A comparison of automation techniques for optimization of compressor scheduling

Compressor selection is one of the primary functions in operation of natural gas pipelines, and a major concern of the task is to minimize operating costs. This study presents a comparison of three automation techniques for compressor selection: mixed integer linear programming, genetic algorithms, and expert systems. In compressor selection, dispatchers often turn on/off compressor units based on the status of the pipeline and the anticipated customer demand. Since a novice dispatcher often performs this task on a trial-and-error basis without any guarantee of optimal operations, it is desirable to develop a decision support system that can select compressors based on the available data. This study presents a comparison of three automation techniques for incorporation into a decision support system. Based on parameter values for one section of the gas pipeline at the St. Louis East area in Saskatchewan, Canada, a comparison of the strengths and weaknesses of the three automation techniques as well as the recommendations they gave are discussed.