(Previously in this series: Part I, Part II)

This week we conclude our look at nuclear power construction costs.

So far we’ve looked solely at conventional nuclear power stations that supply electricity to the grid for residential and commercial use. But civilian power stations aren’t the only sort of nuclear reactor that the US builds. The US Navy has been building nuclear reactors for 70 years as part of its Naval Reactors program. It’s built over 200 nuclear-powered ships and 500 reactor cores, meaning the US Navy has more experience building nuclear reactors than anyone in the world (or second most, if you count the former Soviet Union’s experience building nuclear ships.) Unlike civilian power stations, US naval nuclear-powered ships are frequently (though not always) delivered on-time and on-budget.

As I’ve previously argued, shipbuilding practices and improvements are a useful source of comparison for improvements to the construction process. So it’s potentially useful to compare the civilian and naval nuclear power programs, and see if there are lessons to be learned.

A brief history of US naval reactors

The US Navy had been interested in the potential of nuclear-powered ships, particularly submarines, as far back as 1939; Enrico Fermi gave a presentation to the navy on the potential of nuclear weapons and energy in 1939, shortly after the fission chain reaction was discovered. At the time submarines used a diesel engine (which required oxygen to run) for operating on the surface and for charging the batteries, and a battery-powered electric motor for submerged operation. Battery operation was slow, and limited in duration - after a comparatively short period of time the submarine would be forced to surface. A nuclear reactor, which didn’t require oxygen to generate power, offered the tantalizing prospect of a submarine which could travel submerged at high speeds, and remain submerged for weeks or months at a time.

A joint project to explore the possibility of nuclear reactor began at Oak Ridge labs in 1946, and included a group of naval officers led by Hyman Rickover (who would go on to lead the navy’s reactor program for over 30 years.) By 1947, partly spurred by rising concern about the Soviet Union, plans were underway to build a nuclear-powered submarine. The prototype submarine reactor began construction in 1950 (4 years before the construction of the first civilian reactor at Shippingport, which used a reactor originally designed to power aircraft carriers), and the world’s first nuclear-powered submarine, the USS Nautilus, launched in 1954. The Nautilus confirmed the potential of nuclear-powered submarines, and over the next 20 years the navy would build 60 more, as well as several nuclear-powered surface ships. Because of the advantages of nuclear propulsion, the navy continues to use nuclear power for all its submarines today, though the only surface ships that currently use nuclear power are aircraft carriers.

Civilian vs naval reactors

It’s useful to compare the design philosophies and implementation differences between naval and civilian reactors.

In many ways they’re similar. For one, like civilian reactors, naval reactors are built with an overwhelming focus on safety. The navy follows the ALARA philosophy, and goes to great lengths to minimize exposure to radiation. No military or civilian personnel in the Naval Reactors program have ever received more than 10% of the annual occupational exposure limit (the same limit that the NRC uses), and the average annual exposure of reactor personnel is 0.112 rem per year, compared to the 0.620 rem per year average of the US population as a whole. Naval reactors have operated for over 5000 reactor years, and over 130 million miles of travel, without a major accident.

(This depends somewhat on the definition of “major accident.” The US has had two nuclear submarines, the USS Thresher and the USS Scorpion, sink for reasons that are only partially understood, though in neither case has a reactor failure been implicated. And the navy had at least one submarine, the USS Guardfish, experience a partial loss of reactor coolant that resulted in minor radiation exposure, the records of which have remained classified.)

The methods it uses to achieve this level of safety are similar to civilian reactors. Like the NRC, naval reactors also use a “defense in depth” approach to safety, with multiple overlapping, redundant safety systems to prevent single points of failure. Safety analysis and potential failure modes are based on operating experience (and in some cases draw on NRC recommendations based on civilian experience.) And like civilian reactors, the navy ensures performance requirements are met via an extremely thorough system of quality control and assurance (known as SUBSAFE), where components and installation work are carefully tested and documented at every step in the manufacturing process, to ensure the correct material and part is used in the right place. Welds, for example, are extensively tested with radiographic inspection to ensure they meet specifications. (We see somewhat similar stringency on welding requirements in civilian reactors - of the 3000 people currently working on the Flamanville reactor, 800 are repairing welds that didn’t meet specifications.) And like civilian reactors, this system has a level of stringency similar to the requirements for manned spaceflight - in the early 2000s, NASA did a study to see what safety lessons it could learn from the navy’s reactor program.

