Thermal performance lines of chiller plants of two institutional buildings in Hong Kong

Abstract

Chiller plant electricity consumption of two institutional buildings have been audited and to identify potential problem areas and establish a basis for assessing improvement measures, thermal performance lines for the two plants were derived using multiple linear regression methods. Stepping regression methods and variance inflation factor analysis were used to select the significant weather parameters for use as the independent variables in the performance line models. This yielded models that cannot be used for inter-building comparison because different combinations of independent variables and coefficients are involved in the models for the two plants. Another model equation with a single independent variable, the product of daily mean outdoor temperature and humidity ratio, was found applicable to both plants. Adequacy of the models was verified by examining the residuals and validity of the models verified by data splitting method. Comparing the regression statistics with observations made in the energy audits shows that a thermal performance line may be used as a diagnostics tool on the quality of control and operation of a chiller plant.