The Nuclear Industry’s Really Bad Safety Analysis

Sign up for daily news updates from CleanTechnica on email. Or follow us on Google News!

Before we allow small modular reactors, mini reactors on barges, reactors for making hydrogen, reactors to be set up on the Moon, or just about any nuclear reactors to be built, we should get an explanation from the nuclear industry of why some of its calculations have been so bad. I am talking about numbers that are so bad, off by an order of magnitude, that they are functionally deceptive. And if they are not intentionally deceptive, that is not an excuse. They fool people into thinking things are true when they are not, and they are very much to the advantage of the nuclear industry.

The Japanese knew about tsunamis long ago, but a few of them seem to have forgotten. The Great Wave off Kanagawa, by Katsushika Hokusai. (Printed from 1731-1734,public domain.)

In my experience, nuclear industry numbers come in three flavors. The first of these is just about spot-on, just about 100% of the time. These numbers relate to such things as calculations of energy potentials. If a company says a reactor can provide 1,000 MW of power, you can count on it that the output of that plant will be 1,000 MW.

There is a second type of calculation that the industry provides. This relates to the costs and timelines for construction of new reactors. Calculations in these areas seem to underestimate both by about half. If the figure for a new reactor is $5 billion and its construction is to take 5 years, I would figure on a $10 billion cost and a 10-year timeline. I acknowledge that I have not done a careful study of this, but I have often noted that real world results that were double what was estimated, and I only remember one that was correct. (I am speaking of US and UK reactors here, not those that are built in China, with which I am much less familiar.)

It is the third type of number that really bothers me, however, and this is something I have studied since the Fukushima Disaster. For some purposes regarding safety, the industry numbers are off by an order of magnitude, with its calculations appearing to make the reactors ten times as safe as they are.

I admit that some safety numbers are controversial. There is huge disagreement about how many people died as a result of the Chernobyl Disaster, for example. I will start with that.

I got an email some time back from a nuclear industry advocate. In it he said the explosion of a reactor at Chernobyl proved that nuclear energy was safer than coal or other fossil fuels. I will grant you that our use of coal has caused a large number of deaths, both in the mines and among the general public. But the email was revealing in its very specific number. How it was calculated, I don’t know, but the figure given was 17 deaths.

Though that was the lowest I have seen, other figures from nuclear advocates are equally specific. A list at Wikipedia says provides the old Soviet Union’s official list of 31 people who were killed. I have seen that figure used recently. Elsewhere in the same article, however, it is stated that a total of 60 people died, according to a later source, including later deaths of radiation-induced cancer. Again, the same article speaks of UN estimates of 4,000 people who might have died of health effects in the Ukraine, Belarus, and Russia, with a total of 16,000 when other parts of Europe are factored in, and maybe as many as 60,000 when worldwide effects are considered. Greenpeace has projected death tolls possibly going up to a million.

So what are we to believe? To answer that question, I will say we should not believe any specific figure, but instead note that the low figures from the nuclear industry and the high figures from anti-nuclear activists differ by over four orders of magnitude.

I would also note that the figures from the nuclear industry are not believable, because they do not even admit the possibility of more deaths than the low numbers they specify. And I will not say the high figures are unbelievable in quite the same way, though I would not believe them just because they come from Greenpeace. The point is that the differences themselves are disturbing.

Some time around 1978, I had a chat with two friends who were nuclear engineers. That was before the Three Mile Island meltdown and the movie “The China Syndrome,” both of which happened the following year. I had heard about the China Syndrome and asked them about it. The younger of them told me it was a meltdown, and I would probably never see a meltdown take place in my lifetime, because the likelihood of such an event was so small. In fact, he said, the probability was only one in ten thousand.

“Uh, Fred,” the older engineer said, “that’s one in ten thousand for any one reactor in any one year. If you have a thousand reactors worldwide, we would see one every ten years or so.”

The younger man thought for a couple of seconds, and then he said, “Well, you’ll probably never see one in this country. Or maybe you’ll see just one. But that is it.”

The older nuclear engineer said, “Uh, Fred, we have already had two.”

“I’m talking about reactors supplying power to customers here, not research or military reactors.”

“The two I am referring to were both supplying power to customers. One was the Sodium Reactor Experiment, just outside Los Angeles, which operating commercially despite its name. The other was Fermi 1, just outside Detroit.”

Looking back, the nuclear industry appears to have sided with Fred. The position would seem to be, if you don’t know about it, it didn’t happen.

In those days, the nuclear industry’s safety analysis, which Fred had referred to, said the chance of a “core damage event” (meaning partial or total meltdown) was one in 10,000 per reactor year. Not long after that, a new calculation on newer equipment put the chance at one in 20,000. Now we have newer reactors that are supposed to have a likelihood of one in 50,000, and we are promised future reactors that cannot melt down at all. The great majority of reactors running today have a one in 20,000 likelihood, so I will hope to be pardoned if I apply that figure to all of them for the sake of simplicity.

