriving Eqs.
a wider à priori determined region (region of influence). Essentially (31) and (32) we used our assumption that the region of influence is
the same idea is used in our approach. The size of this region is large enough to prohibit the fuel vapour from escaping from
considered as a free parameter, within a certain range, which can be volume Vg. For sufficiently small timesteps, we can assume that
adjusted to get the best fit with experimental data. In the case of T1 ¼ T0 in Eq. (31) and updated at the next time step.
very large Vg, the coupled solution reduces to the one-way solution. The calculation continues until YvðcellÞ ¼ Yvs . Once this happens,
For small Vg, we can obtain an unphysical solution, since in this case the droplet stops evaporating (BM ¼ 0). Apart from YvðcellÞ , Eqs. (31)
the interaction of gas in Vg and the ambient gas, ignored in our and (32) are used for calculation of physical properties of the
analysis, can play the dominant role. mixture of vapour and air.
We appreciate that this approach does not allow us to perform
a quantitative predictive analysis of mass, momentum and energy 3.3. Heat balance
exchange between individual droplets and gas, but it does enable
us to perform a simple qualitative analysis of the effect of this Assuming that gas temperature inside volume Vg is homoge-
exchange on time evolution of droplet temperature and radius. As neous, the time evolution of this temperature can be described by
shown later, it allows us to explain some discrepancies between the equation:
experimental measurements and the predictions of the one-way
model. Note that this model will be applied to the analysis of d mg cpg Tg
¼ 2pkg Nud Rd Tg Ts ; (33)
experiments similar to those described in [5,20,21], where the dt
droplets’ velocities were determined from the experiments and not
where mg is the total mass of gas (mixture of air and vapour) in
calculated. Hence the momentum transfer between the region of
volume Vg, Nud is the Nusselt number for evaporating droplets
influence and the droplets is not considered in our model. In this
(taking into account the heating/cooling of the vapour).
case, we can approximate Rg as
Eq. (33) was solved at each time step. The value of mg in this
pffiffiffiffiffiffiffiffiffiffi equation is the mass of the mixture of air and fuel vapour as
Rg ¼ Rd þ kD tD ; (28)
calculated at the previous time step. The values of kg and cpg were
where tD is the characteristic duration of the process, calculated for the mixture of air and fuel vapour. The physical
! meaning of Eq. (33) is obvious: this is the mathematical expression
kg of the statement that the energy lost by the gas is spent on droplet
kD ¼ max ; Dv ; (29) heating and evaporation.
cpg rg
Note that in our approach to the coupled solution we implicitly
where the first term in the latter equation describes the heat took into account the transient nature of the process of droplet
diffusivity in volume Vg, while the second term describes diffusion heating and evaporation. At the same time, the expressions for Nu
S.S. Sazhin et al. / International Journal of Thermal Sciences 50 (2011) 1164e1180 1169
and Sh, discussed in Section 2, were derived under the assumption negative then the time step is reduced and the calculations are
that the process is quasi-steady (heat and mass flow rate through repeated. If the ratio of this radius to the initial radius is less
each sphere surrounding the droplet remains the same). Transient than an a priori chosen small number es ¼ 106, then the
effects on Nu have been considered in a number of papers, remaining part of the droplet is assumed to be evaporated with
including [25e27]. As shown in [25], the quasi-steady approxima- all liquid species transferred into the gas phase with the cor-
tion is valid at times much greater than tc hR2d cp rg =ðpkg Þ. For responding decrease in gas temperature. If this ratio is greater
droplets with radii about 60 mm and typical parameters for the than 106 then go to the next step.
experiment considered in our paper, tc z 0.25 ms. The time scales of 15. Recalculate the distributions of temperature and species for the
new radius (e.g. TðRÞ ¼ TðRRd2 =Rd1 Þ ¼ TðRÞ,~ where R
the processes under consideration are much greater than 0.25 ms. d1;2 are
We anticipate that the quasi-steady approximation for Sh is valid droplet radii at the beginning and the end of the time step, R ~ is
after a certain ~t c which is of the same order of magnitude as the the new R used at the second time step, T are the values of
abovementioned tc. The approach which we use is consistent with temperature at the end of the time step).
widely used practice in Computational Fluid Dynamics (CDF) codes 16. Return to Step 1 and repeat the calculations for the next time
[1,24]. For sufficiently large computational cells, only a small frac- step.
tion of mass and heat exchanged between gas and droplets is lost in
the ambient gas.
