Nonlinear contact setup assistant for assembly simulations

Engineers describe a specific and recurring failure mode: they build a large assembly model, define contacts, run it, and either get non-convergence or results they don't trust — because general contact definitions interact with each other in ways that are hard to predict before the run. 'Cannot perform nonlinear analysis of huge assemblies, general contact won't work' and 'non-linear with separating contacts are not yet supported' are not fringe complaints — they describe the messy reality of contact-heavy simulation on production assemblies.

The gap isn't in the solver math. The gap is upstream, in how contacts are defined and prepared before the job is submitted. Most engineers set up contacts manually, face-pair by face-pair, or use the auto-contact detection built into the preprocessor — which is fast but naive, generating redundant contact pairs, mismatched stiffness settings, and incompatible formulations between adjacent regions. By the time the job fails, tracing the failure back to a specific contact definition is slow and expert-dependent.

This persists because preprocessors like HyperMesh or Mechanical's geometry tree are general-purpose tools not designed around the specific workflow of debugging a contact setup before submission. The engineering judgment required to choose between penalty, augmented Lagrangian, and MPC formulations for a given contact pair — and to match stiffness to local mesh density — is locked inside the heads of senior analysts who learned it through years of failures. Junior analysts inherit their setups without understanding the reasoning.

A focused tool that takes an assembly model's contact definition, analyzes mesh density mismatches across contact pairs, flags formulation choices likely to cause divergence given the loading type, and proposes a consolidated contact deck ready to re-import — that's something a simulation lead would use before every major nonlinear submission. The need recurs every time a new assembly or design variant enters the simulation queue, which in production environments is constant.

What to build

Build a standalone preprocessor add-in for HyperMesh and Ansys Mechanical that reads an assembly's existing contact pair definitions, scores each pair for convergence risk based on mesh density mismatch, formulation type, and local geometry curvature, and outputs a revised contact deck with recommended formulation changes and stiffness values, importable directly back into the model.

Where to start

Target Ansys Mechanical users first through the ACT (Ansys Customization Toolkit) extension marketplace, which gives you a legitimate distribution path inside the tool and a user base already accustomed to installing third-party extensions.

The hard part

Getting read/write access to contact deck formats inside HyperMesh and Mechanical requires either undocumented API work or forcing users to export and re-import files manually — the latter is friction that kills adoption even if the analysis is valuable.

How it makes money

Annual license per analyst seat, priced around $1,500–3,000/year, sold directly to simulation team leads who can expense it without going through enterprise procurement — low enough to be a discretionary purchase, high enough to be a real business at 50 customers.