Track validation evidence, assumptions, and acceptance criteria for simulation quality assurance.
Simulation results are only useful if you can trust them. SimPilot's validation framework helps you build and maintain that trust -- tracking assumptions, evidence, acceptance criteria, and reviewer signoff for every simulation project.
Why validation matters
Engineering simulation involves approximations at every level: geometry simplification, mesh resolution, turbulence modeling, boundary condition assumptions. Without systematic validation, errors compound silently. SimPilot makes validation a first-class part of the workflow rather than an afterthought.
The validation framework
Each project maintains a structured validation record with five components:
Intended use
What the simulation is being used for. A pressure drop estimate for sizing has different accuracy requirements than a stress analysis for a flight-critical component.
Assumptions
Documented modeling assumptions: steady-state vs. transient, turbulence model choice, material properties, boundary condition simplifications. Each assumption is tracked with its justification.
Acceptance criteria
Quantitative thresholds that define "good enough." For example: mesh convergence within 2%, drag prediction within 5% of wind tunnel data, maximum stress below yield with a 1.5x safety factor.
Evidence
The results and checks that demonstrate criteria are met. Mesh convergence studies, comparison with experimental data, conservation checks, grid independence plots.
Deviations
Where results don't meet criteria, and the engineering justification for accepting (or not accepting) the deviation.
Auto-validation
After a simulation converges, SimPilot automatically runs a set of quality checks:
Check
What it verifies
Mass conservation
Net mass imbalance across all boundaries is within tolerance
Momentum conservation
Force balance consistency across the domain
Empirical correlation comparison
Results compared against standard engineering correlations (e.g., Moody chart, Nusselt correlations)
Plausibility screening
Physical quantities are within realistic ranges (no negative absolute pressures, no supersonic flow in a low-speed case)
Auto-validation results are attached to the simulation as evidence and flagged in the chat if any check fails.
Auto-validation is non-blocking
Failed auto-validation checks produce warnings, not hard failures. The simulation results are still available -- the warnings help you decide whether to trust them.
Reviewer signoff
For formal validation workflows, simulations can be submitted for engineering review. Reviewers examine the case setup, results, and validation evidence before approving or requesting changes. See Engineering Reviews for the full workflow.