Published: August 25, 2026 | Updated: October 4, 2026
TECHNOLOGY & SYSTEMS
Japan tightened how wind operators must handle lightning twice in two months. On 6 August 2026 the Ministry of Economy, Trade and Industry (METI) revised its interpretation of the technical standard for wind power equipment. That document spells out what an operator must do to meet the legal safety level. Operators must now be able to know at all times whether a strike occurred, and inspect inside the blade with a borescope when they cannot. Then on 25 September the final report on April’s blade failure at Oga, Akita, showed the causal strike had been recorded. It was 129 coulombs, below the turbine’s 300-coulomb stop setting, and the monthly patrol had checked spinning blades by eye. On 28 September METI asked every operator in Japan to stop turbines for monthly blade checks and use telescopes or drones. The stop threshold itself is still each operator’s call.
👉 Japan’s Offshore Wind Technology Roadmap 2026
→
Execution Reality
→
Bankability Test
Installing a detection device used to satisfy the rule. Now the operator has to be able to say whether a strike happened, by any method, and must borescope the blade for any period it cannot account for.
The Oga blade was damaged by a 129-coulomb strike on 20 March 2026, outside the period when detection was switched off. The device recorded it, but the stop setting was 300 coulombs. A record without a stop threshold and an inspection method that can see damage did not prevent the failure.
METI asked all operators to review their safety regulations so that monthly patrols are done with the turbine stopped, using telescopes, telephoto cameras or drones, and are recorded. It still set no coulomb figure for when a turbine must stop.
August 2026: from installing a device to knowing what happened
Two provisions changed, sub-items (ha) and (ni) of Article 7, paragraph 6, item 1(a). Placing the old and new text side by side shows what kind of change this is.
| Provision | Before | After (6 August 2026) |
|---|---|---|
| (ha) Down conductor | Install a down conductor, the cable that carries strike current from the blade tip safely to ground | As before, plus: where conductive materials such as carbon fibre reinforced plastic are used in the blade, that material must be properly electrically bonded to the down conductor |
| (ni) Knowing about strikes | Install a lightning detection device so the turbine can be stopped immediately when a strike occurs | Take measures so the presence or absence of a strike can be known at all times, through detection devices, monitoring cameras, use of weather information or other means. Where those measures fail to establish whether a strike occurred, confirm whether damage exists by inspecting inside the blade with a borescope or equivalent. While the turbine is operating, a detection device must still be installed so it can be stopped immediately |
The (ni) rewrite is the substance. The old provision was satisfied by the presence of hardware. Whether that hardware was working at a given moment was a separate question the rule did not reach. The new provision is satisfied by knowledge. It does not care which method produced the knowledge, but it does care about periods when no method produced any, and it prices those periods at one internal inspection each.
The final sentence was not in the draft. Several commenters asked whether knowing “at all times” meant two or more independent measures were mandatory, and METI responded by adding the clause requiring a detection device during operation. Read together, the new latitude applies mainly to periods when the turbine is stopped or without power.
METI explained the background in its comment responses. On the turbine that failed in April 2026, the lightning detection device had not been operating for a period while the turbine was stopped. Strikes in that window could not be identified. The August text was written to close that gap.
The event was the second blade failure in Akita, following one in May 2025, with the same turbine model and the same maintenance provider. We covered those overlapping conditions when the second failure was reported.
What the August rule left to operators: thresholds and existing assets
Forty-four comments were submitted. Reading the responses, the questions METI settled and the questions it deliberately left open separate cleanly.
| Issue | What operators asked for | What METI answered |
|---|---|---|
| Definition of a strike | Detectable strikes can occur dozens of times a day. Publish a coulomb threshold | No threshold. Operators may set their own, based on equipment structure and lightning characteristics, and record the resulting inspection scope in their safety regulations, the rulebook each operator writes and files with the ministry, provided the basis is reasonable |
| Existing assets | State whether the revision applies, and any transition or grace period | The interpretation has no concept of retroactive application, but each operator is expected to take measures sufficient to meet the ministerial ordinance. No grace period was specified |
| Weather data | Release JMA lightning data free of charge. Nowcast is image data without an API | Operators should obtain weather information themselves if they need it |
| Scope | Limit the rule by turbine rating or proximity to inhabited areas | No limitation applied |
The first row carries the most weight. Until it is settled what counts as a strike, it is not settled when the inspection duty is triggered. Operators set that line themselves and must be able to defend it. The treatment of existing assets sits in the same open space: no retroactive application, and an expectation of compliance with the ordinance, stated together.
