[Bluejay]

Hey folks, first time post.

First thing I should note is I'm looking more for a little advice/pointers in the right direction rather than being spoon-fed answers here.

For my final year project at Uni, I'm looking at how one or more rotating cylinders can be used to 'energise' the boundary flow over an aerofoil to prevent separation at higher AOA. I'm hoping first to validate a model of an ordinary 2D aerofoil with no cylinder, then validate the same aerofoil with one or more cylinders embedded in the aerofoil surface. Luckily I have data to validate against from a guy who wind-tunnel tested some basic models with these cylinders back in the 1980's. The data is based on a Reynold's number of 46200. I'm looking to try and get similar results with CFD, before taking it a little further by playing with various parameters etc.

Prior to this project I had no CFD experience at all, so I've spent the last month or two learning the basics. I have a few questions though and was hoping someone could just point me in the right direction:

1. Am I best using BlockMesh or Salome? I've been playing with Salome, and it seems pretty simple to use. I've only touched on Blockmesh through the userguide tutorials though. My understanding is that Blockmesh is best for a simple 2D aerofoil, however what about for one with rotating cylinders?

2. Simple question I know, but I'd like to test different turbulence solvers for each aerofoil. Will I need to create a new mesh each time I use a new solver or can multiple solvers be used for the same mesh? I've yet to definitively choose my solvers yet, but it's likely I'll use SimpleFoam and I'm looking at K-Kl-Omega. Any thoughts on this and whether there's others I should look at?

3. Finally, any other things I should add to my 'to look at' list? I'm debating between Blockmesh/salome, I'm looking at grid-dependency, and I've completed the majority of the OpenFoam tutorials, so I'm looking now at getting my first case for validating a simple 2D aerofoil going.

Apologies if any of this is a little naive, like I say I'm just finding my feet . Any replies will be greatly appreciated!

[GregorS]

Yes, blockMesh is OK for simple geometries, but when it comes to some more
complex cases, it is not really handy. I don't have any experience with Salome,
but would recommend you to look at Gmsh. It can be used to create hybrid meshes, so you can create structured mesh around airfoil and cylinders to resolve boundary layer and use unstructured mesh elsewhere. This reflects in fewer mesh elements and also saves you some time with mesh generation.

Same mesh can be used for all solvers.

When it comes to turbulence models, you should also look at Spalart-Allmaras and k-omega SST. SA is easy to use and gives satisfactory results, while KW-SST proved successful at predicting boundary layer separation. Another thing when it comes to simulating boundary layer separation is approach to boundary layer treatment. Here you can use high-Re ( 30 < y+ < 300) with use of wall functions as boundary conditions on surface or low-Re (y+ < 1) approach, where you completely resolve boundary layer. Both, SA and KW-SST, models can be used with high-RE and low-Re approach. Low-Re is found to be better at predicting boundary layer separations, but you can have some problems with simulation stability. So it is a good way to first run high-Re simulation and use converged solution as initial condition for low-Re treatment.