5. Experimental setup and input parameters
4. Numerical algorithm
The experimental setup used for validation of the model is the
The numerical scheme outlined below is specifically focused on same as described in [20,21]. Monodispersed and periodically
the application of the model for interpretation of the experimental spaced droplets were generated by the break-up of a liquid jet via
results reported in [5,20,23]. In the experiments reported in these Rayleigh-type instability at the injector outlet. A piezoelectric
papers ambient pressure was constant. The time evolution of ceramic inserted into the injector body triggered the disintegration
droplet velocities was directly measured and these were used as of the liquid jet by vibrations at certain frequencies. The fuel
input parameters of the model at each time step. In contrast to [9], temperature was regulated in the injector body and a thermo-
the effects of droplets on the gas phase were taken into account couple (located near the outlet of the injector) was used to verify
(coupled solution). that the set temperature was maintained the same during the
These are the main steps of the numerical algorithm: experiment. The droplets were injected into a hot air enclosure
equipped with several optical windows for measurements. Air
1. Assume the initial distribution of temperature and mass frac- within the enclosure was preheated up to 400 C. The walls of the
tions of species inside the droplet or use the distributions enclosure were made of thermally isolating materials, and an
obtained at the previous time step (in our case the initial electric wire was inserted into the interior of the walls. The electric
distributions of both were assumed homogeneous). Recalculate current was adjusted so that the Joule heating could compensate
the mass fractions of species into molar fractions of species. thermal losses. A constant air flow was maintained to avoid the risk
2. Calculate the values of liquid thermal conductivity and effec- of progressive saturation of vapour in the enclosure. An almost
tive thermal conductivity of the droplet. Use the values of the uniform air velocity in the cross-section of the enclosure was
droplet velocities from the experimental data. achieved by introducing a layer of metal foam at the air inlet. Air
3. Calculate species partial pressures and molar fractions in the velocity of about 0.3 m/s turned out to be sufficient to deplete
gas phase from the Raoult law. vapour from the enclosure. The relatively large radius of the
4. Calculate maximum of the thermal and mass diffusivities; enclosure (0.05 m) and small diffusion length did not allow vapour
calculate Rg according Eq. (28) with tD ¼ 12 ms (typical transit to diffuse to the walls before it was removed through the holes
time of droplets). drilled at the top of the enclosure.
5. Calculate the interaction volume based on Eq. (26) for the Downstream distance from the injector was converted into time
spherical volume and Eq. (27) for the cylindrical volume where with the help of the space evolution of the droplet velocity measured
d ¼ C 2 Rd, C is the distance parameter. using the Laser Doppler Anemometry (LDA) technique. Droplet
6. Calculate the concentration of vapour of all species in the temperatures were measured using the laser induced fluorescence
region of influence and the value of the Spalding mass transfer technique described in [28,20]. The emissions from two different
number. spectral bands with different temperature sensitivities were utilized.
7. Calculate the values of heat capacity, diffusivity of the mixture The ratio of the intensities of these two bands allowed us to eliminate
of vapour species in the air and species evaporation rates (ei). the effects of parameters that are unknown or difficult to control,
8. Calculate the values of Nusselt and Sherwood numbers for such as the variations in laser intensity, tracer concentration or
isolated droplets. measurement volume during the acquisitions. At the same time this
9. Calculate the values of Nusselt and Sherwood numbers for ratio depends on the temperature, making possible its measurement.
droplets, taking into account the interaction between them Although fluorescent tracers like Rhodamine B or Kitton red are well
(Eqs. (24) and (25)). known for the temperature sensitivity of their fluorescent signal and
10. Calculate the total change of mass of the evaporating droplet their high quantum yield, the fluorescence spectrum of these mole-
during the time step Dt, taking into account the interaction cules can vary significantly with the composition of the liquid
between droplets and the change in liquid density. mixture in which they are dissolved. To account for this effect,
11. Calculate the rate of change of droplet radius. a special method has been developed and described in [29]. This
12. Calculate the distribution of temperature inside the droplet. method is based on the inclusion of a third spectral band. However, in
13. Calculate the distribution of species inside the droplet. the case of the n-decane/3-pentanone mixture, considered in this
14. Recalculate the droplet’s radius at the end of the time step Dt. paper, we found a new fluorescent tracer, Pyrromethene 597-C8, for
In our analysis the droplets never fully evaporated, but the which the fluorescent ratio is almost independent on the volume
program was designed to deal with the case when the fraction of 3-pentanone up to about 15% with only a negligible effect
complete evaporation takes place, if necessary. If this radius is to the sensitivity of the fluorescent ratio on the temperature [30].
1170 S.S. Sazhin et al. / International Journal of Thermal Sciences 50 (2011) 1164e1180
Measurements, using this tracer were therefore performed with only Table 3
two bands. The probe volume (intersection between the laser beams The experimentally observed time evolutions of droplet average temperatures
velocities and diameters for the 95% decane e 5% 3-pentanone mixture droplets
and the detection field of view) was about 150 150 1200 mm3. It (Case 1), 90% decane e 10% 3-pentanone mixture droplets (Case 2) and 85% decane e
was larger than the droplet in order to provide a global excitation of 15% 3-pentanone mixture droplets (Case 3). The contribution of gas velocity is
the whole droplet volume. The signal was averaged over the total ignored in the estimates of Sherwood and Nusselt numbers as in [9]. Time was
time of droplet transit in the probe volume. measured from the moment of injection.
