Felipe dos Santos Couto, SEAS '22, Mechanical Engineering, Columbia University
Supervising Faculty, Sponsor, and Location of Research
Dr. Bolun Xu, Undergraduate Research Involvement Program, The Earth Institute, Columbia University
Abstract
Renewable energy is no longer an interesting possibility for the future but rather an urgent demand in the present to fight the climate crisis. Nevertheless, academia, industry, and public entities still face many challenges to increase penetration of renewable sources on the world energy mix. In this context, storing energy has risen as a key alternative and numerous storage solutions have been developed. The purpose of this study is to better understand how large-scale energy storage systems (ESS) impact electricity prices. We developed interpretable machine learning models and analyzed how supply and demand attributes correlate with price fluctuations – especially, price spikes, which are a major sign of market inefficiency. Optimal Regression Trees and Multiple Linear Regressions were applied to the Southwest Power Pool market data, shedding some light upon the attributes’ sensitivities. We then utilized the sensitivities to estimate price reductions due to energy storage. Results showed a potential reduction, on average, of 19.0% on price spikes. Further studies shall continue to investigate the effects grid-scale ESS on prices. By doing so, we hope to corroborate with the expansion of renewable energy generation and ESS on the grid.
Keywords
renewable energy, energy storage, interpretable machine learning, electricity price model