Cross-solver results comparison for validation engineers

Multiple complaints point to a specific and genuinely puzzling problem: running the same physical model through different solvers — or even through the same solver accessed via different front-ends — produces results that don't match. 'Differences in results when using OptiStruct as the solver compared to HyperMesh as the preprocessing software,' and 'results from SimSolid aren't comparable to what we get in other Altair products.' Engineers who need to validate a design can't tell whether a discrepancy means one solver is wrong, the model setup differs between tools, or the underlying physics assumptions differ between solver methodologies.

This gap persists because it's structurally awkward for any single vendor to acknowledge it. If Altair built a tool that showed SimSolid diverging from OptiStruct on a given geometry, they'd be documenting their own product's limits publicly. So no vendor builds this honestly. The user is left with the only option: run both, manually compare stress contours in different result viewers, and use judgment to decide which to trust — which is not a defensible engineering process.

What doesn't exist is a workflow that takes two result files from different solvers, normalizes the output to comparable quantities (von Mises stress, displacement, natural frequency), and flags the regions and load cases where the two diverge beyond a user-set tolerance — with enough context to diagnose whether the divergence is likely a mesh density difference, a solver assumption difference, or a model setup error.

This is a business because validation is a mandatory step in regulated industries — aerospace certification, automotive FMEA, medical device structural analysis. The engineer responsible for signing off on a simulation result needs a defensible paper trail. Right now they don't have one for cross-solver comparison. That's a compliance gap, not a convenience gap, and compliance gaps get budget.

What to build

Build a desktop app for Windows (where the CAE tools already run) that imports result files from at least two of OptiStruct .h3d, Ansys .rst, and Nastran .op2, maps them onto a common mesh reference frame, computes scalar field differences per node or element, and exports a structured comparison report showing divergence hotspots with the magnitude and location of discrepancy — suitable for inclusion in a design validation dossier.

Where to start

Start with OptiStruct .h3d versus Nastran .op2 comparison only — both use the same underlying element types and the mesh-mapping problem is far simpler, letting you deliver a trustworthy first version to structural analysts who are already familiar with both formats and immediately understand the value without needing to explain interpolation methodology.

The hard part

Mapping results from two different meshes onto a common reference frame — especially when one is a meshless method like SimSolid and the other is a traditional FEM result — requires interpolation decisions that will be scrutinized by the very auditors and DVE leads who are your customers, so a technically wrong shortcut here destroys credibility entirely.

How it makes money

Per-project license for consultancies who bill validation work by program, plus annual site licenses for OEM validation teams who run ongoing comparison workflows across multiple programs simultaneously.