Hi there: Has any CFD scholar over there tried applying Roe's FDS (1981) into Bckward facing flow.

I have problem verifying my results with the experiments. I am not sure that it is due to the buggs in the code or the scheme is not suitable for this case.

Thanks for the help. MJ

Ciao,

you might have a look to the following arcicle. It refers to both Roe-solver as well as to the backward facing step problem.

Qurik, J.J. "A contribution to the great Rieman Solver debate," Int. J. for Num. Meth. in Fluids, Vol. 18, pp. 555-574, 1992.

I found this article very interesting and helpfull when doing our own validation on the same Problems (Roe-Solver and backward facing Step).

Ciao Heinz

Hi there:

I found the paper very informative as a good collection for Roe's shortcoomings. Here I suggets the paper for all the Roe's users. Especially when the author concludes the failure looks recur when grids are aligned with shocks; giveing not enough dissipation ..

But there were nothing about backward facing step; which is a test with viscous terms and recirculationg zones. Moreover, I don't think I will need any of the fixing mentioned in this paper to my Roe solver becuase I have (1)- very low speed, (2)- I have enough recirculating flow field which makes the flow not-aligned with my grids in general which produce enough dissipation.

Did you try backward facing step by Roe scheme?

Thanks

Ciao,

it locks like there wasa some misunderstanding. I was thinking you are simulationg high speed flows, when I refered to Quirks publication.

We did ourself simulation of the typ what quirk calls in this paper diffraction of a strong shock wave around a 90 degree corner (se fig 9-11). We also called this test case backward facing step problem when we did the same simulations, using Roe's solver on a 3-D unstructured grid. I think the 3-D unstructuered grid are a little more diffusive than those used by Quirk. Therefore we did not see such stability problems as Quirk for the Roe solver.

Those simualtion we did by only solving the Euler Equations using Roe's solver. We think viscous terms can be neglected due to the high speed nature of the test case (Mach 5 flow).

In addition our own experience are, that Roe's aproximate Rieman solver is not very well suited for low speed flows, eg. Ma<0.3-0.5. There was already alot of discussion on this in this forum. The headline was density and pressure based methods.

What exactly does your Problem look like, eg flow speed and so on. If you are interessted I can send you some of our results by E-Mail.

Ciao Heinz

Hello I read your post and i would like to read the papaer that Heinz suggested. But i can't find it. If you have that paper in text file. Please share it to me. Thank you very much

suthichock nunthasookkasame

Hi there:

It is a journal paper at the exact address that Heniz has suggested. I don't have text copy of that, but I can fax it to you if you are in Canada.

Regards Mohammad

Hi Thank you for your response. But I am not in Canada. Anyway, Do you have that paper in computer file format ( like microsoft document , or .PDF )? If you have it , please E-mail to me . Thank you. Suthichock Nunthasookkasame