Nuclear Plant Costs – A Look Back and Ahead

TVA's Browns Ferry Nuclear Plant under construction; photo from TVA brochure, Will Davis collection.

TVA’s Browns Ferry Nuclear Plant under construction (Photo from TVA brochure, Will Davis collection)

by Will Davis

Reports of escalating costs for some nuclear plants under construction around the world, while the costs for other plants have not, have led to a call for an examination of the historic trends of nuclear plant construction project costs. This new interest has led to at least one significant new paper. This retrospective introduces the topic to those unfamiliar with it, and shows lessons learned that the industry now incorporates in building nuclear plants.

The beginning, and building a nuclear plant

The earliest nuclear plants in the United States were built with (normally) considerable U.S. Atomic Energy Commission (AEC) assistance, either in waived fuel costs or with the AEC paying for development and/or owning the reactor. There were also a number of so-called “turn-key” projects wherein the vendor of the nuclear steam supply system (NSSS) acted as primary contractor to the buyer (a utility) and provided the entire nuclear plant at a flat, agreed price. In the cases of the turn-key plants, any potential cost overruns were absorbed by the vendor and its contractors. The idea was to get plants built and operating for the purpose of demonstrating their merits (and of course, for the vendors, disclosing any areas in which future plants could be improved).

This process was not considered permanent, and when enough time had passed that utilities became confident and the estimates of the profitability of nuclear-generated electricity were coming in, they began ordering nuclear power plants on their own. (At that time, nuclear plant annual fuel costs were less than half that of coal, measured in cents per kilowatt-hour.) This required a utility to enter into complicated contractual arrangements with an NSSS vendor, an A-E (or architect-engineer) firm to design the plant, and someone to act as constructor. There was no fixed arrangement for these agreements. In some cases, the A-E was expected to oversee the constructor, while in other cases the utility did; in yet others, the utility did not hire a general construction contractor and entered into multiple contracts with various companies to build the plant. In a very different mode, there were utilities that traditionally designed and built their own power plants (TVA and Duke Power), and in the nuclear era they did just that, acting as their own A-E and constructor. Still others, such as Detroit Edison, had enough experience that they could act as their own A-E but hired an outside firm from time to time to provide a check of their process to assure the rate payers (customers) that the project was being well-managed. The variability of the execution of such arrangements on all of the nuclear plants ordered in the United States (whether completed or not) and the ramifications of such variations could (and perhaps should) fill a small book. It should also be noted that when projects were going poorly, these contractual arrangements were sometimes altered mid-project, leading to further delays as newly involved firms came up to speed on new responsibilities.

Construction of any facility that needs tens of thousands of tons of steel, concrete, and other materials built in highly complex and interrelated systems in a rather dense footprint is difficult, and runs the risk of cost overrun if delayed. The type of project, whether municipal stadium or interstate highway or power plant, is irrelevant. Nuclear plants were sometimes singled out as particularly susceptible to overruns due to the added quality assurance and complexity issues not seen in some other large infrastructure projects. When projects were not well managed, they fell behind and cost their owners more than planned or expected. As orders for nuclear plants began to swell in the mid- to late-1960s, it also became apparent that manufacturing of key, large components could not quite keep up with the order volume, and that skilled and capable craftsmen/tradespeople were not available all the time. These factors led to delays. Since the utilities were financing the debt they had taken on in order to obtain working capital to build the plants, the cost went up as the time before the plant would be complete and making money slipped further and further into the future. Delay was, in fact, a critical element in increasing nuclear plant costs, after the earliest plants were completed.

