Optistruct vs Abaqus
Hello everyone,
can anyone tell me the deviation percentage in stress results between optistruct and Abaqus results?. I have compared the results for static analysis and was shocked to see the difference between the solvers stress results of about 9%. Eventhough both the solver results are an approximation, deviation of about 9% is huge. When I compared the P1 Major strains the results were almost similar but results in VonMises stresses gave me a huge difference. Does anyone observed the similar difference in the results. If yes what might be the reasons?. (I have also used the same mesh in both the solvers).
Regards,
Punith Gupta
Answers
-
Hi Punith,
Is the Abaqus run completely linear static or the load is divided?
OptiStruct runs linear STATIC analysis but I think in Abaqus, load gets divided.
0 -
Hi prakash,
it is a normal linear STATIC analysis.
Regards,
Punith Gupta
0 -
Punith,
Can you share both the input decks?
0 -
Prakash,
sorry I cant share those files. But I am sure that only onestep load has been used for static analyis. I know that there is a RAMP option which is onestep load and the other one is you can specify the userdefined function for the load in Abaqus.So load cases are similar.I have also switched on the USE CORNER DATA option in postprocessing. I dont understand why this differences. What do you think? May be the algorithms implimented for computing the VonMises are different?. This is just a guess because when I look at P1 major strains they are almost same.
How much deviation do you expect between the solvers? Any Idea?.
Regards,
Punith Gupta
0 -
Punith,
I did some benchmark between Abaqus and OptiStruct and the resutls are almost similar (difference is fractional)
Did you output the results to corner data in OptiStruct?
0 -
Hi Prakash,
thanks for the reply. There was difference in constraints modelling. Now I got the same results with a small difference. This is just for only one type of wheel and the results had to be compared for the other type of wheels as well. Your answer was helpful to me and I expect the same results now for the other models.
Thanks and regards,
Punith Gupta
0 -
Hi Punith,
I am not sure about the code of both the solvers but the background basics are same /emoticons/default_smile.png' srcset='/emoticons/smile@2x.png 2x' title=':)' width='20'>
0