[Korimbo]

Hello everyone,

I am dealing with an external flow problem - urban flow to be more exact. I ran a simulation using standard k-epsilon turbulence model and it converged very easily and I got very accurate results when compared to experimental data. When I tried to switch to k-omega SST, only thing i changed was that i used specific dissipation rate instead of turbulent dissipation rate for the inlet conditions (i created another field function for the specific dissipation rate which is calculated from the turbulent energy and turbulent dissipation rate that i used in the k-epsilon case). Everything else, including the mesh, was the same. I am using wall functions and y+>30 to resolve the boundary layer near the walls. The residuals actually go very low to about 1e-6 but then start oscillating and the solution clearly hasnt converged.

Does anyone have an idea why k-omega SST could be problematic in that case?

[Peresvet]

Originally Posted by Korimbo

Hello everyone,

I am dealing with an external flow problem - urban flow to be more exact. I ran a simulation using standard k-epsilon turbulence model and it converged very easily and I got very accurate results when compared to experimental data. When I tried to switch to k-omega SST, only thing i changed was that i used specific dissipation rate instead of turbulent dissipation rate for the inlet conditions (i created another field function for the specific dissipation rate which is calculated from the turbulent energy and turbulent dissipation rate that i used in the k-epsilon case). Everything else, including the mesh, was the same. I am using wall functions and y+>30 to resolve the boundary layer near the walls. The residuals actually go very low to about 1e-6 but then start oscillating and the solution clearly hasnt converged.

Does anyone have an idea why k-omega SST could be problematic in that case?

To compare these models, it is necessary to initially have a fine mesh and additionally change the default settings. I would recommend starting the calculation (k - e), and switching to omega-sst in the middle.
For me, all hydrodynamic calculations converge well if the mesh is the same and fine enough for the two models. The mesh for the standard (k - e) model will not work. For small boat sizes, I use about 30 million cells with a standard pool, no extra volume. The prismatic layer is not less than 10 layers, and the initial volume of the water cell near the vessel is less than 0.5 mm3.
No problem. Precision 99.9 between models. Physicality at altitude.
Such tasks for experienced users are at the level of Doctors of alcohol.