The Antidote

Nuclear 1: Seismic risk assessment May 3, 2010

The draft environmental impact report (EIR) for Eskom’s proposed Nuclear 1 project is available for public comment until 31 May 2010. Eskom is trying to get approval to build new nuclear power stations at three sites (Thyspunt in the Eastern Cape, Bantamsklip in the Western Cape and Duynefontein next to the existing Koeberg plant near Cape Town). You can find the entire report here.

Since I have some professional qualifications in the area and am interested in the matter I worked through the Seismological Risk Assessment Specialist Report (Appendix E4). Here are my (rather lengthy) comments and questions which I will be submitting before the end of the month:

Comments and questions regarding the Seismological Risk Assessment Specialist Report of the Draft Environmental Impact Report for Nuclear 1 (Eskom Nuclear Power Station and Associated Infrastructure)

By: Andreas Späth
Qualifications: BSc (Geology) UCT, MSc (Geology) UCT, PhD (Geology) UCT

An unsatisfactory report

This report is incomplete, scientifically flawed and presents, by its own admission, data that are not in conformance with the latest internationally accepted practice. The (anonymous) authors of the report nevertheless proceed to use these insufficient data to conclude that “the seismic hazard does not preclude a standard export NPS (nuclear power station) at any of the proposed sites”. They appear to come to a preconceived conclusion in the face of their own contradictory observations.

The report seems to have nothing to add to and in fact appears to rely on and revisit the results of a previous report – the Specialist Study: Geology and Seismic Hazard, Council for Geoscience Report number: 2007-0277 which was part of the Pebble Bed Modular Reactor Demonstration Power Plant (PBMR DPP) Environmental Impact Assessment and Environmental Management Programme.

Question: Has any new and additional work (theoretical or field-based) been carried out since the PBMR DPP report or is the current report based on the results presented in that report?

Considering the importance of a thorough seismic evaluation of any site for an installation as potentially hazardous as a nuclear power plant (in the words of the report itself: “’Local vibratory ground motion’ resulting from geological-related seismic events (fault rupture), which, in terms of potential consequences, constitutes the most serious geo-scientific threat to a NPS”), this report is highly problematic and its recommendations and conclusions are, at best, questionable.

My comments focus particularly on the assessment of the Duynefontein site, but the majority are generally applicable to all three proposed sites.

Evidence for a major historic earthquake

The report acknowledges that a major earthquake has occurred in the close vicinity of the Duynefontein site in historic time:

“Reliable evidence for a large earthquake with an intensity of VIII, and ML 6.3 (Brandt et al., 2005) having occurred in 1809 within 25 km of Duynefontein comes from historical records of its secondary effects. […] Dames and Moore (1976) concluded that enough circumstantial evidence exists for the presence of a NW striking fault offshore of Duynefontein but that it does not come closer than 8 km to the site. It is however possible that such a fault could pass anywhere between 7 and 10 km offshore of Duynefontein (the inferred Melkbos Ridge fault passes 7.5 from the Koeberg NPS). No new research has been performed to confirm or refute the presence of the fault or its point of closest approach to the site.”

The fact that the source (fault) responsible for this historic earthquake has not been identified with certainty does in no way preclude a recurrence of a similar event in future and neither should the fact that it took place over 200 years ago inspire any confidence since such a time period is practically instantaneous in geological terms.

The important conclusion, studiously avoided by the report’s authors, is that there is no reason to believe that major earthquakes will not happen in the close vicinity of the Duynefontein site again.

The earlier report for the PBMR DPP mentioned above is more explicit:

“Detailed work will have to be undertaken if a new location is chosen on this site.The questions around the 1809 to 1810 seismic events and the existence of the Milnerton fault have to be further resolved.” […]

“Whatever the cause of the earthquake, its effects imply that peak ground
accelerations (for M 6 proximal events) between 0.2 and 0.3g were attained
(Talwani and Gassman, 2000) 11 km south of Koeberg.” […]

“The seismic hazard model should therefore take into account the possibility that a fault capable of producing an event at least equal in size to the 1809 event of inferred M 6.3 magnitude, and with a minimum recurrence interval of about 200 years, is located about 8 km SW of Koeberg.”

