April 01, 1998

WHICH NUCLEAR PLANTS WILL SURVIVE competition? To answer that question, senior managers at electric utilities must know a nuclear plant's true economic potential. Without an accurate understanding of operating economics, a utility might lose a good plant or waste resources on poorer plants that should be closed.

Of course, a shutdown may be appropriate at some plants (em perhaps a few situated in the most competitive regions, or others plagued by poor inherent physical characteristics. However, most U.S. nuclear plants show a significant potential for improvement in operating costs. That fact warrants a closer look (em otherwise, some owners could miss opportunities for improvement and close plants needlessly. Others might cut costs too aggressively or in the wrong areas, incurring expensive downtime because of mechanical or regulatory problems.

To identify opportunities to improve nuclear operations, we conducted a detailed analysis of more than 2,000 reactor-years of operating data. We culled this data both from sources in the public domain, such as U.S. Energy Information Administration and the Federal Energy Regulatory Commission's Form 1, and from privately gathered industry data. Our findings help explain why individual plants vary so dramatically in their economic performance. Our work should also offer managers the information they need to define the most efficient staffing levels at any nuclear power plant.

To be sure, one can improve performance in only two simple ways: boost output or cut costs. Nevertheless, this truism hides a wealth of insight. Our study of long-term cost and output trends reveals some interesting facts.

Industry Trends: Looking for Weak Spots

Figure 1 shows the annual average capacity factors for all U.S. plants since 1971. (Years when plants produced no output are excluded and adjustments were made for plants in their start-up years.) As the figure shows, capacity factors peaked just below 80 percent three times. The first peak occurred before the impact of the National Environmental Protection Act. The second appeared just before the accident at Three Mile Island. The third peak is occurring today. Thus, capacity factor improvement at many plants now promises limited economic potential.

Figure 2 shows (as indices) the four cost elements associated with plant operations, corrected for inflation. Two elements, fuel costs and capital additions, have been falling since the early 1980s. During the past 12 years, fuel costs have fallen by 30 percent; annual capital additions dropped by 60 percent. Both categories are now leveling off at essentially the same point as they were just before Three Mile Island. Thus, like the capacity factor, fuel costs and capital additions show little chance for any further industry-wide improvement.

This lack of opportunity shifts the focus to operation and maintenance (O&M) costs (other than fuel). After reaching a peak five years ago, O&M costs have decreased 10 percent. Nevertheless, maintenance costs are still more than twice what they were before the TMI incident. Operations costs appear almost four times higher. Experience gained in other countries suggests that U.S. nuclear plant maintenance costs are possibly 25 percent too high and operations costs are more than twice what they could be. If this premise is correct, O&M costs hold the promise for significant savings.

About 80 percent of O&M costs are labor-related; any reduction implies a drop in staffing levels. In fact, Figure 3 shows that staffing reductions have already begun. O&M plant labor (contractors plus utility employees) has fallen 10 percent over the past five years. The number of utility employees rose 20 percent over the same period, suggesting that utilities are replacing private contractors with payroll employees.

Nevertheless, this approach cannot continue indefinitely. For 20 years, contractors represented nearly 50 percent of the nuclear labor force. In the last five years, however, this ratio has dropped to 25 percent. Further reductions are possible but will be limited because some cyclical jobs, such as major maintenance and refueling outages, do not justify hiring permanent staff.

Utilities seeking to trim labor costs face the difficult task of cutting back their own employees. The big question is: What specific labor categories can be reduced without jeopardizing safety and efficiency? Too many engineers is inefficient; too few heightens the risk of downtime from mechanical or regulatory problems.

Plant Variations: Crucial to Analysis

The number of people required to operate a plant depends on the type of plant in question. The old custom of using raw industry averages, or even "peer group" averages, to ascertain correct staffing levels can lead to severe economic or regulatory problems. Furthermore, the current vogue of trying to emulate the practices of the economically most-efficient plants could put some plants in jeopardy. Trying to get a Geo Metro to travel as fast as a Corvette makes no sense; neither does expecting a Corvette to get the gas mileage of a Metro. This example may be extreme, but it does expose a flaw inherent in nuclear plant benchmarking.

