Impact of coolant distribution design on server-level thermal management in data centers

被引:0
作者
Khoshvaght-Aliabadi, M. [1 ,2 ,3 ]
Ghodrati, P. [1 ]
Shin, J. Y. [1 ]
Kang, Y. T. [1 ,3 ]
机构
[1] Korea Univ, Sch Mech Engn, 145 Anam ro, Seoul 02841, South Korea
[2] Islamic Azad Univ, Dept Chem Engn, Shahrood Branch, Shahrood, Iran
[3] Korea Univ, Res Ctr Plus Energy Bldg Innovat Technol, Sch Mech Engn, 145 Anam ro, Seoul 02841, South Korea
基金
新加坡国家研究基金会;
关键词
Data center; Thermal management; Server-level; CPU cooling; Cold plate; Optimization; ENERGY EFFICIENCY;
D O I
10.1016/j.energy.2025.136936
中图分类号
O414.1 [热力学];
学科分类号
摘要
A significant portion of data center servers (DCSs) energy is used for CPU cooling, necessitating efficient coolant distribution to optimize thermal management and improve energy efficiency. This study investigates and analyzes various coolant distribution designs for jetted cold plate cooling, with a particular focus on DCSs configured with two and four CPUs. The BBD-RSM approach effectively determines the key design parameters for optimizing cold plate configurations. In addition, the analysis provides guidance on selecting an appropriate coolant mass flow rate to enhance thermal performance and energy efficiency in CPU cooling applications. In the two-CPU case, the design with a smooth cold plate on the first CPU and a finned cold plate on the second achieves the lowest thermal resistance, improving temperature uniformity by 23.5 % and 78.5 % compared to the series and parallel arrangements, respectively. For the four-CPU case, a series design is recommended to achieve lower thermal resistance and improved temperature uniformity, while a parallel design is preferred to minimize pumping power consumption. However, a hybrid configuration that strategically uses finned cold plates results in the optimal design, reducing pumping power consumption by at least 10 % and limiting the temperature difference between the CPUs to approximately 1 K.
引用
收藏
页数:25
相关论文
共 70 条
[1]  
Aiman Y, 2021, Evergreen, V8, P187
[2]  
Aizawa N, 2020, ASHRAE TRAN, V126, P212
[3]  
[Anonymous], 2012, Datacom Equipment Power Trends and Cooling Applications
[4]   Water immersion cooling of high power density electronics [J].
Birbarah, Patrick ;
Gebrael, Tarek ;
Foulkes, Thomas ;
Stillwell, Andrew ;
Moore, Alexandra ;
Pilawa-Podgurski, Robert ;
Miljkovic, Nenad .
INTERNATIONAL JOURNAL OF HEAT AND MASS TRANSFER, 2020, 147
[5]  
Bonner RW, 2008, P IEEE SEMICOND THER, P150
[6]  
Caney N, 2018, C HEAT TRANSF INT SU, P631
[7]   Improving Data Center Energy Efficiency With Advanced Thermal Management [J].
Chainer, Timothy J. ;
Schultz, Mark D. ;
Parida, Pritish R. ;
Gaynes, Michael A. .
IEEE TRANSACTIONS ON COMPONENTS PACKAGING AND MANUFACTURING TECHNOLOGY, 2017, 7 (08) :1228-1239
[8]  
Chang B, C ADV 48U SINGL PHAS, P1
[9]   Application status and prospect of spray cooling in electronics and energy conversion industries [J].
Chen, Hua ;
Ruan, Xiao-hui ;
Peng, Yu-hang ;
Wang, Yan-ling ;
Yu, Chang-kun .
SUSTAINABLE ENERGY TECHNOLOGIES AND ASSESSMENTS, 2022, 52
[10]   Bidirectional symmetrical parallel mini-channel cold plate for energy efficient cooling of large battery packs [J].
Chen, Yiming ;
Chen, Kai ;
Dong, Yuan ;
Wu, Xiaoling .
ENERGY, 2022, 242