Вы находитесь на странице: 1из 10

Microgravity Sci. Technol.

(2010) 22:305314
DOI 10.1007/s12217-010-9187-8

ORIGINAL ARTICLE

A SLICVOF Method Based on Unstructured Grid


Meng Huang Bin Chen Lilong Wu

Received: 29 November 2009 / Accepted: 10 March 2010 / Published online: 31 March 2010
Springer Science+Business Media B.V. 2010

Abstract In this paper, an interface reconstruction


technology similar with SLIC method on structured
grid is developed for the unstructured grid. For a cell
containing the interface, a straight line segment parallel
with one edge of the triangle cell is set to the interface
according to the VOF function values in the three
neighbor cells and volume flux of the particular fluid
is determined by the geometry relation between the
interface line and volume flux on each edge. Finally,
four classical cases are used to test this unstructured
SLICVOF interface reconstruction method including
advection test of a right-angled triangle and a hollowed
circle, Zalesak slotted disk rotation test as a test of
scalar advection, and single-vortex shearing flow test,
which are effective to test interface capturing method.
It was found that the interface transported back by the
reversed velocity field is comparative with SLICVOF
on structure grid. Through the comparison between
two triangle grids generated by Delaunay Triangulation
Method and Bubble Packing Method, it is found that
the grid quality has great effect on the proposed unstructured SLICVOF method. More regular triangle
cells will lead to better interface capturing result.
Keywords VOF SLIC Unstructured grid
Two-phase flow

M. Huang B. Chen (B) L. Wu


State Key Laboratory of Multiphase Flow
in Power Engineering, Xian Jiaotong University,
Xian, 710049, China
e-mail: chenbin@mail.xjtu.edu.cn

Introduction
With the development of space technology for human
space exploration, correct understanding of multiphase
flow especially gas-liquid two-phase flow phenomena
under microgravity is being increasingly essential for
many applications in thermal management, propulsion systems, solid waste management, water recovery,
Environmental Control and Life Support Systems
(Suol et al. 2009). Experimental investigations of twophase flow under microgravity are extremely expensive
because they require special facilities such as drop
tower, zero-g aircraft, returnable satellite, space shuttle
or space station. By contrast, numerical simulation is a
superior tool to reduce the expenditure and has been
the subject of a vast body of research. In space environment, gasliquid two-phase flow sometime occurs
in complicated geometries. For example, the propellant
tank in low gravity involves liquid-vapor two-phase
sloshing flow in a cylinder with spherical end caps
(Chato and Martin 2006). Such problems are difficult to
be numerically simulated by the conventional methods
based on structured grid and new methodology based
on unstructured grid should be addressed, which will
be helpful to reveal the function mechanism of surface
tension, thermocapillary convection, etc.
In the numerical simulation of two immiscible fluids,
an important issue is how to track the motion of the free
interface. The difficulties arise from the discontinuities
of pressure and velocity on the sharp moving interface as well as the presence of unsteady deformation,
breakup and coalescence. Over the last several decades,
mainly two groups of approaches were proposed to
compute the two-phase flow problems with moving
interface: interface tracking (front tracking method)

