[stud-many]

Hey there.

We have to explain differences from cfd-calculations to experements of ~6% polytropic efficiency on a small transonic single-stage diagonal compressor. Our current idea is to use the deformed blade geometry by a hot-cold-transformation (ANSYS Mechanical) for new cfd-calculations. As the flow is supersonic in most operation-points we expect a sensitive behavior on changes in the geometry.

The problem now is, that the only dataI can extract from ANSYS mechanical is STL / facetted faces and I can't really use them for our meshing-strategy (we'd like to use Autogrid5 for meshing as we did before).

Right now I only have to Options in my mind.

--> import an old mesh (cgns) in icem and transform it to the stl-geometry

It's possible to import the blocking from Autogrid5 and to create a new mesh. But I don't know if this really works on the stl-geometry and it would need some effort to work with this geometry

--> calculate the deformation of the *.geom from the deformed stl-geometry. this could be a solution, but right now I think it would need lots of working hours to get extract the right data from millions of node-positions in the stl-file.

I would think, that there has to be an established method as this is a common problem in turbomachinery. Any help is appreciated!

[ghorrocks]

Why can't you just use the built in FSI capability? Why are you doing this all yourself?

[stud-many]

Good point and thanks for your answer. Probably I did not think about this because I never used this. But the problem is, that we used an unstructured mesh for the FEM-Simulations and we need to use a structured mesh for CFD.

The most important aspect to be examined is the correlation CFD <--> deformation due to centrifugal forces (60.000 rpm). I don't know any way to do that automatically.

(my current approach is now allign the deformed stl to the ideal geometry and then deform/project all control-points of the ideal geometry on the stl-surface...that is still not a really nice solution...)

[ghorrocks]

But the problem is, that we used an unstructured mesh for the FEM-Simulations and we need to use a structured mesh for CFD.

This is irrelevant. FSI can handle different mesh types in the different coupled simulations.

Have a look at the ANSYS FSI capability. There are many tutorials on it to get you started.

You don't want to write your own FSI coupling system. It is very complex and I suspect you have better things to do with your time. Use the FSI coupling already available in CFX and ANSYS Mechanical.

[stud-many]

I really don't want to write a FSI myself. I just want to deform my CFD-mesh from a FEM-solution that was calculated with CFD-results. I thought this is the easiest way (I'm still not that familiar with this software).

As far as i understand FSI by now, I need to solve the mechanics and fluid mechanics in one calculation (system coupling, mesh deformation), right? This would be new to me, but it sounds reasonable.

(Thx again for your input!)

Edit:

Got it! THX!!

[stud-many]

Hey there. Again: thanks for your help. But i have (two) new issues with this calculation and i can't really figure out how to solve them (I'm still not that experienced, especially with ANSYS Workbench and Mechanical)

I did a steady-state FSI with one operationg point and the results look promissing. But at this operating point, the fluid flow is not transsonic. To get to more interessting OP's i need the previous result as initial values.

I solve the case on a cluster by submitting the job in a batch-file. In these batch-commands i start ANSYS Workbench and load a journal file which then executes the solving and afterwards saves the project. Do I have to use an initial results file for both systems, Mechanical AND CFX (because of the mesh deformation?)

The other thing is way more important. For my purposes it's fine only to transfer the mesh-displacement. But the calculation took 50hours with 4 Coupling-Iterations a 500 iterations in CFX. I don't think this is really necessary as this is a steady state-case. However I can't find any tutorials for mechanical --> cfx one-way fsi where the mesh-displacement is only calculated at the beginning. Can anybody help here?

My WB-Setup:
WBProject.jpg

The Coupling-System-Setup. After the first calculation i needed to raise the "Number of steps" to two. I don't really understand why but it looks like the first calculation is defined as Step 1 and the new calculation as Step 2?
WBProjectSysCoup.jpg