In the presence of thermal gradients inside droplets, the Case Time (ms) Average Velocity (m/s) Diameter (mm)
temperature obtained, based on this signal, could differ from the Temperature ( C)
volume-averaged droplet temperature. This is related to the fact 1 4.784 28.66 7.77 122.7
that the laser intensity is not distributed uniformly inside the 1 6.105 30.45 7.33 122.7
droplet, mainly due to light refraction at the droplet surface and the 1 7.476 32 7.22 122.5
1 8.897 33.9 6.96 122.7
Gaussian profile of intensity in the laser beams. Also, the receiving 1 10.368 36.8 6.75 123.1
optics collect the fluorescent signal from the droplet interior with 1 11.889 40.1 6.44 123.3
a spatial distortion. These effects could be taken into account with 2 3.86 25.6 9.34 126.1
the help of an efficient ray tracing code, based on the geometric 2 4.9 26.67 9.04 125.7
2 5.97 28.17 8.6 125.6
optics approximation [31]. This analysis, however, is beyond the
2 7.07 29.9 8.25 125.5
scope of this paper. We would just mention that the 2cLIF 2 8.2 31.88 8.04 125.4
measurements allow us to estimate the volume-averaged droplet 2 9.36 34.74 7.86 125.5
temperature with systematically more weighting in the zones near 3 3.86 26.3 9.2 127.7
the centre of the droplet rather than at its edge. We cannot quantify 3 4.9 27.99 8.92 127.6
3 5.97 29.2 8.67 127.5
this effect at this stage and we assume that the measured 3 7.07 30.83 8.38 127.2
temperatures are close to the droplet average temperatures. 3 8.2 32.26 8.04 127.1
Moreover, the forward scattering interferometry (FSI) was used 3 9.36 34.7 7.91 126.6
to determine the droplet diameter with a good accuracy [32]. In the 3 9.36 36.2 7.69 126.3
case of a droplet having diameter of 120 mm, the uncertainty of the
diameter measurement using this method was estimated to be
about 1.2 mm [33]. results considered in this paper. Remembering that this estimate
Three sets of experimental data referring to the mixtures was made for the maximal sD, and is rather crude by its nature, we
of decane and 3-pentanone will be used in our analysis. These are 95% can take the actual value of this parameter a bit less than 780 in the
decane e 5% 3-pentanone, 90% decane e 10% 3-pentanone and 85% coupled solution, but not too small to affect the assumptions of the
decane e 15% 3-pentanone mixture droplets, which will be referred model.
to as Cases 1, 2 and 3 respectively. Droplet initial diameters, ambient The calculations were performed for rhVg =Vd ¼ 600 (coupled
temperatures, distance parameters and injection frequencies for solution) and r ¼ 5 104 (it was shown that for this r the results
these cases are presented in Table 2. The experimentally observed predicted by the coupled solution are indistinguishable from the
time evolutions of droplet average temperatures, velocities and results predicted by the one-way solution, described in [9]). Twenty
diameters are presented in Table 3. terms in the series in the analytical solution for temperature and
species were used in the calculations (the sensitivity of the results
with respect to the choice of the number of terms will be investi-
6. Results gated in Section 7). As in [9], in both cases the temperatures at the
centre of the droplets, the surface of the droplets and the average
A crucial parameter which needs to be considered in the droplet temperatures were calculated. The values of ambient gas
coupled solution is the volume of gas where the interaction temperature and droplet velocities were the input parameters of
between droplets and gas is taken into account, or even more the model. Note that although the ambient gas temperature
important the ratio of this volume and the volume of the droplet remained constant during each experiment, in the case of the
rhVg =Vd . For spherical and cylindrical cells these volumes are coupled solution gas temperature inside the volume Vg was allowed
linked with parameter Rg via Eqs. (26) and (27) respectively. A to change with time. This temperature plays the same role as the
crude estimate of this parameter is given by Eq. (28). Taking sD in ambient temperature in the case of the one-way solution. As in [9]
this equation equal to 12 ms (characteristic transit time of droplets), the calculations started at the moment when droplets were first
Rd ¼ 61.35 mm (see Table 2), d ¼ 463.806 mm (based on the distance observed. The initial droplet temperatures were assumed equal to
parameter equal to 3.78 (see Table 2)), and estimating kD as the measured temperatures of the first observed droplets. As in [9],
3.61 105 m2/s, we obtain Rg z 7.2004 104 m z 720 mm. This it is assumed that initially there is no temperature gradient inside
gives us r z 1615 for the spherical cell, and r z 780 for the cylin- droplets.
drical one. The latter is more appropriate for the experimental The time step in the calculations was taken equal to 105 s, and
the number of points along the radius inside droplets, where the
temperatures were stored, was taken equal to 2000. The latter
Table 2
number controls the accuracy of the calculation of the spatial
Droplet initial diameters, ambient temperatures, distance parameters and injection derivatives of the temperature and species mass fractions, used in
frequencies for the 95% decane e 5% 3-pentanone mixture droplets (Case 1), 90% boundary conditions for T and Yli (see Eqs. (2) and (16)). The reason
decane e 10% 3-pentanone mixture droplets (Case 2) and 85% decane e 15% 3- for this choice of the time step and the number of points will be
pentanone mixture droplets (Case 3).
discussed later in Section 7.
Parameter Case 1 Case 2 Case 3 The plots of time evolutions of the temperatures at the centre of
Initial diameter 122.7 mm 126.1 mm 127.7 mm the droplets, the surface of the droplets and the average droplet
Ambient temperature 370 C 374 C 374 C temperatures, predicted by the coupled and one-way solutions for
Distance parameter 3.78 3.84 3.8
the 95% decane e 5% 3-pentanone mixture droplets (Case 1), are
Injection frequency 20600 Hz 20600 Hz 20500 Hz
shown in Fig. 2. As follows from this figure, all three temperatures
S.S. Sazhin et al. / International Journal of Thermal Sciences 50 (2011) 1164e1180 1171
Fig. 4. The same as Figs. 2 and 3 but for the 85% n-decane 15% 3-pentanone mixture
Fig. 2. The time evolution of the droplet surface, average and central temperatures (Ts, droplets with initial diameters 127.7 mm, homogeneous temperature 26.3 C injected
Tav and Tc), as predicted by the coupled and one-way solutions, and experimentally into an ambient gas at constant temperature equal to 374 C (see Tables 2 and 3).
