KEPLER: An Integrated Simulating System for Efficient Industrial Design and Optimization

Abstract

This paper describes the fundamental numerical schemes, objectives, structure and current capabilities of the KEPLER simulating system. This computational analysis and design software collection is aim to be designed and developed to solve wide range of industrial simulations in analysis and optimization tasks for industrial designers. KEPLER integrates pre-processing(automatic mesh generation), simulation and post-processing by using the "In-memory database" technology, which provide one-step solution for complete simulations without switching between different softwares/tools, also enables real-time reading/editing configurations and extracting results during the simulation by multi-users which could increase the working and collaboration efficiency dramatically. KEPLER provides unsteady solutions for compressible and incompressible flows(CFD) based on innovative high order Kinetic Energy Preserving flux reconstruction method with space-time expansion as time marching method(STE-KEP-FR). FR method is a localized stencil numerical scheme which constructs high order discretization completely in-cell, this method is very efficient for its simple differential formulation. The KEP method preserve the order of accuracy and enhance the nonlinear stability of FR discretization without increasing distinct computing cost. The STE method enables time-accurate local time-stepping, which means each cell use its local optimal time-step instead of the globally smallest time-step during the time-marching process, and the local time-step could be adaptive for stability and any other dynamic process. Because of the locality of STE-FR-FR method on both space and time, time-adaptive non-conformal and overset mesh are using seamless and efficiently on KEPLER, which enable the abilities for simulating complex moving/deforming geometries and wide range of dynamic flow problems. The computing efficiency, accuracy, stability of convection calculations, and overset performance is investigated in this work which are all critical issues for the ability and Competitiveness of Kepler code.