How to Verify an FE Model
Verification confirms that the numerical model was built correctly. This guide provides the complete verification workflow, from unit consistency through mesh convergence to hand calculation correlation.
1. The Engineering Task
Verify that the FE model was built correctly — that the units are consistent, the mesh is converged, the loads and constraints are properly applied, and the results agree with independent hand calculations.
verification-validation
Verification asks "did we build the model correctly?" Validation asks "does the model represent reality?" Both are needed. This guide covers verification.
2. When to Use This Method
Every analysis used for an engineering decision. The depth of verification scales with the consequence: a quick unit-and-reaction check for screening, a full documented verification for certification work.
3. What You Need Before Starting
- The completed FE model with results
- A hand calculation for at least one location with a known analytical solution
- The unit system used in the model (length, force, mass, stress, density)
- A mesh convergence plan (see How to Perform a Mesh Convergence Study)
4. Step-by-Step Method
- Unit consistency: confirm that all units are compatible. The most common error is density in the wrong system — check that density × gravity × volume produces the correct weight
- Reaction force check: sum all reaction forces and compare with applied loads. For static equilibrium, they must balance to within 0.1%
- Deformed shape review: examine the deformation plot. Does it match the expected physical behaviour?
- Hand calculation correlation: extract the FEA result at a location with a known analytical solution (beam deflection, plate stress, natural frequency). Agreement within 5–10% is typical for a well-meshed model
- Mesh convergence: run at least three mesh densities and confirm the quantity of interest is stable (see How to Perform a Mesh Convergence Study)
- Sensitivity check: vary a key parameter (mesh size near the feature, contact stiffness, boundary condition stiffness) and confirm the result is not unduly sensitive
- Document: record all verification checks in the analysis report
5. What to Check
| Check | Method | Acceptance |
|---|---|---|
| Units | Verify density × g × volume = weight | Exact |
| Reactions | Sum reactions vs applied loads | Within 0.1% |
| Deformation | Visual + hand calc comparison | Physically plausible |
| Hand calc | FEA vs analytical at known point | Within 5–10% |
| Mesh convergence | Three mesh densities | < 5% change on last refinement |
| Sensitivity | Vary one parameter | No large change in result |
6. How to Interpret the Result
A verified model is one where all checks pass and the results are documented. The verification does not prove the result is correct in absolute terms — it proves the model was solved correctly. Validation against test data is a separate step. Report the verification checks alongside the engineering results.
7. Common Mistakes
- Confusing verification with validation — verification checks the maths, validation checks the physics against reality
- Skipping the hand calculation because "the software is reliable" — the software is reliable, but the input may be wrong
- Not checking units — mixed units are the single most common source of incorrect FEA results
- Reporting mesh convergence without showing the convergence ratio
- Documenting verification informally — for certification work, it must be in the analysis report
8. Further Reading
See the Verification, Validation & Model Quality Knowledge category for V&V theory. See How to Perform a Free-Body Check on an FE Model and How to Compare FEA with Hand Calculations for detailed sub-procedures.