Why fuel cells can't get started

February 15, 2008 11:01 AM EST

First National of Nebraska, number 3 on Computerworld's ranking this week of the nation's most green IT operations, has been using fuel cells to provide up to 340 kilowatts of clean power to its data center for more than eight years. So why hasn't the technology caught on? No one has overcome the cost and complexity issues yet, although one startup's efforts look promising.

First National uses fuel cell technology based on a phosphoric acid process that produces electricity for about 15 cents per killowatt hour. The national average for grid power is 10 cents per kWh, and you don't have the issue of maintaining your own power plant. First National can get power from the grid for about 5 cents per kWh, but the high availability aspect of fuel cell technology allows the company to meet uptime requirements for its credit card processing operations without investing $100 million in a backup data center. "It's a huge, huge number, says CIO Ken Kucera. "I'm not a Chase or Citibank that can [afford] three or four data centers." For Kucera, fuel cell technology was part of a high availability, "seven nines" design necessary to allow it to compete with those much larger players with deeper pockets.

Related Articles:
Top 12 Green IT Companies

In the eight years since First National opened its data center, no other business has chosen to run its data center on fuel cells, according to Peter Gross at consultancy EYP Mission Critical Facilities Inc. Given the progress IT has seen in the last eight years, you'd think that fuel cell technology would have evolved to the point where it was cost effective for at least a small segment of the data center market.

That hasn't happened for several reasons. The biggest remains cost.

Because the technology doesn't respond well to sudden changes to power demands, backup systems are needed. "If there is a chiller starting up or some other device that requires a lot of additional power, the fuel cell, because of the chemical reaction required, doesn't respond fast enough and the voltage tends to collapse rapidly," say Gross. To deal with that, large flywheel generators are used to provide transitional power until the fuel cells can get up to speed. "These are fairly expensive devices that need to be placed in series with the fuel cells," Gross says. And they take up a lot of space. The entire system at First National takes up about the same square footage as a tennis court, he says.

Using flywheels is more environmentally friendly than using chemical batteries in UPSes. When compared to UPSes, flywheels don't generate hydrogen, they're easier to maintain and take up less space. But they stor less energy and provide backup power for a much shorter duration, since stored energy is released while the flywheel spins down.

Maintenance is also expensive on fuel cell systems, Gross adds.

Several companies are working on alternative fuel cell technologies. Perhaps the most well known is the proton exchange membrane technology used in some automobiles. But the most promising for data center power, says Gross, is Bloom Energy's solid oxide technology. He says the company is close to producing a fuel cell capable of delivering about 100 kilowatts of power at a price that could make it commercially viable in a few years.


The challenge with the solid oxide process is that the reaction produces temperatures in the range of 600 to 1,000 degrees Centigrade, and that has produced problems with corrosion. Bloom claims to have overcome those issues and says it has been able to mass produce its design. The first tests will likely take place in California and New York, where tax incentives and high utility rates create a proving ground on which to refine the design and start building manufacturing economies of scale. If all goes well, Gross says Bloom could be offering a fuel cell system that produces electricity at a comparable cost to utility power within the next two years.