[srotheram]

I'm performing a transient analysis of a hole pattern seal in order to extract rotordynamic coefficients, the problem is similar to a journal bearing with a whirling rotor orbit. The shaft whirling orbit is simulated using a periodic circular mesh motion

I've been using a method described in a paper which correlates with test data. The boundary conditions are the same but my values for fluid forces on the shaft are an order of magnitude too high compared to the published data.

I've checked the boundary conditions, geometry size, obtained a copy of the original ccl file for the motion of the shaft and used meshes of different sizes, densities and y+ from 1 to 50. All with the same result.

Any suggestions?

[ghorrocks]

Your mesh is very coarse, when it comes down to resolving the flow in each of the little cylinders. Your mesh will need to be much finer than that to resolve anything accurately.

But this simulation does not lend itself to modelling the whole thing like you have done. You will find the mesh required to do so is huge. If I was doing this I would try to derive a function where the conditions on a single cylinder can be modelled, then you do a ODE or PDE to solve it over the entire geometry.

[srotheram]

Glenn,

Thanks for the feedback.

I've also tried the model without the small cylinders included, so basically an annulus with a whirling shaft and continue to get fluid forces on the shaft which are an order of magnitude too high.

[urosgrivc]

Is using a ciclic simetry condition a problem in your case? if so why?
for rotation of the shaft it is posible to use tangential wall velocity.

You could use a wery small angle section of your geometry in your case, that way you could use a lot finer mesh and use an interface where it repeats itselve.

[ghorrocks]

First look at the FAQ on accuracy: https://www.cfd-online.com/Wiki/Ansy..._inaccurate.3F

Let us know how you go with the standard procedures to get an accurate simulation.