A. The vexing history of nuclear reactor costs
The current cost controversy cannot be fully comprehended without placing it in the context of the history of reactor costs in the U.S. The cost of electricity generated by nuclear reactors in the United States has been a vexing problem for almost half a century.(*16) Touted as producing power that would be “too cheap to meter,”(*17) 240 reactors were ordered in about a decade from the late-1960s to the late-1970s.(*18) If all of the reactors had been completed on time, well over half of all power generated in the U.S. by the mid-1980s would have been from nuclear reactors.(*19)
Things did not work out that way. The “great bandwagon market” for nuclear reactors, as it came to be known, sputtered badly. Construction delays and cost overruns, as well as regulatory changes, drove the cost of reactors up dramatically.(*20) “Too cheap to meter” quickly became “too expensive to build.” More than half of all the orders for reactors were cancelled. Many of the projects had incurred significant costs,(*21) setting up lengthy fights over who should pay for facilities that were never used to supply electricity.(*22) The cost overruns were also reviewed in lengthy, contentious state regulatory prudence proceedings, where the failures of management to control costs and to provide power at as reasonable cost were investigated.(*23) As a result, no orders for nuclear reactors were placed in the U.S. after 1977. The last reactor brought on-line in the U.S. was completed in 1996. Construction on that reactor had begun in 1974.
The vexing nature of the cost of nuclear reactors has reemerged in what is now being called the “nuclear renaissance.” Less than a decade after the last reactor was brought on-line, nuclear reactors were back in the news and at the center of public policy debates with calls for large subsidies to promote nuclear technology. Along with a number of other factors, very low cost estimates put forward by the industry and academics and funded by the Department of Energy helped to create the illusion of a nuclear renaissance. Those studies certainly gave the Department of Energy an opportunity to broadcast headlines such as “University of Chicago: Nuclear Power Competitive with Coal & Natural Gas.”(*24) The initial cost projections, however, have not held up.
Much like the initial cost projections from the earlier round of nuclear reactors, projected costs escalated rapidly. By 2008, projected costs were three to four times higher than the initial cost projections in 2001-2004. Estimates that had put the cost of nuclear reactors as low as 6 cents per kilowatt hour (kWh) have been joined updated by estimates that put it as high as 30 cents.
Placing the ongoing conflict over projections of nuclear reactor costs in historic perspective takes on special importance. The management failure that Forbes refers to was much more than just the inability to execute massively complex construction projects. It was, first and foremost, a failure of analysis, a failure to distinguish hope and hype from reality.
For nearly a quarter of a century the theology of nuclear power – unchallenged and unchallengeable – was accepted by a variety of diverse interests to advance a variety of diverse causes. Rarely did those who seized on nuclear power as a means to their ends know its actual economic and technical status. Instead, the information available to them was part of a catechism whose basic function was to answer infidels and sustain the faith of the converted. The result, a circular flow of self-congratulatory claims, preserved the discrepancy between promise and performance.
Systematic confusion of expectation with fact, of hope with reality, has been the most characteristic feature of the entire 30-year effort to develop nuclear power.
The identification of promise with performance began in the United States. The economic “analyses” which controlled discussion during the critical early years of light water commercial sales had nothing to do with the detached confrontation of proposition with evidence which we think of as analysis. The public agencies with putative responsibility for facing the facts had neither the means nor the motivation to respond critically to the nuclear industry’s propaganda; they could only sanctify it. This they did with notable eagerness.(*25)
B. Too cheap to meter becomes too expensive to build
The rapid escalation of cost projections for new reactors in recent years raises major concerns, especially in light of the history of cost escalation in the nuclear industry. The last time the industry tried to ramp up production in the U.S., costs skyrocketed. From the mid-1960s to the mid-1970s, a small number of turnkey reactors were brought on-line. From the mid-1970s onward, more than 200 reactors were ordered, but half of them were never completed (see Figure IV-1).
Source: Completed: Koomey, Jonathan, and Nathan E. Hultman, “A Reactor Level Analysis of Busbar Costs for US Nuclear Plants, 1970-2005,” Energy Journal, 2007.; Cancelled: "Cancelled Nuclear Units Ordered in the U.S.", http://clonemaster.homestead.com/files/cancel.htm
The reactors that did make it on-line proved to be much more costly than originally projected. Figure IV-2 shows the increase in projected and actual costs by the date of commencement of construction for completed reactors, expressed as a percentage of the projected cost of the initial reactors. That is, Figure IV-2 uses the projected costs of the 1966-1967 reactors as the base and expresses all future projections and actual costs as a percentage of that base. This captures the fact that not only were projected costs increasing, but actual costs were increasing faster than projected costs.
