The submission by The Australia Institute (TAI) to the South Australian Nuclear Fuel Cycle Royal Commission includes discussion of time taken to build nuclear power plants. The mean time shown is 9.4 years. The conclusions drawn by TAI are:

“If Australia begins to develop a nuclear power industry, build times are likely to be long enough that renewables and storage will be well established long before completion, invalidating any realistic business case which is, if European nuclear power profitability and current levelised cost of energy is any guide, already tenuous”.

This analysis is flawed. The given mean, 9.4 years, is mathematically correct based on the data shown. However TAI have not interrogated and analysed the data adequately in order to provide a robust conclusion about potential nuclear build time for Australia.

The data set for this finding is shown by TAI in Table 1 on page 8 of the submission. I have reproduced it here.

Country Units Mean time Min. time Max. time










South Korea








































Useful research needs to clearly define the question under investigation and then both interrogate and analyse the available data set. Let’s assume the question here is “How long does it take to build nuclear reactors?” in the context of informing discussion in South Australia.

In this data set covering 40 reactor builds in nine nations there are six obvious outliers at the high end with builds of 24-years or greater. There is a gap of 13 years build time between the lowest of these outliers and the next longest reactor build (11.2 years). These outliers will skew the mean to the high end.

What should be done with them? They could be excluded. Given (a) the gap between the outliers and the rest of the data set (b) that they are relatively small in number and (c) three of the data points are the only build in a given nation, exclusion seems reasonable prima facie.

It would be better to interrogate the outliers, understand what caused them, and either exclude on a clear justification or attempt to correct the data for any identified confounding factors.

What are confounding factors? Borrowing a definition from epidemiology:

“Factors that can cause or prevent the outcome of interest, are not intermediate variables, and are not associated with the factor(s) under investigation. They give rise to situations in which the effects of two processes are not separated, or the contribution of causal factors cannot be separated, or the measure of the effect of exposure or risk is distorted because of its association with other factors influencing the outcome of the study“.

In this data, understanding the outcome of interest (nuclear build time) from the available data may well be confounded by factors that have nothing to do with nuclear build time (i.e. not associated with the factors under investigation). Could that be the case in this data? Let’s look at the outliers.

  • Argentina: Like many economies in South America, Argentina has experienced repeated economic and social upheaval. I discuss this example further below
  • Romania: Romania was part of the Warsaw Pact, under dictatorial rule for over forty years, which ended in violent revolution in 1989,
  • Iran: No comment required. Including Iranian nuclear build in this dataset without question is clearly untenable.
  • Russia: The collapse of the Soviet Union in 1989 was probably the greatest social, political and economic upheaval of the second half of the 20th century

In all of the outliers, the metric of build time, measured from start of construction to end, has been grossly confounded by social, political and economic upheaval. These data points tell us very little about nuclear build time. They may tell us something about the impact of such upheaval on large infrastructure projects. It would be an interesting question to examine, separately.

One might try to correct this data for the confounding factors. One way might be by specifically measuring time in active construction as opposed to calendar time between start and finish. Take the Kirchner reactor in Argentina (Atucha 2). It was not a 33 year build as suggested by the data used by TAI. It started in 1981, in the last days of a military dictatorship. It proceeded slowly due to lack of funds until 1994 when it was suspended. The Argentine economy collapsed in default in 2001, began recovery a few years later and has been quite strong since then. Atucha 2 was revisited in 2006 as part of a strategic plan for the nuclear sector in Argentina, and reached full power in February 2015.

So, with all that interesting history one could attempt to apply a corrected figure for Atucha 2 and the other outliers. However for the sake of simplicity I will exclude these data points as too confounded to inform the question being asked. When these six data points (of 40) are excluded, the mean drops from 9.3 years to 5.8 years, a 3.5 year difference. With those exclusions and the relevant justification, the data set now appears to me to be more informative and representative of the potential Australian situation: a stable nation that may embark on a nuclear build program.

No one could seriously contend that a mean time of 5.8 years works against the argument for nuclear as a deployable source of clean energy. This is particularly true when considering the quantity and reliability of that supply.

In a metric pioneered by Geoff Russell and further developed at The Breakthrough Institute, nuclear has been decisively shown to be the fastest pathway to adding new energy anywhere, ever. For example South Australia’s wind sector (which I support) has incrementally (but, by most observations, quite aggressively) developed over a 12-year period to now provide a variable supply of about 3,000 GWh per annum. A single CANDU reactor would deliver about 5,500 GWh per year in a reliable, dispatchable form. They may deploy differently however one cannot form a “time to deploy” argument against nuclear without forming the same argument against wind.

In coming years there will be more data. Most of the builds coming from China and South Korea will pull that mean down. The current builds in Europe will pull it up. The current builds in the US will hover around the current 5.8 year mean.

