NRA details Hamaoka ground-motion 'tuning'; Chubu books ¥370bn loss

Japan's Nuclear Regulation Authority (NRA) has set out what its inspectors found in Chubu Electric Power's ground-motion work for the Hamaoka Nuclear Power Station, and the list runs past the hand-picked "representative waves" that sank the plant's restart applications. A secretariat report to the Commission's October 28, 2026 meeting says Chubu arbitrarily set the period at which it joined two calculation methods, so that results exceeding its design spectrum were left out, and arbitrarily lowered the acceleration at chosen moments of one design simulated ground motion, Ss2-D. On October 29, Chubu booked a ¥370.3 billion special loss: a ¥304.9 billion impairment covering the full book value of its nuclear generation fixed assets, plus a ¥65.3 billion write-down of nuclear fuel.

The findings arrive after the applications they concern are gone. Chubu withdrew the units 3 and 4 applications in a request dated September 25, 2026, and the NRA closed them without ruling on them (Chubu files Hamaoka 3 and 4 withdrawal; NRA issues no disposition). The wave-selection practices that started it are in our report on the investigation committee's findings.

What the inspectors found

The first finding concerns how two methods were stitched together. For fault-model ground motions, Chubu combined the statistical Green's function method, suited to short periods, with the wavenumber integration method, suited to long periods, joining them at a "connection period". For the assumed intraslab earthquake beneath the site, the basic setting was 0.67 seconds, but for some source models Chubu ran several connection periods and arbitrarily picked ones in which wavenumber-integration results exceeding both the Green's function results and the Ss-D design response spectrum were not reflected, the NRA report says. The review documents gave only the basic source model's connection period.

For an earthquake on the A-17 fault, Chubu set the connection period at 5 seconds, so the wavenumber-integration results at periods of 3 seconds and below were filtered out, and only the filtered spectrum went into the review documents, according to the report. Inside Chubu's nuclear civil engineering department, it adds, people had said the evaluation should follow the site's ground properties and that the review documents should state what was being done; neither point led to a correction or an explanation in the review.

The second finding is about a single wave. Simulated ground motions matched to a response spectrum were made by generating many candidates with random phases (900, for example) and choosing one that met the fit conditions. For Ss2-D, the design spectrum for the east side of the site around unit 5, where ground motion is notably amplified, Chubu lowered the acceleration at moments it had identified as having a large effect on the facilities, confirmed the altered wave still met the fit conditions, and after another facility-impact check made it final, the report says. The changes set the acceleration to 0.8 to 0.95 times its earlier value, at several points in time. Chubu called the operation "tuning". The NRA notes that Chubu's own investigation has not yet established the details of how the simulated motions were selected.

Third, emails and other records confirm that internal helpline reports about the ground-motion results were made in 2019 and 2021 and that Chubu did not handle either appropriately, matching what Chubu told the regulator in its second report on September 14.

The report also finds gaps in how Chubu's quality management system, required by its safety regulations, covered the work. There was no individual work plan for the design basis ground motion; instructions to contractors went out in meetings and emails without specifications, and almost none of them were recorded. Management reviews of the restart review project were conducted from the standpoint of putting the schedule first, including when the design basis ground motion would be fixed, rather than nuclear safety, and an employee safety-culture survey carried comments hinting at the matter that did not lead the company to it.

Inspectors have been to Chubu's head office 13 times since January 26, 2026. Next, they will interview the contractors commissioned for work on the simulated ground motions and check related documents. The NRA will then compile the inspection results, have its SERP (significance evaluation and regulatory measures) meeting reach a provisional evaluation, and put it to the Commission. The report gives no date.

Chubu writes down the plant

Chubu says the misconduct and the withdrawal have raised uncertainty over restart timing and safety-upgrade construction, leaving its future business scenario hard to identify. For the accounts it assumed a longer shutdown, lower sales from generation after any restart and higher safety-upgrade costs; on those assumptions, future cash flows fall short of the book value of the nuclear business's fixed assets. Even the capital it spends maintaining the nuclear facilities from the third quarter on (October onward) is expected to be booked as impairment, about ¥10 billion of it, bringing the full-year special loss to about ¥380 billion.

Chubu now forecasts a net loss attributable to owners of ¥215.0 billion for the year to March 31, 2027, against the ¥160.0 billion profit it forecast on July 29, 2026.

The same day, Chubu filed its answer to the additional report order the Minister of Economy, Trade and Industry issued on March 31 under the Electricity Business Act, signed by its new president, Minoru Yasui. The summary adds five measures drawn from the investigation committee's recommendations, among them an outside committee to evaluate culture reform (planned for December 2026), compliance staff such as people with legal or risk-management experience at the head of its Nuclear Governance Promotion Department, and periodic technical audits of ground-motion work by outside experts with no interest in Chubu or the work. A survey of the final review documents for units 3 and 4 found 10 instances that read as Chubu instructing analyses by methods of doubtful validity; none is the representative-wave selection itself, and Chubu says it is still examining the facts and whether they had any effect.

Hamaoka's seismic work is not Chubu's only recent improper case, by its own account. The full report says the organizational problems the committee identified are common to others at the company, naming three: accounts with some business partners on part of Hamaoka's safety-upgrade construction left unsettled for long periods, demolition quantities rewritten in reports on units 1 and 2 to NuRO, Japan's spent-fuel reprocessing and decommissioning organization, and a cost-calculation error that left some customers of retail arm Chubu Electric Power Miraiz paying too much for electricity. The regulator's own verdict, through the SERP evaluation, is still to come.