[SKennedy_1]

Hello everyone,

My problem is: I have a rotor, which is partially contained by the bore walls, but also has a radial discharge port.

I normally simulate using a fluid-fluid interface between the rotor and discharge port, and assume adiabatic walls for the bore.

However I now want to include heat transfer between the gas and the bore walls, so now need to add a fluid-solid interface...this ultimately means the rotor side will have a fluid-fluid interface and a fluid-solid interface....and it's not possible to assign the same side of the domain to more than one interface.

Oh, I should also mention the rotor is moving, as the mesh changing I can't simply split the rotor wall.

Hope I've explained that Ok.
Any help appreciated.

Stuart

[ghorrocks]

You do not need to model the solid to include heat transfer. You can simply assign a heat transfer boundary condition to the fluid domain face (eg convective) and it will have heat transfer.

The usual approach for modelling lobed impellers like this is using immersed solids. Are you using this approach? If not why not?

[SKennedy_1]

Hi ghorrocks,

Thanks for you response. I may be slightly misunderstanding what is required to undertake the simulation. I was applying similar principles of modelling a simple fluid flowing through pipe problem, in which I have modelled the solid and applied and HTC boundary condition to external wall - to eventually export to thermal analysis.

In my problem, I am modelling a screw compressor, so I was looking to replicate the above problem, except the gas being compressed isthe fluid and the casing is the pipe.

Regardless I will look to apply your suggestion.

Re immersed body - I understand this only to be used for incompressible fluids and any literature I have read indicates this is not suitable for screw compressor modelling.
regards,
Stuart

[ghorrocks]

You can impose the HTC boundary on the fluid domain with no need for a solid domain. Do you need the solid domain for any reason? If it does not contribute significantly then do not model it.

Immersed solid - OK, sounds like immersed solid is not appropriate. So how are you modelling the motion of the rotors?

[SKennedy_1]

OK - cool. I included the solid domain as this will be used in the next stage of thermal and subsequently for structural. I understand the HTC to be a function of temperature, therefore if this is varying in the compression chamber, I was unsure if applying it as a BC would be appropriate.

I really need to get my head into a few more tutorials and HT theory!!

I create the deforming fluid domain in 3rd party software and import a new grid position via a user library at each time step. Quite neat - but takes a will to compute.

Cheers,
Stuart

[ghorrocks]

OK, if you need the thermal profile for subsequent analysis then yes, you need to model the casing. But you can impose the same thermal conditions on a solid domain as a fluid domain, so keep that in mind for future analysis.

OK, so you are using the pre-defined mesh approach. Yes, this is likely to be very slow as it will need to do results interpolation every new mesh. If that is every time step then this simulation is likely to be extremely slow, and the interpolation is likely to significantly smear the variables. I would not be surprised if these problems make the approach unviable.

I think you will find CFX is not really a suitable CFD code for this analysis. You will probably have more luck with Fluent as it has overset mesh options and a few others which CFX does not have. Those approaches are likely to be better than what is possible in CFX.