[DaveR]

Hi all,

I am new to openFOAM and have been trying a range of 2D and 3D cases. Currently, I am attempting to model vortex shedding over a 2D sphere at low Reynolds number. I have been hoping to get the lift and drag coefficients out for validation of my methods.

My methodology has been to create a 3D sphere utilising snappyHexMesh - centered at the origin - and then to define my blockmesh to intersect only one half. When this has meshed, I use extrudeMesh to extrude the intersecting half 1 cell in the z-direction.

To fix Re = 100, my parameters are set to:
L = dia of circle = 0.07 m,
u = 0.021 ms^-1
nu = 1.48e-05 m^2s^-1

I have tried solving using both pisoFoam and icoFoam set up for a laminar flow.

Vortex shedding is well illustrated, however my drag coefficient is consistently low at ~6e-03.

I understand there is a comflics tutorial on YouTube, so I re-ran my simulation set up setting the variables in the same way (l = 2, u = 1, nu = 2e-02) and get the exact same results.

Adjusting all simulation parameters result in essentially the same output for the drag coefficient, so I feel as if I've exhausted all simulations options other than my mesh - which is relatively crude. I cannot for the life of me get the boundary layers to grow however as soon as the mesh is somewhat refined, which I feel is the cause of the discrepancy. Furthermore, if I increase the levels of my refinement too far it crashes my computer - could this be indicative of my relatively low computer spec (3rd gen i3), or a problem with the dictionaries? (I initially wanted a refinement cylinder in the wake region to better characterise the flow, however this made it impossible to mesh at any reasonable refinement without crashing).


I've attached my case below if anyone could be so kind to take a look and offer any pointers in the right direction. Apologies if I've missed anything glaringly obvious - I'm still learning!

https://drive.google.com/folderview?...Vk&usp=sharing


Best Regards,

Dave

[matejfor]

Mesh Quality forever. If the mesh is no good, you cannot expect good results.

2D sphere is actually cylinder, right?

Assuming your Re number is low enough to have only 2D vortex shedding, the mesh quality is crutial. You approach of snappy mesh extrusion might work, if the mesh on a cut is really nice.
Advanced tricks you can use to build nice prism. layers are described for OpenFOAM-v3+ : http://www.openfoam.com/version-v3.0...isedMeshMotion


Maybe nicer mesh can be obtained using block-structured mesh built with blockMesh. You can create edges of your block as arcs so it is possible.

And not to stop there, 3rd option is to use foamyQuadMesh to create 2D surface mesh and then extrude. foamyQuadMesh will create body fitted mesh. See tutorials in: $FOAM_TUTORIALS/mesh/foamyQuadMesh

happy foaming
M.

[DaveR]

Matejfor,

Thank you kindly for your response.

I managed to fix the problem. For some reason the mesh wouldn't snap properly to the sphere. Not sure how you'd get around this for modelling in 3D, but in the case of 2D modelling a cylinder produced much better snapping (and viscous layers were present).

In terms of creating a cross section, I ultimately made a 2D blockmesh opposed to using the extrudeMesh facility. This prevented the explosion in cell count that was previously crashing my PC.

Regards,

Dave