Parallel implicit unstructured grid Euler solvers

We present fast and highly scalable parallel computations for a number of important and fundamental matrix problems on distributed memory systems (DMS). These problems include matrix multiplication, matrix chain product, and computing the powers, the inverse, the characteristic polynomial, the determinant, the rank, the Krylov matrix, and an LU- and a QR-factorization of a matrix, and solving linear systems of equations. Our highly scalable parallel computations for these problems are based on a highly scalable implementation of the fastest sequential matrix multiplication algorithm on DMS. We show that compared with the best known parallel time complexities on parallel random access machines (PRAM), the most powerful but unrealistic shared memory model of parallel computing, our parallel matrix computations achieve the same speeds on distributed memory parallel computers (DMPC), and have an extra polylog factor in the time complexities on DMS with hypercubic networks. Furthermore, our parallel matrix computations are fully scalable on DMPC and highly scalable over a wide range of system size on DMS with hypercubic networks. Such fast (in terms of parallel time complexity) and highly scalable (in terms of our definition of scalability) parallel matrix computations were rarely seen before on any distributed memory systems.

We present fast and scalable parallel computations for a number of important and fundamental matrix problems on distributed memory systems (DMS). These problems include computing the powers, the inverse, the characteristic polynomial, the determinant, the rank, the Krylov matrix, and an LU- and a QR-factorization of a matrix, and solving linear systems of equations. These parallel computations are based on efficient implementations of the fastest sequential matrix multiplication algorithm on DMS. We show that compared with the best known time complexities on PRAM, our parallel matrix computations achieve the same speeds on distributed memory parallel computers (DMPC), and have an extra polylog factor in the time complexities on DMS with hypercubic networks. Furthermore, our parallel matrix computations are fully scalable on DMPC and highly scalable over a wide range of system size on DMS with hypercubic networks. Such fast and scalable parallel matrix computations were not seen before on any distributed memory systems.

The complex nature of the ocean environment requires advanced computational acoustic models to gain insights into the dominating physical factors controlling the underwater sound propagation and scattering in the ocean. In this study, we explore the “ab-initio” approach to solve wave equations with numerical algorithms that can be implemented in distributed memory parallel computers. The goal is to improve the calculation speed so long-range global scale hydroacoustic wave propagation can be studied more efficiently and effectively. Two major algorithms: the finite difference time domain (FDTD) method and the Parabolic Equation (PE) method are investigated. Two PE-based numerical models are considered. One is the Split-Step-Fourier Parabolic Equation (SSFPE) model using split-step Fourier schemes in three-dimensional (3D) environments. The second PE model is a multi-frequency implementation of the Range-dependent Acoustic Model (RAM) that computes two-dimensional (2D) sound pressure fields. One of the primary challenges in global-scale hydroacoustic propagation modeling with the “ab-initio” approach is the demand for significant computational resources on both memories and computational speeds. To address this, we employed parallel computing techniques using directive-based programming languages, such as OpenACC and XscalableACC, to leverage the power of multiple Graphics Processing Unit (GPU) systems and Central Processing Unit (CPU) cluster computers. Our performance evaluation revealed substantial speedup gains. For the FDTD method, we achieved approximately 3.5-fold and 4-fold speedups with four GPUs and four CPU cluster nodes, respectively. With the 3D SSFPE model, we obtained a speedup of 3.7-fold using four CPU cluster nodes. A significant result was observed with the 2D broadband multi-frequency RAM model, where we achieved a 110-fold speedup compared to one core original implementation. These results demonstrate the promising potential of massively parallel computers for ocean acoustic propagation modeling and highlight the significant performance benefits of utilizing parallel computing techniques. Our findings emphasize the importance of efficient computational strategies.

