FE Model Validation Checklist
Review prompts for model validation against test data — strain-gauge correlation, modal correlation, measurement uncertainty and acceptance criteria.
Purpose
For validating that the FE model represents the real physical structure adequately for its intended use.
Correlation Planning
- Are correlation acceptance criteria defined before comparing results? — What level of agreement is required for the decision?
- Is the model validated for the specific use case — not just "it matches test"? — Valid for displacement may not be valid for local stress
- Is the test configuration documented — boundary conditions, load introduction, instrumentation? — Test must be understood before correlating
Strain-Gauge Correlation
- Is FE strain extracted at the exact gauge centre location and orientation? — Not the nearest node
- Is the gauge integration length accounted for? — Gauge averages strain over its active length
- For rosette gauges, is each arm compared independently? — Do not average
- Is the comparison at the correct load level, with linearity verified? — Scale if test was at a different load
Modal Correlation
- Are natural frequencies compared (typically within 5%)? — First 5–10 modes or up to the frequency of interest
- Are mode shapes compared using MAC (target > 0.8)? — Frequency alone is insufficient
- Are the test boundary conditions consistent with the FE model? — Free-free test vs constrained model will not match
Uncertainty and Over-Tuning
- Is measurement uncertainty accounted for (typically 2–5% for gauges)? — A 3% match may be within test uncertainty
- Has the model been adjusted to match test without physical justification? — Adjusting 10 parameters to match 5 data points is curve-fitting
- Is the validation evidence documented with acceptance status? — Validated, partially validated, or not validated