Early estimates of escalating cost

The AEC published WASH-1082, “Current Status & Future Technical & Economic Potential of Light Water Reactors”¹ in March 1968—years before the Calvert Cliffs decision that would force nuclear plant owners to develop complicated and expensive Environmental Impact Statements, and a decade before the Three Mile Island accident, which led to an incredible increase in regulatory requirements. Even in that early document, nuclear plant costs were seen to be increasing—but some of these increases were being felt in any kind of power plant, be it nuclear, coal, or oil. WASH-1082 tells us that between March 1967 and March 1968, nuclear plant costs had increased “on the order of 15-20 percent” based on the following factors:

• Escalation of labor and materials (approximately 6 percent)

• Increase in cost of money and projected lengthening of construction schedules

• Added cost of structures and systems associated with the evolutionary interpretation of safety requirements

• Increase in contingency allowances in anticipation of design changes resulting from safeguards interpretation leading to alteration of and/or addition to systems and structures significantly advanced in construction

This last factor—changes made while the plant is well along, or which affect structures or systems already installed—became a major sticking point. While “moving the goal posts” was a common complaint about regulatory-derived field changes, a study done by Theodore Barry in the early 1970s² showed that in the period before TMI only about 12 percent of change notices were derived from regulatory requirements. In many cases, difficulties with project management caused changes to have to occur, as plant construction was not well coordinated between subcontractors and trades. In many cases, the lack of skilled trades, which were in very high demand, led to inspection failures, and work already performed had to be redone. While there were some instances where late delivery of large components (reactor vessel, turbine generator, steam generators) delayed a project, this was statistically unusual (especially as import of such components increased and domestic capacity stepped up) and delays that drove up costs were more often related to other factors, as we will see.

This illustration shows the average expected schedule for nuclear plant construction in 1971; it is taken from an original copy of US AEC WASH-1174-71, "The Nuclear Industry 1971."  The 84 month point at far left is the decision to build a nuclear unit.  The volume notes that this schedule was a composite from many industry sources, and that individual sites continued to vary.

This illustration shows the average expected schedule for nuclear plant construction in 1971. It is taken from an original copy of US AEC WASH-1174-71, “The Nuclear Industry 1971.” The 84-month point at far left is the decision to build a nuclear unit. The volume notes that this schedule was a composite from many industry sources, and that individual sites continued to vary. This is referred to as a “72-month schedule.” Many nuclear plants exceeded this, but many others beat it. It is significant to note that this schedule does not include development of an Environmental Impact Statement, which would soon be mandated. The “Calvert Cliffs” decision forcing this immediately put about 50 new large sets of paperwork on the AEC’s review table and seriously delayed review of all aspects of new nuclear plant construction for some time.

Lessons learned from the first nuclear era

Glenn Williams, who worked for two different A-E firms during the first nuclear build, exclusively provided ANS Nuclear Cafe with reflections on the causes of increased nuclear plant costs in the United States, and delay in building nuclear plants. We present his discussion in his own words (in italics):

On Cost Escalation: In all of the plants in which I worked, the cost issues were largely the same. The two biggest cost drivers were either schedule delays or rework.

Time is money. Delaying critical path activities cost owners additional interest, overhead expenses, and lost revenues. Delays also exposed the project to unforeseen escalation and maintenance costs.

Rework frequently caused triple damages. The first damage was the original labor, materials, and schedule costs that were planned and sequenced for the original work. That investment was lost when management decided rework was required. The second damage was unplanned labor, materials, and schedule costs associated with removal of defective work. The third damage was the labor, materials, and schedule costs required to replace the original work, which was out of sequence, involving difficult conditions and using inefficient procedures. Rework frequently involved critical path construction, which meant that rework delayed the plant’s commercial operation.

On Contracts: I had the rare opportunity to work on a two-unit nuclear construction project that relied on fixed-price contracts. It was a disaster. Contractors loved the fixed price environment; they waited for scope changes and hammered the owners when any change request was needed. In a typical month, there were hundreds of pending change requests, which seemed to compound. If approved, one change request created more change requests. In the end, owners were forced to pay contractors hundreds of millions of dollars to buy out their fixed price contracts and convert them to cost-plus contracts.

That experience taught us that cost-plus construction was the prudent approach. It was particularly appropriate for “first of a kind” construction.

Lessons Learned: The first lesson was about planning. There was a tendency for managers to confuse planning with scheduling. Consequently, managers would publish schedules without reference to a road map. Without a detailed road map, any path would guarantee that the owners would be confronted with schedule delays and cost overruns.

