节能探索 | 谷歌在数据中心如何实现24x7全时碳中和-1

关注并将「DeepKnowledge」设为星标

向全天候无碳运营迈进的谷歌数据中心：发展与洞察-1

Moving toward 24x7 Carbon-Free Energy at Google Data Centers: Progress and Insights -1

近年来，谷歌已成为世界上最大的可再生能源采购商，仅在2017年，就从专门为谷歌建造的太阳能与风力发电场采购超过70亿千瓦时的电力（约等于罗得岛州每年消耗的电量）。这使谷歌能够通过直接采购可再生能源来满足全年耗电量：谷歌是同等规模中首家采取此项行动的企业。

In recent years, Google has become the world’s largest corporate buyer of renewable energy. In 2017 alone, we purchased more than seven billion kilowatt-hours of electricity (roughly as much as is used yearly by the state of Rhode Island1 ) from solar and wind farms that were built specifically for Google. This enabled us to match 100% of our annual electricity consumption through direct purchases of renewable energy; we are the first company of our size to do so.

实现 100% 的可再生能源采购目标是谷歌一项重要里程碑，伴随着运营规模的增长，我们将继续加大可再生能源的采购。然而，这也仅仅是个开始。它标志着一个实现更大、更长期挑战的开端：实现谷歌全天候无碳化运营。 

Reaching our 100% renewable energy purchasing goal was an important milestone, and we will continue to increase our purchases of renewable energy as our operations grow. However, it is also just the beginning. It represents a head start toward achieving a much greater, longer-term challenge: sourcing carbon-free energy for our operations on a 24x7 basis.

应对这一挑战，谷歌需要采购充足的无碳能源去随时满足总用电需求。这种做法看起来与现状明显不同，尽管我们大规模采购可再生能源，但仍涉及碳基能源。谷歌的每一处基础设施仍像其他电力消费方一样连接到其区域电网：各地区的电力组合通常包括一些无碳资源（如风能、太阳能、水力发电、核能），但也包括煤、天然气和石油等碳基资源。因此，我们依赖这些基于碳的资源，尤其是在风速或阳光不足时，以及在无碳能源获取有限的地方。无论是否无碳，全天候供电是使谷歌能够持续提供谷歌搜索结果、YouTube视频播放、云平台服务等的动力。 

Meeting this challenge requires sourcing enough carbon-free energy to match our electricity consumption in all places, at all times. Such an approach looks markedly different from the status quo, which, despite our large-scale procurement of renewables, still involves carbon-based power. Each Google facility is connected to its regional power grid just like any other electricity consumer; the power mix in each region usually includes some carbon-free resources (e.g. wind, solar, hydro, nuclear), but also carbon-based resources like coal, natural gas, and oil. Accordingly, we rely on those carbon-based resources — particularly when wind speeds or sunlight fade, and also in places where there is limited access to carbon-free energy. Carbonfree or not, around-the-clock electricity is the fuel that enables us to continuously deliver Google search results, YouTube video plays, Google Cloud Platform services, and much more without interruption.

为了解决碳基能源的使用问题，我们会在太阳能和风能富余的地区和时间段采购剩余的可再生能源。例如，在美国中西部等地采购更多的风能，以抵消我们在亚洲可再生能源采购的短缺。在一些地方，我们采购日间冗余的太阳能，以补偿在夜间使用碳基能源。通过这种方式，我们采购充足的可再生能源去匹配每年的总用电量。我们的采购可直接导致更多的可再生能源被补充到电网中（比如，我们会坚持更高的项目标准），并减少二氧化碳排放。以全球范围和年度为基准，我们采购的太阳能和风能会将我们用电消耗的碳元素全部中和。然而，这仍是一个不完美的解决方案。谷歌将致力于打造一个让每处谷歌设施都始终与无碳能源不间断适配的未来。 

To address our use of carbon-based energy, we buy a surplus of renewable energy in regions or hours when solar and wind power are abundant. For example, we buy larger amounts of wind energy in places like the US Midwest to offset our lack of renewable energy purchases in Asia. And in some places, we buy additional solar energy during the day to compensate for our use of carbon-based energy at night. In this way, we purchase enough renewable energy to match our total annual electricity use. Our purchases directly result in more renewables being added to the grid (i.e. we adhere to high standards of project additionality2 ) and reduce CO2 emissions. On a global and annual basis, our purchases of solar and wind energy zero out the entire carbon footprint of our electricity use. Yet this is an imperfect solution. We want to build a future where each Google facility is always matched — around the clock — with carbon-free power.

