After a first article “Will nuclear power plants resist climate change?” , here is our second article on safety by Michaël Mangeon, associate researcher EVS – Environnement Ville Société (UMR 56000).

Climate change impacts the intensity and frequency of extreme hazards that raise safety issues for nuclear facilities. In the latest IPCC report, and in particular chapter 12 on regional impacts, the projected changes in natural hazards highlight several key elements for the geographical areas of France (the Central and Western European regions and, for its southernmost region, the Mediterranean).

Natural hazard projections

• A global increase in temperatures and hot extremes ;
• Sea level rise in all European regions, except the Baltic Sea, at a rate close to or greater than the average global sea level rise;
• An increase in winter precipitation in Northern Europe. In summer, a decrease in precipitation is projected around the Mediterranean, spreading to the northern regions. An increase in extreme precipitation in Western Europe ;
• An increase in river flows in Western Europe and thus in flooding;
• An increase in aridity and forest fires in the Mediterranean ;
• A decrease in cold spells and snow cover;
• A decrease in extreme wind in the Mediterranean ;

These natural hazards are taken into account in the safety demonstration of nuclear installations because they can lead to equipment failure or disablement, as we saw with the accident at the Fukushima Daiichi plant in 2011 following an earthquake and then a tsunami. Nuclear safety specialists use the generic term “external hazards of natural origin” or “natural hazards” to refer to natural hazards and their consequences.

With climate change, the safety of nuclear installations in the face of these natural hazards, some of which will become more intense and/or frequent, raises questions for politicians, associations and society as a whole. This issue is also beginning to be addressed by experts who are independent of the public authorities.

Capturing extreme events through statistics to ensure the safety of nuclear facilities

In France, the industrial company (in our examples we use mainly EDF, the operator of the nuclear power plants producing electricity) must demonstrate the safety of its installations to the nuclear safety authority (ASN) and its technical support, the Institute for Radiation Protection and Nuclear Safety (IRSN). The risk to society and the environment from a nuclear accident requires consideration of a higher level of risk than is defined for other industrial sectors.

A global target for natural hazards: 1 in 10,000 per year

The level of risk to be retained for the protection of nuclear power plants is a political decision, which can be translated into a return period and/or a probabilistic target. For example, for the risk of flooding (sea and river), scenarios have been defined in a 2013 guide from the French Nuclear Safety Authority (ASN) to protect nuclear power plants against floods with a 1 in 10,000 chance of occurring per year. This probabilistic target is sometimes mirrored by a return period of 10,000 years, known as a decamillennial.

This one in 10,000 chance of occurring per year has now become a European reference for all “natural hazards” with the publication of a guide by the European Regulators Association (WENRA). By way of comparison, for land use planning and civil security purposes, the 100-year hazard (one in 100) is often used as a reference, particularly in the context of the development of Flood Risk Prevention Plans (PPRi).

In this sense, the target of one risk in 10,000 per year for nuclear power plants is exceptional and is intended to avoid ‘black swans’, events of very low probability but with potentially dramatic consequences.

How can we evaluate and calculate these events that we have never observed and measured “in reality”?

To answer this question, French nuclear safety specialists use a variety of methods, with a strong emphasis on statistical methods and in particular extreme value theory. The idea is to arrive, from observed and measured data, by a statistical extrapolation, at one or more values of extreme hazards never observed, in order to have a usable numerical value (a flow, a wind speed, a temperature…) for a quantitative evaluation. Ultimately, this value can be used to protect the power plant, for example by means of a dyke of appropriate height for the case of flooding.

Examples of statistical calculation limitations and methods to overcome them

The complexity with these desired hazard levels is that they have often (and fortunately) never been observed in real life and are therefore based on calculation. The use of these statistical methods is problematic because confidence in a statistical extrapolation is dependent on the number of observations available, which is often limited (a few decades of data recorded and therefore available for calculation).

With limited data, nuclear safety specialists cannot therefore “extrapolate” directly to hazards with a return period of 10,000 years because the uncertainties in the result would be too great. They therefore do a ‘best guess’ (up to 100, 200, even 1000 years for floods) and then mobilise hazard mark-ups and/or combinations to arrive at extreme scenarios, which they imagine to cover the annual probabilistic target of one in 10,000 risks.

Climate change and nuclear safety: towards a paradigm shift?

These statistical methods, which were used in the 1970s for the design/construction of nuclear power plants without taking climate change into account, are based on the knowledge of past events and assume that the process is ‘stationary’ over time. To put it simply, nuclear safety specialists then imagined that the conditions for the occurrence of extreme hazards would not change over time. For some extreme hazards, the assumption of constant conditions over time no longer holds due to climate change.

Since the 2000s, developments to take account of climate change have taken place…

Over the last few years, these methods have gradually evolved. This is particularly the case for two of the hazards that are impacted by the effects of climate change.

