The frames are still standing. That is the first thing you notice.
Rails in neat rows across a coastal field, bolted to their posts, perfectly aligned.
The first several bays of each row are empty.
What was on them is in the grass, folded and torn, with glass through it.
The damage stops partway down and the rest of the row is untouched.
That pattern is not random. It is the signature.
Why an array comes apart from its edge inward
The load that governs a ground mounted array is not the wind pushing on the face of it. It is the wind pulling the modules up.
Air moving across a tilted row separates at the leading edge and slips underneath, and what it leaves behind on the underside is suction.
That suction is not spread evenly. It is worst along the windward row and worst of all at the corners, where the flow has an edge to peel around.
Design codes know this and apply heavier load factors to perimeter zones for exactly that reason.
The other half is how a module is held. Most are gripped by clamps bearing on the frame, and a clamp holds by friction and torque rather than by anything passing through the panel.
Lose one module at an edge and the row behind it is suddenly a windward row too, which is how a failure walks inward and why the wreckage sits in a band.
What is actually on the headland
The site is near Cemaes on the north coast of Anglesey, 190 acres of it, rated at 49.9 megawatts and built to supply somewhere around 9,500 homes.
Consent was granted in 2017 to a smaller developer, and the project was later bought by the renewables arm of a large European utility.
Construction started in the summer of 2022 and the first electricity followed that year.
So the solar farm was roughly two years old when the storm arrived, which is worth saying because it has been widely described as newly commissioned.
A wind turbine elsewhere on the island was destroyed the same night.
Generation stopped, recovery work ran into the following year, and the operator said it would investigate once the site was safe to assess.
The gust numbers, and which one belongs here
The storm crossed on the sixth and seventh of December. A red warning was issued for west Wales and both coasts of the Bristol Channel.
Anglesey sat outside that band, under amber.
The station at Valley on the island recorded a peak gust of 79 miles an hour, its strongest December gust since 1997.
Three Welsh stations beat it comfortably that night. Capel Curig reached 93, Aberdaron 92, and Aberporth 85.
A figure of 96 miles an hour has followed this story around. That reading came from Berry Head in Devon, several hundred miles away, and was the highest December gust in the national record.
Attaching it to a field in north Anglesey is the error that makes the failure look inevitable when the local number does not.
Why the local reading settles nothing
A gust record is a point measurement, taken at one mast, at one height, over open ground chosen for consistency.
It is not the load a structure five or ten miles away experienced.
Exposure, upwind roughness and the shape of the ground all change what arrives at a given row, and a coastal site with nothing between it and the sea sees more than an inland instrument suggests.
Which means the honest reading is not that the array failed below its rating. It is that nobody measured the wind where the array stood.
Hardware behaves the same way under hail, where the difference between a wrecked plant and an intact one came down to one motor doing its job.
The official record of that night is public and detailed. No root cause for this site has been.
What is left in the field
Hundreds of modules torn from their rails is not only a repair bill. It is a pile of laminated glass and aluminum with nowhere obvious to go.
Purpose built recycling capacity for exactly this material is only now reaching scale, and storm debris arrives faster than any plant was sized for.
The wider question is whether perimeter detailing on large sites is being specified for the coast it is actually on.
That is answerable, but only by whoever holds the clamp torque records and the wind study, and none of that has been released.
Until it is, the case sits where local reporting left it, with residents describing the wreckage and an operator promising a review.
The frames are still standing. That remains the useful detail, because whatever let go, it was not the steel.
