Dear.
I have an issue with my simulation on ABAQUS. I simulate a crack in a composite laminate [0/45/0] using cohesive surfaces. The crack is therefore explicitely modeled using a traction-separation law and a Benzeggagh-Kenane progressif damage criterion.
My issue appears when ABAQUS calculates the quadratic stress criterion. Indeed, when I calculate manually the criterion using CPRESS and CSHEAR or S12 and S22, I find the right value. However at the nodes before last in the thickness (cf figure), the criterion is over-evaluated by 7%. It leads to the beginning of damage at these nodes instead of the center of the ply.
My questions are :
Why the value is not right only at theses nodes ?
How the quadratic criterion is calculated (which variables are considered) ?
How to fix this problem ?
Thanks.
PS : The figure shows only the 45° ply and the fracture surface as soon as quad criterion meets 1.
PS2 : Periodic conditions are applied around the cell, Surface-to-surface discretization, thinner meshes show the same issue at the before last nodes.
Hi,
Did you read the value from nodes or centroid of elements? I am not sure but may be the problem is referred to the integration point in surface elements.
Hi,
Thanks for answering. Actually I read values at nodes, CPRESS, CSHEAR and CSLIP are variables from elements nodal and the cquad criterion is found at unique nodal. It is may be due to averaging element nodal variables or boundary discontinuities.
I solved a part of the issue by meshing finer at the boundary (using bias ratio to keep the same number of elements in the thickness). With an element small enough, the standard error becomes negligible. However I still wonder how CSQUAD is calculated and how variables used are extracted.
apart from fine meshing you need to look at the values of stiffness i.e. knn, kss, ktt. these values are usually high then default values taken by abaqus. you can manually put these values using E/t...where E is modulus and t is adjacent thickness. for typical epoxy the values may range from 10^12 to 10^15 or greater. this may increase your computational time too.
- Badges
- Science topic