Navigant Research Blog

Microsoft Deploys Fuel Cells into the Core of World’s First Gas Data Center

— October 12, 2017

Fuel cells have been used to power data centers for years, with players including Apple, eBay, and Equinix all making big investments in the technology. But while most fuel cells power data center facilities from the outside, Microsoft just built a pilot data center with the fuel cells installed right on the racks. This is a shift that could radically simplify future data center infrastructure and improve energy efficiency in these energy-hungry facilities. The big investments noted above notwithstanding, fuel cells have only captured a small fraction of data center market share. New types of deployments like Microsoft’s data center could help drive fuel cells toward the segment’s mainstream.

A Unique Fuel Cell Application

The unique design routes natural gas piping directly to the server racks, which could help eliminate a significant amount of electrical wiring, gear, and controls typical to data centers. A photo from Microsoft’s blog post depicts at least five devices that appear to be fuel cells positioned atop the rack. At an assumed 5 kW-10 kW per rack, the 20 racks likely represent a load of 100 kW-200 kW. The deployment is a good fit for fuel cells since they can be readily scaled in size to match load. That is, a given system can add or remove individual cells or stacks to precisely match demand, a feat not possible with more monolithic alternatives like generator sets (gensets) or microturbines.

There are some potential challenges with this configuration. Installing that much fuel cell support infrastructure (exhaust flue, gas piping, and controls, etc.) could impose significant cost on installations, and maintenance on all those systems could be more taxing than on a single multi-megawatt system installed outdoors. And gas-powered systems generally face the challenge of gas grid outages. Though these are rarer than electric grid outages, they represent a concern—especially in seismic zones like those on the US West Coast. When an outage occurs, many data centers still rely on diesel backup generators since the fuel can be stored onsite. Despite these challenges, this type of deployment shows promise, thanks to ongoing fuel cell technology improvements and the low cost of natural gas.

New Players Enter the Arena

Microsoft mentions project partners McKinstry, a design-build construction firm, and Cummins, an engine and genset manufacturer. Though the fuel cell provider is not noted, Cummins teamed up with UK-based Ceres Power Holdings PLC to develop solid oxide fuel cells for data centers under a Department of Energy (DOE) award in 2016. The award specifies a minimum efficiency of 60% and a capacity of 5 kW scalable to 100 kW. That efficiency is slightly below the 65% (lower heating value) efficiency listed by Bloom Energy, which has largely dominated data center fuel cell deployments to date—though its systems are larger. Regardless of the approach, the high efficiency and consistent energy output of fuel cells is a good match for data centers at large.

While the current design operates on natural gas, a modified future system using pure hydrogen storage could help zero-carbon data centers incorporate intermittent renewable power. That is, the intermittency of renewables like solar PV has historically limited adoption on data center sites, which form a consistent load. If, however, that PV or wind system could generate hydrogen using an electrolyzer in a power-to-gas configuration, the energy could be stored to consistently power the data center via fuel cells. These types of innovations could represent a massive opportunity. According to Yole Développement, data centers used 1.6% of global power production in 2015 and are anticipated to grow to 1.9% in 2020. By any measure, the opportunities in this space loom large.


Hydrogen in Microgrids: Diverse Business Models Begin to Emerge

— September 7, 2017

Hydrogen has long held promise as an energy carrier, though electrolyzer and fuel cell technologies have so far not broken into the mass market—largely due to high costs and infrastructure challenges. As those technologies continue to get cheaper and more efficient, they present intriguing possibilities for hydrogen in one unexpected application: microgrids.

Microgrids, whether grid-tied or remote, rely on local power generation. While solar PV, wind, and other renewables capture many headlines, fossil-fueled distributed generation (DG) accounts for more capacity than any other—40% of the total—among the microgrids tracked in Navigant Research’s Microgrid Deployment Tracker 2Q17. Fossil-fueled DG is often selected since it can provide dispatchable power for long periods and can generally store energy-dense fuel onsite. These facts also hold true for hydrogen. For longer duration storage, hydrogen often outperforms batteries by a significant margin without the emissions associated with fossil fuels.

Emerging business models are setting the stage for hydrogen to play multiple and significant roles in microgrids. Some of these business models are briefly described below.