Like with civilian reactors, naval reactor requirements have evolved as experience has been gained operating reactors, and have increased in stringency following accidents - the SUBSAFE program, for instance, was instituted following the loss of the USS Thresher in 1963. Every operational issue, no matter how small, gets documented (a GAO analysis in 1991 found that the navy’s 9 land-based prototype reactors had generated more than 12,000 documented operation issues since their inception), and informs subsequent requirements and procedures. However, naval reactor design has always been extremely conservative - the USS Nautilus, for instance, despite being built in the 1950s, was able to meet the much stricter radiation standards in place in the 1970s.

Naval reactors also show a somewhat similar philosophy to construction flexibility as civilian reactors. While for most of the ship, the shipbuilder has some leeway to decide the best means and methods for building the ship, and can make adjustments to the design as necessary, drawings for the reactor and related systems are marked “non-deviation”, which must be followed exactly and can’t be changed without approval. Over time, larger and larger numbers of design documents have been marked “non-deviation.”

And naval reactors show a level of design reuse that is often desired but seldom achieved in civilian plants. For instance, the S5W reactor was used on 98 different submarines of 8 different classes between 1959 and 1979. And though the Virginia class has undergone several design evolutions since its inception, all versions use the S9G reactor.

However, there are also many differences between naval reactors and civilian ones.

Some of these differences are physical. Naval reactors are smaller and produce less power than a commercial power station, which makes managing the risk from loss-of-coolant accidents easier. The largest naval reactor, the A1B, generates an estimated 380MW of electricity and driveshaft power, and submarine reactors generate less than 100MW, compared to 1100MWe or more for commercial power stations. Modern naval reactors (it’s speculated) are often designed to use a convection-based cooling system, allowing the reactor to operate without the need for cooling pumps.

Naval reactors also use highly-enriched uranium for fuel, a denser power source that allows ship reactors to operate without refueling for extremely long periods of time - current reactors are designed to last the entire life of the ship without being refueled. This means that unlike civilian reactors, Naval reactors don’t need to be designed for regular refueling, and refueling one is an involved process. When the Ohio class submarine’s service life was extended from 30 to 42 years, it included a refueling overhaul which took 2+ years to complete (though this included many other items beyond just refueling.)

While civilian power plants will (ideally) operate near their capacity for their entire life (and often can’t be easily ramped up or down), naval reactors spend most of their time operating at a fraction of their potential output, and are designed to quickly ramp up or down (and be turned off and on) as conditions dictate. And these operating conditions are much more extreme than a civilian reactor will be subject to - beyond the potential conditions created by an accident, a naval reactor must operate deep underwater, and may be subject to battle damage (potentially causing shocks, equipment failures, exposure to high-pressure water, etc.)

There are also conceptual differences. One major one is the extreme focus on training as a key element of safe reactor operation. Though naval reactors are designed to be tolerant of operator errors (once source described them as needing to be “sailor proof”), properly trained personnel are considered a key component of reactor safety. From “Rickover and the Nuclear Navy”:

Recognizing that it was impossible to design equipment that would never fail and equally impossible to devise procedures that would cover all contingencies, the only logical course was to train the operator so that he would have a thorough understanding of the plant and its capabilities. That was the reason Rickover had worked out a comprehensive interview system - in which his senior staff took part - to select officers who were intelligent, capable of understanding complex phenomena, willing to undergo rigorous training, and able to grasp the essential element that in confronting a technical problem there could be no equivocation or evasion. A reactor operator had to be able to integrate the information flowing to him and use his knowledge of the plant to handle the situation. He could not depend on memorizing procedures. He had to know.

