Air-Flow Simulation of a Savonius Rotor

Submitted by Dorfner_C on February 6, 2014 - 09:48

Problems

Hello everybody,

I have designed a Savonius Vertical Axis Wind Turbine on a CAD programm. Now I want to test it on CAEDIUM to simulate the airflow, the pressure, the forces and the moments. I have defined an flow-volume and a inlet/outlet, the surfaces are walls etc. The simulation of the airflow and the pressure works very well, but I have problems with the forces and the moments. The results are far away of which is possible. Also I want to know if there is any possibility to simulate that the Savonius Rotor rotates because of the airflow. Are there any compareable examples?

I am grateful for any little help,
Thanks

Is your simulation converged?

Submitted by symscape on February 6, 2014 - 11:21.

I'll assume that you have an objective measure of convergence through a residual monitor and a force monitor, as demonstrated in the tutorial "Flow Over a Double-Sided Membrane". If this is not the case then you need to create the monitors and re-run the simulation to get an idea of convergence.

If the forces are not realistic then most likely there are areas where the velocity is not realistic either. Apply Vector Fields->U (velocity) on the Results Tool Palette as a Color Map onto the flow volume (iso-surfaces) to identify where the problem is located - it might reveal an issue with the geometry or the mesh.

Do you have adequate resolution of the blades, especially the curved surfaces? If not try applying the Accuracy tool on the Physics Tool Palette to those faces.

The moments (torque) calculated in Caedium assume the center of rotation is at the origin.

Before considering a rotating case I suggest you determine the cause of the problem in your stationary case.

To consider an idealized rotating case you can use a multi-volume flow domain with a Moving Reference Frame (MRF) similar to the approach in the tutorial "Rotating Paddle using Moving Reference Frame".

Hello, Thank you for your

Submitted by Dorfner_C on February 7, 2014 - 02:06.

Hello,

Thank you for your quick answer. Now I try to increase the accuracy of the curved surfaces. Is the origin of an imported step-file the same as the Caedium origin for the moment calculation? How can I change the axis system for the moment calculation?

Thanks

Transform your geometry to the origin

Submitted by symscape on February 7, 2014 - 07:23.

Yes, Caedium preserves the location of geometry in an imported file.

The easiest way to change the axis system is to translate your geometry to the origin. Otherwise you will have to create a new results using New->Result, moment = F x (XYZ - newOrigin), where:

x = cross product
F = Vector Field->F (force)
XYZ = VectorField->XYZ (location)
newOrigin = new vector constant with length units, set to your alternate origin

For more on new result creation see "Fluid Flow Around a Sphere: Theory Comparison".

It's best practice to always create geometry at the origin in your CAD system and in Caedium to maintain accuracy.

dynamic simulation

Submitted by Dorfner_C on February 17, 2014 - 03:03.

Hello everyone,

I simulated my savonius rotor in a stationary case. Now I want to calculate the moments on my three blades in a dynamic case, but I don´t know how to start. The example of the rotating paddle with the moving reference frame isn´t very helpful. My savonius rotor has three blades moved by 120 degress in a circle. I want to rotate the blades by a constant angluar velocity and a constant wind speed and then calculate the moment on the blades.

Thank you for any reply

Multi-volume flow domain

Submitted by symscape on February 17, 2014 - 08:14.

To perform a Moving Reference Frame (MRF) simulation you have to create a 2 volume flow domain. One volume, in your case a cylinder, must enclose the rotating components and will be designated the MRF. The other volume, a much larger cylinder, needs to enclose the first and must geometrically and topologically match the boundaries on the first embedded volume. For the matching internal boundaries use the Faces->Connect tool on the Geometry Tool Palette to connect the 2 volumes. This then requires a multi-volume simulation as described in the rotating paddle tutorial, in terms of Substance assignment to a volume group and boundary condition assignment. Note you apply the Conditions->Angular Velocity tool on the Physics Tool Palette to the MRF volume.

The MRF simulates an instant in time for the position of the blades. Imagine spinning the turbine and being able to measure the forces at a certain blade orientation. If you want to get the moment variation for a full cycle then you will need to run a series of simulations where you incrementally vary the onset (free stream) air velocity through 360 degrees - hence the need for an outer cylinder as in the tutorial "Flow Over a Double-Sided Membrane".

Keep in mind that MRF is a lower order approximation to the actual air flow around rotating blades compared to a full moving mesh simulation.

Symscape