The build program of the UAE will pull the mean down and that’s interesting for Australia. That is a new nuclear nation, not an established nuclear nation, delivering outcomes among the best in the world.

All of this information provides important guidance for Australia, provided we look not only at the numbers but also behind them.

Numbers are useful and they can also be misused. The role of the researcher and analyst in nuclear remains a crucial one. TAI need to apply more rigour to be taken seriously in this space.


  1. Re implementation of a nuclear program in a new country, I have recently added to the WNA info paper on UAE the following para:
    Before the UAE implemented its nuclear power program from 2008, it was considered that such new programs would be developed sequentially and slowly. The UAE demonstrated that it is possible to proceed faster by doing a number of things in parallel, by using experienced expatriates initially and transitioning to local expertise over time, and by committing to an experienced reactor and power plant builder with a track record of on-time and on-budget performance. It is on track to have its first reactor operational within nine years, using KEPCO expertise in running it.

    1. Thanks for the great research and outreach work you do Ian, and for providing authoritative public resources for reference.

      I know you are as busy as the rest of us, but how about expanding your outreach to Twitter?

      Best wishes,

  2. Given strong opposition to the construction of a nuclear reactor in Australia, I think ten years is bravely optimistic. Sure, from first site clearing to last water tap installation may very well take less than that, but the time in courts, conducting an EES, then another one, developing our own special snowflake of a regulatory authority, these would take a very long time indeed. For just about every reactor (so apart from the setting up of the regulatory authority).

    Not wanting to be gloomy…

    1. Depends if we’re talking application to criticality, or construction to criticality. I can see 10 years for application to criticality, but construction could be less.

      This all depends on the regulatory process of approvals. Whether it is a US version where objections can be raised during construction delaying the build (increasing financial risk), or a UK version where all the objections are done at the planning stage before construction begins.

  3. According to their website The Australia Institute is the country’s most influential progressive think tank. Given their sloppy analysis of the time to build a nuclear power plant, all I can say is “Heaven help us if this is true”.

    The Institute is funded by donations from philanthropic trusts and individuals. No details is given of the donors but I wonder if they have the “Greenpeace” problem where they doggedly act as an anti-nuclear organisation because many of their donors don’t want nuclear power. If this is the case then it is an unfortunate approach to being the most influential progressive think tank.

    An analysis performed as exposed by Ben, can only come from an organisation who wants to get the answer they are looking for, not the truth.

    I suspect that The Australia Institute will live to regret their bias when, in ten to twenty years time, Australia is falling well behind the rest of the world in greenhouse gas emissions because the rest of the developed world is using and building nuclear power plants, not burning fossil fuels, to meet the energy needs of their countries. If this think tank really believes that we can power Australia using only wind and solar with a bit of existing hydro then heaven help us indeed!

  4. Very much appreciated.

    Robert Pritchard Executive Director Energy Policy Institute of Australia Level 23 Governor Macquarie Tower 1 Farrer Place Sydney NSW 2000 Australia

    Tel +612 9252 8900 | Mob +61 413 755 616

    The Energy Policy Institute of Australia is an independent, apolitical, technology-neutral energy policy body. The Institute advocates a secure investment climate to ensure that Australia remains internationally competitive.

    From: Decarbonise SA Reply-To: Decarbonise SA Date: Wednesday, 26 August 2015 1:46 PM To: Robert Pritchard Subject: [New post] How long does it take to build a nuclear plant? Another look at The Australia Institute Decarbonise SA posted: “The submission by The Australia Institute (TAI) to the South Australian Nuclear Fuel Cycle Royal Commission includes discussion of time taken to build nuclear power plants. The mean time shown is 9.4 years. The conclusions drawn by TAI are: “If Australi”

  5. Maybe it’s a case of you get what you pay for. if you want a power source that does the same job as coal minus atmospheric pollution you have pay big bucks upfront and wait a few years for it. Wind and solar have been generously mandated and subsidised since 2001 yet they provide less than 10% of our electricity and relatively less than hydro did in 1975. Our power sector emissions are now increasing again suggesting wind and solar haven’t risen to the challenge.

    It would be better if large plant could be built quicker. The Areva EPR fails on that count. I understand offsite prefabrication of AP1000 components in the US hasn’t been smooth so far but perhaps will get better. Opponents point to cost reductions in wind and solar but apparently that won’t happen to nuclear. Still not cheap if you need to overbuild several times over to cover adverse weather output. Also the gas backup that enables intermittent generation won’t get cheaper if eastern Australia exports 70% of gas production as LNG.

  6. Ben:

    This is the wrong fight.

    It is undeniable that solar farms and wind farms can be built quickly in Australia, whereas nuclear power plants can’t – thanks to the need for a legislative framework and inevitable warfare with the anti-nuclear movement.