observed temperatures for the 95% n-decane 5% 3-pentanone mixture droplets with
initial diameters 122.7 mm and homogeneous temperature 28.66 C injected into an
ambient gas at constant temperature equal to 370 C (see Tables 2 and 3). Zero time The plots of time evolutions of the gas temperatures in the
corresponds to the start of injection.
region of influence for all three cases, as predicted by the coupled
solution, are shown in Fig. 5. In the case of the one-way solution,
under consideration are well separated, as in the cases, considered this temperature remained constant. As can be seen from Fig. 5, the
in [9]. Also, the coupled solutions show visibly slower rates of drop in the gas temperature during droplet heating could reach up
droplet heating, compared with the predictions of the one-way to about 70 C and this leads to a visible reduction of the rate of
solution. All experimental plots lie between the average droplet increase of droplets temperatures predicted by the coupled solu-
temperature and the temperature at the centre of the droplet tion. This drop in gas temperature is expected to produce much
predicted by the coupled solution. stronger effect due to the coupled solution, compared with the
The same plots as in Fig. 2 but for the 90% decane e 10% addition of fuel vapour to the region of influence.
3-pentanone mixture droplets (Case 2) and the 85% decane e 15% The plots, similar to those shown in Figs. 2e4, but for droplet
3-pentanone mixture droplets (Case 3) are shown in Figs. 3 and 4 radii, are shown in Figs. 6e8. As one can see from these figures,
respectively. The trends of all curves in these figures are rather neither coupled nor one-way solutions can predict the observed
similar to those shown in Fig. 2. In both cases experimental plots lie trends in the time evolution of droplet radii. At the same time, one
between the average droplet temperature and the temperature at can see from these figures that the deviation between the predicted
the centre of the droplet predicted by the coupled solution, as in the and observed droplet radii is less than about 0.5 mm in most cases
case shown in Fig. 2, except three experimental points in Fig. 4. (less than 1%), which is within the margins of experimental errors.
Even in the latter case, however, the deviation between the pre- Hence, the deviations between the predicted and observed droplet
dicted temperatures at the centre of the droplet and the ones radii cannot undermine the validity of the model.
obtained experimentally is well within the experimental error. In Finally we present four sets of plots, which cannot be validated
all three cases, the one-way solution predicts unrealistically high against experimental data at the moment, but show the underlying
average droplet temperatures, well above those observed experi- physics of the processes involved. The temperature distribution
mentally. Hence, the effects of coupling need to be taken into inside the 95% decane e 5% 3-pentanone mixture droplets (Case 1)
account for the accurate prediction droplet temperatures. at three moments of time, as predicted by the coupled and one-way
Fig. 3. The same as Fig. 2 but for the 90% n-decane 10% 3-pentanone mixture droplets Fig. 5. The time evolution of gas temperature in the region of influence for three
with initial diameters 126.1 mm, homogeneous temperature 25.6 C injected into an mixtures of n-decane and 3-pentanone. The droplets and the ambient gas parameters
ambient gas at constant temperature equal to 374 C (see Tables 2 and 3). are the same as used in Figs. 2e4.
1172 S.S. Sazhin et al. / International Journal of Thermal Sciences 50 (2011) 1164e1180
Fig. 6. The plots of Rd versus time predicted by the one-way and coupled solutions for
the same conditions as in Fig. 2.
Fig. 8. The plots of Rd versus time predicted by the one-way and coupled solutions for
the same conditions as in Fig. 4.
solutions, is presented in Fig. 9. As can be seen from this figure, at all
times the temperature increases with the distance from the centre
of the droplets, as expected. The temperature gradient is less influence into ambient gas. The predicted mass of n-decane is
important for case predicted by the coupled solution compared always greater than that of 3-pentanone, despite higher volatility of
with the one-way solution. The deviation between the tempera- 3-pentanone, compared with n-decane (cf. Fig. 10). This is related to
tures in all areas of the droplets, predicted by coupled and one-way the fact that the original amount of n-decane in droplets is 20 times
solutions, increases with time. more than the amount of 3-pentanone.
The plots similar to those shown in Fig. 9, but for the mass The time evolution of masses of air and the total mass of the
fraction of n-decane, are presented in Fig. 10. As can be seen from mixture of air and vapour in the region of influence is shown in
this figure, the mass fraction of n-decane in the areas close to the Fig. 12. The increase of the mass of air with time is related to the
surface of the droplet, predicted by coupled and on-way solutions, decrease of temperature in the interaction volume with time (cf.
increases with time. This is related to higher volatility of 3-penta- Fig. 5) at constant pressure. Note that for a fixed temperature, part
none, compared with n-decane. As in the case of temperature, of air should have been removed from the interaction volume by
shown in Fig. 9, the rate of increase of the mass fraction of n-decane, the evaporating fuel (see Eq. (31)).
predicted by the coupled solution, is slower than the one predicted
by the one-way solution. The deviation between the results pre- 7. Accuracy versus CPU efficiency
dicted by these solutions increases with time. This can be related to
the fact that in the coupled solution, vapour in the region of influ- In the analysis, reported in our earlier paper [9], we were con-
ence is partially saturated and the temperature in this region is cerned with the accuracy of the results, and chose the maximum
lower than that in the ambient gas. number of terms in the series in the analytical solutions for
The time evolution of evaporated masses of 3-pentanone, temperature and species mass fractions. However, remembering
n-decane and the total vapour mass in the region of influence is that perhaps the most important practical application of the model
shown in Fig. 11. As one can see from this figure, evaporated masses of described in [9] would be its potential implementation into
both substances increase with time. This can be related to the fact that computational fluid dynamic (CFD) codes, and application of the
no condensation takes place during the period under consideration, latter to simulation of realistic engineering processes, finding
and to our assumption that no vapour escapes from the region of a reasonable compromise between the accuracy of the model and
Fig. 9. The plots of T versus R=Rd for three moments of time after the start of calcu-
Fig. 7. The plots of Rd versus time predicted by the one-way and coupled solutions for lations predicted by the one-way and coupled solutions and for the same conditions as
the same conditions as in Fig. 3. in Fig. 2.
