Computationally efficient solution of a compressor load sharing problem using the alternating direction method of multipliers

摘要

This article presents an efficient implementation of a compressor load sharing optimization problem. In a typical gas pipeline application several compressor units are interconnected and operated together to meet station set point requirements. Load sharing optimization solves the problem of distributing the load among several compressors while at the same time satisfying operational constraints, user constraints and minimizing the consumption of the station. The structure of the problem is special and can therefore be exploited by formulating and solving the problem using the alternating direction method of multipliers (ADMM). The potential of this approach is assessed through simulations using field data from a large compression station with variable-speed gas turbine driven gas compressors. Despite being based on code generation, the ADMM implementation is about ten times faster than a state-of-the-art active set method while at the same time having a much lower memory footprint.