Benchmark of computational hydraulics models for open-channel flow with lateral cavities

Area of Science:

• Hydro-environmental engineering
• Computational fluid dynamics
• Open-channel flow modeling

Background:

• Diverse computational models exist for hydro-environmental engineering, including 2D shallow water, 3D turbulence (RANS, LES), and particle-based methods.
• Accurate prediction of mean velocities and free-surface dynamics in complex flows is crucial for engineering applications.

Purpose of the Study:

• To conduct a first-of-its-kind comparison of six distinct computational hydraulics fluid dynamics models.
• To evaluate model performance in predicting mean velocities and free-surface dynamics in open-channel flows with lateral cavities.

Main Methods:

• Comparison of six models: Iber+, HO-SWM, GBVC, OpenFOAM (RANS), Hydro3D (LES), and DualSPHysics (SPH).
• Application to two benchmark cases involving open-channel flows with symmetric lateral cavities.
• Quantification of computational cost relative to model complexity.

Main Results:

• Shallow-water models capture large-scale in-cavity flow structures but show wider shear layers and higher main channel velocities compared to 3D models.
• Three-dimensional models (RANS, LES, SPH) provide improved mean velocity predictions against experimental data.
• Most models capture the transverse standing wave; shallow-water models match theoretical values, while 3D models slightly overestimate it.

Conclusions:

• Three-dimensional models offer superior accuracy for mean velocities in these specific flow conditions.
• Model choice involves a trade-off between predictive accuracy and computational cost, with complexity exhibiting logarithmic cost growth.
• Further research can refine model selection for hydro-environmental engineering applications based on specific needs.