Accepted Manuscript: Chemical Engineering Science
Accepted Manuscript: Chemical Engineering Science
Accepted Manuscript: Chemical Engineering Science
PII: S0009-2509(17)30760-1
DOI: https://doi.org/10.1016/j.ces.2017.12.023
Reference: CES 13958
Please cite this article as: S. Rieks, E.Y. Kenig, Modelling and numerical simulation of coupled transport phenomena
with phase change: mixture evaporation from a rectangular capillary, Chemical Engineering Science (2017), doi:
https://doi.org/10.1016/j.ces.2017.12.023
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Modelling and numerical simulation of coupled
transport phenomena with phase change: mixture
evaporation from a rectangular capillary
Abstract
∗ Correspondingauthor
Email address: eugeny.kenig@upb.de (Eugeny Y. Kenig)
2
change (Haroun et al., 2010; Ganguli & Kenig, 2011; Marschall et al., 2012;
Bothe & Fleckenstein, 2013).
Bassano (2003), Burghoff & Kenig (2006) as well as Hayashi & Tomiyama (2011)
worked on the numerical simulation of the species transfer influence on the
35 movement of phase interfaces. They used different assumptions to simplify
their approaches, e.g. equal densities in both phases or negligible influence of
phase-change-related terms in the species transfer boundary conditions at the
interface. Also, Fleckenstein & Bothe (2015) considered species transfer and its
influence on the interface movement in two-phase flows. However, their model
40 depends on an heuristic correction factor to calculate species fluxes across the
interface.
Heat and species transfer with two-way coupling has only been considered by
few groups. Tanguy et al. (2007) and Sáenz et al. (2014) presented numerical
simulations including species transfer in non-isothermal systems. Their inves-
45 tigations of species transfer were limited to one of the two phases. Qin &
Grigoriev (2012) presented an approach for the simulation of momentum, heat
and species transport in two-phase systems with two-way coupling depending
on empirical coefficients. In our recent paper, we developed an approach free of
empirical or heuristic parameters and gave a first validation using several one-
50 dimensional cases (Rieks & Kenig, 2018). In the present work, the validation
of this approach is extended to two-dimensional problems in order to provide a
more solid validation.
Both the mathematical model and its numerical treatment are based on the
55 approach suggested by Rieks & Kenig (2018). Here, we only focus on the most
important aspects. Backgrounds for the approach are available in Sedov (1972),
Slattery (1972) and Leal (2007).
Our model is formulated for a two-phase Newtonian fluid system, with two non-
reacting components in each phase. Thermal viscous dissipation and thermo-
3
60 dynamic excess properties as well as external body forces (as, e.g., gravitation)
are neglected. Fluid properties are assumed to be constant within each phase.
The Volume-of-Fluid (VOF)-method is applied to capture the movement of the
phase interface. In this method, an indicator function αl is introduced, rep-
resenting the liquid volume fraction. αl is advected within the computational
65 domain using the following advection equation:
∂αl ṁ
+ ul · ∇αl = δ (1)
∂t ρl
Eq. 1 is formulated for phase-change systems (Fleckenstein & Bothe, 2015).
The quantity ṁ denotes the interfacial mass flux and δ is the Dirac δ-function
that indicates the interface position. The method presented by Rusche (2002)
is used to avoid numerical diffusion at the interface. The cells that are cut by
70 the sharp interface are identified using the function β defined as follows:
kΩk = Ωl − Ωv (3)
4
80 for Ω = {ρ, cp , λ, µ}. The one-field formulated continuity equation (Tryggvason
et al., 2011) is written as follows:
1 1
∇ · u = ṁ − δ (5)
ρv ρl
Momentum transport is described with Eq. 6 (Fleckenstein & Bothe, 2015):
∂(ρu)
+ ∇ · (ρuu) = ∇ · T + fσ δ (6)
∂t
where T denotes the stress tensor
2
T = µ[∇u + (∇u)T ] − (p + µ∇ · u)I (7)
3
and fσ = σκnint represents the surface tension force (Tryggvason et al., 2011).
