PHYSICS-AWARE AND MARKET-RESPONSIVE OPTIMIZATION FOR HYDROPOWER STORAGE: A CONVERGENCE-BASED DECISION SUPPORT FRAMEWORK

Abstract

The increasing penetration of variable renewable energy into modern power grids has fundamentallyaltered the operational context of hydropower, necessitating a shift from conventional load-peaking strategies toward adaptive, market-responsive dispatch. This thesis develops and evaluates the foundational components of a physics-aware, market-responsive optimization framework for hydropower storage, demonstrated through application to DeGray Dam a 74 MW United States Corps of Engineers (USACE)-operated facility within the Midcontinent Independent System Operator (MISO) South region and marketed by Southwestern Power Administration (SWPA). The framework integrates two independent analytical components coupled through a composite multi-objective function balancing economic reward maximization against physical risk penalization. The market component employs a walk-forward validated statistical model forecasting daily maximum Locational Marginal Prices (LMP) using publicly available MISO Market Operating Margin (MOM) reports and Energy Information Administration (EIA) commodity price data across a 978-day dataset spanning January 2023 through September 2025. Ridge Regression was selected as the production model after nine-month holdout validation, achieving an aggregate RMSE of 0.284 on the log scale and demonstrating superior robustness in high-volatility regimes. The physical component employs an Hydrologic Engineer Center-Hydrologic Modeling Software (HEC-HMS) simulation calibrated to the Caddo River watershed, translating dispatch schedules into daily reservoir storage trajectories and quantifying risk penalties for energy-in-storage violations. Two operational scenarios were evaluated: a conservation baseline and a market-dispatch strategy triggering four-hour full-turbine releases when the predicted LMP exceeded the 75th percentile threshold ($65.45/MWh). The market-dispatch scenario generated $1,415,531 in additional revenue on 62 dispatch days at a capture rate of 54.8%, while incurring a $455,506 risk penalty for energy storage of 144 days below the storage threshold of 50%. Under a base-case weighting of λ1 = 0.60, market dispatch dominated the conservation baseline by a composite margin of $667,116. Sensitivity analysis identified a crossover at λ∗1 ≈ 0.25, indicating the dispatch strategy is preferred across a wide range of operator risk tolerances. The results establish a quantitative foundation for convergence-based decision support in federal hydropower operations, with direct applicability to other grid operator facilities and broader relevance to hydropower systems operating in organized energy markets.