Naval reactor training focuses sourcing highly qualified candidates, and instilling in them deep technical knowledge, the individual’s responsibility to maintain safety, and the discipline to maintain proper procedures day in and day out. This includes extensive upfront training (reactor operators essentially go to a second, naval school for nuclear engineering after they get an engineering degree), as well as ongoing training to ensure skills and knowledge stay sharp. Following the Three Mile Island accident, Naval Reactors was asked to prepare a statement on what civilian reactors could learn from the navy’s long history of safe reactor operation. Of the 111 pages prepared, 88 were devoted to training.

Another major difference is that with nuclear vessels, emphasis is often placed on maintaining the industrial base that can construct them - it’s recognized that nuclear submarines (along with the reactors that power them) are incredibly complex artifacts, and only a few firms and people possess the required skills, and those capabilities may deteriorate if not deliberately maintained. This gets mentioned repeatedly, for instance, in RAND’s report on lessons learned from various nations’ submarine programs:

Both the technical community—the civilian and military engineering directorates and laboratories, test centers, and centers of excellence that support submarines—and the industrial base that designs, builds, and maintains submarines must be sustained at some level so they can provide the required capabilities when needed. This is particularly important for the submarine industrial base because submarine design and construction requires specific skills that cannot be sustained by surface ship programs.

For instance, the second and third Seawolf subs were built largely to maintain continuity in the industrial base until the next submarine program began. And the Virginia submarine program was structured with both Electric Boat and Newport News building portions of each sub partly to ensure multiple shipyards maintained the ability to build nuclear submarines.

More generally, these both reflect the importance of continuity - ensuring that knowledge of how to build and operate nuclear technology is maintained and passed down.

Nuclear submarine cost management

While the navy has often managed to control costs and schedule on nuclear ship construction, it hasn’t always done so. So it’s useful to look at programs that differ in how well they controlled costs, and see how they compare (as well as how they compare to civilian reactor construction.)

Unsuccessfully controlled costs - the Seawolf class

The Seawolf class submarine program began development in the early 1980s. Its goal was to produce an attack submarine with greater capabilities than the then-current Los Angeles class, to counter the increasing capability of Soviet subs. It was designed to incorporate numerous technical advances, including the use of new materials (HY-130 steel, later changed to HY-100 steel) that would allow it to dive deeper without adding significant weight.

However, the Seawolf program was beset with problems. The contract was split between two shipyards (Electric Boat and Newport News), which had different production processes, CAD systems, part numbering schemes, and design details. This created significant coordination issues as drawings and information required translation between the two shipyards (exacerbated by the fact that the shipyards were reluctant to cooperate for fear of “sharing secrets”), causing project delays and adding costs.

In addition, the requirements for welding HY-100 steel turned out to be poorly understood. Numerous welds that met specifications were later found to be faulty, requiring extensive rework. This issue alone caused a 1-year delay in project completion.

On top of this, construction on the first sub in the class began when design was only 10% complete. Inevitably, design requirements changed as development proceeded, causing further rework, delays, and cost overruns. There were also difficulties finding vendors who could meet the fabrication requirements of the components, forcing the shipyards to fabricate many of them themselves (at increased cost.) The initial submarine in the class (the USS Seawolf) would ultimately take 6 years to build.

Ultimately the collapse of the Soviet Union changed the rationale behind the Seawolf class, and (combined with higher than expected costs) resulted in the program ending with just 3 of the proposed 29 subs built, eliminating the possibility of learning from experience and reducing costs over time.

Overall, the issues experienced by the Seawolf program aren’t dissimilar from those seen on civilian power plants:

Delays and cost overruns caused by frequently changing design requirements
Rework caused by an inability to meet the necessary quality requirements
Component sourcing difficulties
Coordination difficulties making it difficult to reliably get work done
Inability to learn from experience and improve over time

Successfully controlled costs - the Virginia class

Development of the Virginia class attack submarine began in the 1990s following the fall of the Soviet Union and the end of the Seawolf program. The Virginia class was deliberately designed to a) learn lessons from the Seawolf program, and b) be inexpensive to produce (the goal was a submarine inexpensive enough that the navy could afford to build 2 per year, which would help to maintain the industrial base.)

