A Computational Approach to the Real Option Management of Network Contracts for Natural Gas Pipeline Transport Capacity

Commodity merchants use real option models to manage their operations. A central element of such a model is its underlying operating policy. We focus on network contracts for the transport capacity of natural gas pipelines, specific energy conversion assets. Practitioners commonly manage these contracts as portfolios of spread options. Although computationally fast, we show that this approach in general is heuristic because of the suboptimality of its operating policy. We propose a different computationally efficient real option approach based on an optimal operating policy, integrating linear optimization and Monte Carlo simulation. We use our approach to benchmark the spread option approach in a numerical study based on market data and realistic instances. We find that our approach can substantially improve on the contract valuations computed by the spread option approach, especially for contracts with flexibility in their allowed capacity usage. We also show that the optimal operating policy of contracts with this flexibility is of the greedy type, whereas the one of contracts without it in general is not. However, we observe that greedy optimization yields a near-optimal operating policy for the latter type of contracts with a substantially reduced computational effort. A version of our model was recently implemented by a major international energy trading company. Our model can also be employed to efficiently estimate the contract value sensitivities (the "Greeks"), which merchants use for financial hedging purposes. Potentially, our work has wider significance for the merchant management of other commodity conversion assets with payoffs determined by solving capacity constrained optimization models.