Flow around jet engine
 

Hi !
I'm Sylvain, French student, and I'm now in a training period in an jet engine engineering company.

My goal is to have lift and drag forces on the pod which contain the engine.
I tried a few different ways :
-empty pod (nothing inside, you have juste a revolved airfoil)
-inlet and outlet on the engine (by pressure, massflow or normal speed)
I'm not able to find the drag I'm supposed to have, about 70N, I find 200N in some case, 400 in other, and a lot of different values, depending on the turbulence model I use etc.

I'm trying a High Resolution SST, with a large mesh (the pod is abbout 1m long and 1m diameter), and I find something like 300N drage. I supposed it's caused by ma poor mesh, so I refined it, I found 100N. I decided to put some layers to have a low Y+ (40layers on 5mm), but I have some convergence problems, the model oscilating around 10e-3. If I stop the run, I obtain 200N.
The domain is large (10meters cube), my B.Cs are ok.

I tried hundreds of differents way, but nothing seems to be good, If I find an approched result, I refine the mesh, and it's very different, sometimes more, sometimes less... I'm totally lost, I have searched in the FAQ, I have searched airfoil tutorials, but can't find.
I would like to give you more details, but I don't know what is very important, it's very complicated ! I don't have skills in CFD, I have to move in groping... How can you said "ok, that's the good result" if you have no idea about it ? :)

If you can help me please...

Thanks a lot.

Sylvain

If you have large separations the SST model may not be too suitable, you may need a LES style model.

This sounds ilke the sort of difficulty you can find with poor mesh quality. You will need to improve mesh quality.

Thank you for your answer !
Ok I will try with a LES model.
It will be difficult to improve the number of elements to improve mesh quality, the computer is powerful but after 10millions elemets, the run don't want to start... Do you think I shoold reduce the domain to increase the mesh size ?
Bye !
Sylvain

You need to run sensitivity analyses on everything important, including boundary proximity. The fact that your results are all over the place says you are not close to an accurate solution yet, so you probably have several issues to solve.