306

Microgravity Sci. Technol. (2010) 22:305314

Fig. 1 Line segment


approximation of the actual
interface: a K > 0 and
bK<0

interface
interface

F1

F1

F3

and interface capturing methods (level set and VOF


methods). Each method has its own advantages as well
as drawbacks, and no method is applicable to the wide
range of possible two-phase flow phenomena. In the
front tracking method (Unverdi and Tryggvason 1992;
Tryggvason et al. 2001), an Eulerian grid is used to solve
the NavierStokes equation for the two immiscible
fluids and the interface is explicitly represented by the
Lagrangian tracking and redistribution of a number of
marker points. This method can calculate curvature in
high accuracy and consequently preserve the sharpness
of the interface. However, it cannot easily handle significant topology changes of the interface, for example
breakup or coalescence of bubbles or droplets.
Nevertheless, capturing method can deal with topological change of the interface easily. Level set method
(Osher and Sethian 1988; Sethian and Smereka 2003)
uses distance function to implicitly capture the interface
efficiently. It has advantages in computation of curvature but disadvantages in mass conservation. As an alternative, VOF (volume-of-fluid) method defines a volume fraction function F to track the interface implicitly.
Once obtain F in each cell by the advection of flow
field, the interface can be reconstructed by different
approximation schemes: SLICVOF (simple line interface calculation) scheme (Noh and Woodward 1976),
SOLAVOF (Hirt and Nichols 1981), PLIC (Piecewise
Linear Interface Calculation) scheme (Youngs 1982),
and other higher order differencing schemes. Although
the three-dimensional implementation is not so easy,
VOF is still a popular method to track interface because
it can preserve mass in a natural way.
The methods mentioned above originally resolve
fluid motion by a structured grid and the applications
are therefore restrained to simple domain. For geometrically complex domain, body-fitted grid is one choice
but it is usually limited to relatively simple geometries.
Methods based on unstructured grid with triangular
(2D) and tetrahedral (3D) cells are highly flexible to

F2

F3

F2

adapt the domain boundaries and the computational


expense could be saved through the local refinement
to a certain extent. Recently, there is a growing interest to develop interface tracking/capturing methods
on unstructured grid. Gloth et al. (2003) presented a
front tracking method on unstructured grid with a local
level set formulation for the topological treatment of
fronts. Although the results of isolated discontinuities
have not been impaired, the concept of this method
is not being rigorously conservative. Unstructured grid
level set methods proposed mostly based on finite element methods, except Herrmann (2008) proposed a
balanced force refined level set grid method, which
solves the NavierStokes equations on an unstructured
grid and track the interface location by a refined level
set method on an auxiliary, high-resolution, equidistant
Cartesian grid.
As to the VOF method, Gao (1999) developed
a finite elementVOF method for unstructured grid.
Zhao et al. (2002) discretized the VOF equation
by a high-order characteristics-based finite volume
scheme and a matrix-free implicit dual time-stepping
scheme utilizing fully unstructured (triangular) meshes.
Shahbazi et al. (2003) developed a second order
accurate a piecewise linear remapping volume tracking
algorithm on triangular meshes. Ji et al. (2005) validated the similar LagrangianEulerian Re-map (MLER)
method by single vortex test and complex deformation

Fig. 2 Decomposition of
velocity vector

Microgravity Sci. Technol. (2010) 22:305314

307

Fig. 3 Different situation for the calculation of volume flux in a triangular cell: a case 1, b case 2, c case 3, and d case 4

field test. This method is not easy to implement because


it contains a Lagrangian phase, a reconstruction phase
and a remapping phase. Yang et al. (2006) developed
analytical relations for reconstructing piecewise linear interfaces and volume fractions in triangular and
tetrahedral grids, i.e., PLIC method in triangular grids.
They proposed an adaptive coupled level set and
volume-of-fluid method for 2D problems on unstructured triangular grids. Wang et al. (2009) also developed a coupled level set and volume-of-fluid method
for sharp interface simulation of plunging breaking
waves, in which the interface is reconstructed via a
VOFPLIC scheme and the Level set function is redistanced based on the reconstructed interface.
To date, it will be of great theoretical importance and
practical value to develop efficient VOF method based
on the unstructured grid with high accuracy to capture
the interface in geometrically complex domain. In this
paper, we will develop an interface re-construction
technology on unstructured grid similar with structured
SLIC method. For a cell containing the interface (also
called a mixed cell), a straight line segment parallel with
one edge of the triangle cell is set to the approximation

interface according to the volume fraction of the liquid


in the three neighboring cells. Volume flux is determined by the geometry relation between the interface
line and the normal velocity on each edge.
Numerical Algorithm
Discretization of VOF Function
The VOF function f is a continuous function defined in
the calculation domain and it denotes the exact distribution of volume fraction for the two fluids. The volume
fraction F, an integrally averaged quantity, is defined as
the portion of the area (2D) or volume (3D) of the cell
filled with the specific phase fluid. Therefore, all cells
can be classified into three types: full with specific phase
(F = 1), empty with specific phase (i.e., full with the
other phase, F = 0), and interfacial cell (0 < F < 1). In
2D, the continuous VOF function f and the discretized
volume fraction F are related as follows

