Minimization of Power Usage in a Compressor Station with Multiple Compressors

In natural-gas transmission, a compressor station often consists of multiple types of compressors. For any given suction pressure, discharge pressure, and total mass flow, an operation manager needs to decide the set of compressors to work, the rotation speed of each working compressor, and the natural-gas flow allocation among all the working compressors. In this paper, two solution approaches are investigated and compared: mixed-integer linear programming and dynamic programming. Both approaches require the discretization of actual volumetric flow. Numerical results reveal that when the discrete interval is 0.1m3/s, dynamic programming takes less computation time than mixed-integer linear programming does. However, this order is reversed when the discrete interval is 0.01 or 0.001m3/s.