85 Eq. 8 represents a one-field formulated heat transport equation (cf. Juric &
Tryggvason (1998) and Rattner & Garimella (2014)).
∂(ρcp T )
+ ∇ · (ρucp T ) + ∇ · q − Λṁδ + kṁcp T k δ
∂t
∂T ∂Teq
= βρcp ( − ) (8)
∂t ∂t
The heat flux vector is defined as q = −λ∇T . The terms on the right side of
Eq. 8 are used to establish thermal equilibrium at the interface: The (ṁδ)-field
is determined by iteratively adjusting its values until the right side of Eq. 8
90 approaches zero. This means that (ṁδ) is only non-zero in cells with β = 1. As
a result, the temperature T at the interface is fixed at Teq . The dependence of
the equilibrium temperature Teq on pressure and concentrations is determined
by equations common in vapour-liquid equilibrium treatment; the phase equilib-
rium is described based on the NRTL-model, which takes mixture non-idealities
95 into account (Poling et al., 2001). The corresponding equations can be found in
Appendix A.
The treatment of the species transfer is based on the approaches presented by
5
Fleckenstein & Bothe (2015) and Sáenz et al. (2014). We define an ”overlap-
ping” indicator function αk,ol for phase k:
100
αk,ol is 1 in cells fully or partially filled with phase k. Thus, αl,ol and αv,ol are
both equal to 1 in cells that contain the interface. The distribution coefficient
K = K(T, p, t) denotes the concentration ratio at the interface.
cl = K cv (10)
∂(αv,ol cv )
+ ∇ · (αv,ol cv usp,v ) + ṁsp δ = 0 (12)
∂t
110 Note that the algebraic sign of (ṁsp δ) differs in Eq. 11 and 12, because a species
is transported from one phase into another one, causing species loss in the first
and species gain in the second phase.
Here, the species transport equations are presented for a liquid and a vapour
phase. Clearly, gas-liquid and liquid-liquid systems can be considered with the
115 same set of equations as well.
The described equations were implemented into OpenFOAM
R and solved se-
quentially. Our code was developed based on the solver ”poreFoam” (Raeini
et al., 2012), with implemented VOF-method, continuity and momentum trans-
fer equations. The ”poreFoam” code can be accessed online (Raeini et al., 2016).
6
120 ”poreFoam” also includes filters to reduce the impact of spurious (i.e. unphys-
ical) currents. Compared to other front-capturing methods, the ”poreFoam”-
approach showed superior suppression of spurious currents, so that we chose it
for our simulations.
The phase-change-related source term in the advection equation, Eq. 1, is well
125 known to cause numerical instabilities. To solve this problem, Hardt & Won-
dra (2008) developed a useful numerical method in which (ṁδ) is modified using
smoothing, cutting and rescaling steps. The related equations were implemented
into an OpenFOAM
-solver
R (”evapVOFHardt”) that is also available online
(Kunkelmann, 2011, 2017).
130 We combined the ”poreFoam” code with our own implementation of heat and
species transfer. Further, to be able to capture the phase change, we applied the
code related to the numerical method by Hardt & Wondra (2008) from ”evap-
VOFHardt”. The structure of our OpenFOAM
simulation
R code is explained
in detail in Rieks & Kenig (2018).
7
Section 1).
150 It is worth noting that any reasonable experimental validation of the full model,
both two- and three-dimensional, would require measurements of locally resolved
temperature and concentration fields. However, such data can only be obtained
by non-invasive measurement techniques that are still under development, and
thus, relevant experimental data is not yet available.
