DEVELOPER TOOLS

The Computational Materials Institute (CMI), a unit of the Cornell Theory Center (CTC), has deployed the world's fastest Windows-based high-performance computing cluster dedicated to basic research in computational materials science. The cluster has 17 Dell PowerEdge 1855 enclosures, each holding 10 blades. Each blade has two 3.6 GHz Xeon EM64T processors, 4 GB of RAM, and 292 GB disk. The nodes are connected via Giagabit Ethernet and a Force10 E12000 switch. The cluster is running Windows Server 2003 (SP1), Enterprise x64 Edition. Grants from the National Science Foundation and the Air force Office of Scientific Research provided the funding needed to secure this leading edge computing power. "With this new capacity, we will be able to sustain our leadership in collaborations and link scientists at some of the top research organizations in this country," said Anthony Ingraffea, CTC associate director, CMI coordinator, and Dwight C. Baum Professor of Engineering at Cornell. "The computational power available from parallel, cluster computing is one of the fundamental components of advancing research and achieving scientific breakthroughs." The new equipment will support innovative research into metallic and composite material and structure damage simulation to help predict more accurate life expectancies for air and spacecraft. With the additional computing and memory support of the new cluster, researchers will be able to produce models that illustrate unprecedented geometric detail and physics-based behavior. "Organizations have had to be overly conservative when deciding to ground aircraft, because existing computational resources have lacked adequate processing and memory performance," said Ingraffea. "Results from our modeling will provide more complete and reliable data that can be used to improve the quality and efficiency of commercial and government air and space programs." The cluster will be used in two existing project areas: a DARPA-funded research effort designed to improve the operational readiness of America's military, and NASA and Air Force-funded investigations into damage tolerance of advanced spacecraft thermal protection systems using novel metallic, composite, and functionally graded structural systems. The cluster will support these research initiatives because it can handle simulations of realistic structures under extreme loading conditions. In particular, the cluster will accommodate three-dimensional structural-scale models exposed to extreme thermo-mechanical loadings involving high temperatures and high stresses, and sub-millimeter scale polycrystalline metallic models with thousands of grains with representation of constituent particles within the grains. “The additional computational capabilities represented by the new cluster will be essential to achieving the goals of the Structural Integrity Prognosis System [SIPS], namely the prediction of structural viability of military aircraft,” said Dr. John Papazian, the Northrop Grumman Principal Investigator for DARPA’s SIPS. “The new cluster will permit visualization of the detailed microscopic process that control metal fatigue, significantly enhancing military readiness.” Rapid and responsive access to space has been identified as an essential future strategic capability on which the U.S. Air Force will be heavily reliant. Risk-quantified, durable, hot structures hold the key to realizing the kind of strategic mission capability needed to protect national interests. The new computing cluster at CMI provides a necessary ingredient for attaining the fundamental understanding needed to design and field the necessary game-changing aerospace structural solutions. “Physics-based structural simulations are the only way to ensure that the vehicles we build can indeed, repeatedly and reliably, survive the heat loads accompanying re-entry and long-duration, high-speed, atmospheric flight conditions, and simultaneously handle the aerodynamic loads associated with such missions,” said Dr. Ravi Chona, Director of the Structural Sciences Center of the Air Force Research Laboratory. “The challenge has always been doing both of these things using one integrated, well-understood, optimized, structural concept.” “Material scientists are enjoying a revolution in their ability to measure properties and to create test specimens at the micron-scale level in these materials, using such techniques at orientation imaging microscopy and ion beam machining,” said Ingraffea. “One can now see the progression of damage at this scale over time. With these observations, we are now poised to formulate new theories to explain this previously unobservable behavior. This new parallel machine will serve two crucial roles in this science: first, it will allow us to determine the stress and strain conditions associated with the observed damage evolution with acceptable accuracy. Such determinations are essential to the formulation of physics-based explanations of observed behavior. Second, it will test the proof of our understanding and ability to control damage mechanisms when we encode our new theories and ask them to predict behavior under different conditions through simulation.”