Data center sustainability: Beyond PPAs
Data centers are being driven by customers to reduce carbon footprints, consumption of energy, water, raw materials and the amount of waste they produce.
Various metrics are being used to measure progress on these fronts.
Green leases are being used to document commitments.
Data center customers want transparent and quantifiable measurements of sustainability. Some customers are required to comply with mandatory environmental and infrastructure codes. Others are vying for third-party published awards and rankings, while some have published goals for their organizations on carbon neutrality and other sustainability targets.
Some data center customers remain who value price over sustainability, but a recent survey by 451 Research found that almost a third of multi-tenant data center representatives said all of their customers want contractually binding commitments to efficiency and sustainability, while another 44% said most of their customers want this.
Sustainability commitments are being written into green leases.
A green lease uses standard lease clauses that provide for the management and improvement of the environmental performance of a building to align financial incentives and sustainability goals between a landlord and a tenant. Green lease provisions may incentivize energy and resource efficiency investments, streamline renewable energy procurement and support sustainable building certifications.
Some of the most common efficiency metrics to measure the sustainability and efficiency of a data center include the following: PUE (power usage effectiveness, a metric of energy efficiency), WUE (water usage effectiveness, a measure of how efficiently water is used) and CUE (carbon usage effectiveness, a metric of operational sustainability).
These metrics were developed by the nonprofit industry consortium The Green Grid. The Green Grid is currently developing a new IUE metric (infrastructure utility effectiveness) to determine how much design infrastructure capacity an operational data center is able to use. The goal of this new metric is to identify stranded capacity and opportunities to improve efficiency. These and other metrics may be incorporated into green lease provisions, or used as benchmarks for corporate sustainability goals.
In addition to green lease terms between a data center landlord and tenant, data centers can apply for certification under green building rating systems.
In the United States, the most widely used green building rating system is the Leadership in Energy and Environmental Design (LEED) developed by the US Green Building Council. Data centers may also be certified for energy efficiency under the national data center energy efficiency information program by Energy Star, part of an initiative by the US Environmental Protection Agency and the US Department of Energy. Only data centers in the top 25% in energy performance may receive Energy Star certification. Certifications on energy efficiency and sustainability are also available through government programs in other countries, as well as third-party certification providers. BREEAM certifies sustainability ratings for buildings and other infrastructure projects, both during construction and for in-use facilities, in 89 countries. CEEDA (Certification of Energy Efficiency for Data Centers) is a global organization with certification standards designed specifically for different types of data center facilities.
Top industry players also compete for a number of sustainability awards and rankings. These include accolades from the following organizations: GRESB (the Global ESG Benchmark for Real Assets, a nonprofit that evaluates ESG performance data), the Carbon Disclosure Project (a nonprofit that runs a global disclosure system for investors, companies, cities, states and regions to manage their environmental impacts), RE100 (an initiative of global businesses, aiming for a transition to 100% renewable electricity), Sustainalytics (an independent global provider of ESG and corporate governance research and ratings), the US EPA's green power partnership (a voluntary program that encourages organizations to use green power as a way to reduce the environmental impacts associated with conventional electricity use), REBA (the Renewable Energy Buyers Alliance, a group of energy buyers aiming to achieve a zero-carbon energy future) and the Data Center Coalition's energy committee (the energy committee of the trade association for the data center industry).
Advances in Design
Focus on sustainability in design can start as early as site location.
Electing to build a data center on a brownfield, rather than greenfield, site may be considered to facilitate re-use or re-purposing of existing building infrastructure. Proximity of public transportation may be factored in to cut down on carbon emissions from vehicle traffic.
Increasingly, data center owners and operators are assessing and selecting construction and internal infrastructure materials based on environmental performance and implications such as carbon footprint, hazardous chemicals and recycling or refurbishing opportunities.
Data center design and construction is taking into account "embodied" carbon (carbon emissions resulting from construction supply manufacturing, transportation and installation) as part of the effort to have a more accurate measurement of the data center's true carbon footprint. Building Transparency, a nonprofit, manages the embodied carbon in construction calculator (EC3) — a free tool providing access to supply chain data and measurements on embodied carbon in construction materials. Use of EC3 is being promoted in the data center industry. The EC3 tool allows a user to see the carbon and price impact of electing different choices in building materials.
Technology advances in building materials help make these choices possible.
In one example, CarbonCure, a concrete material, is now used for tile walls that frame data centers. Concrete's durability and strength are ideal for industrial construction, but the production of cement requires the use of substantial energy, and the actual chemical process emits very high levels of CO2. CarbonCure takes CO2 produced by large emitters such as refineries and chemically mineralizes it during the concrete manufacturing process to make greener and stronger concrete. The process reduces the volume of cement required in the mixing of concrete, while also permanently removing CO2 from the atmosphere.
