Evaluating seasonal chiller performance using operational data

Chillers are the largest consumers of electricity in modern cooling systems, typically accounting for 40 % of the total energy consumption in commercial and industrial building cooling systems. An accurate online evaluation of chiller performance is crucial for various applications, such as chiller sequencing and predictive maintenance. Due to the absence of a performance metric that adapts to specific system characteristics and an evaluation process that incorporates predictive uncertainty, this study introduces a new metric, the Adaptive Chiller Performance Value (ACPV), and proposes a novel four-step evaluation procedure. ACPV assigns weights based on the proportion of typical operating conditions, identified from the chiller operation data, ensuring that the evaluated performance aligns closely with the system's specific features. The proposed four-step process considers the selection of confidence levels for prediction intervals and the potential impacts of predictive uncertainty during the evaluation. The reliability of the proposed method was validated through a case study conducted on a multi-chiller cooling plant of an operational data center. The complete evaluation process was implemented, and the results indicated that the proposed evaluation framework provides both intuitive and statistically significant insights. Moreover, it can be easily adapted to diverse cooling systems, demonstrating its significance in developing chiller sequencing strategies and facilitating predictive maintenance, with the ultimate goals of energy savings and carbon emission reduction.