Coordinated Operation of Centrifugal Compressors in Series via Model Predictive Control
This paper investigates the control challenges in the operation of centrifugal compressors in series configuration, which arise from the interactions of the different control loops for process and anti-surge control. The investigation is carried out with simulations, where two compressors in series are considered to represent the simplest possible case for systems analysis. The simulations are based on a well-known nonlinear compressor model to represent the plant behavior as close to reality as possible. The paper starts with the derivation of linear transfer functions from step response analysis, which are used to develop both advanced decentralized control solutions using loop decoupling schemes as well as a model predictive controller (MPC). The various alternative control strategies are then investigated by introducing a typical downstream disturbance case that pushes the compressors towards surge conditions while creating interactions among all control loops. The simulation results are analyzed to determine the advantages and the shortcomings of the different control strategies. The MPC based approach stands out as a high performance alternative to traditional control schemes relying on simple feedback and feedforward elements.