The following progress has been made on decommissioning work:
- Chooz A: the reactor was shut down in 1991 and nuclear dismantling began in 2007 after the dismantling decree was issued. The final stage of dismantling began in 2016 and involves segmentation, conditioning and removal of reactor vessel internals, followed by dismantling of the vessel itself. These operations are due to be completed in 2024. The dismantling decree requires them to be followed by a period of surveillance of the runoff water from the cave for twenty years, meaning that declassification of the facility would occur in 2047;
- UNNG reactors: these six reactors were shut down between 1973 and 1994 and received their dismantling decrees between 2008 and 2010 (except for Chinon A1 and A2). Fuel removal and circuit draining have been completed for all these reactors, and dismantling operations are in process for the conventional and nuclear buildings in the periphery of the “reactor caissons”. Following the ASN’s decision of 2020, applications for dismantling permits will be submitted for all these reactors in 2022, to obtain new decrees allowing continuation of dismantling operations according to an “in-air” strategy. Opening of the top part of the first UNGG reactor caisson – Chinon A2 – is expected in 2033: the initial extractions of vessel internals and graphite blocks are due to start in 2040 and last 14 years. In parallel, the other UNGG sites are finalising their work and operations to put the sites into a safe storage configuration (by 2035). A safe storage configuration state means that 80% of surfaces have been dismantled and the reactor caissons awaiting dismantling are safe: this will allow sufficient progress on the first reactor in this series to gain experience and ensure safety for the other five operations. Opening of the caissons after the first UNGG decommissioning is scheduled to take place in or after 2055;
- Creys Malville: this plant was shut down in 1998 and received its dismantling decree in 2006. The following key stages have been completed: removal of the fuel, dismantling of the machine room, drainage of the circuits, processing and elimination of the sodium used for cooling in all circuits, filling the reactor vessel, opening and extracting the vessel caps, and the start of dismantling of the core vessel cap (which weighs several hundred tonnes). The next stages are dismantling the vessel internals (due to be completed in 2026), electromechanical dismantling in the reactor building, then decontamination (dismantling should end in 2038);
- Brennilis: this plant was shut down in 1985 and received a partial dismantling decree in 2011 allowing dismantling of all installations peripheral to the “reactor block”. The following key stages have been completed: removal of the fuel, dismantling of the machine room, the fuel building, auxiliary buildings, heat exchangers and the effluent treatment station. The next stages are examination of the application for full dismantling authorisation, with a view to obtaining a dismantling decree in 2022 that would enable EDF to dismantle the reactor block(the end of these operations is currently forecast at 2040).
28.4 Provisions for last cores
These provisions cover the future expenses resulting from scrapping fuel that will only be partially irradiated when the reactor is shut down. It is measured based on:
- the cost of the loss on fuel in the reactor that is not totally spent at the time of final reactor shutdown and cannot be reused due to technical and regulatory constraints (“front-end” expenses);
- the cost of fuel processing, and waste removal and storage operations (“back-end” expenses). These costs are valued in a similar way to provisions for spent fuel management and long-term radioactive waste management.
These unavoidable costs are components of the cost of nuclear reactor shutdown and decommissioning. As such, they are fully covered by provisions from the commissioning date and an asset associated with the provision is recognised. In a decision of 11 December 2020, France’s Council of State challenged the tax-deductibility of the consequences of immediate recognition of a provision for dismantling of the last core (“front-end” last core expenses) (see note 14).
In 2020 after the Fessenheim plant was definitively shut down, €99 million of the provision for last cores, concerning the two reactors at Fessenheim, was reversed with a corresponding reduction in the inventories of non-irradiated fuel in the reactor at the time of the shutdown, and in parallel, provisions for spent fuel management and long-term radioactive waste management were recognised for the cost of processing this fuel and storage of the waste that will result.
28.5 Discount rate, inflation and sensitivity analyses
28.5.1 Calculation of the discount rate and inflation
Until 30 June 2020, the discount rate was based on the sliding 10-year average yield on French OAT 2055 treasury bonds which have a similar duration to the obligations, plus the spread of corporate bonds rated A to AA.
As of 31 December 2020, the methodologies used to determine the discount rate changed as follows:
The discount rate is now based on an interest rate curve, which comprises asovereign yield curve constructed on year-end market data for liquid horizons (OAT bond 0-20 year curve) and then converging, using an interpolation curve, towards the very long-term rate UFR (Ultimate Forward Rate) – with yields that become close to the UFR after 50 years – plus a curve of the spread of corporate bonds rated A to BBB. Based on the disbursements expected to meet nuclear obligations, a single equivalent discount rate is deduced by applying the discount rates from the interest rate curve constructed in this way to each flow as appropriate to its maturity. This single discount rate is then applied to the forecast disbursement schedules for thecosts of the obligations, to determine the provisions.
The UFR was defined by the European Insurance and Occupational Pensions Authority(EIOPA) for very long-term insurance liabilities that will involve disbursements beyond market horizons. The UFR calculated for 2020 is 3.51%. This is used in the calculation methodology, in compliance with the decision by the French authorities, which in the ministerial order of 1 July 2020 amending the order of 21 March 2007 on secure financing of nuclear expenses (see below) changed the formula of the regulatory ceiling for the discount rate, such that it now refers to the UFR instead of the arithmetic 48-month average of the TEC 30-year rate. The UFR is considered more relevant for nuclear provisions in view of the very long-term maturities. The sovereign yield curve indicates rates in a range of [-0.6%;0.2%] for outflows between 0 and 20 years, [0.2%; 3.2%] for outflows between 20 and 50 years, and a rate moving towards 3.51% for outflows after 50 years.
This change in calculation methodology for the discount rate provides the best assessment of the time value of money with regard to nuclear provisions, which are characterised by very long-term disbursement outflows, well beyond market horizons. This assessment is largely achieved through:
- use of an interest rate curve based on observed year-end market data with liquid horizons, converging over non-liquid horizons towards a very long-term rate with no cycle effect (instead of an average rate concerning a single duration corresponding to the average duration of the obligations), i.e. yield data for all the maturities associated with nuclear provisions;
- use of a very long-term rate (calculated UFR) produced by an independent body and now adopted by the French authorities in setting the formula of the regulatory ceiling, to take account of long trends in yield movements, in coherence with the distant disbursement horizon;
- a change in the references of the bond spread to include corporate bonds rated A to BBB by ratings agencies, in order to construct a robust spread curve since there are few AA-rated bonds, particularly on long maturities, whereas most“Investment Grade” bonds are BBB-rated bonds and the great majority of them have longer maturities.
