ASHRAE论文：数据中心制冷能耗和投资成本

yi321yi 2023-07-20 发布于上海

展开全文

数据中心能源和投资成本考量

Data Center Energy And Capital Cost Considerations

July 19, 2023 BY DAVE QUIRK, P.E., MEMBER ASHRAE; HAMED YASSAGHI, PH.D.; TOM DAVIDSON, P.E., MEMBER ASHRAE

译者说

数据中心建设成本高，运行期间能耗高且维护工作量大，空调制冷系统是保证数据中心安全运行的核心保障之一；制冷系统在数据中心的整体投资占比较高，运行期间能耗约占数据中心总能耗的20%~40%，同时空调制冷系统与气候特征有很强的关联，充分利用自然冷源，降低机械制冷，设计合理的空调制冷系统架构可以有效地降低投资成本及运行能耗；数据中心全生命周期投资成本、能源利用及安全稳定运行需要结合气候特征寻找平衡点。

ASHRAE数据处理环境热指南(第5版)1中关于服务器可靠性数据分析部分，以及对数据中心的能源消耗和投资成本的影响部分未被充分利用。这个可靠性分析被定义为x-因子，是第5版热指南第2.4.3节“服务器可靠性趋势与服务器进口环境温度”的重点内容。

An underused portion of ASHRAE’s Thermal Guidelines for Data Processing Environments, 5th Edition,1 is the analysis of server reliability data and the impact this can have on both the energy consumption and the capital cost of a data center. This reliability analysis is called x-factor and is the focus of Section 2.4.3, Server Reliability Trend Versus Ambient Temperature, of the Thermal Guidelines’ 5th edition.

x-因子可靠性指标到底是什么？其首次出现是在热指南的第 3 版中2。如文档所述，“由于需要考虑诸多的变量和场景，TC 9.9 采取的方法是初步建立故障率（x-因子）为1.0的基线，这个基线反映了在ITE设备恒定进口温度为20°C (68°F) 下的平均故障概率。”再基于制造商的数据创建一个表格，以显示其他信息技术设备 (ITE) 进口温度下的相对故障率（具有下限、平均值和上限）。该表之前已在 ASHRAE Journal3 中转载，但在表 1 中显示为更新版。

What exactly is the x-factor reliability metric? It first appeared in the 3rd edition of Thermal Guidelines.2As stated in the document, “Since there are so many different variables and scenarios to consider, the approach taken by TC 9.9 was to initially establish a baseline failure rate (x-factor) of 1.0 that refmected the average probability of failure under a constant ITE inlet temperature of 20°C (68°F).” Using manufacturer’s data, a table was then created to show the relative failure rate (with lower, average and upper bounds) at other information technology equipment (ITE) inlet temperatures. This table has been previously reproduced in ASHRAE Journal3 but is shown in Table 1 as a refresher.

本专栏文章的目的是重新审查x-因子，并着眼于潜在的降低能源消耗和节约投资成本。

The purpose of this column is to reexamine x-factor with an eye toward the potential energy and capital cost savings.

数据中心国际专家培训

ATD设计课程

没有做过运维的设计师就无法设计出好用的数据中心吗？ATD课程将彻底解决这个问题

AOS运维课程

运维管理的直接目标是优秀的运维人员而非设备本身；全球权威的AOS运维管理专家课程将带您深刻理解运维管理的本质

ATS管理课程

正确和系统地了解Tier分级体系会提升数据中心的项目投资回报、减少业务中断风险，ATS课程将全面带您学习Tier知识，帮助您有效提升企业的运营指标和内外部客户的满意度。

点击图片查看课程排期

扫码回复【uptime培训】了解课程详情

数据中心冷却技术与当地气候结合

Alignment of Data Center Cooling System With Local Climate

虽然考虑到x-因子的设计可以显著地提升投资成本和能源成本优势，但重要的是要理解，在不考虑机械制冷情况下的方案初衷，并不意味着让IT设备在允许范围之外运行。这意味着在某些气候条件下需要机械制冷（可能还有加湿和/或除湿）持续保持运行在允许范围内（A2、A3 或 A4，视情况而定）。即使在最极端的设计工况下，也可以通过集成节能装置实现部分冷却。在这种情况下，IT 负载的部分制冷需求可以采用自由冷却技术，从而降低投资成本和能源消耗。

While designing with x-factor considerations can yield significant capital and energy cost advantages, it is important to understand that the intent of operating without mechanical refrigeration does not mean that one should operate outside of the allowable operating range of the IT equipment. This means that mechanical cooling (and possibly also humidification and/or dehumidification) will be required in certain climates to stay within the Allowable (A2, A3 or A4 as applicable) range at all times. It may be that partial cooling can be accomplished via integrated economizer even during the worst-case design condition. In this instance, the cooling system can be designed for only a portion of the full IT load and associated reduction in capital costs and energy consumption.

