A Comparative Study of Model Predictive Control Approaches for Surge Mitigation in Centrifugal Compressors

Surge is a phenomenon that occurs in centrifugal compressors characterized by a sudden and severe drop in pressure and flow rate. To avoid the phenomena, an anti-surge valve (ASV) is installed in parallel with the compressor. The valve opening is manipulated by a control system to partially recycle the flow to the compressor inlet. Current industrial solutions for anti-surge control are based on a proportional-integral controller (PI). This approach presents several challenges, like parameter tuning, lack of constraints and difficulty in handling nonlinear dynamics. Therefore, the traditional solution is a suboptimal control that can potentially lead to surge events or unneeded ASV opening. One effective approach is the use of model predictive control (MPC), which offers the advantage of considering future system states and constraints, enabling proactive control actions to prevent surge. This study presents a comparative analysis of different MPC approaches while considering practical surge scenarios. The MPC formulations are implemented and tested using a dynamic model of a single stage gas compressor. The performance analysis involves inducing disturbances through rapid valve closures on the suction and discharge sides of the compressor. A comparison on how to deal with parameter retuning on real plant is performed. The results offer guidance for selecting the most suitable approach for centrifugal gas compressors in practical applications.