One comment was blunt. It noted that the operator in the reference case was already inspecting monthly, so changing inspection cycles could not be expected to prevent recurrence. METI answered that the revision aims to make lightning protection more effective and that it would act further as investigations produced findings. In August, that exchange was left open.
25 September: the strike was recorded, and fell below the stop setting
On 25 September the operator of the Wind Kingdom Oga wind farm in Oga, Akita (four 1,870 kW turbines) submitted its final report to METI’s working group on natural-disaster countermeasures for electrical equipment. The causal strike fell outside the unmonitored window of 5 August 2025 to 14 January 2026 that had prompted the August revision.
On 20 March 2026, a 129-coulomb strike hit the blade away from its receptor, the metal fitting set into the blade surface to take lightning. Coulombs measure the electric charge a strike delivers. The strike penetrated the blade shell, and internal discharge and pressure weakened the tip joint and surrounding structure.
The detection device logged the strike. The turbine’s automatic stop, however, was set at ±100 kA or 300 coulombs, so the turbine kept running and no post-strike inspection followed. The next monthly patrol checked the blades by eye, from the ground, with the rotor turning. The damage was missed. On 12 April the blade met its highest wind load since the strike and failed.
The August revision addressed the hours when the detector was off. The final report describes a failure that happened while the detector was on and recording. A record that never breaks would not, on its own, have prevented it. What was missing was a stop threshold that acted on the record, and a way of looking that could find the damage.
28 September: METI’s request to every wind operator
Three days after the final report, METI’s Electricity Safety Division wrote to all wind power operators in Japan and asked them to review their safety regulations, filing amendments where needed. The letter is a request, and the technical standard itself was not changed again. Set side by side with what happened at Oga and the operator’s own fixes, it reads as follows.
| Item | At Oga | Operator’s fixes (25 Sept) | METI request (28 Sept, all operators) |
|---|---|---|---|
| Stop threshold | Auto-stop at ±100 kA or 300 C. A 129 C strike did not stop it | Lowered to ±1 kA or 1 C, the device’s lowest setting | Set the strike level that stops a turbine appropriately. Without a detector, stop ahead of approaching lightning using weather data or an equivalent |
| After a strike | No inspection | At ±20 kA or 50 C and above, maintenance contractor inspects the blade, including a continuity test | Check for damage after every stop |
| Monthly patrol | Rotor turning, checked by eye | Turbine stopped, checked with telephoto camera or drone images | Unless no new strike could have hit since the last inspection, stop the turbine, check the blade surface with telescopes, telephoto cameras or drones, and keep records |
| Numerical threshold | None | 1 C and 50 C, set by the operator | None given |
The table answers the comment left open in August. Monthly patrols did happen at Oga. What needed to change was how they looked. Every element of the request, stopping the rotor, using an optical tool, keeping a record, changes the content of the patrol, and none of it changes how often the patrol happens.
The number that decides when a turbine stops was again left out. The Oga operator chose the device’s floor of 1 coulomb. For the rest of the fleet, the level will be whatever each operator writes into its safety regulations and can justify.
14 of 29 blade failures were lightning, and surface checks have limits
Lightning gets this attention for a reason. In METI’s tally of blade failures, compiled from accident reports for fiscal 2021 to fiscal 2025, lightning caused 14 of 29, or 48%, with fiscal 2025 figures still provisional. METI also notes many cases where blades fitted with receptors were struck on the surface instead, which is the Oga pattern.
Stopping the rotor and looking through a lens still leaves damage that does not show on the surface. In the same material, analysing failures traced to manufacturing defects, METI observes that cracks and delamination cannot always be detected by external inspection alone. That is why the Oga operator added continuity tests, done by rope access technicians or by drone, which pass a current through the down conductor to confirm it is unbroken. Where a test fails, a fibrescope locates the fault inside the blade.
Looking inside turns things up. In the emergency inspection of the same turbine model, 211 turbines had been inspected internally as of 14 September, and 25 showed anomalies such as corrosion, gaps or melting at down conductor and receptor connections. In DeepWind’s reading, the stopped external patrol is a floor, and internal checks need to run as a separate programme.
