"Parallel High-Order Finite-Volume Methods with Adaptive Mesh Refinement for Physically-Complex Flows"
Prof. Clinton P. T. Groth
Institute for Aerospace Studies, University of Toronto Toronto, Ontario, Canada
The development and application of highly-scalable and efficient parallel high-order finite- volume methods with local solution-dependent adaptive mesh refinement (AMR) are described. The proposed solution methodology has been designed for the solution of multi-scale physically-complex flows having both disparate and anisotropic spatial and temporal scales on high-performance, multi-processor, distributed-memory, computer architectures and combines a family of robust and accurate high-order central essentially non-oscillatory (CENO) spatial discretization schemes with a block-based anisotropic AMR and an efficient parallel implementation of fully-implicit time-marching schemes. The CENO scheme is a hybrid approach that avoids some of the complexities associated with essentially non-oscillatory (ENO) and weighted ENO schemes and is therefore suited for application to two- and three-dimensional mesh with irregular and unstructured topologies. The anisotropic AMR method uses an unstructured binary tree and hierarchical data structure to permit local refinement of the grid in preferred directions as directed by an error-based refinement strategy. A dual-time-stepping like approach is combined with various second- and high-order implicit temporal discretization schemes and a Newtwon-Krylov-Schwarz (NKS) iterative solution method is used to solve the system of nonlinear algebraic equations arising from the combined temporal and spatial discretization procedures in a fully-coupled manner. Applications will be discussed for a range of two- and three-dimensional problems including high-speed compressible flows of gases and plasmas as as well as both laminar and turbulent reactive flows. The computational performance of the parallel high-order AMR schemes are assessed and their potential for the efficient and accurate simulation of physically-complex flows will be demonstrated.
Prof. Groth is a computational fluid dynamicist with expertise in parallel adaptive mesh refinement (AMR) finite-volume schemes for compressible and turbulent reactive flows. He has contributed significantly to the use of AMR in both computational fluid dynamics (CFD) and computational combustion modelling. He is a leading researcher in high- performance computing and the development of reliable and robust large-eddy simulation (LES) techniques for numerical combustion modelling of premixed, non-premixed, and partially-premixed flames. He has also applied high-fidelity numerical techniques to the study of laminar flames at elevated pressures. He is the author or co-author of nearly 60 journal articles, more than 140 conference papers, and 50 invited scholarly presentations. He is also a past President and member of the Board of Directors of the Computational Fluid Dynamics Society of Canada and a current member of the Scientific Committee of the International Conference on Computational Fluid Dynamics. Professor Groth’s awards include: atural Sciences and Engineering Research Council of Canada Discovery Accelerator Supplement Award (2014); Royal Academy of Engineering, Distinguished Visiting Fellowship Award, University of Cambridge (2013); and Premier’s Research Excellence Award, Ontario (2001).