Roger Z. Ríos-Mercado
Graduate Program in Systems Engineering
Universidad Autónoma de Nuevo León
Abstract: In the gas industry there are three important stages which are production, transportation and sales. In particular, in the transportation phase, the objective is to transport gas from some production or storage centers to different distribution centers at the least cost possible. To achieve this, compressor stations are placed at some points of the network to keep the gas moving. The function representing the incurred cost of fuel at each compressor unit (centrifugal type), which is installed in the network, is typically nonlinear, nonconvex, and difficult to evaluate computationally. Optimization algorithms for this problem have to evaluate this function many times. Due to this, a better approach may be using approximation functions, which are easier to evaluate than the real function. In this paper, we perform a computational evaluation with several approximation functions over a set of collected data from nine compressors units. The results confirm that one these proposed functions does a very good job at approximating the real function.