학술논문
데이터센터의 발열량에 따른 경제적인 냉방부하에 관한 연구 / A Study on the Economic Cooling Load according to Amount of Heat Generating in the Data Center
Document Type
Dissertation/ Thesis
Author
Source
Subject
Language
Korean
Abstract
데이터센터는 주로 인터넷, 클라우드, 소셜네트워크(SNS) 서비스 등을 통해 주고받는 각종 메시지와 자료, 사진, 동영상 등을 보관 및 처리하기 위해 24시간 연속적으로 가동되고 있으며 이로인해 사용하는 전력량도 급증하여 일명 ‘전기먹는 하마’ 라고 칭하고 있다. 특히 2014년을 기준으로 국내에 대략 113개의 데이터센터가 존재하고 있으며, 이들 센터에서 사용하는 전력량은 연간 26억 kWH 정도 이르는 것으로 추정되고 있는데, 이러한 전력 사용량은 원자력 발전소 1기의 연간 발전량의 1/3에 해당되는 양으로 데이터 센터의 전력 사용량 절감에 대한 요구가 증대되고 있다. 이로부터 국내외 정보기술(IT) 기업들의 에너지 다이어트 경쟁이 점차적으로 치열해 지는 경향이고, 풍력과 태양광 등을 활용한 신재생에너지 사용 비율의 증대와 더불어 외기 도입을 통해 실내를 냉방시키는 외기 도입 시스템, 폐열을 회수하여 활용하는 폐열회수 시스템, 겨울철 찬공기를 이용하여 냉각수를 냉수로 활용하는 프리쿨링 시스템 등 각종 에너지 절약형 냉각기술 개발 및 적용에 매진하고 있는 실정이다. 따라서 본 연구에서는 데이터센터에서의 발열량을 파악하고 경제적인 냉방부하 운전을 유도하기 위해 서울시 금천구에서 가동중인 데이터센터 1곳을 대상으로 실내 발열량을 측정하고 이를 통해 적정 냉방부하 운전을 산정하였다. 또한 데이터센터의 특성상 건물 규모, 장소가 다양하고 서버의 발열부하 역시 다양하여 서로 상이한 조건에서 발생하는 측정치의 오차를 감소시키기 위하여 대표 모델로서 데이터센터 내의 전산실로 한정하여 선정하였다. 측정한 자료에 의하면, 상기 데이터센터에서의 설정온도 및 설정풍량에 대한 실내온도 분포 및 압축기 적산시간은 전산망 기술기준에 관한 규칙 제13조에서 권장하는 온도범위 기준을 대체로 만족하였지만 압축기 적산시간에 있어서 다소 상이한 결과를 초래하기도 하였다. 아울러, 서버의 배치 및 수량에 따라 실내온도 분포가 약간의 차이가 초래되었으며, 각 개소마다 항온항습 설정치를 달리하여 운영한다면 보다 효율적으로 전산실 실내 발열부하를 제거할 수 있는 것으로 파악되었다. 결과적으로, 현재 운영중인 데이터센터 내에서 보다 효율적으로 에너지 절약을 실현하기 위해서는 각 데이터센터의 현재 서버 가동상태 및 발열량을 검토해 보고 최적의 상태를 유지시킬 수 있는 냉방부하 용량을 산정하여 이를 적용하면, 향후 보다 에너지 절약적으로 가동이 가능할 것으로 사려된다.
Data Centers are operated 24 hours a day in order to store and process various messages, data, pictures and videos that are exchanged mainly through Internet, cloud system and social network services (SNS). Therefore, their power consumption has skyrocketed in recent years. They are even regarded as ‘The Biggest Energy Hog’. In particular, there are about 113 Data Centers in Korea as of 2014. The electric energy consumed by these Centers is estimated to reach 2.6 billion kWH per year, which is equivalent to one third of the annual power output of a single unit of a nuclear power plant. Hence, there is a growing demand for reducing the power consumption of these Data Centers. In this regard, many domestic and overseas Information Technology companies are competitively reducing their energy consumption. Not only they attempt to use more renewable energy sources such as wind power and solar power but they also focus on developing and applying a variety of energy-saving cooling technologies, including outdoor air utilization system for cooling indoors through outdoor air, waste heat recovery system for recovering and utilizing waste heat and pre-cooling system for using coolant as cold water through cold air in winter. In this study, I therefore measured the indoor calorific value at the Data Center located in Geumcheon-gu of Seoul and calculated the appropriate cooling load operation in order to identify the calorific value of data centers and help them achieve economical cooling load operation. Moreover, I limited the scope of this study to the data processing room of a Data Center in order to reduce the error of measured values that might be caused in different conditions because the Data Centers greatly vary in size, location and server heat load. According to the measured data, the indoor temperature distribution and compressor estimation time as to the set temperature and air volume at the above-mentioned Data Center generally satisfied the temperature range recommended by Article 13 of the Regulations on Computer Network Technology Standards. However, the result on the compressor estimation time in this study was slightly different. In addition, there was also a slight discrepancy in the indoor temperature distribution depending on the arrangement and quantity of servers. I identified that it would be possible to remove the heating load in a computer room in a more efficient manner by varying the set point of constant temperature and humidity in each center. In conclusion, it will be possible to achieve energy savings at the data centers by reviewing the current server operation status and calorific value of each data center and calculating and applying a cooling load capacity that could maintain the optimal condition.
Data Centers are operated 24 hours a day in order to store and process various messages, data, pictures and videos that are exchanged mainly through Internet, cloud system and social network services (SNS). Therefore, their power consumption has skyrocketed in recent years. They are even regarded as ‘The Biggest Energy Hog’. In particular, there are about 113 Data Centers in Korea as of 2014. The electric energy consumed by these Centers is estimated to reach 2.6 billion kWH per year, which is equivalent to one third of the annual power output of a single unit of a nuclear power plant. Hence, there is a growing demand for reducing the power consumption of these Data Centers. In this regard, many domestic and overseas Information Technology companies are competitively reducing their energy consumption. Not only they attempt to use more renewable energy sources such as wind power and solar power but they also focus on developing and applying a variety of energy-saving cooling technologies, including outdoor air utilization system for cooling indoors through outdoor air, waste heat recovery system for recovering and utilizing waste heat and pre-cooling system for using coolant as cold water through cold air in winter. In this study, I therefore measured the indoor calorific value at the Data Center located in Geumcheon-gu of Seoul and calculated the appropriate cooling load operation in order to identify the calorific value of data centers and help them achieve economical cooling load operation. Moreover, I limited the scope of this study to the data processing room of a Data Center in order to reduce the error of measured values that might be caused in different conditions because the Data Centers greatly vary in size, location and server heat load. According to the measured data, the indoor temperature distribution and compressor estimation time as to the set temperature and air volume at the above-mentioned Data Center generally satisfied the temperature range recommended by Article 13 of the Regulations on Computer Network Technology Standards. However, the result on the compressor estimation time in this study was slightly different. In addition, there was also a slight discrepancy in the indoor temperature distribution depending on the arrangement and quantity of servers. I identified that it would be possible to remove the heating load in a computer room in a more efficient manner by varying the set point of constant temperature and humidity in each center. In conclusion, it will be possible to achieve energy savings at the data centers by reviewing the current server operation status and calorific value of each data center and calculating and applying a cooling load capacity that could maintain the optimal condition.