Decentralized model predictive control of a multi-evaporator air conditioning system

• A control architecture is proposed for coordinating the temperature control of multiple zones cooled by a multi-evaporator vapor compression system.
• The resulting architecture is decentralized and modular, despite the coupled nature of the system dynamics.
• The control architecture relies upon model predictive control's ability to respect system constraints in order to maintain safe, efficient operation.
• Evaporators or compressors can easily be added to the supervisory controller without having to develop new controllers for existing components.

Temperature control of multiple zones with a multi-evaporator vapor compression system is a common problem in modern air conditioning. Due to the coupled system dynamics, standard decoupled controllers can interfere with each unit′s performance. This paper proposes an architecture that is decentralized and modular, avoiding competing controllers and the practical difficulty of implementing a centralized controller. A model predictive control (MPC) supervisor calculates evaporator cooling and pressure setpoints for each zone, balancing temperature regulation with energy efficiency; these setpoints are tracked by local level controllers, which rely upon MPC's ability to respect constraints in order to maintain safe, efficient operation.