The value of energy storage depends on how the firm operates the storage to capture price fluctuations (both seasonal and random) under physical constraints (energy storing and releasing rates, storage capacity). In many cases, these constraints bind operational flexibility, which makes finding an optimal policy through stochastic dynamic programming a difficult task. To overcome computational difficulty, heuristic policies based on static optimization are widely used in practice. We analyze the gap between the heuristic policies and the optimal policy, and develop a new optimization approach that can substantially close the optimality gap without exacerbating the computational burden significantly. We will report the numerical results with actual data from the natural gas market. We will also discuss the possibility of enhancing storage value by integrating granular spot market operations into forward market operations, where the firm can capture not only price differentials in the forward market but also the spot-forward price differentials.
Group for Research in Decision Analysis