From Human to Height-Field: Predictive Shotcrete Simulation with a Physics-Informed Particle System

This paper presents a validation study of a real-time shotcrete simulation model based on a physics-informed particle system. The model emulates concrete spray dynamics by replaying empirically recorded nozzle trajectories from expert and novice operators in a controlled virtual environment. We evaluate its ability to capture skill-related differences in key sprayability metrics, including material adhesion, rebound, and cohesive failure. The results reveal statistically significant distinctions between expert and novice performance, confirming the model’s sensitivity to operator expertise. The simulation also demonstrated strong repeatability and robustness despite inherent stochasticity. While the lack of physical ground truth and the absence of the material's lateral motion are noted limitations, the findings support the model's potential use in training and skill assessment. Future work will focus on improving material modeling to enhance realism and predictive accuracy.