Our January forecast, revisited: drones entered through the request
In January 2026, analysing the Arayahama failure, DeepWind made two predictions. The first, that electrical bonding between the down conductor and CFRP would be clarified, appeared in the revised (ha) in August.
The second, that drones and robotic cameras would become a de facto obligation, did not appear in the August text, which chose borescope inspection through an opening. The 28 September request does name drones, alongside telescopes and telephoto cameras, as tools for the monthly patrol. They are one option among several, and operators decide whether to use them. External surveying and internal inspection now sit side by side with different jobs.
👉 Case Study Deep Dive: The Arayahama Blade Failure Analysis
Offshore, the wait before a stopped inspection is what costs
Neither the August revision nor the September request mentions offshore or at-sea installations. Both respond to onshore failures. The rules nonetheless cover wind power generation equipment as a category, and offshore projects sit inside it.
Onshore, a stopped patrol means a short pause in output while someone looks. Offshore, the work starts with getting eyes within range of the blade. A trial by Akita Offshore Wind and KDDI Smart Drone from 23 to 26 March 2026 shows one route. Covering four offshore turbines at Noshiro and Akita ports, the companies mounted an automated drone dock on a vehicle, commanded flights remotely from an office, and inspected the blades from land. Each turbine took 30 minutes to an hour, measured from the vehicle’s arrival at the launch point to the end of the inspection. The stated aim was to cut the time turbines stay stopped for blade inspection.
The trial has limits. The four turbines are fixed-bottom units inside port areas, and the blades were turning slowly or briefly paused. Land-based inspection works only when flight conditions allow. Whether the same approach reaches projects further offshore, or floating turbines, is not yet shown.
The direction is still clear. If monthly stopped checks become routine, the cost of each stop will depend more on how long it takes to begin the inspection than on the minutes the rotor stands still. A drone flown from shore shortens that wait. Continuity tests and internal checks still need people on the turbine. Who performs that work, and how much of it moves to remote operation, is the question our July analysis of O&M automation set out.
👉 Can Marine Drones Cut Offshore Wind O&M Costs? Japan’s OPEX and Bankability Test
Offshore, two design choices carry the cost: avoiding inspections that need not happen, and finishing the ones that must quickly. If strike data keeps recording while a turbine is stopped or without power, borescope inspection stays exceptional. A lower stop threshold means more stops and more checks, so access to remote, shore-based visual inspection decides how long each stop lasts. Where the detector draws power, and at what coulomb level the turbine stops, are specification choices that feed straight into annual O&M (operation and maintenance) effort.
August made operators prove they knew about every strike. September set the minimum standard for how they look for damage. The number that decides when to stop is still theirs.
The Oga final report shows that a record alone does not prevent a failure. The 129-coulomb strike was logged. The turbine kept running, and the monthly patrol looked at a turning rotor by eye. Preventing a repeat needs three things in sequence: a record of the strike, a threshold that turns the record into a stop, and an inspection method that can find damage once the turbine is stopped. August dealt with the first, and September with the third. The second remains in each operator’s safety regulations.
That threshold is now the working question for operators and maintenance contractors. A lower setting moves toward safety and multiplies stops and inspections. Offshore, each of those means moving people or equipment within reach of the blade. Which coulomb level an operator chooses, and how quickly it can close out the inspection that follows, will likely shape O&M cost and staffing for Japan’s fleet more than any further rule text.
Related DeepWind Articles
- Case Study Deep Dive: The Arayahama Blade Failure Analysis
- Japan’s First Offshore Wind Tower Factory, Akita Blade Failure, and MLIT’s Floating Wind Installation R&D Program
- Can Marine Drones Cut Offshore Wind O&M Costs? Japan’s OPEX and Bankability Test
- Japan’s First Fatal Wind Turbine Accident: Causes and Safety Measures
- Japan’s Offshore Wind Technology Roadmap 2026
DeepWind Weekly tracks Japan’s offshore wind market beyond headline announcements, focusing on execution risk, cost structure, project viability, supply-chain constraints, and policy implications.
Subscribe to receive weekly intelligence on Japan’s offshore wind market.
Why Isn't Floating Wind Bankable in Japan Yet?
The synchronized supply-chain bottleneck, quantified node by node, with a full sensitivity ranking and a bankability ladder of scaling conditions. Includes Simulator Professional (β) access to reproduce every figure on your own assumptions.
See the full analysis →