Nuclear may (or may not) be expensive, but it's much faster to build than renewables!

This is the first of two posts, and deals with the speed of building nuclear reactors. The next post will deal with costs.

Build speeds and bullshit: the Australian Energy Council

It's a recent article from the Australian Energy Council's Hamish Fitzsimmons and purports to show how slow nuclear is to roll out. His argument rests heavily on a reproduced graphic from a report by the Institute for Energy Economics and Financial Analysis (IEEFA) claiming to show that reactors take "more than twice as long to build as the owners projected". I note that the IEEFA website talks about its "non-partisan evidenced-based" approach.

Here's the graph:

At the right hand end are four US reactors; and yes they are taking a very long time, being the first reactors built in decades in the US. But, beyond that, I'd be betting that the US will never finish a new reactor until the US Nuclear Regulatory Commission (NRC) is replaced by something not totally dedicated to the obstruction of the industry, as it has been for the past four decades. I'll say about more about the NRC when I discuss costs in the companion post to this piece.

The graph also shows 7 Chinese, 1 Finnish, 1 French and 1 British example; with the unproven NuScale SMR estimated build time on the far left. The French, the Finnish and two of the Chinese reactors in the graph are EPRs. Could it be that this reactor design simply sucks? As it happens, yes.

The EPR is a wonderful example of what Fred Brooks called the "Second System Effect", where you throw every feature, including the kitchen sink into a new design in a grandiose display of power and hubris. There are four independent cooling systems! We put hundreds of people thousands of metres up in the air in planes with dual safety systems; why would nuclear reactor need four. The consequences of aircraft failure are certainly cheaper than reactor failures, but far more deadly. The EPR design is so devilishly complicated that not even the production engineering geniuses in China can build it quickly. The designers themselves have acknowledged this and have long been working on a better design.

Brooks was a computer scientist but the Second System Effect describes something universal in the human psyche. But not all Chinese nuclear plants are EPRs.

But before looking at the Chinese nuclear build speeds, let's think about how to compare the rate of building nuclear reactors and solar farms. Here's a first rough guide. Construction on the Bungala Solar farm began in April 2017 and the plant seems to have been completed in November 2018. That's 19 months to build something that supplies 570 GWh of electricity per year, but only when the sun is shining. The UK Hinkley Point C looks like a slow build, it may take 10 years, or more. But it will supply 45 times more electricity annually than Bungala and supply it on request, not just when the sun is shining.

Now let's look more carefully at the Chinese examples. Why choose those 7 reactors? What about the rest? Below is a graph showing the construction time from the start of construction to the grid connection for the 54 Chinese reactors currently in operation. The data is from the same source as the graph above, but without the "non-partisan evidence-based" cherry-picking of the IEEFA.

The median build time of these reactors was 5.6 years with an average of 5.9 years. I've colour coded the bars by reactor size. The two lightest bars are EPRs and yes, they took almost a decade to build, but other reactors also had construction delays. The "small" Shidao Bay 1 reactor is the first Chinese SMR, I'll say much more about it in the costing post. There are big plans for this reactor, spending time tuning the design isn't surprising.

Note the pair of numbers on the end of each bar below (n:r). The first number is the size of the reactor in Bungala-units. So the first reactor, Hongyanhe-6 will generate 15 times the electricity annually of Bungala. The second number, 3.6 in this case, is how much faster the nuclear build is (per Bungala unit) than the actual Bungala build. I.e., Building 15 Bungalas in 6.7 years is 3.6 times faster than building 1 in 19 months ... it's about 5.2 months per Bungala. Only one reactor has a ratio below one ... meaning slower than Bungala.

Keep in mind that it's really not very sensible to compare dispatchable nuclear power with solar power. That's like comparing a Tesla that you own and can use whenever you choose to a car that runs only when the sun is shining and will slow down instantly if a cloud obscures the sun.

Of course, a robust comparison would consider that powerplants of all kinds can be built in parallel. How can a comparison take this into account? That's the topic of the next graphic.

A more robust comparison

The next two charts are complicated but I assure you, they will repay the effort put into understanding the details!

How were they built? BP's Statistical Review of World Energy is published annually and has details on energy (not just electricity) for all technologies going back decades. They are one of two big primary data sources (the IEA being the other). The following graph is built by looking at growth of electricity by source technology and converting to an amount per person in each country. Then I look through a rolling 20 year window to find the period with the most growth per person for each country; then I pick the top 20 countries. Here's the result. There is more explanation to follow, but first, just have a look.

