Older Archives

The economic loss of the early retirement of nuclear power plants

February 15th, 2012 by François Lévêque, Ecole des mines de Paris

2011 will remain as one of the year that shaped the fate of nuclear power the most. The Fukushima accident has had profound consequences on energy policies in several countries, and even France and its unique system have not escaped the controversy. But if the debate focuses mostly on an assessment of a partial or total phase-out with 2030 as a global deadline, there is few economic appraisal of the optimum pace for a nuclear power plants’ retirement. Closing a nuclear power plant earlier than expected, even though the operator complies with all the safety requirements, would mean getting rid of a still valuable asset, and a global loss for the economy.

What is the financial difference between the several phasing-out paces proposed in France? This is what we try to determine here by assessing the cost of closing nuclear power plants (NPP, hereafter) 30 or 40 years of operation compared to the 50 years lifespan that the French nuclear safety authority is expected to grant to EDF.

First of all, it is important to notice that the issue of the economically sound age to close a nuclear power plant is to a large extent decoupled with the choice of pursuing or exiting from nuclear power. The economic age to close nuclear plants has to be addressed whether the nuclear fleet is replaced by wind farms, gas power plants, or new nuclear builds. A government could decide to abandon nuclear power generation while authorizing a maximal life span for the existing reactors to the extents safety requirements are complied with and it is economically profitable to run them. In other words, one can end nuclear power without wasting money in retiring plants too early.

The German case

The analysis of the German situation is a good beginning to assess the cost of an early closure. With 23% of its electricity generation based on nuclear power, Germany presents a far different electricity mix than France. With more ambitious environmental policies, the development of renewable energies has already begun for a long time. The fact that the country still relies on national fossil reserves, such as coal, and has been characterized by a strong opposition to nuclear allowed Germany to consider its nuclear phase-out as soon as 2002, under the Red-Green government. In 2010, however, when the German government “Energy Concept” set up a framework for a long term transition toward a 60% share of renewables in its electricity generation mix by 2050, the government backed out a little: to ensure the transition, nuclear power was to be used as a technology bridge, and it was decided to delay the phase-out initially planned for 2022, with an average 12-year extension of the plants’ lifespan.

The Fukushima accident, less than a year after, deeply changed the rules of the game. In a very fast policy reversal, the government temporarily shut down the seven oldest nuclear plants, and a few months later, set again 2022 as the date of the last plant’s closure, also deciding not to open again the eight reactors closed after the moratorium or for (pre-Fukushima) maintenance.

Consequently, in only one year the nuclear fleet decreased from 17 to 9 power plants. This shift required very fast decisions to ensure the energy supply, and other decisions will follow to progressively adapt the German energy system to the ending of nuclear power generation in 2022. The cost of this overall transition from one scenario to the other is however not so clear. Over the last six months, many figures were mentioned: RWE threw a €250 billion figure whereas Siemens estimated the early phase-out at a €100 billion cost, in case of a balanced replacement of nuclear power with gas and renewable energies. Some stakeholders provided more precise studies on the subject. This is the case, for example, of the EWI/GWS/Prognos report, commissioned by the Germany Ministry of Industry.

Delivered in 2011, the report estimates the additional cost resulting from a shift from the “Energy Concept” scenario to a complete phase-out by 2022. In this study, nuclear electricity generation is substituted with a mix of coal and natural gas and, a major shift in the electricity import/export balance. In 2022, the closure of the remaining NPPs is expected to lead to a 110 TWh electricity shortage, which will be substituted with 50 TWh from gas, 20 TWh from the import/export balance and 40 TWh from coal. The estimation of the resulting shift in costs is derived from these hypotheses. From the “Energy Concept” longer exit scenario, the bill inflates respectively by €30/MWh and €13 billion due to the increase in variable costs and imports. This amount is then weighted by maintenance and operation costs as well as a high level of retrofit investments expected in case of a nuclear continuation, with respectively €18 and €9 billion. In total, the direct additional cost of a 12-year earlier exit is assessed in this study at a €16Bn figure. Compared to results from other studies, such as Enervis Energy Advisors’ (€20Bn) or IER Stuttgart’s (€42Bn), this is a relatively low assessment, mostly due to the high level of retrofit investment, between €600 and 1200 million per reactor.

Extension to the French case: scenarios, assumptions and results

Any attempt to assess the cost difference, hence the resulting loss, between an early and an economically appropriate exit in the French case implies several context points that allow us to transcript the German assessment to the French one.