Another greener replacement material is a natural fiber-filled polypropylene, derived from jute fibers, that has been developed for use in parts inside and outside the rack, including adapters, bus-bar covers and other mechanical parts inside the server.
In order to contribute to joint innovations and the common goals of increasing sustainability, maintaining efficiency in production and promoting favorable economics, information on innovative and sustainable materials is shared across the data center industry through forums, including the nonprofit Open Compute Project, an organization that shares designs of data center products and best practices.
In addition to siting, materials fabrication and construction management, data center design is advancing to increase overall sustainability and efficiency.
Modular data center construction prevents over-building upfront and allows incremental capacity to be added as needed over time. This avoids the construction of massive data centers where a single-tenant cloud provider or multi-tenant colocation owner may require years to work up to using the building's capacity. This also has the added benefit of allowing for faster time to market. A modular system facilitates design of components to have their own sustainable lifecycles.
High-density construction leverages the space in a data center more efficiently and packs more computational power into a smaller amount of real estate. Advanced cooling technologies are key in high densities, given the increased heat generated by components packed closely together. Where prior construction required raised floors or room-level air cooling in order to cool servers, new approaches such as cooling through cold plates and tubes and immersion cooling technology eliminate the need for hot air and cool air aisles and major HVAC systems pumping air in and out of the data center. Immersion cooling seals servers and other electronic equipment in a dielectric liquid with a boiling point lower than water. In single-phase immersion cooling, the fluid is cycled out to a heat exchanger where it is cooled and then cycled back to the immersion tank. Two-phase immersion cooling allows the fluid to boil and the resulting vapor to condense on a heat exchanger inside the immersion tank. These advanced cooling technologies reduce both energy and water consumption.
Use of Technology
Sophisticated technologies, such as artificial intelligence and machine learning, are being used in data center design and operation to increase both sustainability and efficiency.
Technology is used in data centers to monitor energy consumption, water use, temperature, humidity and peak demand cycles, and it can serve in various functions, including smart temperature and lighting controls, rainwater reclamation, waste heat recycling and efficient cooling.
AI and machine learning technology can identify where equipment is wasting electricity, hot spots inside a data center and anomalies or performance issues in processes or equipment. This information can be used to minimize downtime and increase output, and it can train deep neural networks to optimize data center performance and increase efficiency. Prognostic artificial intelligence can forecast future events such as surges in demand or temperature changes and adapt system variables accordingly. Use of prognostic AI can prevent a data center from going beyond its operating constraints while also ensuring it operates as efficiently as possible.
Google has used machine learning in its data centers to control cooling systems, as developed by its artificial intelligence research group, DeepMind. Every five minutes, the cloud-based AI pulls a snapshot of the data center cooling system from thousands of sensors and feeds it into deep neural networks to predict how different combinations of potential actions will affect future energy consumption. The AI system then identifies which actions will minimize energy consumption while satisfying a set of safety constraints. Those actions are sent back to the data center, where the actions are verified by the local control system and implemented. Use of this AI resulted in a 40% drop in energy used for cooling and a 15% reduction in overall energy consumption.
The use of artificial intelligence not only increases the density of data center components by packing more computing power into each piece of equipment, but it also allows for more tailored cooling solutions to combat any rise in heat resulting from increased density.
Many technology majors have developed their own bespoke systems suited to their data centers. Huawei created several AI-based systems, including iCooling, iPower and iManager, that allow for intelligent thermal management, an increase in data center availability and a reduction in inefficiencies. Equinix developed IBX SmartView, a bespoke data center infrastructure management software-as-a-service tool to analyze the operations of its data centers.
Some of the largest industry players are also undertaking research and testing in pioneering technologies as they work toward lowering the carbon output of data centers. Microsoft is currently working on initiatives that include underwater data centers for edge computing scenarios with controlled environments that may also extend the life of servers, data storage using DNA and holograms to house immense amounts of data, and containerized data modules that use satellite broadband connectivity, allowing capacity to be deployed in geographies with no fiber infrastructure.
Sustainable water strategies for data centers include both sourcing and design.
Water is primarily used in data centers for cooling. About 40% of the power consumed by a data center may go toward air-conditioning. Use of water-based evaporative cooling technologies has been a common tool to reduce power consumption. This reduction in power consumption comes at the expense of increased water usage. Data center operators are stepping up their efforts to reduce reliance on water supplies, as global sustainability movements increase the focus on water as a finite resource.
