Airflow uniformity optimization for modular data center based on the constructal T-shaped underfloor air ducts. (5th June 2019)

• Record Type: Journal Article
• Authors: Zhang, Yiwen; Zhang, Kai; Liu, Jinxiang; Kosonen, Risto; Yuan, Xiaolei
• Is Part Of: Applied thermal engineering. Volume 155(2019)
• Page Start: 489
• Page End: 500
• Publication Date: 2019-06-05
• Subjects: Airflow uniformity -- Modular data center -- Constructal theory -- CFD; Heat engineering -- Periodicals; Heating -- Equipment and supplies -- Periodicals; Periodicals; 621.40205
• DOI: 10.1016/j.applthermaleng.2019.04.025
• Languages: English
• ISSNs: 1359-4311

Abstract

Highlights: Constructal T-shaped underfloor air ducts network (CTUADN) is proposed. Proposed CTUADN greatly improves uniformity of air distribution in data centers. Dimension of optimal air supply terminal structures is obtained. CTUADN integrated rack modules is the most effective way to improve air uniformity. Abstract: Maldistribution of underfloor airflow rate causes the nonuniform cooling power distribution of server racks in large-scale data center. In this regard, the constructal theory improves the uniformity and efficiency of airflow distribution by optimizing the geometric design of fluid distribution network. Thus, the design of a T-shaped underfloor air duct network based on the constructal theory was proposed for airflow uniformity optimization, and the concept of modular data center based on the constructal duct system was introduced in this study. The optimal air structure for the supply terminal was analyzed among 8 configurations by using CFD simulations. And then the optimal terminal was employed in an expanded modular data center with four rack modules and the constructal air ducts system. Finally, the rectifiers and ceiling vents were deployed in the model for improving the airflow uniformity in further. By designing the constructal T-shaped underfloor air duct network with the optimal structure and associated models, the airflow uniformity in the modular data center with 4 modules was improved, and the standard deviations of overall inlet and outlet air temperature of each rack were only 0.98 and 0.56.