First, with 75% of its electricity generation based on nuclear power, few renewables and no fossil reserves, the plausible short-term option will differ. The replacement of nuclear power plants will likely be ensured by a combination of imported electricity and gas. Like in Germany, the assumption is made that the development of renewable energies will go on as currently expected. A short-term massive replacement by renewable power is unlikely, as it remains a more expensive source of energy and retains some challenges for the electricity grid. Moreover, given the current context of security of supply, and the rise of the German and Italian imports, we will assume that a rise in imports is hardly possible. As a result, our estimation will make the oversimplifying hypothesis that the current nuclear fleet (58 reactors, without considering the Flamanville EPR) will be replaced by new gas power plants, whatever the nuclear fleet closing pace. This assumption can be considered as a conservative assessment of the costs for the early retirement of NPPs, as it does not include new investments in gas network and all the other technological options than gas (including most actions of energy efficiency) could be roughly considered as more costly.


One must now focus on the closure pace. It will result on a combination of decisions from the Autorité de Sûreté du Nucléaire (ASN, hereafter) and EDF. Every ten years, after the decennial control, the ASN gives for each plant a series of safety requirements that have to be implemented by EDF if it wants to keep operating the plant. After the first thirty years of lifespan are reached, the ASN will, following the control, set the requirements to allow a lifespan extension for another decade. These measures could require major investments (for instance, they are currently assessed at €650M per reactor). EDF will then either make the investments, pushing the lifespan to 40 years, or close the plant (closure after 30 years), depending on whether the investments for prolongation are worth. The same will happen after the fourth decennial control to ponder a possible extension to a 50-year lifespan. The €650 million invested at the third decennial control could be enough or not to push to fifty years. The following scenario tree is therefore considered:

As a result, we focus on the analysis of four scenarios. In the first one (scenario “50”), EDF decides to go on with the extension investments once (€650M), but is allowed to push the extension of the whole nuclear fleet, that is 58 reactors, to a 50-year lifespan. In the second one (scenario “50+”), EDF has to realize new investments at the fourth visits to push its fleet to 50 years (650M+650M). In the third one (scenario “40”), the second safety investment series is not worth to be made, and a closure after a 40-year lifespan is planned for the entire fleet. The last one (scenario “30”) assumes that safety investments are not worth and every power plant closes after thirty years. The general assumption is made that whatever the amount, the safety requirements to push the lifespan to a higher-than-fifty value are not profitable or simply not possible. Finally, one must add to the lifespan extension investments the latest requirements of the ASN, made after the Fukushima disaster. In that respect, a €150M investment per reactor is expected to be made between 2012 and 2014.

An early retirement of NPPs means here that reactors are closed because of a political decision although the ASN has given its approval to increase their lifespan under certain conditions and the investment costs to comply with these conditions are lower than the benefits provided by the increase in duration. The calculation we want to make consists in computing the costs of closing the reactors at 30 years whereas they could have been economically operated 10 years or 20 years more, or closing them at 40 years although they could have been economically operated 10 years more.

In the case of an earlier closure, nuclear plants will then be replaced earlier by gas power plants, which will make costs heavily differ from one scenario to another. The assessment of these differences requires specific assumptions regarding the specific costs of each scenario. The figures we took into account can be found in the following table; they are based on an OECD/IEA study (Projected Costs of Generating Electricity 2010). Note also that indirect costs (e.g., effects on job creation and destruction) are not taken into account.


The costs of a nuclear early termination in terms of production costs’ comparison are easy to draw. Closing the reactors ten or twenty years earlier than economically sound would mean two things: i) not capitalizing on the most profitable years on the plant, once the construction is fully amortized (in the above estimation, only the operating costs of NPPs – 25€/ MWh – and prolongation investments are considered) and ii) replacing the existing nuclear power fleet by a more expensive power generation capacity. Given the higher variable costs of gas compared to nuclear power generation, and the fact that construction and dismantling investments will occur in any case, the longer nuclear power plants are operated, the less the invoice inflates. In the ”30” and “40” scenarios, the last nuclear power plant closes respectively in 2030 and 2040, giving way to a massive fleet of gas plants, subjects to higher fuel prices and C02 emission costs, which will make the additional cost grow. For each reactor, the €650 million required for a minimal 10-year extension is then worthwhile after two years, the additional variable cost being assessed at €300 million per reactor-year. The economic loss resulting from an earlier closure is therefore clear: With only one series of extension investments, after the third decennial decades for each reactor, the total resulting loss is assessed to respectively 245 and 145 billion Euros for a 20-year or 10-year earlier exit, the reference scenario giving 50 years as the average lifespan. And even with additional investments needed to reach the 50-year lifespan objective (another 650 million per reactor after the fourth visit) the loss is still assessed at a high value, between 107 and 213 billion Euros.

