Stanford University Department of Aeronautics & Astronautics Aerospace Computing Laboratory

Numerical Methods

The principle focus of the ACL is the development and application of numerical techniques in the desgn of aerospace products. The basis of these numerical techniques lies in the application of multigrid methods pioneered by Proessor Jameson in the past decades. These methods are being used to solve mathematical models of fluid flow ranging from the linearized potential flow equations to the fully non-linear unsteady Navier-Stokes equations. The computational efficiency of these techniques have made them the de-facto standard in the aerospace industry. These codes have been used to analyze and design vehicles ranging from sailboats to commercial airliners.

Optimization Methods

Professor Jameson has also been a forerunner in the development of adjoint theory as applied to optimization methods. The group has integrated this research with the solver technology to provide state of the art design tools. Typically the solution and optimization algorithms change the shape of the designs automatically to minimize a figure of merit preselected by the designer. The codes eliminate costly design cycles, speeding development and hence minimizing costs. In addition, the codes ensure that new designs perform optimally through a wide variety of conditions.

Parallel Computing

With the advent of parallel computing, scientists have witnessed tremendous reductions in computation time over more classical supercomputing technologies. Historically, the ACL has been committed towards developing algorithms that can be applied in these parallel computing environments. The Department of Energy recently started the Acclerated Strategic Computing Initiative (ASCI) which is dedicated towards developing the science of parallel computing. As one of 5 universities participating in this project, Stanford's goal is to complete a fully unsteady Navier-Stokes simulation of a jet turbine engine. From its library of software, the ACL has already developed a turbine analysis code capable of analyzing multiple blade rows of a turbine or compressor. Work is continuing on merging this code with compressor simulations to complete a full engine analysis.

Multi-Disciplinary Optimization

Description not yet available.

High-order discretization unstructured methods
Spectral Difference method utilizes the concept of discontinuous and high-order representations to achieve high accuracy as well as conservation. It combines the salient features of structured and unstructured grid methods to achieve high computational efficiency and geometric flexibility.