An Improved Basket of Spread Options Heuristic for Merchant Energy Storage

Abstract

Practitioners use the basket of spread options (BSO) heuristic to model merchant energy storage as a portfolio of spread options and forward sales. This method solves a linear program to obtain the composition of this portfolio and its associated BSO policy. Sequential reoptimization of this model yields the rolling BSO (RBSO) policy. It is known that this policy can struggle when storage is fast but otherwise it performs well, often being near optimal. Further, the objective function of the BSO linear program is a lower bound on the value of the BSO policy associated with a feasible portfolio. To attempt to obtain an improved RBSO policy, especially for fast storage, this paper proposes a BSO heuristic that replaces this function with a weighted average of itself and a simple upper bound on the value of a given BSO policy. On a set of known natural gas storage instances, limited optimization of the value of the weight that defines this average yields improved RBSO policies. The observed improvement is substantial and modest, respectively, when the original RBSO policies perform poorly, which occurs on some fast storage instances, and on average. Moreover, setting the averaging weight equal to 0.6 leads to RBSO policies that virtually match the performance of the best considered RBSO policies. This BSO heuristic is thus as easy to use in practice as the original BSO heuristic.