A simulation model of chiller system containing four sets of chillers is built in Energyplus 7.0 simulation software. Based on the simulation data of chiller system, a data-mining tool - hierarchical cluster method is applied in chiller system to analyze the operation performance of chiller system. Using a built-in analytical tool in SASS v17.0, hierarchical cluster method is made on the five main operating variables of chiller system over a year. There are three clusters produced by hierarchical cluster method, which represent three typical operating conditions of chiller system of a whole year. Through the analysis of the characteristics of clusters of main operating variables, the operating conditions of each chiller can be illustrated, which matches with the control strategy implemented in the chiller system. By the analysis of COPs of each chiller in different clusters, the potential of energy saying for the chiller system under different clusters can be illustrated. Besides, it is found that the Refrigeration Operation Energy Efficiency Ratio (ROEER) of one chiller over one year can be represented by the COPs of one chiller in different cluster centers, which demonstrates that the cluster results are reasonable. (C) 2016 Elsevier B.V. All rights reserved.