EVALUATION OF THE SUITABILITY OF EMPIRICALLY BASED MODELS FOR THE PREDICTION OF ENERGY PERFORMANCE OF LIQUID CHILLERS

This paper presents the results of comparing the suitability of six reported chiller performance models to be used for energy analysis. The data adopted to validate the models comprise of two broad classes including constant and variable chilled water flow rates. There are 373 sets of data in total. The optimal regression coefficient for each empirically based model is obtained by using the least square method. Criterions for evaluating the model suitability are based on the coefficient of variation of root-mean-square error (CVRMSE). The results show that among the six empirically based performance models for liquid chillers chosen in the study, the multivariate polynomial regression model possesses the best prediction accuracy. Its average CVRMSE value is around 0.66%. The results of the study could be served as a reference for the selection of empirically based models for liquid chillers when applied in the energy analysis of buildings.