Remote Microgrids: Hydrogen Displacing Diesel

A new Chilean microgrid developed by Enel, with support from Electro Power Systems (EPS), is showing that hydrogen can fill the same role as diesel, but without the emissions associated with the latter. Remote microgrids have historically depended on diesel gensets, often because many days’ worth of fuel can be stored onsite. While batteries are generally too expensive for multiday storage durations, hydrogen tanks can be easily scaled, independent of the peak power demand.

According to EPS, this type of model is quickly becoming commercially viable. Some reasons include capital and operating cost declines, tighter emissions regulations across the globe, and an eagerness to bypass the diesel value chain across hazardous terrain in remote areas.

Microgrids Exporting Hydrogen

The developers of the Stone Edge Farm microgrid in California had a challenge: despite having excess onsite electricity production from PV and other sources, they faced hurdles in exporting that power in an economically viable way. For example, some of the hurdles to exporting into the California Independent System Operator (CAISO) market include reaching the minimum threshold of 0.5 MW and meeting the ISO’s resource implementation requirements, which include building an onsite meteorological station and control platform. Since these presented significant barriers, the developer looked to another product to export from the microgrid: hydrogen. A bank of onsite electrolyzers turns excess electricity into hydrogen, which then fuels the onsite Toyota Mirai fuel cell vehicles and can also feed the microgrid’s fuel cell bank to generate power.

Islands: Hydrogen as Local Energy Commodity

Many islands are dependent on diesel fuel for both transport and electricity, since it has historically been the cheapest large-scale energy carrier available. However, in places like Hawaii, the appeal of hydrogen is growing thanks to concern over climate change and a growing need to store the high output of intermittent renewables—often using power-to-gas schemes (for more information, see Navigant Research’s Power-to-Gas for Renewables Integration report). In addition, the captive nature of the vehicles helps alleviate the infrastructure problem since relatively few stations are needed. ENGIE, a member of the Hydrogen Council, has been bullish on hydrogen as a future fuel. The company is helping to build an island microgrid based around hydrogen technologies near Singapore. More projects are sure to be announced as the technologies continue to improve.

Thanks to cheap renewables and improving electrolysis technology, hydrogen’s outlook is getting better. Due to the challenges with major fueling infrastructure rollouts, Navigant Research anticipates that hydrogen development will be focused in small geographic areas through 2020. Fitting, then, that hydrogen should find a foothold on the small scale of microgrids.


What Would a Perma-Eclipse Do to Solar Power?

— August 15, 2017

On August 21, a total solar eclipse will captivate millions of observers across the United States. Early on its 1,800 mph path across the country, the moon’s shadow will block 5.6 GW worth of solar power plants in California, the top solar state. The California Independent System Operator (CAISO), the state’s grid operator, is well prepared to respond with increased flex-ramp usage and regulation service procurement—essentially a combination of demand management and flexible natural gas and hydropower units. CAISO is aided in part by lessons learned from the 2015 eclipse in Europe, which has higher renewables penetration than the United States.

The eclipse reminds us that the sun’s rays can experience volatility beyond known daily and annual cycles and begs the question: what would happen if the sun stopped shining? Though the question may sound alarmist, it is not entirely trivial. A significant impact event would have solar-blocking potential, with impacting objects above 1 km (about half a mile) in diameter potentially ejecting large masses of pulverized rock into the stratosphere. Solar-blocking geoengineering projects, while intentionally limited in scope, are specifically designed to block the sun’s rays. Movie buffs will remember that humanity scorched the sky and purposefully blocked out the sun to battle solar-dependent robots in The Matrix trilogy.

Solar PV accounted for just about 2% of global electricity production in 2016 but was also the world’s leading source of additional power generating capacity. With some grids anticipating 30%, 50%, or higher eventual PV penetrations, the potential degree of vulnerability is significant—though the probability of diminished insolation is low.

Utility-Scale Solar PV Generators and Path of August 21 Solar Eclipse

(Source: US Energy Information Administration)

A Portfolio Approach

The appeal of solar PV, especially when combined with storage, is undeniable. A clean, distributable, and increasingly inexpensive energy source, solar PV will be a crucial source of power globally. But, much like a contrarian stock market investor, it is worthwhile to look beyond the hype to see what risks loom. To use another stock market analogy, asset diversification is important on the electric grid.

