数据中心封闭冷通道机柜竖直方向流量分配不均匀性的研究

  • 打印
  • 收藏
收藏成功


打开文本图片集

关键词:机柜冷却气流分配;二维建模;封闭冷通道

中图分类号:TK121文献标志码:A

DOI:10.7652/xjtuxb202511008 文章编号:0253-987X(2025)11-0084-11

Abstract: To investigate the influence of diferent enclosed cold aisle parameters and cabinet flow characteristics on the cooling airflow distribution in data center cabinets,a two-dimensional flow model suitable for enclosed cold aisle scenarios is proposed. The model systematically examines the distribution trends of cooling airflow in cabinets under varying flow resistances and cabinet flow characteristics. The two-dimensional model is located on a plane parallel to the cabinet inlet, and a negative source term based on cabinet flow characteristics is introduced in the governing equations to describe the airflow passing through the cabinets. Simulation results for different cold aisle parameters and cabinet flow characteristics reveal that when flow resistance dominates in the cold aisle,airflow tends to concentrate at the bottom of the cabinets,increasing the risk of hotspots at the top. Conversely,when inertia dominates the flow in the cold aisle,more airflow is directed toward the top,leading to potential hotspots at the bottom. Steeper cabinet flow characteristic curves contribute to improved vertical airflow distribution uniformity. A twodimensional simulation of an actual data center cold aisle configuration demonstrates that appropriately increasing the flow resistance in the enclosed cold aisle can reduce the range of the flow uniformity index(FUI) to 83.0% of the baseline scenario. Additionally,increasing the flow resistance within the servers to steepen their flow characteristic curves further reduces the FUI range to 55.4% of the baseline scenario. This study provides a quantitative reference for evaluating the impact of enclosed cold aisle parameters and cabinet flow characteristics on improving flow distribution uniformity.

Keywords: data center thermal management;two-dimensional model; cold aisle containment

随着计算与网络技术的快速发展,AI、5G、物联网、云计算等新兴技术的计算、处理、存储与共享均依赖于数据中心[1]。(剩余15884字)

monitor
客服机器人