[FudgeBalls]

Hello, I've a uni task which is to predict how much power will be generated by a wind turbine. I've simulation problems:

Firstly, I'm struggling to simulate the rotation of the turbine. I've tried steady simulations with a rotating reference frame but they produce 3-5 too much power (and greatly exceed Betz). I've tried using an overset mesh with DFBI, one axis free rotation, in this case I began spinning the turbine at design but it just slowed down and stopped, as if no lift was being created at all. Finally I've attempted to use DFBI superposed rotation which is currently iterating so I'm yet to see those results.

Has anyone successfully modeled a turbine? I've read a few reports and they state that both methods that have failed for me do work, but of course they don't go into methodical detail :/

I wonder if the rotating reference frame doesn't work because I'm adding energy into the system?

I'm using the value of torque (shear+pressure) present on the rotor and multiplying it by the angular speed to calculate power. This I believe works. Works better than when I attempted to compare total pressures anyway... point is, is this the usual method?

Thanks in advance!

[fluid23]

Both approaches should work. It sounds like you have a more fundamental problem. How confident are you in your mesh? 9 times out 10 that is the root of these sorts of issues. If you can, provide some shots of the mesh for the community to view and comment on.

Also make sure your units are correct. That could be why the MRF results are off.

[FudgeBalls]

Mesh is a cylinder. Has about 1.2 mil cells, Radius is about 2r and length is about 5r.

Static region:Velocity inlet, symmetry planes, pressure out.
Rotating region: wall, overset mesh

I'm aware that I need a larger mesh to get the far field but I was attempting to set up the simulation correctly before aiming for accurate results. Is this the problem?

I've attached two shots of the mesh at the rotor and one attempting to show the scale of the mesh.

[fluid23]

The proximity of your boundary conditions could be impacting your results. 1.2 mil cells also seems low for a 3 bladed wind turbine. I am guessing you aren't refining your wake either? I would revisit your mesh if I were you, although, it is hard to see most of the details in the images you provided.

[FudgeBalls]

After seeing your last reply I meshed up a 5mil version, 8r length, 3r width. Some new pictures should help show my mesh. Wake has some refinement. I'm currently running two sims on this newer mesh but they seem to have fairly similar results at the moment. What I mean by this and my greater than expected results are:

(Basic calculation)
P = 0.5 * roe * V^3 * A * Cp

roe = 1.225
V = 5 (m/s)
A = pi * 10^2 = 314.2
Cp = 0.5 (generous)

P = 12'000 W

(CFD)
P = T * w

T = torque = 15'000 (-ish, depending on specific simulation)
w = angular speed = 20rpm = 2.1 rad/s (found from design point)

P = 31'500W


If you're interested in the background at all, I'm looking into small portable turbines hence the low radius and wind speed.

[fluid23]

This does look a little better. I think your far-field boundaries are still a little coarse, but you can always refine those later. I don't think those are the route of your issues.

Can you show a screenshot of your torque report? Also, what are the units on your torque value ,N-m?

I am surprised to hear you call a 20m diameter turbine 'small and portable'. Your tip radius (not diameter) is 10m, right?

[FudgeBalls]

Attached is the what the current simulation has got so far in terms of residuals and the moment on the rotor. 12KNm is more reasonable; but at 20rpm I should be expecting about 5KNm.

Yeah, 9.75m radius. Idea is that it can sit inside a 40ft shipping container (12.2m in length) when folding blades are implemented. So it's portable, but just not in your backpack kind of portable.

[fluid23]

Ok. Gottcha. Is this your MRF model or something else? Without being able to dig into your model, my answers are limited. If you haven't done so already, you might work the tutorials just to make sure you aren't missing something.

I will note that 200 iterations seems awfully low and your residuals have not converged. Try letting it run out until your residuals flatten and see what happens.

[fluid23]

Also, you should check your wall y+ values to makes sure your near wall mesh is appropriate. That can really screw things up.

[FudgeBalls]

Yes, that was using an MRF, steady simulation.

Since my last post I've checked the y+ values and having a value of ~100 should be alright for K-e for a high Y+ approach. I've attached a few pictures of a simulation I've ran recently, it has a larger mesh, larger volumetric controls etc, pushing 5.5 mil cells.

Since I'm still getting 15.5KNm I'm meshing up an even larger mesh as we speak but there hasn't been strong evidence for the torque value to drop with this increase in mesh size (yet) so I'm not convinced this next mesh will do much better.

I have followed tutorials so I'm fairly positive that I've not missed a fundamental step. The one concern I have is, is it possible I've set up the report incorrectly? See attached in my next reply.

[FudgeBalls]

using two just to be test things out. rot is rotating region (wall only), cylinder 3 2 faces is the turbine as a part. X direction is the direction of the freestream, reference pressure is 1Bar but that only affects the value by 1Nm anyway. [0, 0, 0] is found at the centre of the turbine.

[bridgeovertroubledwater]

hi i am trying to do the same thing with a wind turbine now, did you create two subtracts one for the blades and rotating region and then one with the domain and rotating region?