Conduction in rotating mesh

Niels Linnemann

Hello I an simulating heat transfer in a rotary regenerator used for preheating combusting air. The preheater can be considered a big circular disk with many plates in the radial direction which draws heat by convection in the hot side and releases heat in the cold side. http://www.faculty.virginia.edu/riba...s/image023.gif

I have modeled the preheater as a porous zone with a UDF file for specifying heat transfer from and to the "non" existing solid in the porous zone.

The problem is that in this way I have to specify a fixed wall temperature which isn't the case since it is rotating so there will be a temperature gradient in radial and rotational direction.

In order to update the temperature profile I use the source-terms from the porous case in another case with a pure solid conduction problem with modification to the conduction term in order to simulate the porosity.

Both cases are rotating with 0.088rad/s, but when i simulate the solid case I get +5000K on one of the sides and -1K on the other side of the centerline with temperature gradient parallel to the centerline (they should be angeled). In then thought well maybe the energy balance is not fulfilled since I use guessed values for the surface temperature in the first run. Then I tried making a simple model with source terms that are positive above the centerline and negative below the centerline, this way the source term and volume exactly match so energy balance is fine. BUT the temperature profiles are still parallel to the centerline which shouldn't be the case when the solid is rotating. Increasing or decreasing the rotational velocity does not change anything although you see a change in the porous case.

I have not found anything in the manual or the web that addresses this kind of behavior.

The solid case is a pure conduction (circular disk) problem no flow and adiabatic walls, with an UDF file mapping the source terms to the geometry.

The Porous case is a pure flow problem with an UDF file to simulate convection to a non existing solid.

I have tried updating the surface temperature directly in the porous case but in this way the conduction is not simulated.

If anyone have a solution or another approach that could be used please let me know.