8
0,10
0,09
0,08
0,07
bubble radius [mm]
0,06
0,05
0,04
0,03
analytical solution, superheat 1K
0,02 analytical solution, superheat 0.5K
numerical solution, superheat 1K
0,01
numerical solution, superheat 0.5K
0,00
0 0,001 0,002
time [s]
Fig. 1. Analytical (Scriven, 1959) and numerical solution for the process of bubble growth:
bubble radius as a function of time for pure water at 1atm with superheatings of 0.5K and
1K
Further on, close to the phase interface, the initial liquid-side temperature field
180 was smoothed, building a steep profile, so that the initial bubble growth rate
matched the analytically determined bubble growth rate in Scriven (1959). In
Fig. 1, the analytically and numerically obtained bubble radius is plotted as a
function of time.
The agreement between the analytical and numerical solution is satisfactory for
185 both superheating values. The indicator function, the temperature field and
pathlines at t = 0.001s are shown in Fig. 2 for a superheating of 1K.
It was observed that due to numerical errors, the bubble did not remain per-
fectly round during the simulation. The horizontal and the diagonal bubble
radius showed a deviation of about 0.5% from the average radius, which can
190 be considered as acceptable. As visible on the right side of Fig. 2, the bubble
shape remained round.
Evaporation goes along with a significant volume change due to the density dif-
ference between liquid and vapour phase. This volume change leads to a flow
within the liquid phase, as it is clear from the pathlines in Fig. 2. Further, Fig.
9
Fig. 2. Numerical solution for the process of bubble growth: indicator function (on the right),
temperature field and pathlines (on the left) at t = 0.001s for 1K superheating
195 2 shows the radial temperature profile, which develops due to heat transport
through the liquid to the phase interface.
Both cases were simulated using a rotationally symmetric 2D-grid. The spatial
210 resolution of the grid close to the phase interface was ∆x = 1.5·10−5 m for which
grid independence was reached. The temporal resolution of the simulations was
∆t = 1 · 10−3 s.
10
Table 1
Parameters for validation cases with diffusive species transfer of acetone between an aqueous
and a toluene phase (Engberg, 2016)
interface
0,6
0,4
0,2
0,0
0,0 0,5 1,0 1,5 2,0
position [mm]
Fig. 3. Concentration of acetone diffusing across a curved phase interface at t = 10s for
case 1, plotted along the radial coordinate of the droplet; results were determined using our
approach and the method of lines (Engberg, 2016)
For the two considered cases, Engberg (2016) presented an exact numerical
solution, determined by the method of lines. We used the results of Engberg
215 (2016) as a reference solution for the validation of our simulations. Both our
results and his solutions are shown in Fig. 3 to 6, and a fully satisfactory
agreement between them is established for both species transport directions,
which means that the transfer of acetone is captured very well.
Fig. 7 shows the two-dimensional concentration fields for case 2 after 10s and
220 20s. It can be seen that the simulated concentration fields are rotationally
symmetric. This confirms that no significant simulation errors arised because
11
1,0 method of lines
(Engberg, 2016)
this work
concentration [kg/m³] 0,8 interface
0,6
0,4
0,2
0,0
0,0 0,5 1,0 1,5 2,0
position [mm]
Fig. 4. Concentration of acetone diffusing across a curved phase interface at t = 20s for
case 1, plotted along the radial coordinate of the droplet; results were determined using our
approach and the method of lines (Engberg, 2016)
0,8 interface
concentration [kg/m³]
0,6
0,4
0,2
0,0
0,0 0,5 1,0 1,5 2,0
position [mm]
Fig. 5. Concentration of acetone diffusing across a curved phase interface at t = 10s for
case 2, plotted along the radial coordinate of the droplet; results were determined using our
approach and the method of lines (Engberg, 2016)
12
1,0 method of lines
(Engberg, 2016)
this work
0,8 interface
concentration [kg/m³]
0,6
0,4
0,2
0,0
0,0 0,5 1,0 1,5 2,0
position [mm]
Fig. 6. Concentration of acetone diffusing across a curved phase interface at t = 20s for
case 2, plotted along the radial coordinate of the droplet; results were determined using our
approach and the method of lines (Engberg, 2016)
13
Fig. 7. Concentration field of acetone diffusing across a curved water/toluene interface at
t = 10s and t = 20s for case 2, determined using the approach presented in this paper
14
Fig. 8. Simulation results for the species transfer of dissolved acetic acid out of a steadily
sinking water droplet in cyclohexane; concentration fields at t = 5s and t = 10s determined
with the model by Haroun et al. (2010) (a) and our approach (b)
1 This assumption is reasonable, since only small movements of the interface were consid-
ered.
