In an attempt to achieve net zero emissions, tech companies have strongly considered fuel cell technology as a greener and more efficient alternative to energy conversion than the traditional combustion methods of power plants. Over the past several years, the data center industry has experimented with centralized fuel cells through simulation and pilot installations. There are a few reasons to consider fuel cells for data center power production, the first of which is its reliability when connected to the natural gas grid. Reliability of individual fuel cells themselves is mediocre at best; however, when connected to the natural gas grid their reliability improves dramatically since fuel cells can operate indefinitely as long as they are provided sufficient fuel and air. In addition, because data centers will require multiple fuel cell systems, adding a few redundant ones can easily account for any individual fuel cell failures. The reliability of fuel cells is heavily influenced by the reliability of the gas grid, which is known to be high, exhibiting greater than five nines reliability, much higher than three nines for the electric grid. The second benefit is that gas distribution within a data center is much cheaper than the high voltage switch gear, transformers, and copper cables required to connect to the electric grid. If fuel cells are placed closer to the power consumption units , then data centers can easily eliminate the power distribution system, including the backup power generation system. This is highly favorable because the electrical infrastructure accounts for over 25% of the capital cost for state-of-the-art data centers. The third benefit is that fuel cells are environmentally friendly. Although initial phases of introducing fuel cells into the data center would require that they operate on natural gas, even with this source of fuel, fuel cell emissions are much cleaner and far more efficient than those from traditional combustion methods. Carbon dioxide emissions have the potential to be reduced to 49%,heavy duty propagation trays nitrogen oxides by 91%, carbon monoxide by 68%, and volatile organic compounds by 93% when compared with a combustion cogeneration plant.
With respect to architecture, there are several designs for incorporating fuel cells into data centers: at the utility power level, rack level, or server level. The utility power level combines groups of fuel cells to achieve a power rating on the order of megawatts to replace the traditional electric utility power input, therefore disconnecting the data center from the electric grid, or operating in parallel with the grid. The rack level design uses fuel cells with a power rating on the order of kilowatts to power one or a few nearby server racks, eliminating the entire power distribution network within the data center and replacing it with a fuel distribution network . The server level design integrates small fuel cells on the order of watts into the servers, which is similar to the rack level design but eliminates the short distance power cabling needed from the rack level fuel cells to the servers – helping to minimize DC transmission losses. Reliability can be maximized using the server level design because fuel cell failures would only affect a single server; however, it is worth noting that smaller fuel cells may not be as energy efficient and cost effective as their larger counterparts. Microsoft has envisioned a new concept for their data centers, aptly labelled as the ‘stark’ design. They proposed a direct generation method that places fuel cells at the server rack level, inches from the servers. The close proximity allows for the direct use of DC power without the large capital cost, potential for failures, and efficiency penalties associated with AC-DC inversion equipment. As a result, power distribution units, backup power generation equipment, high voltage transformers, expensive switch gear, and AC-DC power supplies in the servers can be completely removed from the data centers. Previous analysis and experiments have shown that low cost, low greenhouse gas, high reliability , and high efficiency can be achieved by using mid-sized fuel cells at the rack level, directly supplying DC power to the servers, and effectively replacing the power distribution system in a data center with a gas distribution network. Data centers are facilities that contain information technology devices used for data processing, storage, and communications in addition to the infrastructure equipment required to operate them as reliably as possible.
The infrastructure equipment typically consists of specialized power conversion and backup equipment and environmental control equipment . Within the data centers, the volume storage servers and cooling system infrastructure are by far the largest consumers of electricity; together they account for over 70% of current energy demand. As servers become more powerful, more power is needed to run and cool them. Therefore, the biggest consumer of square footage in data centers is not by servers but by the power infrastructure. From an outside perspective, Microsoft’s data center in Tukwila, Washington looks like a nondescript sprawl of beige boxlike buildings arranged to be inconspicuous like remote warehouses. Like most data centers, the Microsoft Tukwila facility comprises a sprawling array of servers, load balancers, routers, fire walls, tape-backup libraries and database machines, all resting on a raised floor of removable white tiles, beneath which run neatly arranged bundles of power cabling. To keep the servers cool, Tukwila has a system of what are known as hot and cold aisles where cold air seeps from perforated tiles in front and is sucked through the servers by fans, ultimately expelled into the space between the backs of the racks and ventilated out and away. Tukwila can be thought of as less a building than a giant machine built for computing. Ranging from small computer server rooms to mammoth server farms, data centers now house more than several millions of computer servers. Even more startling is that in 2013 alone, roughly 3 million server rooms used enough electricity to power all households in New York City for 2 years – equivalent to the annual output of 34 large coal-fired power plants. According to a McKensey Quarterly report, the annual CO2 emissions of data centers will reach approximately 1.54 metric gigatonnes by 2020, which could make IT companies among the biggest greenhouse gas emitters. This would be equivalent to the amount of electricity generated by 50 large coal-fired power plants – each with 500 megawatts of capacity – emitting nearly 150 million metric tons of CO2 emissions per year. There continues to be several persisting issues slowing the progress of energy efficiency in data centers: comatose servers, peak provisioning, limited deployment of virtualization technology, failure to power down unused servers, and shortsighted procurement practices.
