Following the 16 July Niigataken Chuetsu-Oki earthquake, TEPCO’s Kashiwazaki-Kariwa – the world’s largest nuclear power station, with five 1100 MWe boiling water reactors (BWRs) and two 1356 MWe advanced boiling water reactors (ABWRs) – will remain shut down, pending the resolution of some difficult and troubling seismic safety issues.

This is going to take some time. And even for a utility the size of Tokyo Electric Power Co, the forced closure, for an as yet undetermined period, of around 8.2 GWe of nuclear plant (which is not far short of China’s entire current nuclear installed capacity) is no small matter.

TEPCO estimated its maximum system demand for the summer (assuming no heat wave) to be 61.1 GW and said it had been “able to secure 62.14 GW without Kashiwazaki-Kariwa.” However the expectation is that TEPCO’s profits will be cut by around 80% (about 2 billion dollars) this financial year as a result of the closure.

Once again a Japanese nuclear utility finds itself providing a dramatic illustration of one of the downsides of large centralised generation. Advocates of distributed renewables will also be able to point out once again that it is not just wind power that can be said to suffer from the problem of intermittency – large centralised plants, notably nuclear units, can also sometimes be intermittent, but on a grand scale.

There is also the catastrophic effect all this will have on public confidence in nuclear power – fragile at the best of times – and which TEPCO has been working earnestly to restore in the wake of the maintenance and inspection record falsification fiasco of five years ago.

From information put out by TEPCO and NISA (the Japanese nuclear safety regulator), as well as the excellent report of an International Atomic Energy Agency (IAEA) seismic safety expert mission, which visited the site on 6-10 August, the events of 16 July are pretty clear, less clear are the implications for the seismic protection of nuclear plants, both in Japan and elsewhere.

The good news is that the three units (3, 4 and 7) that were operating at full power at the time of the Richter 6.8 earthquake, as well as unit 2, which was in the process of being started up, tripped automatically and shut down safely as they are designed to do in the event of a seismic disturbance being detected at the site (units 1, 5 and 6 were already down for their annual outages). And there appears to have been no damage to safety related systems.

However, there was a good deal of damage to other systems in and around the plant. NISA says “64 troubles” were found and investigated. Particularly notable were:

• A fire in the unit 3 in-house transformer that lasted two hours, until it was extinguished by the local town fire brigade, revealing inadequacies in the on-site fire fighting system. The fire was caused by sparks from a short circuit (due to movement of the foundations) igniting oil that had leaked from the transformer.

• A small (1.2 cubic meter) spillage of mildly radioactive water (90 000 Bq in total) due to the spent fuel pool water of unit 6 “sloshing” (the IAEA mission’s term) about during the earthquake. This somehow found its way outside (via a defective cable penetration, which allowed it to drip into a discharge water pit in a non-controlled area, whence it was pumped into the sea). However the estimated dose is miniscule, about 2×10-9 mSv (compared with around 2.4 mSv from natural sources annually, about 0.19 mSv for a Tokyo/New York round trip by air and a Japanese statutory limit of 1 mSv for doses to members of the public).

• A small leak of radioactive iodine and other particulates into the air via the main stack, amounting to a total of 400 million Bq discharged, with an estimated dose of 2 x 10-7 mSv. The source was the main turbine condenser and, according to the IAEA report, “the root cause was the delay by the operator in manually stopping the turbine gland steam ventilator.”

Other incidents arising from the earthquake included drive shaft failure in the ceiling crane of unit 6, water spillages on the operating floors of several units, fluorescent light fittings falling from control room ceilings (causing a minor shoulder injury to one of the operators in the case of unit 6), falling platforms in spent fuel pools, HVAC diffusers becoming detached, tipping of cabinets, fire suppression piping failure causing flooding, failure of heat exchanger anchorages, failure of a condenser seawater connection, damage to/significant displacement of ducts connected to main exhaust stacks and damage to roads.

The power plant was in fact subject to much stronger seismic effects than were assumed in its design, suggesting the existence of active faults nearer the site than had been thought. The epicentre of the earthquake was 16 km from the plant, with a hypocentre at a depth of 17 km. The peak ground acceleration recorded at the site was 680 Gal (the rather strange unit that seismic engineers seem to use), which is 0.68 g. But the plant was only designed to cope with 273 Gal, ie 0.27 g, which is basically the seismic design basis commonly assumed in current reactor designs (see the table on p 16).

Thankfully, key systems proved very robust.

As the IAEA mission reports, there was no loss of offsite power, which has generally been found to occur at about 0.25 g or above (a testament to the ruggedness of the Japanese T&D network under earthquake conditions).

Inside the plant, the IAEA report also notes that “safety related structures, systems and components … seem to be in a much better general condition than might be expected for such a strong earthquake” – perhaps due to the combined effect of “conservatisms introduced at different stages of the design process.”

But this is probably more due to luck than judgement. The basic issue is that the seismic inputs that a plant might be expected to see are being underestimated. Well before the 16 July earthquake there was clear recognition by the Japanese regulators that seismic safety issues needed to be revisited, and indeed Chubu Electric’s Hamaoka 1 and 2, which started up in the 1970s and are located in an area where an earthquake of magnitude Richter 8 or more is thought to be possible, have been shut down for an extended period for seismic modifications. There have also been, within just the past couple of years, two other instances where Japanese nuclear units have experienced earthquakes exceeding their design basis response spectra (again without damage to safety systems).

In 2006 a revision of the country’s 1978 seismic guidelines was completed, a five year long process triggered by a Richter 7.3 earthquake in 2000, which was found to have originated where no active fault was thought to exist. These guidelines set out improved methods for reviewing seismic design, including identification of faults and evaluation of seismic hazard, and starting in late 2006, are being used to re-evaluate seismic safety at every nuclear reactor in the country. The events at Kashiwazaki-Kariwa underline the importance of completing these evaluations as soon as possible, and then reappraising the new guidelines as a matter of urgency.

Twenty years ago – MPS September 1987

“An order from an American electrical utility to supply and install 12 gas turbine units to convert the Midland nuclear power plant in Michigan has been won by Brown Boveri, Switzerland.”

“A new operating company named ASEA Brown Boveri will be formed with effect from 1 January, 1988.”

“Following the successful operation of two pilot scale wave power projects in Western Norway, a larger station is now being planned”