LIQIANG
2004-11-05
In doing the example case, I found that for case NACA0012 in the package, when attack angle is set to 30 degrees and free stream is 0.8 Mach, the case not run normally, and another situation attack angle is set to 3.0 degrees and free stream is 3.0 Mach, the case does not run normally either, can any one find the reason of this problem, and help me, is there some flaws in the source code???
Joseph H. Morrison
2004-11-09
I have not had a chance to run these cases. Can you please provide me with additional details on what you are seeing in the solution? The most useful items would be plots of the convergence and contour plots of pressure for both cases.
I expect that the 30 degree Mach 0.8 case is unsteady. You have the airfoil at a very high angle of attack for this Mach number. There is already a relatively strong shock on the upper surface at alpha of 3 degrees. At 30 degree angle of attack you have to accelerate the flow around the nose of the airfoil even more. I expect an even stronger shock. Then the flow has to turn around the trailing edge. I expect that there will be unsteady shedding going on in this case, but I have not run it.
The Mach 3.0 case may not be appropriate for this grid. At Mach 3.0, I expect that the bow shock stands off from the body. The shock will be very misaligned with this shock and it will slow convergence and may show oscillatory convergence as the shock bounces around in the grid. You may need to reduce the CFL to improve convergence.
Joe
LIQIANG
2004-11-14
I have tried the sample case again, and found that for this case of 3.0 degrees attack angle and free stream of 3.0 Mach, without multigrid method, the case can run with first order correctly. Using multigridmethod, when transfering from coaser grid to finer mesh, this will be error on function sqrt(negative value),
another, in the future version, can the anthor improve the order of the difference scheme, such as the ENO or WENO scheme, which can be easily incorporated into the flus difference.
This work is very great, thank you Joe
Joseph H. Morrison
2004-11-16
Does the M=3 alpha=3 case run using the second order scheme? Where does the sqrt(-) value occur in the grid transfer (I expect that it occurs at/near the shock). This can be fixed by improving the grid transfer operator. There are several papers on multigrid on high speed flows to look to for guidance.
I have not planned on implementing ENO or WENO but I agree that it would be useful. It would require additional transfers at block boundaries and some reworking of the storage to accommodate this. Is anyone interested in working on this?
I greatly appreciate the kind words. I hope that you find ISAAC useful. If you publish any work using ISAAC, I would like to add a reference to your work on the website.
Joe
sj_151
2007-09-09
hi all,
I am facing the following metric errors after generating the grid in a 32-bit compiler.
METRIC: ERROR-> Volume <= 0 at BLOCK,LEVEL,I,J,K = 1 1 2 3 2-6.125E+00
METRIC: ERROR-> Volume <= 0 at BLOCK,LEVEL,I,J,K = 1 1 2 5 2-6.125E+00
METRIC: ERROR-> Volume <= 0 at BLOCK,LEVEL,I,J,K = 1 1 2 7 2-6.125E+00
METRIC: ERROR-> Volume <= 0 at BLOCK,LEVEL,I,J,K = 1 1 2 9 2-6.125E+00
METRIC: ERROR-> Volume <= 0 at BLOCK,LEVEL,I,J,K = 1 1 2 11 2-6.125E+00
METRIC: ERROR-> Volume <= 0 at BLOCK,LEVEL,I,J,K = 1 1 2 13 2-6.125E+00
and similar error statements.Can anyone please tell if this is because of the grid or is this a storage problem.I have maximized all storage.
Sj