In this paper, the high-order compact gas-kinetic scheme (CGKS) on three-dimensional hybrid unstructured mesh is further developed with the p-multigrid technique for steady-state solution acceleration. The p-multigrid strategy is a two-level algorithm. On the high-order level, the high-order CGKS is used to evolve both cell-averaged conservative flow variables and their gradients under high-order compact initial reconstruction at the beginning of next time step. On the low-order level, starting from the high-order level solution the cell-averaged conservative flow variables is evolved by a first-order scheme, where implicit backward Euler smoother is adopted for accelerating the convergence of steady-state solution. The final iterative updating scheme becomes numerically simple and computationally efficient. The effectiveness of the p-multigrid method is validated in both subsonic and supersonic flow simulations in two- and three-dimensional space with hybrid unstructured mesh. One order of magnitude speedup in the convergence rate has be achieved by the approach in comparison with the explicit counterpart.
A Nonlinear Multigrid Steady-State Solver for Microflow
LMAM & School of Mathematical Sciences, Peking University(北京大学数学科学学院);HEDPS & CAPT, LMAM & School of Mathematical Sciences, Peking University(北京大学数学科学学院)
Acceleration of an algebraic multigrid pressure solver using graph neural networks
Faculty of Mechanical Engineering, Delft University of Technology, The Netherlands(荷兰代尔夫特理工大学机械工程学院);Maritime Research Institute Netherlands, The Netherlands(荷兰海事研究院);Department of Aeronautics, Imperial College London, United Kingdom(英国伦敦帝国理工学院航空系)