Convergence behaviour of partitioned methods for conjugate heat transfer problems
Please login to view abstract download link
Conjugate heat transfer (CHT) considers the simultaneous solution of heat transfer in adjacent solid and fluid domains. If solved with a partitioned approach, temperature and heat flux are exchanged to achieve a thermal equilibrium at the interface. Gauss-Seidel iterations, however, are only stable for a limited range of Biot numbers because temperature and heat flux are exchanged unmodified. In order to extend the stability range of Gauss-Seidel iterations, static relaxation, Aitken relaxation or quasi-Newton methods can be used. These methods are tested on a heated flat plate problem, both in steady state and transient simulations, and their convergence behaviour is analysed. Using temperature as input for the fluid solver and heat flux as input for the solid solver converges best for steady state simulations, while the opposite is most stable for transient cases. Furthermore, the quasi-Newton method converges slightly better than the relaxation methods in transient simulations with rapid changes at the interface.