The reactors commenced in 1966-1967 actually cost twice as much to build as originally estimated. The reactors commenced in 1968-1969 were projected to cost slightly more than the reactors commenced in 1966-1967, but they actually cost over three times as much as the projected costs of the reactors commenced in 1966-1967. Performance got worse, not better, over the decade.(*26)
Source: Energy Information Administration, 1996, "An Analysis of Nuclear Power Plant Construction Costs, January 1, 1986".
The learning that usually lowers initial costs has not generally occurred in the nuclear power business. Contrary to the industry’s own oft-repeated claims that reactor costs were “soon going to stabilize” and that “learning by doing” would soon produce cost declines just the opposite happened. The magnitude of cost underestimation was as large for reactors ordered in the early 1970s as it had been for much earlier commercial sales.(*27)
On average, the actual costs for each reactor were almost three times higher than the original projection for that reactor. The final cohort of reactors cost seven times as much as the projected cost of the original cohort. In short, the first round of nuclear reactors went quickly from being “too cheap to meter” to being “too costly to build.”
Source: Energy Information Administration, 1996, "An Analysis of Nuclear Power Plant Construction Costs, January 1, 1986".
Figure IV-3 overlays the recent cost projections on the historical pattern completed reactor costs. It uses the estimates from 2001 as the base and then expresses all subsequent estimates as a percentage of that base. For each of the two year cohorts the graph shows two projections, one based on the average of the mid-point estimates for all of the studies completed in that year; the other based on the average of all projections in that two year cohort. The initial 2001-2002 midpoint estimates averaged about $1,761 per kW. The initial 2001-2002 estimates for all projections were about $1,775 per kW. The midpoint and the all estimates track closely until 2009, when a number of high estimates pull the all estimate average up. The estimate based only on midpoints for 2009 was $6,500. The estimate based on all projections for 2009 was over $8,000. Interestingly, one of the high estimates for 2009 comes from an independent analyst and one comes from a utility. The increase in projected prices falls about half way between the projections from the 1960s and 1970s and the actual increases in that period.
C. The importance of construction periods
In the 1960s and 1970s, one of the major causes of the cost increases and missed projections was the inability of the industry to deliver reactors on time (see Figure IV-4). Large capital costs, sitting on the books, generated capital charges and a rate shock when the utilities finally finished the reactor. These charges cumulate, creating more and more expensive power.
Source: Nuclear Energy Economics and Policy Analysis, "The Effects of Inflation in Engineering Economic Studies," February 18, 2004.
By the end of the construction cycle that was started in the 1960s, the projected construction time increased by 50%, from just over 4 years to just over 6 years -- but actual construction periods were almost 10 years. In other words, actual construction time at the end of the cycle was more than twice as long as the original projection. The correlation between construction periods and overnight costs is strong for both completed reactors and projections for future reactor costs.
For the completed plants the length of the construction period explains just over half the variance in overnight cost projections. For the future projections, the length of the construction period explains almost two thirds of the variance in overnight cost projections.
We are now witnessing a dispute over the projected construction periods. Some analysts project construction periods of five or six years, while others project construction periods of ten years or more.(*28) Figure IV-7 shows the year-by-year construction expenditures in two recent studies with longer construction periods. Severance is for a two-unit project; Moody’s is for a single unit.
Source: Moody’s, "New Nuclear Generating Capacity: Potential Credit Implications for U.S. Investor Owned Utilities", May 2008. p. 8; Severance, Craig A., "Business Risks and Costs of New Nuclear Power", January 2, 2009. p. 35.
D. A range of cost estimates
Given this history, the initial low cost projections and their recent updates should be viewed with suspicion. Figure IV-8 shows the relationship between overnight and busbar costs for two different sets of cost estimates in the “nuclear renaissance” period. The bottom panel presents the estimates since 2008. The two low cost estimates can be readily explained. The CRS study relied on the utility overnight costs and then applied a utility finance model. The MIT II study is the update of the 2003 MIT study, which was optimistic then and remains so. Wall Street and independent analysts provide much higher estimates. The high outliers are from the Severance study. The exhibit also includes an estimate of busbar costs based on the CEC utility cost of generation model.
The relationship between overnight costs and busbar costs is predictable. The MIT and CRS estimates appear to be low both because the overnight estimates are low and because they translate overnight costs into busbar costs at a lower rate. With overnight costs of about $4,000, the busbar costs in the CEC model are about 12 cents per kWh. The MIT II and CRS costs are about 3.5 cents lower. Thus, 12 cents per kWh would appear to be a lower bound. The Moody’s estimate of about 15 cents is the midpoint. Harding’s 2009 mid-estimate is 17.3 cents. Several of Harding’s 2009 estimates are above 20 cents. Even adjusting for the unique costs that Severance includes, his estimates are above 20 cents as well. The range of reasonable estimates appears to be 12 cents to 20 cents, with a mid point of 16 cents.