Most of our energy ultimately comes from the sun, and this is especially true of today’s zero-carbon resources. Wind energy is partially driven by daily solar cycles and experienced a 10% decline during Europe’s eclipse. Hydropower, a flexible generation resource that will help ramp during California’s eclipse, is also driven by the sun’s ability to evaporate water. Biopower, another important carbon-neutral dispatchable resource, is driven by the sun, though on the longer scale of months to years. Compared to solar power, each of these should be less directly affected by potential solar-blocking phenomena. Meanwhile, nuclear, geothermal, tidal, and carbon-captured fossil fuel power are not dependent on the sun’s rays. A vague threat to the availability of solar energy does not suggest these should be adopted en masse. However, some consideration should be given to adopting a diversified, risk-mitigated portfolio of generation.

What would happen if a heavily solar-dependent Earth suddenly lost that energy source? Our collective gaze would undoubtedly turn from the sky back to the ground—to the likes of nuclear, geothermal, and for the quickest fix, fossil fuels. Being prepared ahead of time with a diversified, efficient, and clean energy mix could help mitigate that risk.

Still, this month’s eclipse will affect the US grid little since fossil fuels still account for most of the national power supply. For now at least, we can use plenty more renewables to diversify our energy portfolio.


Can California Wield Energy Storage to Grid’s Advantage?

— July 25, 2017

The California Public Utilities Commission has proposed a ruling that could require thousands of energy storage projects to be more responsive to the dynamic grid as the state continues to grapple with integrating intermittent renewables. The ruling would apply to projects funded by the Self-Generation Incentive Program (SGIP), which underwent a major overhaul this year in a bid to grow energy storage.

In the theme of continual refinement, the ruling proposes to improve projects’ grid support, one of the three key policy goals of SGIP. It would require systems to operate under dynamic tariffs like critical peak pricing or time of use (TOU) or participate as an aggregated demand response or distributed energy resources (DER) product that is bid into the California Independent System Operator’s (CAISO’s) wholesale markets. The goal is to make DER more reactive to real-time changes on the electric grid. The effect on the California’s storage market could be significant, with SGIP likely supporting most of the state’s gigawatt-sized industry through 2020.

Industry Weighs In

The ruling requested comments from all interested parties. Utilities, vendors, and others have weighed in, revealing some key themes:

  • Storage revenue predictability impacts: Many storage deployments primarily monetize by demand charge management, a practice that could become more complex under the new rules. As storage revenue streams continue to be a moving target, this ruling could add complexity to vendors attempting to reliably model the profitability of their projects.
  • Some customers are ineligible: For example, community choice aggregator and direct access customers don’t have access to all the programs and tariffs available to investor-owned utility (IOU) customers. Regarding aggregation in the CAISO market, commentators noted that, while there is a great deal of promise, stakeholder engagement is still in early phases, which could present hurdles to broad adoption.
  • Challenges with existing tariffs: Some expressed concern that existing tariffs and programs may not directly align with SGIP goals. For example, certain TOU customers are guaranteed grandfathered TOU time periods for up to 10 years, which may or may not incentivize battery storage operations to align with SGIP.

New Regulatory Constructs Needed

Many opined that new tariffs or programs are needed to truly get storage systems to align with grid priorities (and carbon emissions mitigation, another of SGIP’s goals). Some predicate their position on whether new TOU rates are rolled out effectively. Others point out that aggregation of storage into virtual power plants may get easier as more DER providers gain approval. And toward the goal of limiting emissions, integrating real-time marginal grid emissions into tariffs would be a major step toward fulfilling SGIPs (and the state’s) carbon emission reduction goals. To that end, companies like the non-profit WattTime are pushing to make such real-time data available, actionable, and ready to implement into tariffs.

The outcome of this ruling remains to be seen, as comments are still being considered. However, one thing is clear: California will continue to push for aggressive integration and aggregation of responsive DER in its quest to develop an advanced and distributed electric grid.


Blog Articles

Most Recent

By Date


Clean Transportation, Digital Utility Strategies, Electric Vehicles, Energy Technologies, Policy & Regulation, Renewable Energy, Smart Energy Practice, Smart Energy Program, Transportation Efficiencies, Utility Transformations

By Author

{"userID":"","pageName":"Adam Forni","path":"\/author\/aforni1","date":"10\/20\/2017"}