CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Two-phase problem (Vof?) (http://www.cfd-online.com/Forums/main/7866-two-phase-problem-vof.html)

 Raj July 25, 2004 16:15

Two-phase problem (Vof?)

I need help in solving a two phase problem. I have a closed tank(3-d, complicated geometry)that is half filled with liquid and has ullage space(with air) above it. After 4 hours, it is drained. I am using CFD-ACE to model flow and heat transfer. What is the most appropriate way to solve this? I tried to use VOF, but it requires a very small time step which is not feasible for my problem. Can anyone suggest what to do here?

 Eugene de Villiers August 3, 2004 09:31

Re: Two-phase problem (Vof?)

Cant you increase or add VOF subcycles to reduce the CFL limitations of the overall timestep? In a good VOF code you should be able to run the interface at a courant number of near one, provided you subcycle the phase fraction transport equation about 4 or 5 times.

Let me know if you need a reference.

Eugene

 Raj August 3, 2004 10:40

Re: Two-phase problem (Vof?)

Thanks Eugene -

Yes, can you please send me the references. Also, I dont understand what do u mean by adding/increasing VOF sub-cycles. I am using CFD-ACE.

 Eugene August 3, 2004 13:11

Re: Two-phase problem (Vof?)

Well, I was using this code: www.nabla.co.uk for my VOF calculations of spray atomization, there are some descriptions on the website. I do not have any knowledge of CFD-ACE unfortunately.

By subcycling I mean you run the momentum-pressure solver at one timestep. Then after the PISO loop has converged for that time, you solve the phase fraction transport equation at a different timestep that is an integer fraction of the main timestep. Example: main timestep = 1 second. Using the converged velocity field, solve the phase fraction transport with a timestep of say 0.25 seconds. By doing this 4 times you arrive at a phase fraction distribution at the same time as the current momentum solution.

You can make it more accurate by storing old fields and using time interpolated velocity at the different subcycles. Since your fluid fraction is being transported at a much smaller courant number than your momentum you can dodge some of the stability issues involved with the sharp density gradient at the interface. Of course this isnt a fix all: you cant have too large a courant number difference between the two (at a given location) or you'll run into a different set of problems.

Hope this helps

 All times are GMT -4. The time now is 14:36.