[Flo27]

Hi guys,

I'm simulating a Darreius rotor combined with a Savonius rotor.
I made a unstructured mesh with a structured area at both sides of the rotating interface, so that the mesh at the interface matches perfectly each time step.

I ran the simulation for about 1000 time steps and noticed a serious problem at the transition from the structured area to the unstructured area in the rotating body. (See Picture)

My boundary conditions are:

Transient Simulation:
-Time steps: 2.5e-05s
Rotating:
-4000 rpm

Stationary:
-Velocity: x = 15 m/s

Advection Scheme:
-Central Difference (Bounded CDS)

Transient Scheme:
-Second Order Backward Euler
-Automatic Timestep Initialization

Turbulence Numerics:
-High Resolution

I really don't know whats wrong and I would be very happy if someone could help me.

Kind regards,
Flo27

[ghorrocks]

the mesh at the interface matches perfectly each time step.

Why did you do that? How does that help? GGI interfaces are pretty good and should not produce any significant diffusion on a reasonable mesh. So matching mesh across the interface should be unnecessary.

What is the serious error? The ring of high velocity at the interface?

If so, try this:
* do you see it when you plot velocity (not velocity in stn frame)
* Check you are not plotting hybrid values
* do you see it when you use other differencing schemes such as Hi-res or hybrid?
* did you have an initial condition where the rotating bit was initialised with a rotating velocity (ie zero velocity)?
* Are you sure you have correctly put an interface between the rotating the stationary bits? what type of interface did you use anyway? I am suspicious you have not done this properly as matching nodes over interfaces is not usual procedure.