Fuel cells and microgrids would seem to an obvious combination, but they haven’t been paired as much as their compatibility would suggest.

There are only 116 MW of fuel cells deployed globally in microgrids, according to Navigant Research’s Microgrid Deployment Tracker. That compares with 2,827 MW of diesel power, 1,842 MW of solar power, and 1,641 MW of combined heat and power units installed in microgrids.

Fuel cells have many characteristics that make them good candidates for microgrids. They have high reliability, low emission rates, low noise levels and a relatively small footprint. But they also have drawbacks that hinder their deployment in microgrids. They are sensitive to slight variations in voltage, which makes them “temperamental,” said Peter Asmus, a principal research analyst with Navigant.

Fuel cells are also very expensive compared with some other options, said Katrina Westerhof, an analyst on Lux Research’s distributed generation team. But as deployments expand, costs could come down and fuel cells could be more frequently installed in microgrid settings.

Bloom Energy hopes to be in the front of that change. The Sunnyvale, California company is pushing to create a niche for fuel cells in microgrids. Bloom, in partnership with Constellation and the city of Hartford, recently announced an 800-kW microgrid project in Connecticut that will use its fuel cells.

“Fuel cells are more consistent and reliable than other forms of distributed generation,” said Asim Hussain, vice president of marketing at Bloom. “They are good when mission critical reliability is needed.”

“There are three drivers for the use of fuel cells in microgrids: reliability, sustainability and economics” Ben Chadwick, director of distributed energy development for Constellation, said. The Hartford project is in the city’s Parkville neighborhood and, when it comes online in the third quarter, it will power a school, library, senior center and health center during normal operations and provide emergency power, if there is an outage.

The city wanted reliability and sustainability at an economic cost, Chadwick said. To accomplish that with solar power and energy storage would require a large amount of storage and a small solar farm somewhere, which was not feasible in the urban setting. In the Hartford project, the microgrid is “right at the heart of the load,” where, Chadwick said, fuel cells have an advantage because of their relatively small footprint and low decibel levels.

The Parkville project is one of several being done under an agreement between Bloom and Constellation signed in August 2015 under which Bloom is supplying 40 MW of fuel cells and some of the equity financing. Constellation is developing the projects, and is responsible for finding the customers, signing the power purchase agreements and lining up financing.

Constellation has already secured customer offtake agreements for all 40 MW of the fuel cells and is about half way through the installation of those units at “hundreds of sites” in California, Connecticut, New Jersey and New York, Chadwick said. The company aims to have them online by year end with the investment tax credit for fuel cells expires.

The partnership is the second time around for both companies. They struck a deal about one year earlier under which Constellation bought 21 MW of Bloom fuel cells.

Driven by resilience

The use of fuel cells in grid-tied microgrid is motivated by resiliency, said Westerhof. Fuel cells’ ability to provide baseload power is one of their key attractions, she said, but the downside is their “enormous” capital costs. A Bloom solid oxide fuel cell costs about $7/watt to $8/watt while a diesel generator costs about $1/watt, she said.

That cost difference makes fuels cells less attractive in the market for backup power and more suited for microgrids that are looking for baseload power, usually critical installations such as at schools, government facilities and military bases.

In the long run, Westerhof is more bullish on solar-plus-storage for backup power because the prices of those technologies are falling more rapidly than are fuel cell costs.

Hussain shuns the comparison of fuel cells with solar-plus-storage. For one, he sees the technologies as compatible but, more importantly, they serve different roles. Fuel cells are able to provide baseload power, so the more apt comparison is with diesel generators, he said. On a per kWh basis he said Bloom fuel cells are more efficient then diesel generators with about half the emissions of a fossil generator.

A fuel cell is like a battery. It generates electricity from an electrochemical reaction, but unlike a battery, a fuel cell needs a constant supply of fuel, namely, hydrogen. As long as it has fuel, a fuel cell can keep producing energy.

Hydrogen is the most abundant element in the universe, but it is rarely found unalloyed. The most common sources of hydrogen are from “splitting” water into its constituent components or the less energy intensive method of “reforming” natural gas.

There are at least six different classes of fuel cells. They are generally classified according to the nature of the electrolyte, either solid or liquid, and the material and fuel that they use. Each fuel cell type also has its own operational characteristics, offering advantages for particular applications.

Three technologies rank at the top of commercial viability scale put together by the National Renewable Energy Laboratory: proton exchange membrane (also known as polymer), solid oxide and molten carbonate.

As with many emerging technologies, companies tend to specialize in a single version of the technology. Bloom, for instance, specializes in solid oxide technology.

Because they can be scaled from hundreds of kilowatts up into the megawatt range, solid oxide and molten carbonate fuel cells are most suitable for microgrid applications, Kevin Harrison, senior engineer with NREL, said. They are also most suitable for baseload, rather than backup, applications because they operate at high heat and have long ramp up times compared with diesel engines or gas turbines.

Proton exchange technology, on the other hand, can ramp quickly, which is why it is a frequent choice for use in electric vehicles.

Molten carbonate fuel cells operate at around 650 degrees Celsius. Solid oxide cells operate as high as 1,000 degrees C. The high operating heat of solid oxide and molten carbonate technologies gives them high efficiencies, around 60% and as high as 80% if the heat is captured and put to productive use.

If efficiency were the sole criteria, fuel cells could easily compete with gas-fired turbines and would have the advantage of producing roughly half the emissions. But fuel cells are eight or nine times as expensive as gas turbines on a per watt basis, Adam Forni, senior research analysts at Navigant, said.

Hussain argues that per watt cost comparisons vary by state and by configuration, and he cites other metrics, such as “total cost of ownership” which can work for companies that require high reliability and put a premium on sustainability, such as the 9.75 MW fuel cell system Bloom is installing in Utah for eBay.

Hussain also said Bloom is working on system tweaks such as configuring their fuel cells to share a microgrid DC inverter as a way to gain efficiencies and, further down the road, looking at capturing the CO2 output from its fuel cells.

Another driver of fuel cell deployments are incentives and subsidies. Fuel cells are most economic in states that provide subsidies, such as California, Connecticut, Massachusetts, New Jersey and New York, Westerhof said.

Incentives are important, said Forni, but “the real future for fuel cells is to bring down the costs.”