Massive supercomputing grant to be used in water management research

November 25, 2008

Patrick Reed, associate professor of civil engineering at Penn State, has been awarded a large resource allocation (LRAC) grant on the largest open science supercomputer in the world.

"It's a new scale of computing with 62,000 processors," Reed explained. "My award of 1.79 million computing hours represents the most selective level of resource allocations."

Reed's project, "TeraGrid-Enabled Hydrologic Systems Monitoring, Prediction and Management Under Uncertainty," will make use of the Texas Advanced Computing Center's Ranger Supercomputer and 65 terabytes of storage space at the San Diego Supercomputing Center.

Though the use of supercomputing in research isn't new, what sets Reed's project apart is the sheer size of computing that's being devoted to the project.

"It's a new level of computing that not many folks have explored," he said. "We are going to be able explore the observation, prediction and management of water in whole new ways. This grant will be using new models that physically represent what's out in the world and what's evolving. We will focus on evaluating those models and then combining them with new, multi-objective decision support tools to identify where new observations could improve our predictions and use the improved predictions in managing our water resources."

To better illustrate his work, Reed says to imagine a tree. One end of the tree consists of branches, which symbolize all of the ways we simulate future water systems. At the other end are the roots, which symbolize our various design objectives, such as costs, risks, reliability and so on. The supercomputer serves as the tree's trunk, supporting the linkage of our design objectives to our predictions of the future.

Reed's project will give researchers new ways to explore tradeoffs for water systems costs, risks, vulnerabilities and uncertainties using data from huge numbers of simulations.

"It's a new way to do design when you have lots of interacting variables and lots of objectives," the civil engineer said.

He explained that most decisions aren't made with just one variable in mind. For example, even though cost is a consideration in any project, an administrator also may need to factor uncertainties or risks into his calculation. Reed said that his work adds those other objectives into the calculations to better understand the key challenges and benefits for projects.

"The supercomputer is the connective piece. This facilitates our ability to connect advances in our ability to predict with advances in our ability to manage."

Reed serves as the principal investigator on the project, which lasts through September. His co-principal investigator is Reed Maxwell, assistant professor of geology and geologic engineering at the Colorado School of Mines.


(Media Contacts)

Last Updated March 19, 2009