Dwindling economic competitiveness has plagued the nuclear power industry for
some years. In the industry's early years, some reactors were completed for less than $100 million. Experience gained overseas (often in projects with American partners) provides sobering evidence that nuclear reactors can still be built at low cost in short periods of time. But not here in the United States, where rising construction and O&M costs have become industry bywords.
What lies behind this decline?
For the last 20 years, the U.S. nuclear industry has devoted the bulk of its energies largely to regulatory issues. We have not developed the types of economic analyses that would be taken for granted in the oil, securities, or automobile industries. True, certain notable power-plant improvements in recent years have trimmed fuel costs and boosted plant output. And other moves are now underway to reduce operating and maintenance (O&M) costs. But to prioritize and optimize these cost reductions, we need a better understanding of what has happened to costs in the U.S. nuclear industry.
In this article we present some cost findings drawn largely from publicly available data for 109 U.S. nuclear power reactors. The data sources include such agencies as the Nuclear Regulatory Commission, Energy Information Agency, Federal Energy Regulatory Commission (FERC), Utility Data Institute, Electric Utility Cost Group, and Institute for Nuclear Power Operations. The study applies multivariate, nonlinear regression analysis to O&M costs from 1975 to the present, and construction costs from 1966 through 1989.
Perhaps the most interesting finding concerns plant size. Rather than a single universe, with various cost trends showing positive or negative correlations to plant size, we found three separate universes that exhibit entirely different cost behaviors, each with its own "best" and "worst" of class.
Our study analyzes all of the U.S. nuclear power plants, ranging from the single-unit Ginna (New York) plant of 470 Mw(e) to the Palo Verde (Arizona) plant of 3 reactors of 1,270 Mw(e) each ("e" denotes electrical energy output, as opposed to thermal output). All of these findings are derived from analyzing the complete industry data itself, avoiding the historical sampling problems of analyzing subsets of the reactor data.
Instead, using data from every plant in the industry, our study isolates and precisely quantifies the factors that drive nuclear power plant operating costs and performance. It examines 29 different operating criteria in both costs and performance, including O&M, fuel, capital additions, capacity factors, thermal efficiencies, outage durations, labor requirements by major functional area, wages rates, and productivity. The study also analyzes safety-related criteria, such as SALP (systematic assessment of licensee performance) ratings and safety system failures.
Caveat: We do not claim any direct hands-on experience in operating nuclear power plants. Nevertheless, we see this lack as a distinct, dispassionate advantage. What the study has found are the true cost relationships derived from the actual operating data of all U.S. plants; these are not theoretical exercises, nor do they involve data sampling, so often abused in earlier studies. Moreover, the relationships are not complicated and, in fact, make intuitive sense.
Six key variables determine the behavior of all performance criteria for U.S. nuclear plants:
1) Time (specific year of operation)
2) Number of units
3) Reactor type (Pressurized [PWR] or Boiling Water [BWR])
4) Plant size in megawatts (design electric rating, or "DER")
5) Plant vintage (issue date for operating license, or "OL")
6) Regional location.
These variables allow one to construct analytical models that will describe the behavior of 90 to 95 percent of all operating plants, within a confidence interval of 6 to 8 percent. Moreover, the analytical technique allows an observer to isolate the impact of each variable from all others. This method gives a simple visual understanding of what is currently happening in the industry, not unlike operating control curves.
s In 1980, the year following the Three Mile Island (TMI) accident, O&M costs rose 55 percent for single-unit plants, but only 35 percent for dual units. Thus, "site-related" nonreactor costs were primarily impacted.
s Since TMI (1979), "reactor-related" costs (predominantly O&M) have doubled, while site-related costs (such as administration and support services) have increased 15-fold. This escalation in costs is especially striking in comparison with the French nuclear program.
s From 1987 to 1992, FERC operating costs at single-unit plants increased by 27 percent, while at dual-unit plants they escalated 44 percent (em primarily due to increased labor requirements. Thus, some of the historical cost advantages of dual-unit plants over single-unit plants are eroding.
s More recently, analyses show that FERC maintenance costs across the industry fell by almost 10 percent between 1993 and 1994. At the same time, FERC operations costs fell a mere 3 percent. However, comparisons to the French system suggest that the latter area offers the greatest potential for economic improvement.
2. Findings Related
to Number of Units
s Dual-unit plants still possess distinct economic advantages. On a per-unit basis, single units range from being 20 percent more expensive for maintenance, to 60 percent more expensive for plant administration. Other operating criteria lie between these two extremes.
3. Findings Related
to Reactor Type
s Over the years, BWRs have proven consistently more expensive to operate than PWRs. This cost disadvantage is gradually being reduced, but still ranges from 15 percent higher for O&M to 25 percent more for fuel costs.
