Towards High-Fidelity Simulation and Analysis of Hypersonic Air-Breathing Weapon Systems using Large-Scale GPU Computing Systems
Kessler, David (Naval Research Laboratory)
There is an immediate Navy and DoD need for applied research in the area of robust and reliable propulsion systems for hypersonic weapon platforms to enable time-critical, extended-range precision strikes and maintain the Navy's global reach and dominance of force. While traditional rocket engines represent one potential path toward meeting this need, air-breathing ramjet and scramjet (supersonic combustion ramjet) engines have the potential to enable longer-range, higher-payload missions or faster time-to-target than a comparable rocket-propelled vehicle.
In an air-breathing engine, mixing between the fuel and oxidizer takes place in a high-speed turbulent flow environment for which the flow residence time in the combustion chamber is too short for chemical reactions to proceed to completion without some method of flame stabilization. A principal technical roadblock is the lack of a fundamental understanding of the coupled physics that control this stabilization process. High-fidelity simulations have the potential to provide a wealth of information regarding the underlying combustion stability and performance characteristics of concept engines.
Performing such simulations requires resolving an extremely large range of length and time scales, and full-scale simulations including detailed chemical kinetics of tactical hydrocarbon fuels are still beyond the scope of the fastest computing systems available today. The influx of GPU devices into the high-performance computing space (and the increase in computational throughput relative to traditional CPU architectures that they provide), however, has shifted our concept of the scope and scale of simulations that we can reasonably expect to perform. While this paradigm shift comes with a price in software design and programming overhead, simulation tools that have made this investment are poised to demonstrate leap forward capabilities.
In this presentation, we will give an overview of one such simulation tool, NRL's JENRE® Multiphysics Framework, and describe our efforts to-date aimed at simulating high-speed air-breathing combustors. We will focus in on the use of this tool on current and emerging gpu architectures and describe how we will leverage an ongoing HPCMP-sponsored Dedicated High Performance Computing Investment (DHPI) program to demonstrate the applicability of high-fidelity, scale-resolving simulations to flight-relevant systems.