Is nuclear power a solution for Australia?

| September 6, 2023

This is an update on an article originally written in 2019. The most surprising change since then is the lack of progress in global nuclear energy. Although there are many more proposed nuclear plants and SMR designs, output from nuclear energy last year was less than in 2019 and new capacity barely exceeded retirements, while SMRs are still nine years away.

People often ask why Australia does not have nuclear power because we have a lot of uranium. The first part of the answer is that the question is like asking why don’t we make semiconductors, we have a lot of sand which is the ore for silicon or perhaps even simpler why don’t farmers control the bread market? The value of refined uranium ore (yellowcake) that we export is around 10% of the finished fuel assembly and in turn the fuel is less than 10% of the cost of nuclear power so in effect we control about 1% of the cost.

Another refrain is that the rest of the OECD is expanding nuclear why don’t we. The problem with this argument is that outside China, despite much vaunted openings in the US and Finland and plants under construction in the UK, France, Pakistan, Bangladesh, Turkey etc, world nuclear production is falling. Global nuclear power share of generation peaked in 2002 at 17% of electricity produced. Output peaked a few years later in 2006 but growth was already slower than overall demand, so its share had dropped to 14.7%. Since then, nuclear output in 2022 was slightly less than 2006 while overall demand had grown 44% so last year nuclear had fallen to 9.2% of total electricity production.

It is likely that this year nuclear power will be up a little, but growth will still be less than overall demand so market share will continue to decline. No nuclear plants ordered since 2020 will come on line this decade and so far connections are barely outnumbering retirements. Even if net additions accelerate, but demand growth continues at 3.5%, by 2030 nuclear power will be barely 7.5% of supply. In contrast wind and solar which in 2006 was 5% of nuclear output is growing so rapidly that it passed nuclear in 2021 and is on track to be 45% greater than nuclear this year and provide 13.5% of global demand. By 2026 wind and solar will double nuclear output.

Load Following

A key issue with nuclear is relatively poor load following. Power can be ramped from 50-60% to 100% and back reasonably quickly (hours not second or minutes) but it badly affects their lifetime and economics. Due to the high fixed costs and the effect of cycling on increasing maintenance and shortening life, it can be shown that a nuclear plant running at 60% capacity has twice the lifetime cost of power of the same plant running at 90%.

Therefore, nuclear plants always need support for fast demand changes. In France and Japan this is provided through hydro and pumped hydro and or export/import. France imports at peak and exports off peak. Japan has 27 GW of pumped hydro for 53 GW of nuclear and it ramped coal and gas plants up and down to minimise load changes on nuclear. The North American method is to limit nuclear power to around 15-20% of supply so that even at minimum demand, all available plants can run at or near capacity.

In the Australian situation, none of these options work, we can’t export, we don’t have enough hydro and as minimum grid demand is now starting to drop below 11 GW, the maximum number of 1.1 GW plants that we could keep running at 75%CF is about 10-12. If two or three of those happened to be offline at once on a high demand day, we suddenly have a full-blown power crisis. Victoria had a power crisis when it lost 1,700 MW of coal, what would happen if it lost 2,200 MW of nuclear.

In 2015 five out of five Swiss reactors were offline for 10 hours, in October 2017, 21 of 58 French reactors were offline for about six weeks. 40% of Belgian capacity has been offline regularly over the last few years. In June last year, Swiss nuclear power ran for ten days at 13% of normal output at one stage supplying 4.3% of demand vs an average of 42%. Last year for the whole year French nuclear produced 22% less than the year before and 34% less than 2005. In 2018 Belgian nuclear power ran to 80% below normal for eight weeks. How do you replace that, – two hundred Snowy IIs?

Load Variation

A further problem for all grids is the growing variability of the load. In the past, the base load was 40-50% of the maximum load. In Victoria and SA now, minimum load is less than 20% of maximum load and trending lower (minimum load in SA has been less than 5% of maximum). That means that while one or two coal or nuclear plants might run at 80% capacity factor, on average the whole generation fleet can’t average much above 45-55%.