S.S. Sazhin et al. / International Journal of Thermal Sciences 50 (2011) 1164e1180 1173
Fig. 10. The plots of the n-decane mass fraction Yl;ndecane versus R=Rd for three Fig. 12. The plots of the total mass of air and the mixture of air and vapour in the
moments of time after the start of calculations predicted by the one-way and coupled region of influence versus time predicted by the coupled solution for the same
solutions and for the same conditions as in Fig. 2. conditions as in Fig. 2.
its CPU efficiency becomes more important than its accuracy where F stands for either temperature or mass fraction, N is the
alone. The accuracy and CPU efficiency of the model, depending on number of terms in the series in the solution to the equation for
the number of terms in the series in the analytical solutions for temperature or mass fraction, i refers to time steps, imax is the
temperature and species mass fractions, is investigated in maximal number of timesteps used in calculation (this number
this section. We focus on the analysis of the 95% decane e 5% 3- varied from about 700 to about 800). The errors were calculated
pentanone mixture droplets (Case 1). The conclusions referring to relative to the values obtained when 100 terms in the expression
other mixtures are essentially the same as the ones for the for the temperature or mass fraction of n-decane were used. Higher
abovementioned mixture. order terms did not produce any effects on the results.
In our previous paper [34] we have shown that modelling of The plots of relative errors of the temperatures at the centre of
mono-component droplet heating and evaporation, based on the droplets (Tc), the surface of the droplets (Ts), the average droplet
the analytical solution of the heat conduction equation inside the temperatures (Tav), predicted by the coupled solution, and the CPU
droplet, is more CPU efficient and accurate than when based on the requirement versus the number of terms in the series in the
numerical solution of this equation. We anticipate that the same expression for droplet temperature (n) are presented in Fig. 13. 100
conclusion remains valid for multi-component droplets of the type terms in the expression for the mass fraction of n-decane (Eq. (79)
considered in our paper, when the species diffusion inside droplets in Appendix 2) were taken. All temperatures were measured in C.
is taken into account. Calculations were performed on 3000 Hz CPU, 3 GB RAM work
The relative errors were estimated based on the following station. As follows from this figure, the errors of calculating all
equation: temperatures, are practically equal to zero when the number of
Pi ¼ imax terms n is close or larger than 20. Hence, 20 terms were used in the
i¼1
ðjFN¼n FN¼100 j=FN¼100 Þ analysis presented in Section 6. The CPU time increased by about
Error ¼ ;
imax 50% when the number of terms in the expression for temperature
increased from 20 to 100. This increase of the number of terms,
however, does not lead to any increase in the accuracy of the results
Fig. 13. The plots of relative errors (in percent) of calculation of droplet surface,
Fig. 11. The plots of the evaporated masses of n-decane, 3-pentanone and the total average and central temperatures (Ts, Tav and Tc) and CPU time versus the number of
evaporated mass versus time predicted by the coupled solution for the same condi- terms in the solution to the temperature equation with the fixed number of terms in
tions as in Fig. 2. the solution to the species equation equal to 100 for the same conditions as in Fig. 2.
1174 S.S. Sazhin et al. / International Journal of Thermal Sciences 50 (2011) 1164e1180
Fig. 14. The plots of relative errors (in percent) of calculation of liquid n-decane mass Fig. 16. The plots of errors (in percent) of calculation of liquid n-decane mass fractions
fractions at the surface of the droplet, average mass fraction of n-decane and its mass at the surface of the droplet, average mass fraction and the mass fraction at the centre
fraction at the centre of the droplet (Ys, Yav and Yc) and CPU time versus the number of of the droplet (Ys, Yav and Yc) and CPU time versus the number of terms in the solutions
terms in the solution to the species equation with fixed number of terms in the to the temperature and species equations (these numbers are assumed equal) for the
solution to the temperature equation equal to 100 for the same conditions as in Fig. 2. same conditions as in Fig. 2.
S.S. Sazhin et al. / International Journal of Thermal Sciences 50 (2011) 1164e1180 1175
Acknowledgements
Fig. 17. The plots of the predicted droplet surface temperatures versus NR for different The authors are grateful to the European Regional Development
timesteps for the 95% n-decane e 5% 3-pentanone mixture (Case 1) at 5 ms after the Fund Franco-British INTERREG IVA (Project C5, Reference 4005) for
start of calculations. the financial support of this project.
