Ansys CFX deformable mesh 1

[xlxc34]

Hi,

I am working on a simulation tool for wind turbine blade fatigue testing.

The gif on this page is worth a thousand words...

http://www.talentfactory.dk/en/tour/manu/bladtest.htm

I have written a beam element FEA code in MATLAB that allows a transient analysis to be performed. It works great, but in order to get the model to match up to physical test results I have to use a very high drag coefficient for the blade when simulating the damping that occurs due to air resistance. I have a colleague who is a CFD expert and he has agreed to do a bit of CFD analysis, however he is new to Ansys CFX. As he is working on this problem more or less in his spare time, I was wondering if anyone has any pointers for the analysis I've outlined below?

Aerofoil with 1m chord length moving up and down with simple harmonic motion. The frequency is 1Hz and the amplitude is 1m. The analysis is 2D.

Any help would be much appreciated!

 

[xlxc34]

I should probably also mention that the air around the blade would be completely static if the blade were not moving.

 

[btrueblood]

First, cool pic, and a star just for that (beats reading the 1000-word description

How high is the Cd you need to assume? A wild guess would put it at about 1.5 to 2.0...but there may be some effect of the entrained air column.

Does your FEA model with the Cd damping effect give the right phase between input (displacement or force, whichever you are using) and the output deflection. Then I'd say you're close enough...

How do you handle correlating the quasi-static fatigue test to real world (rotating blade and the increased apparent blade stiffness?

 

[btrueblood]

Hmm, a few more minute's thought.

1. 2d model should keep a fine grid for several chord lengths on both ends of the foil. Outer grid should model the building space (ceiling, walls and floor) of the test area.

2. There may be some significant pumping effect going on, a-la a japanese hand fan. I.e. some energy may be getting dumped into flow being pumped along the blade axis. This wouldn't be captured by a 2d model.

 

[xlxc34]

Thanks a lot for those tips... In answer to your questions...

I need to use Cd 4.5... this I suspect is due to the fact that as the blade travels down it pulls air with it, so when it goes up it is meeting air that is still travellling downwards (if you catch my drift). As F is proportional to v^2 this could account for a drag coefficient that high (or perhaps I'm clutching at straws).

Yes, the drag force is 180 degrees out of phase with the input force from the excitation equipment. So thats one less thing to worry about.

Finally, the blade resonant frequency is only about 5 times the rotational frequency in service so it would take about 7 years to test a typical blade at the loads it sees in its life. What they do is look at the fatigue damage that occurs in service using Miner's law and then increase the loads so that the same amount of damage occurs after about a million cycles (instead of 150 million).

Thanks again for the advice, do you know if there are any settings that I should be looking at in CFX related to meshing, time step etc? I'll just be passing them straight onto my colleague as I have no clue about CFD.

 

[xlxc34]

Don't seem to be able to edit posts... the displacement lags the input force by 90 degrees.

 

[btrueblood]

No clue on the time step, other than it will be an issue. As always, I would keep refining time steps until some kind of convergence appears, or the runtime goes to unacceptably high limits.

Agree with you, first order effect is entrained airflow meeting the blade on its return swing, average Cd should increase over a steady state value. 4.5 doesn't seem all that unreasonable given that effect, plus the spanwise pumping effect alluded to earlier.

Displacement...should be 180 degrees from force? Drag force will be proportional to velocity, i.e. should be 90 degrees lagging relative to force...or am I screwing up my math...