Greetings everyone,

I am currently working on a Lagrangian–Eulerian phase method to simulate the effect of sand particles on riverbed erosion, incorporating hull–bank interaction. The simulation includes DFBI, VOF (air and water as Eulerian phases), DEM (for sand particles), segregated flow, and the k-epsilon turbulence model.

The computational domain is 70 meters in length and 30 meters in width (with half-domain symmetry applied), and the ship model is 18 meters long. Boundary conditions consist of a velocity inlet, pressure outlet, smooth wall boundaries, and a symmetry plane.

The issue I am facing is that the simulation becomes significantly time-consuming due to the DEM sub-stepping.

Here are some simulation parameters:

• Flow velocity: 5 m/s
• Total cell count: 521,826
• Time step size: 0.001 s
• Maximum inner iterations per step: 2
• RAM of the system used - 48GB
• 14 cores, 2300Mhs

I would greatly appreciate any suggestions to improve the computational efficiency of this setup.