'Best' algorithm for CFD on unstructured 2d grids?
 

Dear All,

I have used CFX and other commercial codes, and have a good background in scientific programming. Now, mainly for my own interest and education, I wish to develop my own (toy) code for 2-d CFD on an unstructured grid. I hope to implement k-epsilon and maybe k-omega models within this code....

So my question is:

1) Do you know of a good, robust algorithm for simulating turbulent flow on an unstructured 2-d grid?

2) Can you point me towards a good reference which describes it?! One that gives enough detail to replicate the method if possible?

I would prefer to work with an algorithm using collocated variables (pressure and velocity at same nodes) but I am open minded and can be convinced otherwise by good evidence.

All references would be very welcome - but I would particularly welcome information about algorithms that are well-proven and general purpose, even if they are not state-of-the-art.

Very many thanks for your help.

Best regards, andy2o

Re: 'Best' algorithm for CFD on unstructured 2d gr
 

Why would you want to do turbulent flow in 2d ? There are so many source codes available. You can also browse through the cfd-wiki for a list of codes available for free and then decide on how you want to structure your code. You cannot take a turbulence model for example and say its general purpose since all the numerical models have their pros and cons. The better option would be to decide the application that you have in mind and have a look at what kind of CFD codes have been used to analyze the problem.

Re: 'Best' algorithm for CFD on unstructured 2d gr
 

Harish,

Thanks for your reply...

"Why would you want to do turbulent flow in 2d ? There are so many source codes available. You can also browse through the cfd-wiki for a list of codes available for free and then decide on how you want to structure your code."

I want to understand the algorithms associated with finite volume based CFD on an unstructured grid - not as a pure numerical analyst, nor as a CFD user, but somewhere inbetween. Too many of the issues to do with mesh quality measures, higher-order upwinding on unstructured meshes, etc. are still slightly mysterious to me after reading through the CFX theory guide and the other references I have at the moment. So now I am looking for better, more detailed references, as my way of understanding it is to look at the details of the implementation. I have looked through the codes (thank you for pointing out the wiki - I found some that I was not aware of from other searches). There are some interesting codes to study - perhaps I will code a solver, of my own, perhaps not.

Either way, I would still very much welcome references to algorithms for the discretization of the Navier-Stokes equations and turbulence models on an unstructured grid!!! :)

"You cannot take a turbulence model for example and say its general purpose since all the numerical models have their pros and cons. "

I'm sorry - I worded my post badly, and my aim was not clear.... I know that different turbulence models have pros and cons. That's not what I was trying to ask about :(

What I would like to find information about are numerical discretization schemes (upwinding, linearisation of non-linear terms, treatment of the source terms, Laplacian, how to apply wall-function boundary conditions) applied to a range of k-omega and k-epsilon models. The exact details, of course, are different for the k-epsilon model and the k-omega model, but I'm pretty sure there are some general rules about the formation of a good numerical discretization for these families of equations.

Any references woulod still be very welcome.

Best wishes and thanks for your reply, andy2o

Re: 'Best' algorithm for CFD on unstructured 2d gr
 

i could tell you that i understand most of implementation details of navier stokes and turbulence. And the reason i understand it is i follow closely documentations of following three solvers: Fluent, CFX and StarCCM+.

the best that you can read is starCCM+ manual, it tells you in detail how the things are implemented. No need to write solvers to do it. Just go through it. (Beware the starting version's manual had lots of typos in it, some of them are improved in new version, but you might encounter them).

Re: 'Best' algorithm for CFD on unstructured 2d gr
 

Take a look at the book of Blazek. It has lot of practical details.

Look at the notes of Tim Barth of NASA. You will find them on his website.

You can also look at the manuals of some CFD codes like NSC2KE and Typhon.

Re: 'Best' algorithm for CFD on unstructured 2d gr
 

Thanks for your reply. The Star CCM+ manual sound good - I'll see if I can get access to them, but it seems quite unlikely. :(

Regards, andy2o

Re: 'Best' algorithm for CFD on unstructured 2d gr
 

Many thanks. The NSC2KE code has some good information, and I will look for that book by Blazek (it's not cheap on Amazon though!) - I miss having a university library for this sort of thing...

regards, andy2o