Meeting corporate renewable power targets

Abstract

Prominent companies have committed to procuring a percentage of their power demand from renewable sources by a future date. Long-term financial contracts with renewable generators, known as corporate power purchase agreements (CPPAs), are popular to meet such a renewable power purchase target (RPPT). By analyzing a simplified three-stage model, we show that the generation capacity contracted via a CPPA is more nuanced to structure optimally compared to traditional long-term power contracts due to the interplay between price and supply uncertainties as well as the RPPT. We subsequently propose a Markov decision process (MDP) to formalize rolling-power- purchase policies used in practice, that is, the construction of dynamic CPPA portfolios to meet an RPPT. The optimal MDP policy is intractable to compute but possesses the following key properties: (i) its decisions account for stochastic prices and supply, (ii) it captures the timing flexibility to enter CPPAs, and (iii) it can sign CPPAs with different tenures. We develop forecast-based reoptimization heuristics and a novel information-relaxation based reoptimization approach that sacrifice and approximate, respectively, the first property of the MDP policy and capture the remaining properties. We perform an extensive computational study on realistic procurement instances to uncover managerial insights related to procurement costs, the control of risks arising from supply uncertainty, the relevance of CPPAs as markets evolve, and the near-optimality of rolling power purchases from our information-relaxation based procurement heuristic.