COMMENTARY

fact, it is known from earth science that a mega-thrust earthquake occurring undersea can potentially provoke a tsunami. Indeed, the most recent example is the 2004 Indian ocean earthquake and tsunami, although in that case the earthquake was totally undersea (i e, no peak ground acceleration). Therefore, earthquakes and follow-up tsunamis are not independent events contrary to what was considered in the design of the plant. As a result, the Fukushima-Daiichi nuclear disaster is partly due to the fact that knowledge from earth science had not been understood enough at the time (or were not available) of the power plant design, rather than simply the fact that a quasi-impossible natural “disaster” actually happened.

In addition, the PRA methods used for taking into account some natural events can be criticised. Indeed, in the particular case of earthquakes for instance, there is a high uncertainty when performing the seismic hazardous analysis (e g, the maximal seismic magnitude which can occur in a certain zone of seismic activity) since it depends on earth science practices. This uncertainty can become so high for the very large earthquakes at very low probabilities that it has an impact on the overall probabilistic risk assessment of the plant’s design.[2] Worse, as the hazard estimate and structure ruggedness play an equal part in protection against seismic risk and in assessment of safety in case of a seismic event, the probabilistic risk assessment of nuclear power plants is based on uncertainty. It can be thus considered useless to try to improve the structure ruggedness or the component-level safety if the hazard estimate remains affected by such high uncertainty. In fact, there is a systemic problem with nuclear power plant’s design as, given this uncertainty, it is impossible to use probabilities (neither frequencies nor conventional ones, nor Bayesian), even if we want to assume that we can define an extreme natural event (which is impossible).

As a result, there is always a risk that an unpredictable event that exceeds the design assumptions will occur – which has been the case with the 9.0-magnitude Tohoku earthquake. This possibility of

Economic & Political Weekly

EPW

exceeding assumptions has been unfortunately verified during the current nuclear crisis in Japan as Massachusetts Institute of Technology (MIT) experts acknowledge that “nuclear power plants are designed for earthquakes and hurricanes, and in some places tsunamis. But these were unbelievably large”.[3] This demonstrates that the assumptions taken into account in the design of the Fukushima-Daiichi reactors were not conservative enough to protect them against such a natural event. More generally, there is no reason to say that assumptions will never be exceeded again or that another similar combination of natural events will never happen again, even with a very low probability of occurrence on paper. This demonstrates the existing systemic uncertainty that cannot be avoided with nuclear design.

Argument #2: “New reactor designs would stand such natural events”.

The introduction of the above discussed probabilistic approach to risk assessment to the nuclear industry is due to Norman C Rasmussen, a former professor of nuclear engineering at the MIT in the US. In 1975, he headed the publication of a report for the Nuclear Regulatory Commission (often called the “Rasmussen Report”).[4] This report received worldwide attention as it established the formal discipline of PRA, whose methods are now used routinely in nuclear power plant safety assessment. According to the Rasmussen report, the risk of a nuclear power plant failure was low, with a core damage accident occurring only once in every 20,000 years of operation in the US – one reactor running for one year counting as a year of operating experience.[5] But in 1979 – only four years after the Rasmussen report was published – a partial meltdown occurred at the Three Mile Island 2 reactor in Pennsylvania, when the nuclear industry in that country had fewer

than 500 years of operating experience. A new study ordered by the Nuclear Regulatory Commission reassessed the risk and estimated it at one meltdown per 1,000 years of reactor operation, 20 times more frequent than assumed in the Rasmussen report. This was the first “lesson learned”, facilitating an

improvement of the PRA-based design of nuclear power plants.

Core Damage Frequency

Nowadays, the current core damage frequency (CDF) of the current generation II reactors is said to be between about 5×10-5 per reactor-years or one core damage for every 20,000 reactor years (as expected by Rasmussen in 1975 for the US) in Europe [6] and one for every 50,000 reactor years (or 2×10-5) in the US.[7] With about 440 nuclear reactors currently operating worldwide, this corresponds to one core damage every 45 to 100 years and more. However, with three new core damage accidents at Fukushima-Daiichi nuclear reactors 1, 2 and 3 (with Three Mile Island in 1979 and Chernobyl in 1986), we have had five core damage accidents in less than 40 years. In fact, a core damage has happened every eight years on the average in the world since 1970, corresponding to the beginning of the operation of generation II reactors (very few generation I reactors remain today). This shows a large discrepancy between the safety announced by the nuclear industry and the safety actually measured.

The same day that the three core damages at Fukushima-Daiichi were confirmed,[8] experts at the MIT published an article saying that “new nuclear power plants may not have failed in Japan” (Bullis 2011 in note 2 above). But we obviously cannot know whether or not another reactor would have sustained the same natural events unless a complicated simulation is performed (which also requires us to know exactly the sequence of events that happened at Fukushima). Therefore, comparing one feature to another between different designs is a simplistic approach that does not capture the dynamic sequence of events, which is the one that matters in such nuclear accidents.