A 6M digital twin for modeling and simulation in subsurface reservoirs

An efficient parallel-computing method for modeling nonisothermal multiphase flow and multicomponent transport in porous and fractured media

Ray tracing is a well known technique to generate life-like images. Unfortunately, ray tracing complex scenes can require large amounts of CPU time and memory storage. Distributed memory parallel computers with large memory capacities and high processing speeds are ideal candidates to perform ray tracing. However, the computational cost of rendering pixels and patterns of data access cannot be predicted until runtime. To parallelize such an application efficiently on distributed memory parallel computers, the issues of database distribution, dynamic data management and dynamic load balancing must be addressed. In this paper, we present a parallel implementation of a ray tracing algorithm on the Intel Delta parallel computer. In our database distribution, a small fraction of database is duplicated on each processor, while the remaining part is evenly distributed among groups of processors. In the system, there are multiple copies of the entire database in the memory of groups of processors. Dynamic data management is acheived by an ALRU cache scheme which can exploit image coherence to reduce data movements in ray tracing consecutive pixels. We balance load among processors by distributing subimages to processors in a global fashion based on previous workload requests. The success of our implementation depends crucially on a number of parameters which are experimentally evaluated. © 1997 John Wiley & Sons, Ltd.

In this paper high order time integration schemes are investigated for the integration of uid-structure interaction (FSI) model problem with deforming domains. The integration algorithm consists of a partitioned scheme in which a combination of implicit and explicit (IMEX) high order Runge-Kutta time integration schemes is used: the uid and structure systems are decoupled at their interface and both systems are integrated using the implicit system under a given boundary condition at the interface. The inuence of this boundary condition (the coupling term) is integrated using the explicit Runge-Kutta scheme. The uid dynamics are written in the arbitrary Lagrange Eulerian formulation to cope with the deforming uid mesh. An expression is proposed for the mesh face velocities of the deforming mesh such that the discrete geometric conservation law is satised. For the model problem numerical investigations are made into the eciency of these high order time integration schemes compared to lower order methods. The IMEX scheme retains the order of the implicit and explicit schemes and obtains the same accuracy as a second order staggered scheme for less computational work.

Currently, there are two approaches to implementing large-scale ice sheet simulations. One approach, which is based on the use of an evenly-spaced structured grid, is to simulate the entire ice sheet at the highest resolution required. The other approach is to model the ice sheet at different levels of resolution using unstructured meshes. Structured grids lend themselves quite well to partitioning on multi-core, distributed memory parallel computers, leading to good computational efficiency, load-balancing and reduced communication costs. The disadvantage is the significant computational resources required to model a system of this size at a high enough resolution to correctly model regions of the ice sheet undergoing rapid change. Unstructured meshes can reduce such costs by allocating more computational resources to areas of the ice sheet whose dynamics are evolving quickly and fewer resources to areas evolving more slowly. However, unstructured meshes do not easily lend themselves to effective decomposition on distributed memory parallel computers, and can experience poor cache utilization, significant load imbalance, high communication costs, and increased synchronization times. In this paper, we describe our alternative approach based on embedded simulation, which provides a degree of multi-resolution capabilities to structured grids while maintaining many of the computational efficiencies that come with the structured approach. We provide experimental results focusing on the scalability characteristics and cache efficiency of each approach as a function of increasing problem size and core counts. We show that the embedded approach scales very well across all problem sizes and core counts, and that the unstructured approach scales well at smaller problem sizes and core counts but degrades significantly as the problem size becomes large.

The unstructured triangular mesh is successfully applied to the steady simulations of the slider due to its flexibility, accuracy and mesh efficiency in capturing various complex ABS rails and recess wall regions. This paper introduces a new implicit algorithm with second order time accuracy for the time-dependent simulations of the slider dynamics based on the unstructured triangular mesh. The new algorithm is specially developed for the unstructured triangular mesh and the finite volume method. It is applied to simulate the load/unload details based on the 9-D model, such as the influences of the limiter position on the flying height during the unload process, and the development and the 3-D profiles of the air bearing pressure during the load process. Because of the mesh efficiency of the unstructured triangular mesh and the flexibility of applying greater time steps, the simulation times are significantly shorter than those of the structured rectangular mesh. The simulation results based on the algorithm are in good correlation with the experimental results.

