Since its inception in 2018, the annual CSIRO Gencost (GC) report has become a major media event in Australia. Few are interested in the esoteric details of electricity generation, but there is always interest and controversy over GC’s treatment of nuclear energy; banned here since 1988. GC always finds nuclear energy to be much more expensive than renewables, but, despite this, nuclear support keeps growing and opposition keeps shrinking. Five recent polls (1,2,3,4,5) have shown that those strongly opposed to nuclear energy are now a smallish minority (below 20 per cent). A recent 2024 Lowy poll, with a sample size of 2,600, found that even 46 per cent of Green voters support nuclear energy (either strongly or somewhat).
Enter GC 2024. This report purports to compare electricity generation costs of different technologies by massaging numbers to provide a fair comparison; a level playing field. Each year as the GC report drops, the anti-nuclear hard core erupts in a chorus of “See we told you so! It’s too expensive and too slow anyway”.
GC is used by the Australian Energy Market Operator (AEMO) Integrated System Plan (ISP) to design a pathway for our electricity generation and distribution systems between now and 2050.
The concept of levelising the cost of different technologies is intrinsically flawed when the technologies have large operational and technical differences. So here’s how the GC and the ISP work together to bolster the case of the 20 per cent who are still strongly opposed to nuclear energy.
First, think about the cost of providing transport in a city. To keep things simple, we’ll ignore the roads and just think about the vehicles, which would correspond to generators in an electricity system. If the city was entirely served by public transport, you would simply add up the cost of all the buses, trams and trains. But what if a majority of the population went and bought a car? Sticking with the same calculation and ignoring the cars would falsely imply that the cost of transport had been slashed. It’s intuitively obvious that the total transport price has increased; massively. I’m not making judgement on the convenience of the two systems, just the price.
That’s the first big GC/ISP trick. They presume everybody spends many thousands on their own infrastructure but then excludes that from the cost of the system as a whole.
Second, how do you compare something like a nuclear reactor that can run for 60 years with renewables which don’t?
Consider the age-old choice between renting and buying.
You could compare the levelised cost, per room, over the first 30 years of both renting and buying. But that ignores the second 30 years or so, when home ownership shines brightest. After 30 years of paying off your mortgage, you have a valuable asset where you can live both rent and landlord free until you can’t get your walking frame over the threshold.
Buying renewables is very much like renting. At the end of 30 years of renewable ownership, all you have is a widely dispersed collection of detritus which would, at best, require a large and costly effort to collect and recycle; assuming this was economically viable. But reactors are there for the long haul. Some 67% of Australians have bought or are buying a home. Many of the 33% would like to join them. GC was designed for the rest.
Gencost24 (GC24) has an FAQ section in which they discussed what they revealingly called a “perception” of cheap electricity prices in countries with nuclear plants. They wax lyrically, but irrelevantly, about the difficulties of comparing prices between countries; but forgot the obvious. Namely, countries which invested in nuclear 30-50 years ago now have an incredibly valuable asset base which just keeps pumping out electricity long after the mortgage is paid.
That’s the second big GC/ISP trick.
How much do these tricks distort the comparison?
GC put the cost of a large nuclear reactor at $8.5 billion, with a build time for the first one of 15 years (including regulatory creation). Under the Step Change scenario, the ISP suggest we will have 144 gigawatt-hour (GWh) of consumer batteries on the grid by 2050, with 88 GWh of them installed after 2040.
GC estimates the current cost of these batteries as $1455/kWh (GC p.58), about double that of the grid scale batteries. That exposes the cynical nature of the GC/ISP cost shifting. Putting 144 GWh of storage at scale on the grid for everybody’s use would be like providing public transport and cost half as much, but it wouldn’t enable GC to ignore the costs. It would also be considerably safer. A big utility battery fire is expensive, but rarely dangerous. But a battery fire in a house or block of flats? That’s a very different issue.
Do the maths, 144 GWh of batteries at $1455/kWh is $209 billion. Of that, 88 GWh, will be installed after 2040, for $128 billion. A detailed analysis would take a stab at guessing cost reductions, but $209 will still be a minimal estimate because GC also hides all the costs of household photovoltaic (PV) panels. In addition I’m conservatively ignoring the costs of all of the batteries installed before 2030 which will need to be replaced by 2050. In summary, even without all the tedious details, two things are clear, renewables are incredibly expensive, and we won’t be finished with the buying of them by 2040, which some would argue would make nuclear redundant.
Now let’s imagine we take that $128 billion due to be spent on batteries after 2040, and invest it on nuclear energy now. At $8.5 billion per reactor, the supply chains could be sorted by 2040 (and possibly before) and we could have 15 reactors rolling off them during the 2040s and go a long way towards setting Australia up for clean energy security through to 2100. With $200 billion, we’d get more than 23 reactors.
It’s worth noting that the real issue with building nuclear power isn't the time for individual reactors, but maximising the parallelism in the supply chain so as to minimise the fleet build time. But again, I can’t help thinking like a homeowner, instead of a renter and clearly, for CSIRO, home owner thinking is anathema.
Clarification on parallelism
Some people have asked me to clarify the final paragraph. Here it is.
Suppose you start your first nuclear reactor at the beginning of some year. You want to think about the skills and people involved so that you can begin your second one in 6 months time. Having two entirely separate teams isn't sensible. Then you want the third reactor beginning 6 months after that. Again, this is all about skill sharing and project management. Planning that kind of fleet building program isn't the same as planning a single reactor in isolation. If you look at the Japanese nuclear rollout, this is exactly how they did it. They had about 9 reactors underway in parallel at some points. You need to find the bottlenecks that will limit your parallelism and either expand them or fit you effort to match them. If you can start each new reactor 6 months after the previous one then you can build a fleet of 10 in 15 years. The principle is no different from planning a fleet of wind or solar farms. The big difference with the nuclear build is that there are fewer bottlenecks. That will be the subject of another post ... stay tuned.
Brilliant stuff Geoff, many thanks