Hello,
i was trying to use a superelement approach to consense 80% of my FE-Model to Analyse 1 (and lateron multible) very local loads.
Internal Superelements seemed promissing, but i dont know if my setup is correct.
See the attached .hm and .fem files. Run them to get a sense of the output.
Multiple questions :
- Are the internal Superelements (defined by the SESET) the GRidpoints that compose the superelements (in my understanding the elements that are to be rationalized / condensed)?
- Is the condensation step slower than solving without superlement-setup (does condensation-computetime scale badly when going to e.g. 500k superelements / 50k residual )
- Does one have to define 2 subcases for 1 condensation with static loadstep that is only accounted for in the second subcase ? (1st subcase in .hm file is linear static subcase already)
Why internal superelements?
- i dont need to cut out my model into multible parts per load point→ just redefine the SESET
- with normal static condensation (Direct Matrix Input (Superelements) ) it seems to be nessesary to create multible models for the condensed and residual parts. This wouldnt be feasable if refining or remeshing, nevertheless moving the loadpoint to a different position