What staffing levels are appropriate for which plants? Our most important finding indicates that the optimal number of employees is related to a plant's size and age and vintage and number and type of reactors. Furthermore, these relationships, which can be predicted to within plus or minus 5 percent, vary significantly among labor categories. Cost and staffing levels achieved by the most economic plants will not work for a plant with unfavorable characteristics; doing so will almost certainly lead to severe long-term problems. Even a plant with the best characteristics may not be able to achieve the economies of a plant near the top of the industry.

For many years, the industry assumed that because every plant in the U.S. was different, they could not be compared with one another. In recent years this bias has reversed. Now, the implicit assumption is that good management practices can somehow overcome all differences. The truth lies between these extremes. Certainly, management can significantly influence plant economics, but some plants have far greater potential for improvement than others. Which plants? What benchmarks should be used?

Unlike in France, no two U.S. plants are of identical size and vintage and have the same number and type of reactors. However, the wide variation in physical characteristics of U.S. plants makes it easier to conduct economic analysis. With the right mathematical techniques, one can isolate precisely the individual economic effects of plant size, vintage and geography. Once this breakdown is done, one may calculate the economic potential for a nuclear plant with any given set of characteristics. This approach is equally valid for costs, labor requirements, safety criteria and plant performance.

The analytical challenge comes in isolating the individual impact of several factors acting together. This task can prove especially difficult where the relationships cannot be approximated by simple, continuous, linear relationships (as with nuclear power plants). That explains why utility analysts using "Multiple Regression" programs have met with limited success in unraveling these relationships. More complex relationships call for a nonlinear, iterative form of "Variance Analysis," as employed here.

Potential Savings: Size, Vintage, Reactor Type

To illustrate how plant characteristics influence performance, we studied the behavior of total non-fuel costs (em O&M plus capital additions (em as a function of capacity factor, overall plant size and vintage, number and type of reactors (dual- or single-unit) and plant location. The analysis was based on individual annual costs of all 71 plants (106 reactors) operating from 1993-1996. Several surprising results emerged.

First, costs appear virtually unrelated to the level of power production (i.e., capacity factor). Except for the impact of price escalation, costs appear constant for any plant depending on its specific characteristics of size, vintage, etc. (Only fuel costs appear directly related to kilowatt-hours produced.)

Figure 4 isolates the impact on total plant costs of the size of the plant in megawatts. Note that for every $52 required to run a 450-MW plant, $130 is needed to run a 1,200 MW plant. However, plants are not members of a single population but fall into one of three size categories: small plants (425-700 MW), medium plants (775-975 MW) and large plants (1,025-1,275 MW). Ninety-five percent of all U.S. plants fall within one of the three tight envelopes shown in the figure.

Rather surprisingly, the operation of U.S. plants shows no economies of scale; the three populations follow closely the line representing a constant dollar/MW relationship. Furthermore, there is an additional 10-percent economic deterioration within each group due to size. When the three cost elements are analyzed separately, it is predominantly the behavior of maintenance and capital additions costs that cause this phenomenon. This cost-size relationship has proven consistent for many years and is therefore unlikely to be easily changed.

Figure 5 shows the impact on total costs due solely to plant vintage, where vintage is defined as the date the plant received its full-power operating license. Under normal circumstances, one might expect to see a slight increase in costs as a plant ages, reflecting higher maintenance needs. As can be seen, however, that is not true here. Instead, "middle-aged" plants appear abnormally expensive. Older plants show a lower cost; plants placed on line most recently now approach the lower cost levels achieved by the older group. What does this curve indicate?

Nuclear plants can be analyzed as falling into one of three vintage groups. The first group contains plants licensed by the Atomic Energy Commission before the formation of the Nuclear Regulatory Commission in 1974. The second group includes plants whose construction was disrupted by the ever-changing regulatory requirements imposed after passage of the NEPA, and following the Browns Ferry fire and the TMI event. The last group includes those plants that came on line during the progressively more stable regulatory environment that prevailed after 1982.