谷歌有多种令人信服的理由去追求向无碳能源的转变：我们将努力在气候变化方面起引领作用作为当务之急。作为一个庞大的电力消费者，谷歌致力于最大限度地减少自身的环境足迹；作为一家业务不断增长的企业，我们同样重视可再生能源的本益分析和财务确定性。在保持每年100%的可再生能源采购目标的同时，我们认识到电力行业完全脱碳需要全天候的无碳能源供应。为实现这一长期目标，对于将无碳能源从当今全球电力供应组合中重要但有限的要素提升到为我们的运营提供充分动力，并最终上升为向整个电网提供充足动力的资源而言非常重要。 

Google has multiple compelling reasons to pursue a shift to carbonfree energy sources: we strive to lead on climate change as a business imperative; we are a large electricity consumer that seeks to minimize our environmental footprint; and we are a growing business that prizes the cost-effectiveness and financial certainty of renewable power sources. In parallel to maintaining our annual 100% renewable energy purchasing goal, we recognize that fully decarbonizing the electricity sector requires 24x7 carbon-free energy. Pursuing this long-term objective is important for elevating carbon-free energy from being an important but limited element of the global electricity supply portfolio today, to a resource that fully powers our operations and ultimately the entire electric grid.

为指导谷歌的发展，我们已着手研究在数据中心园区中如何发展全天候无碳电力的方式。在谷歌的总用电量中，数据中心用电占到绝大部分。迄今为止，我们取得的进展主要得益于签署风能和太阳能采购协议（PPA），以实现每年100% 可再生能源的使用目标。谷歌对能源效率的长期承诺也发挥了至关重要的作用：通过最大限度地降低电力需求，我们减少了与消费相匹配所需的无碳能源量。谷歌数据中心的平均能耗约为行业水平的一半，我们还将继续采取创新策略来提升效率。 

To guide Google’s progress, we’ve analyzed how we are tracking toward 24x7 carbon-free electricity at our data center campuses, which account for the vast majority of Google’s total electricity consumption. Our progress to date has primarily resulted from signing wind and solar power purchase agreements (PPAs) in pursuit of our annual 100% renewable energy match goal. Google’s longstanding commitment to energy efficiency has also played a vital role: by minimizing our electricity needs, we have reduced the amount of carbon-free energy required to match our consumption. The average Google data center uses half as much energy as a typical data center, and we continue to employ innovative strategies to improve efficiency.3

为指引我们实现全天候无碳能源使用的目标，确定我们目前在每个地区所处的位置是至关重要的。本文的分析标志着我们第一次深入研究这个问题，而这正是朝着无碳化发展未来所迈出的重要一步。 

To orient us toward achieving 24x7 carbon-free energy, it’s important that we ascertain where we stand today in each region. The analysis herein marks the first time that we have examined this question in depth and is an essential step in working toward a carbon-free energy future.

在本文中，我们讨论了我们最初的全天候无碳化运营努力的三个方面： 

In this paper, we discuss three aspects of our initial 24x7 carbon-free energy efforts:

评估给定数据中心与区域全天候无碳能源匹配程度的框架

A framework for evaluating how well a given data center is matched with regional 24x7 carbon-free energy

谷歌在无碳能源方面取得进步（以几处具有代表性的数据中心为例）

Examples of Google’s progress on carbon-free energy at a representative selection of data center campuses

指引我们实现全天候无碳化运营的观点

Insights that can guide our journey toward 24x7 carbon-free energy

我们将无碳能源定义为在发电过程中不直接排放二氧化碳的能源。这包括太阳能、风能、地热能、水力发电、生物质能及核能等可再生能源。未来，我们的框架可以扩展到其他技术，如碳捕获和储存，这些技术尚未大规模应用，但可以从其他来源实现无碳发电。 

We define carbon-free energy as any type of electricity generation that does not directly emit carbon dioxide. This includes renewables like solar, wind, geothermal, hydropower, and biomass.4 Nuclear power is also carbon free. In the future, our framework can be extended to other technologies, such as carbon capture and storage, that are yet to be deployed at scale but could enable carbon-free power generation from additional sources.

通过分析细粒度能源数据，我们可以测量给定数据中心的每小时用电量与区域电网每小时无碳供电的一致性。第一个无碳能源组合是谷歌通过长期PPA从一个地区的风能和太阳能项目采购的电力。 第二个无碳能源组合是来自更广泛的区域电网的电力，如核能、水力发电及谷歌尚未直接签约的可再生能源。 

By analyzing granular energy data, we can measure how a given data center’s hourly electricity use coincides with hourly carbon-free electricity supply on the regional grid.5 This carbon-free electricity consists today of two components. The first carbon-free component is electricity that Google buys, through long-term PPAs, from wind and solar projects in a region.6 The second carbon-free component is electricity coming from the broader regional power grid via sources like nuclear, hydropower, and renewables that Google has not directly contracted.