Hydro and gas plants run between 1% and 25% of the year. That is acceptable because they can start and stop quickly, but nuclear plants can’t, so once nuclear penetration gets above 15-20% of annual supply, storage is needed to absorb the excess when demand is low and fast response is needed to meet the peaks when demand is high.  (See Japan above) Thus, just like wind and solar, nuclear must have matching fast response reserves.

However, the experience of France, Belgium, Britain, Canada, and Switzerland have shown that although nuclear outages are less common than renewable outages, the duration is far longer, so to achieve 99.9994% reliability that we aim for, a nuclear system may need slightly less backup power than a wind/solar system but in the order of ten times as much energy storage.  For example, from March to December last year France would have drained two hundred and thirty Snowy IIs, to make up for the fall in its nuclear output. In June 2022 Switzerland needed an extra 3GW/240 GWh over ten days

Emergency Support

Then there is the question of emergency support. There are many types of backup but basically, we can talk about short, medium and long term. Short term, often confused with inertial response is what happens when there is a sudden load or supply change.

Every system must have enough “spinning reserves” i.e. underutilised capacity, running on the grid in case of a supply (generation or transmission) failure. Therefore, a 1GW generator needs 1 GW of spare capacity in the system. They can’t just be ticking over, if the reserve plants are idling and load comes on instantly, it is just like dropping the clutch on a manual car, the system stalls.

That is exactly what happened during the blackout in South Australia. The Quarantine plant that was being paid to be on standby, failed when it tried to start too quickly. That means if there is a single 1.1 GW nuclear or even coal or CC gas plant on the system, there must be multiple gas, hydro, coal, nuclear plants running at 35-65% capacity with at least 1GW spare capacity. Thus the other plants must be contributing 1.5-2.5 GW.

So, for example in SA with a single 1.1 GW plant running, a minimum of 3-4 GW generators should be running and producing at least 2.5 GW depending on the mix of backup plants. SA minimum demand is 200 MW and falling, so on balmy nights with light wind SA would have to export 3.2-3.6 GW two to three times the current interconnect capacity.  On windy spring afternoons even more. That assumes the other states can absorb it.

If there are 15 or 20 gas, coal or hydro generators up to say 300 or 600MW, when there is a fault at one unit, the others can cover it and depending on the size of the fault it might be fixed in minutes or hours. A nuclear plant takes hours to shut down and many hours to restart so for example a simple steam leak which just required a 4 hour repair on the Pilgrim plant near Boston required closure for 3-4 days. Thus even 20 hour pumped hydro system will not provide enough backup.

If the plant is tripped and does not shut down according to proper procedure, a phenomenon called Xenon poisoning means that it can take 72 hours to get up to full power and in some cases, it has taken 3-4 weeks. Finally, most nuclear plants shut down for four to six weeks every 12-36 months for refuelling and sometimes six months to three years for major overhauls, so again large capacity long duration reserves are required.

Thermal Capacity

Apart from huge backup requirements, a further difficulty is that in hot weather the capacity of all thermal plants is reduced.  So, when we look at a northern hemisphere plant rated at 1,100 MW at 15C, at 42C in Australia it will deliver between 950 and 1,000 MW therefore we must spend even more money to get peak capacity.

A particular issue with nuclear is cooling water, A single 1.1 GW nuclear plant, (the most common size) needs 20-25 GL of water per annum. that is about 7% of the water supply for Melbourne, a fully nuclear system would need 4-5 plants to supply Melbourne so increasing fresh water demand by 30-35%. The plant could be cooled by seawater, but that requires more pumps and much higher maintenance due to fouling and corrosion.

The nuclear plants at Barrakah each pump 100 tonnes of water per second through their cooling systems and the hot water exiting the plants tends to create dead zones in the sea for 100s of meters around the outlets. Fouling will be much more common here than in Northern countries because higher water temperatures mean much faster growth of marine life.


