数据中心多场景冷却系统失效研究
Research on multi-scenario cooling system failure in data centers
摘要：
根据机柜通道布局（开放或封闭）、空调回风温度设定值、空调失效台数、空调失效部件等方面划定7种冷却失效场景，选取某中低密度机房为研究对象，对7种场景进行建模及CFD仿真，研究其冷却系统失效后的机房温升情况。研究结果表明：7种场景下设备达到极限温度45 ℃的时间均超过6 min，在此时间内数据中心的柴油发电机系统及其后级负载设备可以全部启动，但多台空调故障短时间内无法全部处理，因此无计划停电导致冷却系统失效的风险在安全可控范围，而多台空调故障引起冷却失效仍存在风险。在抵御冷却失效风险能力方面，由强到弱依次为：机柜封闭热通道、开放通道、机柜封闭冷通道；空调仅压缩机失效强于空调压缩机、风机均失效，前者机房温升到设备极限温度的时间比后者长近13 min；空调回风温度设定值越低，抵御失效风险能力越强。
Abstract：
According to the terms of open or enclosing aisles, set-point of room air temperature, the number and the equipnent component of cooling failure system, defines seven cooling failure scenarios. Taking a medium-low density computer room as a research object, establishes the model, performs CFD simulation and analyses room air temperature rise after cooling failure in data centers. The results indicate that the time reaching the limit temperature (45 ℃) of the IT equipment is more than 6 minutes. During this period, the diesel generator system and the IT equipment of data centers can all be started, while the malfunction of multiple air conditioners cannot be handled completely. The risk of cooling failure caused by non-planned power failure is within the safe and controllable range, however, the risk of cooling failure caused by multiple air conditioners failure still exists. In terms of the ability to withstand the risk of cooling failure, the enclosing hot aisle is super to open aisle, followed by the enclosing cold aisle. The cooling failure only caused by the compressor of the air conditioner has stronger ability than the cooling failure caused by both the compressor and the fan, the former can withstand nearly 13 minutes longer than the latter. The lower of the set-point of room air temperature, the stronger ability to resist the risk of cooling failure.
Keywords:cooling failure, date center, computational fluid dynamics (CFD), enclosing aisle, room air temperature rise, multi-scenario