Parallel processing solves FEA problems faster

Using DDM (domain decomposition methodology) codes, non-linear finite element solver MARC K7.3 divides solid models into domains which are solved in parallel on separate processors. Results are then re-integrated to create a combined analysis solution for the entire model.This makes it easy to throw more processing power at a problem, either through multi-processor workstations or, appealingly, through several computers networked together.

The result is a shorter processing time for a given problem, but it also means you can solve highly complex problems reliably on computers that might simply have crashed previously when presented with such a model. The company reckons that, with the new system, problems that could only have been solved overnight previously could now be completed in a few hours.

Importantly, parallel processing could be a key enabler for analysing whole assemblies where previously you might only have been able to do component analysis.