DOE (Design of Experiments) studies systematically explore how input parameters affect simulation outcomes. Instead of running one-off simulations and guessing which parameter matters, you define the space and let SimPilot map it methodically.
Single-variable sweeps
Vary one parameter while holding everything else constant. This is the simplest way to understand a parameter's effect on your results.
"Run this pipe flow at velocities 5, 10, 15, 20, and 25 m/s"
SimPilot creates 5 simulations, each identical except for the inlet velocity. Results are presented as a parameter-response curve showing exactly how the output changes with the swept variable.
Multi-variable sweeps
Vary multiple parameters simultaneously to capture interaction effects:
"Sweep Reynolds numbers 1000, 5000, 10000 with turbulence models k-epsilon and k-omega-SST"
SimPilot computes the Cartesian product of all parameter values. In this example, 3 Reynolds numbers x 2 turbulence models = 6 combinations. Each combination runs as an independent simulation.
The total number of combinations is capped by your plan. If the product exceeds your cap, you'll be asked to reduce the parameter space. See
Batch & Parameter Sweeps for plan-specific limits.
Design space coverage
For larger parameter spaces, SimPilot supports structured sampling strategies:
| Strategy | Best for | Description |
|---|
| Full factorial | Small spaces (< 50 runs) | Every combination of every parameter level |
| Latin Hypercube | Medium spaces | Space-filling design that covers the domain evenly with fewer runs |
| One-at-a-time | Screening | Vary each parameter independently to identify dominant factors |
The AI selects the appropriate strategy based on the number of parameters, their ranges, and your compute budget -- or you can specify one explicitly.
Results
After all experiments complete, you receive:
Parameter-response tables
A structured table with every input combination and its corresponding output metrics. Sort by any column to identify trends.
Main effects
How much each individual parameter influences the output, averaged across all other parameter levels.
Interaction effects
Where two parameters together produce an effect that differs from the sum of their individual effects. Critical for identifying non-obvious coupling.
Auto-generated visualizations
The study dashboard includes chart types selected for the data:
- Line plots: Parameter vs. response curves for single-variable sweeps
- Bar charts: Side-by-side comparison of output metrics across runs
- Tables: Full parameter-response matrix
- Parallel coordinates: Visualize all parameters and responses simultaneously -- trace individual runs across axes
- Heatmaps: 2D parameter interaction matrices showing output intensity
- Scatter plots: Pairwise parameter-response relationships
- Surface plots (3D): Response surfaces over 2-parameter domains
- Contour plots: 2D slices through the parameter space
- Parallel coordinates with interactive filtering to drill into regions of interest