For floods of the “Submersion marine” type

First of all, for the nuclear sites of the seaside power plants currently in operation, the flooding scenario updated in 2013, and then following the supplementary safety assessments (SSA) after the Fukushima accident, includes the following combination:

• A surge (a rise in the sea surface which is usually due to the effect of a meteorological depression) of a millennial magnitude (1 in 1000/year risk);
• A tide of coefficient 120 (maximum coefficient);
• Consideration of climate change for mean sea level with a 20 cm increase for global warming, based on IPCC projections to cover developments over the next 10 to 20 years. After the Fukushima accident, certain structures, systems and components forming a “hard core” are protected against 50 cm higher sea levels. For the EPR reactor under construction in Flamanville, the “setting” or level of the platform on which the reactor is built must take into account the foreseeable evolution of the sea level until 2080.

These different elements of the scenario are reassessed at each review, every 10 years.

Significant heatwaves have led to regulatory changes…

The heatwave of the summer of 2003 and then that of 2006 showed the possible vulnerabilities of nuclear power plants, with numerous exemptions granted for thermal discharges into watercourses, but also exceeding the maximum temperatures imagined at the design stage of the reactors for certain systems and equipment.

Although the consequences for nuclear safety were nevertheless limited, a new reference system known as “extreme heat” was put in place by EDF in order to re-evaluate the safety of the installations for higher outside air and cold source temperatures, likely to be reached by 2042, taking account of global warming. Indeed, on the subject of temperatures, the influence of climate change is very important.

…On risk assessment

After 2003, a 30-year return period was sought using a statistical calculation based on data from the past, but for which there is already a clear trend towards higher temperatures due to climate change. This trend is then mathematically extended for decades to come. Today, at the request of the IRSN and the ASN, EDF must take into account an exceptional 100-year temperature, based on the same type of calculation, to cover climatic changes until the next safety review of the site. Finally, for the EPR site under construction in Flamanville, EDF uses a different method and mobilises climate model outputs with projections to 2080-2100.

The target of one in 10,000 risks per year seems difficult to justify for the “hot” aggression. In this sense, EDF has added a flat-rate increase of 2°C to these calculations in an attempt to move towards this target.

…and on the protection of nuclear power plants

From a safety point of view, the checks of a nuclear power plant against “hot spots” cover two aspects:

• High outside air temperatures lead to an increase in temperature in the rooms of nuclear power plants, which can disrupt the operation of equipment, cause premature ageing of the equipment and even render it unavailable. In this respect, EDF ensures that the air temperature in the premises is in line with the permissible temperatures of the equipment in them. For example, after the 2003 heat wave or in the context of periodic safety reviews, such as IRSN explainsThe performance of the heat exchangers cooling the water of the safety systems with water from the cold source was increased, self-contained air conditioners were installed, cold batteries were added to certain ventilation systems… “.

• High temperatures of cold source water can affect the cooling capabilities of nuclear power plant systems. If the water temperature in the cold source is too high, EDF can implement load shedding (i.e. stop the cooling of some non-essential systems) to maintain the cooling of essential systems. On the new “EPR 2” reactor project, in addition to the main cold source (sea or river), EDF has added a diversified cold source which consists of a water basin cooled autonomously by an air cooler. In addition, during heatwaves (but not only), cloggers (such as algae or fry) can proliferate and clog the cold spring. Drums, screens and other floating devices are installed to protect the plants from these cloggers.

Finally, special rules for the graduated operation of reactors (watch, vigilance, pre-alert and alert phases) are provided for by EDF in the event of “hot weather” to prevent, detect and control the consequences of high air and water temperatures on the operation of installations.

But, despite the use of statistical methods and the implementation of material and organisational countermeasures, can we be certain that the plants will never be affected by extreme hazards?

The occurrence of real events and scientific advances call for modesty and caution

The use of statistical methods may, however, be questionable in the light of observed reality and, in particular, of events, whether or not related to climate change, that have affected nuclear power plants.

The shock of the flooding of the Blayais nuclear power plant in 1999

During the Martin storm at the end of December 1999,the Blayais power plant was flooded by waves which, pushed by the swell, passed over the power plant’s dyke, even though the latter was designed to withstand a thousand-year surge and a tidal coefficient of 120. The effect of swell had not been taken into account on this estuary site. This event shows the complexity of capturing the full effects of a hydrometeorological situation.

In addition, crisis management was complex due to the damage caused by the storm. Finally, this event highlighted certain limitations of the statistical methods usually used, in particular the difficulty of taking into account extreme events, known as “outliers”, which, because of their rarity and intensity, can make statistical extrapolations very uncertain.

Some aftermath of the Blayais flood

Additional methods (consideration of historical events not recorded in the databases or of larger geographical areas) are now being developed to extend the database of extreme events. The Blayais flood was classified as a level 2 incident on the INES scale, which ranks nuclear events from 0 to 7 (Chernobyl and Fukushima were classified as level 7).

The flooding of the Blayais is a major media and political shock. For safety specialists, this event will lead to a long overhaul of flood regulations.