    HOWEVER, renewable energy does not NEED to be used to generate electricity; decades of misguided climate change activism has focused on reducing CO2 emissions by bringing wind farms and solar farms onto the electricity grid, but what if we focused on using renewable energy to decarbonise TRANSPORT ahead of electricity?

    I am following Germany’s RWE power-to-gas plant closely. If the renewables lobby learnt to see the benefit of focusing on using PV solar farms to generate hydrogen, so many problems could be solved:

    – hydrogen fuel for vehicles – this could be used almost immediately.
    – hydrogen fuel for planes – it would take longer.
    – hydrogen gas to replace conventional gas in peaking gas plants, and heating for homes.

    MEANWHILE, nuclear power can do what it’s good at – focus on replacing coal plants to provide base load electricity.

    CCS can be developed at the same time and retrofitted to coal plants where possible.

    This approach would end the fighting between all the various interest groups.

    – Climate change deniers have never been opposed to nuclear power and even are warming to the idea of solar power.
    – The nuclear lobby would find nothing to argue about in this plan.
    – The renewables lobby should be happy to see investment in P2G and PV solar farms.
    – The anti-wind lobby should be happy as there would be no need for bird- and bat killing, noisy, ugly wind farms.

    It’s good for the economy too – jobs created in P2G R&D, PV solar farm construction, and a local nuclear power industry.

    I think this is the answer. IF P2G works – but if they’ve cracked 86% efficiency at RWE?

    1. The obvious point is that if Germany is so advanced on power-to-gas how come they are building new coal fired power stations? See under Methane in this article
      Several factors are multiplied to get a 31% round trip efficiency. The 86% claim may have omitted some of these factors. I don’t have an EROEI figure for P2G but I doubt it is the 7 or more needed for a major energy source.

  7. All of the coal plants were approved years ago, whereas these breakthroughs in P2G are happening now.

    As I said it’s a big IF – IF viable P2G is closer than we think. If not, then I fall back believing that building nuclear power should be prioritised first, and R&D into technologies like storage and nuclear fusion second.

    I agree with you that using P2G to backup the grid is a silly idea. No one really has any idea how you would put storage plants onto the grid and use them to reduce CO2 emissions.

    But why would you put hydrogen storage plants on the grid in the first place if you could instead produce fuel for vehicles or planes or gas that can be burnt in conventional gas plants? If the renewables people solve this problem, great.

    Best of all, renewables plants can be built in the locations where solar resources are best. We wouldn’t need to build wind farms right next to towns but could instead focus on building PV solar farms at locations where there’s maximum sunlight.

    Nuclear would still required to replace the coal plants because it will take decades to be able to produce enough hydrogen to decarbonise all of transport AND replace all conventional gas. And advanced nuclear would still be required to fission up all of the stockpiles of nuclear “waste”.

    There is no reason for the nuclear lobby and renewables lobby to be fighting to decarbonise the electricity grid.

    1. Interesting ideas Alex.

      I have no conflict between using all these technologies. I don’t know how you took that impression from this post, however to reassure you I am not putting technologies in opposition.

  8. Ben,

    No, when I said “the wrong fight” I meant it is pointless denying that it’s going to take a long time to build nuclear power plants in Australia. It will.

    I take your point that these guys seem to be exaggerating how long it takes just to BUILD NPPs. But even so, it seems inevitable that it will take decades to build a nuclear power industry in Australia. We need to wait for the Royal Commission to finish; legislate federally; legislate in SA; train engineers; set up Australia’s version of the NRC; and so on.

    None of this changes the fact that building loads and loads of solar and wind capacity – e.g. Labor’s 50% RET – without any clue if or how this can ever be integrated into the electricity grid – is terrible policy. Roger Pielke Jr’s idea of a carbon tax that raises revenue that is invested into R&D makes far more sense to me.

    HOWEVER, if those gunning for 100% renewables could be persuaded to give up on grid-connected renewables and focus instead on off-grid production of hydrogen, everyone could get along I think.

    Who would lose in this scenario? No more wasted renewables capacity; climate deniers won’t object to hydrogen powered cars or planes; they won’t object to nuclear power; the growing anti-wind lobby will be immediately onside; the solar power industry should be supportive; CCS for coal plants should make the fossil fuel interests happy.

    Aside from Big Wind, there are no losers – nothing to argue about.

    1. Well I understand that better now.

      No, decades is not inevitable, and the sooner people stop saying it the sooner we can pull together to create a different outcome. Yes, it will take time. No, there is no excuse for the shoddy analysis in the submission which feeds into the “decades” narrative. Yes, when the work is done it produces and amazing amount of energy and it worth it.

      If you can convince the 100 % renewables crowd of ANYTHING, let me know ;).

      I think what you are discussing is interesting, keep on interrogating the concept.

      1. As I said earlier, the UAE experience looks like being 9 years from policy to power, starting well behind where Aust is now, with ANSTO, ARPANSA and a U mining industry already.

