Outages from lightning strikes: anticipate the downtime

Lightning does not pick the hour. It picks the asset that was exposed.

For whoever operates renewable generation or transmission, lightning is the fastest and most expensive threat. It gives no time to react. It hits the turbine blade or the line, trips protection, and the team finds out from the machine offline or the lost load, not from the forecast. The decision to shut down preventively or mobilize maintenance usually happens after the strike, when the damage is already done.

Brazil is a particularly demanding scenario for this. There are about 78 million lightning strikes a year, the highest strike incidence in the world, and the country operates the 5th largest wind fleet, much of it without hyperlocal monitoring. Globally, strikes cause around US$ 100 million a year in damage to wind farms, and lightning accounts for 60% of blade losses (Vaisala Lightning Report 2024). Each lightning incident on a turbine generates, on average, 30 hours of downtime, plus the blade repair (ECN / European Commission).

In transmission and distribution, the picture is just as critical: lightning strikes account for 40% to 60% of unplanned outages in the Brazilian power sector, a share that climbs even higher in regions with high soil resistivity (SENDI/SNPTEE, 2016). That makes lightning one of the leading causes of line tripping in the country — not a rare event, but a recurring risk every storm season.

The strike is treated as bad luck. And bad luck, in this case, is measurable and can be anticipated.

The invisible cost: the downtime nobody sums

The strike you see is the stopped turbine or the tripped line. What it moves underneath rarely ends up in the same tally.

Each strike on a blade is generation downtime, costly repair, lost production window and asset degradation that shortens its life. Each strike on a line is unavailability time, variable-penalty exposure and risk of equipment damage. Add the emergency crew mobilization and the maintenance window consumed by the repair, and the number stops being noise. This cost arrives fragmented — as blade repair, as downtime, as variable penalty — and is rarely summed as what it was at the source: a strike that was not anticipated on the right asset.

Fixing this is urgent because lightning season returns every year, and Brazil leads the world ranking for strike incidence.

Why weather forecasting does not protect the asset

The confusion that costs money: weather forecasting is not climate intelligence.

Public forecasting says there will be a storm in the region, at a scale of about 25 km. It does not say whether turbine 47, on top of the hill, or the line crossing that stretch, has high strike risk between 2 pm and 6 pm, nor what O&M should do beforehand. For an operations center, "storm in the Northeast" does not shut down a machine or position a crew.

Climate intelligence starts somewhere else. It begins with business knowledge: which turbine is most exposed, which line cannot lose redundancy, which asset concentrates risk and which maintenance window is safe. Next comes hyperlocalization, because lightning hits the asset, not the region, and knowing the risk is at turbine 47 changes the shutdown decision. Layered on top is situational awareness: what took the machine offline last season, how the strike hit the blade, which downtime could have been avoided with a preventive shutdown. The end result is not a storm alert — it is the probability of stoppage per asset, with window and action protocol.

A bulletin reports the storm. Climate intelligence protects the asset.

The path to anticipate the strike

Anticipating means turning the forecast into an operations and maintenance decision, before lightning hits the asset. It also means preparing the team for different horizons: from an imminent strike, in minutes, to convective storms and wind ramp events that can be flagged days ahead. The path for energy has six steps:

1. Define risk through business knowledge. Register each turbine, substation and line as a monitored asset, with coordinates, hub altitude and the operational limits that trigger risk.

2. Forecast with lead time and hyperlocalization. Cross the storm condition with the hyperlocal forecast per asset. Lightning and gusts above cut-out come with 1 to 6 hours of lead time; imminent strike, nowcasting in 15 minutes to 2 hours.

3. Understand protocols, impacts and resources. For each risk level, know when to apply the preventive-shutdown protocol, when to suspend work at height and what each scenario costs.

4. Alert the right owner. O&M and the operations center receive the probability of stoppage per asset, with window and protocol, like "turbine 47, high strike risk between 2 pm and 6 pm, preventive-shutdown protocol recommended."

5. Trigger the predefined action. With lead time, you can protect the turbine, suspend work at height for safety and position the crew in the safe window, without improvisation.

6. Audit. Record the forecast risk, the action and the outcome. The trail calibrates the model and produces a report for insurance, variable penalty and compliance.

The practical difference: reacting costs a damaged blade, downtime and a crew chasing the strike. Anticipating costs a preventive shutdown on the right asset and a protected crew.

The proof from those who already anticipate

This is not theory. A North Sea offshore wind farm operating in severe conditions came to rely on hyperlocal 10-day forecasting per turbine with i4cast®, reaching 27 times ROI and 2,749% accumulated value compared to the global ECMWF model, with fewer unnecessary mobilizations and more maintenance windows captured.

Start with your most exposed asset

You do not need to cover the whole farm or grid to start. Start with the turbine or line that suffers most from strikes. Map the assets, cross them with lightning history and see where anticipation protects downtime and asset life.

Request a free climate exposure diagnosis per asset → Request diagnosis

FAQ

Can you forecast strike risk per turbine or per line? Yes. Each turbine, substation and line is registered as a monitored asset. The model crosses hyperlocal forecast, the event fingerprint and operational thresholds to generate a probability of stoppage per asset.

How much lead time do I get? Lightning and gusts above cut-out come with 1 to 6 hours. Imminent strike, nowcasting in 15 minutes to 2 hours. Convective storms and wind ramp events, 6 to 120 hours.

Does it serve generation and transmission and distribution? Yes. In generation, it protects the blade and reduces downtime. In transmission and distribution, it helps protect the asset and reduce unavailability minutes — especially given that lightning accounts for 40% to 60% of the sector's unplanned outages.

Is this weather forecasting? No. It is probability of stoppage per asset, with an action protocol and an audit trail for reports and insurance — not a storm bulletin.