Improved energy management of chiller systems by multivariate and data envelopment analyses

The operation of chiller systems accounts for the major proportion of electricity consumption in commercial buildings. This paper considers using multivariate and data envelopment analyses to facilitate the energy management of chiller systems. The system studied contains five sets of chillers, pumps and cooling waters and it operates for an institutional building. Based on a huge set of operating data, multiple linear regression was used to correlate the system coefficient of performance (COP) with a set of climatic and operating variables. Data envelopment analysis was then employed to calculate the scale, technical and overall efficiencies. These three efficiencies were further examined to ascertain which controllable variables caused a decrease of system COP. The results show that the existing energy management gives a technical efficiency of 0.76 and fine-tuning the controllable variables could achieve an electricity saving of 5.34% in relation to the existing operation. The significance of this study is to demonstrate a systematic approach to examine which operating variable should be fine-tuned to improve system performance with higher technical efficiency.