A mounting system has to let the array move without letting it go

The array is not the same size in July as in January

Aluminium has a comparatively large coefficient of thermal expansion, and a rail on a roof experiences a temperature range from below freezing on a clear winter night to well above air temperature in summer sun. Across a long run of rail, that adds up to a movement measured in millimetres rather than in fractions of one.

Millimetres sound trivial until you consider what they are acting against. If both ends of a long rail are clamped rigidly to fixings that are themselves anchored to a rafter, the expansion has nowhere to go and the resulting force has to be absorbed somewhere — by bending the rail, by loading the fixing, or by working at the roof covering.

Nothing dramatic happens on any single day. It happens a few thousand times.

Which is why some fixings are deliberately allowed to slide

Mounting systems handle this with expansion provision: a joint that permits longitudinal movement, or a fixing point that allows the rail to slide while restraining it in other directions, at intervals specified by the manufacturer. Longer runs get more of them, and the interval is a design figure rather than a judgement made on the roof.

Modules themselves are clamped, and the clamps have a specified position along the frame and a specified torque. Both matter. A clamp placed outside the permitted zone loads the module frame in a way it was not tested for, and a glass laminate that is over-constrained is a glass laminate under stress.

The failure this produces is not a collapse. It is cracked cells, a module that quietly underperforms, and occasionally a broken frame corner years after installation, none of which is easily attributed back to the mounting.

Torque is a specification, not a feeling

Every bolted connection in a mounting system has a stated torque, and both directions of error cause problems. Undertightened, the joint relaxes and the connection loosens as the array cycles thermally and as wind works at it. Overtightened, the thread deforms or the aluminium yields, and a fixing that has yielded has already lost most of its holding capacity.

Stainless fasteners into aluminium have a further property that catches people out, which is a tendency to gall — to seize partway through tightening as the surfaces cold-weld together. A seized fastener is neither properly tightened nor removable, and it is a common finding when someone tries to service an old array.

A crew using a torque wrench is a good sign, and it is one of the few things visible from the ground during installation.

Dissimilar metals on a wet roof

Put two different metals in contact where water can bridge them and the less noble of the two corrodes preferentially. A roof supplies the water reliably, and the assembly involves aluminium rails, stainless fasteners, sometimes steel roof components and occasionally copper somewhere in the earthing.

Well-designed systems handle this by keeping certain combinations apart, by using isolating washers, or by selecting materials that sit close enough together to be compatible. Poorly designed or improvised ones do not, and the consequence appears over years as a white bloom around a fixing and a joint losing its grip.

Copper against aluminium is the combination to avoid particularly, and it can arise inadvertently where a bonding conductor is terminated directly onto a rail without a suitable interface. This is one more reason the earthing arrangement belongs to a qualified installer.

Flat roofs move for a different reason

A ballasted array on a flat roof is not fixed to the structure at all in many designs, so thermal movement of the frame translates into the whole assembly creeping across the surface over years, particularly on a membrane with a slippery finish. Systems intended for this use include friction and restraint provisions for exactly that reason.

The membrane below also matters, since a rigid frame sliding on a single-ply covering will eventually wear it, and protective mats or bearing pads are standard. Where an array is added to a flat roof without them, the first sign of trouble is usually water rather than a mechanical problem.

What is worth asking about

Ask whether the mounting components are all from one system and installed to that manufacturer’s instructions. Mixing rails from one supplier with clamps from another is common, cheap and outside anybody’s tested arrangement, and it is where clamp positions and torques stop being defined.

Ask also what the expansion provision is for the rail runs on your roof. The answer will be a number of joints at a stated spacing, and an installer who has that answer at hand has thought about the roof rather than repeated a standard layout.

It is a dull area with no marketing attached to it, which is precisely why it is worth a question. Nobody will volunteer it.