PurposeA computational fluid dynamics code for the calculation of laminar hypersonic multi‐species gas flows in chemical non‐equilibrium in axisymmetric or two‐dimensional configuration on shared and distributed memory parallel computers is presented and validated. The code is designed to work efficiently in combination with an automatic domain decompositioning method developed to facilitate efficient parallel computations of various flow problems.Design/methodology/approachThe baseline implicit numerical method developed is the lower‐upper symmetric Gauss‐Seidel scheme, which is combined with a sub‐iteration scheme to achieve time‐accuracy up to third‐order. The spatial discretisation is based on Roe's flux‐difference splitting and various non‐linear flux limiters maintaining total‐variation diminishing properties and up to third‐order spatial accuracy in continuous regions of flow. The domain subdivision procedure is designed to work for single‐ and multi‐block domains without being constrained by the block boundaries, and an arbitrary number of processors used for the computation.FindingsThe code developed reproduces accurately various types of flows, e.g. flow over a flat plate, diffusive mixing and oscillating shock induced combustion around a projectile fired into premixed gas, and demonstrates close to linear scalability within limits of load imbalance.Research limitations/implicationsThe cases considered are axisymmetric or two‐dimensional, and assume laminar flow. An extension to three‐dimensional turbulent flows is left for future work.Originality/valueResults of a parallel computation, utilising a newly developed automatic domain subdivision procedure, for oscillating shock‐induced combustion around a projectile and various other cases are presented. The influence of entropy correction in Roe's flux‐difference splitting algorithm on diffusive mixing of multi‐species flows was examined.

Measurement and modelling of distributions of data communication times is commonly done for telecommunication networks, but this has not previously been done for message passing communications on parallel computers. We have used the MPIBench program to measure distributions of point-to-point MPI communication times for two different parallel computers, with a low-end Ethernet network and a high-end Quadrics network respectively. Here we present and discuss the results of efforts to fit the measured distributions with standard probability distribution functions such as exponential, lognormal, Erlang, gamma, Pearson 5 and Weibull distributions.KeywordsWeibull DistributionParallel ComputerParallel ProgramInterarrival TimeCommunication TimeThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

The production of realistic image generated by computer requires a huge amount of computation and a large memory capacity. The use of highly parallel computers allows this process to be performed faster. Distributed memory parallel computers (DMPCs), such as hypercubes or transputer-based machines, offer an attractive performance/cost ratio when the load balancing has been balance and the partition of the data domain has been performed. This paper presents a parallel ray tracing algorithm for DMPC using a Shared Virtual Memory (SVM) which solves these two classical problems. This algorithm has been implemented on a hypercube iPSC/2 and results are given.

Parallel Iterative Methods with Factored Preconditioning Matrices for Solving Discrete Elliptic Equations

Parallel approximate factorization method for solving discrete elliptic equations

