[quarkz]

Hi,

I have been trying to run a hypersonic double cone problem with turbulence modelling. Using komega-sst, the omega diverged after 3-4 time steps. I changed to another model and the same thing happens. So I tried to run my case using my previous laminar solution, which completed w/p problem.

To reach steady state, I ran until t=0.005 in the laminar case. So I tried to use the laminar solution at t=0.0002 to start my turbulent case, and let it run till 0.005.

Is this a reasonable or correct way to simulate a turbulent case?

Another online study suggested running at a low Re 1st, then increasing the Re to the actual one as time proceeds. Is that ok too?

Thanks.

[LuckyTran]

Both are okay. I would think the laminar solution is preferred over the solution at wrong Reynolds number. I don't get why you don't use the lastest laminar solution as the initial condition for your turbulent case.

A better solution though is to try and figure out what set of schemes is making it blow up, assuming it's not something simple like providing the correct initial k and omega fields.

[quarkz]

Hi LuckyTran,

Thanks for the reply. My case converges to a steady solution. So I thought that if I start with a laminar steady solution, the new turbulent steady solution may be too similar to the laminar one.

I already tried to mix different schemes, like upwind etc. However, it still blows up. There are simply too many possible combinations and I have tried quite a few. Hence, I decided to use this mtd. It seems to be working (not diverging) but I have yet to anaylze the results.

[juliom]

I always use the same approach when I have problems with stability. I start with a case with a lower Reynolds and increase the Reynolds little by little. The big problem with CFD in general are the BC and IC. With turbulence models this becomes more important.

[quarkz]

Hi juliom,

I'll try your mtd too. Thanks!