As is always the case, we cannot say whether these differences are inherent, or driven by different regulatory treatment. An examination of foreign BWRs and PWRs might help answer some of these questions. For now, at least, we can quantify these differences and monitor their behavior over time.
4. Findings Related
to Unit Size
The single most important finding regarding size is that U.S. plants do not behave as one single population.
s There are three distinct groups of plant data: small (470-700 Mw), medium (780-970 Mw), and large (1040-1270 Mw). These three groups behave similarly but independently of each other.
In general, within each group, plants overlap one another and exhibit non-economies of size. Furthermore, the largest plants in a smaller size group cost more in absolute dollar terms than smaller plants in the next higher size range. This phenomenon is associated with every single one of the 29 performance criteria studied, with the exception of utility wage rates.
The capacity factor analysis shows that as plant size increases within each group, capacity factors fall proportionately. In other words, increasing plant capacity within any of the three groups will significantly increase the absolute cost of operations without increasing electrical output. Thus "stretching" plants within each group has proved very uneconomical (see sidebar).
5. Findings Related
Vintage, as measured by OL date, also shows somewhat unexpected behavior. Intuitively, one would assume that age would render a plant less efficient and more costly to operate (em like the family car. This is not the case.
s The most efficient plants came on line prior to 1974 and after 1982. Those plants that received OLs from the time the NRC was formed to a few years after TMI are appreciably more expensive to operate today than other plants.
s Difference in vintage accounts for as much as 20 percent in administrative costs, 75 percent in engineering costs, and more than 100 percent in safety system failures and capital additions (year in and year out).
The reasons behind these large percentage differences are not known. The heavy hand of regulation is a suspect, both during construction and in current operating requirements. Can this be mitigated over the remaining life of these plants? The industry would do well to study this area to seek generic solutions and possible regulatory relief.
6. Findings Related
Once the effects of the other variables such as size, type, and vintage have been accounted for, strong and consistent regional patterns become obvious.
s Regional differences vary by as much as 50 percent for total labor requirements, 80 percent for O&M costs, and 250 percent for capital additions, year in and year out.
Whether these patterns reflect natural characteristics such as climate, demographics such as urbanization, or different regional regulatory interpretations cannot yet be determined. The differences are both real and large. They are not due to wage rate differences, which are impressively consistent across the industry.
Plants in areas with high spending patterns could learn a great deal from plants located in the most efficient regions. Such knowledge would allow the development of realistic and achievable staffing standards for any plant, based on U.S. experience. However, U.S. plants would still need further improvements before they could compete successfully against foreign reactor programs.
* * *
Until recently, utilities did not possess a quantifiable basis on which to estimate operating costs or what energy output to expect. Our new findings cast an entirely new light on the U.S. nuclear industry. Nevertheless, these findings only show us "what" is happening to the industry in terms of cost behaviors. They do not explain "why."
The U.S. nuclear industry has survived an amazing gauntlet of problems unparalleled in the global experience within nuclear energy. We need to place into context some of the problems faced by the industry in relation to the findings from its own cost data.
This country boasts almost 2,000 reactor-years of commercial nuclear operating experience that could be leveraged to the benefit of all U.S. utilities. If nothing else, the industry can use its experience to quantify historical and current problems and highlight the areas of greatest economic potential.
Some costs are still too high and the ranges of cost and performance are too wide. Some are within the control of management; others are not. Our analyses can quantifiably make such distinctions and should be added to management's arsenal. t
Michael R. Fox has a PhD in physical chemistry from the University of Washington, and 29 years of experience in the nuclear energy field (em both at the Hanford plant and the Idaho National Engineering Laboratory. He is a member of the American Nuclear Society (ANS) and served as national chairman for the ANS public information committee from 1990 to 1992. Before coming to the United States in 1977, Jay Maidment worked for 20 years in Europe and Africa for a multinational mining-finance company. Throughout his professional career his main interest has been the application of operations research techniques to solve practical problems. For the past 15 years he has used his expertise to analyze and describe the economic behavior of the U.S. nuclear power-plant program from construction through operations. Maidment has a BS in physics and applied mathematics.
Stretching Reactor Capacity
"Stretching" describes the vendor practice of upgrading reactor capacity as more units are sold to subsequent clients. Our study found that stretching causes major economic variation across the whole spectrum of plants.
For instance, on a per-megawatt-installed basis, the size of a plant can account for as much as a 40-percent difference in materials costs, a 100-percent difference in engineering costs, and more than a 200-percent difference in safety system failures.
We have not tested whether the same effects occur if a utility simply upgrades an existing plant. However, our findings would allow for such a determination.
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.