The myth of nuclear reliability

“No amount of experimentation can ever prove me right; a single experiment can prove me wrong.” – Einstein

From the examples below it is clear that the claim of nuclear reliability is wrong.

As a student in the late 1960s I watched a training film which assured us that nuclear power was available every hour of every day and it would soon be “too cheap to meter” (Chairman of US Atomic Energy Commission 1954). In the 1970s, the nuclear story began to unravel, nuclear plants were nowhere near as reliable as claimed or as cheap.

Nuclear plants must be shut down for maintenance and refuelling, which takes 4-6 weeks every 1-3 years. An additional complication is that due to the huge thermal mass and the risk of fuel containment failing with rapid operating changes, a nominal two-hour repair of a simple wiring fault required a 48–72-hour power down/power up process. Consequently, in the 1970s and 1980s, nuclear availability was in the 70-80% range, not the claimed 95%.

Later, it was realised that in an emergency shutdown due to an external issue such as a turbine fault, transformer overheating, loss of transmission, etc., xenon gas was generated within the reactor and stopped the nuclear reaction. Xenon, which itself is radioactive, must be carefully and thoroughly extracted from the reactor before restart. After the Great Northeast blackout in the US, some reactors took two weeks to return to service, but 48 to 72 hours is typical.

International Experience

Then it became clear that the benefits of a common design had their downsides. In Canada in 1994, it was discovered that their design led to premature failure of cooling tubes, so 8 of 22 reactors were shut down. It took until 2014 for production to fully recover. Then by 2014, a series of upgrades began with one to three reactors offline for 24-30 months each. 2023 output was still 16% down on 1994. If Canada’s nuclear plants had matched 1994’s output till last year, they would have generated 3,100 TWh over 29 years. In fact, they supplied 2,620 TWh, they lost the equivalent of 4.5 years production

France had a similar experience in 2020 and 2022.

The British Magnox reactors had problems with graphite blocks. There wasn’t a single year where nuclear output in the UK was above 86% of capacity. For a week in January this year, nuclear power ran at 50%  This year, British imports of electricity are running at about 30 TWh/y, almost identical to the decline in nuclear output since 2016 and half decline since 1998.

In Switzerland in 2015, for a brief period, all five reactors were offline. For ten days in 2022, they ran at 13% capacity. Luckily, they have 15 GW of hydro and 46 interconnectors to other countries to back up their 3GW of nuclear power, and in that period solar supplied almost three times as much energy as nuclear.

In Belgium in 2015, output was down 46% on the 2000-2012 average. Worse, for six weeks late in 2018, four of five reactors were offline and for the whole second half of 2018 nuclear output was 61% down on historical levels. Fortunately for Belgium, they burned a lot of cheap gas. More significantly, they imported an average of 24% and up to 44% of their electricity for the half.

Before turning to the big producers, France and the US, let’s check the latest nuclear champion, Finland. In winter 2023/24, nuclear power ran faultlessly but load varies, so its contribution varied between 23% and 48% of the load. But by March for long periods, imports were larger than nuclear power output. In May, nuclear output was 40% down on January.

France

While France is a nuclear success story, it is not without significant problems. Annual nuclear output peaked in France in 2005 at 450 TWh, 79% share of generation and 81% capacity factor.  By 2016-17, problems began to appear, and nuclear output dropped below 400 TWh; by 2020 output was around 350 TWh and 67.5% market share. Then in 2022, disaster struck. A new form of stress corrosion was discovered in Civaux-1, which was only 20 years old.

Further, a record drought meant cooling water was restricted at another six reactors halving power output there, even after a temporary suspension of environmental regulations. Soon half of France’s reactors were offline. The result was that in the midst of the global gas crisis, France’s 2022 nuclear output was 182 TWh below 2005. That is the equivalent to 520 Snowy IIs.  Relative to the NEM, the reduction is equivalent to quadrupling our 2022 gas output and completely draining seventy Snowy IIs.

The NEM was in near crisis when coal output fell by 4.8% between winter 2021 and winter 2022. In July to September 2022, French nuclear output was down 43% from historical levels, so instead of exporting 14% of its electricity for those months, it imported 10%. Who will we import from?

USA

While the US nuclear system is more productive, with 93% Capacity Factor, it also has 870 GW of fossil fuels and hydro/pumped hydro and import capacity to back up the 97 GW of nuclear. That is equivalent to increasing our existing coal, gas and hydro capacity by 50% to back up 7 GW of nuclear.

Alternatively, US nuclear power works because it only supplies 18% of US grid electricity from 91 reactors. If we only want 20% of grid supply from nuclear, that means just four or five conventional reactors. With 43 coal generators, we still have problems when there are clusters of outages.

If a large number of reactors is required so that the loss of three to five at once, as has happened in Belgium, Switzerland, Sweden and Canada, is not a problem, we need a number about where the US is now, meaning roughly a hundred 80-120 MW reactors. But even then, it is no guarantee. In April 2023, nuclear output in the US was down 23% on January, and by the way, the only 120MW reactor that is even in a planning stage is the Chinese ACP100.

In Korea, which is often cited as the reference, output fell by 19% between 2015 and 2018. Similarly, in Sweden, 20% down from 1991 to 1993, 22% down 1999 to 2000, 33% from 2004 to 2009. In Japan, from 2002 and until 2003, down 27% and again by 18% from 2006 to 2008, all before Fukushima.

Annual combined wind and solar output has never fallen by that much in any country on earth.

In conclusion, a feasible number of nuclear reactors in Australia would not guarantee reliability, regardless of cost, and a practical number would require about ten times as much backup as a practical amount of wind and solar.