Parallel Computational Fluid Dynamics 2003 (eBook)

Cover 1
Preface 5
Acknowledgements 7
Table of Contents 9
Mathematical Modeling using Supercomputers with Parallel Architecture 17
Introduction 17
Turbulence and Hydrodynamic Instabilities 18
Supersolver 21
Applications 21
Gas Dynamics (CFD) 21
Hydrodynamic Instabilities 22
Seismic Data Processing 23
Safety of Housing and Industrial Constructions under Intensive Dynamic Loadings 23
Nonlinear Contact Shell Dynamics 23
Computer Models in Medicine 23
Conclusion 24
References 24
A Review of Lattice Boltzmann Models for Compressible Flows 35
Introduction 35
Conventional LB Model 36
Adjustable Speed of Sound Models 36
Full Thermal LB Models 38
Adaptive Velocity Compressible LB Model 39
Concluding Remarks 44
Acknowledgement 44
References 44
DNS of the Generation and Propagation of Sound 45
Introduction 45
Numerical Method 46
Results and Discussion 46
Variation with Body Shape 46
Active Control of Sound 49
Conclusions 51
References 51
CFD on the BlueGene/L Supercomputer 53
System Hardware 53
Operating System Architecture 54
Programming Models 55
Control System 55
Advances in Parallel CFD 55
The BlueGene/L and CFD 56
References 56
Software Frameworks for Integrated Modeling 57
Integrated Modeling in River/Coastal Engineering 57
Component Based Software Development (CBD) 58
Component Based Approaches for Large Scale Modeling 59
Communication Mechanisms for the Components 61
Specification of Components' Interfaces 61
Dealing with Interpolation and Other Data Conversions 62
Conclusions 63
References 64
Parallel Algorithms of Direct Solving the Boltzmann Equation in Aerodynamics Problems 65
Introduction 65
Numerical Methods and Formulation of the Problems 66
Parallel Implementation 66
Results 68
Conclusion 1
References 72
Aitken Acceleration of Dirichlet - Neumann Algorithm for Flow in Porous Media with Discontinuous Permeability 73
Introduction 73
Physical Equations 74
Finite Element Resolution 74
Dirichlet-Neumann Algorithm 75
Convergence 75
Aitken Acceleration of the Dirichlet-Neumann Algorithm 76
1-D Framework, 2 Sub-Domains 77
2-D Framework, 2 Sub-Domains 78
2-D Framework, 4 Sub-Domains 79
Conclusions 80
References 80
Parallel Algorithm for Numerical Simulation of 3D Incompressible Flows 81
Introduction 81
Quasi-Hydro-Dynamic Equations 81
Parallel Implementation 83
Computational Results 85
References 86
Big Unstructured Mesh Processing on Multiprocessor Computer Systems* 89
References 95
Parallel Conjugate Gradient Preconditioning via Incomplete Cholesky of Overlapping Submatrices 97
New Theory of the Conjugate Gradient Method 97
Parallel Preconditioner 98
The Idea of the Preconditioner 98
Formal Description of Parallel Preconditioner 99
Reordering and Block Splitting 100
Parallel Implementation 100
Numerical Experiments 101
Conclusion 104
References 104
Coupling Scheme for Continuum and Parallel DSMC Parts to a Numerical Solution of a Two-dimensional Subsonic Problem 105
Introduction 105
Statement of the Problem 106
Velocity Simulations for the Incoming Molecules 107
Evaluation of Convergence for the Present Coupling Scheme 109
Some Numerical Results of the Simulation 110
Conclusions 112
Acknowledgments 112
References 112
Parallel Iterative Methods with Factored Preconditioning Matrices for Solving Discrete Elliptic Equations 113
Introduction 113
The Preconditioning Matrices of PVICCG, PVMICCG, AND PATMCG 114
Parallel Methods for Solving 2D Problems on Orthogonal Grids 114
Parallel Methods for Solving 2D Problems on Locally Refined Grids 116
Parallel Methods for Solving 2D Problems on Uniform Triangular and Unstructured Triangular Grids 117
Parallel Methods for Solving 3D Problems on Orthogonal Grids 119
Conclusion 120
References 120
A Local Grid Refinement Algorithm on Modern High-Performance Computers 121
Introduction 121
Numerical Procedure 121
Grid Refinement Method 122
Application 125