1
Fi =
f (x, y, t) ds
(1)
Si
Si

Fig. 4 Four test cases: a advection test of a right triangle, b advection test of a hollowed circle, c Zalesak slotted disk rotation test, and
d single-vortex shearing flow test

308

Microgravity Sci. Technol. (2010) 22:305314

where Si is the area of the ith cell. For a given flow
field u, the evolution of VOF function f is governed
by following advection equation because this function
must be conserved as it is convected with the fluid
f
+ (u ) f = 0
t

(2)

This equation can be recast in integration as follows:


f ds +
S

3




u nk f dl = 0

k = 1, 2, 3

(3)

k=1

where k denotes the each edge of the triangular cell i, l


is the edge length, and nk is the normal vector on edge
k. Here non-staggered grid is used, i.e., all variables are
stored in the center of the triangular cell. Assuming
u nk = uk , we can get the following equation by the
temporal discretization



3 




n+1
n
f dS +
f uk dt dl = 0 (4)
f dS
Si

k=1

Si

lk

Substituting Eq. 1 into Eq. 4, we obtain


Fin+1 Fin =

3

1 
f uk dt dl
Si

(5)

k=1

As a result, the change of volume fraction F in current


triangular cell depends on the solution of the RHS
terms, which can be explained as the mass transportation rate through each boundary. They can be solved if
the mass transportations between each cell are worked
out.
Fig. 5 Unstructured grid:
a BPM (7,600 cells) and b
DTM (7,600 cells)

Line Segment Approximation of the Actual Interface


After advection of volume fraction, interface reconstruction techniques are essential to reconstruct an
interface approximately through the given volume fractions in every grid cell inversely. Similar with SLIC
VOF, we use a sequence of straight line segments
aligned with the grid to approximate the interface inside the triangular cell in geometric way. Then the
problem will be how to locate the line segment in each
cell.
Since the volume fraction of each cell is known at
the beginning of every time step, we could judge the
position of the line segment in a cell according to the
volume fraction of neighboring cells and that of itself.
Consider an arbitrary triangular grid cell of a given
volume fraction F and assume the volume fractions in
its three neighboring cells satisfy F1 > F2 > F3, we
can locate the interface and specific phase location
through the judgment of the following criteria:
K = (F1 F2) (F2 F3) = F1 2F2 + F3

(6)

Typically, there will be two cases corresponding to K >


0 or K < 0 shown in Fig. 1, where the dash and solid
lines inside the central interface cell are the actual and
approximation interface, respectively. Shadow region
is the specific phase fluid, while the region filled with
cross lines is the other phase fluid. Figure 1a shows
the case K > 0, which means F1 is larger than F2
and F3, so the specific phase inside the central cell
should be near the F1 cell and we can arrange the
line segment (approximation interface) parallel with
the common boundary between the central and F1 cell.
Figure 1b shows the other case K < 0, which means

