Onsite fuel cell testing can reveal challenges that laboratory evaluations alone may not capture: site integration, changing electrical loads, useful heat recovery and the reliability of supporting equipment. A U.S. field program launched in 1983 put 40-kW phosphoric acid fuel cell (PAFC) power plants into a range of operating environments, providing practical evidence for utilities and other organizations assessing distributed generation.
The program involved the Gas Research Institute (GRI), the U.S. Department of Energy, utilities and an industrial participant. Its findings offer a useful case study in how to structure technology demonstrations—and how to interpret performance beyond a single headline measure.
Why test fuel cells at customer sites?
The field test was one part of a broader onsite fuel cell program. Its purpose was not simply to check whether a packaged plant could generate electricity. The project also aimed to evaluate the technology in real applications, identify requirements for an energy service business and pass operational and market feedback back to technology developers.
That feedback loop matters in industrial automation and power-system integration. A power plant has to work with the facility’s electrical and thermal demands, its utility connection and the equipment surrounding the core generating system. A test across varied sites can expose issues that a controlled test of the fuel cell stack might not reveal.
In this program, 42 plants were installed at 36 sites in 18 market segments. Some served building loads; others supplied heat for processes such as swimming pool heating, food processing, fuel preheating, plant propagation and aluminum parts processing. The variety helped test different combinations of electrical demand and heat use.
Installation and operating configurations
Participating utilities selected sites and took responsibility for site design and installation, using either in-house engineers and crews or outside engineering and construction firms. The reported average preparation time was five weeks before delivery, followed by four weeks to complete installation. The average design and installation cost for these demonstration units was $1,000 per kilowatt.
Operating arrangements varied. Eight sites ran independently of the utility grid in a load-following mode. At the other 28 sites, the plants were connected to the grid and operated at a constant 40-kW output; the utility supplied building demand above that level. These configurations gave the program experience with both isolated operation and grid-connected operation.
Thermal integration varied as well. Recovered heat was used for domestic water, space heating and process applications. Most units were installed outdoors, while six were located indoors, including installations in hotels and an exposition pavilion. In other words, the field work considered not just the generating equipment but also where and how it could fit into an operating facility.
What the field test found
The overall project goals were 300,000 operating hours and 55–65% availability. The program exceeded 300,000 operating hours and reported overall availability of 63%. Fourteen plants achieved availability of 70% or more, while 13 operated for more than 8,000 hours. The longest uninterrupted operating run was 2,165 hours.
Electrical performance generally met expectations. According to the reported results, most plants operated at or above the electrical efficiency specification, and testing also confirmed the specified high part-load electrical efficiency. This is relevant when assessing distributed generation: performance at less than rated output can matter when a unit is serving a variable facility load.
Thermal output told a more qualified story. Measured heat recovery averaged about 10% below specification. Even so, the plants produced approximately one thermal kilowatt-hour for each electrical kilowatt-hour generated. The result illustrates why cogeneration assessments should consider both outputs and how well recovered heat can be used at the site.
Balance-of-plant reliability matters
The main power plant components—the fuel processors, fuel cell stacks and inverters—were reported to have experienced few operating problems during the test. Many of the issues instead involved ancillary equipment, including pumps, valves, fittings and controls.
Two recurring concerns were the stack coolant system and electrical or electronic controls. The coolant system tended to plug after roughly 2,000 operating hours, prompting shutdowns for cleaning and flushing about every 1,500 to 2,000 hours. Control failures contributed to grid disconnects and forced outages. These findings were communicated to the technology development effort, where improved prototype designs were being tested.
For engineers and plant managers, the operational lesson is clear: reliability depends on the complete system, not only the electrochemical stack. Maintenance intervals, instrumentation, control behavior and the quality of the grid interface all affect how a generating unit performs in service.
Applying the lessons to project evaluation
A practical evaluation of onsite generation should therefore examine several dimensions together:
- Load match: Determine whether the unit will follow facility demand or operate at a set output alongside grid supply.
- Heat use: Identify actual thermal loads and whether recovered heat can be used consistently.
- Site integration: Account for installation work, location, electrical interconnection and supporting systems.
- Operating performance: Track availability, efficiency at partial load, run duration and forced outages.
- Maintenance needs: Include balance-of-plant equipment, controls and cleaning requirements in the service plan.
The historical field results do not, by themselves, establish the performance of later fuel cell designs or current commercial offerings. They do show the value of testing across real facilities and feeding operational evidence into design and commercialization work.
Conclusion
The 40-kW PAFC demonstration combined diverse sites, grid-connected and load-following operation, and a range of thermal applications. It met its operating-hour and availability goals, while also identifying gaps in heat output and recurring issues in coolant and control systems. For decision-makers assessing distributed power generation, the practical takeaway is to evaluate the integrated plant in its intended operating environment—and to treat installation, thermal use and maintainability as core parts of the system design.