On the water supply front, a data center district may include water treatment plants that allow the data centers to cool their servers using local bodies of water or wastewater from municipal water systems or shared cooling solutions, such as district chilled water and river water cooling loops. Google has data center districts in Virginia, Washington and Texas that offer "grey water" feeds providing recycled wastewater to industrial customers. Availability of water supply alternatives may also factor into data center siting.
On the design front, an increasing number of data center providers are choosing cooling systems with minimal or no need for water. Where temperatures are low enough, fresh-air cooling can be employed. In hotter climates, innovative coolant solutions, refrigerant economization, water-to-the-chip technology, immersion cooling and rear-door chilling units provide effective cooling without high water consumption. Data centers that continue to use water for cooling are incorporating rainwater recovery strategies that capture rain from huge roofs or parking lots and store it on site, reducing potential burden on local water systems.
Facebook has deployed the StatePoint liquid cooling system, a new evaporative cooling system using a liquid-to-air energy exchanger where water is cooled as it evaporates through a membrane separation layer. The StatePoint liquid cooling requires less water than a typical indirect cooling system by using air to cool water instead of using water to cool air. Based on testing for several different locations, it is anticipated that StatePoint liquid cooling can reduce water usage by more than 20% for data centers in hot and humid climates and by almost 90% in cooler climates, in comparison with previous indirect cooling system technologies.
The Green Grid nonprofit industry consortium is currently developing a total-cost-of-ownership calculator for liquid cooling data centers to forecast the benefits of incorporating liquid cooling in a data center environment. This tool is planned to be flexible enough to integrate different liquid cooling technologies in variable IT environments and will provide an assessment of cost.
Facebook and other data center majors also invest in circular systems that reuse water as many times as possible before releasing it to wastewater treatment plants. Amazon and others are treating water themselves as they re-use it in their data centers. Data center majors have also taken on water restoration projects in water-stressed regions to promote long-term sustainability of the local watersheds. Recently, Microsoft announced that it will be "water positive" in all of its direct operations by 2030, meaning that the company will reduce the volume of water it uses and replenish more than it consumes.
Water consumption is on its way to becoming as visible a sustainability factor for data centers as energy consumption. Many actors in the data center industry are tracking their water use and reporting it in their performance and sustainability metrics. Some, such as Amazon, have developed their own water efficiency metrics. Others, such as Digital Realty Trust, a real estate investment trust that invests in carrier-neutral data centers and provides colocation services, analyze and track their water-scarcity risks using the World Resource Institute's AqueductTM tool to inform data design decisions and water conservation project selection.
Data center owners, operators and customers also consider waste materials and waste heat in evaluating sustainability and efficiency.
The ability to document a net-zero waste-stream impact has the potential to emerge as a meaningful metric for data center service providers, as customers consider the entirety of their data center's sustainability programs.
Data center owners and operators consider the lifecycle and carbon footprint of materials and components in how hardware is designed, operated and decommissioned and aim to re-use, recycle and generally divert materials and waste from landfills. This requires an understanding of the daily inflow of materials and outbound flow of goods and services compared with the subsequent outflow of material that is reclaimed, repurposed, recycled or disposed of as waste.
In order to address the sustainability of data centers comprehensively, data center designers also look to minimize the amount of toxic-laden electronics that end up in landfills and to generally eliminate the use of hazardous, scarce or difficult-to-create materials altogether.
A key priority is tracking the environmental impact of construction components, including the carbon footprint of the manufacturing of materials and a "reverse-logistics" process to track the waste stream and disposition of debris. Data center owners and operators have found opportunities to reuse hardware within their own data centers. When equipment reaches its end of life at that data center, data center owner-operators engage third parties who remarket components to find a second useful life. Some data center equipment suppliers specifically focus on refurbishing existing parts in an effort to promote waste reduction. When remarketing or refurbishing is not an option, recycling partners break down equipment for reclamation and recycling.
Managing packaging for equipment that is shipped to a data center is also an important facet of comprehensive waste stream accountability.
Waste heat from data centers can be recycled on-site and used in energy production, and waste heat not recycled on-site can be used by neighboring businesses or homes for heating. Data centers are increasingly becoming heat providers. Amazon's corporate headquarters in Seattle has been kept warm by using waste heat from a non-Amazon 34-story data center in a neighboring district. Various northern European countries are directing data center waste heat into district energy systems for reuse. Stockholm's Data Parks hope to use waste heat from data centers to heat 10% of the city by 2035. Facebook is directing waste heat from its data center operations in Denmark to heat nearby homes in Odense. Notre Dame University is using waste heat from a data center to heat a campus greenhouse. Syracuse University's data center transfers excess cold and hot water for air conditioning and heating to a nearby building. While the United States is not on par with Europe, California will be mandating waste heat recovery for data centers in its building efficiency standards.