Analysis check

[feaplastic]

Hello,

I ran a geometric nonlinear analysis, my results are quiet ok. But i want to ensure analysis has been done in a right way. For example in another thread, I read something like strain energy density must be converged for a correct analysis. But now, I could not find that thread abt point load.

Anyone can explain about methods available to crosscheck the analysis results. Even basic tips would be also fine since my knowledge is poor.

Thanks

feaplastic

 

[xerf]

1. Perform the analysis with elements size=h and max_time_increment=dtmax

2. Perform the same analysis (keeping the same element type) with with elements size=h/2 and max_time_increment=dtmax/2

3. Check the convergence of maximum displacements in the 2 cases.

4. If the displacements field (primary dependent variable)is accurate then validate the secondary dependent variables such as stress and strain fields. The convergence of secondary dependent variable with mesh refinement is typically slower since their computation relies on the primary dependent variables (i.e. displacement fields) and uses additional approximations. The assesment of stresses/strains can be more subjective. You can look at the stresses at integration points of elements in the regions of interest. Strong variations of strains/stresses at neighbouring integration points is a bad sign and indicate that a more refined mesh should be used in that region in order to obtain more accurate stress/strain fields.

5. Depending on the problem and element type used, you should check the results of typical FEM artifacts such as hourglassing, shear locking , volumetric locking.

 

[feaplastic]

Thanks xerf. I performed analysis with different mesh sizes and checked my results. I dont see any erratic results. Anyhow, I dont have an idea about your 5th tip. I tried in abaqus manual to find out without success. If possible, give me some links to learn. Thank you once again.

If my memory is right, User Mr.Prost wrote abt strain energy density convergence. Hope he will see this thread and reply.

feaplastic