Energy management of chiller systems by data envelopment analysis

摘要

Purpose - The operation of chiller systems could account for considerable electricity consumption in air-conditioned buildings in subtropical regions. The purpose of this paper is to consider using data envelopment analysis (DEA) to facilitate management of their energy performance. Design/methodology/approach - A system serving an institutional building was studied, which contains five sets of chillers, pumps and cooling towers. The building has a total floor area of about 25,000m~2 and comprises classrooms, lecture theatres, offices and laboratories. The scale, technical and overall efficiencies defined in DEA were calculated based on the correlation between the output variable - system coefficient of performance (COP) - and the input variables - load factor and temperatures of chilled water and condenser water. The efficiencies were further examined to explain how outside air temperatures and controllable variables affect the system performance. Findings - The paper reveals that existing energy management gives a technical efficiency of 0.85 and fine-tuning the temperature-related variables could achieve an electricity saving of 14.8 per cent. Research limitations/implications - The improved COP predicted by DEA is related only to fine-tuning of the input variables concerned. An increase of COP by other advanced controls or system upgrades should be assessed based on robust system modelling techniques. Yet the extent of COP improvements helps investigate energy management opportunities requiring no or insignificant capital investment on existing systems. Practical implications - A systematic approach to performing energy management of a chiller system is proposed. The DEA helps examine which operating variable should be fine-tuned to achieve the highest possible performance. Originality/value - It is an under researched area to consider using scale and technical efficiencies in DEA to explain energy management of chiller systems and to estimate the highest achievable performance.