Microgravity Sci. Technol. (2010) 22:305314

309

F3 is smaller than F1 and F2, this time the specific


phase inside the central cell should be away from the F3
cell and we can arrange the line segment parallel with
common boundary between the central and F3 cell.
The exact location of this line segment will be decided
by the current volume fraction F (area ratio) in the
central cell.
Calculation of Volume Flux
Once we know the location of the line segment for the
approximation of the actual interface, it is necessary
to compute volume fluxes by geometric calculation of
the new volume fraction F n+1 which can be obtained
by solving Eq. 5 for each cell edge by edge in a geometric way. As show in Fig. 2, firstly we decompose the
velocity vector into three components perpendicular to
each edge to get uk , and it is positive when points to the
outside of the cell. We only calculate positive volume
flux for current cell so as to ensure that the volume
flux of the specific phase through every edge will be
considered and none will be reconsidered. For a cell
fully filled with specific phase fluid (F = 1), the volume
flux can be computed directly without the consideration
of the interface. However, for those cells containing
Fig. 6 Advection test of a
right triangle with velocity
field (1, 1)

BPM

T/2

interfaces, it is necessary to take the interrelation between volume flux and interface location into account.
Figure 3 illustrates four different situations by the ubiety of the fluid distribution and the current edge lk .
We intercept a length of uk dt from edge lk to get a
net region (filled with cross pattern) in each situation,
then the area of the this region will be the volume flux
through edge lk to outside. If a cell is fully filled with the
specific phase fluid, the volume flux can be calculated in
the way similar with Fig. 2a. If the cell is empty, nothing
shall be done. Substituting the calculated volume fluxes
on each edge into Eq. 5, the new volume fraction F n+1
can be solved for the reconstruction of interface in the
next time step. Repeat this process, the time-dependent
reconstruction of interface can be accomplished for
every time step.
Implementation of Surface Tension
For incompressible two-phase flow with surface tension
effect under microgravity, the governing equations are
as follows:
u
=0
x

(7)

DTM

Structured

310

Microgravity Sci. Technol. (2010) 22:305314

u
+ i u u = P + i 2 u + F s
t

(8)

F s = n

(9)

where the subscript i denotes gas or liquid phase, F s is


the surface tension force which is non-zero only at the
location of the phase interface, is the surface tension
coefficient, is the delta-function and n is the local
interface normal vector. The local surface curvature
can be calculated by (Zhao et al. 2002):


f
(10)
= n =
| f |
The above equations are closed with the following constitutive relations for the density and dynamic viscosity:
= fl + (1 f ) g

(11)

= f l + (1 f ) g

(12)

In order to implement surface tension, the Continuum


Surface Force (CSF) method (Brackbill et al. 1992) can
be used to convert the interfacial surface force into a
corresponding volume force, which can be included in
the discretized NS equations as follows:
F s = n = (x) 

(x) 2 (x)


l g l + g

Fig. 7 Advection test of a


hollowed circle with velocity
field (1, 1)

(13)

BPM

T/2

Table 1 Error measurement of advection test (case 1)


Grid

Cell number

Er

Eg

Em

Structured
BPM
DTM

64 64
7,600
7,600

7.33E-02
8.66E-02
1.25E-01

3.15E-03
3.97E-03
5.61E-03

6.86E-04
2.25E-03
6.48E-02

Once the volume fraction F is convected by Eq. 5 and


the curvature is estimated by Eq. 10, the interface can
be easily captured based on the above NS equations
which will be solved by high-resolution finite volume
solver on unstructured collocated grid (Kim and Choi
2000). Together with unstructured grid generation technology, it will be a promising tool to simulate surface
tension or thermocapillary convection effect under microgravity in complicated geometry.
Validation of the Algorithm
As shown in Fig. 4, four classical cases are tested to
confirm the validation of our method to capture the
fluid interface. Figure 4a and b are a right-angled triangle and a concentric ring in advection flow, respectively.
Figure 4c is a Zalesak model in rotating flow and
Fig. 4d is a circle in shearing flow. Those test cases are
usually applied together to show a relatively complete

DTM

Structured

Microgravity Sci. Technol. (2010) 22:305314

311

Table 2 Error measurement of advection test (case 2)


