Casting intelligence.
In seconds.
See metal fill, cooling fronts move, and a casting deform. Ace-Mini predicts casting fields from part geometry, gate placement and gravity—so you can explore a setup before committing to a full solver run.
Try the prediction lab →01 / More ideas. Less waiting.
Compare the average time to generate casting fields with the recorded runtime of a leading industry casting solver, across the complete comparison set.
02 / See the physics, side by side
Pick a part and press Play. The leading industry casting solver appears on the left; Ace-Mini appears on the right. Every physics view shares the same geometry, with linked cameras and matching color scales.
03 / Agreement you can measure
Average model-to-solver agreement across the complete comparison set. Region agreement measures liquid/air or solid/liquid classification; overlap measures how closely the predicted regions occupy the same space.
Explore detailed field measurements
Relative error = total absolute deviation ÷ total solver reference magnitude × 100, averaged across comparisons. Displacement uses vector magnitudes; temperature uses Kelvin. Zero-reference cases are N/A. Errors can exceed 100%.04 / Small model. Spatial predictions.
A geometry encoder describes the part. A spatial field network combines that representation with gates, gravity and time to predict filling, temperature, solidification and defect indicators. A second network predicts deformation through cooling and demolding.