Of course, we do not pretend these figures are precise and coined in the marble. A sensitivity analysis on the main hypotheses (e.g., price of gas, level of CO2 tax, level of investments to increase the reactor lifespan…) would show significant variations. However, when compared with the evolution in Germany, our assessment presents figures in the similar range of those offered by EWI/GWS/Prognos, which ties in our estimates, given the relatively similar assumptions made on investments and variable costs. The Energies 2050 report issued yesterday by the French ministry for energy also provides same orders of magnitude.

There is therefore little doubt that an early closure of the French nuclear fleet will cost more than 100 B€, an extra cost that ultimately will have to be passed to the French consumers.

Michel Berthélémy, Sébastien Douguet and François Lévêque, Mines-ParisTech

Send this to a friend

One Response to “The economic loss of the early retirement of nuclear power plants”

  1. Samuel Bordenave Says:

    Your study gives an original point of view on the total cost of an early retirement of the NPPs. It has simplified many aspects, but this is clearly stated, and what is more, most of the simplifications (such as not including renewables) tend to lessen the cost, rather than inflate it.

    However, I would like to comment on some of the data you use in your model, and some issues that are not fully addressed.

    First, it strikes me that you haven’t taken into account commodities forwards for 2050.
    You give a figure of 7€ CO2 cost for CCG. New CCG produce approximately 0.4 tons of CO2 per MWh of gas. With the CO2 ton trading at nearly 6.5€ these days, this would mean a CO2 cost of 2.6 €/MWh. But more interesting is the fact that the EU has set a target for carbon price at around 50€ for 2030. This would mean a CO2 cost of nearly 20€!
    Similarly, the price of long-term gas contracts in Europe should reach 34 €/MWh in 2030 (source: UFE, “Electricité 2030”), against 24€ as of today. With new CCG having an efficiency of 0.55, this would mean that fuel cycle cost would go from 43 €/MWh to 61 €/MWh. This means we can expect the total cost of electricity production from CCG to have increased of amost 40€ in 2030. This is approximately 65% higher than its present value. We can probably expect an even bigger increase for 2050, which should not be disregarded.

    My second comment is about capital expenditure and WACC issues.
    Your table (just before the “Results” title) does not deal with CAPEX. However, these costs are essential to evaluate the production costs of NPPs. Should the reader assume that the CAPEX costs are worth ARENH minus O&M and fuel cycle? It would then give a value of approximately 42 – 17 = 25 €/MWh. Is this what you refer to in the Results when you say “in the above estimation, only the operating costs of NPPs – 25€/ MWh – and prolongation investments are considered“? But then are you referring to O&M or to CAPEX?

    Similarly, there is no mention of CAPEX for the new CCGs, which is more troubling. If your study makes the assumption that total building costs are paid overnight, then it does not include the costs of capital in the results, which is strange since your aim is to evaluate the total costs the consumers will have to pay.
    You might say that this doesn’t matter since this will again inflate the price, as the rest of your simplifications. But your study ends in 2050, and CCG are typically amortized in 30 years. So what about the plants built after 2020? Some of the capital costs should not be included in your calculus, and this could lessen the final cost.
    In any case, you should specify what your assumptions about capital costs and amortization are.
    The same is true for life extension costs. Do you consider that they are paid overnight, and that there is no CAPEX? I agree that it is not really clear whether these should be included in O&M costs or as extra capital costs. But in my opinion you cannot spend 54 * 650 millions overnight, without having to find capital to finance it. Amortization should typically take place during 10 years (length of life extension). On the one hand it may increase the costs of the 40, 50 and 50+ scenarios. On the other hand, for NPPs turning 50 less than 10 years before 2050, your study should not consider the costs that would be amortized after that date.

    In a certain way, your study gives the impression that every euro spent over the period is equal, be it in 2050 or in 2012. However, any expense has to be financed at WACC rate, and this means a difference in term of NPV.
    The evaluation of WACC is really important when it comes to such capital-intensive structures as NPPs. With the economic outlook in Europe and the difficulties many companies have to increase their capital, WACC is likely to increase for electricity producers. A good illustration of this is the fact that the IEA used two different level of WACC in its “projected costs of electricity production”, one at 5% and one at 10. At 5% discount rate, the CAPEX of an NPP is 30 €/MWh, and at 10% it almost reaches 70 €/MWh!

    To try and clear this issue, a good idea would be to calculate the NPV of all different scenarios, using an assumed and probable WACC (around 7% or 8%), as well as a probable amortization period for each overnight cost. This would diminish the total costs (as they would be amortized over time) but in my opinion this would allow a more objective comparison between your scenarios.

Leave a Reply

You must be logged in to post a comment.