Grid

Cell number

Er

Eg

Em

Structured
BPM
DTM

64 64
7,600
7,600

2.55E-01
9.21E-02
9.63E-02

2.41E-02
8.68E-03
9.05E-03

2.62E-04
5.00E-03
3.07E-02

estimate to an interface capturing method. The first two


cases are used to test the capability of the method in
advection flow, the third one is to test that in rotating
flow and the last one in shearing flow.
The cases are described in the following four illustrations. The calculation domain is a 1 1 square
and the values of f denote the distribution of the
fluids. We operate a reverse process after a forward
process with the same steps and check the deformation
by the comparison of the final and the beginning fluid
distribution, so that the effect of the method can be
estimated.
The velocity fields are set for test a, b, c and d as
follows

u = 1, v = 1

u = 1, v = 1
u = (y 0.5) , v = (x 0.5)
(14)

u
=

cos

0.5)
sin

0.5)
,
(x
(y

v = sin (x 0.5) cos (y 0.5)


Fig. 8 Zalesak slotted disk
rotation test

BPM

T/2

In order to compare our method with structured VOF


SLIC method and check the influence of grid quality
on the result, totally three different grids are used in
this work: (1) unstructured grid by Bubble Packing
Method (BPM, Shimada and Gossard 1995) with 7,600
cells; (2) unstructured grid by Delaunay Triangulation Method (DTM, Blazek 2001) with 7,600 cells; (3)
64 64 structured grid. The unstructured grids used
in this work are shown in Fig. 5. Apparently, BPM
can generate more regular cells than DTM and we
can check the influence of cell regularity on the final
results.
Two different interface tracking errors measurement
criteria Er and Eg as well as the relative mass conservation error Em are conducted to test the accuracy of
our method (Zhang and Liu 2008). Er is the relative
distortion defined as:

 e
 f (T) f n 
i
i
i
Er =
 0
fi

(15)

where fie (T) is the exact solution of volume fraction


for ith cell at the end of the test t = T and fin is the
computational result after n time steps, fi0 is the initial

DTM

Structured

312

Microgravity Sci. Technol. (2010) 22:305314

Table 3 Error measurement of rotation test (case 3)


Grid

Cell number

Er

Eg

Em

Structured
BPM
DTM

64 64
7,600
7,600

4.61E-02
5.18E-02
1.21E-01

2.02E-02
2.30E-02
5.38E-02

3.59E-04
6.65E-03
5.67E-02

solution. Eg is the absolute difference or geometrical


error between the exact and the computed results:



Eg =
Si  fie (T) fin 
(16)
i

where Si is the area of the ith cell. Finally, the relative
mass conservation error is described as follows:


 e
 n

 f (T)
f
i
i 

i
i
Em =
(17)
 0
fi
i

Figures 6 and 7 show the advection test of a right


triangle and a hollowed circle with velocity field (1,
1), respectively. In these figures, each column shows
initial state (0), state after half revolution (T/2) and
state after one revolution (T) for results on different

Fig. 9 Single-vortex shearing


flow test

BPM

T/2

grid. The advection of a right triangle is a good test


to compare the result by unstructured and structured
grid because it has two legs aligned with coordinates
and a hypotenuse which is supposed to behave better
on unstructured grid. It can be seen from Fig. 6 that for
unstructured grid by BPM and DTM, the deformation
of the hypotenuse is good as we expected, while BPM
is better than DTM for the two legs. As to the result
on the structured grid, the two legs show very small
deformation. However, the deformation of hypotenuse
is rather serious and the final shape is like a stair. For
the advection of a hollowed circle, BPM and DTM
almost keep the original circular shape although in
DTM there is numerical diffusion in the outside circle.
However, the final shape on structured grid becomes
rectangular for both outside and inside circle.
The interface tracking errors and the relative mass
conservation error are measured by Eqs. 79 and shown
in Tables 1 and 2. For the advection test of a right
triangle, structure grid shows the smallest errors. The
results of BPM are better than DTM and in the same
order with structured grid. For the advection test of
a hollowed circle, the results on structure grid have
smallest relative distortion error Er and relative mass

