Search the Community
Showing results for tags 'sensitivity'.
Found 3 results
Hy, I would like to calculate a special safety value from stress responses, and using it in topology optimization. The test of this calculation is started with a simple problem. DRESP1 case 1: - constaint: DRESP1 (von mises stress response, one selected element) - objective: DRESP1 volume DRESP2 case 1: - only response: DRESP1 (von mises stress response, one selected element) - constaint: DRESP2, (equation response) is calculated the following equation a(x)=x and using the DRESP1 von mises response - objective: DRESP1 volume In this case the changes on design variables are the same, and the results are the same. Than i changed the response element quantity to 2 and the DRESP1 and DRESP2 case results are different I have tried to change constraint screening, nothing has changed I have tried to look on sensitivity values, but no stress values included in *.asens files or *_topol.h3d files I have tried to compare it with displacement response/constraint and they are good, so the DRESP1 and DRESP2 with 2 response nodes has the same results. Is this problem connected to Stress-NORM aggregation ? How can i turn ON Stress-NORM aggregation in DRESP2 case? How can i examine the stress response sensitivities? Did i miss someting in the setup? I have attached the 2 different *.hm and *.fem files Thanks, CSR dresp1_simp.fem dresp1_simp.hm dresp2_simp.fem dresp2_simp.hm
Hi everybody, I am a new user, and these are my questions: How can we modify the direct method to adjoint method? How can we access and print the values of built-in adjoint-based sensitivities at each iteration? Also, Could we change the sensitivities and feed them to OPTISTRUCT instead of the default re-built ones? I am looking forward to hearing from you.
Hello, The Topology Optimization in Optistruct by default uses the Direct method for the Sensitivity analysis. I am predicting the result variation using the Adjoint method. It would be of great help to me if I know the way that I can make the solver use the Adjoint method of Sensitivity anslysis. Any help is appreciated! Thanks!