15
Table 2
Varied parameters of the case study on mixture evaporation: contact angle Θ and the super-
heating at the beginning of the simulation ∆T0
270 superheating of the wall at the beginning of the simulation (∆T0 ). The varied
parameters of the three case studies are given in Table 2.
The simulations discussed in sections 3.1 and 3.2 as well as the one-dimensional
simulations presented in Rieks & Kenig (2018) were performed with relatively
fine grids which allowed grid-independent solutions to be obtained. For the
275 simulations described in this section, a time step of ∆t = 1·10−5 s was necessary
to avoid spurious currents. The physical process time to simulate was about 5s,
so that a significant number of time steps were necessary. Since the number of
equations to be solved to capture two-way coupled momentum, heat and species
transport was also substantial, it was only possible to use coarse grids in order
280 to avoid excessive simulation times. The grid we used consisted of square cells.
In order to evaluate the influence of the grid, we simulated case 1 with different
∆x. Fig. 9 shows the mass fraction profile of ethanol for case 1 at t = 2s as a
function of the distance to the heated wall along the centre vertical line.
It can be seen that a grid with ∆x = 2.5 · 10−5 m, the used resolution for the
285 simulations presented here, was not fine enough to reach fully grid-independent
solutions. Consequently, the following simulation results cannot be considered
as fully accurate. However, they appear to be accurate enough to describe the
essential features of the complex interplay of coupled transport phenomena in
binary two-phase systems. The simulation results for case 1 (25◦ contact angle,
290 4K superheating) are shown in Fig. 10 to 14 as well as in the supplementary
Video 1 in Appendix B.
As it is visible in Fig. 11, heat is transported from the bottom to the liquid-
vapour interface. At the interface, liquid evaporates. Because of the curved
16
0,85
0,80
0,75
we [-]
0,60
0,55
0,0000 0,0005 0,0010 0,0015 0,0020
distance from the heated wall [m]
Fig. 9. The influence of the grid: mass fraction of ethanol along the centre vertical line as a
function of the distance from the heated wall at t = 2s; Θ = 25◦ , ∆T0 = 4K
Fig. 10. Simulation results for the evaporation of ethanol/water from a rectangular capillary:
indicator function α after 0s, 2.5s, 5s and 7.5s; Θ = 25◦ , ∆T0 = 4K (case 1)
17
Fig. 11. Simulation results for the evaporation of ethanol/water from a rectangular capillary:
temperature after 0s, 2.5s, 5s and 7.5s; Θ = 25◦ , ∆T0 = 4K (case 1)
Fig. 12. Simulation results for the evaporation of ethanol/water from a rectangular capillary:
mass fraction of ethanol after 0s, 2.5s, 5s and 7.5s; Θ = 25◦ , ∆T0 = 4K (case 1)
Fig. 13. Simulation results for the evaporation of ethanol/water from a rectangular capillary:
magnitude of the flow velocity after 0s, 2.5s, 5s and 7.5s; Θ = 25◦ , ∆T0 = 4K (case 1)
18
Fig. 14. Simulation results for the evaporation of ethanol/water from a rectangular capillary:
pathlines for t = 3s; Θ = 25◦ , ∆T0 = 4K (case 1)
meniscus, the distance from the interface to the heated wall is a function of the
295 horizontal position. That is why the temperature gradient between the wall and
the interface is highest in the area close to the symmetry axis. Thus, the amount
of liquid that evaporates is at its maximum at the lowest point of the meniscus.