Fortunately, Microsoft has taken these issues seriously and has worked with vendors to reduce power use when processors are idle. Dileep Bhandarkar, a distinguished engineer at Microsoft who oversaw the company’s server hardware architecture in 2011, says, “It used to be [that] an idle server would be [at] 50% of the power . We’ve pushed that down to about 30%”. Despite this achievement in reducing power consumption of idle servers, many servers remain “comatose” and no longer needed. If half the savings potential from energy efficient best practices are realized, then America’s data center industry could slash their electricity consumption by as much as 40%. In the last decade, there have been dramatic advances in data center design. Previous methods of using outside air directly to cool servers has been replaced by computer room air conditioning systems, evaporative cooling methods have replaced absorption chillers, and power over subscription is used to better utilize power capacity. The data center industry measures their building system efficiency using the power utilization effectiveness measurement. The PUE measurement must consider all the equipment necessary to maintain the daily operation of the data center after receiving power from the utility grid. A diagram of the major components considered in a PUE calculation is shown in Figure 5. A PUE of 2.0 means that for every watt of IT power consumed, an additional watt is required to cool, distribute power to the IT equipment, and operate the data center. A PUE closer to 1.0 means nearly all of the energy is used for computing – the ideal case. Industry-wide,vertical cannabis the PUE ratio between overall facility power consumption and the power used by servers has improved from a 2.0 to a best practice of 1.11. Despite this significant achievement, the fundamental data center power infrastructure, consisting of transformers, power distribution units, uninterruptible power supply systems, and backup generators has changed very little. The power distribution chain starting from the utility grid to the server racks is illustrated in Figure 6. This power system remains necessary to deal with the high-voltage AC power utility grid and its relatively low reliability of three nines . In an attempt to reduce the PUE measurement of Microsoft’s data centers, Microsoft has decided to pursue alternative means of generating power in order to reduce the power infrastructure required to maintain high reliability. According to the United States Environmental Protection Agency , adapting distributed generation methods in data center design could achieve great energy savings, significant environmental benefits, and high power reliability. Therefore, Microsoft has chosen to focus their research on a novel and promising distributed generation technology: high temperature fuel cell systems . They have proposed to place the HT-FCs at the rack level, significantly close to the servers. This new design layout limits the failure domain to just a few dozen servers instead of the entire data center like what would typically occur if the electric grid were to experience failure. Placing the HT-FCs at the rack level would allow Microsoft to eliminate the backup generators, high-voltage transformers, expensive switch gear, transfer switches and AC-DC conversion systems within a data center.
Currently, less than half of the interior square footage within a data center is dedicated to the electrical infrastructure; therefore eliminating this can significantly shrink the physical space requirements. Additionally, shifting the UPS and battery backup functions from the data center into the server cabinet can reduce the power losses from the multiple AC-DC conversions that occur between the utility power grid and the data center equipment . Currently, over 85% of the world’s energy needs are met and will continue to be met in the coming decades through the consumption of fossil fuels . These fossil fuels have become a very reliable fuel resource that is used to produce power through combustion processes , which transforms the fuel’s chemical energy into other forms of energy for everyday use . The relative ease and low cost of producing power through combustion-based processes has made developed and developing countries extremely reliant on the availability and access to the limited fossil fuels on the planet. Yet, the benefits of combustion-based energy generation do not exist without the consequences of harmful criteria pollutant and greenhouse gases emissions. Combustion emissions continue to remain the major source of urban air pollution leading to respiratory health problems and the primary source of greenhouse gas emissions contributing to global warming. Nevertheless, energy conversion and combustion emissions continue to increase simultaneously as the world’s population increases because combustion-based power production is quick and easy to set up compared to alternative methods of energy conversion.As the inadvertent effects of climate change become a very grim reality in this lifetime, great strides have been made for reducing the pollution emitted by combustion sources thanks to scientific and public awareness of federal and state governments. The first major step toward climate change legislation was made when dense visible smog in many of the nation’s cities and industrial centers circa the 1960’s prompted the United States Environmental Protection Agency to pass the Clean Air Act in 1970 at the height of the national environmental movement. With this legislation, the states were now required to adopt enforceable plans to achieve and maintain air quality meeting the air quality standards for the six common “criteria pollutants”. Following with the federal mandate, California implemented the most progressive power generation legislation in 2001 with the introduction of the California Public Utilities Commission , Self-Generation Incentive Program . The SGIP provided incentives to support existing, new, and emerging distributed energy resources, such as wind turbines, recycling waste heat, microturbines, gas turbines, fuel cells, and advanced energy storage systems.