Recently, the new coalition government of the Netherlands looked across its decarbonization portfolio, realized that it had failed to meet renewables targets, and so announced that it would build two nuclear power reactors with 1-1.6 GW capacity each. And the government is claiming that it will have them running in 2035, but hass only outlined costs through 2030 of €5 billion ($5.5 billion).
The Netherlands’ plan does have a couple of things going for it. The country actually has a small, 50-year old, 485-MW nuclear reactor at Borssele, and they are apparently going to build the new reactors on the same site. They’ve also extended the life of the very old reactor, which has people understandably concerned. So they have operational experience with nuclear, albeit with a very different technology with considerably different operational characteristics, predating as it does most computerization of control systems. They have already jumped through the International Atomic Energy Agency’s 28 or so major hoops. They already have the seven overlapping, somewhat concentric layers of security from international to internal site high-security areas in place and know what is required. The combination puts them ahead of countries that don’t have existing nuclear reactors, and ahead of projects attempting to site reactors in a new location.
What doesn’t the Netherlands have or know about these reactors?
They don’t know what technology they will use. Some reports say that they will stick with third-generation nuclear technology, which sounds conservative until you realize that Hinkley in the UK, Flamanville in France, Vogtle and Summer in the US, and Olkiluoto in Finland were all third generation AP1000s and European Pressurized Reactors (EPR), and all have suffered massive cost and budget overruns.
They don’t have any trained, certified, or security cleared design or construction resources. The requirements for nuclear design and construction resources are substantially higher than for wind, solar, and other generation options. High security clearances are required for a vastly greater percentage of nuclear construction resources than for other forms of electrical generation, especially as they’ll be doing construction on a running nuclear site. Many people in non-nuclear trades such as boilers, turbines, electricians, and the like who would be acceptable for a wind farm, solar farm, or hydro project will not pass the filters for nuclear projects. In fact, many utility-scale construction projects employ vast numbers of unskilled day laborers that they pick up off street corners at the beginning of the day and drop off again at the end.
They don’t have a significant nuclear engineering program in any of their universities. The nuclear chair in TU Delft retired a decade or so ago and was never replaced. There’s a professor of nuclear engineering at the school, J
, and he’s clearly excited by this opportunity and hoping that the chair will be re-established with him sitting in it per his public comments.They have no one who has ever led and run the construction of a nuclear plant. The people who built Borssele are dead or retired to Spain or Portugal, one assumes.
They don’t have a primary contractor, and that’s much more of a problem than it was a decade ago. The three major countries that are building or attempting to build nuclear reactors in other jurisdictions are Russia, China, and France. Russia has made itself an international pariah and clearly wouldn’t pass basic security checks. China has been politically blackballed because it’s stopped being a cheap manufacturer of consumer goods and become instead a major economic competitor which has surpassed the US by several measures and is set to surpass it by most of the rest by 2035. And then there’s France, which has proven to Europe and the world that it is incompetent to build new nuclear reactors, and has had problems operating its own.
The Netherlands doesn’t have a plan, just an aspiration. They don’t have a schedule, just a notional target that is close enough to 2030 to sound good. They don’t have a budget, they have a number that they think that they can sell. There’s just so much failure inherent in this proposal that it’s like asking a flatland triangle to successfully build the Pyramid of Giza. Where to start?
Let’s lead in with Professor Bent Flyvbjerg’s Top-Ten Behavioral Biases in Project Management. As a reminder, Flyvbjerg’s recent book with Dan Gardner, How Big Things Get Done, has a chart of project categories by least to most likely to meet cost projections, with nuclear construction as 23rd of 25 categories of over 16,000 megaprojects. That was the first publication of the breakdown, and he and his team have subsequently provided an updated version with numbers, in worst to best order.
As the data shows, 55 nuclear construction projects globally had cost overruns greater than 50%, and the average of those projects were 204% overruns, which is to say that they cost three times more than budgeted for. Flyvbjerg attributes this to fat-tailed risks inherent in nuclear projects, but it starts very early, which is where the biases come in.