Consequently, at this point, the mass fraction of ethanol (we ) decreases most
significantly (Fig. 12). As ethanol represents the low boiling component of the
300 considered mixture, decreasing we at the interface results in a slow rise of the
interfacial temperature, cf. Fig. 11. Again, this process is most pronounced at
the lowest point of the meniscus. As shown in Fig. 10, the liquid level within
the capillary stays nearly constant. However, in Video 1 in Appendix B, the
small movement of the interface due to evaporation is clearly visible.
305 Fig. 14 shows pathlines for t = 3s for both phases. The flow is caused by the
significant volume change during evaporation. In the vapour phase, the flow
velocity is highest close to the symmetry plane, due to two reasons. First, the
evaporation is strongest at the lowest point of the meniscus, and second, the wall
friction causes low flow velocities close to the wall. The pathlines are approxi-
310 mately perpendicular to the interface at the lowest point of the meniscus, what
is typical for evaporation. Closer to the wall, the flow is significantly influenced
by friction, and thus, the trajectories are not perpendicular to the interface.
A slow flow within the liquid phase evolves due to evaporation with a constant
19
Fig. 15. Simulation results for the evaporation of ethanol/water from a rectangular capillary:
temperature after 0s, 2.5s, 5s and 7.5s; Θ = 25◦ , ∆T0 = 2K (case 2)
contact angle at the wall. Because the phase change occurs mostly at the lowest
315 point of the meniscus, the latter tends to move fastest at the symmetry plane.
At the same time, the surface tension is significant and forces the meniscus
to keep its round shape. Consequently, the meniscus sinks uniformly, i.e. the
surface of the liquid maintains a similar shape. This is possible if liquid flows
towards the area with strongest evaporation. This flow can be seen in Fig. 14.
320 Fig. 13 shows the temporal development of the flow velocity magnitude in both
phases. At the beginning of the simulation, heat is transported through the
liquid to the interface. Thus, evaporation starts, and the average flow veloc-
ity within the vapour phase rises quickly. However, since the temperature at
the interface also rises due to evaporation of ethanol (cf. Fig. 11 and 12),
325 the temperature difference between the heated wall and the meniscus decreases.
Consequently, the flow velocity also reduces with time.
When simulating flows that are significantly influenced by capillary forces, un-
physical (spurious) currents represent a well-known problem. These currents
typically appear as random, chaotic and fluctuating flows close to the phase
330 interface. Fig. 13, Fig. 14 and Video 1 in Appendix B show that with the help
of the method by Raeini et al. (2012), spurious currents could be sufficiently
avoided in our simulations, as the flow is not disturbed by this type of influence.
We now take the results of the simulation with Θ = 25◦ and ∆T0 = 4K (case
1) as a reference for the further studies.
20
Fig. 16. Simulation results for the evaporation of ethanol/water from a rectangular capillary:
mass fraction of ethanol after 0s, 2.5s, 5s and 7.5s; Θ = 25◦ , ∆T0 = 2K (case 2)
Fig. 17. Simulation results for the evaporation of ethanol/water from a rectangular capillary:
magnitude of the flow velocity after 0s, 2.5s, 5s and 7.5s; Θ = 25◦ , ∆T0 = 2K (case 2)
21
Fig. 18. Simulation results for the evaporation of ethanol/water from a rectangular capillary:
indicator function α after 0s, 2.5s, 5s and 7.5s; Θ = 60◦ , ∆T0 = 4K (case 3)
335 The simulation results for case 2 (Θ = 25◦ and ∆T0 = 2K) are shown in Fig. 15
to 17 and Video 2 in Appendix B. Here, the wall superheating is smaller than in
the reference case. Consequently, as visible in Fig. 15, less heat is transported
to the interface. Thus, the evaporation is slower, resulting in lower flow veloc-
ities (Fig. 17) and a less significant change of the mass fraction of ethanol in
340 the region close to the interface (Fig. 16). However, the simulation for case 2
shows qualitatively similar results as those for case 1. For instance, the shape of
the meniscus is the same as in case 1, and hence, evaporation is still strongest
close to the symmetry axis. This results in rising temperatures (Fig. 15) and
decreasing ethanol mass fractions (Fig. 16) that are most pronounced in this
345 region. Consequently, the flow velocity shows the highest values close to the
symmetry axis (Fig. 17). Further, the temporal maximum in the flow velocity
magnitude develops for the same reasons as described for case 12 . This maxi-
mum is, however, less significant than in case 1. Because the change of the mass
fraction at the interface is smaller, the rise in the interface temperature is also
350 smaller, and consequently, there is only a marginal change of the evaporating
mass flux between 2.5s and 7.5s.
