Validation
612 benchmarks, published
Closed-form solutions, published experimental data and cross-checks against two independent solvers. Every result and every deviation is in the manual.
Benchmark categories
| Category | Cases | Reference | Max deviation |
|---|---|---|---|
| Beam elements, linear | 88 | closed form | 0.01% |
| Plate and shell, linear | 104 | closed form / NAFEMS | 0.4% |
| Geometric nonlinearity | 61 | published benchmarks | 1.1% |
| Material nonlinearity | 74 | experimental data | 3.8% |
| Modal and dynamic | 92 | closed form | 0.2% |
| Consolidation | 48 | analytical / published | 2.4% |
| Slope stability | 39 | published case studies | 2.9% |
| Code checking | 106 | worked examples | exact |
About validation
Yes, the whole suite runs on every release candidate and the results are published with the version. A benchmark that moves by more than its stated tolerance blocks the release.
Yes — the model files are in the examples package, with expected results and tolerances.
That remains the engineer's responsibility, and the report generator is built to make it auditable rather than to hide it.
Yes, per version, in the same manual. A solver with no listed limitations is a solver whose limitations have not been written down.