（表1）

对数据中心设备规模和投资成本的影响

Impact on Data Center Equipment Sizing and Capital Costs

x-因子最初是为了减少甚至消除数据中心机械制冷的理念而设计的。基于x-因子实施变温度控制，主要是降低冷却系统成本，从而降低整体投资成本。对于本专栏文章，我们研究了冷冻水系统的权衡，不过也可以对直接或间接外部空气冷却或其他系统进行类似的分析。

The x-factor was originally developed to exploit the concept of reducing or even eliminating mechanical cooling for data centers. The primary capital cost savings associated with implementing variable supply temperature control based on x-factor is the reduced cost of the cooling system. For this column, we examine the trade-offs for a chilled water system, though similar analyses could be performed for direct or indirect outside air cooling or other systems.

利用冷塔的冷水系统的累计逼近度（冷通道送风温度与室外湿球温度之间的温差）在 8.3°C 至 13.9°C（15°F 至 25°F）范围内。由于冷塔在高湿球温度时其效率更高，因而累计逼近度通常在设计工况附近最低。是否配置冷水机组，其系统的投资成本取决于两个主要参数：

Chilled water systems using cooling towers have an aggregate approach temperature (temperature difference between cold aisle supply temperature and ambient wet-bulb temperature) in the range of 15°F to 25°F (8.3°C to 13.9°C). The aggregated approach temperature is typically lowest near the design condition due to the higher efficiency of cooling towers at high wet-bulb temperatures. The capital cost of a system with chillers vs. a system without chillers depends on two main parameters:

· 取消冷水机组（机械制冷）相关配置带来的成本节约；

· Cost savings associated with eliminating the chillers (mechanical cooling);

· 由于数据中心（机房）在相对高送风温度情况下导致服务器风量增加而带来的成本增加。

· Cost penalty associated with increased airflow requirements in a warm data center due to server fan speed increases.

取消冷水机组相关配置的总投资成本节约可量化为1300 美元/冷吨（370 美元/千瓦），但初投资支出的合理范围与设备类型、地理位置、系统架构和劳动力价格等有关。对于1,000 kW（1 MW）规模的数据中心，配备两台350 rt（1231 kW）冷水机组、N+1冗余，两台冷水机组相关的初投资为：350 rt × 2= 700 rt × 1300 美元/rt = 910,000美元。

The overall capital cost savings associated with the elimination of the chillers can be normalized at $1,300/ton ($370/kW) , but a significant range in CapEx is possible depending on the equipment type, physical location, system topology and labor availability. For a 1,000 kW (1 MW) data center with two 350 ton ( 1231 kW) chillers and N + 1 redundancy, the capacity and CapEx of installed chillers would be 350 tons × 2 chillers = 700 tons × $1,300/ton = $910,000.

在未配置冷水机组的数据中心中，需要增加风量要求的原因是，数据机房每年有一段时间机柜运行温度在允许范围的上限附近。从图1可看出，服务器的风量需求会随着机柜进风温度升高而显著增加，可高达18℃至20℃（65°F 至 68°F）基线值的 250%。更高的风量将导致更多的CRAH 空调数量或其他空调形式的末端需求。如果假设风量增加要求为180%，则需要 1.8 倍的CRAH空调单元数量。假设额外风量的投资支出约为4.50美元/cfm（9.54 美元/L·s，1cfm=1.7m3/h ）。对于本文中提到的案例，需要增加104,820 cfm（49,470 L/s）的风量，相当于471,690美元的成本增加。将这两个值相加，可为具有1,000 kW（1 MW）ITE负荷的数据中心建设净投资成本节约438,310美元。

The reason airflow requirements need to increase in a data center without chillers is that the data hall will be operating near the top of the allowable range for a significant number of hours per year. Figure 1 shows that the airflow requirement for servers can increase significantly with higher inlet temperatures, to as much as 250% of the baseline value corresponding to 65°F to 68°F (18°C to 20°C). This higher airflow will result in the need for more CRAH units or other forms of terminal air distribution. If we assume that the airfl ow rate increase requirement is 180%, 1.8 times as many CRAH units would be needed. The CapEx of additional airfl ow is assumed to be about $4.50/cfm ($9.54/L·s) . For the example provided in this column, an increase in airfl ow of 104,820 cfm (49 470 L/s) is needed, corresponding to a cost of $471,690. Adding these two values together yields a net CapEx savings of $438,310 for a data center with a 1 MW ITE design load.