2 The field of the indicator function is not shown here, because it hardly changes and is
almost identical in cases 1 and 2. The phase distribution for case 2 can be seen in Video 2 in
Appendix B.
22
Fig. 19. Simulation results for the evaporation of ethanol/water from a rectangular capillary:
temperature after 0s, 2.5s, 5s and 7.5s; Θ = 60◦ , ∆T0 = 4K (case 3)
Fig. 20. Simulation results for the evaporation of ethanol/water from a rectangular capillary:
mass fraction of ethanol after 0s, 2.5s, 5s and 7.5s; Θ = 60◦ , ∆T0 = 4K (case 3)
Fig. 21. Simulation results for the evaporation of ethanol/water from a rectangular capillary:
magnitude of the flow velocity after 0s, 2.5s, 5s and 7.5s; Θ = 60◦ , ∆T0 = 4K (case 3)
23
Finally, simulation case 3 is analysed. Fig. 18 to 21 and Video 3 in Appendix
B show the simulation results for Θ = 60◦ and ∆T0 = 4K. In simulation
case 3, the contact angle at the wall is higher than in case 1, resulting in a
355 lower curvature of the meniscus (Fig. 18). Consequently, the distance between
the heated wall and the interface is more uniform than in the reference case.
Thus, the temperature field (Fig. 19) and the ethanol mass fraction field (Fig.
20) along the interface (from left to right) are also more uniform than in the
reference case, due to a more evenly distributed evaporation. The same is true
360 for the flow velocities (Fig. 21), but still, the flow is fastest in the centre of the
domain because of the high friction at the wall.
4. Conclusions
24
to a wide variety of process engineering problems. A major novelty of our ap-
proach is that it considers all three transport phenomena in a fully coupled way,
i.e., the two-way coupling of momentum and heat transfer, of momentum and
species transfer as well as of heat and species transfer is taken into account. This
385 corresponds to a number of real systems with phase change, e.g. in evaporation,
condensation and distillation.
It should be mentioned that the wetting behaviour in the mixture evaporation
simulations was modelled with a static contact angle. A more realistic approach
for this boundary condition would be desirable, e.g. the use of a dynamic con-
390 tact angle. Furthermore, the computational effort to solve the full system of
transport equations and boundary conditions in the presented form is quite sig-
nificant even for two-dimensional problems. Much higher computing capacities
are necessary for 3D simulations and refined discretisation. This appears to be
the task for the future.
395 5. Acknowledgement
Equilibrium data for the system ethanol/water (e/w) were obtained using
400 the following equations (Poling et al., 2001): the Antoine equation (Eq. A.1),
Dalton’s law (Eq. A.2), the equilibrium relation (Eq. A.3) and the NRTL-
model-related equations (Eq. A.4 to Eq. A.7).
