A physics-guided self-adaptive chiller sequencing controller of enhanced robustness and energy efficiency accommodating measurement uncertainties

Abstract

For multi-chiller systems commonly applied in commercial buildings, a reliable chiller sequencing control strategy makes a crucial contribution to ensure robust and energy-efficient operation. However, the commonly used chiller sequencing control strategy often deviates from expectations significantly due to common sensor measurement uncertainties encountered in practice. To address this problem, this study proposes a physics-guided chiller sequencing control strategy that improves the system's robustness and energy efficiency by adaptively adjusting chiller switching thresholds to accommodate sensor measurement uncertainties. First, a physics-guided fault detection and diagnosis (FDD) supervisor is developed to diagnose the fault types associated with each chiller-ON event under the corresponding switching thresholds. Subsequently, based on the identified fault type, a self-adaptive switching threshold supervisor is developed to adaptively adjust the chiller switching thresholds (i.e., key parameters for determining the chiller stages) for mitigating the adverse impacts resulting from the sensor measurement uncertainties. The test results show that the proposed control strategy can significantly enhance the robustness under negative measurement uncertainties and save the total system energy consumption by up to 7.46 % without sacrificing robustness under positive measurement uncertainties.