Interpret and Validate Thermal-Fluid Results
Use this guide after a thermal or thermal-fluid run completes. The review connects field visualizations with traceable engineering evidence and determines whether the result is ready for scenario comparison, further calibration, or a field test.
Prerequisites
| Requirement | Check |
|---|---|
| Completed run | The expected execution and audit stages are complete and the result can be opened. |
| Frozen identity | Scene, CAD, Model Asset, component, baseline, solver, and parameter versions are recorded. |
| Comparison basis | Baseline and changed scenarios use a common domain, result region, units, and time window. |
| Validation evidence | Measurements, reference cases, equipment specifications, or an approved acceptance method are available. |
| Reviewer | The responsible engineering owner is assigned and understands the intended decision. |
Review sequence
Confirm result identity
Match the result to the frozen scene, CAD, Model Asset, component set, baseline, solver profile, and parameter set. Confirm that the job completed its audit stage and that the result checksum matches the report or evidence record.
Read the fields
| Result | What to inspect |
|---|---|
| Velocity | Direction, recirculation, short-circuit paths, stagnant regions, jet spread |
| Pressure | Relative pressure between zones, supply and return behavior, unexpected blockage |
| Temperature | Hot regions, gradients, supply-to-return change, transient recovery |
| Heat flow | Heat-source response, transport direction, removal path, stored energy |
| Probes and sections | Comparable values at fixed points, lines, planes, or equipment regions |
| Scenario KPIs | Baseline and changed result calculated over the same scope and time window |
Use consistent color ranges and camera positions when comparing field images. Preserve the slice location, plane normal, sampling resolution, and time or averaging window.
Check conservation
Review inlet and outlet mass flow, heat added, heat removed, and energy accumulated in the domain. Small residuals may reflect discretization and averaging, while a large or growing imbalance often indicates an incorrect boundary, unresolved opening, unstable run, or incomplete averaging window.
The evidence record should state the balance method and accepted tolerance for the project. A screenshot alone is insufficient because it does not preserve the calculation scope.
Check convergence and sensitivity
Compare at least one representative result across suitable grid, time, or averaging settings. Focus on the decision metric rather than requiring every cell to match.
- confirm that the run reaches a stable or intentionally transient regime;
- compare a finer grid where geometry or gradients are important;
- vary uncertain boundary values within an approved range;
- compare repeated runs when turbulence or runtime behavior is stochastic;
- record whether the scenario ranking remains stable.
Compare with evidence
Use an applicable validation source:
- a standard benchmark or analytical case;
- an independent engineering result with compatible assumptions;
- measured velocity, temperature, pressure, power, or flow;
- equipment specifications and commissioning records;
- a controlled field test for the selected operating change.
Align units, coordinates, sensor location, time window, and boundary conditions before calculating error or drawing a conclusion.
Decision record
An accepted review should contain:
- the engineering question and selected scenarios;
- frozen input identities and result checksums;
- conservation, convergence, and sensitivity findings;
- measured or reference comparisons;
- known uncertainty and the approved use of the result;
- reviewer, decision, and next action.
Failure handling
| Finding | Next action |
|---|---|
| Balance error grows over time | Check boundaries, domain closure, source units, and solver stability |
| Hot region moves when the grid changes | Refine the affected geometry and repeat the comparison |
| Visual field and probes disagree | Check slice position, interpolation, sampling region, and time window |
| Scenario ranking changes under small input changes | Calibrate the uncertain boundary and collect additional measurements |
| Physics runtime differs while the field is stable | Validate contact, collision, mass properties, and time step in that runtime |