B
log10 (p0,sp ) = A − (A.1)
C +T
25
2
x2w τew exp (−aew τew )
exp (−awe τwe )
ln (γe ) = x2w τwe + (A.4)
xe + xw exp (−awe τwe ) (xw + xe exp (−aew τew ))2
2
x2e τwe exp (−awe τwe )
exp (−aew τew )
ln (γw ) = x2e τew + (A.5)
xw + xe exp (−aew τew ) (xe + xw exp (−awe τwe ))2
∆gew
τew = (A.6)
RT
∆gwe
τwe = (A.7)
RT
In order to obtain the distribution coefficient K (Eq. 10), which represents
a field variable, it is necessary to determine the ratio of the concentrations of
405 ethanol (cl,e /cv,e ) at the interface. ck,e is determined by the following equation:
wk,e
ρk,e
ϕk,e = wk,e wk,w (A.9)
ρk,e + ρk,w
3 As described in section 2, we used constant physical properties within each phase for
the solution of all relevant transport equations. However, assuming constant phase densities
for the determination of the distribution coefficient K may be rather inaccurate, because
cl,e ρl wl,e
K= cv,e
= ρv wv,e
, while both ρk are, in general, influenced by concentration changes. This
is why we consider the relationship of density on component mass fractions here.
26
Table Appendix A.1
Fluid properties of ethanol and water (Linstrom & Mallard, 2016; Kretzschmar, 2016); diffu-
sivities were estimated with methods described by Poling et al. (2001)
R
α w
V R k k,sp
dV
w̄k,sp = (A.11)
V
αk dV
R
α ϕ
V R k k,sp
dV
ϕ̄k,sp = (A.12)
V
αk dV
in which V denotes the total volume of the considered domain. The fluid prop-
erties of pure water and ethanol, the parameters of the NRTL-model and the
parameters of the Antoine equation are collected in Tables Appendix A.1 to
N
Appendix A.3. For surface tension, a constant value σ = 0.04 m is used.
27
Table Appendix A.2
Parameters of the Antoine equation (A, B, C) for ethanol and water (Dortmund Data Bank
Software & Separation Technology GmbH, 2016); the parameters refer to Eq. A.1 with T in
◦C and p0,sp in mmHg.
28
420 Appendix B. Supplementary materials
The supplementary videos to this article are to be found under the link: DOI
Nomenclature
Latin letters
a NRTL-parameter (−)
A parameter of the Antoine equation (−)
B parameter of the Antoine equation (−)
c mass concentration of the light boiling component (kg m−3 )
C parameter of the Antoine equation (−)
cp specific isobaric heat capacity (J kg −1 K −1 )
D diffusivity (m2 s−1 )
fσ surface tension force (P a)
g specific free enthalpy (J mol−1 )
I identity matrix (−)
j diffusion flux vector (kg m−2 s−1 )
K distribution coefficient (−)
ṁ total mass flux across the interface (kg m−2 s−1 )
ṁsp species flux across the interface (kg m−2 s−1 )
M molar mass (kg mol−1 )
n normal vector (−)
p pressure (P a)
p0,sp pure component vapour pressure (P a)
psp partial pressure (P a)
q heat flux vector (W m−2 )
R ideal gas constant (J mol−1 K −1 )
t time (s)
T temperature (K)
T stress tensor (P a)
29
u velocity vector (m s−1 )
V volume (m3 )
w mass fraction (−)
x liquid mole fraction (−)
x position (m)
y vapour mole fraction (−)
Greek letters
425 Subscripts
0 initial condition
e ethanol
eq equilibrium
int interface
k arbitrary phase
l liquid
30
ol overlap
sp species
v vapour
w water
Superscripts
Ω̄ phase averaged Ω
References
Burghoff, S., & Kenig, E. Y. (2006). A CFD model for mass transfer and
435 interfacial phenomena on single droplets. AIChE J., 52 , 4071–4078. doi:10.
1002/aic.11038.
Dortmund Data Bank Software & Separation Technology GmbH (2016). Dort-
mund Data Bank. www.ddbst.com. Accessed: 2016.
Engberg, R. F., Wegener, M., & Kenig, E. Y. (2014). The impact of Marangoni
440 convection on fluid dynamics and mass transfer at deformable single rising
droplets - a numerical study. Chem. Eng. Sci., 116 , 208–222. doi:10.1016/
j.ces.2014.04.023.
31
Fleckenstein, S., & Bothe, D. (2015). A volume-of-fluid-based numerical method
for multi-component mass transfer with local volume changes. J. Comput.