The first bias and most evident here, is strategic misrepresentation, aka lying outright or obfuscating the likely truth in order to get something going. When equivalent projects are looked at, €5 billion is clearly a gross understatement of the real costs, but is also clearly the only number that the government believes it can sell. The projected overnight capital costs of nuclear generation in OECD countries from the International Energy Agency (IEA) and Nuclear Energy Agency (NEA) in their 2020 report was an average of €3.7 billion ($4 billion) per GW, making the 2-3.2 GW projected reactors at €7.4 billion, 50% higher than the number the Netherlands is citing, to €12 billion, 140% higher. And that’s projected costs, not actuals as per Flyvbjerg’s data. Unsurprisingly, given massive cost overruns of all nuclear construction in Europe in recent decades, few are taking the number seriously.
There’s also strong evidence of the optimism bias, as going from no nuclear construction industry to two working reactors in 13 years is radically unlikely, as is the €5 billion price tag.
The planning fallacy is fully in evidence. That’s the one that originally brought Flyvbjerg to my attention, as it’s included in Daniel Kahneman’s Thinking, Fast and Slow. It is the tendency to underestimate costs, schedule, and risk and overestimate benefits and opportunities. It’s such a depressing reality that a long-term collaborator of mine refused to give any estimates at all for months after reading it the first time.
The over confidence bias is clearly in evidence. Anyone looking at the cost and schedule overruns globally in nuclear vs where the Netherlands is starting would not be considering it as the basis for 13% of grid decarbonization in any commitments. This isn’t to say that nuclear is bad, as I consider keeping older plants that can be kept going safely and economically to be a good choice, but building new ones is so fraught with long-tailed risks that complete cancellation is a strong potential outcome, as occurred with Summer in the US, for example. Decades-long delays is also a potential outcome, with the US Watts Bar 2 having been suspended in 1985 and only completed in 2015.
As I pointed out a couple of months ago, the country with the best possible conditions for nuclear build-out, China, is seeing trending to flat nuclear, with about three reactors a year on average getting to grid connection for the past five years, after the 2016 and 2018 peaks of seven reactors. Meanwhile, renewables are continuing to accelerate deployment.
The base-rate fallacy is evident as well. It’s the tendency to ignore generic base-rate information and focus on specific information pertaining to a certain case or small sample. Using Flyvbjerg’s reference class forecasting estimation methodology would look at as complete a list of all nuclear generation projects as it was possible to get, use the median of schedule and budget as the basis for an estimate, and use the average statistical deviance from the mean to inform contingency planning. In the Netherlands’ case, they are clearly not doing that, as the median cost of even European third-generation reactors in the past 20 years is closer to €10 billion per GW of capacity with massive variances in cost and schedule. A planning number would be closer to €20 to €32 billion for the reactors, with significant contingency even after that.
Luckily for the Netherlands, a database of nuclear construction projects with sufficient numbers to do reference class forecasting is available in nearby Copenhagen, so they should be talking with Flyvbjerg and his team to get better planning numbers. Of course, if they did that, then they would run right into the strategic misrepresentation problem. In other words, while they might believe that they are unique and different than other nuclear deployment projects, even they surely know that they are radically low-balling the numbers for political reasons.
Of course, if they do get this off the ground, then they’ll run into one other bias, escalation of commitment, also known as the sunk-cost fallacy. They’ll keep doubling down on bad plans and budgets as costs keep escalating.
It’s remarkable that the Netherlands considers this a viable route, but for a flat windy country, they have clearly not been leaning into renewables as they should have. It’s a desperation play like the Hail Mary pass of US football, whatever that’s called in actual football.
I don’t like paywalls. You don’t like paywalls. Who likes paywalls? Here at CleanTechnica, we implemented a limited paywall for a while, but it always felt wrong — and it was always tough to decide what we should put behind there. In theory, your most exclusive and best content goes behind a paywall. But then fewer people read it! We just don’t like paywalls, and so we’ve decided to ditch ours. Unfortunately, the media business is still a tough, cut-throat business with tiny margins. It’s a never-ending Olympic challenge to stay above water or even perhaps — gasp — grow. So …
Source: Clean Technica