As part of this effort, Virginia deliberately eschewed new, untested technology as much as possible, and avoided trying to push the boundaries of submarine performance. Technology and systems were deliberately chosen on a cost-benefit approach. For newly developed systems that it did include, it used a “try before buy” acquisition strategy, where new equipment was tested first on land or on other ships prior to construction.

Similarly, the Virginia program placed a great deal of emphasis on upfront design work, and attempting to spot and resolve problems before they occurred during construction.

Virginia made extensive use of 3D CAD to experiment with equipment layouts and find potential interferences, in addition to building full-scale mockups to help find potential issues (a method with a long history of successful use in nuclear sub design - previous sub programs often built full-scale wood models of the submarine prior to construction.)
When construction on the first Virginia class sub began, design was 50% complete (compared to just 10% on the first Seawolf class), and Virginia had 80% fewer design errors found during construction than Seawolf. Unlike Seawolf, design requirements were frozen early on to minimize changes required during construction.
Shipyards and vendors were brought in early on in the design process, to ensure constructability was taken into account.
While Virginia also used multiple shipyards to build different components, effort was made to improve the structure. Newport News modified its construction practices to match those of Electric Boat, and sent a team of 35 engineers to learn and understand their methods.

Virginia also took greater advantage of modular construction than previous programs. As we’ve discussed previously, modular ship construction can greatly accelerate ship construction and reduce costs, as it allows construction and verification work to take place in a workshop environment rather than an enclosed hull, and minimizes the amount of difficult overhead work that must be performed. Virginia was able to complete a much larger fraction of work prior to the hull being closed than previous programs.

Emphasis was also placed on simplicity and standardization - Virginia used 80% fewer unique components than Seawolf had used.

As a result of these efforts, the first Virginia class cost ~$2.8 billion to build (in 2004 dollars), compared to an average of $6 billion for each of the three Seawolf classes (though direct comparison is difficult, as the submarines had different capabilities.) [0] However, this was still 50% higher than originally budgeted for. Effort was thus placed on improving the construction process of subsequent subs. The builders managed to cut over 100,000 labor hours out of the construction process by improving and simplifying design and increasing automation. The cost was eventually brought down to less than $2 billion in 2004 dollars (though the Virginia was later upgraded with a “Virginia Payload Module”, which raises its current cost to ~$3.6 billion, $2.6 billion in 2004-dollars.) And though there have been some hiccups, the construction time for Virginia subs has steadily decreased (though recently there have been challenges meeting delivery dates as focus shifts to the new Columbia-class ballistic missile subs.)

Like we saw with Seawolf, the strategies for successful cost and schedule management largely overlap with what’s typically called for to improve civilian plant construction: simplicity of design, stability of requirements, clear lines of responsibility, learning from experience and improving over time.

Comparative cost

However, just because nuclear-powered ships can be built on-time and on-budget doesn’t mean that they can be built cheaply. A nuclear-powered naval vessel is not, and has never been, less expensive to produce than a conventionally powered ship:

A 1961 naval study (by which time the US had built close to 20 nuclear-powered subs and surface ships) estimated that a nuclear-powered surface ship would cost approximately 1.5 times what an oil fired powered ship would cost. Of that extra cost, approximately 1/3rd was from the nuclear fuel, with the rest the cost of the reactor and additional systems.
In 1982, it was estimated that a French nuclear submarine cost approximately 1.7 times what a conventional diesel electric submarine cost, and also had higher maintenance and support costs.
More recently, a CRS report estimated that making a naval surface ship nuclear-powered would add $600-800 million to the cost of the ship in 2007 dollars ($845-$1,127 million in 2022 dollars.)

