Simulation Validation and Accuracy — 3D PyBullet

USE WITH CAUTION

One or more checks are not strong enough to call the model fully validated.

18 of 18 trials saved.

Key metrics

Condition MAPE
18.13%
Median condition error
16.95%
Worst condition error
34.65%
Mean X/Y profile disagreement
1.44°

Condition-by-condition

ConditionTrialsPhysical meanSimulation meanErrorSimulation SD
Unsieved Quartz Sand3 / 329.75°32.08°7.82%1.60°
Quartz 125-250 um3 / 331.35°36.93°17.80%2.34°
Quartz 250-500 um3 / 329.68°34.74°17.07%1.07°
Quartz >500 um3 / 330.83°36.02°16.83%0.90°
Glass Beads - 1 Cup3 / 319.09°21.88°14.59%0.42°
Glass Beads - 2 Cups3 / 319.55°26.32°34.65%1.80°

Interpretation

Internal quality includes repeatability, settling, and orthogonal-profile consistency. External validation compares completed simulation means with the known physical experiment. The physical values do not directly control individual research trial results.

Model limitation

This version uses true 3D rigid-body dynamics. Glass beads are smooth spherical rigid bodies. Quartz sand uses a rough-sphere DEM proxy with high sliding, rolling, and spinning resistance plus near-zero restitution so coarse particles dissipate impact energy like a granular bed rather than behaving like bouncing rocks. Passing this report validates the frozen model only for this specific test set, not for every granular material or apparatus.