July-September 1998 Issue


Reducing Downtime in Nuclear Power Plants

oday, the typical US nuclear power plant spends almost two out of every 18 months shut down for refueling. As owners of such plants face new competition for customers, they are looking for ways to reduce costs; and refueling less often is one option. Working closely with power plant operators, Energy Laboratory researchers have designed reactor cores and operating procedures that would enable power plants to run for up to about four years before needing to refuel. Because of the extra cost of the necessary enriched fuel, adopting a four-year "extended operating cycle" under today's economic conditions would be cost-effective at plants that now experience relatively long downtimes for refueling and forced shutdowns but not at plants that operate more efficiently. A three-year operating cycle requiring less highly enriched fuel would bring savings at many more plants. And if laser-based technology now being developed reduces the cost of enriched uranium, the economics of the extended cycles would improve significantly. Perhaps most important, the MIT team identified strategies that plant operators can use to reduce forced shutdowns and to perform more maintenance procedures while their plants are on-line. The researchers emphasize that any reduction in downtime will not only reduce costs but also prevent possible long-term damage to plants caused by repeated stopping and starting.

To improve their position in the now-competitive electricity market, owners of nuclear power plants would like to reduce the amount of time their plants are not producing power. One reason for such idleness is the need to refuel. Every 18 to 24 months, plants are typically shut down while their operators put in new fuel and perform a variety of maintenance and testing tasks. Such "refueling outages" can mean a month or two of downtime and a serious loss of revenues. The nuclear power company has no electricity to sell. Moreover, to satisfy its customers, it must buy "replacement power" from other electricity generators. If plants could shut down for refueling less often or for a shorter time, the savings could be significant.

For the past three years, an Energy Laboratory team led by Professor Neil E. Todreas has been looking at the technical feasibility and the economic effects of refueling less often. Faculty participants in the study included Professors Michael J. Driscoll, Michael W. Golay, and John E. Meyer. Eleven graduate students contributed to the integrated analysis, as did personnel from Boston Edison Company, the Idaho National Engineering and Environmental Laboratory, PECO Energy, North Atlantic Energy Services Corporation, Studsvik of America, and Yankee Atomic Electric Company (now Duke Engineering & Services). Faculty and students alike went to nuclear power plants to work on-site with utility managers and plant operators.

The potential economic benefit from refueling less often is clear. Suppose a plant shuts down once a year for 40 days of refueling activities. If instead it spends 40 days refueling only once every two years, the average annual downtime for refueling drops to 20 days. Average annual income would increase, everything else being equal.

But everything else is not equal. Perhaps most important, to operate longer without refueling, a plant must be equipped with fuel (uranium) that is more highly enriched than is conventional fuel. Enriching uranium involves increasing the amount of the fissile uranium-235 isotope that naturally occurs. The more highly enriched the fuel, the more costly it is. The economic viability of the extended operating cycle thus depends on whether the savings from postponing refueling offset the added cost of the highly enriched fuel plus the cost of any other technical changes required.

The MIT team's first concern was whether refueling less often would be technically feasible and, if so, what changes would be required. As an example, they considered extending the typical 18-month operating cycle for pressurized water reactors (PWRs) and the increasingly popular 24-month cycle for boiling water reactors (BWRs) to 48 months. They addressed three key questions. Is it possible to design a core that can run that long without replenishing the fuel? Can the testing and maintenance activities now performed during refueling be delayed or performed while the plant is on-line? And can a plant operate that much longer without more and more shutdowns as stressed equipment begins to fail?

The first challenge was to design a reactor core that could generate the required level of electricity for four calendar years and could handle the more highly enriched fuel safely. (The term "calendar years" rather than "operating years" assumes that the plant will sometimes be shut down or operated at less than full power.) Using state-of-the-art computer codes, Professor Todreas and his team designed new cores suited for use in the typical PWR and BWR. Analyses of neutronic behavior and thermal and mechanical fuel performance confirmed that both designs could operate for close to four calendar years without renewing the fuel.

For successful operation, the new cores require fuel that contains more than 7% uranium-235 (U-235) by weight--two to three times more than is contained in the fuel used now. The current regulatory limit for commercial uranium is 5% U-235 enriched. Military and research reactors use uranium that is far more highly enriched, so making the fuel is technically feasible. Commercial use of 7% U-235 fuel would raise regulatory issues and would require changes in procedures and facilities for fuel fabrication, transportation, storage, and disposal. However, the technology to make those changes is proved and available.

Professor Todreas and his colleagues next considered the feasibility of postponing the testing and maintenance procedures that are now typically performed every 18 or 24 months during refueling outages. Working closely with power plant engineers, they analyzed the "surveillance" programs of two typical nuclear plants, one a PWR and the other a BWR. They reviewed the roughly 4000 surveillance activities that are now performed, from checking valve and containment leak rates and instrument calibrations to testing standby safety equipment to make sure it will operate if necessary.

Analysis of those tasks confirmed that many of them could be delayed until the 48-month refueling. Today's more frequent schedule has arisen simply because the plant is shut down for refueling anyway. Other tasks could be performed on-line--a practice that is already growing in the industry and is preferable for the many standby safety systems that should be available during outages. Only about 1% of all the surveillance activities for the PWR and 4% for the BWR could not be delayed or performed on-line. The MIT team and plant personnel outlined possible engineering solutions for most of those activities, for example, by duplicating certain critical systems so that one could run while the other shuts down for maintenance.