The second lesson was about integration. Engineering is accomplished in a logical sequence. Construction work requires a very different sequence. Startup requires yet a third sequence. If the integration between the three phases is not carefully managed, schedules blow up and the owners are faced with cost overruns.

It should be kept in mind that if engineering is not 100 percent complete, construction planning is impossible. Also, a small engineering change can create a tidal wave in construction and startup.

This leads to the third lesson: Everything is integrated. Project management is about scope, schedule, and cost. Change any one of these and the other two will change. As one of my old colleagues once said, “Good, fast, cheap—pick two. You can never have all three.”

The cost of nuclear plants—A new paper, a new look

Jessica Lovering of The Breakthrough Institute decided, as she tells ANS Nuclear Cafe, that the time was right to do a new study on the historic costs of nuclear plants. She found that the majority of anti-nuclear cost-based arguments were built entirely on the U.S. nuclear plant cost experience of the 1970s. “We heard hints that current costs were lower in Asia, but we wanted reliable cost data that we could analyze to understand the differences.” As a result, the study includes most nations with large nuclear builds. A great deal of digging was required to get some of the data, she said.

Lovering said that she found two major surprises when researching the global, historic costs of nuclear energy. First was that every country experienced lowering costs in the early years of nuclear plant construction. Second was that South Korea continues to experience reducing costs, even now. She attributes South Korea’s continued cost reduction in part to the focus on standardized (in fact, duplicate) nuclear plants being built at various locations. (This was realized and implemented in the United States, in the SNUPPS program and also in Duke Power’s “Project 81″ program.)

When asked what the major idea was that she wanted readers to take away from the new study, Lovering told us that she wanted everyone to realize that every country had a different cost experience. “There’s nothing intrinsic to nuclear technology that makes it rise in costs. It’s simply due to the industrial policies and market dynamics in each country at the time.” The paper is available free of charge (see below).

Click here to access the paper “Historical Construction Costs of Global Nuclear Power Reactors.”


(1) “Current Status & Future Technical & Economic Potential of Light Water Reactors.”  WASH-1082.  US Atomic Energy Commission, March 1968.

(2) “Energy Northwest—A History of the Washington Public Power Supply System.” Gary K. Miller.  Energy Northwest 2001.


Will DavisWill Davis is Communications Director and board member for the N/S Savannah Association, Inc. He is a consultant to the Global America Business Institute, a contributing author for Fuel Cycle Week, and he writes his own popular blog Atomic Power Review. Davis is also a consultant and writer for the American Nuclear Society, and serves on the ANS Communications Committee and will serve on the Book Publishing Committee beginning in June. He is a former US Navy reactor operator, qualified on S8G and S5W plants

13 thoughts on “Nuclear Plant Costs – A Look Back and Ahead

  1. George W. Hinman

    I directed a study for the Washington legislature on the cost escalations in construction of the nuclear plants undertaken by the Washington Public Power Supply System, especially plants four and five. We concluded that fast track construction, i.e. undertaking construction before engineering design was complete,or at least almost complete, was a major factor in the delays and cost escalations at the WPPSS undertaking. Interferences encountered during construction led to diversion of engineering to redesign and away from advanced design. The result was delays and cost increases, especially when construction caught up to design and workers had to wait with nothing to do. You pointed out in your write-up that design had to be complete before construction began. Fast track construction should be avoided on first-of-kind complex projects, but it is very difficult to convince the construction firms of that principle. Many of your points are well-taken as well. Only on one the five WPPSS plants was completed.

  2. Will Davis

    Thanks for your comment, George! Believe it or not I am sure I recognize your name, as I have just finished reading “Energy Northwest” by Gary K. Miller and I believe you and your study are mentioned in that book. Certainly the WPPSS experience informed some of what went into this article, although as you well know WPPSS was somewhat of a unique and extreme example. The mention of changing A-E’s in the middle certainly would have WPPSS as a prime example. Thanks again for reading; I really appreciate it.

