CFX converge problem caused by shock waves

littlelz · August 10, 2009, 05:41

dear all,

I have got a serious convergence problem in CFX12 transition simulation.

I am doing the transition simulation on a isolated nacelle using CFX12. The mesh and the CFD model setup are ok. We can get successful results for all the test cases with various operating conditions except the cases with shock waves (in our cases, freestream mach number reaches 0.88). I have done refining the mesh on both nacelle surface and normal to it, but no success. I also follow the suggestion as in this forum (http://www.cfd-online.com/Forums/cfx...r-3d-wing.html), set max continuity loops = 2 with high res solver, however, no success at all. The general trend is no matter what I did, the simulation will converge slowly first, then diverge totally. (I monitor some critical parameters, like Cp, Cf, they can’t be converged enough).

I guess it is a converge problem caused by shock waves. I really appreciate any of your help and advice.

regards,

littlelz

ghorrocks · August 10, 2009, 06:22

Hi,

Is it steady state? Assuming it is:

There are some general tips here http://www.cfd-online.com/Wiki/Ansys...gence_criteria

But in this case I would recommend concentrating on using local timescale factor. Also have you tried a solution which is subsonic, converging the solution, then slowly increasing the mach number to the desired figure? This can also help.

If that does not work then I would consider doing a full transient simulation. They tend to be the last resort when nothing else works. Slow, but the most reliable convergence.

Glenn Horrocks

littlelz · August 10, 2009, 17:47

many thanks, Glenn

always get your help and advice, thank you very much.

the tips in your link I have tried before because there is separation flow in our nacelle case, the flow can't be converged very well in steady state simulation. however, i am still using steady state simulation because we found the convergence problem caused by separation is just a local probloem, which doesn't influence the overall result. as I monitor the Cl and Cd, they converge well.

however, this time the converge problem is caused by shock wave. we have tried M=0.8, M=0.82 they are converging very well. as M reaches 0.88, it can't converge at all. so I am wondering if there is any special tips for shock wave converge problem

anyway, I will try your advice one by one.

many thanks again

littlelz

Timon · August 17, 2009, 09:35

Besides using local timestepping, try enabling "high speed numerics" in compressibility control.

August 10, 2009, 05:41	CFX converge problem caused by shock waves	#1
littlelz New Member Join Date: Mar 2009 Posts: 24 Rep Power: 17	dear all, I have got a serious convergence problem in CFX12 transition simulation. I am doing the transition simulation on a isolated nacelle using CFX12. The mesh and the CFD model setup are ok. We can get successful results for all the test cases with various operating conditions except the cases with shock waves (in our cases, freestream mach number reaches 0.88). I have done refining the mesh on both nacelle surface and normal to it, but no success. I also follow the suggestion as in this forum (http://www.cfd-online.com/Forums/cfx...r-3d-wing.html), set max continuity loops = 2 with high res solver, however, no success at all. The general trend is no matter what I did, the simulation will converge slowly first, then diverge totally. (I monitor some critical parameters, like Cp, Cf, they can’t be converged enough). I guess it is a converge problem caused by shock waves. I really appreciate any of your help and advice. regards, littlelz

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
problem in displaying surfaces in CFX	haho	CFX	1	July 5, 2009 19:25
CFX 11 x64 solver? problem	Attesz	CFX	6	June 7, 2009 08:37
Ansys Workbench (CFX) bucket problem	njsavage	CFX	1	April 30, 2009 09:51
Ansys CFX bucket problem	njsavage	Main CFD Forum	1	April 30, 2009 09:48
Workbench (CFX) bucket problem	njsavage	ANSYS	0	April 29, 2009 17:10

August 10, 2009, 06:22		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	Hi, Is it steady state? Assuming it is: There are some general tips here http://www.cfd-online.com/Wiki/Ansys...gence_criteria But in this case I would recommend concentrating on using local timescale factor. Also have you tried a solution which is subsonic, converging the solution, then slowly increasing the mach number to the desired figure? This can also help. If that does not work then I would consider doing a full transient simulation. They tend to be the last resort when nothing else works. Slow, but the most reliable convergence. Glenn Horrocks

August 10, 2009, 17:47		#3
littlelz New Member Join Date: Mar 2009 Posts: 24 Rep Power: 17	many thanks, Glenn always get your help and advice, thank you very much. the tips in your link I have tried before because there is separation flow in our nacelle case, the flow can't be converged very well in steady state simulation. however, i am still using steady state simulation because we found the convergence problem caused by separation is just a local probloem, which doesn't influence the overall result. as I monitor the Cl and Cd, they converge well. however, this time the converge problem is caused by shock wave. we have tried M=0.8, M=0.82 they are converging very well. as M reaches 0.88, it can't converge at all. so I am wondering if there is any special tips for shock wave converge problem anyway, I will try your advice one by one. many thanks again littlelz

August 17, 2009, 09:35		#4
Timon Member Join Date: Mar 2009 Posts: 44 Rep Power: 17	Besides using local timestepping, try enabling "high speed numerics" in compressibility control.