（表2）

对数据中心能耗的影响

Impact on Data Center Energy Consumption

对于冷水系统，减少（或消除）冷水机组的配置会对一年内的能耗节约产生益处。分析中包含的附加能耗有：(a) 冷却塔风扇能耗增加，以提供全年100%的节能模式冷却，以及 (b) 当服务器风扇转速升高时，需要CRAH空调以更高的风速运行导致所需的风机功耗增加。需要注意的是，在(b)工况时，假设无冷水机组配置的数据中心配备了比基准设计更多的 CRAH空调单元，这样，在一年中仅若干小时冷通道送风温度明显高于20°C (68°F) 的基线 x-因子温度时风机转速较高，除此之外的大多数时间这些CRAH空调单元可以在低转速下运行。因此，无冷水机组设计方案的空调风机年度能耗，通常会低于标准冷水机组设计方案的CRAH 空调单元的风机能耗。准确的权衡点取决于数据中心站点的外部环境湿球温度分布。

For a chilled water system, the energy savings would accrue from the reduction (or elimination) of chiller energy during the course of a year. Parasitic energy to be included in the analysis includes: (a) the increase in cooling tower fan energy to provide 100% economizer cooling for the entire year, and (b) increased fan energy required to operate CRAH units at higher airfmow rates when server fans ramp up. It should be noted that in the case of (b), the chillerless data center is assumed to be equipped with more CRAH units than the baseline design, and these CRAH units can operate at low fan speeds except during the hours of the year when the cold aisle supply temperature is signifjcantly above the baseline x-factor temperature of 68°F (20°C). As such, the annualized cooling air distribution fan energy of a chillerless design will typically be less than the fan energy associated with a standard chiller plant design with fewer CRAH units. The exact tradeoff depends on the ambient wet-bulb distribution of the site in question.

表2比较了基于x-因子变送风温度控制的投资成本和能源支出成本方面的潜在节省空间。作为参考，列出了ASHRAE 90.4-2019的机械负载组件系数（MLC）以及为基准（配置冷水机组）和x-因子（不配置冷水机组）设计计算的MLC。表2给出了基于美国ASHRAE气候分区的典型城市，在场景1（水冷冷冻水基准）和场景2（无水冷冷水机组）的能耗变化通用信息。为确保基线系统与标准设计合理一致，将MLC值与标准90.4-2019中包括的MLC目标值进行比对。基线系统仅在三个气候区略高于标准90.4-2019中提及的MLC目标，而其他所有气候区均略低于标准，结果表明这方面存在合理的一致性。

Table 2 compares the potential reduction in CapEx costs and the potential reduction in energy (OpEx) costs while using x-factor variable supply temperature control. For reference, ASHRAE Standard 90.4-20194 mechanical load component (MLC) factors are also shown, as well as the MLCs calculated for both the baseline (chiller) and the x-factor (chillerless) designs.* Table 2 provides general information on the variation in energy use of Scenario 1 (the water-cooled chilled water baseline) and Scenario 2 (water-cooled chillerless data center) for representative cities in each of the U.S.-based ASHRAE Climate Zones. To make sure the baseline system is reasonably aligned with standard designs, MLC values are compared to the MLC targets contained in Standard 90.4-2019. There was reasonable alignment in this regard, with the baseline system slightly above Standard 90.4-2019 in only three climate zones, and slightly below in all others.