Appendix 1
For all timesteps in the range 106e104 s and NR 1500 the
predicted temperature remains almost the same. For smaller NR, the
Transport coefficients for mixtures
accuracy of calculations clearly deteriorates, especially for the time
step equal to 106 s. The plots, shown in Fig. 17, are similar to those
Liquid thermal conductivity for a mixture
obtained for other cases and other moments of time (these are not
A number of approximate formulae for thermal conductivity in
shown in the paper). Based on these plots, the values of the time
multi-component liquids are reviewed in [35]. We use the Vrede-
step 105 s and NR ¼ 2000 have been chosen. Note that the CPU
veld equation, valid for nonaqueous systems in which the ratio of
time could be further reduced, practically without any detrimental
thermal conductivities does not exceed two [35]:
effects on accuracy, if the time step is increased to 104 s and NR is
reduced to 1000. X 1=2
Note that although the numerical code, used in our analysis, is kl ¼ Yi k2
li ; (34)
based on the analytical solutions both for temperature and species i
mass fractions, it still requires the introduction of time steps and where Yi are mass fractions for species i, kli are thermal conduc-
the spatial mesh inside the droplets. The time steps are required to tivities of species i.
take into account the time dependence of gas temperature in the
zone of influence and the evaporation process. The spatial mesh is Liquid viscosity for a mixture
required for calculating the integrals used in the analytical solu- A number of approximate formulae for dynamic viscosity of
tions, the averaging of variables and postprocessing. multi-component liquids (ml) are reviewed in [35]. It was shown
that in the case of two species, the approach suggested by Grunberg
8. Conclusions and Nissan [36] is the most accurate and convenient in practical
applications:
The earlier reported simplified model for multi-component
droplet heating and evaporation is generalised to take into account X
2
lnml ¼ Xi lnmli þ X1 X2 G12 ;
the coupling between the droplets and the ambient gas. Similarly to
i¼1
the original simplified model, the model takes into account droplet
heating by convection from the ambient gas, the distribution of where G12 can be negative or positive.
temperature inside the droplet, diffusion of liquid species inside the As follows from the example presented in [36], the contribution
droplet, droplet swelling or contraction due to changing average of the term proportional to G12 in this equation is about 1%. In our
temperature, the effects of recirculation in the moving droplets on analysis the liquid viscosity is used for the estimate of the correc-
heat and species mass diffusion within them. The effects of the non- tion factors in effective thermal conductivity and effective diffu-
unity activity coefficient are ignored (the Raoult law is assumed to sivity models. These correction factors are rather weak functions of
be valid) and the interaction between droplets is taken into account liquid viscosity especially for large Peclet numbers. This allows us to
based on the correlation suggested in [20]. The size of the gas ignore the term proportional to G12, and reduce the above equation
volume, where the interaction between droplets and gas needs to to the Arrhenius formula [36]
be taken into account (region of influence), is estimated based on
X
2
the characteristic thermal and mass diffusion scales. lnml ¼ Xi lnmli :
The model is applied to the analysis of the experimentally i¼1
observed heating and evaporation of monodispersed n-decane/3-
pentanone mixture droplets at atmospheric pressure. It is pointed In the case of several components, this equation can be gener-
out that the effect of coupling leads to noticeably better agreement alised to
between the predictions of the model and the experimentally X
observed average droplet temperatures. In most cases, the experi-
lnml ¼ Xi lnmli : (35)
i
mentally observed droplet temperatures lie between the average
and central temperatures, predicted by the coupled solution. The The latter equation is used in our analysis.
1176 S.S. Sazhin et al. / International Journal of Thermal Sciences 50 (2011) 1164e1180
Yli ðt; RÞ ¼ yðt; RÞ þ eðtÞ; (43) Bð1 þ h0 Þ ¼ 0. Since h0 ˛ðN; 1Þ, the latter equation is satisfied
only when B ¼ 0. This leads to the trivial solution v ¼ 0 which is
where the subscript i at e is omitted to simplify the notation.
disregarded in our analysis. Hence, Eq. (51) has no non-trivial
Having substituted (43) into Eq. (15) and the boundary condi-
solutions for p ¼ 0.
tion (42) we can rewrite this equation and the corresponding
boundary and initial conditions in the form:
The SturmeLiouville problem for p < 0
! 2
Assuming that p ¼ l < 0 we write the general solution to
2
vy v y 2 vy deðtÞ (51) as:
¼ Dl þ ; (44)
vt vR2 R vR dt
R R
vðRÞ ¼ Acosh l þ Bsinh l ; (53)
vy a Rd Rd
y ¼ 0; (45)
vR Dl R¼Rd where A and B are arbitrary constants.
The boundary condition at R ¼ 0 (see (52)) implies that A ¼ 0.
yjt¼0 ¼ Yli0 ðRÞ eð0ÞhYli0 ðRÞ e0 : (46) The boundary condition at R ¼ Rd leads to the following equation:
n 0.
Expansion of R in a Fourier series with respect to functions vn
The solutions to Eqs. (71) and (72) subject to the initial condition
Before we are able to use (50) for the solution of Eq. (47) we
(73) can be presented in the form:
need to find the presentation of R as a Fourier series with respect to " 2 #
functions vn: l0
Q0 ðtÞ ¼ exp Dl t ½qi0 eð0ÞQ0
Rd
Zt " #
X
N deðsÞ l 2
R ¼ Qn vn ðRÞ; (67) Q0 exp Dl 0 ðt sÞ ds; (75)
ds Rd
n¼0 0
S.S. Sazhin et al. / International Journal of Thermal Sciences 50 (2011) 1164e1180 1179
" 2 #
l References
Qn ðtÞ ¼ exp Dl n t ½qin eð0ÞQn
Rd [1] S.S. Sazhin, Advanced models of fuel droplet heating and evaporation, Prog.
Zt " 2 # Energy Combust. Sci. 32 (2006) 162e214.
deðsÞ ln [2] M. Klingsporn, U. Renz, Vaporization of a binary unsteady spray at high
Qn exp Dl ðt sÞ ds; ð76Þ temperature and high pressure, Int. J. Heat Mass Transfer 37 (Suppl. 1) (1994)
ds Rd 265e272.
