[VisMart]

Hi guys,

I need your help with my simulation of a 3d wind turbine with Fluent for my final thesis. I have 2 fluids domains, a rotating fluid cilinder containing the rotor, and the rest of the static fluid in the "wind tunnel". In the interface, the solution gives no continuity as you can see at the pics, mostly evident in vorticity (not such in velocity). It is made with MRF (sliding mesh). I don't know how to solve that issue, and I guess it should be something related with the "transmission of information" between both regions in the interface. When I create the interface(with the two cylinder surface, one for the rotor and the other one for the tunnel), appears automatically two "wall regions", one for each cylinder surface, and I don't know if I should change its boundary conditions, or leave them with its default.

Thanks a lot guys

[VisMart]

The case is:

3D, double precision
Steady
Turb = Kw-SST
Rot. vel. = 425 rpm
Vel. inlet = 10 m/s
Press. out = 0Pa
Initialization = Standar from inlet
Sol. methods = SIMPLE, 2nd order,2nd order,2nd order,2nd order.

[CFD-fellow]

Hi
Vorticity is the curl of velocity vector, so the problem is in your velocity distribution. Because of its range its not as sensible as vorticity contours.
Dont change those appeared walls, they are ok.You can increase yor mesh elements or if it is enough, you can broaden your moving zone limits to prevent discontinuity in conntours.But I think it doesnt affect on your general parameters such as rotor moment or drag.In my experience MRF is not enough for your simulation, change to moving mesh(sliding mesh) from the rest of solution.
Regards

[VisMart]

Thak you CDF-fellow,

So, I have to change the case from "steady" to "transient"? Or a moving mesh makes sense with steady case? I am starting with Fluent and I don't know the software enough, so I will try to apply your advices. Maybe the transient solver will diffuse the vorticity through the fluid and the problem will disappear, isn't it? or, It will persist and I have to ignore it? . The mesh is very fine, and the rotor is wide enough, so I don´t think this is the problem.

This case is symmetric, rpm and v.inlet are constants, so I thought the steady solver would be enough. In fact, I just to review my first simulation values of Torque and Axial Force and they seem correct, so you are right, and maybe this is not as important as I though .

Thank you again!!

[CFD-fellow]

Hi vismart

[CFD-fellow]

Hi
Yes change your solver to transient. Moving mesh is more robust and gives closer values to experimental data. Choose a time step so that your rotor rotate one degree.
Regards

[VisMart]

Ok, I will try it and write back. Thanks

[VisMart]

Originally Posted by CFD-fellow

Hi
Yes change your solver to transient. Moving mesh is more robust and gives closer values to experimental data. Choose a time step so that your rotor rotate one degree.
Regards

Hi again,

I am still running the transient simulation, and I have any answer yet for the last quote.

But I have another question about the steady case. As you can see in the pics, there is no continuity in velocity and the streamlines are very weird. The global coefficients (Torque, Axial force) are quite well according to the experimental results, but I am worried about this problem. Is it usual in steady cases? or am I doing something wrong?

Thanks in advance!

[CFD-fellow]

Yes its usual and this subject is described in Fluent documentation. You cant get real streamlines with MRF. How much is the difference between experiment and numerical simulation torque and axial force?

[VisMart]

Originally Posted by CFD-fellow

Yes its usual and this subject is described in Fluent documentation. You cant get real streamlines with MRF. How much is the difference between experiment and numerical simulation torque and axial force?

Tank you CFD-fellow!! the difference is about 3-10%, It depends of the velocity inlet, because I am trying to study cases with design conditions, turbulent wake conditions and stalled conditions.

[gfoam]

May be the streamlines issue is because you are computing them in each reference frame. You have to compute them in a inertial erference frame, so when you plot the streamlines use velocity in a stationary reference frame in CFDPost. I hope this helps you. regards. Gonzalo