In this study, computational fluid dynamics (CFD) software is used to model and simulate a smooth disk rotating in a still fluid. The primary quantity of interest is the convective heat transfer coefficient. Laminar, transition, and turbulent flow regimes are considered. To ensure that the physical characteristics of the situation are properly captured, the velocity and temperature profiles of the fluid are qualitatively and quantitatively compared to results reported in the literature. By analyzing the velocity and temperature profiles, an estimate for the proper size of the computational domain within the CFD software is obtained. If the computational domain is too small, the velocity and temperature profiles are affected; hence, the calculation of the convective heat transfer coefficient is influenced by the size of the computational domain.
Two methods — the global rotation method (GRM) and the local rotation (or sliding) method (LRM) — are used to obtain results for the convective heat transfer coefficient. The global rotation method involves conversion of the specified angular velocity into a mass-distributed force. The local rotation method involves the creation of a secondary body that rotates. The secondary body can cause inaccuracies to appear when discretizing the domain. These inaccuracies cause the local rotation method to be less predictable than the global rotation method.
An extensive and systematic verification and validation study of the two methods is performed with validation achieved by comparing the calculated values to published correlations and measured results. Both methods demonstrate agreement with the expected heat transfer coefficient value by maintaining a difference of less than 3% for low angular velocities. At higher angular speeds, the global rotation method maintains less than a 3% difference, once an appropriate mesh is established and the turbulent parameters are set correctly.
Over the entire range of parameters considered, the global rotation method is best suited to model and simulate a singular disk rotating in still air. The global rotation method does not require the introduction of a secondary body and yields predictable and uniform convergence to published values as the mesh size is reduced.
Relationships between accuracy, grid size, and computational effort are reported. Recommended laminar and turbulent parameter settings are discussed. The results of this fundamental study provide insight and guidance into the proper modeling and simulation of both low-speed and high-speed rotating machinery and equipment.