# Parallel Computational Fluid Dynamics '97

## Recent Developments and Advances Using Parallel Computers

**Edited by**

- D. Emerson, Daresbury Laboratory, Daresbury, Warrington, WA4 4AD, Cheshire, UK
- A. Ecer, Deparment of Mechanical Engineering, Indiana University Purdue, University Indianapolis, IN, USA
- P. Fox, Purdue School of Engineering and Technology, 799 West Michigan Street, Indianapolis, IN 46202-5160, USA
- J. Periaux, Dassault-Aviation, 78 Quai Marcel Dassault, 92214 Saint Cloud, France
- N. Satofuka, Kyoto Institute of Technology, Matsugasaki, Sakyo-ku, Kyoto 606, Japan

Computational Fluid Dynamics (CFD) is a discipline that has always been in the vanguard of the exploitation of emerging and developing technologies. Advances in both algorithms and computers have rapidly been absorbed by the CFD community in its quest for more accurate simulations and reductions in the time to solution. Within this context, parallel computing has played an increasingly important role. Moreover, the uptake of parallel computing has brought the CFD community into ever-closer contact with hardware vendors and computer scientists. The multidisciplinary subject of parallel CFD and its rapidly evolving nature, in terms of hardware and software, requires a regular international meeting of this nature to keep abreast of the most recent developments.

Parallel CFD '97 is part of an annual conference series dedicated to the discussion of recent developments and applications of parallel computing in the field of CFD and related disciplines. This was the 9th in the series, and since the inaugural conference in 1989, many new developments and technologies have emerged. The intervening years have also proved to be extremely volatile for many hardware vendors and a number of companies appeared and then disappeared. However, the belief that parallel computing is the only way forward has remained undiminished. Moreover, the increasing reliability and acceptance of parallel computers has seen many commercial companies now offering parallel versions of their codes, many developed within the EC funded EUROPORT activity, but generally for more modest numbers of processors. It is clear that industry has not moved to large scale parallel systems but it has shown a keen interest in more modest parallel systems recognising that parallel computing will play an important role in the future. This book forms the proceedings of the CFD '97 conference, which was organised by the the Computational Engineering Group at Daresbury Laboratory and held in Manchester, England, on May 19-21 1997. The sessions involved papers on many diverse subjects including turbulence, reactive flows, adaptive schemes, unsteady flows, unstructured mesh applications, industrial applications, developments in software tools and environments, climate modelling, parallel algorithms, evaluation of computer architectures and a special session devoted to parallel CFD at the AEREA research centres. This year's conference, like its predecessors, saw a continued improvement in both the quantity and quality of contributed papers.

Since the conference series began many significant milestones have been acheived. For example in 1994, Massively Parallel Processing (MPP) became a reality with the advent of Cray T3D. This, of course, has brought with it the new challenge of scalability for both algorithms and architectures. In the 12 months since the 1996 conference, two more major milestones were achieved: microprocessors with a peak performance of a Gflop/s became available and the world's first Tflop/s calculation was performed. In the 1991 proceedings, the editors indicated that a Tflop/s computer was likely to be available in the latter half of this decade. On December 4th 1996, Intel achieved this breakthrough on the Linpack benchmark using 7,264 (200MHz) Pentium Pro microprocessors as part of the ASCI Red project. With the developments in MPP, the rapid rise of SMP architectures and advances in PC technology, the future for parallel CFD looks both promising and challenging.

