[Dedok]

Hello, could you help me to solve my problem, please?

I've tried to receive a pressure distribution along bearing face. I have a running shaft, stationary antifriction surfaces of bearing and a gas film between them. So, I’ve prepared an axially symmetric geometry with three wedge-shaped clearances for a CFX calculation. I have a clearance in wedge-shape gap 0.05 mm. In this case, I did the structured mesh in ICEM CFD with a quality of hexa-core elements higher than 0.5, 40 elements through-the-thickness of clearance and 10 elements across the width. I've attached 3 pictures with a geometry, boundary conditions (left and right sides of gas domain connect via Translational Periodicity) and results.

I know that in a real situation it should be their own peak of pressure before each wedge-shaped clearance. I've tried to calculate this task with different boundary conditions, for example, all heat transfer options, turbulence and etc. Nevertheless, the right ones are Isothermal and laminar (it's approved by various references). Why do I receive the same picture with all boundaries? It's always the gradient of pressure: Pmin after the first clearance and increasing of pressure to Pmax across the inner surface before first clearance, ignoring another 2 wedge-shaped microchannels.

How should i fix this issue? Maybe CFX can't calculate right such microchannels or this is due to the enclosed volume and considered length (1 mm), although i received the right picture of Couette-Taylor flow in easier task of coaxial surfaces without wedge-shaped microchannels.

[evcelica]

Why are you using translational periodicity? Why not just symmetry on the front and back instead.
If symmetry is appropriate, then model as only 1 element thick.
I have a feeling this may be related to the periodicity, as that can be difficult to solve. I feel much more so with a compressible fluid.

[ghorrocks]

Can you attach an image of what you intend to model and how you modelled it? Your CCL file would help as well.

I use CFX to model MEMS structures where the dimensions are much smaller than yours, some are sub-micron. Your geometry seems huge compared to mine. CFX works fine, as long as the Navier Stokes equations and boundary conditions assumptions remain valid.

[Dedok]

Because i modulate a task of endless cylinder.

[Dedok]

Originally Posted by evcelica

Why are you using translational periodicity? Why not just symmetry on the front and back instead.
If symmetry is appropriate, then model as only 1 element thick.
I have a feeling this may be related to the periodicity, as that can be difficult to solve. I feel much more so with a compressible fluid.

Because i modulate a task of endless cylinder.

[Dedok]

Originally Posted by ghorrocks

Can you attach an image of what you intend to model and how you modelled it? Your CCL file would help as well.

I use CFX to model MEMS structures where the dimensions are much smaller than yours, some are sub-micron. Your geometry seems huge compared to mine. CFX works fine, as long as the Navier Stokes equations and boundary conditions assumptions remain valid.

I've attached images. In addition, i've recieved 3 peaks before wedge-shaped microchannel only in case when rotational inner surface of shaft was coaxial with outer surface. If i create an eccentricity, for example, shift down the inner surface to stational outer surface, there is only one peak again.

[evcelica]

Endless cylinder should use symmetry planes, and be 1 element thick, swept mesh. This is essentially a 2D simulation.