（表3）

无机械制冷的MLC比基线系统MLC平均低0.09。研究了气候对冷却塔、冷水机组和空气分配装置的影响与能源成本的关系。假设美国所有数据中心站点的能源平均零售成本为0.115美元/kWh。可以看到，气候区1A和2A的潜在能源成本节省最多，因为基线系统冷水机组在节能模式下几乎不运行，因而（场景1）相应的MLC更高。

The chillerless system had an MLC that averaged 0.09 less than the baseline system MLC. Energy costs examined the climate impacts on cooling tower, chiller and air distribution energy. An average annual U.S. retail energy cost of $0.115/kWh was assumed for all sites. Note that the potential energy savings is highest in Climate Zones 1A and 2A because the baseline chillers rarely operate in economizer mode and the (Scenario 1) MLC is thus higher.

表3总结了无冷水机组的数据中心与采用机械制冷的“标准”冷冻水系统相比的优缺点。可以看出，需要考虑诸多的因素，无机械制冷的设计可能并非在所有情况下都是最佳的选择。然而，在已知服务器规格参数和故障率的情况下，无机械制冷的设计存在若干优势。即更低的初投资成本、更低的能源成本、更低的维护成本、更简单和更可靠的冷却系统，以及无全球变暖潜能值（GWP）制冷剂问题等。

Table 3 provides a summary of the advantages and disadvantages of a chillerless data center when compared to a “standard” chilled water system with mechanical cooling. As can be seen, there are many factors to be considered, and a design without mechanical cooling is probably not the best choice in all cases. However, in instances in which server specifjcations and failure rates are known, several advantages exist to the chillerless design. These include lower capital cost, lower energy cost, lower maintenance cost, a less complex and more reliable cooling system and no global warming potential (GWP) refrigerant issues.

与项目和IT设备规范相结合

Alignment With Project and IT Equipment Specifications

热指南中给出的x-因子值基于符合A2、A3和A4环境分级的IT设备。

The x-factor values provided in Thermal Guidelines are based on ITE aligned with Classes A2, A3 and A4.

即使基于x-因子分析A2、A3、A4的ITE设备不需要机械制冷，但如果要安装适配于A1和新H1环境分级的高性能设备，也应考虑增加机械制冷。

If equipment corresponding to high performance Class A1 and/or the new Class H1 is being installed, consideration should be given to the addition of mechanical cooling even if it is not needed for the A2/A3/A4 ITE based on the x-factor analysis.

当大多数IT设备适配A3和/或A4环境分级，通过隔离A1/H1环境分级（以及A2分级，具体取决于气候区和冷却系统设计）的IT设备，再使用x-因子变温度控制仍可显著降低初投资和运行成本。

Segregation of Class A1/H1 equipment (and also Class A2 depending on Climate Zone and cooling system design) should still allow for signifjcant capital and operating cost savings by using x-factor fmoating temperature control when most of the IT equipment falls into the A3 and/or A4 classes.

总结

Summary

十年来，x-因子可靠性数据一直是ASHRAE数据处理环境热指南的一部分。较好地实施变送风温度控制，并与x-因子可靠性数据保持一致应该会降低新建数据中心的建造和运营成本，且对大多数气候区的ITE设备可靠性影响很小。投资成本、能源利用和年度x-因子可靠性确切影响依赖于诸多变量。需要综合考虑气候区、IT设备规格参数、能源利用和投资成本影响，再确定x-因子是否适合您的数据中心建设。以上是关于数据中心设计和运营的可靠性数据的两栏内容的第一篇，在第二篇专栏文章中，将重点介绍既有数据中心设施改造设计时，优化和x-因子分析的潜在用途。

The x-factor reliability data has been part of ASHRAE’s Thermal Guidelines for Data Processing Environments for 10 years. Greater implementation of fmoating supply air temperature control to align with x-factor reliability data should result in lower construction and operating costs for new data centers with only minor impact on ITE reliability for most climate zones. The exact impact on capital cost, energy use and annualized x-factor reliability depends on many variables. It is important to look at climate zone, IT equipment specs, energy use and capital cost impact to determine whether x-factor is a good match for your data center. This is the fjrst of two columns on the reliability data on data center design and operation. The second column will focus on optimization and the potential use of x-factor analysis in design modifjcations to existing facilities.

今日话题讨论

降低数据中心冷却设备投资成本，您还有什么妙招？

留言区写下您对今日话题的见解，截止下周三从留言区评论中精选最相关话题的同学将会获得由山特（SANTAK）独家赞助，价值168元的【SP-BOX系列防雷防浪涌插线板】一个。（中奖名单将在下周三留言区置顶公示）快来留言区发表自己的观点吧！