        1. OK. Thanks, I will look into this.

          I think if we had a strong, pro-nuclear, Labor prime minister – a Jay Weatherill at the federal level – it could start moving. But not just anyone, someone like a Keating who could bring the nation as well as the antinuclear zealots in the party along to support it. But right now the Labor Party has just committed to the grassroots “LEAN” plan of a 50% renewables target. The same crowd are hysterically anti-nuclear. Shorten came out immediately and opposed Weatherill’s Royal Commission. Politically, I can’t see how Shorten can prioritise this, and I don’t think he supports it. Nuclear costs money, and he’s just committed to more of it than he has on renewables.

          Meanwhile, John Howard couldn’t get nuclear onto the agenda in the face of a Labor scare campaign. The moment Julie Bishop brought it up, the Labor frontbenchers went into full scare campaign mode. What chance is there for Tony Abbott to do it?

          I suppose if the Royal Commission in SA makes a strong case for federal reform, perhaps the federal leaders would listen.

  9. I predict that opponents of SA nuclear will insist upon cooling towers for power plant knowing the driest state in the driest continent cannot spare the fresh water. This requirement appears to be causing the closure of Oyster Creek US. There’s hundreds of kilometres of suitable coastline for seawater cooling but somehow every inch of it will have endangered larvae. Plenty of work for activist lawyers not so much for laid off car workers or defence contractors.

  10. But you’re not seriously suggesting that Australia could have a nuclear reactor up and running in ~6 years, are you? And, to discuss confounding factors, aren’t people in Australia proposing we build rather more advanced reactors than most of those in the list above? Don’t you favour 3+ gen, transitioning to 4th? Time to build those is likely to be much higher in the early days, isn’t it? Did you discuss that, and I missed it?

    I don’t think a ~10-15 year timeframe is unreasonable, and the point remains: renewables are cheaper right now, and getting cheaper, while storage shows every sign of being economically solved in far less time than it’s likely to take a new nuclear industry to get started. And nuclear generators plainly do have a history of “outliers” in construction times.

    The UAE experience is instructive, I agree. They decided to go nuclear in 2008, if memory serves. The contract was awarded a year later. It might be generating power from nuclear in 2017, assuming all goes as planned. That’s 8-9 years, and it’s not done yet. They’ve also gone faster by taking the risky option of building a several reactors to the same design – a design that hasn’t actually run yet, so if design flaws are found, that has great potential to cause some pretty severe headaches.

    The first instance of the same reactor design in South Korea looks to have taken ~15 years from project beginning, and ~9 years from the beginning of construction. Again, it’s not due to begin commercial operation until next year, and so might be delayed further yet.

    Another confounding factor going against your low mean is that the table lists projects that have actually been completed. And surely Australia’s regulatory environment would be more similar to European build, rather than China’s. Yet you brushed over the issue of time overruns and poor profitability in Europe without comment.

    You can quibble over the numbers used, that’s fine. It’s a very brief discussion paper and there’s not enough room to go into detail. The point remains: in the time it’ll take to get a nuclear power industry in Australia, there’s every reason to think renewables will have made it redundant (at least for us – the same may not apply to other countries). Nothing you’ve written actually refutes the point.

    Apologies for not going into the other submissions. I’ve run out of lunch break!

    1. “storage shows every sign of being economically solved in far less time than it’s likely to take a new nuclear industry to get started”

      You just made that up.

      And after our little exchange on the Guardian in which you claimed that renewables were cheaper than nuclear “everywhere” and insisted that Lazard LCOE estimates for various technologies in the United States could be transported directly to China, you might find the just released IEA’s Projected Costs of Electricity Generation, 2015 Edition interesting. The Executive Summary is here:

      A couple of key points:

      1. Nuclear is cheaper than coal and gas at low discount rates and $30/tonne carbon price

      2. “Second, while the 2010 study noted a significant increase in the cost of baseload technologies, the data in this report suggest that any such cost inflation has been arrested. This is particularly notable in the case of nuclear technologies, which have costs that are roughly on a par with those reported in the prior study, thus undermining the growing narrative that nuclear costs continue to increase globally”

    2. Dan, let’s be quite clear.

      The numbers in the table? Your numbers. Not mine.

      The analysis? Badly, in fact fatally flawed and you don’t refute.

      Rather than humbly acknowledging this, your comment is an attempt to move the goalposts.

      If what you have said above is the point you wish to make, and the way you wish to make it, then I recommend you retract the submission, re-draft this section and re-submit. I will be happy to critique it afresh. That would be so much the better as right now this flawed analysis is in the public domain.

  11. I’ve often wondered about that study into the fastest paths to electricity production. The 11 years for Belgium and France, both seem to generate electricity in the first year. How do you generate electricity from a nuclear power plant in less than 2 years?

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s