Conclusion 128
References 128
Viscous Gas Flow Simulation on Nested Grids using Multiprocessor Computer Systems 129
Introduction 129
Choice of Numerical Methods 130
Nested Grid Partitioning 131
Test Problems Simulation Results 132
References 134
Domain Coupling with the DOVE Scheme 135
Introduction 135
Method 136
Initialization 137
Communication 137
Results 138
Poisson Solver 138
Flow Solver 140
Conclusions and Future Work 141
References 143
Crash-Free Computational Technology for LEGAK Calculation of Solid Flows on Distributed Memory Multiprocessors 145
Introduction 145
Legak Basics 145
Crash-Free Technology 146
Scheme Parallelization 147
Some Numerical Legak Examples of High-Strain Interfaces 147
Conclusions 152
References 152
On Arbitrary-Order Schemes for Parallel CFD Calculations * 153
Introduction 153
Noncentered (Upwind) Multioperators for Convective Terms 154
Centered Multioperators 155
Numerical Examples 156
References 159
Domain Decomposition Techniques to Solve the Navier-Stokes Equations 161
Introduction 161
The Navier-Stokes Equations 162
Approximation in Space 162
Domain Decomposition Method 163
Splitting Schemes 165
Scheme of Full Decomposition 165
References 167
Parallel Adaptive Solution of the Stokes and Oseen Problems on Unstructured 3D Meshes 169
Introduction 169
Model Problems 170
Quasi-Optimal Meshes and their Properties 171
Sequential Iterative Solution of Saddle Point Systems 173
Parallel Iterative Solution of Saddle Point Systems 175
Conclusion 176
References 177
Numerical Solution of a Tidal Wave Problem 179
Numerical Simulation of Initial Perturbation Growth for Oblique Impact of Metal Plates 189
Introduction 189
Numerical Methods, Problem Setup 189
Computational Results 191
Collision Velocity Effect 192
Simulation of the Metal Interface Instability with Varying Strength Properties 192
Metal Interface Instability Simulation with Varying Melting Heat Values 194
Metal Interface Instability Simulation with Varying Initial Perturbation Wavelengths 194
Variation of the Perturbation Wavelength 195
Conclusions 195
References 195
High Accuracy Simulation of Acoustic Noise from Sources in Pipes 197
CAA Software Features 197
General Problem Formulation 198
High Accuracy Numerical Techniques 198
MacCormack Methods 198
Compact Stencil Differentiation Technique 199
Dispersion Relation Preserving Schemes 200
Boundary Conditions 200
Testing 201
Benchmarks Description 201
Test Results 201
Numerical Results 202
Conclusion 203
References 203
Parallel Computing of 3D Separated Homogeneous and Stratified Fluid Flows Around the Bluff Bodies 205
Numerical Method 205
Homogeneous Flow Around a Sphere 206
Homogeneous Flow Around a 3D Circular Cylinder 210
Stratified Flow Around a Sphere 210
Stratified Flow Around a 2D Circular Cylinder 211
References 212
A Study of a Gravity-Gradient Mixing Properties by the Means of a Direct Numerical Modeling 213
References 219
Parallel DNS for Flow Separation and Transition Around Airfoil 221
Abstract 221
Introduction 221
Governing Equations and Numerical Methods 222
Governing Equations 222
Numerical Methods 222
Problem Definitions 222
Results and Analysis 223
2D Instability Waves and Vortex Shedding 223
3-D Non-Linear Instability and Breakdown 225
Instantaneous Characteristics 227
Conclusion 227
Acknowledgment 228
Reference 228
Parallelization of Phase-Field Model for Phase Transformation Problem 229
Introduction 229
Governing Equations of Phase-Field Model 230
Numerical Methods and Problem Parameters 231
Approach to Parallelization and Results 232
Summary 232
Acknowledgements 234
References 234
DNS and RANS/LES-Computations of Complex Geometry Flows using a Parallel Multiblock Finite-Volume Code 235
Introduction 235
Parallel Multiblock Finite-Volume Code 236
Some Applications of the Code SINF 237
Silicon Melt Convection in a Crucible of a Czochralski Crystal Growth System 237
