To briefly recap, we previously found that nuclear power construction costs in the US and around the world steadily increased from the late 1960s through the 1980s.
Great summary! I just finished my PhD on the construction cost for advanced nuclear projects. You accurately describe many of the problems these projects face in both Part I and Part II. From the megaproject scale and complexity to the high degree of regulatory oversight and indirect costs. However, I think you overstate the impact of regulation on the recent cost overruns. It is correct to say that the primary issues for Westinghouse at Voglte have been change orders, but they were mostly internally driven, not caused by changing regulation. Westinghouse had only completed a fraction of detailed design when they started construction, so there were hundreds of design changes from stud spacings to wall locations to penetrations to tank sizes. Each of these required license amendments, but the change was usually driven by a constructability challenge not a license amendment.
Many times the vendors put themselves into this situation. For example, you mentioned Framatome's challenge with the reactor pressure vessel. Well, many people would have said Le Creusot did not have a large enough forge to manufacture that vessel and vessel head. Le Creusot tried anyway and the forgings had heterogeneities that the regulator found. Similarly, Hitachi was famously successful with the modularization of the ABWR, but they slowed built out their modularity over 6 generations of BWRs. Westinghouse tried to go from 0-100% in one generation with the AP1000, and this stressed the supply chain capability.
There is also a huge challenge with labor market availability. At HPC in the UK, there are 8,000 workers for 2 reactors. The Barakah site in the UAE had over 25,000 workers for 4 reactors at peak. The US and Western Europe simply don't have the construction labor resources to deliver these projects in a timely fashion. And the labor constraints limit their ability to respond to change orders/mistakes/etc.
Fingers crossed the industry has learned a few things when going into the SMRs. Some designs approach of eliminating systems entirely seems the right way to keep costs under control. For example (following it as the first build will be near me by OPG) the BWRX-300 is eliminating the main core circulation pumps, which eliminates the backup and redundant backup pumps, and the drives and instrumentation associated with them, and pipes and valves and all the potential interferences those imply. Trading steel for whole systems seems to be a good approach.
Do that with a bunch of the other systems, and maybe we start to get to something manageable. We will have a workforce that has been doing a whole bunch of reactor refurbishments in the leadup to this as well so I take this project as one of the best chances for a reasonably under control FOAK new build.
Thanks for the deep dig. Having lived in Southern California most of my life I remember the fiasco of the botched revamp of the San Onofre plant. It’s not something that many remain aware of, even though it happened this century and must be a contender for biggest and most embarrassing construction failure of all-time. Any light you can shed on that would be much appreciated. As I recall the blame landed on inferior Chinese steel that corroded rapidly as soon as it went into use, forcing closure of the plant less than a year after the expensive re-model was finished. The tab is still being paid. Is it fair to consider this as evidence of industrial decline in the US? Do tell...
Maybe you could add a gross profit figure/evolution? Does profit scale with total cost? Are companies lobbying to increase safety or to build as fast as possible?
The ALARA discussion (here and in the previous article's comments) is most interesting. "As Low As Reasonably Achievable", where those who define "Reasonably Achievable" are not the ones doing the work, is a microcosm of our times. In our democratic system, irrational public fear easily gains the political upper hand, leading to potentially very bad policies.
Edit: Efforts at non-proliferation of nuclear weapons can also explain the economic failure of nuclear power. I vaguely remember a college professor saying that the Carter administration was extremely concerned about international proliferation of nuclear weapons. If the "normal" technology to produce electricity around the world were nuclear, nuclear materials and weapons would also become more common. Strangling the nuclear power industry was seen as the only way to prevent proliferation.
ALARA may sound like a policy designed to make nuclear expensive, but in practice it doesn't seem to be a significant part of the problem. Enforcement of ALARA mostly just means you have to measure and track worker exposure (and public exposure, but generally there isn't any) and consider if there's a way to do the same job with lower doses.
At the risk of being an ad-hominem, NRC Commissioner David Jaczko's Wikipedia page makes for interesting reading. Prior to being the NRC Chairman he worked for Harry Reid, Nevada senator and opponent of the Yucca Mountain waste repository. "an inspector general report found that he sought unilaterally and improperly to block the Yucca Mountain nuclear waste repository project from advancing"
That may be it , but it’s hard to be certain. I think it may have been as much a case of false advertising and lack of domestic US sources. It’s still hard to fathom. The original stuff worked for 30+ years. This wasn’t a cheap apartment complex put up by a shady developer. Since this installment the article on US Navy nuclear power detailed the impressive quality control that is applied to their projects. San Onofre borders USMC Camp Pendleton. Pretty ironic.
