The modern-day jet engine may be powerful enough to shuttle travellers across a continent in just six hours but it's also unbearably loud—for both the ground crews that work around them and residents within earshot of airports. And while aircraft engineers are developing quieter designs, building and testing these hushed prototypes can run into the six figures. But with the help of the US Livermore National Labs' supercomputer and some open-source modelling software, commercial airliners may soon be whisper quiet.

The 3,000 square-foot Sequoia IBM Bluegene/Q supercomputer at Lawrence Livermore (CA) National Laboratories is among the most powerful parallel computing systems on the planet. It sports over 1.5 million embedded processors 1.6 PB of memory and crunches numbers at a staggering 16.32 PFLOPS. The Sequoia's cores are arranged in a 5D Torus design wherein each core is directly connected to ten others. This greatly reduces latency even with cores two and three connections away. Read/Write functions are handled by these processors as well—some of which tap directly to the system's primary input/output channel through an 11th connection.

While all 1.5 million cores may be necessary to calculate the nuclear weapons simulations that it is normally charged with, Joseph Nichols' research team from Stanford Engineering's Centre for Turbulence Research harnessed just over a million of them for the jet engine research. Working in conjunction with teams from the NASA Glenn Research Centre in Ohio and the US Navy's NAVAIR to develop a quieter jet engine without actually having to build one.

"These runs represent at least an order-of-magnitude increase in computational power over the largest simulations performed at the Centre for Turbulence Research previously," said Nichols "The implications for predictive science are mind-boggling."

The technique is known as predictive modelling and it is an exacting process. The noise that a jet engine produces constitutes less than one percent of the device's total energy output, which means that accurately reproducing them in Computational fluid dynamics (CFD) simulations requires incredibly precise calculations.

"Computational fluid dynamics (CFD) simulations, like the one Nichols solved, are incredibly complex. Only recently, with the advent of massive supercomputers boasting hundreds of thousands of computing cores, have engineers been able to model jet engines and the noise they produce with accuracy and speed," said Parviz Moin, the Franklin M. and Caroline P. Johnson Professor in the School of Engineering and Director of the CTR.

Up next—predictive modelling of which far-flung airport your luggage will incorrectly arrive at.

[R&D - CTR - Wired - Stanford - Image: Lawrence Livermore National Laboratories]