搜索历史清除全部记录
最多显示8条历史搜索记录噢~
全部
  • 全部
  • 产品管理
  • 新闻资讯
  • 介绍内容
  • 企业网点
  • 常见问题
  • 企业视频
  • 企业图册
Search

中文版

English

Data center liquid cooling

Crypto Mining Liquid Cooling

3 MW · H200 GPU Cluster Liquid Cooling Solution

A purpose-built cold plate liquid cooling infrastructure for a 128-node NVIDIA H200 compute cluster, designed to Tier 2 standards with a 60-day delivery commitment. The solution leverages Xinjiang's cold, arid climate to maximize natural cooling hours and achieve industry-leading PUE.

3 MW
Total IT load
128
H200 servers
60 days
Target delivery
Tier 2
Data center class
<1.15
Target PUE

Core system configuration

 
Cold plates (server side)
Custom copper/aluminium cold plates bonded directly to GPU and CPU heat sources per H200 node. Supports single-GPU heat flux ≥ 300 W — over 40% more efficient than air cooling.
 
In-rack manifold system
Per-rack distribution manifolds with quick-disconnect fittings enable hot-swap server maintenance. Stainless steel / copper pipework with standardised valves for future capacity expansion.
 
Row / room-level CDU
N+1 redundant CDU to Tier 2 standard. Dual-loop redundant pump sets with online switchover. Plate heat exchanger isolates primary and secondary loops, protecting server pipework from external contamination.
 
Primary-side natural cooling
Xinjiang's annual mean temperature of ~7 °C enables 8,000+ hours of free-cooling per year via dry coolers. Mechanical refrigeration serves only as a warm-season backup.
 
Cooling fluid
Secondary loop uses deionised water or low-concentration ethylene glycol. Given Xinjiang's extreme winter lows, ≥ 30% glycol solution is recommended for full freeze protection down to −20 °C.
 
DCIM monitoring
Full-chain sensor coverage: fluid temperature, flow rate, differential pressure, leak detection, and electrical parameters integrated into a unified DCIM platform with remote O&M and auto-alerting.

System flow architecture

Primary cooling loop

CDU secondary outlet (cool fluid)

→

Row manifold

→

Cold plate heat absorption (GPU/CPU)

→

Warm fluid return

→

CDU plate HX

→

Repeat

Primary-side heat rejection loop

CDU primary outlet (warm fluid)

→

Outdoor dry cooler

→

Cool fluid returns to CDU

→

Free cooling priority; chiller backup in summer

Residual heat handling

Cabinet heat leakage (<5%)

→

Room air

→

Precision in-row A/C unit
Ürümqi's annual mean temperature of ~7 °C and extreme low of −20 °C provide one of China's longest free-cooling windows. In this climate, the cold plate + dry cooler configuration can sustain a PUE below 1.10 year-round — materially lower operating energy costs compared to equivalent facilities in eastern China.

Key technical parameters

Total servers128 × NVIDIA H200
Per-server design load10kW (8× H200 at full load)
Total IT load~3 MW
CDU redundancyN+1 (Tier 2 standard)
Pump redundancyDual-pump active standby, online switching
Cold plate materialCopper / aluminium (selected by heat flux)
Secondary coolantDI water + 30% ethylene glycol (antifreeze)
Pipework standardStainless steel / copper, quick-disconnect fittings
Leak protectionFull-chain leak sensors + auto shut-off valves
Ambient range−20 °C to +45 °C

 

60-day delivery schedule

1
Day 1 – 7
Design sign-off & procurement launch
Finalise cold plate selection, CDU specification, and piping drawings. Simultaneously initiate procurement of long-lead items: plate heat exchangers, pump sets, and cold plates.
2
Day 8 – 25
Equipment manufacturing & prefabrication
CDU cabinet production, custom cold plate fabrication, manifold pre-assembly, and dry cooler manufacturing. Civil works and embedded pipework conditions verified in parallel.
3
Day 26 – 42
On-site installation & pipework
CDU placement, outdoor dry cooler installation, main room pipework, in-rack manifold fitting, cold plate mounting on servers, and quick-disconnect connections.
4
Day 43 – 53
System commissioning & fluid fill
Pressure and leak testing, system fill and air purge, CDU parameter tuning, flow/pressure balancing, leak detection system validation, and DCIM platform integration.
5
Day 54 – 60
Load acceptance testing & handover
128× H200 servers powered on under full load, monitoring data verification, PUE measurement, documentation archiving, and formal project handover.

Tier 2 compliance · certifications

N+1 redundancy Dual supply loops Online maintainability Full-chain leak detection RoHS CE UL CCS Extreme cold rated −20 °C  tier-1 components

Previous page

None

NVIDIA GB200 Cold Plate Liquid Cooling Solution

Next page

Related Recommendations

3 MW · H200 GPU Cluster Liquid Cooling Solution

Delivered a computing cluster comprising 128 NVIDIA H200 servers featuring cold-plate liquid cooling within 60 days.

NVIDIA GB200 Cold Plate Liquid Cooling Solution

The NVIDIA GB200 NVL2 ultra-high-density compute node utilizes cold-plate liquid cooling technology, is designed to T2 standards, and offers rapid delivery within seven days.

Air-to-Liquid Cooling Retrofit for a Shenzhen Listed Company's Data Center

A Shenzhen-listed company completed the cold-plate liquid cooling retrofit for 16 AI servers in 15 days, reducing the PUE to below 1.15.

WhatsMiner M63S++ Hydro-Cooling Solution

100MW liquid-cooled blockchain mining project in Kazakhstan, compatible with WhatsMiner M63S++ and supporting rapid 60-day delivery.

Shenzhen Lianli Liquid Cooling Technology Co., Ltd

E-mail: sales@lianliai.com

Tel：+86-18002936305

Headquarters Address: 12th Floor, Chaopinsan Technology Building, No.18 Gaoke Avenue, Longgang District, Shenzhen City, Guangdong Province, China

Factory Address: Unit 101, Building 15 (Ersha Shenlian), No. 6 Baolong Avenue, Baolong Community, Baolong Subdistrict, Longgang District

© 2022 ShenZhen LIANLI Liquid Cooling Technology Co., Ltd

Privacy Policy

中企跨境-全域组件 制作前进入CSS配置样式

Telegram：

E-mail：

WhatsApp：

Telephone：

在线客服添加返回顶部

右侧在线客服样式 1,2,3 1

图片alt标题设置： Dongguan LIANLI Electronic Technology Co., Ltd.

表单验证提示文本： Content cannot be empty!

循环体没有内容时： Sorry,no matching items were found.

CSS / JS 文件放置地