  3. Ivan Grosz

    I was in charge with St. Lucie 1, mechanical-nuclear design and involved in other nuclear plant designs.
    When deliveries are on schedule and suitable personal is available in the key compartments, overruns are caused because of interface problems, if not between engineering compartments then between engineering and construction and changes resulting from enhanced safety criteria. Competitive bidding with unscrupulous plant suppliers leads to less than accurate cost projections.

  4. Dennis Mosebey

    It is incorrect that standardization occurred in SNUPPS. It started out that way but as time went on and even to this moment Wolf Creek and Callaway do not maintain identical design especially in non nuclear systems like Circ and Service Water. Callaway has a tower, Wolf Creek made a lake. Even the standardization that did occur did not hold cost down. Wolf Creek started out at 700 million and ended up at just over 2.2 billion! It was 9 years from start to commercial operation and every one of em was a struggle with questions first on Containment Basemat, then TMI backfit and all this with very little anti nuke activity. So the reference to SNUPPS in Jessica’s article should only be taken in context of concept and not actual.

  5. Ivan Grosz

    If I may add to my previous comments. Construction permit slippage turned out to be a major cause of cost overruns. The US permitting process is probably more open ended than in some of the other countries.

  6. Will Davis

    Well, and I’m not speaking for Jessica here, but the point of SNUPPS was to take full advantage of the AEC push for replication or duplication to speed licensing of a project, so that what was duplicated in this project was the power block. It was known from the beginning that the individual owners’ projects would incorporate site-specific changes such as you mention (cooling towers, cooling ponds, service water to body of water, or combination.)

    The SNUPPS plants incurred the same sort of overruns due to regulatory change that hit every other plant at the time – none was immune and some handled it better than others. Just about every plant at that time incurred an increase such as you describe. The plan was intended to decrease the unit cost to any participant by about 10 per cent, or about $100 million on a $1 billion nuclear unit. Even, as I reported some time back, when the cost of Wolf Creek had gone up to $2.9 billion the savings from the SNUPPS program were estimated at $214 million (these are 1985 numbers.) This is not what we might call an enormous saving, but it makes a difference and to some utilties’ boards of directors made sense. I imagine the savings might have been larger than they were had most of the SNUPPS units not been cancelled.

    Sherrell Greene, a consultant who I respect, has said before that “it’s easy to build nuclear plants identical to each other but very very hard to keep them that way for any time,” pointing up the further complication that the dedication to duplication must remain in the long haul to continue saving money from the concept.

  7. Will Davis

    Ivan, thanks for (both of) your comments! They back up what we’ve written and are very helpful. In reading about WPPSS I did see an echo of what you wrote — in other words, that competitive bidding was used to win a contract by lowballing and then money was made back up in droves when the change orders came in. And as to the permit slippage — it would seem more than clear that the Calvert Cliffs decision seriously affected a majority of nuclear plants in this country that were completed, and almost surely had an effect monetarily on many that were cancelled. That review slippage appears pivotal, historically, coming as it did at a time when the power demand forecasts began to level and then tank. Thanks for commenting!

  8. Leslie Corrice

    I personally experienced the massive cost impact of nuclear regulation during my tenure with Perry Nuclear Plant. When TMI happened, the NRC placed a moratorium on nuclear construction, including projects already under way (as was ours). By the time the halt was lifted about two years later, the cost of Perry unit #1 more than doubled and would eventually more than triple! Safety improvements mandated during the delay were almost entirely hypothetical, based on an incident in Pennsylvania that harmed no-one, and never will.

  9. Colin Megson

    Strange how it is accepted even in pro-nuclear circles that nuclear power is expensive.

    What do we pay for – kWh! So a simple bit of arithmetic shows just how much electricity a nuclear plant delivers, compared to electricity from wind turbines.

    This is the kind of information which needs to be presented to politicians, the media and the public at large – not a back-to-the-wall defence of high costs by blaming safety, waste or accident escalators.

