Nilanjan Ray Chaudhuri knows anyone who has experienced an extended power outage understands the effects can go well beyond inconvenient and become outright dangerous.
An assistant professor in electrical engineering at Penn State University and principal investigator on the project, Ray Chaudhuri is working to prevent failures in the power grid and enable a quick recovery when they do occur through his project, “Coupled cAscade Modeling, Prevention, and Recovery (CAMPR): When Graph Theory meets Trajectory Sensitivity.”
“One interesting thing about the power grid is that it is the largest machine man has built because of the geospatial spread,” said Ray Chaudhuri.
Another noteworthy aspect of this system is that it has a physical structure — the wires on pylons — coupled with a cyber infrastructure — the sensors that monitor the power system with the ability to control actions, such as increasing or decreasing generation or shedding loads.
Because of this coupled nature of the cyber and physical elements, when a failure occurs in the physical system, the sensors, which have limited battery backup and must be powered by the grid as well, will fail. When the sensors fail, it is impossible to receive accurate information about the physical system, which in turn leads to the possibility of making poor decisions on how to correct the blackout.
The cyclical nature of this problem is exacerbated by cascading failures, the specific focus of Ray Chaudhuri’s study. Cascading failures in the physical system have a domino effect, with a failure at one point in the system leading to the failure in an adjacent point, and so on.
The nature of the dual element cyber-physical system also leads to another challenge.
“What has happened is that people have started looking into this problem of cascading failure from a purely networking standpoint. But the way failures propagate in reality is something that isn’t reflected in most of the research because it needs domain knowledge in the power area. It’s a different mechanism of cascade propagation than what has been captured in the existing literature. So it’s very much a nascent area of research even now,” Ray Chaudhuri said.
The project has three general thrusts: To understand the mechanisms of this propagation and develop better models for analysis; to establish actions that can be taken to prevent cascading failures; and to restore the power grid as quickly as possible in scenarios where preventing a failure is not possible or successful.
Establishing a better model for statistical analysis, which is part of the first goal of the project, would be an exciting breakthrough, as the existing models are either quasi-steady-state (QSS) models, which can produce inaccurate results in the later stages of blackouts, or fully dynamic models, which is impractical for large-scale statistical analysis.
The new models Ray Chaudhuri and his team are developing are temporally hybrid and spatio-temporally hybrid models which will result in unprecedented accuracy that will enable the realization of the next two goals in the project, namely, the prevention of blackouts and the recovery of the power grid when they do occur.
“I feel really excited that we came up with some neat ideas, and I’m really looking to see if we can make some breakthroughs here,” he said.
While this research could greatly reduce cascading failures with the power grid, the implications of this work extend into other fields, as well.
“Any cyber-physical infrastructure with a reasonably complex system will have similar issues to what we have here,” said Ray Chaudhuri, listing oil and gas as examples of industries that will benefit from this research.
Ray Chaudhuri is collaborating on this research with co-PIs Tom La Porta, the William E. Leonhard Endowed Chair, Evan Pugh Professor and Director of the School of Electrical Engineering and Computer Science, and with Ting He, associate professor of computer science and engineering.
“This project is based on interdisciplinary collaboration between a power systems expert (Dr. Ray Chaudhuri) and two communication network experts (Dr. La Porta and myself),” said He. “As the next-generation power system will be increasingly coupled with its control system for automatic monitoring and control, and the control system, which is a communication network by itself, will also rely on the power system for power supply and piggybacked control signaling (using Power Line Carrier Communication), the joint study of the two systems is the key to unlocking important insights into the cascading behavior of failures. This makes our interdisciplinary team uniquely qualified to perform this research.”