Mesh Convergence

[haolam]

Hi,

Basing on the precision calculation method used in Algor, how do I know when the results have converged? the precision ranges from zero (best) to 5 (worst), but what is the standard acceptance of precision that the results should be based on so that I won't have to do mutiple runs with finer meshes to determine that the solutions have converged. Thanks.

Hao

[PipePakPat]

Precision does not consider the "significance" of the stress for which the precision is based.  For this reason, a poor precision may be indicated in a region of the model that possesses insignificant stresses.  I mesh convergence study requires no less than three different mesh sizes refined in the regions of interest to determine if the stresses converge or continue to increase which would indicate a singularity.  Singularities are addressed by altering the original geometry to include fine details such as filets or rounds or by applying loads or boundary conditions distributed over a consistent area instead of a specific single node.