The 1960s and 1970s may seem like ancient history, but the new proposed cohort of reactors could easily be afflicted with the same problems of delay and cost overruns. Inherent characteristics of large complex nuclear reactors make them prone to these problems. Reactor design is complex, site-specific, and non-standardized. In extremely large, complex projects that are dependent on sequential and complementary activities, delays tend to turn into interruptions. Inherent cost escalation afflicts mega projects, a category into which nuclear reactors certainly fall.(*29)
The endemic problems that afflict nuclear reactors take on particular importance in an industry in which the supply train is stretched thin. Material costs have been rising and skilled labor is in short supply. These one of a kind, specialized products have few suppliers. In some cases, there is only one potential supplier for critical parts. Any increase in demand sends prices skyrocketing. Any interruption or delay in delivery cannot be easily accommodated and ripples through the implementation of the project.(*30)
The severe difficulties of Finland’s Olkiluoto nuclear reactor being built by Areva SA, the French state-owned nuclear construction firm, provide a reminder of how these problems unfold.(*31) Touted as the turnkey project to replace the aging cohort of nuclear reactors, the project has fallen three years behind schedule and more than 50% over budget.(*32) The delay has caused the sponsors of the project to face the problem of purchasing expensive replacement power; the costs of which they are trying to recover from the reactor builder. The cost overruns and the cost of replacement power could more than double the cost of the reactor.(*33)
A description of the process by which the U.S. ended up with hundreds of reactors that were “too expensive to build,” written in 1978, before the accident at Three Mile Island changed the terrain of nuclear reactors in the U.S., bears an eerie resemblance to the past decade in the U.S.:
At the beginning of 1970, none of the plants ordered during the Great Bandwagon Market was yet operating in the United States. This meant that virtually all of the economic information about the status of light water reactors in the early 1970s was based upon expectation rather than actual experience. The distinction between cost records and cost estimation may seem obvious, but apparently it eluded many in government and industry for years…
In the first half of this crucial 10-year period, the buyers of nuclear power plants had to accept, more or less on faith, the seller’s claims about the economic performance of their product. Meanwhile, each additional buyer was cited by the reactor manufacturers as proof of the soundness of their product…The rush to nuclear power had become a self-sustaining process...
There were few, if any, credible challenges to this natural conclusion. Indeed, quite the contrary. Government officials regularly cited the nuclear industry’s analyses of light water plants as proof of the success of their own research and development policies. The industry, in turn, cited those same government statements as official confirmation. The result was a circular flow of mutually reinforcing assertion that apparently intoxicated both parties and inhibited normal commercial skepticism about advertisements which purported to be analyses. As intoxication with promises about light water reactors grew during the late 1960s and crossed national and even ideological boundaries, the distinction between promotional prospectus and critical evaluation become progressively more obscure.
From the available cost records about changing light water reactor capital costs, it is possible to show that on average, plants that entered operation in 1975 were about three times more costly in constant dollars than the early commercial plants competed five years earlier.(*34)
The similarities between the great bandwagon market and the nuclear renaissance, and the fact that utilities not only steadfastly refuse to accept the risk of cost overruns but also are demanding massive taxpayer and ratepayer subsidies to build the next generation of reactors, should give policy makers pause. The one major difference between the great bandwagon market and the nuclear renaissance is that there has been an extensive challenge to the extremely optimistic cost estimates of the early phase, a challenge from Wall Street and independent analysts. It may be impossible to escape the uncertainty of cost estimation, but it is possible to avoid past mistakes.
Reflecting the poor track record of the nuclear industry in the U.S., the debate over the economics of the nuclear renaissance is being carried out before substantial sums of money are spent. Unlike the 1960s and 1970s, when the vendors and government officials monopolized the preparation of cost analyses, today Wall Street and independent analysts have come forward with much higher estimates of the cost of new nuclear reactors. And, because the stranglehold of the vendors and utilities on analysis has been broken, the current debate includes a much wider range of options.
As important as bad analysis was, it might have had little impact if it had not been combined with another critical mistake. The nuclear reactor vendors had delivered a small number of reactors at fixed prices and eaten massive cost overruns. After a few loss leaders were delivered, they shifted tactics. Unwilling and unable to sustain those losses, as the Forbes article put it, the Great Bandwagon Market was impelled by evangelisms, optimism and seemingly irresistible economics… But the suppliers had learned their lesson. The new generation of plants would be built under reimbursable-cost-plus-fixed-fee contracts. Without that, the nuclear power program would probably have sputtered out in the mid-Seventies, when cost lurched out of control.(*35)
The contemporary policy debate takes the effort to insulate utilities from the high cost of nuclear reactors even farther. In addition to a broad range of general subsidies and the cost plus rate treatment, they are seeking large federal loan guarantees and treatment by state public utility commissions that would grant preapproval and recovery of construction costs.