Modeling and Optimization of Industrial Centrifugal Compressor Stations Employing Data-Driven Methods

Abstract

This paper addresses optimal operation of centrifugal compressors operating in a parallel configuration. A group of compressors operates in parallel to increase the supply of a gas. One current practice is to distribute the load equally without considering the fact that the individual compressors have different characteristics and they are in a different health condition due to past hours of operation and non-uniform maintenance plan. Data from past operation is used for generating data-driven models of the compressors. These models and operational constraints of the compressor station are used in an optimization model. The optimization model computes the distribution of both the load and cooling water of each compressor which reduces the operational cost, i.e. power consumption of the motors and purchase of cooling water. The suggested optimization model is applied on a real process case study. The results from optimal operation from optimization show a reduction in the power consumption of the compressor station compared to the actual power consumed in past operation. The magnitude of this benefit ranges between 0.67 up to 2.16 %.