INSPIRE: Optimization Algorithms for Regional Thermoelectric Power Generation with Nonlinear Interference

Research Foundation Of The City University Of New York (Lehman), Bronx NY

Investigators

Abstract

This INSPIRE project is jointly funded by the Algorithmic Foundations program in the Computing and Communications Foundations Division in the Directorate for Computer and Information Science and Engineering, the Environmental Sustainability program in the Chemical, Bioengineering, Environmental, and Transport Systems Division in the Directorate for Engineering, and the Office of Integrative Activities (OIA) INSPIRE program. A thermoelectric power plant's operations can affect its surrounding environment, for example by raising water temperatures, which can be harmful to aquatic life, and so must comply with government regulation such as the Clean Water Act (CWA). It has recently been observed that the effects of power plants' operations can be much longer-reaching and subtler than this, however: the use of this warmer water by a second power plant located downstream can degrade that plant's efficiency, causing it in turn to heat the river water more than it otherwise would have, or even forcing it to shut down in order to comply with the CWA. Such complex dynamics characterizing the joint effects of a region's power plants suggest possible gains from managing plants jointly rather than individually. This analytical perspective prompts consideration of a huge variety of potential benefits to seek and costs to avoid in optimizing regional plant operations, and the interference phenomenon prompts (re)consideration of a number of classical algorithmic problems in the field of combinatorial optimization. This research offers many potential societal benefits in terms of environmental protection, economic savings, energy security, protection from blackouts, public health, and so on. Insights provided by the algorithmic solutions this project develops will be conveyed to decision makers and, if successful, will ultimately lead to improvements in management practices in existing plants and in long-range strategic planning. The project will provide research training for graduate students and will expose undergraduates at Lehman College and CCNY (both Minority-Serving Institutions) to interdisciplinary scientific research. This project will inaugurate the study of a novel class of combinatorial optimization problems. More specifically, it will investigate new variations on classical problems such as knapsack and job scheduling, modified to incorporate a distinctive feature of the motivating application setting, i.e. the *nonlinear interference* that can occur between active power plants. The PI and his team will design efficient (near-)optimal algorithms for these problems in the sense of guaranteed approximation, and, leveraging existing analytical models, they will perform algorithmic engineering studies assessing their algorithms' real-world viability. Finally, using tools from algorithmic game theory, they will quantify and provide a rigorous foundation for the perceived benefits of solving the motivating plant management problems jointly rather than plant-by-plant.