Nuclear cost and water consumption – The elephants in the control room

| December 20, 2019


There are four nuclear plants being built around the world where public information on costs is reasonably reliable.

These are Plant Vogtle in the US (US$27.5bn, 2.2GW), Framanville France (12.4bn+, 1.6 GW), Olkiluoto in Finland (around €10 bn+, 1.6 GW) and Hinckley Point in the UK (₤22 bn+, 3.2 GW).

There are two further plants whose power costs have been published, Akkuyu in Turkey US$127/MWh and Barakah in the Emirates US$110/MWh.

It should be emphasised that none of these costs are the full cost recovery. For example in the British case it is estimated that some $10 bn has been spent by others on upgrading the grid and backup power supplies. In Turkey the cost of the plant is just that, and doesn’t include civil works, grid connections, cooling water supply.

In the US plant Vogtle has benefited from some US$8bn of federal government loan guarantees and an unusual form of financing where customers have paid about 8% premium on their bills for 10-12 years before the plant is to be commissioned.

All of the plants get catastrophe insurance and some security from their government and most have inadequate bond structures for long term waste storage. They also rarely pay for cooling water. Many have preferential supply agreements which will require other cheaper sources of power to turn off to allow the nuclear plants to keep running.

However, even on the published information, nuclear power plants in democracies are running at about A$13m/MW.

In our case we do not have an experienced nuclear workforce, Australian construction costs are higher by 20-30% for large projects –  and there are 5,000 tradesmen on site at Plant Vogtle out of a workforce of 9,000 as nuclear power plants are very large projects.

We do not have the heavy fabrication facilities required, and these cost hundreds of millions to build  For example the Osborne Naval Shipyard design for 1/10th of the throughput of a nuclear fab shop cost $380m.  Even the inspectors would have to be imported.

So it is reasonable to suggest that new nuclear in Australia would cost at least A$16m per MW including subsidised construction finance, resulting in a first day of operation cost of a 2.2 GW plant of A$41 bn. Amortised over 50 years station life at a very low weighted average cost of capital at 5.5% – lower than plant Vogtle – that still works out at about $2.4 bn/yr.

Due to the variability of demand in Australia the plants would be unlikely to be able to achieve a capacity factor above about 80% – halfway between the US and France and higher than Korea.  So over a typical year a two unit 2.2 GW plant would be expected to generate about 15,500,000 MWh meaning the fixed costs per MWh would be $2.4bn/15.5m or $156/MWh.

The daily running costs of US nuclear plants average out at US$40/MWh.  This is lower than France and almost certainly lower than any new nuclear plant in Australia could achieve due to the much larger American skill base, higher utilisation and lower operating temperatures.  The best case for Australia would be A$60+ for maintenace and operation. Thus an Australian nuclear power station could be expected to deliver power at a cost of A$216/MWh.

Now if you use the cheaper Barrakah design at about US$5,300/MW and allow for 15 years of inflation at 1.5% to allow time for the project to come online, and a modest 10% Australian premium, power here could be produced at about A$10.4 bn per GW.

After a slightly lower capacity factor of 75%, about the same as Korea, and a realistic WACC of 6.5% the ammortisation amounts to $107/MWh with a similar A$60/MWh operating and maintenance cost and the total delivered cost of power is a mere A $167/MWh.

This figure aligns closely with the figure quoted by the CEO of the Barrakah plant some years ago at US$110/MWh

The costs of a renewable alternative

It should be noted that many of the arguments about relative costs are based on the figures used in the Finkel report. These are well out of date. Nuclear power has become even more expensive and actual renewable contracts in Australia are down 40-50% on the Finkel figures.

Thus if we dispersed 2 GW of wind $3.6 bn, 1.2 GW of tracking solar $1.8bn, 2 GW of rooftop solar $2.5 bn, 1 GW of waste/biomass/geothermal $2.5bn and 1 GW/15GWh of pumped hydro $1.8 bn and 1 GW/ 2 GWh of batteries $1.2bn across the NEM the total cost would be $13.5 bn.

Annual generation would be 17,500 GWh – more than the nuclear plant – and minimum available output would be 2.5 GW+.  Typical hot day peak demand at 5pm would be about 4GW.

About 30% of generation would go through storage at 85% efficiency, so net output would be around 16,500 GWh.  Some would be curtailed so we can assume a similar annual output to the nuclear plant.

However the operating costs average around $18 and the capital, even if amortised over 30 years are only $59/MWh for a total of $77 including backup.

In summary, for 1/3rd of the investment, in one third of the time, we can get renewable power and backup for 1/3rd of the cost of nuclear power.

Cooling Water

A key issue with nuclear plants is cooling. Because of the cost of shutdowns and the degradation of materials by irradiation, the plants are designed to run at lower peak temperatures (260-320 C) than coal (500-670 C), gas turbines (1,300-1430C) or internal combustion plants (2,000 C).

The thermal efficiency of a plant is directly related to the difference between the peak temperature and the cooling medium – what is termed Carnot efficiency.