,

Published: April 1998

Imprint: North-holland

ISBN: 978-0-444-82849-1

## Contents

- Invited Papers.
**Adaptive Schemes.**A generic strategy for dynamic load balancing of distributed memory parallel computational mechanics using unstructured meshes (A.Arulananthan et al.). Communication cost function for parallel CFD using variable time stepping algorithms (Y.P. Chien et al.). Dynamic load balancing for adaptive mesh coarsening in computational fluid dynamics (T. Gutzmer). A parallel unstructured mesh adaptation for unsteady compressible flow simulations (T. Kinoshita, O. Inoue). A fully concurrent DSMC implementation with adaptive domain decomposition (C.D. Robinson, J.K. Harvey). Parallel dynamic load-balancing for the solution of transient CFD problems using adaptive tetrahedral meshes (N. Touheed et al.). Parallel dynamic load-balancing for adaptive unstructured meshes (C.Walshaw et al.).**Combustion and Reactive Flows.**Convergence and computing time acceleration for the numerical simulation of turbulent combustion processes by means of a parallel multigrid algorithm (A. Bundschuh et al.). Coupling of a combustion code with an incompressible Navier-Stokes code on MIMD architecture (G. Edjlali et al.). Parallel simulation of forest fire spread due to firebrand transport (J.M. McDonough et al.).**Association of European Research Establishments in Aeronautics Special Session.**Comparisons of the MPI and PVM Performances by using structured and unstructured CFD codes (E. Bucchignani et al.). Three-dimensional simulation on a parallel computer of supersonic coflowing jets (O. Louedin, J. Ryan). Navier-Stokes algorithm development within the FAME mesh environment (S.H. Onslow et al.). Partitioning and parallel development of an unstructured, adaptive flow solver on the NEC-SX4 (H. van der Ven, J.J.W. van der Vegt).**Distributed Computing.**Parallel workstation clusters and MPI for sparce systems in computational science (A. Berner, G.F. Carey) Inegration of an implicit multiblock code into a workstation cluster environment (F. Cantariti et al.). Parallel solution of Maxwell's equations on a cluster of WS in PVM environment (U. Glucat et al.). Application of the networked computers for numerical investigation of 3D turbulent boundary layer over complex bodies (S.V. Peigin, S.V. Timchenko).**Unsteady Flows.**Simulation of accoustic wave propagation within unsteady viscous compressible gas flows on parallel distributed memory computer systems (A.V. Alexandrov et al.). Parallel solution of hovering rotor flow (C.B. Allen, D.P. Jones). High accuracy simulation of viscous unsteady gasdynamic flows (A.N. Antonov et al.). RAMSYS: A parallel code for the aerodynamic analysis of 3D potential flows around rotorcraft configurations (A. D'Alascio et al.). Multistage simulations for turbomachinery design on parallel architectures (G. Fritsch, G. Möhres).**Applications on Unstructured Meshes.**Massively parallel implementation of an explicit CFD algorithm on unstructred grids, II (B.L. Bihari et al.). Towards the parallelisation of pressure correction method on unstructured grids (Y.C. Chuang et al.). Parallel implementation of a discountinuous finite element method for the solution of the Navier-Stokes equations (A. Codenotti et al.). Hybrid cell finite volume euler solutions of flow around a main-jib sail using an IBM SP2 (N.C. Rycroft et al.). Development of a parallel unstructured spectral/hp method for unsteady fluid dynamics (S.J. Sherwin et al.). Parallel buidling blocks for finite element simulations: application to solid-liquid mixture flows (D. Vanderstraeten, M. Knepley). Parallel CFD computation on unstructured grids (Y.F. Yao, B.E. Richards).**Parallel Algorithms.**A. domain decomposition based parallel solver for viscous incompressible flows (H.U. Akay et al.). Parallelisation of the discrete transfer radiation model (N.W. Bressloff). Study of flow bifurcation phenomena using a parallel characteristics based method (D. Drikakis, A. Spentzos). Efficient parallel computing using digital filtering algorithms (A. Ecer et al.). Parallel implicit PDE computations: algorithms and software (W.D. Gropp et al.). Parallel controlled random search algorithms for shape optimization (Y.F. Hu et al.). Performance of ICCG solver in vector and parallel machine architecture (K. Minami et al.). Parallel iterative solvers with localized ILU preconditioning (K. Nakajima et al.). Last achievements and some trends in CFD (Y.D. Shevelev). The effective parallel algorithm for solution of parabolic partial differential equations system (S.V. Timchenko). Multioperator high-order compact unwind methods for CFD parallel calcuations (A.I. Tolstykh).