Safe and efficient operation of centrifugal compressors using linearized MPC

Centrifugal compressors are widely used in many industrial applications such as natural gas extraction, refrigeration and gas transport. Since gas compression is both an energy and capital intensive process, it is very important to operate gas compressors in a safe and efficient manner in order to save resources and utilize their maximum capacity. This typically means to operate gas compression systems at working points that are close to surge, an unstable condition that can cause severe damage to the compressor and the associated piping system. Anti-surge control systems aim at preventing the compressor from going into surge due to planned operating condition changes and also due to unexpected external disturbances which can affect the system within a time window of hundreds of milliseconds. Anti-surge control is achieved by recycling or venting-off some of the compressed gas in order to increase the mass flow through the compressor and reduce the pressure ratio. Recently, also the use of electrical drives powering the compressor for enhancing surge avoidance has been proposed. Generally, anti-surge control can be stated as a nonlinear multivariable control problem with output constraints that also present very fast dynamics. In this paper, the potential benefits of an anti-surge controller based on linearized model predictive control (MPC) are discussed. The employed MPC algorithm has been implemented on an industrial controller to meet the real-time requirements and has been tested on a experimental compressor test rig. The control performance of a linearized MPC-based anti-surge controller is compared to that of a traditional anti-surge control approach with controlled discharge pressure and controlled distance to surge in the case of external disturbances.