CFD被广泛用于确定现有或计划中的数据中心的稳态冷却性能。然而，在缺乏紧凑的瞬态服务器模型的情况下，包括服务器热质量影响的全面瞬态建模需要非常大、详细的模型。因此，诸如冷却损失后的瞬态场景通常在CFD中建模，忽略服务器的热质量效应（因此过于保守），或者基于混合理想化建模，不提供局部信息。 此外，迄今为止，在充分混合的模型中捕捉服务器热质量效应所使用的技术是有限的，并且尚未根据实验数据进行验证。我们提出了一个紧凑的瞬态服务器模型，它可以处理所有实际的数据中心用例，并且很容易融入CFD、充分混合和其他数值模型。该黑匣子服务器模型可预测排气温度，以响应随时间变化的环境温度和/或内部加热，还可用于模拟整个机架。该模型引入了“服务器热效率”的概念和一个附加参数，该参数考虑了服务器内部热源和热质量的相对位置。 本文提出了一种实验测量瞬态服务器特性的方法，并报告了样本服务器的初步结果。一个例子强调了紧凑模型的使用及其相对于更近似的技术的好处:ASHRAE交易——第119卷，第2部分，科罗拉多州丹佛。

CFD is widely used to determine the steady-state cooling performance of existing or planned data centers. However, in the absence of a compact transient server model, comprehensive transient modeling which includes the effect of server thermal mass requires prohibitively large, detailed models. Consequently, transient scenarios such as that following a loss of cooling are typically modeled in CFD ignoring thermal masseffects of servers (and are thus overly conservative) or are modeled based on a well-mixed idealization, which provides no local information. Further, the techniques used to date for capturing server thermal mass effects in well-mixed models are limited and have not been validated against experimental data.We propose a compact transient server model that handles all practical data center use cases and is simple to incorporate into CFD, well-mixed, and other numerical models. This black-box server model predicts exhaust temperature in response to time-varying ambient temperature and/or internal heating and can also be used to model entire racks. The model introduces the concept of “server thermal effectiveness” and an additional parameter which takes into account the relative position of server internal heat sources and thermal mass. A method is presented for experimentally measuring the transient server characteristics and preliminary results for sample servers are reported. An example highlights the use of the compact model and its benefits relative to more-approximate techniques