Phys., 301 , 35–58. doi:10.1016/j.jcp.2015.08.011.
Hardt, S., & Wondra, F. (2008). Evaporation model for interfacial flows based
on a continuum-field representation of the source terms. J. Comput. Phys.,
227 , 5871–5895. doi:10.1016/j.jcp.2008.02.020.
455 Haroun, Y., Legendre, D., & Raynal, L. (2010). Direct numerial simulation of
reactive absorption in gas-liquid flow on structured packing using interface
capturing method. Chem. Eng. Sci., 65 , 351–356. doi:10.1016/j.cse.2009.
07.018.
Kunkelmann, C., & Stephan, P. (2009). CFD simulation of boiling flows using
the volume-of-fluid method within OpenFOAM. Num. Heat Transfer, Part
A, 56 , 631–646. doi:10.1080/10407780903423908.
32
Kurihara, K., Nakamichi, M., & Kojima, K. (1993). Isobaric vapor-liquid equi-
475 libria for methanol + ethanol + water and the three constituent binary sys-
tems. J. Chem. Eng. Data, 38 , 446–449. doi:10.1021/je00011a031.
Marek, R., & Straub, J. (2001). Analysis of the evaporation coefficient and
the condensation coefficient of water. Int. J. Heat Mass Transfer , 44 , 39–53.
485 doi:10.1016/S0017-9310(00)00086-7.
Marschall, H., Hinterberger, K., Schüler, C., Habla, F., & Hinrichsen, O. (2012).
Numerical simulation of species mass transfer across fluid interfaces in free-
surface flows using OpenFOAM. Chem. Eng. Sci., 78 , 111–127. doi:10.1016/
j.ces.2012.02.034.
490 Poling, B. E., Prausnitz, J. M., & O’Connell, J. (2001). The properties of gases
and liquids. McGraw-Hill.
495 Raeini, A. Q., Bijeljic, B., & Blunt, M. J. (2016). poreFoam-code for Open-
FOAM. URL: http://www.imperial.ac.uk/earth-science/research/
research-groups/perm/research/pore-scale-modelling/software/
direct-two-phase-flow-solver/.
Raeini, A. Q., Blunt, M. J., & Bijeljic, B. (2012). Modelling two-phase flow in
500 porous media at the pore scale using the volume-of-fluid method. J. Comput.
Phys., 231 , 5653–5668. doi:10.1016/j.jcp.2012.04.011.
33
Rattner, A. S., & Garimella, S. (2014). Simple mechanically consistent formu-
lation for volume-of-fluid based computations of condensing flows. J. Heat
Transfer , 136 , 071501–1–071501–9. doi:10.1115/1.4026808.
505 Rieks, S., & Kenig, E. Y. (2018). Modelling and numerical simulation of cou-
pled transport phenomena with phase change: layer evaporation of a binary
mixture. Chem. Eng. Sci., 176 , 367–376. doi:10.1016/j.ces.2017.10.040.
510 Sáenz, P. J., Valluri, P., Sefiane, K., Karapetsas, G., & Matar, O. K. (2014). On
phase change in Marangoni-driven flows and its effects on the hydrothermal-
wave instabilities. Phys. Fluids, 26 , 024114–1–024114–27. doi:10.1063/1.
4866770.
520 Tanguy, S., Ménard, T., & Berlemont, A. (2007). A level set method for vapor-
izing two-phase flows. J. Comput. Phys., 221 , 837–853. doi:10.1016/j.jcp.
2006.07.003.
Tryggvason, G., Scardovelli, R., & Zaleski, S. (2011). Direct numerical simula-
tions of gas-liquid multiphase flows. Cambridge University Press, New York.
525 Welch, S. W. J., & Wilson, J. (2000). A volume of fluid based method for
fluid flows with phase change. J. Comput. Phys., 160 , 662–682. doi:10.1006/
jcph.2000.6481.
34
Highlights
Both heat and species transfer at the phase interface are captured