**Evaluation of Architecture and Machine Performance.**FLOWer and CLIC-3D, A portable flow solving system for block structured 3D-applications: status and benchmarks (H.M. Bleecke et al.). Delft-hydra - an architecture for coupling concurrent simulators (I.J.P. Elshoff et al.). A 3D free surface flow and transport model on different high performance computational architectures (R. Hinkelmann et al.). Recent progress on numerical wind tunnel at the National Aerospace Laboratory, Japan (N. Hirose et al.). Performance comparison of the cray T3E/512 and the NEC SX-4/32 for a parallel CFD-code based on message passing (J. Lepper et al.). About some performance issues that occur when porting LES/DNS codes from vector machines to parallel platforms (M. Porquie et al.). Microtasking versus message passing parallelisation of the 3D-combustion code AIOLOS on the NEC SX-4 (B. Risio et al.). Parallel performance of domain decomposition based transport (P. Wilders).**Navier-Stokes Applications.**Portable parallelization of a 3-D flow solver (T. Bönisch, R. Rühle). Implementation of a Navier-Stokes solver on a parallel computing system (G. Passoni et al.). Parallel application of a Navier-Stokes solver for projectile aerodynamics (J. Sahu et al.). Incompressible Navier-Stokes solver on massively parallel computer adopting coupled method (K. Shimano et al.).**Industrial Applications.**A Multi-platform shared- or distribute-memory Navier-Stokes code (F. Chalot et al.). Predictions of external car aerodynamics on distributed memory machines (H. Schiffermüller et al.). Industrial flow simulations using different parallel architectures (J.B. Vos et al.).**Software Tools, Mappings and Environments.**On the use of Cray's scientific libraries for Navier-Stokes algorithm for complex three-dimensional geometrics (V. Botte et al.). Automatic generation of multi-dimensionally partitioned parallel CFD code in a parallelisation tool (E.W. Evans et al.). ELMER - an environment for parallel industrial CFD (H. Hakula et al.). Semi-automatic parallelisation of unstructured mesh codes (C.S. Ierotheou et al.). Modelling continuum mechanics phenomena using three dimensional unstructured meshes on massively parallel processors (K. McManus et al.). An object-oriented programming paradigm for parallel computational fluid dynamics on memory distributed parallel computers (T. Ohta).**Turbulence.**Numerical study of separation bubbles with turbulent reattachment followed byu a boundary layer relaxation (M. Alam, N.D. Sandham). Efficient parallel-turbulence simulation using the combination method on workstation-clusters and MIMD-systems (W. Huber). Industrial use of large eddy simulation (C.B. Jenssen). High performance computing of turbulent flows with a non-linear v2 - f model on the CRAY T3D using SHMEM and MPI (F.S. Lien). Parallel computation of lattice boltzmann equations for incompressible flows (N. Satofuka et al.). Numerical simulation of 3-D free shear layers (Y. Tsai). Data-parallel DNS of turbulent flow (R.W.C.P. Verstappen, A.E. P. Veldman). Parallel implicit computation of turbulent transonic flow around a complete aircraft configuration (C. Weber).**Environmental and Climate Modeling.**Parallel computing of dispersion of passive pollutants in coastal seas (S. Chumbe et al.). A parallel implementation of a spectral element ocean model for simulating low-latitude circulation system (H. Ma et al.). Modelling the global ocean circulation on the T3D (C. S. Richmond et al.).**Multidisciplinary and Complementary Applications.**ZFEM: Collaborative visualization for parallel multidisciplinary applications (J.R. Cebral). Development of parallel computing environment for aircraft aero-structural coupled analysis (R. Onishi et al.). A parallel self-adaptive grid generation strategy for a highly unstructured euler solver (K. Warendorf, R. Rühle)Invited Papers. Adaptive Schemes. Combustion and Reactive Flows. Association of European Research Establishments in Aeronautics Special Session. Distributed Computing. Unsteady Flows. Applications on Unstructured Meshes. Parallel Algorithms. Evaluation of Architecture and Machine Performance. Navier-Stokes Applications. Industrial Applications. Software Tools, Mappings and Environments. Turbulence. Environmental and Climate Modeling. Multidisciplinary and Complementary Applications.