Then there is the question of time. There are four companies in the west capable of building nuclear reactors, EDF (France), Toshiba/Westinghouse, Hitachi and Korean KHNP. and their current projects are way over time (4-14 years) and budget. KHNP is heavily subsidised and although apparently cheaper, no-one knows what the real cost of their plants are.

However, it is known that their latest plants on established sites have taken over 10 years to commission from the date of construction start.

Toshiba and Hitachi have both abandoned new approved nuclear plants in the UK after spending billions and not being able to make the finances stand up even at a guaranteed price of £75/MWh + inflation, about A$200/MWh by the time they were scheduled to open. Korean and Chinese manufacturers declined to take over the projects even when offered all the preliminary work and established workforce for nothing. Despite offering almost £1,000m in incentives and co-pays, the UK has yet to conclude a firm contract with EDF to build Sizewell C, twelve years after the site was initially approved.

Creating a Regulatory Commission

Australia doesn’t have the equivalent of the Nuclear Regulatory Commission and that will take 3-4 years to legislate, set up and staff, the UK has 700 experienced staff in their equivalent and their nuclear system supplies the equivalent of 15% of Australia’s electricity demand. Then operating licences typically take 3-4 years to issue, 2-6 year for financing etc and 12-14 years construction hence 2048 is the earliest likely time for the first power from a large-scale nuclear plant.


Then there is the biggie, cost. Eleven existing nuclear plants in the US have closed or are closing before expected because they can’t compete, even just on operating and maintenance costs with gas, wind and solar, the latest is Palisades Michigan. State governments are subsidising a further 3-5 to stay open. The US was building 4 plants, which are way over time and budget and aside from the $5-6b losses by the contractors, the plants are so expensive that full cost recovery will require a lifetime price of around US$135/MWhr. Two of those plants were abandoned after spending US$9bn.

The Hinckley Point power station in the UK is guaranteed a price of A$171/MWhr in 2012 prices, so by the time it opens in 2028 it will be earning about A$300/MWhr. Again, this is on existing site with cold cooling water, better load factors and lower financing costs than Australia. To put it another way at current American/French/British costs, Nuclear is costing about US$12bn/GW which will provide between 7,500 and 8,100 GWhr per year. Gas backup will be about another US$1.5b. If pumped storage is used that would be about $2-4bn/GW

Making the completely unrealistic assumption that we can build a plant for the same price here without the skills, experience or industrial capabilities of the Americans and neglecting 17 years of inflation from the start of Plant Vogtle to the start of a plant here, means a Capex of A$21-23bn for a 1.1 GW plant.  To provide 8,200 GWh per year from a combination of 1.3 GW of wind/ 2.2 GW of solar PV and 2 GW/8GWh of batteries at today’s price will cost A$9.5bn.

The system will have summer peak capacity of 2,100 MW vs 980MW for the nuclear plant and it will have 40-80% response time of 0.3 seconds vs 30-90 minutes for the nuclear/ gas combination. Operating cost according to the US Department of energy is less than half the nuclear plant.  Not only that, if the decision was made this year to proceed, the whole system could be operating by the end of 2028, two or three years before the first sod might be turned for the first nuclear plant.


One Comment

  1. Alan Stevenson

    Alan Stevenson

    September 6, 2023 at 10:25 am

    Like many people I know very little about nuclear power plants. However, as discussed here, a plant requires a tremendous amount of cooling water. The plant operates by producing a lot of heat which is transferred to water in order to make steam for the turbines. That transfer procedure takes two steps – direct transference to water in an enclosed system and then a re-transference to a secondary supply for the generators. This is done to prevent radioactive transfer to the outside world. The core of the reactor is hot enough to melt iron and requires constant additional cooling. This hot water (thousands of gallons per minute) is then dumped into the environment (usually the sea). We are currently having problems with warming oceans. Wherever a reactor is built it will probably be on the East coast or South Australia. The currents will send the hot water up our coast to the Barrier Reef.