The New York Times•
July 26, 2010
Safety Fears Raised at French Reactor
By PATRICIA BRETT
Didier Anger, head of the anti-nuclear action committee Crilan, and a former member of the European parliament, said the bid to block, or at least delay, construction of the EPR reactor had been made in a letter this month from the committee to the French nuclear safety authority, the A.S.N.
Changes to the initially proposed cladding design would affect the operating economics of the reactor and would present "a serious safety problem," Mr. Anger said. The safety regulator had not yet responded, he added by telephone last week.
Areva, the state-controlled French nuclear engineering company, is already battling construction delays and cost overruns at both Flamanville and Olkiluoto, in
Areva said last month that it would book a €400 million, or $516 million, first-half charge against cost overruns on the Olkiluoto project, bringing its total write-downs on the facility to €2.7 billion — close to the €3 billion it estimated in 2005 as the cost of the entire job. The plant was originally due to go online up last year but is now expected to be operational in 2013.
Meanwhile Areva's plant operating partner, Électricité de France, is preparing a safety report on Flamanville, to be filed with the A.S.N. by the end of this year, that will be key to the regulator's decision on whether to allow the French reactor to start working. E.D.F. hopes to start it up in October next year.
"Nuclear safety is E.D.F.'s top priority," a spokeswoman for the company said, adding that any discussion of alleged safety shortcomings in the plant would be "inappropriate" before the release of the report.
The spokeswoman was responding to questions on internal E.D.F. documents, leaked in March to the anti-nuclear coalition Sortir du Nucléaire, that raised concerns over the safety of the reactor core.
Spanning 10 years from 1999 to 2009, the documents, published by the coalition, referred to a major redesign of the core because original plans did not meet safety criteria, and indicated that despite the revamping, some issues remained unresolved. They also documented efforts by E.D.F. to conceal unfavorable data from the A.S.N. while trying to persuade the regulator to relax safety requirements.
The documents referred specifically to Flamanville, but all EPR plants have the same basic design, according to national nuclear regulators in
Areva and E.D.F. say the EPR represents a giant leap forward in terms of reactor design, performance and safety. They say that it can achieve an operating output of 4,500 megawatts — far more than earlier designs, so unit production costs will be lower; that it uses a type of fuel that will extend the fuel cycle; and that it can operate more flexibly, allowing power output to be reduced so that some reactor maintenance work can be carried out at low power, cutting down on shutdowns.
Asked to comment on the significance of safety and performance issues highlighted in the leaked documents, an E.D.F. spokeswoman played down their importance, describing the documents as merely "working papers."
However, Guillaume Wack, director of the A.S.N.'s nuclear power plant directorate, commented in more detail, saying that the documents traced various safety requirements set for the Flamanville plant and the options explored by E.D.F. and Areva to meet them.
Mr. Wack said that the basic blueprint for the EPR design, presented to the regulator in 2004, had been developed through a close 10-year consultative process.
Marketing, however, was never subordinated to that process. In 2000, while consultations with the French safety regulator were still taking place, Areva sold the Olkiluoto project as a turn-key plant to the Finnish utility Teollisuuden Voima. Construction began in 2003, with delivery planned for April 2009, but quickly ran into delays and is now more than three years behind schedule.
In 2006, with the backlog and cost overruns on the Finnish plant already reaching uncomfortable levels, E.D.F. submitted to the French regulator the preliminary safety report required for a construction permit for the proposed Flamanville reactor. Following approval by the regulator, the French government authorized construction in April 2007.
Yet, among the leaked documents, one dated February 2007 — just before the construction of the Flamanville plant was approved — reported that technical studies by E.D.F. had shown that the EPR's original core design had failed to meet safety criteria for a control rod ejection accident at high power.
Control rods, as the name suggests, regulate the nuclear reaction in the reactor vessel, and a control rod ejection accident has a domino effect, causing parts of the reactor to overheat. In a bad enough case, that will break the cladding covering the radioactive fuel rods, causing them to release radioactivity, with potentially disastrous consequences.
