Bayraksan collaborating on $2.1M DOE sustainable energy grant
Ohio State ISE Professor and Associate Chair for Research Güzin Bayraksan is teaming up with researchers at Southern Methodist University (SMU) on a project to contribute leading-edge research in computational mathematics for sustainability, which will include the operation and management of the electric grid.
The Department of Energy (DOE) has awarded $2.1 million in grant funding to “New Abstractions and Randomized Algorithms for Multiscale Stochastic Optimization.” It’s all part of an $8.5 million allocation in funding by the DOE seeking “High-Performance Algorithms Research for Complex Energy Systems and Processes.”
“Integrating large amounts of renewable energy generation poses significant challenges for power system operations,” Bayraksan says. “The reliable delivery of electric power involves multiple layers of operations arranged in a hierarchy.
“For instance, the day-ahead unit commitment problem is solved one day in advance. However, within it, the real-world problem includes short-term unit commitment and hour-ahead economic dispatch decisions. While the day-ahead problem sees a coarse representation of uncertainty, such as the weather forecast for the next day, the hour-ahead problem sees a more refined representation of uncertainty – renewable energy generation in the next hour with significantly refined weather forecasts.”
Bayraksan says she and the lead collaborator, SMU Assistant Professor of Operations Research and Engineering Management Harsha Gangammanavar, both share a mutual interest in the research of randomized algorithms, which help account for unanticipated challenges.
“Such randomized algorithms are needed when decision making problems, which often involve uncertainty and risk, become so large that it is impossible to solve them through traditional algorithms,” she says. “Randomization allows such complex problems (which are beyond our current capabilities) to be solved, but approximately. Statistical methods are then used to validate the results. Another important advantage of randomized algorithms is that they work well in data-driven settings where the data comes in sequentially.”
Bayraksan says the project will entail investigating randomized decomposition-and-coordination-based solution methods and their statistical validation/verification and will provide support in scenario generation for graph-based abstractions for these problems.
In addition to Bayraksan and Gangammanavar, researchers from University of Southern California, University of Wisconsin-Madison, Penn State and Argonne National Laboratory will be participating in the project.
Story by Nancy Richison
