Fluid-implicit particle solvers in a video gaming engine provide new perspectives in hydroinformatics

ABSTRACT

This paper proposes a game-engine-based workflow for modeling flows in hydraulic structures. A vertical-slot fishway is simulated with a semi-Lagrangian PIC-FLIP solver in Unreal Engine's Niagara Fluids. Three discharges (low, average, flood) are compared with physical flume experiments and an Eulerian OpenFOAM model, both serving as references with different error sources rather than ground truth. The OpenFOAM simulations were structured at the same spatial resolution as the Niagara Fluids runs and should be regarded as an approximate numerical comparison rather than mesh-converged benchmarks. The PIC-FLIP model is calibrated against water depth and velocity at four transects by tuning five parameters. Formal equivalence tests confirm water depth agreement within ten percent margins in most cases, where velocity equivalence is limited, particularly at low flow near high-velocity jets, sharp corners in the vicinity of slots, and low-density zones. The video-engine simulations run an order of magnitude faster than OpenFOAM in the hardware environments used here, enabling immersive 3d visualization and post-processing. Despite the absence of a turbulence closure, video game engines can provide fast hydraulic predictions and new, complementary analysis modes, including a demonstrative virtual fish agent and other metrics extracted from logfiles.