CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Level Set Method or VOF (http://www.cfd-online.com/Forums/main/3261-level-set-method-vof.html)

Tareq Al-shaalan March 20, 2001 00:53

Level Set Method or VOF
 
Hello

Which is better method Level Set or VOF to study free surface flow.

I am studying two-phase flow with free surface. There is a vortex in the flow I am studying.


Kike March 20, 2001 04:21

Re: Level Set Method or VOF
 
Dear colleague

The choice will strongly depend on the evolution you prevent for your interface. Each tracking method (Level Set or Heigh of Liquid, VOF, adaptative grids, surface and volume markers) has it own advantages and drawbacks.

Level Set Methods is easy to implement into any finite volume code and do not smear the interface (it has no numerical diffision), but it has serious problems when the interface (z=f(x,y)) becomes a multivalued function.

Volume tracking methods (VOF family) are widely used. It is itself a proof of its efficacy, but you will have some amount of numerical diffusion, deppending on how you reconstruct the interface after convection.

You have a wide spectrum to choose instead of this two. Nevertheless, between them I use VOF, with a special procedure to reconstruct the interface. I had applied this tracking method to liquid films instability and two-phase shear layers instability, with good results.

Try to find this paper (and its related referencces from authors):

"Direct Numerical Simulation of Free-Surface and Interfacial Flow.", Ruben Scardovelli and Stéphane Zaleski, Ann. Rev. Fluid Mech. 1999, 31:567-603.

I hope it helps you, good luck

Kike

Pei-Ying Hsieh March 20, 2001 10:56

Re: Level Set Method or VOF
 
Hi,

There is a group using combined VOF/Level Set method. I have not implemented this method, but, it seems like a good idea.

You can do a search on the Interest and will find some references.

P.Hsieh

John C. Chien March 20, 2001 13:10

Re: Level Set Method or VOF
 
(1). It's going to be very slow for the reader, even if it can be done. (it has been tested one I guess?) (2). I normally do not wait for the banners to appear on the webpage, even for a small picture, it is taking a lot of time to load. (3). You have to consider the average users through average phone lines. (4). You can e-mail to someone to get the answer. But even in that case, most free e-mails have limited disk space to store the data. So, the picture can easily overload the busy e-mail box. (5). When you have problem with mesh generation, it is likely that you have problem with the algorithm used, not the final picture you generate. So, what is your problem?

John C. Chien March 20, 2001 13:13

Err,answer to can I put picture in my post, sorry.
 

kalyan March 20, 2001 13:18

Re: Level Set Method or VOF
 
I tried to keep up with the literature on front/surface tracking methods till a couple of years ago. What I have seen is that the VOF methods are better for 2-phase flows (for reasons pointed out by Kike) but most (atleast among the ones I have seen) of these methods use cartesian meshes. Front reconstruction could be highly problematic on curvilinear or unstructured meshes. Hence, I would suggest using level set methods on such meshes.

A free surface is just a front except that it is not purely convective (it has surface tension). Try to see which method gives you a good representation of the surface tension (geometric curvature).

It is hard to validate codes for free surface flows. Most studies I have come across are qualitative. (Eg. collision for liquid drops, drop impinging on a liquid surface/wall etc.). The most reasonable validation procedure seems to be a grid independence test. There may be known solutions using which one can validate a code but I do not know any.

Mixed level-set/VOF idea also sounds pretty appealing. I have not read the paper yet. If you try it, please let me know how it works.

Tareq Al-shaalan March 21, 2001 01:05

Re: Level Set Method or VOF
 
Hello

I thank you all for your Response.

I am studying free surface in 3D flow. Is there any free subroutine for VOF that will reconstruct the free surface.

Glenn Price March 21, 2001 20:11

Re: Level Set Method or VOF
 
Doug Kothe, Bill Rider & their co-workers at Los Alamos have published a tremendous amount of material on the VOF method along with some comparisons to the level set method (see http://public.lanl.gov/mww/HomePage.html). Williams and Kothe have advanced the piecewise linear VOF method to handle unstructured meshes. See the above site for a PDF copy of a paper on this subject.

As mentioned by another poster, Stephane Zaleski & co-workers have also published some excellent papers on the VOF method.

The main drawback of the level set method is its inability to rigorous conserve mass. Recent works introduce global mass conservation constraints during the re-initialization procedure at each time step. However, this leads to local mass generation/loss in cells around the interface. Nevertheless, the level set method can accurate calculate geometric properties of the interface and is relatively easy to code in its basic form.

Not to exclude front tracking methods, see Prof Gretar Tryggvason's website (http://www-personal.engin.umich.edu/~gretar/) for the state-of-the-art on these methods.

Good luck.

Bubba March 23, 2001 19:33

Re: Level Set Method or VOF
 
I myself am developing a 3-d VOF code. First no there are no freely available 3-d VOF routines. I believe Sussman has an old Level-set code on his web page. I have no idea if it is 2-d or 3-d.

http://euclid.math.fsu.edu/~sussman/

Secondly, I have also been following the advancement of High Resolution methods

(publication from Peric).

http://www.nap.edu/books/0309065372/html/638.html

or the thesis of Ubbink...

http://monet.me.ic.ac.uk/publication...ts.html#Ubbink:1997

These methods diffuse the interface (2-3) cells however if that is the accuracy you are looking for then its the easiest method on ANY type of grid. In fact, in my opinion, if you combine this method with AMR and with a fast solver then you will be able to get solutions just as accurate as any other front tracking scheme...

This code is free and has all the goodies needed:

http://seesar.lbl.gov/CCSE/Software/index.html

Also one should not forget about phase-field methods. I also am intregued by the ghost fluid method of Fidqew (sp?). This seems like a very accurate method, although a bit more expensive.

Check out reports on Levelset and The ghost fluid method here:

http://www.math.ucla.edu/applied/cam/index.html

hmmm..it seems I talk too much :)

good luck..

Tareq Al-shaalan March 25, 2001 00:11

Re: Level Set Method or VOF
 
Hello

Can I use level set method for courser grids?


All times are GMT -4. The time now is 08:26.