In the case of new plants, as shown above, although new safety features are taken into account in the design (i e, reducing the core damage frequency), there will always be a significant uncertainty about whether some assumptions on natural events or component failures will be violated. Statistically, the designed core

COMMENTARY

damage frequency of future generation III

+ reactors is of the order of magnitude of 5×10-7 per reactor-year, depending on the designs.[9] This means that the theoretical nuclear safety would be increased by a factor of 100 with new designs compared to current operating reactors, despite the fact that the generation II core damage frequencies have not even been met for the current reactors. Worse, there is no reason to say that the very low frequency of core damage accidents announced for the generation III+ reactors would actually be met, given the systemic uncertainty affecting core damage frequency estimates as explained before. This actually represents a major limitation on the viability of a large-scale expansion of nuclear energy as an alternative energy source.

Complacency

Therefore, the argument of better safety with new design seems to reflect complacency more than objectivity. Indeed, a good illustration of this complacency towards nuclear energy comes from a recent declaration of French President, Nicolas Sarkozy talking about the design of the new AREVA EPR reactor during the Fukushima nuclear crisis: “The idea of the double wall structure is that if a Boeing 747 crashes on the plant, the reactor is not damaged”.[10] That is true. The double wall structure of the EPR reactor building would withstand such an event and it is part of the new safety features of the future nuclear EPR reactor. But we cannot predict all other threats or mistakes, not just from the outside but also internal to the plant operation. In any case, there is no EPR reactor currently operating in the world. Only five are under construction while there are about 440 plants operating worldwide. In that case, this argument is not relevant at the time of the nuclear energy crisis in Japan. Therefore, we should be very critical about this kind of official discourse as the following political lock-in we face in general seems to apply to nuclear technology:

When we act, we create our own reality. And while you’re studying that reality... we’ll act again, creating other new realities, which you can study too, and that’s how things will sort out. We’re history’s actors…and you, all of you, will be left to just study what we do

– attributed to Karl Rove, former advisor of Georges W Bush.[11]

To conclude, I cannot do anything but to urge you not to take as “truthful” the over-reassuring and non-scientificallybased discourse that tends to minimise the seriousness of the nuclear disaster in Japan or which intends to avoid facing the current problems of nuclear energy by talking of future prospects. The history of humankind is already full of such examples.

The existing systemic uncertainty affecting nuclear power plant design raises the question of whether society is willing to accept continuing with a neverending learning process with potentially high adverse consequences, both to humans and to the environment. It has been argued here that developing new designs will not lead to improved nuclear safety but will simply maintain the techno logical lock-in put in place by the civilian nuclear industry.

References

[1] International Atomic Energy Agency website’s homepage (19 March 2011): http://www.iaea.org/

[2] We consider here the same distinction between low probabilistic “risk” (randomness with knowable probabilities) and “uncertainty” (randomness with unknowable probabilities) as introduced in 1921 by Frank Knight (Risk Uncertainty and Profit) in the field of Economics.

[3] Kevin Bullis (2011): Newer Nuclear Reactors Might

Not Have Failed in Japan, Technology Review, Massachusetts Institute of Technology. URL: http://www.technologyreview.com/energy/ 35100/?nlid=4239&a=f

[4] Norman C Rasmussen et al (1975): Reactor Safety Study: An Assessment of Accident Risks in US Commercial Nuclear Power Plants, WASH-1400 (NUREG-75/014), Rockville, MD, US: Federal Government of the United States, US Nuclear Regulatory Commission, doi: 10.2172/7134131.

[5] Matthew L Wald (2003): Dr Norman C Rasmussen, 75, Expert on Nuclear Power Risk, July 28, The New York Times. URL: http://www.nytimes.com/ 2003/07/ 28/us/dr-norman-c-rasmussen-75expert-on-nuclear-power-risk.html

[6] B A Leurs and R C N Wit (2003): Environmentally Harmful Support Measures in EU Member States, Report for the European Commission, p 137. URL: http://www.mng.org.uk/gh/resources/EC_env_ subsidies.pdf

[7] John Gaertner et al (2008): Safety and Operational Benefits of Risk-Informed Initiatives, Electric Power Research Institute, p 3, footnote 3. URL: http://mydocs.epri.com/docs/Cor porate Documents/SectorPages/Portfolio/Nuclear/ Safety_and_Operational_Benefits_1016308.pdf

[8] The three partial core-meltdown have been confirmed by TEPCO and announced by the French ASN authorities on 15 March 2011 (Press Release No 9). URL: http://www.asn.fr/

[9] E g, UK EPR maximum core damage frequency is equal to 6.1×10-7 per reactor-year (http://www. epr-reactor.co.uk/ssmod/liblocal/docs/V3/Volume%202%20-%20Design%20and%20 Safety/2.R%20-%20Probabilistic%20Safety%20 Assessment/2.R.1%20-%20Level%201%20Probabilistic%20Safety%20Assessment%20-%20v2. pdf).

