Applying Genetic Algorithm to Optimal Loading for Hybrid Chiller Systems

• Main Authors: Jian-He Huang, 黃建和
• Other Authors: Yung-Chung Chang
• Format: Others
• Language: zh-TW
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/39d3cm

碩士 === 國立臺北科技大學 === 能源與冷凍空調工程系碩士班 === 104 === With current innovation and development of chillers, there are various available chiller units on the market. Now the air conditioning systems are mostly designed as mixing chiller with large-capacity unit collocating small-capacity unit, while large-capacity unit usually are centrifugal and small-capacity unit are screw. The operation mode of this combination is mainly a subjective decision of site operators. During the peak period or with crowds they will launch large-capacity unit, otherwise launch small-capacity unit. Judged by operator is not the best operation method, because after a period of operating, due to efficiency decreasing of heat exchanger, piping position difference, shortage of refrigerant, uneven water pumps supply, inconsistent booting order and operation time and others, various chiller units have differences in performance. This study takes a large public museum as subject, applying Screw Chiller and VSD Screw Chiller, as well mixing centrifugal chiller with human subjective operations. To realize the characteristics of running chiller unit, it must be monitored and data parameters be recorded in long term, and then built a power consumption model of chiller using regression analysis. However, different types of chillers are with different limitations. Screw Chiller needs to be segmented to distinguish load factors, while paying attention to the lowest limit of Hertz of its inverter of VSD Screw Chiller; the surge effect of Centrifugal Chiller under the low loading is needed to be noted. Finally, the study uses genetic algorithm to calculate the load optimization of chiller unit which meets the spatial loads, then figure out the operation mode of chiller unit with minimum power consumption and improving efficiency of subject judgments. Results show that, Genetic Algorithm is a better way than subjective judgments and operation. When the spatial load is during 95% to 55%, the efficiency of energy saving is 4.6% to 24.4% respectively, and average efficiency of energy saving is 14.6%.