A towering inferno in microcosm, lasting for fractions of a second, the swirling mix of air and atomised fuel as it explodes in the cylinder (turbulence helps the mixing process between the fuel and oxidiser) would seem to defy analysis. Nevertheless the mix is governed by the same fluid dynamics equations that quantify smoke swirling from, say, a chimney, where large swirls of smoke spin off chains of smaller ones. Analysis of very large numbers of such complex equations requires massive computing power, and a team led by mechanical engineers Joseph Oefelein and Jacqueline Chen of Sandia National Laboratories (Sandia) is using the Jaguar supercomputer at Oak Ridge Leadership Computing Facility (ORLCF) for just this purpose.
LES reveals how fuel from a state-of-the-art injector mixes with air inside an engine cylinder
(Image credit: Joseph Oefelein and Daniel Strong, Sandia National Laboratories)
Supercomputing Helps Engine Manufacturers
R&D professionals seeking to optimise cycle efficiency as they design, for instance, a new marine diesel engine, use desktop computers to simulate actual events in the cylinder during the combustion stage, but manufacturers do not have the time or resources to develop predictive models that have been validated against benchmark experiments. This need is being met by the Scandia researchers.
The Jaguar, a Cray XT5 supercomputer at the Oak Ridge, New Mexico, ORLCF facility has power enough to take on the task, with a ’Gee-whizz’ peak speed of 2.33 petaflops, which means it can solve over two thousand trillion calculations per second. The unmatched speed, memory and bandwidth of this supercomputer enables researchers to solve the equations that simulate turbulent combustion.
Jaguar Supercomputer
(Image courtesy of the Oak Ridge Leadership Computing Facility (OLCF), Oak Ridge National Laboratory)
Analysis of Turbulent Combustion – Two Approaches
The two Sandia researchers employ entirely different but complementary approaches: Chen investigates how flames stabilise, extinguish and re-ignite, employing a direct numerical simulation code (DNS) to simulate the finest micro-scales of turbulent combustion on a three-dimensional grid. Oefelein though, uses a large eddy simulation (LES) code to capture large-scale mixing and combustion processes within the geographic boundaries of an engine cylinder; computing the effects on combustion, for example, on varying the lift of an exhaust valve.
Whereas DNS depicts the fine-grained detail, LES starts at large scales and works its way down. The researchers claim that in combination LES and DNS running on petascale computers like Jaguar can provide a nearly complete picture of combustion processes in engines.
A comprehensive article, dated 12, September 2011, ’Simulating Turbulent Combustion Speeds Design of Power and Propulsion Devices’ is available on the ORLCF website for readers interested in the detail of this research.