Session: 17-01: Poster Presentations

Paper Number: 168354

168354 - Decomposing a Large-Scale Linear Program for Power Sector Capacity Planning 

Abstract: 

The Regional Energy Deployment System (ReEDS) is an open-source power sector capacity expansion model developed at the National Renewable Energy Laboratory (NREL). ReEDS is used for analysis at NREL and by the wider public, including for producing NREL’s Standard Scenarios reports, analyzing impacts of policy choices and legislation, and determining optimal deployment of generation and storage technologies over long planning horizons. Currently, some configurations of the ReEDS model, a large-scale linear program, can take hours to solve. To address this, we seek to exploit the mathematical structure of the model to identify ways to decrease solve time. For example, spatial and temporal divisions in the model imply that there may be sets of loosely coupled constraints, an opportunity to use Dantzig-Wolfe decomposition. We first present a full mathematical formulation for a specific configuration of the ReEDS model, providing a foundation upon which new features can be added to the model. We then describe strategies for performance improvements, including decomposition, scaling, restructuring of the model constraints and variables, and algorithmic tuning. We show runtime improvement through implementation of these strategies.

Presenting Author: Alex Derenchuk Colorado School of Mines

Presenting Author Biography: Alex Derenchuk is a PhD student in the Advanced Energy Systems program at Colorado School of Mines, working with Dr. Alexandra Newman. He received an M.Eng. in Operations Research and Analytics in 2024, a B.E. in Environmental Engineering in 2020, and an A.B. in Earth Sciences in 2019, all from Dartmouth College. He also has 4 years of work experience in energy economics consulting, specializing in price forecasting, asset valuation, grid decarbonization, and market design for resource adequacy.