Probabilistic FEA Best Practices
Best practices for probabilistic finite element analysis — model verification before propagation, input definition, automation, result interpretation and reporting.
Before starting
- Verify the deterministic FEA model before adding uncertainty
- Identify which inputs are genuinely uncertain — not every input needs a distribution
- Assess computational budget — how many FEA runs are feasible?
During the analysis
- Automate model generation, execution and result extraction
- Test the automation on 3-5 cases before the full campaign
- Monitor for solver convergence failures — discard failed runs and investigate
- Use LHS for efficiency if more than 50 samples are needed
After the analysis
- Check output distribution for physical plausibility
- Assess convergence — are results stable with more samples?
- Perform sensitivity analysis to identify dominant inputs
- Document the complete chain from question to result