[10]Yahoo News (15 March 2011), Pour Nicolas Sarkozy “Pas Question de Sortir du Nucléaire”. URL: http://fr.news.yahoo.com/68/20110315/ tsc-pour-nicolas-sarkozy-pas-question-de04aaa9b.html

[11] Ron Suskin (2004): Faith, Certainty and the Presidency of George W Bush, 17 October 2004, The New York Times, URL: http://www.nytimes.com/ 2004/10/17/magazine/17BUSH.html

'PJCPEKPI %QORGVGPEKGU QH #FQNGUEGPVU CPF ;QWVJ

# .KHG 5MKNNU #RRTQCEJ

VJ 0QXGODGT

%CNN HQT #DUVTCEVU

+P VJG UV EGPVWT[ VJG QDLGEVKXG QH GFWECVKQP HQT CNN OWUV DG IGCTGF VQYCTFU GPJCPEKPI ECRCDKNKVKGU GPNCTIKPI EJQKEGU CPF FGXGNQRKPI CIGPE[ D[ DWKNFKPI FKHHGTGPV FKOGPUKQPU QH YGNN DGKPID[DWKNFKPIUGNHKOCIGCPFUGNHYQTVJYJKEJKPVWTPJGNRKPFKXKFWCNUVQDGNGUUXWNPGTCDNG VQ VJG XCTKCVKQPU YKVJKP C IKXGP EQPVGZV 'FWECVKQPCN EQPVGPV OWUV VJGTGHQTG DG TGXKGYGF VQ TGOQXG VJG UVGTGQV[RGU CPF CIGQNF PQTOU VJCV RGTUKUV 'XGT[ CURGEV QH GFWECVKQP OWUV VJGP YQTM VQYCTFU ſIJVKPI CNN V[RGU QH RQXGTV[ KPENWFKPI TGFWEKPI XWNPGTCDKNKV[ D[ DWKNFKPI RQVGPVKCN CPF KPETGCUKPI CIGPE[ CPF YGNN DGKPI QH KPFKXKFWCNU CPF UQEKGVKGU 4CLKX)CPFJK0CVKQPCN+PUVKVWVGQH;QWVJGXGNQROGPV
4)0+;KUQTICPKUKPITF+PVGTPCVKQPCN %QPHGTGPEG QP.KHG 5MKNNU'FWECVKQPKPEQNNCDQTCVKQPYKVJ 4GUGCTEJ%QOOKVVGGQP5QEKQNQI[QH ;QWVJ
4% QH +PVGTPCVKQPCN 5QEKQNQI[ #UUQEKCVKQP
+5# FWTKPI VJ 0QXGODGT CV 4)0+; 5TKRGTWODWFWT

6JGRTGUGPVEQPHGTGPEGEQOGUWRYKVJHQNNQYKPIGZENWUKXGIQCNVQFKCNQIWGQPVJGPGYFKOGPUKQPU QH .KHG 5MKNNU 'FWECVKQP 6JG[ CTG 6Q GZRNQTG PGY XKUVCU QH .KHG 5MKNNU 'FWECVKQP 6Q DTKPI VQIGVJGT VJGQTGVKECN FKUEQWTUG CPF GORKTKECN EQPVTKDWVKQPU QP .KHG 5MKNNU 'FWECVKQP 6Q FKUEWUU EJCNNGPIGU QH .KHG 5MKNNU 'FWECVKQP YKVJ TGHGTGPEG VQ XWNPGTCDNG ITQWR 6Q FQEWOGPV CPF FKUUGOKPCVG VJG RTQEGGFKPIU HQT VJG WUG QH UVCMGJQNFGTU

6JG HQEWU QH TF +PVGTPCVKQPCN %QPHGTGPEG QP .KHG 5MKNNU 'FWECVKQP KU VQ DTKPI XCTKQWU UVCMG JQNFGTU KPVQ C RNCVHQTO VQ FGDCVG CPF FKUEWUU KUUWGU CPF KPPQXCVKXG RTCEVKEGU CUUQEKCVGF YKVJ VJGEQPHGTGPEGVJGOG+PVGTGUVGFUEJQNCTUOC[UGPFVJGKTCDUVTCEVQPQTDGHQTGVJ#RTKN +PVKOCVKQP QH UGNGEVKQP QH CDUVTCEV YKNN DG UGPV VJ #RTKN (QT HWTVJGT FGVCKNU TGHGT VJG YGDUKVG JVVRYYYTIPK[FIQXKP

%QPVCEV T # 4CFJCMTKUJPCP 0CKT NUGTIPK[FIOCKNEQO /QDKNG 0Q

march 26, 2011 vol xlvI no 13

EPW

Economic Political Weekly