[mbranag]

I am creating the fluid film portion of a fixed geometry hydrodynamic bearing. I have used an external code to determine the eccentricity of the rotor. I have attached a picture of the fluid domain. I have been able to run the model when laminar and match it up with the external code which has been experimentally validated. The bearing is a submerged bearing.

The fluid is isothermal. The external wall of the cylindrical-shaped domain is model as a stationary wall representing bearing pad surface. The internal wall is model as a rotating wall, which is rotating 120,000 rpm about the eccentricity of the rotor (representing the rotor). The axial edges of the bearing and the output (right end in the attached photos) are all modeled as 0 psi 'Opening Pressure and Direction' boundary conditions. The inlet boundary has been modeled several different ways. It is modeled as a 0 psi pressure boundary condition in the external code. I have modeled it as a 0 psi 'Total Pressure' and 'Static Pressure' and saw no difference. I have modeled the inlet as an opening with 0 psi 'Opening Pressure and Direction' and also saw no difference. I have also modeled the inlet as a normal direction velocity that is has a linear relationship to the radial distance from the eccentricity of the rotor.

The issue that I am seeing is that when I add turbulence to the model I am getting a drop in the pressure profile. This can be seen in the two different excel plots for the pressure at the axial center of the film. This disagrees with what has been seeing experimentally. The pressure should increase as the flow becomes turbulent. This study started out at 8000 rpm where the flow was actually laminar and I have increased the rotational speed of the rotor to 120,000 rpm to ensure that the flow should be turbulent. I have used the k-Omega, BSL Reynolds Stress, and Zero Equation. At 8000 rpm I was seeing no difference in any of the turbulence models. At 120,000 rpm I am seeing some differences but all of the still have less pressure than the laminar case.

If anyone has any ideas on how to overcome this issue, that would be wonderful.

[mbranag]

I have been looking at the theory for specifically the k-Omega model. Looking and multiple different contours for the eddy viscosity, it never goes negative. Therefore this should be modelling an increase in viscosity which should increase the pressure. I am not sure how this decrease in pressure is occurring from a theoretical stand point.

Thanks!!!

[ghorrocks]

What does a literature review say on this issue?

I would have a close look at the simulations to try to find out where the difference is coming from. As you said, turbulence models are dissipative so they should only make the pressure drops larger, not smaller. So see if the streamlines are different, are they short circuiting, or something like that.

Also: Have a look at the length scale of the turbulence. As your domain is so thin you may well have significant anisotropy. This means the two equation models will not be suitable. And I don't think RSM has models for anisotropy based on wall proximity either. You might need to do a LES approach to capture the anisotropy.

[mbranag]

Literature and discussions with engineers in industry all say that the pressure should increase with turbulence in hydrodynamic bearings.

What variables do you think would be particularly helpful in understanding this phenomena. The streamlines all look as you would expect. They enter through the leading edge of the bearing and leak out the trailing edge and axial edges of the bearing.

This problem should be solvable in CFX as I have seen studies in literature that were able to correctly predict the pressure rise using CFX.

[ghorrocks]

Please read my post #3 again. It mentioned a few issues to look at.