Where the PBMR DPP report called for ”detailed work” that “will have to be undertaken” to resolve “the questions around the 1809 to 1810 seismic events and the existence of the Milnerton fault”, the current report acknowledges that “no new research has been performed to confirm or refute the presence of the fault or its point of closest approach to the site”.

Question: Why has no new detailed work been done to resolve this issue as recommended by the previous PBMR DPP report?

Poor quantitative data

The report is remarkable for its extreme sparsity of quantitative scientific data. While qualitative and observational data is, of course very valuable, for a report of as much significance as this one, the presentation of more measurable quantitative information would have been of paramount importance.

The only quantitative measure evaluating potential seismic activity presented in the entire report is the peak ground acceleration or PGA expected at the three sites during future earthquakes. The values given are:

Thyspunt 0.16g,
Bantamsklip 0.23g, and
Duynefontein 0.30g.

The report assures us that:

“None of these exceed the PGA of 0.3g typically used in the seismic design of NPSs, although the values for the Bantamsklip and Duynefontein sites are close, or at this threshold.”

The fact that this seismic hazard assessment includes only a single quantitative measure is worrying.

Question: Why were no additional quantitative parameters of seismic risk assessment determined?

The report acknowledges that the PGA values quoted for the three sites were determined using outdated methodology (Parametric-Historic PSHA, or Parametric-Historic Probabilistic Seismic Hazard Analysis) which does not conform to the latest internationally accepted practice:

“Parametric-Historic methodology previously employed for SHA [seismic hazard analysis] of these sites, does not include all the aspects recommended in the latest regulatory guides for NPPs [nuclear power plants]. As a result, the ground-motion values calculated using the Parametric-Historic PSHA are not directly comparable in a meaningful manner to those calculated using a PSHA as defined in RG 1.208 and needs to be confirmed. A new and advanced Probabilistic Seismic Hazard Analysis (PSHA) will therefore be undertaken, that will follow the latest internationally accepted practice, and in particular, will conform to the requirements of a Level 3 study as defined in the SSHAC Guidelines (Budnitz et al., 1997).”

The only quantitative assessment of seismic risk included in the report is thus not in conformance with accepted international standards.

Question: When will “a new and advanced Probabilistic Seismic Hazard Analysis (PSHA)” following “the latest internationally accepted practice” be undertaken?

For what they are worth, the PGA values cited in the report are quoted to an accuracy of two decimal places. While the report makes a point of highlighting that the study of seismic phenomena is subject to substantial degrees of uncertainty, associated with both the apparent randomness of the physical processes involved in earthquakes and the seismic waves they generate (this is referred to as aleatory uncertainty) and the lack of sufficient data and knowledge (epistemic uncertainty), no indication of the statistical uncertainty associated with the quoted PGA values is given.

Without knowledge of the associated uncertainty, however, these values become questionable. This is particularly significant in the case for the Duynefontein site for which the quoted PGA value of 0.30g is identical to the value “typically used in the seismic design of NPSs”. The mathematical error associated with the PGA values may, for all we know, take them significantly above this “typical” threshold. Not quoting mathematical error limits with the only quantitative measure cited in the entire report is scientifically sloppy at best.

Question: What are the errors associated with the PGA values quoted in the report?

It should also be noted that reference to a “typical” value for “the seismic design of standard export NPSs” is not particularly enlightening, especially considering the fact that the specific design for Nuclear 1 is yet to be decided upon.

Internationally acceptable data will only be available pending further study and the report acknowledges that the PGA values quoted may increase based on this additional research:

“The findings presented here still needs to be confirmed by a more rigorous PSHA and may increase or decrease these values.”

This is particularly worrisome for all three sites, but especially for the Duynefontein site where the PGA value is already on the threshold “typically used in the seismic design of NPSs”. According to the report’s own observations the PGA values for all three sites may potentially rise above the 0.3g threshold once internationally acceptable methodology is used to recalculate these parameters.

