Illustrative: Boundary Condition Mismatch Between Model and Test
How a test fixture provided more rotational restraint than the FEA model assumed, producing a 40% discrepancy in measured vs predicted stress — and what the correlation revealed.
Case type
Illustrative analysis failure. This case demonstrates the common correlation problem of boundary condition mismatch between the FEA model and the physical test.
1. The system or structure
A cantilever beam component tested in a static load fixture. The FEA model used fully fixed (encastre) boundary conditions at the root. The test fixture clamped the beam root between two steel plates.
2. What failed?
No physical failure. The correlation failed: the strain gauge at mid-span measured 40% less strain than the FEA predicted. The discrepancy was traced to the test fixture providing additional rotational restraint that was not represented in the model.
3. The mechanism
The FEA model assumed a fully fixed root (all DOF constrained). The test fixture, consisting of two clamping plates with multiple bolts, provided substantial but not infinite rotational restraint. The fixture was stiffer than the idealised fixed condition in a different way: it distributed the restraint over a finite length rather than at a single plane.
- FEA prediction: encastre at the root plane — zero rotation at the root
- Test fixture: clamped over a 50 mm length — the effective fixed plane was further inboard, shortening the cantilever
- Effect: the shorter effective cantilever produced less bending moment at mid-span — approximately 40% less strain
- Corrected model: the constraint was applied over a 50 mm length matching the fixture. The correlation improved to within 5%
9. Engineering lessons
- Test fixtures rarely match idealised boundary conditions — the fixture stiffness and geometry must be represented in the model
- Strain gauge correlation identifies boundary condition mismatches — a systematic discrepancy is often a boundary condition issue, not a material or load issue
- When correlation fails, investigate the boundary conditions first — they are the most common source of discrepancy
- The corrected model should represent the test configuration, then be re-run with the in-service boundary conditions for the design assessment