rotor compresser R37 modification #662
Ryupuriwatt
started this conversation in
General
Replies: 2 comments 2 replies
-
(1) Did the adjoint converge? (2) I suggest you try a different optimizer, like SLSQP. (3) In the 2nd screenshot, nuTilda=1e-3 is for setting normalizing value for adjoint, it is not used in flow simulations. |
Beta Was this translation helpful? Give feedback.
1 reply
-
Have you tried SLSQP? IPOPT sometimes has problem converging constrained shape optimization. |
Beta Was this translation helpful? Give feedback.
1 reply
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
hello profressor PingHe, I would like to know what can I do when the optmization objective(Torque) reduce very little(0.24%). I modify from rortor37 to rotor67, using same objective and constaint. which parameter is good for making the toque reduce to more lower value. Is it because residual is not low enough my case all residual is below 1e-5 but I see in the profressorr tutorial profressor set it as 1E-8. I have try to set fvSolution and fvScheme but I can not get as low as profressor set up. Eventhough my case can not get to below 1E-8 but shock wave can be captured. so I really wonder how to make it reduce. For this case how much reduction can be treated as good and I have make a validation the total pressure ratio error is 10% since we have no rotor tip is it ok for 100k Mesh. because I have try 100k 300k 400k 500k and the more higher resolution like 300k 400k 500k can get the presure residual down to just only 1E-4 which is not good enough for capturing the shock wave.
I have another question I see profressor set niuTiuda very high is it help in convergence. If set it like this will solution will be realistic since normal flow condition is around 1e-5 so I am curious
Beta Was this translation helpful? Give feedback.
All reactions