在评估数据中心的每小时用电量与无碳能源的匹配程度时，我们的框架首先考虑与谷歌在特定区域与PPA相关的可再生能源发电情况。如谷歌在给定时间内购买的区域可再生能源电力大于等于数据中心的电力负荷，则数据中心将100%匹配该时段的无碳能源。但如果谷歌的区域可再生能源发电在一定时间内不足以与数据中心的负载相匹配，那么剩余的用电需求将由该地区的电网提供。图1显示了数据中心用电如何在给定小时内与区域无碳能源供应保持一致的几种场景。 

In evaluating how well a data center’s electricity consumption is matched on an hourly basis with carbon-free energy, our framework first considers the renewable energy generation associated with Google’s PPAs in a particular region.7 If regional Google-purchased renewable generation in a given hour is equal to or greater than a data center’s electricity load, then the data center is matched 100% with carbon-free sources for that hour. However, if Google’s regional renewable generation is insufficient in a given hour to match the data center’s load, then the remainder electricity is attributed to the region’s grid mix. Figure 1 shows several possible scenarios for how data center electricity use can line up with regional carbon-free energy supply in a given hour.

在任意给定的小时内，数据中心的能源组合采取以下形式之一：

In any given hour, a data center’s energy profile takes one of the following forms:

在任意时间，各地区的电网组合一般由无碳电力和碳基电力组成。电网组合可能因地区或时间而相差较大。例如，由于北欧国家和美国太平洋西北地区拥有丰富的水力发电，电网组合中无碳电力比重较大；而亚洲和美国东南部的部分地区主要由化石燃料发电站供应电力，碳基电力比重较大。同样，太阳能超容的地区往往在白天比夜间拥有更多的无碳电力。

In any given hour, a region’s grid mix usually consists of some carbonfree power and some carbon-based power. The grid mix can vary greatly from region to region, and also from hour to hour. For instance, the Nordic countries and the US Pacific Northwest have a highly carbon-free grid mix due to abundant hydropower, while parts of Asia and the US Southeast have a largely carbon-based mix due to fossil power plants. In the same vein, regions with outsized solar capacity often exhibit a more carbon-free grid mix during daytime than nighttime.

当数据中心每小时能源消耗量与区域无碳发电量（无论是来自谷歌采购的可再生能源还是其他电网）完全匹配时，数据中心的碳足迹将降至最低。图2描绘了2017年1月芬兰某数据中心完全无碳化的一天，其每小时用电量与谷歌在该地区采购的风能100%相匹配。在一天中的大多数时间，我们的区域风力能源采购协议提供的能量大约是匹配数据中心负载所需的能量的两倍。 

The carbon footprint of a data center is minimized when its hourly energy use is fully matched by regional carbon-free generation, either from Google-contracted renewables or the broader grid mix. Figure 2 depicts a fully carbon-free day in January 2017 at our Finland data center, where hourly electricity consumption was 100% matched with Google-contracted wind energy in the region. In most hours of the day, our regional wind PPAs produced approximately double the energy required to match our data center load.

案例：芬兰某数据中心（2017年1月）

Example from Finland data center on a day in January 2017

但是，通常没有足够的无碳能源在任意小时内与数据中心的每小时电力负载完全匹配。这里有几点可能与谷歌或电网有关的因素，可能会导致特定时间内无碳能源的短缺。谷歌方面的因素可能包括该地区缺乏谷歌的可再生能源采购协议：由于风速较低或阳光微弱，我们现有的PPA的产量不足;或是缺乏储存的无碳能源来填补空白。电网方面的因素可能包括区域电网组合未能完全无碳化，或抑制无碳能源供应的其他技术和监管限制。在发生无碳能源短缺时，我们将数据中心每小时用电量的一部分归因于区域电网的碳基电力。

However, there is often not enough carbon-free energy available to match 100% of a data center’s hourly electric load in a given hour. There are several possible factors — related to Google or the grid — that can cause an hourly shortfall in carbon-free energy. On the Google side, factors may include a dearth of Google renewable PPAs in the region; insufficient output from our existing renewable PPAs, due to low wind speeds or faint sunlight; or a lack of stored carbonfree energy to fill the gaps. On the grid side, factors may include a regional grid mix not being fully carbon free, or other technical and regulatory constraints that inhibit carbon-free energy supply. In the event of a carbon-free energy shortfall, we attribute a portion of the data center’s hourly electricity consumption to carbon-based power from the regional grid.

