UC Berkeley

This dissertation studies renewable energy systems, addressing the pressing challenges posed by their inherent uncertainty and the need for adapted control and forecasting strategies. With the increasing reliance on renewable sources to mitigate climate change impacts, under- standing and optimizing these systems becomes crucial. This work presents new methodologies in four areas, each contributing to the efficient management and integration of renewable energy resources in the evolving energy market landscape.The first chapter introduces a novel approach to the siting, sizing, and bid scheduling of a price-maker battery in a nodal wholesale market. This framework, developed through mixed-integer optimization, demonstrates how strategic positioning and sizing of battery storage can influence prices and alleviate congestion, turning a profit in markets traditionally challenged by high capital costs. By applying this model to New Zealand, the study offers practical insights into maximizing battery storage profitability in nodal markets. In the second chapter, the focus shifts to the modeling and state estimation of Lithium- Sulfur (Li-S) batteries using a Piecewise Affine (PWA) system. Addressing the complexity of differential algebraic equations in standard battery models, this chapter presents a more efficient approach for real-time state estimation, crucial for applications demanding high energy density. The integration of a PWA model within a moving horizon estimation frame- work significantly improves state of charge estimates. This chemistry is particularly useful for sectors like electrified aviation and heavy-duty transport. The third chapter explores the dynamic field of Airborne Wind Energy Systems (AWEs) control using Information-Directed Sampling (IDS). This control strategy adeptly balances the exploration-exploitation trade-off in a partially observable environment, enhancing the efficiency of AWEs in harnessing wind energy at varying altitudes. By implementing an IDS controller and validating it with real-world data, this research offers a new design of AWEs control, contributing to the more effective utilization of airborne wind energy. Finally, the fourth chapter introduces ’MapeMaker’, an open-source software package creating probabilistic scenarios for renewable power production. This tool is capable of generating scenarios that reflect both historical forecast accuracy and potential future advancements in forecasting technologies. By providing a means to create realistic alternative scenarios based on historical data, ’MapeMaker’ is a useful tool in renewable energy planning and operations, particularly valuable for simulation-based analysis methods such as stochastic unit commitment and economic dispatch. In summary, this dissertation presents a comprehensive collection of methodologies and tools that contribute to the optimization and integration of renewable energy resources in modern electricity markets. Through uncertainty-aware control strategies, forecasting techniques, and the development of practical tools, this work addresses the critical challenges of managing uncertainty in renewable energy systems.