0 [3] P.L.C. Lage, C.M. Hackenberg, R.H. Rangel, Nonideal vaporization of dilating
binary droplets with radiation absorption, Combust. Flame 101 (1995)
where n 1. 36e44.
[4] D.J. Torres, P.J. O’Rourke, A.A. Amsden, Efficient multi-component fuel algo-
The final solution rithm, Combust. Theory Model. 7 (2003) 67e86.
[5] C. Maqua, G. Castanet, F. Lemoine, Bi-component droplets evaporation:
Having substituted (56), (61), (75) and (76), into (50), we can
temperature measurements and modelling, Fuel 87 (2008) 2932e2942.
present the final solution to Eq. (15), satisfying boundary condition [6] J. Tamim, W.L.H. Hallett, Continuous thermodynamics model for multicom-
(16) and the corresponding initial condition in the form: ponent vaporization, Chem. Eng. Sci. 50 (1995) 2933e2942.
[7] M. Burger, R. Schmehl, K. Prommersberger, O. Schäfer, R. Koch, S. Wittig,
(" " # Droplet evaporation modelling by the distillation curve model: accounting for
1 l 2
Yli ¼ ei þ exp Dl 0 t ½qi0 ei ð0ÞQ0 kerosene fuel and elevated pressures, Int. J. Heat Mass Transfer 46 (2003)
R Rd 4403e4412.
[8] W.A. Sirignano, G. Wu, Multicomponent-liquid-fuel vaporization with
Zt " # # complex configuration, Int. J. Heat Mass Transfer 51 (2008) 4759e4774.
dei ðsÞ l 2 R [9] S.S. Sazhin, A. Elwardany, P.A. Krutitskii, G. Castanet, F. Lemoine, E.M. Sazhina,
Q0 exp Dl 0 ðt sÞ ds sinh l0
ds Rd Rd M.R. Heikal, A simplified model for bi-component droplet heating and evap-
0 oration, Int. J. Heat Mass Transfer 53 (2010) 4495e4505.
" " 2 # [10] N. Nikolopoulos, A. Theodorakakos, G. Bergeles, A numerical investigation of
X
N
ln the evaporation process of a liquid droplet impinging onto a hot substrate, Int.
þ exp Dl t ½qin ei ð0ÞQn J. Heat Mass Transfer 50 (2007) 303e319.
n¼1
Rd [11] G. Strotos, M. Gavaises, A. Theodorakakos, G. Bergeles, Numerical investiga-
Zt " 2 # # ) tion on the evaporation of droplets depositing on heated surfaces at low
dei ðsÞ ln R Weber numbers, Int. J. Heat Mass Transfer 51 (2007) 1516e1529.
Qn exp Dl ðt sÞ ds sin ln ; [12] G. Strotos, M. Gavaises, A. Theodorakakos, G. Bergeles, Numerical investiga-
ds Rd Rd tion of the cooling effectiveness of a droplet impinging on a heated surface,
0 Int. J. Heat Mass Transfer 51 (2007) 4728e4742.
(77) [13] S.S. Sazhin, P.A. Krutitskii, I.G. Gusev, M.R. Heikal, Transient heating of an
evaporating droplet, Int. J. Heat Mass Transfer 53 (13e14) (2010) 2826e2836.
where Qn, qin, l0 and ln (n 1) are defined by Eqs. (69), (74), (55) [14] S.S. Sazhin, P.A. Krutitskii, I.G. Gusev, M.R. Heikal, Transient heating of an
and (60) respectively; the subscript i at e has been restored. evaporating droplet with presumed time evolution of its radius, Int. J. Heat
Mass Transfer 54 (5e6) (2011) 1278e1288.
Note that Expression (77) contains the term which exponen- [15] H.S. Carslaw, J.C. Jaeger, Conduction of Heat in Solids. Clarendon Press, Oxford,
tially increases with time. This, however, will not lead to an 1986.
unphysical solution to Eq. (15), since this equation is valid only for 0 [16] E.M. Kartashov, Analytical Methods in Heat Transfer Theory in Solids. Vys-
shaya Shkola, Moscow, 2001, (in Russian).
Yli 1. Once Yli reaches one of its limiting values it will remain [17] B. Abramzon, W.A. Sirignano, Droplet vaporization model for spray combus-
equal to this value. tion calculations, Int. J. Heat Mass Transfer 32 (1989) 1605e1618.
If we apply Solution (77) to individual short timesteps, we can [18] G. Continillo, W.A. Sirignano, Unsteady, spherically-symmetric flame propa-
gation through multicomponent fuel spray clouds. in: G. Angelino, L. De Luca,
ignore the time dependence of dei ðsÞ=ds during this time step and W.A. Sirignano (Eds.), Modern Research Topics in Aerospace Propulsion.
assume that Springer-Verlag, New York, Berlin, 1991, pp. 173e198.
[19] P. Atkins, J. de Paula, Atkins’ Physical Chemistry, seventh ed. Oxford University
dei ðtÞ de ðtÞ Press, 2002.
¼ i he0i : [20] V. Deprédurand, G. Castanet, F. Lemoine, Heat and mass transfer in evapo-
dt dt t¼0 rating droplets in interaction: influence of the fuel, Int. J. Heat Mass Transfer
53 (2010) 3495e3502.
