[PiotrekL1]

Hello

I have tried simulation of the simple NACA symmetrical airfoil in 2D. That looks nice but my goal is to build VAWT type H-Darrieus on my engineering dissertation. I finished Inventor training. I can make a 3D model this is no problem but I dont know how to simularte it. I am wondiering if:

- there is a possibility to set friction/load of the turbine generated by the parts( like electric generator) and then put those data , set initial speed( H darrieus cannot do self-start, needs to be spinned up ) and then get for example maximum turbine RPMs when providing details of the wind( like speedd, airpressure..)

- I read about simulation and ( I dont know if I get it correctly) I need to set RPMs of the rotor and I will get a possible lift force generated and I need to check if this lift force will produce enough torque at particular wind speed and if it is sufficient to power the generator.


To be honest I dont know simulation technique of it. I persume it is quite complicated task bur it will be good to know which technique is the easiest one.

[marktm250xc]

This will be an extremely challenging application, especially is this is a very large, full-scale industrial VAWT. The long thin airfoils make it very mesh intensive and the transient aspect makes it resource and time intensive. You

The relatively thin airfoils rule out motion to due masking challenges. The alternative is a rotate region which envelops the VAWT. The rotate region has an option for free-spinning, but the type of VAWT you mention is not free-starting. So, you can use an initial condition to set a starting rotation while the freestream air velocity kicks in.

You will also need to experiment with timestep. With only 2 blades, a blade-to-blade (time required for blade to reach next position or 180deg) is way too large. Estimate the steady-state rpm and then try starting with a 1/2 blade or 1/4 blade and work your way down from there.

[matt.brown.ae]

We did some consulting work on a Darrieus type wind turbine and tried to use Simulation CFD to analyze the 2D rotor performance in an attempt to validate our MSTT models. After much effort and debate we were told by Autodesk support that it was not possible due to the mesher. We found that the rotating region needed to implement the transient analysis had a huge influence on the behavior of the flow. It was essentially pulling the whole mass of air around the rotor around with the blades creating a giant 'whirlpool' (aka vortex). It was not representative of a VAWT flow field.

[ApoloV]

Originally Posted by matt.brown.ae

We did some consulting work on a Darrieus type wind turbine and tried to use Simulation CFD to analyze the 2D rotor performance in an attempt to validate our MSTT models. After much effort and debate we were told by Autodesk support that it was not possible due to the mesher. We found that the rotating region needed to implement the transient analysis had a huge influence on the behavior of the flow. It was essentially pulling the whole mass of air around the rotor around with the blades creating a giant 'whirlpool' (aka vortex). It was not representative of a VAWT flow field.

The main thing here that has to be considered is the timestep. With rotating regions in large air domains if the mesh size on the rotating region is dramatically smaller than the surroundings the default timestep sizing (blade-to-blade or even 3º/step) may be too large.
To maintain proper continuity and linking between the Absolute Reference Frame (air domain) and the Relative Reference Frame (rotating region) a smaller timestep must be used.

Typically to do this we would want no more than 3-4 elements worth of rotation of the Relative frame to the Absolute frame.

As Mark mentioned it can be mesh sensitive to capture the detail of the thin airfoils and have that transition to the surroundings but ultimately it will come down to the timestep size if your rotating region will appear as a "whirlpool" or properly capture the details through it.

I recently just outlined some of this for another user on the Autodesk SimCFD Forums

As advised the other user, You can show this phenomenon as well as get a handle on the timestep size required by taking a 2D slice and running that first (on the XY plane) then you can take that timestep understanding to the 3D and run it out, however I would expect that in some cases the mesh size and timestep required may take these out to be longer running analyses.