A protective system is only as good as the thing that has to move.
Software can decide correctly, a forecast can arrive in time, an operator can send the command, and none of it matters if a gearbox stalls.
That is not a theoretical worry.
It happened on one night in Texas, in one small section of one array, while everything around it did exactly what it was told.
The result is the cleanest test anyone has run on this question.
So what is the command supposed to achieve?
Turning a direct hit into a graze
Damage from a hailstone does not depend on how fast it falls. It depends on how much of that speed is pointed into the glass.
Momentum arriving at an angle splits in two: a component perpendicular to the surface, which does the breaking, and a component along it, which does almost nothing.
Tilt the panel toward vertical and the perpendicular share collapses.
Tempered glass helps that along, because it fails from concentrated point loads rather than distributed ones, and a glancing strike smears the contact across a wider area instead of driving it into one spot.
The panel has not become stronger.
The load has become smaller, and the stone leaves most of its energy in a skid down the face.
Which is why a single axis tracker already owns the fix: it rotates all day anyway, so protection costs a command, not hardware.
The county and the machines standing in it
The ground here is Fort Bend County, southwest of Houston, which now carries more than 1.2 gigawatts of utility scale solar.
Four large arrays sit inside roughly nine miles of each other.
They are built the same way: modules bolted to a torque tube, the tube driven by a gearbox and motor at intervals down the row.
One of the neighbors is rated at 270 megawatts. The array that took the damage runs 350 megawatts across 3,300 acres, enough for something like 62,000 homes on a hot afternoon.
Nothing separates them technically in any way that would predict the outcome.
They differ in what their trackers were doing when the sky opened.
Two storms after lunch and one at two in the morning
Two major hailstorms crossed the county after midday on March 15 of 2024, carrying stones of roughly one to three inches with gusts near 51 miles an hour.
A third came through in the early hours of March 16, which is a rare time for hail and a difficult time to have somebody watching radar.
Engineers set the severity threshold at 2.5 inches, the size expected about once in 500 years at that location. All three storms cleared it, and the worst hit array saw stones above four inches.
Three neighboring projects had stowed their rows at 52 degrees.
Two of them came through with no direct hail damage at all.
The third lost a few dozen modules, and the review located them precisely: an area where a tracker motor fault had prevented complete stow, with hail and wind blown objects doing the rest.
What one motor does not prove
Angle is not the only variable in play, and the comparison does not isolate it.
Module thickness, front glass construction, frame design and tracker stiffness all feed into whether a given panel survives a given stone, and none of those were held constant across four separately procured projects.
The protocol also has hard boundaries. A fixed tilt array cannot stow at any price, because there is nothing to turn.
A stow needs a forecast trigger that fires early enough, motors that answer, and a decision made at an hour when nobody wants to make one.
It costs generation for every hour the rows stand on end, which is a real argument on a summer afternoon and a trivial one at two in the morning.
All of that sits against the 25 year promise sold with every module, which rests on a certification floor rather than any hail rating, and against the tracker handling now being taught at dedicated training sites.
The part of the system nobody budgets for
Tracker motors are maintained on a production logic.
A stuck row costs a fraction of a percent of a plant’s output, so it sits on a list and gets attention when a crew is passing.
That arithmetic is correct almost every day of the year and completely wrong on one of them.
On the night the stow is called, a motor that will not turn is not a rounding error in generation. It is the only glass that breaks.
Which turns drive maintenance from a yield question into an insurance question, and those two live in different budgets at most operators, as the forensic review lays out.
A hail stow is cheap, and the actuator it depends on is cheaper still, as trade coverage of the case noted.
That is the awkward part of the result.