例如，图3显示2017年8月，我们在智利基利库拉的数据中心，由于光照变化，该中心经历了几个小时的无碳能源短缺。在阳光明媚的时候，谷歌在该地区所采购的太阳能农场产生的能量足以与数据中心负载完全匹配：然而，在阳光不充足的时候，数据中心与无碳能源的匹配度低于100%，转而依赖于电网供电组合及其碳基燃料。尽管午间太阳能产量过剩超过"抵消"了当天碳基能源的总贡献，但数据中心仍未能实现24小时的100%无碳匹配。 

As an example, Figure 3 shows a day in August 2017 at our data center in Quilicura, Chile, which experienced several hours of carbonfree energy deficit due to sunlight variability. During sunny hours, a Google-contracted solar farm in the region produced more than enough energy to fully match data center load; during less sunny hours, however, the data center's match with carbon-free energy fell below 100%, relying instead on the grid mix and its carbon-based content. Although excess midday solar production more than “offset” the day's total contribution from carbon-based energy, the data center failed to achieve around-the-clock 100% carbon-free matching.

案例：智利某数据中心（2017年8月）

Example from Chile data center on a day in August 2017

图2和3描绘了数据中心在给定24小时内的无碳配置组合。但是，我们的能源挑战比任何一天都严峻：我们希望在一年中的每一小时将每个数据中心与无碳能源相匹配。为了捕捉这种更广阔的视野，我们可以使用一张热图，可视化一年中的每小时（全部8760个小时），其中更环保的时间表示更高的无碳能源匹配速度。 

Figures 2 and 3 depict the carbon-free profile of a data center for a given 24-hour period. But the scope of our energy challenge is bigger than any single day: we want to match each data center with carbon-free energy in every hour of the entire year. To capture this more expansive view, we can use a heat map that visualizes each hour of the year — all 8,760 of them — where greener hours represent a higher rate of carbon-free energy matching.

图4中的热图描绘了一个理想化的数据中心，该数据中心在一年中的每小时都能与区域无碳能源100%匹配。图形自左到右分别是1月1日与12月31日；图形自顶部到底部则是从每天午夜开始直至当日结束。 

The heat map in Figure 4 depicts an idealized data center that is matched 100% with regional carbon-free energy in every hour of the year. Going from the rectangle’s left to right, January 1st is the left edge and December 31st is the right edge; going from the rectangle's top to bottom, each day begins at midnight and elapses downward until the day is done.

一座由无碳能源全时段供电的数据中心

A data center that is 100% matched with carbon-free energy around the clock — 24 hours a day, 7 days a week, 365 days per year

相比之下，图5描绘了一座现实中的数据中心（我们将在下面进一步探讨），该数据中心在任何时间都无法与无碳能源100%匹配。其能源使用组合每天都在变化，并在正午显著达到无碳能源的最高占比——这与太阳能利用率相吻合。

By comparison, Figure 5 depicts a real-world data center (which we’ll revisit further below) that fails to reach a 100% match with carbon-free energy in any hour. Its energy profile varies throughout each day, though clearly achieves the highest rates of carbon-free energy matching during midday hours — coinciding with the availability of solar energy.

该数据中心的正午用电量与当地太阳能供应量相匹配，使得其无碳电力占比在这该时间段内达到最高。

This data center’s midday electricity use is matched with regional solar energy, making its carbon-free profile highest during those hours

通过将每小时用电量和每小时区域供电量相结合，前述框架与可视化方法可以量化给定数据中心在任何时间尺度上与无碳能源的匹配程度。 

By combining data on hourly electricity consumption and hourly regional electricity supply, the foregoing framework and visualization methods can quantify how well a given data center is matched with carbon-free energy on any time scale.

翻译：

王璐琦

上海上证数据服务有限责任公司 运营管理部助理经理

DKV（Deep Knowledge Volunteer）精英成员

公众号声明：

本文并非原文官方认可的中文版本，仅供学习参考，不用于任何商业用途，版权归原文及作者所有，文章内容请以英文原版为准。中文版未经公众号DeepKnowledge书面授权，请勿转载。

扫描下方二维码关注我们

更多惊喜 等你发现

公众号 | 视频号 | 头条号 | 西瓜视频 | 新浪微博