CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Main CFD Forum (
-   -   Fluid-Void Simulations (Any Softy) (

Fluid-Void Simulations September 1, 2008 21:30

Fluid-Void Simulations (Any Softy)
I have to do some calculations involving polymers of very high viscosity. I tried multiphase calculations with air but the solution is too instable. Is there any software that could perform fluid-void type calculations. I think this is best way to perform calculation on my case.

Ted September 2, 2008 00:50

Re: Fluid-Void Simulations (Any Softy)
solavof, nasavof2d, nast2d

CK September 2, 2008 01:03

Re: Fluid-Void Simulations (Any Softy)
If you would like to use commercial CFD code, then you may consider Flow-3D or Fluent.

Fluid-Void Simulations September 2, 2008 01:52

Re: Fluid-Void Simulations (Any Softy)
First thanks for reply.

solavof, nasavof2d, nast2d

Out of these two of them are 2d codes. And solavof is vof based code.

I have Fluent and I tried using it for VOF simulations. Becuase of very high viscosity ratios solution soon becomes unstable. Fluent also mentions that this could be big problem.

This is why i am thinking a solver that only solves for one phase would be good idea.

So it seems there is no commercial solver that solves fluid void type problems.

Fluid-Void Simulations September 2, 2008 01:54

Re: Fluid-Void Simulations (Any Softy)
i got to see flow 3d , i will check it.

Ted September 2, 2008 04:36

Re: Fluid-Void Simulations (Any Softy)
Flow3D have sinlge fluid option too! (it's nothing else 3dextension of solavof)

i think that high viscosity ratio is not so serious but high density ratio is difficult to handle, it could be coped by limiting time step, though using implicit time integration, have u tried this?

PS: fluent is bit slow for 3d + vof, single phase flow3D is very fast in contrast

otd September 2, 2008 08:55

Re: Fluid-Void Simulations (Any Softy)
There is also a nasavof/3d out there.

Ted September 2, 2008 09:31

Re: Fluid-Void Simulations (Any Softy)
nasavof3d? it is an asysimetric solver (z, r, theta), did u know!?, and it's almost useless in practice!

otd September 2, 2008 13:44

Re: Fluid-Void Simulations (Any Softy)
Don't know about its utility. I was told that NASA terminated funding on the code when the Challenger was lost. The money was needed elsewhere so the code was never thoroughly tested.

What does "asysimetric" mean? Asymetric (that is, 3-dimensional), or axisymmetric?

The code is not limited to axisymmetric. The coding clearly shows a 3-dimensional grid and 3-dimensional indexing on 3 velocity components, plus the pressure and free-surface configuration.

Ted September 2, 2008 14:48

Re: Fluid-Void Simulations (Any Softy)
sorry for typos, i was in hurry.

i mean a full 3D but in 3d cylinderical coordinate, your grid should be a part of such grid to use code,

e.g., 3d cylinderical tank, or pipe ...

Fluid-Void Simulations September 4, 2008 02:31

Re: Fluid-Void Simulations (Any Softy)
Guys thanks very much for all your replies.

I did some experiments with Fluent and it seems with Presto scheme, the solution remains stable other than that if i chose any other pressure interpolation scheme solution becomes unstable.

But for me, Fluent is useless because my calculations involve few rotating parts and handling that is not fluent's forte. (I could only ran few test cases).

I looked at flow-3d but at the moment my company is not interested in investing in another software.

Good thing is, i used particles to track fluid and solved this problem in my own code. Which so far has given me some solution that looks good.

I am trying to understand VOF and single fluid solution so that i could impliment it in my code and do away with particle to track fluid in solution domain.

Anyway, thanks again for all the replies. It was big help.

Ted September 4, 2008 02:48

Re: Fluid-Void Simulations (Any Softy)
if u look for some implementations, i did and can share with u (PLIC-VOF and also MAC method) in fortran 90

Fluid-Void Simulations September 4, 2008 20:48

Re: Fluid-Void Simulations (Any Softy)
ted, thanks for your offer. I would indeed like to see how it is implemented. If i could implement it successfully i would put it online for people to use it. As the program i have could work with unstructured grids and as cartesian grid based solver for general moving bodies.

For some reasons (company issues) I could not put my name and email on cfd-online so i would make a dummy email so that you could contact me. Or you could give some of your email to make contact.

Fluid-Void Simulations September 5, 2008 00:33

Re: Fluid-Void Simulations (Any Softy)
Ted, my email is

could you send me code there. Thanks for all the help.

Ted September 5, 2008 01:49

Re: Fluid-Void Simulations (Any Softy)
download it from here:

no documentation and support is available, however commented code and easy to learn. if you actually need you can learn how to deal with

good luck

Fluid-Void Simulations September 5, 2008 02:15

Re: Fluid-Void Simulations (Any Softy)
Hi, thanks again, I had a quick look. And i think i found out one mistake that i had been doing. (i would call it blunder).

! Solve: dc/dt = -div(uc) + c div(u) (c changes only at cells ! where grad(c) is not zero)

Well i did not add c div(u) part.

Your code is nice little code and it is easy to understand.

I just wondering, why do you need to write a code for VOF. Learning or you need to apply it for some calculations.

For example, i do need to do calculations for research related to work.


Ted September 5, 2008 06:46

Re: Fluid-Void Simulations (Any Softy)
regarding to vof equation: there is no bug, note that there r two versions of vof in my code, split and unsplit

regarding to unsplit one that you notify: not that we solve incompressible flow but numerically so div u != 0, so we have to do divergence correction to keep conservation of vof tracer. if you refer to literature you find more.

but i try to describe briefly there:

vof eq.: c_t + u . grad c = 0

u . grad c = div(cu) - c div(u)

so replace we have:

c_t + div(cu) - c div(u) = 0

commonly we solve first this: c_t + div(cu) = 0

then do divergence correctin, is it ok?

> I just wondering, why do you need to write a code for VOF

i'm a researcher, and it was early part of my research in 6 years ago, i do not use this code currently and just offer it to learners.

All times are GMT -4. The time now is 11:44.