Like plant size, 95 percent of all U.S. plants fall into one of the three intervals shown in the figure. The 15-30 percent cost increase for plants that came on line between 1975 and 1982 is due primarily to higher maintenance costs and capital additions. Since plants of newer vintage exhibit cost characteristics more closely akin to the older group, the relationship between cost and vintage can be seen as remaining relatively constant. The cost variation must be due far more to issues associated with the time when a plant was built than to the plant's chronological age.

As one would expect, dual-unit plants (plants with two reactors placed in close proximity) appear consistently more economic than single-unit plants. However, since the late 1980s, there has been a steady deterioration in the economics of dual-unit plants: They were once 30 percent cheaper per unit than singles, now they appear to be only 15 percent cheaper. Also, plants with pressurized water reactors (PWRs) are consistently more economic than those with boiling water reactors (BWRs). This advantage currently ranges from operations costs that are 15 percent lower, to requirements for annual capital some 30-percent lower than for singe-unit plants.

Finally, some small savings are evident at those utilities that operate more than one plant.

Geography and Staffing Levels

Today, the annual cost of operating a nuclear plant in the United States (excluding fuel and amortization) ranges between $50 million and $250 million. While the four physical plant characteristics already discussed explain 80 percent of this variation, another major factor exists: the region in which the plant is located. Understanding this influence offers potential for significantly improving the overall economics of U.S. plants. For instance, staffing levels in the most efficient parts of the country should act as benchmarks for the rest.

Figure 6 shows the residual impact on plant costs once the effects of plant size, age and number and type of reactors have been removed. The most expensive plants are located in the Southwest and Northeast, which is not surprising since, overall, costs are higher in these areas. What is interesting is the degree to which they vary.

For example, operating a plant in California costs almost twice as much as an identical plant would cost in the Carolinas. This result is not due to differences in wage rates. Analyses show that utility wage rates vary by less than 15 percent across the country. Thus, an identical plant in California requires approximately 50 percent more people to operate it than in the Carolinas. Presumably both plants are equally safe, so why are the additional people needed?

How much of these regional variations are due to local regulatory requirements, union labor practices or the cost of other forms of electricity? Nuclear power plant managers should consider these questions, particularly those now facing deregulation at their utilities. However, just comparing oneself to the economically most efficient plants gives a distorted picture. Not surprisingly the most efficient plants are dual-unit PWRs built before 1974 or after 1984, with capacities at the lower end of each size group. For a utility owning a single-unit BWR, comparisons appear uninformative.

Bonneville Power Administration used this analysis to decide whether to shut its

Washington Public Power Supply System's WNP-2 plant (BPA pays 100 percent of the costs of this plant and receives all output). To identify possible improvements, 30 different cost, labor, safety and performance criteria were analyzed. Over the past three years, this work helped avoid premature shutdown and cut operating costs 35 percent and staffing 25 percent.

Other utilities interested in safely and efficiently reducing their costs should consider a similar approach to the one described here. This analysis can lead to realistic improvements in cost and output while minimizing the potential for long-term regulatory or performance problems.

Finally, from an industry perspective, we now can determine which plants perform "best." From these plants much can be learned. If economic competitiveness cannot be accomplished quickly, early retirements of managers, staff and nuclear power plants will rapidly follow after deregulation.

Jay Maidment is an operations research consultant with broad international experience. Geoffrey Rothwell, Ph.D., is a senior research associate with the Department of Economics at Stanford University and chairs the committee on "Evaluation of Plant Technical Performance" of the Nuclear Power Division of the International Atomic Energy Agency. Both have been associated with U.S. nuclear power for almost 20 years.

Articles found on this page are available to Internet subscribers only. For more information about obtaining a username and password, please call our Customer Service Department at 1-800-368-5001.