Valuation of compound real options for co-investment in residential battery systems. (15th July 2022)

• Record Type: Journal Article
• Authors: Ma, Yiju; Chapman, Archie C.; Verbič, Gregor
• Is Part Of: Applied energy. Volume 318(2022)
• Publication Date: 2022-07-15
• Subjects: Distributed energy resources -- Battery storage -- Rooftop PV -- Home energy management -- Policy function approximation -- Real options valuation; Power (Mechanics) -- Periodicals; Energy conservation -- Periodicals; Energy conversion -- Periodicals; 621.042
• DOI: 10.1016/j.apenergy.2022.119111
• Languages: English
• ISSNs: 0306-2619

Abstract

Abstract: The process of evaluating investments by electricity utilities to subsidize residential battery installations in distribution networks is challenging, because finding efficient investment plans involves reasoning over various sources of uncertainty and consideration of different sizing, location and timing options. This work proposes a novel financial framework using real options valuation (ROV) to solve the optimal investment problem for subsidizing residential battery installations, considering compound options including the option to delay the investment and its subsequent option (expand). We incorporate random variations in PV generation and load behaviour, making use of Monte Carlo (MC) power flow analysis that incorporates battery scheduling profiles. Specifically, the ROV framework solves stochastic power flow analysis with battery scheduling in a forward-looking MC model, and determines the optimal plan for executing the options that maximize the investment value using backward induction over all MC paths. However, this approach to MC becomes computationally prohibitive when battery operational profiles are generated by optimization-based home energy management. To reduce the computational burden, we use policy function approximation to provide fast battery operational profiles, drawing on machine learning methods that reduce MC computation time by 95%. Compared to standard financial analysis, the proposed model allows ROV to capture all uncertainties while allowing utilities to make contingent decisions to maximize the benefits from subsidizing residential battery storage. Highlights: A real options valuation framework is proposed to evaluate battery investment. A Markov decision process is used to model the optimal battery investment problem. Computationally efficient Monte Carlo analysis that incorporates battery schedules. Policy function approximation is trained to emulate battery scheduling. Making contingent decisions on executing options increases investment value.