Experimental study of pump-driven chip-level two-phase cooling system for data centres

Author:

Wang Leixin1ORCID,Cheng Hao1,Yang Tongzhi1,Yuan Weixing1,Ren Kexian12

Affiliation:

1. Laboratory of Ergonomics and Environment Control, School of Aeronautic Science and Engineering, Beihang University, Beijing, China

2. Department of Building Science, School of Architecture, Tsinghua University, Beijing, China

Abstract

In this study, we proposed and designed a pump-driven chip-level two-phase cooling system (PCTCS) and experimentally demonstrated that it can directly cool central processing units in data centres (DCs). The operation of the PCTCS involves an internal refrigerant cycle (IRC) and an external water cycle (EWC). Within the IRC, multiple mini-channel evaporators enable the dissipation of heat from the CPUs in the servers through a plate heat exchanger to initiate the EWC, which is followed by the dissipation of the heat to the environment. We performed experiments to systematically analyze the effects of key operation parameters on PCTCS performance of the IRC. The results indicated that the average and the highest core temperature of the CPUs were 76.3°C and 79.0°C, respectively, under extremely harsh heat dissipation conditions. The key driving component of the PCTCS is a refrigerant pump with high energy efficiency ratio (EERp) defined as ratio of heat dissipated by the power of the refrigerant pump. The EERp was 17–18 when the CPUs were at 0% load and 38–42 and 68–76 when the CPUs were at 50% and 100% loads, respectively. Moreover, optimizing the flow resistance of the IRC can improve the performance of the PCTCS.

Funder

Xiang Yang Solergy Technology Co., Ltd

Publisher

SAGE Publications

Subject

Public Health, Environmental and Occupational Health,Building and Construction

Cited by 1 articles. 订阅此论文施引文献 订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3