Back then, I had a friend who worked on the first X-Ray telescope satellite. One day he was quite excitedly showing off a transistor. "You'll never see this again in your lifetime", he said. Did I mention that he was showing me a tin can transistor? Even in the 1970s, they were pretty common. "It's a failed MILSPEC transistor", he explained. He had never seen one before. He had never met anyone who had seen one before. He had been debugging some component of the satellite all night, and when he could find no other conceivable problem, he finally decided to test the transistor. Ignoring the markings, it looked just like any other transistor, but a MILSPEC, military specification, transistor was not just an ordinary transistor. It had been drawn from a traceable batch. It was overspecified for its job. It had been tested for thousands of hours at high and variable load. What else can I say? He was right. I've seen lots of transistors since then, but I've never seen a failed MILSPEC transistor again. (It obviously cost a whole LOT more than the transistors one got in ten packs at Radio Shack.)
There was an earthquake down in Mexico City, and a lot of supposedly earthquake resistant buildings had collapsed. I asked a friend of mine who had been an architect in Mexico City and was working on his M.Arch degree. He had worked, straight out of school, at a construction site in Mexico City. His job was accepting deliveries of concrete. Being straight out of school, he actually tested the contents of the arriving trucks. Some of them, he said, had perfectly good concrete in them, but an awful lot of them didn't pass the test. Their loads were watered down or otherwise substandard. Sometimes a truck contained just water and some rocks. He rejected a lot of them and was reprimanded. The suppliers had won the bids for supplying the concrete, but having paid off the right people, they had to accept the loads and hope for the best.
Quality and reliability can get expensive. If you may be fighting a war, building something that cannot be repaired or operating a potentially dangerous system, you may have to pay a premium to get something that works right when its abused or in the face of partial failure. All that expensive tracking and documentation is sometimes the only approach, and even all that is sometimes not enough. We've seen enough cases of forged documents, faked Xrays and bogus structural tests. Unfortunately, I don't know a much better way to do it.
Great summary! I just finished my PhD on the construction cost for advanced nuclear projects. You accurately describe many of the problems these projects face in both Part I and Part II. From the megaproject scale and complexity to the high degree of regulatory oversight and indirect costs. However, I think you overstate the impact of regulation on the recent cost overruns. It is correct to say that the primary issues for Westinghouse at Voglte have been change orders, but they were mostly internally driven, not caused by changing regulation. Westinghouse had only completed a fraction of detailed design when they started construction, so there were hundreds of design changes from stud spacings to wall locations to penetrations to tank sizes. Each of these required license amendments, but the change was usually driven by a constructability challenge not a license amendment.
Many times the vendors put themselves into this situation. For example, you mentioned Framatome's challenge with the reactor pressure vessel. Well, many people would have said Le Creusot did not have a large enough forge to manufacture that vessel and vessel head. Le Creusot tried anyway and the forgings had heterogeneities that the regulator found. Similarly, Hitachi was famously successful with the modularization of the ABWR, but they slowed built out their modularity over 6 generations of BWRs. Westinghouse tried to go from 0-100% in one generation with the AP1000, and this stressed the supply chain capability.
There is also a huge challenge with labor market availability. At HPC in the UK, there are 8,000 workers for 2 reactors. The Barakah site in the UAE had over 25,000 workers for 4 reactors at peak. The US and Western Europe simply don't have the construction labor resources to deliver these projects in a timely fashion. And the labor constraints limit their ability to respond to change orders/mistakes/etc.
Fingers crossed the industry has learned a few things when going into the SMRs. Some designs approach of eliminating systems entirely seems the right way to keep costs under control. For example (following it as the first build will be near me by OPG) the BWRX-300 is eliminating the main core circulation pumps, which eliminates the backup and redundant backup pumps, and the drives and instrumentation associated with them, and pipes and valves and all the potential interferences those imply. Trading steel for whole systems seems to be a good approach.
Do that with a bunch of the other systems, and maybe we start to get to something manageable. We will have a workforce that has been doing a whole bunch of reactor refurbishments in the leadup to this as well so I take this project as one of the best chances for a reasonably under control FOAK new build.
