[stuw]

Good morning all,

After 5 years or so messing about in big tanks of water I've recently returned to the joys of CFD. I'm getting on reasonably well, but have a problem with a simulation using the ANSYS CFX 13 immersed solid model. I suspect there is a simple solution to the problem, but having read the documentation and searched I cannot find it.

I'm using the model to run a transient simulation of a tidal turbine, in which the blades rotate at a set speed around an axis in the flow direction. After meshing in ICEM, the domain and body of the turbine are set as one domain, and the blades as a second domain, set as an immersed solid. This works well and the blades rotate as expected, but the blades also remain in their original position regardless of the timestep. This is the problem.

In CFX Pre, the outer surface of the blade appears in the static domain as "Default Domain Default", which causes the problem. However, and this is what I'm asking for help with, I cannot work out how to set the model to remove the outer blade surface from the static domain. I initially thought that this was to do with the setting of the blade surfaces in ICEM as an internal wall or not, but this doesn't affect the outcome.

I have attached a couple of screenshots - one showing the result in Post (a slightly different model but same result) and one showing the setup in Pre with the problematic part of the static domain highlighted.

I wonder if the solution is to somehow model the blade within a cylindrical domain, or whether I need to set up a domain interface, but I hope somebody more experienced might point me in the right direction.

Thanks very much,
Stu

PS. The domain is small and the mesh coarse because this is a preliminary model I'm running on a laptop, I won't be using any results from this model!

[ghorrocks]

You have made a mistake I am seeing a lot on the forum since the introduction of immersed solids. I wish ANSYS made it clearer where immersed solids is an appropriate model and where rotating frames of reference is appropriate.

You should be modelling this as a rotating frames of reference simulation, not an immersed solid. RFR is far more accurate and simpler to implement than immersed solids. Have a look at the rotating machinery tutorial examples for how to set these sort of models up.

[stuw]

Hi ghorrocks,

Thanks for your reply and sorry it's taken me a couple of days to respond. I will do as you suggest and use a rotating frame of reference simulation instead of the immersed solid model.

Maybe I missed somewhere a description of when the immersed solid model is and isn't suitable, but from what I read it did appear to offer at least one option for my modelling. I will continue reading about this too I would like to understand why it is not suitable.

Thanks again,
Stuart

[ghorrocks]

Immersed solid can model this, it is just not a good choice for most cases. The reasons are:
* Immersed solid uses momentum sources to impose the solid body. These momentum sources need a bit of tuning to work properly otherwise you get flow leaking through the blade. RFR on the other hand has the blade as a surface in the mesh so no leakage is possible.
* Immersed solid uses the volume mesh and has no ability to selectively refine the mesh to resolve the blade boundary layer. RFR can handle refined meshes in the blade boundary layer. As most blades require accurate boundary layer modelling to work this is critical.
* On the other hand, immersed solids can mesh the whole domain with a simple high quality mesh whereas a high quality mesh in RFR requires careful work and some skill at meshing.
* And immersed solids can handle motions more complex than rotation, whereas RFR can only do rotation.

[stuw]

Thanks again Glenn, I can see now that I would've hit problems with the immersed solid method as I continued. I'm getting on well using RFR now.