DTM

Structured

Microgravity Sci. Technol. (2010) 22:305314

313

Table 4 Error rotation measurement of shearing flow test


(case 4)
Grid

Cell number

Er

Eg

Em

Structured
BPM
DTM

64 64
7,600
7,600

1.35E-01
6.34E-02
3.07E-01

1.69E-02
7.92E-03
3.84E-02

5.77E-02
2.25E-02
2.66E-01

conservation error Em but largest geometrical error Eg ,


which means curved interface will be captured better
with less deformation but worse mass conservation on
unstructured grid than structure grid. Among unstructured grid, the results of BPM are better than DTM for
all these three errors.
Figure 8 shows the Zalesak slotted disk rotation test,
which rotates a slotted circle in a pure rotation velocity
field and it will return to its initial location after a full
revolution of 2 rotation. Each grid shows numerical
diffusion and BPM result looks best. The diffusion in
outside boundary is more serious for DTM, while the
diffusion in inside boundary is more serious for structured grid. As to the error measurement results shown
in Table 3, the interface tracking errors by structured
grid and BPM are in the same order and better then
DTM, while the relative mass conservation error by
structured grid is the smallest, which means the mass
conversation ability of structured grid is best.
Figure 9 shows the single-vortex shearing flow test,
the evolution of an initially circular fluid body placed
in the single vortex flow. The velocity field will stretch
the fluid body into a filament that spirals toward the
vortex center and then shrink the body back into the
initial circle after a full revolution of 2 evolution. This
test shows numerical diffusion by BPM is smallest best
because not only some flotsams appear in the half and
full revolution but also the final shapes are apart from
initial shape for structured and DTM unstructured grid.
It can also be seen from Table 4 that all the three
errors by BPM grid are the smallest since the cell shape
matches the rotating velocity field well, not only for the
deformation but also for the mass conservation ability.

Conclusion
A new interface reconstructing method for VOF based
on triangular unstructured grids is presented in this
paper. In this method, line segment is set parallel with
one edge of certain triangular cell containing interface through the judgment of a new criteria about the
volume fractions in the three neighboring cells of the
current cell. Once the location of the line segment

approximation of the actual interface is decided, the


new volume fraction in next time step can be computed
through the geometrical calculation of volume flux.
In this way, the time-dependent reconstruction of interface can be accomplished for every time step. The
algorithm is examined under four different evolutionary test cases: (1) advection test of a right triangle,
(2) advection test of a hollowed circle; (3) Zalesak
slotted disk rotation test and (4) single-vortex shearing flow test. In order to compare our method with
structured VOFSLIC method and check the influence
of grid quality on the result, totally three different
grids are used in this work: (1) 64 64 structured
grid; (2) unstructured grid by Bubble Packing Method
with 7,600 cells and (3) unstructured grid by Delaunay
Triangulation Method with 7,600 cells. The evolution
results and error measurements of these four cases on
three different grids show that this method is generally
capable to deal with all kinds of simple flows with
comparative accuracy with similar method on structured grid. In shear flow cases, the numerical diffusion
and errors by BPM are even smaller than that of
structured grid best. At the same time, the quality of
the unstructured grid has great effect on the interface
reconstruction. The more regular the grid is (by BPM),
the better results can be obtained by this method.
Acknowledgements The authors would like to acknowledge
the joint support of Natural Science Foundation of China (No.
50976087), Specialized Research Fund for the Doctoral Program
of Higher Education (No. 20090201110001) and Program for New
Century Excellent Talents in University (NCET-07-0661).

