Selective reducede integration: possible improvements
Current selective reduced integration is structured as a separate split_type
named AMOR_SELECT
. First possible improvement is to integrate selective reduced integration passing a flag directly to the split_type
named AMOR
.
Always in the current version, in equilibrium assembly for AMOR_SELECT
, within the element loop, we do the following:
- Check if we need to activate selective reduced integration (using a threshold on p_field)
- If no, classical full integration loop for volumetric + deviatoric contributions (1 GP loop)
- If yes, we do full integration for the deviatoric component (1 GP loop) + we do reduced integration for the volumetric component (1 GP loop) Hence, we now write 3 GP loops. As an alternative we can split it in just 2 lopps: -Full integration for deviatoric compomenent + volumetric if the reduced integration is not activated -Reduced integration for volumetric component if the reduced integration is activated