Integration of Pumped Hydro Energy Storage and Wind Energy Generation: Structural Analysis and Algorithms
Emre Nadar, Bilkent University
We study the energy generation and storage problem for a hybrid energy system that includes a wind farm and a pumped hydro energy storage (PHES) facility with two connected reservoirs. The operator decides in real-time how much water to pump or release in the PHES facility, how much energy to generate in the wind farm, and how much energy to buy or sell. We model this problem as a Markov decision process under uncertainty in streamflow rate, wind speed, and electricity price. We establish the optimality of a state-dependent threshold policy under positive prices: The state space of the problem can be partitioned into several disjoint domains, each associated with a different action type, such that it is optimal to bring the water level of the upper reservoir to a different state-dependent target level in each domain. Once the optimal amount of water that should be pumped or released is found, we can immediately derive the optimal amount of wind energy that should be generated. Leveraging our structural results, we construct and test three heuristic solution methods for data-calibrated instances in which the price can also be negative: policy-approximation, profit-approximation, and problem-approximation methods. While the policy-approximation method provides virtually optimal solutions about four times faster than the standard dynamic programming algorithm, the problem-approximation method yields instantaneous solutions with an average optimality gap of 3.35%. The existence of natural inflow in the upper reservoir – the major source of structural complexity – improves the profits by 19.9% on average.
Emre Nadar is an assistant professor in the Department of Industrial Engineering at Bilkent University. He holds a Ph.D. degree in Operations Management from Carnegie Mellon University and a B.S. degree in Industrial Engineering from Bilkent University. His research focuses on supply chain management and sustainable operations. He studies problems such as managing inventories in multi-item supply chains, controlling new-product diffusion in closed-loop supply chains, and managing renewable resources in energy supply chains. Dr. Nadar employs the theory of Markov decision processes to model and analyze such problems. His doctoral work on assemble-to-order production systems received the POMS College of Supply Chain Management Best Student Paper Award, and became a finalist in the INFORMS George Nicholson Student Paper Competition and the MSOM Society Student Paper Competition.
Friday, December 31, 2021, 4.00 pm - Zoom Meeting