Massively Separated Flow in a Model of a Gas Turbine Exhaust Diffuser 239
Turbulent Flow through a Single Row of Square Bars 240
Acknowledgements 242
References 242
Numerical Methods and Parallel Algorithms for Study of Burgers' Turbulence * 243
Introduction 243
Numerical Methods and Algorithms 244
Inviscid Limit 244
Finite Viscosity Case 246
Results and Discussion 247
The Structure Function Scaling Analysis 247
Checking the Extended Self-Similarity (ESS) Property 247
Conclusions 249
References 250
Numerical Study of Gravitational Turbulent Mixing in Alternating-Sign Acceleration 251
Setting up 3D Computations 251
Results of 3D Computations 252
Comparison with Results of Calculations by the k-Epsilon Turbulence Model 255
Conclusion 256
References 256
Development of a CFD-Based Model for Photo-Reactor Simulation 259
Introduction 259
Model Development 260
Kinetics 260
Radiation 261
Hydrodynamics 261
Reactor Performance Calculation 261
Results and Discussion 263
Concluding Remarks 266
Acknowledgment 266
References 266
An Unstructured Parallel Solver for Multi-Phase and Reactive Flows in Internal Combustion Engines 267
Introduction 267
Numerical Methods 268
Code Capabilities 269
Parallel Performances 270
Gasoline Direct Injection Engine Simulations 271
Conclusion 272
References 273
Grid Scheduler with Dynamic Load Balancing for Parallel CFD 275
Introduction 275
Requirement to the Scheduler 276
Scheduler Design 277
Experiment 279
Conclusion 281
Acknowledgement 281
References 282
Aitken-Schwarz Method for CFD Solvers 283
Introduction 283
A General Framework for the Aitken-Schwarz Method 284
A CFD Application 288
References 291
A Grid-Based Flow Analysis and Investigation of Load Balance in Heterogeneous Computing Environment 293
Introduction 293
Governing Equations & Numerical Schemes
Investigation of Parallel Efficiency in the Grid 294
Results 296
Dynamic Analysis of Strap-on Boosters in Multi-Stage Launch Vehicle 296
Application of a Load Balance Algorithm 297
Conclusion 299
Acknowledgement 299
References 299
Krylov Method Algorithms Adapted to the Grid Computing 301
Motivations for the Grid Computing 301
A Navier-Stokes Solver Example 302
A Multigrid Solver with a Coarse Grid Domain Decomposition Methods 303
How to Delay the Dot Product in Krylov Methods ? 305
Convergence Results 306
Conclusions 307
References 308
Parallel Computing in Grid Environment 309
Introduction 309
Background on Grid Computing Tools Used 310
Secure Communication Tool in Grid Computing: Globus 310
A Parallelization Library: GPAR 310
Parallel Tasks Distribution and Management Tool: DLB 310
Running Parallel Jobs on Grid 311
Fault Tolerance Results 311
Applications with Grid Job Scheduler between Clusters 313
Jobs in the Globus Environment without the Current Grid Job Scheduler 316
Conclusion 316
Acknowledgement 316
References 316
Parallel Computations in Problems of Climate Modeling* 317
Introduction 317
Communication Kernel 318
Eulerian Dynamical Core 320
OpenMP Implementation of the INM Ocean Model 322
Acknowledgments 323
References 324
New Automatic Code Radiation Earth with Parallel Computing 325
References 329
Hydrothermodynamic Model for Sea Surface Layer and its Realization on the Distributed Computing Cluster 331
References 333
Mathematical Modeling of the Multidimensional Radiation Transfer Problems with Parallel Computing 335
Introduction 335
Mathematical Statement of a Problem 339
Conclusions 342
References 342
Development of a High Resolution Parallel Ocean Circulation Model on the Earth Simulator 343
Introduction 343
Basic Equations 343
Parallelization and Vectorization 344
The Barotropic Equation Solver 345
Computation Performance 345
Computation Results 346
References 348
A Comparison Study of Computational Performace between a Spectral Transform Model and a Gridpoint Model 349
Introduction 349
Model Description 351
AFES 351
NICAM 351
Comparison of Computational Performance 351
