Distributed Extremum Seeking Control for a Variable Refrigerant Flow System

The variable refrigerant flow (VRF) system features the regulation of refrigerant flow to achieve individualized heating and cooling for multi-zone operation. However, such flexibility in configuration presents large variations in system characteristics and thermal loads. Performance of model based control and optimization would thus be limited. The extremum seeking control (ESC) has recently been considered as a model-free control solution of maximizing the efficiency of multi-split VRF system. In this paper, a saddle-point dynamics based distributed ESC strategy is applied to optimize the indoor unit (IDU) superheat setpoint for minimizing the IDU fan power provided the satisfaction of thermal comfort. The ODU runs a stand-alone single-input ESC implemented with the suction pressure setpoint as control input and the ODU power output as the feedback. The proposed distributed ESC strategy is evaluated for the all-cooling operation mode of a VRF system that consists of one outdoor unit (ODU) and four IDU's. Simulation study is performed with a Modelica based dynamic simulation model of the VRF system using the distributed ESC and the centralized multi-input ESC in Dong et al. (2016). Simulation results show that the distributed ESC strategy achieves comparable performance as the centralized ESC in terms of steady-state and transient performance.