A three-dimensional unsteady compressible viscous flow solver has been developed, and applied to a low speed centrifugal compressor configuration. Good agreement with experiments have been obtained for a compressor tested at NASA Lewis Research Center. Comparisons with measured surface pressures, and performance data are given. For this compressor, stall was found to occur in the diffuser region. It is demonstrated that this stall may be eliminated, and stable operation may be restored by the use of bleed valves located on the diffuser walls. INTRODUCTION In many army applications (tanks and rotorcraft) small gas turbine engines are often used. The mass flow rate in these engines is small (below 10 Ibs/sec). The compression is provided by one or two centrifugal compressor stages, sometimes in combination with an axial compressor stage. Centrifugal compressors are well suited for handling such low mass flow rates. They are rugged and compact. However, the geometric complexity of the blades and the twisting blade passages give rise to a host of aerodynamic problems such as unsteadiness, establishment of cross flows, flow separation at the trailing edge of the compressor blade, separation over diffuser guide vanes, and unsteady meandering motion of a 'tip vortex' that originates in the vicinity of the gap between the rotor and the casing. In many instances, this unsteadiness leads to rotating stall and surge. Considerable interest exists in the engine community in understanding and eventually controlling these phenomena, particularly compressor stall and surge. Computational fluid dynamics methods provide an efficient way of studying the complex flow phenomena within centrifugal compressors. Thanks to the availability of post-processing and scientific visualization software, the computed flow fields can be extensively analyzed. It is also convenient to test within a numerical model various control strategies such as bleeding, energizing of incoming flow using jets, excitation /wiggling of guide vanes, etc. Passive mechanisms such as slots, slats and flaps can also be easily studied within the scope of a numerical method. These simulations are thus a useful and economical first step to detailed flow control experiments. Finally, CFD simulations provide a wealth of numerical and flow physics information which are useful in developing reduced order models. For these reasons, a 3-D CFD code capable of analyzing centrifugal compressor configurations has been developed by the present investigators. Rather than solving the flow equations in a rotating coordinate system, the present calculations are carried out in an inertial system, with grid velocity terms used to simulate a rotor. Inflow and outflow boundary conditions appropriate for internal flows in general and centrifugal compressors in particular are used. This paper is organized as follows. The flow solution algorithm is first described. Next, results in the form of velocity fields, surface pressure data and performance map are presented for a NASA low-speed centrifugal compressor. The performance of this compressor for off-design low mass flow ratio conditions is subsequently studied. Methods for improving the compressor performance during such conditions are discussed. NUMERICAL FORMULATION The 3-D unsteady compressible Reynolds averaged Navier-Stokes equations are solved numerically using a time marching scheme. This involves solving the governing equations at each time step by marching in time from an initial flow condition with appropriate boundary conditions. The governing equations are cast in a curvilinear body-fitted coordinate system (£,T|,£) and solved using an Alternating Direction Implicit (ADI) scheme 1. This implicit scheme has been validated for a number of fixed wing and rotary wing configurations both in steady and unsteady flight conditions^'*. The partial derivatives in the governing equations are calculated using finite difference approximations. The Graduate Research Assistant, Student Member AIAA. * Regents' professor, Associate Fellow AIAA. Copyright© 1997, American Institute of Aeronautics and Astronautics, Inc. inviscid flux terms are calculated using a third order upwind scheme and the viscous stress terms are calculated using standard central difference representations. The details of the upwind formulation are given by Roe. The time derivative is approximated using a two point first order backward difference. The finite difference representation of the governing Navier-Stokes equations at time level 'n' is components at the inflow boundary were set to zero, assuming that there was no swirl at the inlet. It is possible within the framework of the present boundary conditions to prescribe swirl, non-uniform total pressure and total temperature at the inlet. At every time step, the speed of sound a and the normal component of velocity un were found at the inflow boundary from the relations: AT 2a 2a — u = n — u n = I 8 f R + S S + S , T (1) Interior where Aq* is the change in the flow property vector q during adjacent time levels and At is the time step. The operators 8£, 5r| and 8^ are the standard central difference operators. The viscous terms are evaluated explicitly and the inviscid flux vectors F, G and H are calculated implicitly at time level n+1. A three factor approximate factorization is used to factorize the implicit coefficient matrix operator into block tri-diagonal matrices. The factorized system of equations is solved using the Thomas algorithm. Turbulence effects are modeled using the SpalartAllmaras model. INITIAL AND BOUNDARY CONDITIONS The centrifugal compressor is shown in figure 1. The boundary grids are also shown for a single blade passage. The volume grid is made of three segments: inlet region, the rotor, and the diffuser. The following boundary conditions were applied: Periodicity: In the present study, the flow through only a single blade passage has been studied, assuming blade-toblade periodicity. At the periodic boundaries, the flow properties are computed by averaging the properties on either side of the periodic boundary. Solid walls: At the solid walls, the no slip boundary condition was used. The velocity on the walls of the inlet and the diffuser were set to zero, while the velocity for grid points on the compressor blades and the shaft were set equal to Q x r. The pressure, density and temperature values at the solid surfaces were extrapolated from the interior using relationships such as 9p/3n=dT/9n=3p/3n=0, where n represents the direction normal to the solid surface. Inflow Boundary: At this boundary, the stagnation temperature T0 and the total pressure p0 were assumed known. In the simulation of a compressor drawing the air from the atmosphere, these quantities may be set based on ambient conditions. In a multistage compressor analysis, these quantities will be known at the downstream boundary of the previous stage. The tangential (vand w-) (2)