Portable Parallelization of Industrial Aerodynamic Applications (Popinda): Results of a Bmbf Project

ISBN-13: 9783322865786 / Angielski / Miękka / 2012 / 233 str.

Anton Sch Ller

Opis

This book contains the main results of the German project POPINDA. It surveys the state of the art of industrial aerodynamic design simulations on parallel systems. POPINDA is an acronym for Portable Parallelization of Industrial Aerodynamic Applications. This project started in late 1993. The research and development work invested in POPINDA corresponds to about 12 scientists working full-time for the three and a half years of the project. POPINDA was funded by the German Federal Ministry for Education, Science, Research and Technology (BMBF). The central goals of POPINDA were to unify and parallelize the block-structured aerodynamic flow codes of the German aircraft industry and to develop new algorithmic approaches to improve the efficiency and robustness of these programs. The philosophy behind these goals is that challenging and important numerical appli cations such as the prediction of the 3D viscous flow around full aircraft in aerodynamic design can only be carried out successfully if the benefits of modern fast numerical solvers and parallel high performance computers are combined. This combination is a "conditio sine qua non" if more complex applications such as aerodynamic design optimization or fluid structure interaction problems have to be solved. When being solved in a standard industrial aerodynamic design process, such more complex applications even require a substantial further reduction of computing times. Parallel and vector computers on the one side and innovative numerical algorithms such as multigrid on the other have enabled impressive improvements in scientific computing in the last 15 years."

Spis treści

1 Overview.- 1.1 Basis, Goals and Results of POPINDA.- 1.1.1 Introduction and Summary.- 1.1.2 Background.- 1.1.3 Basis.- 1.1.4 Approach and Ideas.- 1.1.5 Results.- 1.1.6 Reasons for the Success of POPINDA.- 1.1.7 Impact and Outlook.- 1.2 POPINDA — the Industrial Qualification.- 2 Parallelization and Benchmarking.- 2.1 Unified Block Structures — the Basis for Parallelization.- 2.1.1 Requirements for the Parallelization of Large CFD Codes.- 2.1.2 Parallelization Strategies.- 2.1.3 Parallelization of Block-Structured Flow Solvers within the POPINDA Project.- 2.1.4 Basic Concept of Block Structure.- 2.1.5 Standardization of Production Codes.- 2.2 The High-Level Communications Library CLIC.- 2.2.1 Introduction.- 2.2.2 Overview on Functionality of the CLIC-3D.- 2.2.3 CLIC-3D Design Issues.- 2.2.4 Analysis of the Block Structure.- 2.2.5 Distribution of Alteration Rights on Block Boundaries.- 2.2.6 Special Communication Requirements on Block-Structured Grids..- 2.2.7 Creation of Node Processes and Mapping of the Blocks.- 2.2.8 Special Communication Tasks Performed on Node Processes.- 2.2.9 Parallel Output.- 2.2.10 Global Operations over Ail Node Processes.- 2.2.11 Future Tasks to Be Realized by the CLIC-3D.- 2.3 Porting CLIC from PARMACS to MPI.- 2.3.1 The Objective.- 2.3.2 The Conversion.- 2.3.3 Schematic Representation of Conversion by Means of PM2MPI.- 2.3.4 The GMD Conversion Tool PM2MPI.- 2.3.5 Tools for Conversion.- 2.3.6 Further Developments and Improvements.- 2.3.7 Results.- 2.4 FLOWer.- 2.4.1 Governing Equations.- 2.4.2 Spatial Discretization.- 2.4.3 Time Integration.- 2.4.4 Acceleration Techniques for Steady Calculations.- 2.4.5 Exchange of Solution Data at Block Boundaries.- 2.4.6 Parallelization of the FLOWer Code.- 2.5 NSFLEX-P.- 2.5.1 Governing Equations.- 2.5.2 The Navier-Stokes Solver NSFLEX-P.- 2.6 Benchmarks and Large Scale Examples.- 2.6.1 Benchmarks.- 2.6.2 Large Scale Examples.- 3 Algorithmic Aspects.- 3.1 Singularities of Block-Structured Meshes — a Special Parallelizable Approach.- 3.2 Dual-Time Stepping Method.- 3.3 Scalability of Parallel Multigrid.- 3.3.1 Introduction.- 3.3.2 LiSS — a Package for the Parallel Solution of Partial Differential Equations.- 3.3.3 Multigrid Treatment of Block Boundaries.- 3.3.4 The Solution on the Coarsest Grid.- 3.3.5 Conclusions.- 3.4 Convergence for Increasing Numbers of Blocks.- 3.4.1 Introduction.- 3.4.2 Test Cases.- 3.4.3 FLOWer.- 3.4.4 LiSS.- 3.5 New Smoothers for Higher Order Upwind Discretizations of Convection-Dominated Problems like the Euler Equations.- 3.5.1 Introduction.- 3.5.2 The Discretization and the Solution Method.- 3.5.3 Fourier Analysis Results.- 3.5.4 Numerical Results.- 3.5.5 Conclusions.- 3.6 Krylov Subspace Acceleration for Linear and Nonlinear Multigrid Schemes.- 3.6.1 Introduction.- 3.6.2 The Krylov Acceleration for Linear Multigrid Methods.- 3.6.3 The Krylov Acceleration for Nonlinear Mnltigrid Methods.- 3.6.4 Conclusions.- 3.7 Multiple Semi-Coarsening for 3D Singularly Perturbed Scalar Partial Differential Equations.- 3.7.1 Introduction.- 3.7.2 The 3D Solution Method.- 3.7.3 3D Numerical Results.- 3.7.4 Conclusions.- 4 Adaptive Local Refinements.- 4.1 Why to Use Adaptive Grids?.- 4.1.1 Future Applications.- 4.1.2 Idea.- 4.1.3 A Simple Example.- 4.1.4 Multigrid on Adaptive Grids.- 4.1.5 Refinement Criteria.- 4.1.6 Discretization at Boundaries of Refinement Areas.- 4.1.7 Problems of the Parallelization for Block-Structured Multigrid.- 4.2 Self-Adaptive Local Refinements Supported by the CLIC-3D Library.- 4.2.1 Introduction.- 4.2.2 Overview of CLIC Functions Supporting Self-Adaptive Local Refinements.- 4.2.3 Adaptive Multigrid (MLAT) on Block-Structured Grids.- 4.2.4 The Refinement Criteria of CLIC-3D.- 4.2.5 Creation of a New “Refined” Block Structure.- 4.2.6 Transfer of Grid Function Values.- 4.2.7 Future Tasks to Be Realized for Local Refinements by the CLIC-3D.- 4.3 Load-Balancing Strategies.- 4.3.1 Introduction.- 4.3.2 Communication Model.- 4.3.3 Example.- 4.4 Experiences LiSS.- 4.4.1 Introduction.- 4.4.2 Applications and Results.- 4.5 Experiences FLOWer.- 4.5.1 Local Refinement Procedure.- 4.5.2 First Results of the Local Refinement Procedure.- 5 Special Aspects and Related Activities.- 5.1 Software Engineering and Software Quality Issues.- 5.2 Real Applications on Parallel Systems — the RAPS Initiative.- 5.2.1 Summary.- 5.2.2 Background.- 5.2.3 Benchmarking Parallel Computers.- 5.2.4 The RAPS Approach.- 5.2.5 Exploitation of Results.- 5.2.6 Industrial Impact and Knowledge Flow.- 5.3 MEGAFLOW.