Importantly the earlier PBMR DPP report mentioned above provides some additional and very significant information:

“The maximum possible earthquake for this region calculated by the Parametric-Historic Procedure (Kijko and Graham, 1998, 1999) is expected to be M 6.60±0.3 and the deterministically calculated Peak Ground Acceleration is 0.27 g ±0.14”

Question: What is the reason for the discrepancy between the PGA value of 0.27g quoted in the PBMR DPP report and the PGA value of 0.30g quoted in the current report? Was the PGA value of 0.30g in the current report determined using different methodology and/or data than the value of 0.27g from the PPMR DPP report?

The error quoted in the PBMR DPP report is significant since it pushes the potential PGA value to a maximum of 0.41g which is substantially higher that the threshold of 0.3g “typically used in the seismic design of NPSs” according to the current report. If the error associated with the PGA value of 0.30g presented for Duynefontein in the current report is comparable, the maximum PGA value would be pushed to an even higher value of 0.44g.

Mitigation measures

According to the report, mitigation measures against the effects of earthquake activity at the proposed sites would include the fact that:

“The geotechnical and structural civil engineers shall assign the appropriate “seismic design criteria” for the design of utilities, including on-site and off-site water reservoirs.”

It is, however, entirely unclear how these engineers would be able to determine what would constitute “appropriate” seismic design criteria based on data that do not conform to international standards. References to additional future studies are of little use in the context of assessing the value of the current report.

The need for a scientific peer review

Considering the importance of a thorough assessment of seismic risk for the siting of a nuclear power plant, this report should urgently be subjected to a rigorous peer review process during which it (as well as all available raw data and detailed expositions of any methodologies employed) would be scrutinised by several independent and disinterested seismologists and geotectonicists of international reputation. As it stands the report expresses merely the opinion of its authors and it would not even be considered for publication in an internationally recognised scientific journal. Why should it be sufficient or acceptable as a basis for a decision that may affect the lives and livelihoods of several generations and the health of large stretches of the natural environment?

Unscientific conclusion

Considering the information and observations presented in the report itself, the key conclusion drawn by its authors comes as a major surprise. It would appear that they arrive at a predetermined conclusion regardless if it contradicts some of their own observations noted in the report and some of which I have discussed above.

The key conclusion of the report is as follows:

“Based on current knowledge, the three localities under review are considered suitable locations for standard export NPS’s following the extensive Nuclear Siting Investigation Programme (NSIP). To date no geological evidence has been found that would halt the development of a NPS at any these sites.”

Even on superficial inspection, however, this conclusion is disingenuous and scientifically flawed. An honest assessment of the actual available data presented in the report is more accurately contained in the following paragraph:

“the available data indicate that the Thyspunt site has the lowest seismic risk of the three proposed NPS sites, and from a seismic point of view, Thyspunt is the preferred site of the three proposed NPS sites. Furthermore, in the light of the uncertainty as to whether the revised PSHA will result in significantly different PGA values, Thyspunt is the site with the biggest seismic margin to accommodate changes to this value.”

The report uses outdated methods to calculate a single quantitative measure to assess the seismic risk associated with each of the three proposed sites. The best that can be done using this flawed scientific data is to rank the three sites in an at best semi-quantitative manner – thus the identification of Thyspunt as the preferred site. These data cannot, however, be used to assess the actual seismic risk at the sites on the basis of internationally acceptable scientific methods (even at Thyspunt, the flawed data cannot be used to discount the possibility of possible significant seismic risk). The key conclusion cited above is therefore all but meaningless until such time as data that are consistent with international scientific norms are available.

Quite contrary to the official conclusion, the precautionary principle – a principle that must be applied when assessing technology as sensitive and potentially dangerous as nuclear energy – suggests that on the basis of the information presented in the report, the Duynefontein site should not be considered as a suitable site for a nuclear power plant until scientific evidence suggesting otherwise can be presented.