The summer heat wave of 2019: feedback from nuclear safety

More recently, during the heatwave in the summer of 2019, the exceptional temperatures of the heatwave aggression of the “hot weather” reference defined in 2003 were exceeded for the Paluel, Penly and Gravelines sites. These sites in the north of France had never recorded such temperatures. The statistical calculation had therefore not led to the anticipation of this event.

Of course, in a real situation, exceeding the values defined by the calculation does not always lead to a nuclear incident or accident. For example, safety critical equipment does not shut down when its maximum allowable temperature is exceeded. A series of lines of defence, which specialists call defence in depth and margins, must be in place to avoid the worst in the event of extreme natural hazards.

In addition, following the Fukushima accident, a “hard core” consisting of existing and new equipment protected against extreme events (earthquake, flood and tornado) was set up to manage extreme situations. However, despite these safeguards, an accident is always possible and various events have shown that the calculationsneed to be revised regularly and the methods adjusted. For crisis management, EDF’s impressive ‘Force Action Rapide Nucléaire’ is prepared to deal with extreme situations.

Other extreme hazards that are not (yet?) covered by regulations in relation to climate change

Similarly, some extreme hazards, such as river flooding, rainfall (or storms, even though There is no consensus on the effects of climate change on storms in France), are not yet specifically treated in the calculations as a function of climate change. Recent evidence, over the last few years, of changes in the intensity and/or frequency of these hazards has been followed by IRSN specialists and industrialists (EDF in particular), even if work on the most extreme hazards is still rare. At EDF, A climate watch is also set up to monitor these developments.

As we have seen, the assessment of natural hazards and the management of their consequences focuses on sudden, single events that directly impact the facility. Today, industrialists, IRSN and ASN consider that nuclear power plants have margins against these hazards, particularly with the implementation of the hard core after Fukushima.

But are extreme hazards the only risks associated with climate change that can affect nuclear power plants?

Reflections on nuclear safety in a world in crisis

In addition to natural hazards, the impact of climate change on nuclear safety is still under discussion. What impact will climate change have on the stability of our political, economic, social and industrial systems? And in this sense, in a world of multiple crises, how can the safety of nuclear installations be guaranteed?

How can we imagine the impact of climate change on nuclear safety beyond the issue of natural hazards?

To reflect on these issues, the use of qualitative studies, foresight work and even science fiction (as the army does with the Red Team Defence) on the state of the world in the more or less near future constitute interesting work perspectives. Nevertheless, these issues appear to be much more complex to understand methodologically for nuclear safety specialists.

These aspects may concern what specialists call “organisational and human factors (OHF)”, which are regularly found to be one of the main causes of nuclear incidents or accidents, but they may also mobilise the human and social sciences, a field that is still little exploited in nuclear safety issues.

Some open questions about nuclear safety in a world in crisis

At this stage, many questions remain unanswered. For example, if specialists know how to apprehend “spontaneous” and quantifiable events, how can we imagine the consequences on safety of systemic, lasting and uncertain crises? And, for such scenarios, on what criteria can we agree that safety is sufficiently demonstrated? Do the probabilistic targets used to evaluate extreme hazards still make sense for this type of scenario, which can indirectly impact the entire nuclear socio-technical system over sometimes long periods of time?

Other questions may also arise: how can facilities be maintained in times of sustained economic crisis, shortages of equipment and materials, major social and/or political unrest or even armed conflict? How can we think about the trade-offs between nuclear safety and electricity production in times of extreme tension on the electricity grid? In short, how can a nuclear risk management system cope with a systemic crisis whose contours are unclear and whose intensity and impacts remain uncertain?

Here are some additional ways to explore the subject

In order to think about nuclear safety in a world in crisis, we are conducting, for example, historical analyses of nuclear safety in the Eastern Bloc countries after the collapse of the USSR, in particular the case of the Bulgarian Kozloduy plant.

Similarly, the Covid crisis can provide feedback on how to manage safety (and operations) during a crisis. While the ASN considers that the level of safety and radiation protection has remained satisfactory and that those in charge of nuclear activities have been able to adapt to the situation, it also considers that the context remains uncertain and evolving.

Finally, although this Covid crisis seems to have been well managed from a safety point of view, it has nevertheless led to the postponement of certain maintenance work and, ultimately, to the unavailability of reactors in the winter of 2021-2022. In addition to these maintenance delays, reactors have been shut down due to faults, fortunately without any significant impact on electricity supply at this stage.

Conclusion

Nuclear specialists are working on a daily basis to cope with extreme natural hazards by gradually integrating new knowledge related to climate change.

Nevertheless, certain future crises, with uncertain contours, seem difficult to grasp by the calculation methods traditionally used by safety specialists. In this sense, it is not a question of calling into question the work that has already been done for many years, but of pushing for a broadening of the framework of reflection, and thus of equipping ourselves for the future with capacities for projection, creativity and even invention that will undoubtedly be necessary for the preparation of future crises.

Obviously, these reflections must go beyond the sole framework of nuclear safety and question our overall resilience capacities, at a time when France is considering launching major industrial and energy programmes.