Cornell Theory Center (CTC) announced today, Oct. 26, that the Power Systems Engineering Research Center (PSERC), a consortium of universities led by Cornell University, has received $10 million for research on minimizing failures in complex networked systems, such as electrical, communications and distribution systems.

The award is from the Electric Power Research Institute (EPRI), a power industry research group, and the Army Research Office under the two organizations' Complex Interactive Network/Systems Initiative.

Other members of the consortium include the University of California at Berkeley, the University of Illinois at Urbana-Champaign, George Washington University, Washington State University and the University of Wisconsin at Madison.

According to Gail McCarthy, director of strategic science and technology at EPRI, the award springs from concerns by EPRI and the military about their reliance on very complex networks that could be vulnerable to disruption. As one example, the electric power transmission grid, which currently interconnects electric utilities through more than 672,177 miles of lines, is becoming more complex because restructuring of the industry has dramatically increased the number and complexity of electric power transactions and transmissions.

In addition, the increased reliance on communication and computing networks overlaid on the electric grid is heightening security and reliability problems.

"We are looking for revolutionary, interdisciplinary research that leads to the prevention of widespread interactive, cascading network failures," says McCarthy. "This is a key objective for both organizations."

Robert J. Thomas, Cornell professor of electrical engineering, director of PSERC and principal investigator on the project, notes that the participating institutions have been collaborating through PSERC. Collectively, they bring broad expertise to bear on power systems issues, including dynamic modeling and analysis, protection, nonlinear system theory and control, hybrid systems, failure prediction and large-scale system theory.

"We believe that the issue of faults and failure within large distributed networked systems presents interesting and important challenges to 21st-century mathematics," says Thomas.

Mathematical theories will be tested in an application to be created by the team using computer simulation of a model of the future electric power system. This model will contain not only the grid, but also the communication and computing system overlays needed for distributed control and information flow, as well as for economic data.

"When it comes to complex networks, we know that some degree of failure is inevitable. So the approach we are taking is a fail-safe/safe-fail one," says Thomas. The researchers, he says, will develop methods designed to reduce the frequency of failures (fail-safe) and to limit their effects (safe-fail) in overlapping networks.

The problem of providing reliable power is one that pushes the limits of computing and communication technology. Hundreds of utility companies are interconnected in the electric transmission grid. A typical control center monitors 10,000 or more locations, with updates coming in every 2 seconds on about 4,000. Their computing and communications systems receive data from sensors and controls on more than 4,000 conditions, such as an overload on one part of the grid, which are considered in calculating contingencies for dealing with an outage. The future electric power grid will have to contend with even greater complexity and will have to be able to reconfigure network flows in response to economic and other pressures.

One of the greatest concerns to both the power industry and the military is the cascading failure of a complex, integrated network system. A recent example of a cascading failure occurred in 1996 on the West Coast, when relays in the grid failed, propagating and amplifying a local outage into a cascading fault that led to a region-wide blackout.

Consortium researchers will use the Cornell Theory Center's high-performance computing and visualization resources. These include one of the largest academic IBM RS/6000 POWERparallel Systems and a three-wall virtual reality environment. "Modeling complex networks is a perfect example of a challenging problem requiring CTC resources and interdisciplinary and inter-institutional expertise to have practical results of benefit to society," says Thomas F. Coleman, CTC director.

PSERC is a National Science Foundation (NSF) Industry-University Cooperative Research Center, intended to maintain and increase the ability of the United States to develop and commercialize new technology in the global marketplace; it also is funded by industrial partners. CTC receives funding from Cornell, New York state, National Center for Research Resources at the National Institutes of Health, NSF, Department of Defense and corporate partners.