Computational Conditions 351
Resolution Correspondence 352
Results 352
Discussion 353
Conclusion 355
Acknowledgment 355
References 355
Solving Economic and Environmental Optimal Control in Dumping of Sewage with a Flexible and Parallel Evolutionary Computation 357
Introduction:Numerical Model and Optimisation Problem 357
Flexible Evolution Agent with Parallel Evolutionary Computation 358
Application Example 360
Conclusions 362
References 362
Bremsstrahlung Account in Photon Transport 365
Introduction 365
TTBIAS Model 366
Annihilation Model 367
Comparative Calculations 368
Parallelization 371
Acknowledgement 371
References 371
Background of Computations for Mathematical Models, based on Conservation Laws Systems and Applications to Fluid Dynamics 373
Introduction 373
Difference Scheme for Boltzmann Type Equations and Direct Monte-Carlo Simulation in Physical Kinetics 375
Functional Solutions Theory 376
References 379
Parallel Simulation of Laser-Induced Shock Wave Reflection Transition * 381
Introduction 381
Problem Formulation and Modeling Issues 383
Problem Formulation 383
Numerical Techniques 383
Parallel Simulation 384
Validation 384
Results 384
Conclusions 387
References 388
Application of Parallel Programming Method for 3D MHD Computer Simulations of Magnetic Reconnection in Plasma 389
Introduction 389
Mathematical Model 390
Computational Algorithm 390
Multiprocessor Implementation of Computational Algorithm 391
Results of Computer Simulation 393
References 395
Solution of Wave Equations for a Moving Charge in the Media with Discontinued Dielectric Properties 397
Introduction 397
Maxwell's Equation for Moving Charge in Homogenious Medium 398
Electromagnetic Potentials of Point Charge in Case of Media with Discontinued Dielectric Properties 401
Transition Radiation of the Carge at Crossing Interface between Media with Different Dielectric Properties 403
Flux of Electromagnetic Radiation of Moving Charge Crossing the Interface 405
Conclusion 407
References 407
A Code Coupling Application for Solid-Fluid Interactions and Parallel Computing 409
Introduction 409
The CFD Solver 410
The CSD Solver 411
Coupling of Codes 412
Metacomputing Test Case 412
Conclusion 413
Acknowledgement 413
References 413
Flow Vision Software: Numerical Simulation of Industrial CFD Applications on Parallel Computer Systems 417
Introduction 417
Numerical Methods 418
Parallel Implementation for Small Number of Processes 419
Parallel Implementation for Large Number of Processes 422
Numerical Results 423
Conclusion 423
References 424
Object-Oriented Parallel CFD with JAVA 425
Introduction 425
Java for Scientific Computing 426
Multi-Threaded Matrix/Vector Product in Java 427
Multi-Threaded Multiplicative Schwartz Overlapping Method for a Stabilized Navier-Stokes Formulation 429
Conclusion 431
References 432
Development of Scalable Parallel Programs in ParJava Environment* 433
Introduction 433
Brief Description of ParJava Environment 434
ParJava Tools Supporting the Development of SPMD-Programs 435
Instrumentation and Profiling 436
Symbolic Execution 436
Trace Visualization 437
Related Works 437
Conclusion 438
References 438
Object-Oriented Software for Domain Decomposition Methods with Local Adaptive Refinement 441
Introduction 441
Finite Element Object-Oriented Framework 442
Domain Decomposition 443
Adaptive Refinement 446
Load Balancing 446
Parallel Distributed Components 446
Example 447
Conclusions and Future Research Directions 448
References 448
Large Data Volume Visualization on Distributed Multiprocessor Systems /* 449
Introduction 449
Compression of Iso-Surface 450
Compression of Zones of Unambiguity 450
Compression by Consecutive Excluding of Units 453
References 454
Programs LUCKY & LUCKY_C - 3D Parallel Transport Codes for the Multygroup Transport Equation Solution for XYZ Geometry by PmSn Method
Supercomputer MBC-1000M and Efficiency of Parallel Calculation 455