    This is one of the blog posts I’ve done in attempting to get this message out:

  10. Jim Hopf

    Sorry for the delayed comment/response.

    One general reaction I have to all the above information about contractual and management problems is that it seems to be an argument for SMRs. Assembly line construction of a product with a large amount of forward book order would seem to ameliorate most of those problems. The assembly line would make the modules at a steady pace, with a dedicated, highly experienced work force that is (merely) repeating a fully developed process.

    My other reaction to all the discussed management and contractual problems is to wonder exactly how or why nuclear is so much different than the other energy sources (e.g., gas turbines) that it is in competition with. Do they not have many of these same problems? If not, why not? Is there something unique to nuclear that makes it more vulnerable to these problems? And what, exactly, was different about the other nations that didn’t experience cost escalations?

    As you may be well aware, I tend to blame ever-ratcheting regulations and requirements, along with uniquely onerous component fab QA requirements (NQA-1) for most of these problems.

    The article often mentions rework of components and structures as being a source of escalating costs. The article states that regulatory changes were only responsible for a small fraction of rework, but also states that a lot of rework was done because of “inspection failures”. I have to wonder if the impeccable, and unique, fab QA requirements that only apply for nuclear is largely responsible for this. Would they have failed inspection at a fossil plant?

    Another point is that cost overruns are one thing, but the initial projected cost is another. Nuclear plants were projected to be more expensive from the get go, and nuclear regulations and fab QA requirements are likely part of the reason. I’ve always heard that “nuclear grade” components cost several times what the non-nuclear grade equivalent would. That has to be a huge factor in nuclear’s current high costs. And I wonder if the same is true in countries that don’t have expensive nuclear.

    Perhaps it’s in the Breakthrough report, but I’d be very interested to see how total nuclear plant staffing levels have changed over the decades. Staffing level is a good proxy for overall operation costs (other than fuel). Unless wages massively increased? And operation costs are the key issue with respect to an existing plant being able to stay open. If staffing levels have increased, it’s hard to think of a reason other than ever-increasing requirements.

  11. RTK

    Jim, gas turbine plants definitely do experience rework (I’ve seen some mistakes that really makes one wonder what people were thinking…or not thinking). When looking at the QA requirements for any power plant, one needs to know where the different requirements apply. For gas turbine plants, the “higher” QA requirements (often ASME BPVC Section I) only apply to a relatively small portion of the overall work (e.g., the steam plant and associated equipment, piping, etc.). For a nuclear plant, the “higher” QA requirements (both ASME BPVC Section III and NQA-1) apply to a (much) larger portion of the overall work. While I do not work on the current plants under construction, I would venture a guess that the NQA-1 requirements are a big driver of the increased costs (simply because NQA-1 applies to many more systems/components/equipment than ASME Section III applies too). Simply looking at the quantity difference between where the “higher” QA requirements apply between gas turbine plants and nuclear plants should reveal part of the significant cost difference.

    Since I have the pleasure of working with an ASME BPVC Section I, Section III, and NQA-1 quality assurance programs, my experience is that, in practice, Section I is much weaker than Section III (e.g., in terms of the thoroughness of the inspections, the Insurance oversight [don’t get me started there], etc.). That is not to say that work performed to a Section I QA program is unsafe, but more things “pass muster” in a Section I program than would “pass muster” in a Section III program.

    After having said all of the above, however, I still believe that a properly implemented QA program will help reduce the costs of plant construction. I do not think that it is the requirements of, say, an NQA-1 program that are the problem, but rather it is how those requirements are interpreted and implemented (e.g., by Engineering in their specifications for materials or lack of tolerances of variation in materials). Even on simple cycle gas turbine projects, a poorly implemented quality assurance program (or no QA program other than a book on the shelf) will result in increase rework and wasted time (I am including the concept of good planning – as described in the article – as part of a good QA program). I guess I am trying to say that there are many aspects of a QA program, other than just material requirements, that actually help reduce errors and control costs resulting in a net benefit to the project.