[DiegoLopez]

Hello,

I would like to solve a flow past two rotating cylinders of finite span. My goal for this simulation is to gain understanding for the relationship between incoming flow speed, cylinder rotational speed and forces acing on the body for a situation where the cylinders are placed on a free stream.

The domain is shown schematically here:

domain.png.

The diameter is 3 meters and the span is 25 meters.

I have meshed this using ICEM and achieve a good quality mesh, although with a high aspect ratio due to the refinement applied to the cylinder walls. I attach a few images of the mesh.

The material for the simulation is ideal gas air, the operating pressure is 101325 Pa.

I am setting the inlet BC to velocity inlet, with an x-component of 40 m/s. The outlet is set to pressure-outlet with 0 gauge pressure.

Every domain wall other than the cylinders is set to specified shear = 0.

The cylinders are set to rotational wall (same direction vector for both) with a speed of 40 rads/sec.

The turbulence model is k-w SST with inlet turbulence intensity 5% and turbulent viscosity ratio of 10.

I am solving this on steady-state using second order upwind spatial discretization and pseudo-time method.

On the hybrid initialization step, I get the following warning message:

Stabilizing ads-0 to enhance linear solver robustness.

Before the first iteration of the solver, I get this warning:

WARNING: The wall motion has a significant normal component on 15128 faces of face zone 13. (one of the cylinders
The maximum angle between the velocity and face surface is 89.4 deg at (-1.456e+00, 5.641e+00, 2.500e+01).
AMD EPY The tolerance for this check (currently 20.0 deg) is controlled by the rpvar 'wall/vnormal-tolerance'.

n6 Please check your setup. This warning will not be written again.

Which is strange, because I am only imposing a rotation to the wall, as shown here
wall_bc.jpg

During the solve, the following messages are frequently printed:

Stabilizing temperature to enhance linear solver robustness.
Stabilizing pressure coupled to enhance linear solver robustness.


The solution eventually develops negative k:

Negative k in 28714 cells after linear solve.

And it diverges.

I am not sure why I am not able to get a correct solution. Could anybody give me a hand on this?

Thanks in advance

mesh_top.jpg
mesh_cylinder_zoom.png
mesh_tip.png