Abstract
We formulate the merchant trading of energy in a network of storage and transport assets as a Markov decision process (MDP) with uncertain energy prices. Because of the intractability of this MDP, we develop heuristics and both lower and dual (upper) bounds on the optimal policy value estimated within Monte Carlo simulation. We achieve tractability by applying linear optimization in novel ways for approximate dynamic programming: (i) Iterative extensions of least squares Monte Carlo techniques based on value and continuation function approximations (V and CFAs) that are separable and non-separable, respectively, and piecewise linear concave in the inventory levels of the storage assets; (ii) a generalization of a commercially available deterministic reoptimization heuristic that implicitly uses an analogous CFA; and (iii) a perfect information dual bound based on a separable and linear VFA. Using realistic natural gas instances, we establish that (i) non-separable CFAs yield superior operating policies compared to separable VFAs, and near optimality can be obtained by reactively accounting for price uncertainty, which has both computational and modeling advantages; and (ii) separable and linear VFAs provide near tight dual bounds. The applicability of our methodological developments transcends our specific application.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have