The Main Position for the Parallel Technology 456
Geometrical Module and Means to Control Calculated Model 457
Method to Accelerate Iterative Process 458
Efficiency as Function of Parameters for Calculated Model 459
Conclution 462
References 462
Multiobjective Asynchrone Parallel Genetic Algorithm for Reentry Trajectory Optimization* 463
Introduction 463
Problem Description 464
Solution Method 465
Parallel Implementation 466
Parallel Asynchrone Floating-Point GA 467
Multiobjective Parallel Asynchrone GA 468
Results and Discussions 468
Conclusion 469
References 469
Combined Model OpenMP& MPI for Parallel Computations in Gas Dynamics Program
An Example to the Numerical Processing of Nonisothermal Flow of non- Newtonian Fluids 479
Introduction 479
Isothermal Case 480
The Non-Newtonian Fluid Chosen (CEF Model) 480
Material Functions in Steady-state Shear Flows 481
Explicit Expressions of the Dimensionless Governing Equations 481
Nonisotermal Case 482
Temperature Effects 482
Conclusions 482
References 483
A Spectral-Element Model of Mantle Convection 485
Abstract 485
Model 485
Simulation Results 486
Acknowledgments 486
References 487
Parallelization of a Modern CFD Incompressible Turbulent Flow Code 489
Introduction 489
Governing Equations and Implementation 490
Turbulence Model 490
Parallelization 492
Results 494
References 494
Direct Numerical Simulation of Turbulent Natural Convection Flows Using PC Clusters 497
Introduction 497
Governing Equations and Numerical Methods 498
Parallel Algorithms 498
Introduction 498
Target Problems and Domain Decomposition 499
Fourier Decomposition 499
Direct Schur Decomposition 500
Schur Complement Overview 500
Benchmark Results and Discussion 501
Turbulent Thermal Driven Cavity 503
References 504
Solution of Computational Fluid Dynamics Problems on Parallel Computers with Distributed Memory 505
Introduction 505
Simulation of Low-Mach-Number Flows 505
Gases of Usual and Weak Rarefied Density 505
Dense Gases and Liquids 507
Numerical Simulation of Hydrodynamic Phenomena in a Fast Contact Reactor 509
Conclutions 511
References 511
Simulating Microfluidic Devices on Parallel Computers 513
Introduction 513
The Properties of Fluids at the Micro-Scale 514
Governing Equations 515
Numerical Methodology 516
Modelling Flow in a Rapid Mixing Device 517
Modelling of a Continuous Flow Device 518
Concluding Remarks 519
Acknowledgements 519
References 519
Multilevel Parallelization Strategy for Optimization of Aerodynamic Shapes 521
Introduction 521
Solution of Optimization Problem 522
Multilevel Parallelization Strategy 523
Analysis of Results 525
Conclusions 527
References 527
Parallel Computation of Micro Fluid Dynamic Problems 529
Introduction 529
Kinetic Approach 530
BGK Model Equation 530
Non-Dimensionalization 531
Reduced Distribution Functions 531
Numerical Procedure 532
Numerical Results 533
Flow Past NACA0012 Airfoil 533
Micro-Channel Flow 534
Parallel Computation of Flow Past Micro-Sphere 535
Conclusions 535
References 535
Numerical Modeling of External 3-D CFD Problems on the Parallel Computers and Aerodynamic Shape Optimization 537
Variables of Designing, Restriction and Methods of Optimization 537
Methods of Governing Equations Solution 539
Surface Generation 539
Grid Generation Procedure 540
The Basic Algorithm of Parallel Implementation 540
Numerical Tests and Results 541
The Conclusion 543
References 544
Numerical Simulations of Supersonic Combustion of Methane-Hydrogen Fuel in an Experimental Combustion Chamber 545
Introduction 545
Computational Means and Models 546
PC Cluster 546
MSD Code 546
LCSR Chemistry Model 547
LAERTE Combustion Chamber 548
Results 549
MSD Performance 549
LAERTE Simulations 550
Acknowledgement 552
References 552
Equations of State of Matter at Extreme Conditions 553
Introduction 553
Eos Problems 553
Eos of Metals 554
Eos in Numerical Modeling 555
Conclusions 557
References 558