GPU-Based CFD Analysis for Modeling Complex UAV Flight Scenarios

Period of Performance: 02/10/2009 - 10/02/2009

$99K

Phase 1 SBIR

Recipient Firm

EM Photonics, Inc.
51 East Main Street Array
Newark, DE 19711
Principal Investigator

Abstract

Recent advances in computing have produced special-purpose hardware comprised of vector processors streamlined for high-performance computations. Development of these devices, such as graphics processing units (GPUs) and the Cell Broadband Engine (Cell Processor), has been advanced by the videogame industry. In an effort to increase flexibility and enter new markets, vendors have increased platform usability and opened up underlying hardware constructs to general computing uses. In this project, we focus on leveraging the computational power of commodity graphics hardware to develop a high-performance computational fluid dynamics (CFD) tool capable of rapidly and accurately modeling aircraft interaction with naval vessels, or the dynamic interface (DI). This project is divided into two areas: implementation of a CFD formulation able to simulate complex DI scenarios on programmable graphics technology, and the development of an integration framework that will tune the system specifically for DI problems. Phase I consists of building a prototype flow solver in hardware, and also developing the plan for integration to be carried out during Phase II. This plan will describe an overset grid scheme to enable moving body simulation, coupled flight dynamics with control feedback for real-time flight simulations, and system scalability to increase problem size and computational throughput.