Over the years, we have accumulated very roughly 20,000 reactor years of nuclear power. That means that the number of core damage events should have been about one. But we have not had one. We have had one such event at Chernobyl, in the USSR; three at Fukushima, in Japan; three in the US, including the two already mentioned plus Three Mile Island; two at Saint-Laurent, in France, on separate occasions; one at Jaslovské Bohunice, Czechoslovakia; one at Chapelcross, in Scotland.

Please note that I have only included reactors supplying energy commercially to the grid in this list. Some of the more notorious disasters, such as the Windscale fire in England, the Lucens reactor in Switzerland, and various military reactors are not included. Also please note that there are possibilities that meltdowns at some commercial reactors may have gone unreported or been reported as other types of problems; I am not convinced that the incident at Greifswald, in East Germany, did not include core damage. And who knows what might have happened in the USSR that we were never told about?

Where one core damage event should have been expected, at least 11 happened. Clearly, in practice the safety analysis calculations were off by over an order of magnitude. I don’t see how this can be disputed.

We should ask why this happened and how it can be corrected. Again, studying this over the years, I have come to the conclusion that there is a common thread among all these events, and it is something that was intentionally overlooked in the calculations. That one thing is the human factor, the likelihood that human failure will cause a meltdown.

I use the word ‘intentionally,’ intentionally. Looking into the calculations, I realized (and please correct me, if you can show I am wrong) that human failure cannot be quantified, and nothing is included among the numbers unless it can be quantified. In other words, you cannot calculate human failure, so you leave it out of whatever formulas you prepare. Doing so, you ignore the biggest cause of nuclear failure in your safety analysis.

If you want to counter that the Fukushima Disaster did not result from human failure, I will thank you. This is a great example of both human failure, and how it can be ignored in a manner that is very hard to notice, unless a spotlight is shone on it.

The Fukushima Daiichi plant was protected by a 5.7 meter (18.7 feet) sea wall. The Tohoku tsunami that hit it, which arose from the most powerful earthquakes ever recorded in Japan, was 14.5 meters (47.6 feet) tall at the plant. To get a grasp on what this means, remember that a tsunami of this type is not an ordinary wave that quickly crashes and withdraws, it is a rise in the water level than keeps going on for a period of time, possibly several minutes. So for a period that must have felt terrifyingly long to those who witnessed it, water kept rising until it was nearly 8.8 meters (28.9 feet) higher than the sea wall. Videos of the Tohoku tsunami show the effect.

The nuclear industry’s response to this could be summed up in the question: Who could have predicted such a thing? But that fails utterly to recognize some facts. The sea wall was clearly too low, based on the region’s history.

To start with, the earthquake was the most powerful on record, but the records only go back to 1900. The 14.5-meter wave at Fukushima Daiichi was not the highest that the tsunami produced. The highest wave it had was 40.5 meters (133 feet) in Iwate Prefecture. But we should recognize this was not really the unique event it seemed. Limiting our look to other tsunamis on the northeastern coast of Japan, we could count the 1896 Sanriku Earthquake, which had waves of up to 30 meters (100 feet), and the 1933 Sanriku Earthquake, which had waves of up to 28.7 meters (94 feet).

Both of these events happened less than 100 years before the Fukushima power plant was designed in the early 1970s. They should have been taken into account for construction of sea walls. In fact, they were taken into account for building a 15.5-meter (51 feet) flood gate at Fudai, Iwate, a village of roughly 2,600 people. The waves were higher there than at Fukushima Daiichi, but they barely topped the flood gate, and no one it protected was killed.

So why did Fukushima Daiichi have a 5.7-meter sea wall? I would say it was clearly a matter of human failure. Though I have to admit that reality might have been very different, I can picture someone walking into a room with a group of engineers in it, and asking how tall the sea wall was going to be. And the answers comes back as a question, “What is the budget?”

I have never seen the nuclear industry address the question of why their safety analysis calculations are off by an order of magnitude. Until they do, and I think it should include a gigantic mia culpa, my own choice would be not to allow any of their fantastical designs to be built.

Have a tip for CleanTechnica? Want to advertise? Want to suggest a guest for our CleanTech Talk podcast? Contact us here.

Latest CleanTechnica.TV Videos

CleanTechnica uses affiliate links. See our policy here.

George Harvey

A retired computer engineer, George Harvey researches and writes on energy and climate change, maintains a daily blog (, and has a weekly hour-long TV show, Energy Week with George Harvey and Tom Finnell. In addition to those found at CleanTechnica, many of his articles can be found at

George Harvey has 78 posts and counting. See all posts by George Harvey