Two-stage stochastic programming approach for gas allocation network under uncertainty.

Abstract

Pipelines enable enormous amounts of various products (fluids) to be transported from supply nodes to demand nodes. They have traditionally been recognized as the most efficient and secure method of transferring gases. Uncertainties in the parameters may develop in the actual world for a variety of reasons which reduces the efficiency of the gas allocation network (GAN). The amount of available gas cannot be forecasted exactly due to the uncertainty in gas supply and shared usage by other demands. Design of GAN is carried out in a two stage manner: installation and operation. During operation, there is always a chance of change in number of sources or demands than that of initial design phase. In this paper, to deal with such discrete uncertainties, a two-stage stochastic programming approach for GAN is developed. The first stage represents the installation and commissioning of supplying nodes in order to satisfy the demands, and the second stage represents the actual allocation network under different available supply stations scenarios. Illustrative examples are presented to demonstrate the proposed solution procedure and annualized investment for the examples are calculated. The calculated annualized investment is 12%, 28% and 38% less than the worst case solutions. Result explains the benefits of the model in reducing investment costs while incorporating such discrete uncertainties.