Thanks for the deep dig. Having lived in Southern California most of my life I remember the fiasco of the botched revamp of the San Onofre plant. It’s not something that many remain aware of, even though it happened this century and must be a contender for biggest and most embarrassing construction failure of all-time. Any light you can shed on that would be much appreciated. As I recall the blame landed on inferior Chinese steel that corroded rapidly as soon as it went into use, forcing closure of the plant less than a year after the expensive re-model was finished. The tab is still being paid. Is it fair to consider this as evidence of industrial decline in the US? Do tell...
Sounds like the problem was with Chinese steel probably chosen to cut costs.
Maybe you could add a gross profit figure/evolution? Does profit scale with total cost? Are companies lobbying to increase safety or to build as fast as possible?
Once again a fantastic job. Thanks for the work. Looking forward to Part 3.
The ALARA discussion (here and in the previous article's comments) is most interesting. "As Low As Reasonably Achievable", where those who define "Reasonably Achievable" are not the ones doing the work, is a microcosm of our times. In our democratic system, irrational public fear easily gains the political upper hand, leading to potentially very bad policies.
Edit: Efforts at non-proliferation of nuclear weapons can also explain the economic failure of nuclear power. I vaguely remember a college professor saying that the Carter administration was extremely concerned about international proliferation of nuclear weapons. If the "normal" technology to produce electricity around the world were nuclear, nuclear materials and weapons would also become more common. Strangling the nuclear power industry was seen as the only way to prevent proliferation.
ALARA may sound like a policy designed to make nuclear expensive, but in practice it doesn't seem to be a significant part of the problem. Enforcement of ALARA mostly just means you have to measure and track worker exposure (and public exposure, but generally there isn't any) and consider if there's a way to do the same job with lower doses.
At the risk of being an ad-hominem, NRC Commissioner David Jaczko's Wikipedia page makes for interesting reading. Prior to being the NRC Chairman he worked for Harry Reid, Nevada senator and opponent of the Yucca Mountain waste repository. "an inspector general report found that he sought unilaterally and improperly to block the Yucca Mountain nuclear waste repository project from advancing"
so, basically the whole thing is fubared and there's no point in trying to unfubar the situation.
the industry needs to figure out its own problems before it goes begging to the governments for perpetual public subsidies.
That may be it , but it’s hard to be certain. I think it may have been as much a case of false advertising and lack of domestic US sources. It’s still hard to fathom. The original stuff worked for 30+ years. This wasn’t a cheap apartment complex put up by a shady developer. Since this installment the article on US Navy nuclear power detailed the impressive quality control that is applied to their projects. San Onofre borders USMC Camp Pendleton. Pretty ironic.
Two stories come to mind, both from the 1970s.
Back then, I had a friend who worked on the first X-Ray telescope satellite. One day he was quite excitedly showing off a transistor. "You'll never see this again in your lifetime", he said. Did I mention that he was showing me a tin can transistor? Even in the 1970s, they were pretty common. "It's a failed MILSPEC transistor", he explained. He had never seen one before. He had never met anyone who had seen one before. He had been debugging some component of the satellite all night, and when he could find no other conceivable problem, he finally decided to test the transistor. Ignoring the markings, it looked just like any other transistor, but a MILSPEC, military specification, transistor was not just an ordinary transistor. It had been drawn from a traceable batch. It was overspecified for its job. It had been tested for thousands of hours at high and variable load. What else can I say? He was right. I've seen lots of transistors since then, but I've never seen a failed MILSPEC transistor again. (It obviously cost a whole LOT more than the transistors one got in ten packs at Radio Shack.)
There was an earthquake down in Mexico City, and a lot of supposedly earthquake resistant buildings had collapsed. I asked a friend of mine who had been an architect in Mexico City and was working on his M.Arch degree. He had worked, straight out of school, at a construction site in Mexico City. His job was accepting deliveries of concrete. Being straight out of school, he actually tested the contents of the arriving trucks. Some of them, he said, had perfectly good concrete in them, but an awful lot of them didn't pass the test. Their loads were watered down or otherwise substandard. Sometimes a truck contained just water and some rocks. He rejected a lot of them and was reprimanded. The suppliers had won the bids for supplying the concrete, but having paid off the right people, they had to accept the loads and hope for the best.
Quality and reliability can get expensive. If you may be fighting a war, building something that cannot be repaired or operating a potentially dangerous system, you may have to pay a premium to get something that works right when its abused or in the face of partial failure. All that expensive tracking and documentation is sometimes the only approach, and even all that is sometimes not enough. We've seen enough cases of forged documents, faked Xrays and bogus structural tests. Unfortunately, I don't know a much better way to do it.