Lower temperature means lower efficiency, as less of the heat energy is converted into work and more is removed by the cooling system. So for a given amount of electrical energy delivered, more cooling is required in a nuclear plant. Furthermore the warmer the cooling water or air the more coolant is required.

Thus the Barrakah plants require 100 tonnes of Gulf seawater per second for each generator. In higher latitudes with seawater temperatures in the range of 2-12C, water requirements can still be 40-60 tonnes per second per GW.

Just to put that in perspective Melbourne Water supplies 15-20 tonnes per second to the entire Metropolitan area of almost 5 million people. Even so, the water temperature is raised by 7-10C which is enough to kill any fish larvae unfortunate enough to be sucked into the cooling intakes.

It is enough to change the local environment for all sea life, so finding a suitabable site is very difficult. There are currently no nuclear plants operating using warm seawater for cooling although Barrakah is soon to be commissioned.

The problem there is not just the temperature but the accelerated rates of corrosion and biofouling which will mean the heat exchangers need to be larger, pumping losses will be higher and maintenance bills higher still.

Perhaps the area near Portland in Victoria might work, but then the 500kV line would have to be triplicated to carry away the power, further adding to the cost. Plants at the edges of the grid have a whole lot of other issues so a South Australian plant would be extremely difficult to integrate.

On land in very cold climates, a small number of air cooled plants have been built but the offset is that about 5% of the output of the power plant is used to run the fans. However in warm climates it is virtually impossible to run an air cooled nuclear power plant.

It would require in the order of 450-500 tonnes of air per second to be moved over the heat exchangers per GW of electrical output. At typical air velocities for cooling fans that would have a fan area of 75,000 square meters or if each fan was the cross section of a shipping container, 17,000 fans.

In other cases straight through cooling is used from large rivers or lakes. The Murray at the South Australian border is often down to 9 GL/day or even less. 9 Gl/day is about 105 tonnes/second, and so a single unit nuclear power plant located on the Murray would often need the entire flow to cool it, while heating the water by 8-12 C.  This is obviously an environmentally impossible situation.

That is why cooling towers are the most common cooling method because they are the most efficient. Evaporating water carries away much more heat than liquid flows.  In typical Australian conditions the nuclear plant would evaporate between 20 and 24 GL per year per GW so a two unit 2.2 GW plant like Plant Vogtle currently under construction in the US would need 44-50 GL/ year.

That is more than the 4.7 GW of coal in the Latrobe Valley and almost 30% more than the entire demand served by Barwon Water which includes 320,000 people and all their business homes, parks and gardens. At current spot prices for irrigation water that would be an additional cost of $50m per year.

In summary, in a hot dry continent like Australia, providing cooling water for a nuclear power plant would prove a huge cost and distortion to the water industry.

There are many other issue with nuclear power, including a lack of flexibility, large and long duration backup requirements for refueling and outages and large spinning reserve requirements, but these can be explored at another time.

A closer look at Barrakah

There are a range of risks with all nuclear designs, but the business risks assoctiated with the Barrakah style APR 1400 seem even larger than most.

The Barrakah plants were supposed to progressively come on line in early 2017 but they have yet to generate power. This delay is adding US$1.2-2 bn per year to the eventual liabilities that have to be paid off.

They are designed for an 18 month refueling cycle – unlike the AP1000 at plant Vogtle which has a 3 year refueling cycle. This means lower lifetime capacity factors and higher backup requirements with gas or pumped hydro. The design goal is 90% availability.

They have largely been built with very low finance costs from both Korea and the Emirates together with cheap expat Indian and Pakistani labour which significantly understates their real cost of construction.

The Barrakah plant is a 4 unit plant, which allows useful economies of scale, and there is nowhere in Australia where a 4 unit plant can safely be intergrated into the grid.

Recent problems with the single unit 750MW Kogan Creek generator in Queensland have shown that the grid can be destabilised with the failure of a single unit.  As demand is gradually falling, a single unit of that magnitude is even harder to manage. The APR 1400 units are 1,350-1,400 MW so would be even more difficult to integrate into the grid.

These reactors have not yet been shown to work in a hot environment so their reliability is unknown, in fact there is only one other reactor of this type operating in the world with two more under construction.

The Moorside project in the UK which was to use KEPCO designs has been abandoned and plans for two more units in Korea have been frozen. KEPCO was offered all the development work already done on the Oldbury and Wyfla plants in the UK and did not take them up.

These plants came with billions of pounds worth of development work already done, project teams and permits in place and an offer from the UK government of a guaranteed 75/MWh + inflation for 25+ years.

There is a reasonably held belief that the price was artificially supported by the previous Korean government which viewed nuclear technology as a new export industry and this project as a flagship demonstrator. In contrast the current Korean government was elected on an anti-nuclear program and has pledged to build no more plants after the current two units under construction are completed.

There are some doubts about the level of safety in the design and a new design, APR1400+ was developed to reduce the possibility of a melt down. However no plants of this design have been ordered. So which one would you choose?