This allows us to simplify Expression (77) to: [21] T. Kristyadi, V. Deprédurand, G. Castanet, L. Lemoine, S.S. Sazhin,
(" " #" !# A. Elwardany, E.M. Sazhina, M.R. Heikal, Monodisperse monocomponent fuel
1 l0 2 R2d 0 droplet heating and evaporation, Fuel 89 (2010) 3995e4001.
Yli ¼ ei þ exp Dl t qi0 Q0 ei ð0Þþ e
2 i
[22] S.S. Sazhin, W.A. Abdelghaffar, E.M. Sazhina, M.R. Heikal, Models for droplet
R Rd Dl l0 transient heating: effects on droplet evaporation, ignition, and break-up, Int. J.
# X " " 2 # Thermal Sci. 44 (2005) 610e622.
N
R2 R ln [23] C. Maqua, G. Castanet, F. Grisch, F. Lemoine, T. Kristyadi, S.S. Sazhin, Mono-
þQ0 d2 e0i sinh l0 þ exp Dl t disperse droplet heating and evaporation: experimental study and modelling,
Dl l0 Rd n¼1
Rd Int. J. Heat Mass Transfer 51 (2008) 3932e3945.
" !# # ) [24] S. Tonini, M. Gavaises, A. Theodorakakos, Modelling of high-pressure dense
R2 R2 R
qin Qn ei ð0Þ d2 e0i Qn d2 e0i sin ln : ð78Þ diesel sprays with adaptive local grid refinement, Int. J. Heat Fluid Flow 29
Dl Dl Rd (2008) 427e448.
l n l n [25] S.S. Sazhin, V. Goldshtein, M.R. Heikal, A transient formulation of Newton’s
cooling law for spherical bodies, ASME J. Heat Transfer 123 (2001) 63e64.
If we ignore the time dependence of ei then Expression (78) can [26] S.S. Sazhin, P.A. Krutitskii, S.B. Martynov, D. Mason, M.R. Heikal, E.M. Sazhina,
be simplified to Transient heating of a semitransparent spherical body, Int. J. Thermal Sci. 46
( " # (2007) 444e457.
1 l 2 R [27] S.S. Sazhin, I.G. Gusev, P.A. Krutitskii, M.R. Heikal, Transient heating of
Yli ¼ ei þ exp Dl 0 t ½qi0 ei ð0ÞQ0 sinh l0 a semitransparent spherical body immersed into a gas with inhomogeneous
R Rd Rd temperature distribution, Int. J. Thermal Sci. 50 (2011) 1215e1228.
" # ) [28] P. Lavieille, F. Lemoine, G. Lavergne, G.,N. Lebouch, Evaporating and com-
XN
ln 2 R busting droplet temperature measurements using two color laser induced
þ exp Dl t ½qin eð0ÞQn sin ln : fluorescence, Exp. Fluids 31 (2001) 45e55.
n¼1
Rd Rd
[29] C. Maqua, G. Castanet, N. Doué, G. Lavergne, F. Lemoine, Temperature
measurements of binary droplets using three color laser-induced fluores-
(79)
cence, Exp. Fluids 40 (2006) 786e797.
[30] V. Deprédurand, P. Miron, A. Labergue, M. Wolff, G. Castanet, F. Lemoine,
The assumption that the time dependence of ei can be ignored
A temperature sensitive tracer suitable for two-colour laser-induced fluo-
during the time step will be used in our analysis. This allows us to rescence thermometry applied to evaporating droplets, Meas. Sci. Technol. 19
base our calculations on Eq. (79). (2008) 1e12.
1180 S.S. Sazhin et al. / International Journal of Thermal Sciences 50 (2011) 1164e1180
[31] G. Castanet, A. Delconte, F. Lemoine, L. Mees, G. Grehan, Evaluation of [35] B.E. Polling, J.M. Prausnitz, J.P. O’Connell, The Properties of Gases and Liquids.
temperature gradients within combusting droplets in linear stream using two McGraw-Hill, 2000.
colors laser-induced fluorescence, Exp. Fluids 39 (2005) 431e440. [36] L. Grunberg, A.H. Nissan, Mixture law for viscosity, Nature 164 (1949)
[32] G. Knig, K. Anders, A. Frohn, A new light-scattering technique to measure the 799e800.
diameter of periodically generated moving droplets, J. Aerosol Sci. 17 (1986) [37] W.A. Sirignano, Fluid Dynamics and Transport of Droplets and Sprays. Cam-
157e167. bridge University Press, Cambridge, 1999.
[33] V. Deprédurand, Approche expérimentale de l’ évaporation de sprays de [38] S.S. Sazhin, T. Kristyadi, W.A. Abdelghaffar, M.R. Heikal, Models for fuel
combustibles multicomposant, Thesis of the Institut Polytechnique of Lor- droplet heating and evaporation: comparative analysis, Fuel 85 (2006)
raine, France, 2009. 1613e1630.
[34] S.S. Sazhin, W.A. Abdelghaffar, P.A. Krutitskii, E.M. Sazhina, M.R. Heikal, New [39] S.S. Sazhin, P.A. Krutitskii, W.A. Abdelghaffar, S.V. Mikhalovsky, S.T. Meikle,
approaches to numerical modelling of droplet transient heating and evapo- M.R. Heikal, Transient heating of diesel fuel droplets, Int. J. Heat Mass Transfer
ration, Int. J. Heat Mass Transfer 48 (19e20) (2005) 4215e4228. 47 (2004) 3327e3340.