Hi Dear:

I wrote a 3-D LES code. In linear terms I use Roe scheme

,the viscous term I use Smag model with central

defference and explicit Mac Cormack scheme. But the result

is always unstablity because the density is negative in

the same region. My CFL number is 0.01. What's the problem?

So you are doing compressible LES?

What you mean by Roe's scheme in linear terms. Solving the convecting terms? The Smag terms won't be the problem. Roe's with Rieman solvers will not work very good, use HLLEM or HLLC instead, much better. Which type of flow do you have? I will advice to use implicit, try staring with a simple euler firts order scheme to see if it works. First stability, then accuracy.

Hi Salvador:

Thanks for your suggestion. Actually you are right.

I did compressible LES. Because I want to use

precondtioning so I used Roe scheme. The flow is air.

I will try a simple euler firts order scheme to see if it

works. Thank you.