NSF Award
2433820
The project aims to find new algorithmic methods for the resilient integration of renewable energy into the electrical grid as well as for making transportation more sustainable. For example, in the traditional energy grid, when renewables produce a surplus of energy, such surplus generally does not affect the operation of traditional power plants. Instead, renewables are throttled down, or the surplus is simply ignored. However, in the future, when most of the power is generated by renewables, this will not be tenable. Rather, traditional power plant output needs to be throttled down or switched off in response to less predictable renewable supplies. On the demand side, power usage has also become less predictable with the switch from gas or electric-resistive heating to heat pumps, independent solar panels, the adoption of electric vehicles, and more diverse working hour patterns. With transportation, the transition to electric and autonomous vehicles, as well as multi-modal transportation, is underway and fueled by a smart grid built around renewable energy. Rather than using statistical methods, this project pursues a game-theoretic approach. In the online setting, one imagines the input to be created by an omniscient adversary who knows the code of the online algorithm and strives to defeat the algorithm. An online algorithm with good competitiveness gives a performance guarantee relative to the best that could be done if one knew the future. Thus, it makes good decisions even in situations where the input derives from unusual or unexpected circumstances, such as supply chain disruptions due to disasters. The project will also create more sustainable solutions through algorithms related to batching and server problems in datacenters. The project will also train students and broaden participation in computing as it is hosted at the University of Nevada, Las Vegas, a minority-serving institution ranked as one of the most diverse universities for undergraduates. <br/><br/>Power plant cycling can be avoided by the obvious method of not cycling a unit, and that may include staying on at a loss; this tradeoff is modeled by abstracting the problem as a power-down problem in the online competitive setting. The project seeks to design novel online competitive algorithms for the power-down problem. A "decrease and reset" scheme is considered where competitiveness is relaxed by a small amount to allow for online competitive algorithms that are close to optimal yet with better performance in many types of request sequences. The project also includes consideration of new potential-guided methods, as well as the study of continuous power-down problems. The use of online models opens a new approach to benefit future transportation systems. Models for car-sharing systems, for example, are inherently online. A problem complementary to both power-down as well as transportation is the server problem. The project includes the study of several open questions around online server problems. Competitive algorithms for the delayed server problem for modeling car sharing, battery consolidation systems and traffic signal control are also sought. The investigator heads the University of Nevada Las Vegas Center for Information Technology and Algorithms; the award contributes to further establishing and nurturing the important work being done at this center.<br/><br/>This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
