home  |  about us  |  contact  

This page is part of the CSMD web archive and is not maintained.
Please visit csm.ornl.gov/newsite for the latest CSMD information.


Home > PC >

Originally appeared in March 10, 2006

ORNL Researchers Will Tap 2.5 Million Supercomputer Hours

More than 2 million hours of processing time on some of the world's most powerful computers could help Oak Ridge National Laboratory researchers make significant strides in astrophysics, fusion and materials science.

Two of the three ORNL projects funded through the Department of Energy's recently announced Innovative and Novel Computational Impact on Theory and Experiment (INCITE) program involve collaborations with universities, industry and other national labs. The third belongs solely to ORNL and is led by Phani Nukala of the Computer Science and Mathematics Division.

With 1.5 million processor hours, Nukala and colleague Srdjan Simunovic hope to gain insight into how materials fracture, which despite decades of study remains a fundamental problem of science and engineering. Nukala and Simunovic will perform large-scale three-dimensional simulations of lattice networks in order to understand the size effect on material strength and the scaling laws of fracture.

"Our goal is to develop methods to design high-strength fracture-resistant materials," Nukala said. "The insight that we gain is crucial to synthesizing tough ceramics and other materials that have fracture toughness properties comparable to those of natural materials such as sea shells, which exhibit phenomenal fracture strength and toughness properties despite the brittle nature of their constituents."

Research on synthesizing such tough materials has taken on increased importance because these materials can be used in various defense-related applications, including for lightweight body armor. In addition, the research will help to "develop a prognostic methodology to forecast impending material systems failures," Nukala said.

The project was awarded 1.5 million processor hours on the IBM Blue Gene at Argonne National Laboratory.

In another project, David Schultz and Predrag Krstic of ORNL's Physics Division were part of a team that won 650,000 processor hours to examine in unprecedented detail the interactions of individual atoms, molecules and photons. This information could ultimately help solve one of the fundamental challenges of designing fusion reactors.

"The knowledge we gain through these new simulations will allow advances in our understanding of larger systems of interest to DOE science missions encountered in plasmas, gases and solids," Schultz said. "For example, the ability to model and ultimately improve the operation of magnetic-confinement fusion reactors requires a multi-scale approach in which plasma phenomena are simulated on relatively long time scales while much finer grain interactions must be tracked on times more on the order of femtoseconds - one-millionth of a nanosecond."

This project, led by Auburn University, received computing time at DOE's Pacific Northwest National Laboratory.

In the third ORNL project, Mark Fahey of ORNL's National Center for Computational Sciences is working with a team led by General Atomics to facilitate calculations describing the physics of fusion reactors.

"These computer simulations will allow scientists to understand new features of the complex behavior of the hot swirling fuel, or plasma, inside the donut-shaped reactor vessel, or tokamak," Fahey said.

Previous computer simulations have retained the physics of the heavier particles (ions) but ignored the complex turbulent motion of lighter particles (electrons) that may sometimes cause the plasma to cool too quickly. By understanding this process, scientists can more carefully plan the design and operation of future fusion devices.

The project was awarded 400,000 processor hours on the Cray X1E at ORNL's National Center for Computational Sciences.

Recently, DOE's Office of Science announced the awarding of 18.2 million hours of processor hours, or supercomputing time, to 15 teams. The allocations were made under the INCITE program, now in its third year of providing resources to computationally intensive research projects in the national interest.


Copyright 1993-2006, HPCwire. All Rights Reserved.
Mirrored with permission.