1st order vs 2nd order
Hi I'm simulating flow throught 2 parallel plates with inclined inlet and outlets. I tried the second order upwind momentum discretization (and also QUICK) but it couldn't converge very well. For example the average velocity in a plane was fluctuaing at around 1% of its maximum value. But when I switched to the first order upwind, this fluctuation is reduced to 0.2%.
I know first order can converge more easily and 2nd order is more accurate. But will a 2nd order solution that is partially converged be more accurate than a 1st order one? Are there any ways to help the 2nd order solution converge? I have reduced the momemtum under relaxation to 0.3 but to no avail. I was wonder if I should make some changes to the multigrid entries. Any suggestions? 
Re: 1st order vs 2nd order
Try to increase the pressure underrelaxation factor to 0.6 or so.
Also consider a grid refinement. Regards, ap 
Re: 1st order vs 2nd order
Unfortunately that didn't work. In fact the integral of the pressure in the system is also fluctuating and can't converge.

Re: 1st order vs 2nd order
Check your grid elements. If they are high skewed that could be the possible reason. Luca

Re: 1st order vs 2nd order
Some of the elements are indeed skewed but I can't do much about it because of the geometry of the domain. Can I work around this by using skewness correction in SIMPLEC or PISO? If so, what is a common value to try?

Re: 1st order vs 2nd order
Just found of that PISO dosen't work. I tried it with and without skewness correctiona and used the QUICK scheme with it.

Re: 1st order vs 2nd order
You should try to correct your grid...I had your same problem and when I remeshed the problem was solved. Luca

Re: 1st order vs 2nd order
I remeshed taking care in reducing the number of skewed elements. It dosen't seem to work. As before, the residuals can't get below 1e3 when I used 2nd order upwind. Pressure is also fluctuating.

Van leer second order and AUSM+
Hi, I've got a fortran code for solving 2D compressible flow over an airfoil with the method of first order van leer flux vector splitting . I have to change it to second order van leer flux vector splitting method . Can you please help me on how to do it? It's also possible for me to solve this project with the method of AUSM+ . Are you familiar with this method? can you help me with it please?
Thanks a lot in advance 
All times are GMT 4. The time now is 20:29. 