以粒子群最佳化方法分析冰水主機群之節能運轉策略：以南部某科技廠為例

Strategic Analysis of a Chiller Plant Operation for Reducing Energy Consumption Using Particle Swarm Optimization: A Case Study of a High-Tech Factory in South Taiwan

10.6306/JITE.202407_(17).0010

曾禧濃(Hsi-Nung Tseng)；張志彰(Chih-Chang Chang)

科技廠房 ； 空調系統 ； 性能曲線 ； 負載率 ； 冰水主機負載最佳化 ； high-technology building ； air-conditioning system ； chiller performance curves ； cooling load ； Chiller load optimization

工業科技教育學刊

17期（2024 / 07 / 01）

198 - 214

繁體中文;英文

研究背景：近年來，台灣高科技廠房快速發展，使得工業用電急遽增加，同時也帶來了能源消耗和碳排放問題。為因應全球暖化，「2050淨零排放」目標已是全球共識，各企業紛紛積極推動永續治理，以減少碳排放。依照我國經濟部能源局民國111年能源消費百分比概況，全年工業部門類別電力佔總能源能耗56.1%，以某科技廠能源流向表為例，可發現工廠用電3－4成來自空調系統，企業若能有效提高空調使用效率，將有助減少碳排放，邁向淨零減碳目標。研究目的：本研究將依據廠房空調負荷量及冰水主機性能曲線的結果針對南部某科技廠冰水主機群的運轉進行綜合分析，並利用粒子群演算法分析該廠不同性能冰水主機群能耗最小化條件下的冰水主機運轉組合及負載率，求得冰水主機負載的最佳化，使總耗電量達到最小化。期望透過粒子群演算法的分析找到冰機最佳操作點，並在各種性能條件下實現冰水主機群的節能運轉提供較佳的冰水主機運轉策略，以減少該廠房的空調能耗。研究方法：該廠房內冰水主機群包含螺旋式冰水主機400RT×1台，離心式冰水主機800RT×2台，共計2,000RT，冰水出水溫度設定為7.5℃，供應廠房生產空間使用。本研究採用超音波流量計、溫度計及電力分析儀，針對2023年10月每小時的冰水流量、冰水入出水溫度、冷卻水入出水溫度及主機耗電功率進行量測，以得出廠房實際冷卻負荷的變化。進一步地，依據每台冰水主機在不同負載率下的性能曲線及廠房實際冷卻負荷，應用粒子群最佳化方法進行冰水主機群節能運轉策略分析。研究結論：廠房2023年10月份空調冷卻負荷量測結果之平均值為1032RT，晚間與日間的冷卻負荷差異並不大，冷卻負荷的變化主要受到生產時間、生產線規劃及氣候等因素影響。根據每小時的廠房冷卻負荷量測數據，以及每台冰水主機之部份運轉性能曲線，透過粒子群最佳化方法針對5種冰水主機群運轉組合的最小能耗進行分析，並計算出冰水主機群需求達到最小能耗條件的各冰水主機負載率變化情形，最後結果顯示10月份冰水主機群的最佳運轉組合為：在冷卻負荷低於1600RT時（10月份冷卻負荷最大值未超過1600RT），800RT定頻與變頻離心冰水主機為皆開啟狀態，400RT定頻螺旋冰水主機無需啟動。上述運轉組合在最小能耗的運轉條件下，月總耗電量為366,879 kWh，相較於平均負載運轉，節省13,480 kWh，節電率達到3.58%。

Background: In recent years, Taiwan's high-tech factories have witnessed a significant increase in industrial electricity usage, raising concerns about energy consumption and carbon emissions. With the global consensus on "2050 Net Zero Emissions," enterprises are actively pursuing sustainable practices to reduce carbon emissions. According to Taiwan's Ministry of Economic Affairs, the industrial sector accounted for 56.1% of total energy consumption, with electricity as the primary source. For example, in one tech factory, 30-40% of electricity usage is attributed to air conditioning. This will help reduce carbon emissions and propel enterprises towards their net zero carbon emission goals. Purpose: This study conducts a comprehensive analysis of chiller system operation in a technology plant in southern Taiwan based on air conditioning load and chiller performance curves. It uses particle swarm optimization algorithm to minimize energy consumption by optimizing chiller operation combinations and load ratios, aiming to reduce total electricity consumption. The goal is to find optimal operating points for chillers and provide better operation strategies for energy-saving chiller operation, ultimately reducing the facility's air conditioning energy consumption. Methods: The chiller system in the plant includes one unit of 400 RT screw type chiller, two units of 800 RT centrifugal type water chiller, leaving evaporator water temperature setpoint 7.5 °C. This study used an ultrasonic flowmeter, electronic thermometer, power quality analyzer, we conducted a research and analysis measurement of the hourly chilled water flowrate, leaving and entering chilled water temperature, leaving and entering cooling water temperature and energy consumption of the main equipment between July and December of 2023, to calculate colling load change of plant, based on the performance curves of each chiller under various load rates and the actual cooling load of the facility, the particle swarm optimization algorithm was applied to identify the minimum energy consumption points for different combinations of chillers. This facilitated an analysis of the operational strategy for the chiller system. Conclusions: The average air conditioning cooling load measurement in the factory for October was 1032 RT, with minimal differences observed between nighttime and daytime cooling loads. The variation in cooling load is primarily influenced by production time, production line layout, and climate. Utilizing hourly measurements of the factory's cooling load and partial performance curves of each chiller unit, we conducted an analysis using Particle Swarm Optimization to determine the minimum energy consumption among 5 combinations of chiller unit groups. We calculated the variations in load ratios required for each chiller unit within the chiller unit group to achieve the minimum energy consumption condition. The results indicate that the optimal operational combination for the chiller unit group in October is as follows: when the cooling load is below 1600 RT (with the maximum cooling load for October not exceeding 1600 RT), both the 800 RT fixed-speed and variable-speed centrifugal chiller units are operational, while the 400 RT fixed-speed screw chiller unit remains inactive. Under these conditions of minimum energy consumption, the monthly total electricity consumption is 366,879 kWh, resulting in a savings of 13,480 kWh compared to average load operation, with an energy saving rate of 3.58%.