Investment in Network Expansion Under Uncertainty.

A description of two optimization models for augmenting the capacity of a network over which a homogeneous commodity must be shipped. The models determine the investment that minimizes the sum of the investment cost plus the cost, on the average, of meeting uncertain requirements. These requirements include both shipping costs and penalty costs for unsatisfied demand over the augmented network. Investment decisions must be made under a budget constraint and when requirements on a subset of the arcs are known only in the form of a probability distribution. The first model, a two-stage linear program under uncertainty, is for a two-period horizon: the present, when investments must be made and demand is unknown; and the future, after which demand is observed and shipments take place. The second model is for the multiple-period horizon. It is also shown that these models can be extended to the multicommodity case. 33 pp. Ref.