Given that it can raise the cost of the entire ship by such a substantial amount, this suggests that a nuclear power plant is many times more expensive than a conventional (oil fired, gas turbine, diesel, etc.) power plant on a ship.

And beyond the costs of the ships themselves and the crew to operate them, fielding a nuclear force requires the navy to maintain an organization of 5000+ people in the Naval Reactors organization and its research arms (though it’s unclear to what extent the costs of a new nuclear vessel include these costs.)

The argument for nuclear ships in the navy has never been “nuclear is a cheaper source of power” (though it theoretically could be if oil prices rose enough). Instead, proponents have argued the benefits of nuclear propulsion (unlimited operation at high speeds without needing to worry about refueling logistics) are worth the additional costs:

On 2 January 1963, in a letter clearly intended for publication, Hayward wrote that his experience with the Enterprise [the US’s first nuclear-powered aircraft carrier] off Cuba and in the Mediterranean convinced him that the advantages of nuclear propulsion in surface combatant ships far outweighed the extra costs. The Enterprise was outperforming every carrier in the fleet. Her planes were easier and cheaper to maintain because they were not exposed to corrosive stack gases. The ruggedness and reliability of the propulsion plant gave her a high sustained speed and the ability to maneuver readily that enhanced air operations. In her first year the ship had 10,000 landings, a record no other carrier had achieved. Hayward strongly believed that nuclear propulsion would be badly needed in the years ahead. For that matter he was deeply disturbed that the navy was not exploiting every technological advance fully. Weighing the advantages of technology in dollars and cents now could cost victory later.

For submarines, the case for nuclear power was always very strong (though even today you see it argued that the US should build fewer nuclear subs and more, cheaper, diesel electric subs.) But for surface ships the advantage was less stark, and there was substantial debate in the 1960s and 70s between the navy, the Department of Defense, and congress over the degree that surface ships should be nuclear-powered.

Obviously, these subsidiary benefits largely don’t apply to civilian power stations - commercial nuclear power lives and dies on whether it’s cheaper than other methods of generating electricity.

It’s not obvious to me whether we should expect nuclear power to be cost competitive with other sources of power on a naval vessel. I would expect the efficiency penalty for a smaller-shipboard power plant compared to a larger, commercial power plant to be roughly similar for both nuclear plants and other power sources. And like with commercial power stations, I would expect (all else being equal) there to be substantial savings associated with removing the requirements and equipment for fuel storage and handling (on an Enterprise-sized aircraft carrier, for instance, it was estimated that changing from oil-fired to nuclear power allowed the storage of 7 air squadrons instead of 6, and a 50% increase in storage space for aircraft fuel and ammunition.) In practice, as with civilian power stations, it seems like the costs of the safety measures offset these benefits.

How much should this cost difference (which hasn’t decreased despite many years of experience with nuclear shipbuilding) inform our thinking about civilian nuclear power? It’s hard to say. It could be argued that the high levels of safety required for naval nuclear reactors don’t apply to civilian reactors, though it’s not obvious to me that the difference in safety requirements for nuclear and non-nuclear naval vessels will be greater than those for nuclear and non-nuclear civilian power stations. But it does suggest that making nuclear power cost effective isn’t a simple matter of requirements stability and learning from experience.

These posts will always remain free, but if you find this work valuable, I encourage you to become a paid subscriber. As a paid subscriber, you’ll help support this work and also gain access to a members-only slack channel.

Subscribe now

Construction Physics is produced in partnership with the Institute for Progress, a Washington, DC-based think tank. You can learn more about their work by visiting their website.

You can also contact me on Twitter, LinkedIn, or by email: briancpotter@gmail.com

[0] - I found conflicting estimates for the cost of the Seawolf class. Wikipedia and the National Interest says $3 to $3.5 billion, though the claim is unsourced (and that may be early 90s dollars, which would be ~$4-4.5 billion in 2004-dollars). RAND states the cost of the program was $18 billion. I’ve used the RAND numbers here.

Construction Physics

Why are nuclear power construction costs so…