The final technical task was to consider how adopting the extended cycle might affect the forced outage rate (the percentage of time a plant is down due to equipment failure and other problems). Most US nuclear plants experience forced outages. "Record runs" up to 20 months have occurred--but rarely. Certain equipment has been designed for a limited lifetime, so running a plant nonstop for 48 months could well cause the rate of forced outages to increase. Steps must be taken to prevent that outcome.

To identify the common causes of forced outages in today's plants, the researchers examined the US Nuclear Regulatory Commission's database of PWR and BWR plant operation between 1989 and 1995 as well as records from two specific BWR plants run by PECO Energy. As expected, the results showed that most problems were caused by failures not in the nuclear reactor and safety-related systems but in other systems, such as the turbine and its control system and the feedwater system. The research team described strategies that would enable plant operators and designers to reduce such failures.

Very few data exist on failures that occur after about 20 months. Therefore, the researchers recommend that plant operators undertake "age exploration," a procedure now under way at nuclear installations run by the US Navy. After upgrading all the components and systems that have known problems, operators run their plant--with appropriate monitoring and safety shutdown systems in place--until something fails. They then repair and upgrade the failed equipment and resume operation. In time, the weak links in the plant will be gone. The up-front cost of this procedure will be appreciable, but the long-term payoff will be a significant decrease in the forced outage rate.

The researchers' conclusions suggest that a four-year operating cycle is technically possible. However, the economic benefits proved less straightforward than expected. The technical analysis suggests that the switch to the longer operating cycle would be accompanied by changes other than just the added fuel cost. In particular, for the longer cycle to be effective, forced outages must not become more frequent. Plant operators would have to ensure that their plants were extremely well run and well maintained--a task with an unexpected impact. In such a smoothly running plant, refueling could no doubt be accomplished more quickly. With the refueling time shorter, buying more costly uranium to refuel less often might not provide a net economic gain.

To assess the economics of switching to the longer operating cycle, the researchers developed an economic model incorporating all the key factors. They made realistic assumptions for the costs of enriched uranium, replacement power, material and personnel during outages, and so on. They then examined how the net annual cost would vary with the frequency of refueling, the time required to refuel, and the forced outage rate.

Sample results for the PWR are shown in the figure below. The horizontal axis shows the number of days spent refueling; the vertical axis shows net annual cost. The two dashed lines present results assuming an 18-month operating cycle at forced outage rates of 6% and 2%. The two solid lines present results for an extended cycle of 41 months (a cycle the researchers found technically feasible for PWRs) and the same two outage rates. The results show that, regardless of the forced outage rate, it costs more to refuel every 41 months rather than every 18 months if the time required for refueling is shorter than about 65 days.

Based on such data, the economics of the extended cycle for today's average plant look disappointing. As a reference case representing today's conditions, the researchers assumed an 18-month operating cycle, a refueling time of 49 days, and a forced outage rate of 6%. They then considered a switch to the extended cycle along with a reduction of the refueling time to 42 days and of the forced outage rate to 3%--improved operating conditions that the US fleet of plants is likely to achieve. Such a change would decrease costs at the average PWR only slightly and would actually increase costs at the average BWR.

The researchers also examined a more modest 36-month operating cycle for the PWR, and projected savings proved greater. Such a cycle would require less fuel enrichment. Analyses showed that the 36-month cycle with the improved operating conditions defined above would be $11 million per year less expensive than the 18-month reference cycle without the improved operating conditions (estimated at $110 million per year). If, however, the 18-month cycle is accompanied by the improved operating conditions, it breaks even with the 36-month cycle. Overall, the researchers conclude that--under present economic conditions--the best means of lowering costs is to stick with the standard operating cycle and minimize both the time needed to refuel and the forced outage rate.

Several external changes could alter the results of the MIT analyses. For example, new technologies such as laser isotope enrichment could reduce the cost of highly enriched uranium. The extended cycle would then become more cost-effective--perhaps even preferable. On the other hand, competition in the new regulatory environment could cause the cost of replacement power to drop, lowering the cost of frequent refuelings.

The study has provided a variety of useful results. It has produced analytical tools that can help plant operators assess what operating cycle is best for their specific situation. It has identified ways for today's operators to reduce the frequency of forced outages and to accomplish many of their surveillance activities on-line. And it has focused new attention on the noneconomic impacts of shutting down nuclear plants. Avoiding forced and refueling outages is still not a top priority for many US plants. Operators assume that forced outages will occur periodically, and they use them as a chance to inspect their plants from top to bottom. Yet outages themselves are known to create technical problems. Operating records show that forced outage rates are typically high for several months after plants have shut down and restarted. Perhaps more important, the process of cooling down and heating up a plant, draining and refilling certain systems, and so on can adversely affect both materials and chemicals inside the plant. Temperature changes are made slowly, and other steps are taken to minimize damage. But no one knows the engineering consequences of such practices over the long term.

Professor Todreas believes that clarifying those long-term consequences should be a top priority for the nuclear power industry. If the long-term impacts of shutdowns are better understood and if technical improvements make forced outages less common, the nuclear industry may begin to place a higher value on less frequent refuelings and on uninterrupted operation in general--an attitude that Professor Todreas believes would improve both the near-term economics and the long-term engineering performance of today's nuclear power plants.

Neil E. Todreas is KEPCO Professor of Nuclear Engineering; Michael J. Driscoll is professor of nuclear engineering, emeritus; and Michael W. Golay and John E. Meyer are professors of nuclear engineering. This research was funded by the Idaho National Engineering and Environmental Laboratory University Research Consortium. Further information can be found in references.