References
Blazek, J.: Computational fluid dynamics: principles and applications. Elsevier Science. Ltd, UK (2001)
Brackbill, J.U., Kothe, D.B., Zemach, C.: A continuum method
for modelling surface tension. J. Comput. Phys. 100, 335354
(1992)
Chato, D.J., Martin, T.A.: Vented tank resupply experiment:
flight test results. J. Spacecr. Rockets 43, 11241130 (2006)
Gao, D.M.: A three-dimensional hybrid finite element-volume
tracking model for mould filling in casting processes. Int. J.
Numer. Methods Fluids 29, 877895 (1999)
Gloth, O., Hanel, D., Tran, L., Vilsmeier, R.: A front tracking
method on unstructured grids. Comput. Fluids 32, 547570
(2003)
Herrmann, M.: A balanced force refined level set grid method for
two-phase flows on unstructured flow solver grids. J. Comput. Phys. 227, 26742706 (2008)
Hirt, C.W., Nichols, B.D.: Volume of fluid (VOF) method for the
dynamics of free boundaries. J. Comput. Phys. 39, 201225
(1981)

314
Ji, C., Wang, Y., Wang, J.: A novel VOF-type volume-tracking
method for free-surface flows based on unstructured triangular mesh. China Ocean Eng. 19, 529538 (2005)
Kim, D., Choi, H.: A second-order time-accurate finite volume
method for unsteady incompressible flow on hybrid unstructured grids. J. Comput. Phys. 162, 411428 (2000)
Noh, W.F., Woodward, P.: SLIC (simple line interface method).
Lect. Notes Phys. 24, 330 (1976)
Osher, S., Sethian, J.A.: Fronts propagating with curvaturedependent speed: algorithms based on HamiltonJacobi formulations. J. Comput. Phys. 79, 1249 (1988)
Sethian, J.A., Smereka, P.: Level set methods for fluid interfaces.
Annu. Rev. Fluid Mech. 35, 341372 (2003)
Shahbazi, K., Paraschivoiu, M., Mostaghimi, J.: Second order
accurate volume tracking based on remapping for triangular
meshes. J. Comput. Phys. 188, 100122 (2003)
Shimada, K., Gossard, D.C.: Bubble mesh: automated triangular meshing of non-manifold geometry by sphere packing.
In: Proceedings of ACM Third Symposium on Solid Modeling and Applications, pp. 409419. Salt Lake City, Utah
(1995)
Suol, F., Maldonado, O., Pino, R., Gonzlez-Cinca, R.: Design of an experiment for the study of Bubble Jet interactions in microgravity. Microgravity Sci. Technol. 21, 9599
(2009)

Microgravity Sci. Technol. (2010) 22:305314


Tryggvason, G., Bunner, B., Esmaeeli, A., et al.: A front-tracking
method for the computations of multiphase flow. J. Comput.
Phys. 169, 708759 (2001)
Unverdi, S.Q., Tryggvason, G.A.: A front-tracking method for
viscous, incompressible, multi-fluid flows. J. Comput. Phys.
100, 2537 (1992)
Wang, Z., Yang, J., Koo, B., Stern, F.: A coupled level set and
volume-of-fluid method for sharp interface simulation of
plunging breaking waves. Int. J. Multiph. Flow 35, 227246
(2009)
Yang, X., James, A.J., Lowengrub, J., Zheng, X., Cristini, V.:
An adaptive coupled level-set/volume-of-fluid interface capturing method for unstructured triangular grids. J. Comput.
Phys. 217, 364394 (2006)
Youngs, D.L.: Time-dependent multi-material flow with large
fluid distortion. In: Morton, K.W., Baines, M.J. (eds.) Numerical methods for fluid dynamics, p. 24. Academic Press,
New York (1982)
Zhang, Q., Liu, P.L.-F.: A new interface tracking method: the
polygonal area mapping method. J. Comput. Phys. 227,
40634088 (2008)
Zhao, Y., Tan, H.H., Zhang, B.: A high-resolution characteristics-based implicit dual time-stepping VOF method for free
surface flow simulation on unstructured grids. J. Comput.
Phys. 183, 233273 (2002)

Вам также может понравиться