One of the safety requirements for the EPR is that no more than 10 percent of the rods should break during an accident. But simulations reported in the February 2007 document were described as "particularly unfavorable," pointing to "very significantly higher" damage levels of 20 to 30 percent.
According to the documents, E.D.F. engineers had concluded by 2007 that the entire core of the reactor should be reconfigured. Also, as a temporary fix, the EPR's operational flexibility should be abandoned "until after licensing, and the implementation of new changes in the methodology or the safety criteria so as to ease the operating constraints," according to a May 2007 document.
Leaked E.D.F. technical documents dating from 2007 and 2009 repeatedly argued the case for the A.S.N. to ease its safety criteria in this domain.
Sylvie Cadet-Mercier of the Radioprotection and Nuclear Safety Institute, or I.R.S.N., which provides scientific support and technical analysis to the security regulator, said in an e-mail message that E.D.F. had been proposing a new methodology "based on the identification of different limiting cases on which safety studies are conducted."
In the scenario defined as a serious accident, the zirconium alloy used to clad the radioactive fuel will melt, producing hydrogen, which — if it comes into contact with the core's cooling water — can cause an explosion violent enough to breach the reactor vessel and release radioactive materials beyond the containment building.
According to a leaked document dating from April 2004, E.D.F. has tried to find a safer cladding material. However, all alternative materials examined so far have failed to meet expectations, Ms. Cadet-Mercier said. Several other "remedy" materials are still under study, she added.
Radiation, especially in accident conditions, can cause equipment to malfunction. Equipment required to ensure reactor safety must be qualified to keep working under accident conditions involving the breaking of 10 percent of the fuel rods. But according to an April 2009 document, some equipment used in the EPR design would fail if 1 percent of the rods were to break, leaving the plant vulnerable in an accident.
If a control rod ejection accident happened at low power, the EPR's automatic shutdown mechanism could fail to operate, it indicated.
Mr. Wack said he was unaware of those data, which might not have been communicated to the nuclear regulator. The April 2009 document said that, in the absence of any existing methodology to study the problem, the E.D.F. should "prepare an internal document" on possible solutions, that would "not be transmitted to the A.S.N.".
EPR applications in Britain and the United States have led to requests for further information or tests on key issues including fuel design, equipment quality and building code standards. But the French regulator made no such requests before approving the Flamanville plant — partly, Mr. Wack said, out of recognition that "the cost of detailed testing would be too high for the industry to finance without the certainty of having at least one launch order."
The A.S.N. will soon issue limitations on radioactive and other wastes for the Flamanville reactor, based on the 2006 application, Mr. Wack said. Yet, changes to the core have modified and increased the amount and types of radioactive and chemical waste produced, a May 2007 paper showed.
If updated data, reflecting these changes, were not available, the regulator would set the limits on the basis of the information at its disposal, Mr. Wack said. "They will have to make do," he said.
Both E.D.F. and the security regulator insist that safety is a top priority. Yet Mr. Wack acknowledged that the use of mixed oxide, or MOX, fuel — a blend of plutonium and uranium oxides, for which the EPR reactor is designed — is intrinsically more risky than the uranium oxide fuel used in most nuclear plants. Because it is more highly radiated, it is more complicated to use, store, transport and manage in case of an accident, he said.
"In case of a burst fuel rod cladding, the released fission products will be more difficult to manage, more numerous," he said.
Using MOX is a purely economic decision, based on the need to reduce
E.D.F. papers dating from 2007 pinpointed MOX fuel use as partially responsible for poor safety test results relating to reactor operations in both high-power and very low-power conditions. Operations in those conditions would not be allowed unless E.D.F. could demonstrate that the safety criteria were met, Mr. Wack said.
That, however, raises the issue, mentioned by Ms. Cadet-Mercier, of how to determine the worst-case scenarios to be taken into account.
Nuclear operators would like to modify the existing methods to give more weight to statistical risk assessment, to argue that some scenarios are so improbable that they can be discounted. The problem with that, however, is that the probability of events is determined somewhat arbitrarily by industry experts, Mr. Wack said. The data must be very closely scrutinized, he said, because "we realized that if you change one little coefficient in one area, you completely change the results."