In rotating cavities the driving temperature difference for heat transfer is not easy to define or estimate. Traditionally, some reference temperature, here called bulk temperature, is used. This bulk temperature is closely connected to the heat transfer coefficient. In order to determine these characteristics, the assumption that the wall heat flux is linearly proportional to the temperature difference between wall and inlet air, is used. The slope is equal to the heat transfer coefficient and the x-intercept gives the difference between bulk temperature and inlet temperature. The validity of this assumption is thoroughly investigated by solving the Reynolds averaged Navier-Stokes equations for compressible, axisymmetric flow with a low Reynolds number k-ϵ-model. Rotational and buoyancy effects, which may introduce a non-linear relationship and also affect the local bulk temperature, are all taken into account in the CFD model.
Three different cases were investigated: one simple corotating disk cavity; one simple rotor-stator cavity, and finally one real engine application cavity. The rotational Reynolds numbers, mass flow rates and temperature differences were varied.
Results indicate that the Linear assumption is valid for a range of wall temperatures but not for regions where the local wall temperature affects the flow field, e.g. in corners. Furthermore, when the flow field undergoes a drastic change, new heat transfer characteristics must be determined, or be used with care. Since the heat transfer coefficient and bulk temperature are uniquely determined by the flow field, and not by the local wall temperature, it is not necessary to make a coupled, continuous calculation of the flow field and thermal distribution in the structure.