Each row heading has the rank, the name of the country, the end year of the 20 year period and the population. So the top ranked technological electricity roll out was Iceland between the years 1995 and 2015. Iceland rolled out 24 megawatt-hours per person of hydroelectricity and (in row 2) another 15 megawatts per person of geothermal in this 20 year period. Iceland is a small country and it only takes a couple of large multi-national projects to achieve such things. Iceland is an outlier and of not much use as a guide to what is possible in the rest of the world.

Norway is also blessed with abundant and easily coopted hydroelectric resources, so of limited value as a model.

Next we have the nuclear rollouts of Sweden, France, Finnland and Belgium; with the UAE's gas rollout splitting them.

There are no renewable rollouts until we get to Sweden in 13th position with her wind+solar build in the 20 years to 2020. Sweden's appearance also in position 17 shows that most of that rollout was wind. The two rows can't be added together. I just wanted to treat wind+solar as a category in its own right as well as the individual components.

There are 6 countries with nuclear rollouts with the rest of the top 20 dominated by fossil fuels; 5 gas and 3 coal.

What are the asterisks (*) and solid dots? The asterisks are the amount of clean electricity in the country, measured as MWh/person/year. The non-solid dots are the current electricity supply (including both clean and fossil fuel electricity). Most countries are in the range of 10-20. These countries are all rich first world countries. The solid dots are amounts of electricity consistent with the IEA NetZero by 2050 plan. You can see again how big an outlier Iceland is. With just 330,000 people, it makes more aluminium than Germany.

Sweden is the only country to come close to the rollout speed of nuclear power, and it's wind+solar rollout has been less than half the rate it rolled out nuclear between 1975 and 1995.

Clearly, those people who think nuclear power is too slow to build are recommending something which is even slower; wind+solar. Keep in mind that the wind+solar rollouts to date have been the easy bit of the task. The more challenging aspects of dealing with grids with 80-90 percent wind+solar are yet to come. With nuclear, there are no challenging integration issues.

And again, this time with feeling

Here's the same chart but with a few modifications. First, the fossil fuel technologies have been removed.

The dots are the same as before. The shaded bar shows the amount of clean energy required to hit the IEA NZ2050 target. After each bar, I've calculated how many times the country will need to replicate its fastest 20 year rollout. So if Finland at row 6, will need to repeat it's 20 year nuclear build 5.8 times to get to the IEA target. France will need to repeat it's nuclear build 1.6 times to hit the target.

The graph makes it clear how hard it will be to hit the IEA targets. It also makes it clear how pathetically slow the wind+solar rollouts have been. Innumerate journalists love to quite percentage growth figures without realising how silly this is. It's easy to grow by 100 percent when you are trivially tiny. After 20 "big" years, the global solar industry is still trivially tiny despite all those rapid growth stories. The truth is that solar has been incredibly slow. Here are the numbers again ... in case people missed them. They are based on the 2020 BP stats, but the big hitters are still the brown and grew blobs. It's time OWID updated this!

Hitting peak electricity technology rollout speeds

Clearly, most countries, other than countries which are already nuclear, are starting from a very low base of clean electricity. How do we all build clean electricity fast?

The SMR strategy is to build reactors in factories (or shipyards in the case of Thorcon). Renewables have shown us one folly of such a strategy. How many countries have solar panel factories? Very few. Many countries have ceased production as cost conscious consumers vote for the cheapest products. The factories and supply chains have become bottlenecks. We can't afford to let costs dominate our clean energy build strategies. Speed has to be prioritised over cost. Which doesn't mean costs don't matter, it just means we have to be sensible and not "penny wise pound foolish" as my mum used to say :)

The sheer volume of material required for renewables creates a second bottleneck; exacerbated by the need for batteries. We will need batteries, but we shouldn't waste them.

The old bespoke nuclear build strategy might look slow, but it could possibly have fewer production bottlenecks. The Rolls Royce "SMR" is a mix of distributed and factory production that is very interesting. Some parts are made in on-site factories. That looks like it might maximise parallelism. The Thorcon route is to build reactors in shipyards because these are incredibly well suited to rapid throughput of high quality steel fabrication. We will need all the good ideas we can find. What we don't need is to waste time on technology that is simply cheap, nasty and slow.

Currently, too many people are preoccupied with plugging the holes in renewable intermittency rather than recognising that it is simply bad technology; you can add batteries to a solar panel, but it's still just a solar panel.

The next post is on costs.

Postscript ... 10 year rolling window

A tweet suggested that I'd deliberately chosen a 20 year rolling window to avoid the recent "rapid" rates of renewable growth. Not true. Here's a chart using a 10 year rolling window and the top 30 countries.