Computational fluid mechanics and massively parallel processors
Abstract
An abstract is not available.
- ... Our performance prediction model is based on the work done by Clement and Quinn 6] tional uid dynamics (see, e.g., 7, 17, 2, 13, 20]). But unlike any previous work in this area, we have introduced a methodology for automatic generation of eecient parallel programs for FEM applications on a wide spectrum of parallel and distributed architectures. ...We describe an architecture-adaptable methodology for the parallel implementation of finite element numerical models of physical systems. We use a model of time-dependent ocean currents as our working example. The heart of the computation is the solution of a banded linear system, and we describe an algorithm based on the domain decompositionmethod to solve the banded system. The algorithm is represented in a divide-and-conquer framework facilitates easy implementation of various algorithmic options. The process is straightforward and amenable to automation. We demonstrate the validity of this approach using two radically different target machine, a workstation network and a supercomputer. Our results show very good speedup on both platforms.
- Parallel processing has changed the way much computational physics is done. Areas such as condensed matter physics, fluid dynamics, and other fields are making use of massively parallel computers to solve immense and important problems in new ways. Simulating wave propagation is another area that has benefited through the use of parallel processing. This is graphically illustrated in this article by various numerical simulations of ultrasonic pulses propagating through solids carried out on a massively parallel computer. These computations are accompanied by visualizations of the resulting wavefield. The calculations and visualizations, together, can be completed in only seconds to several minutes and compare well with experimental data. The computations and parallel processing techniques described should be important in related fields, such as geophysics, acoustics, and mechanics.
- We describe an architecture-adaptable methodology for the parallel implementation of finite element numerical models of physical systems. We use a model of time-dependent ocean currents as our working example. The heart of the computation is the solution of a banded linear system, and we describe an algorithm based on the domain decomposition method to solve the banded system. The algorithm is represented in a divide-and-conquer framework. The well-defined algebraic structure of this framework makes it relatively easy to predict the performance of the algorithm with high accuracy. Performance prediction makes possible architecture adaptability. In the case of the domain decomposition method described in this paper, performance prediction enables us to make decisions about granularity and the number of processors to use for individual steps of the algorithm. The divide-and-conquer framework facilitates easy implementation of various algorithmic options. The process is straightforward an...
- The mathematical bases of vortex methods for the numerical solution of flow problems are explored in an analytical review, with an emphasis on techniques for which convergence proofs have been published. Sections are devoted to basic methods for two-dimensional inviscid flows, refinements of these methods, methods for two-dimensional viscous flows, and methods for three-dimensional inviscid and viscous flows. Five problems for which convergence proofs are lacking are listed and briefly characterized.
- We describe an interactive, animated graphics environ ment that can be used to visualize the results of com putational fluid dynamics simulations. With this sys tem, the user can visualize flow past bodies or in chambers, using a variety of techniques borrowed from laboratory experiments, including dye injection, smoke visualization, and bubble wires. The environ ment provides the capability of three-dimensional rota tion of the flow under study, as well as placement of internal flow probes that can read location, velocity, and secondary quantities (such as vorticity and pres sure), which can be calculated from available data as the visualization unfolds. The system is implemented on a Connection Machine 2 attached to a framebuffer. As input, it accepts discretized data from a three- dimensional fluid flow calculation. Data should be provided on a three-dimensional grid at discrete time steps, and internal bodies and flow boundaries are de scribed using discrete panels. Input parameters are menu-driven, and images are updated at five frames per second. A prototype, second version of the system is implemented on a Connection Machine 5 using the Application Visualization System.
- We present a numerical technique to approximate the solution of a simplified model of turbulent combustion. The method, which is particularly suited for flows at high Reynolds number, uses random vortex element techniques coupled to a flame propagation algorithm based on Huyghens' principle. We use this technique to analyze combustion in open and closed vessels. In the first experiment, we model a flame propagating in a swirling, viscous flow inside a closed square. Our results show the growth and development of counterrotating turbulent eddies and their effect on the flame. In the second experiment, we model turbulent combustion within a channel, in which flow enters through a slit at one end. Results detail the effects of exothermicity and viscosity on the speed and shape of the burning front.
- We describe a new graphics environment for essentially real-time interactive visualization of computational fluid mechanics. The researcher may interactively examine fluid data on a graphics display using animated flow visualization diagnostics that mimic those in the experi mental laboratory. These tools include display of moving color contours for scalar fields, smoke or dye injection of passive particles to identify coherent flow structures, and bubble wire tracers for velocity profiles, as well as three-dimensional interactive rotation and zoom and pan. The system is implemented on a data parallel super computer attached to a framebuffer. Since most fluid visualization techniques are highly parallel in nature, this allows rapid animation of fluid motion. We demonstrate our interactive graphics fluid flow system by analyzing data generated by numerical simulations of viscous, in compressible, laminar and turbulent flow over a back ward-facing step and in a closed cavity. Input param eters are menu-driven, and images are updated at nine frames per second.
- The design and application of vortex methods for the numerical solution of flow problems are discussed in an analytical review. Topics addressed include the Navier-Stokes and vorticity transport equations, approximate Lagrangian particle-trajectory approaches, numerical approximations to the equations of motion for free-space inviscid and incompressible viscous flows, boundary conditions, and fast summation methods. Particular attention is given to the treatment of vortex sheets, vortex pairing, mixing and shear layers, two-fluid problems, external and internal flow problems, reactive flows, and three-dimensional problems. Techniques for measuring the accuracy of a random vortex simulation are examined, and basic convergence results are summarized.
- A numerical method for solving the time-dependent Navier–Stokes equations in two space dimensions at high Reynolds number is presented. The crux of the method lies in the numerical simulation of the process of vorticity generation and dispersal, using computer-generated pseudo-random numbers. An application to flow past a circular cylinder is presented.
- We have designed and built the Orrery, a special computer for high-speed high-precision orbital mechanics computations. On the problems the Orrery was designed to solve, it achieves approximately 10 Mflops in about 1 ft3of space while consuming 150 W of power. The specialized parallelarchitecture of the Orrery, which is well matched to orbital mechanics problems, is the key to obtaining such high performance. In this paper we discuss the design, construction, and programming of the Orrery. Copyright © 1985 by The Institute of Electrical and Electronics Engineers, Inc.
