[Fernando R.]

Hi. I am trying to simulate the performance of a vertical axis wind turbine using the CFX. The geometry of the rotor, its rotational speed and the wind conditions are based on the parameters utilized in experimental tests on field.
Since the simulation was based on those experimental tests, I was supposed to obtain similar results. The problem is that, while the real turbine provided more than 10kW of power, the one that I've simulated only consumed power to rotate (generated negative torque). I would like to know what I'm doing wrong.

Information about the domain geometry, mesh and boundary conditions are summarized bellow. Images are attached too.

The rotor has straight blades. To perform the simulation, I've used only half of the rotor (cutting with a plane normal to its axis), being imposed the symmetry boundary condition on the plane of the cut. The simulation was transient.

DOMAIN GEOMETRY
As seen on the Figures 1 to 3, I've created two domains: one enclosing the region of the rotor and a bigger one for the surroundings. The inlet would be on the semicylindrical surface of the surroundings domain. I've chosen it to be semicylindrical to reduce a bit the number of elements.

MESH
I've used a tetragonal, path independent mesh in both domains. In the rotor's I've used a "body of influence" sizing in the region of the blades. Besides, I've applied inflation with the blades surface as boundaries. Figures 4 and 5 show a bit of the mesh.

BOUNDARY CONDITIONS
Surroundings domain
As said above, the "Inlet" was the semicylindrical surface of the surroundings domain. The speed was given by cartesian coordinates. The opposed surface was set as "Outlet", with 0 Pa as average static pressure. On the sides and on the top the surfaces were set as "Free Slip Wall". The bottom surface was set as "Symmetry".
Rotor Domain
The bottom of the rotor domain was set as "Symmetry", and the surface of the rotor (blades, axis, supporting arms) were set as "Smooth Wall". The domain was set as "Rotating" with constant speed.
Interface
The interface between the rotor and surroundings domains (a circle and a cylinder wall) was set as "Fluid Fluid" type. The model was "General Connection" with "Transient Rotor Stator" Frame Change/Mixing Model. The Pitch Change was set as "None".

[singer1812]

These are general statements based on your pics alone:

Domain seems small front to back. Mesh seems like it needs work.

Are you using ICEM to get the mesh?
Size transitions seem abrupt, which is what you get using Octree.
Inflation layer doesnt seem adequate.

Try generating volume mesh using Octree using an appropriate expansion, blow away the volume mesh, and generate a delauney with an appropriate expansion. You will get a much smoother transition. Generate the prisms on the delauney.

This geometery also looks pretty simple. You should be able to get a decent hex mesh on it. You can try that also...

[Fernando R.]

I've attached an image of the half rotor geometry, if it's of interest. The blades has the NACA 0018 profile, with the tips made with a rotation of the profile. The arms are elliptical and the axis is a simple cylinder.
For the mesh I've used the default program that comes when I select the CFX on the Workbench. I've used the path independent method because I thought doing so would reduce the number of elements and reduce the time required to create the mesh.
About the domain, how much would the height, length and width be, relatively to the turbine diameter and half it's height (since I'm simulating half rotor)?
About the suggestions you gave me about creating the mesh, could you tell briefly how can I do it in the default meshing program associated with the CFX?