A sandwich panel arrives finished: structure, insulation and finish in one piece. Which means everything still left to get wrong is in the joints, the fixings, and one decision made upstream — what the core is made of.
This is the system that made industrial construction as we know it possible: two metal skins with an insulating core bonded between them, which together carry themselves and span from one purlin to the next. It goes up in days rather than weeks, needs no continuous scaffolding, and is waterproof and insulated in a single operation. There are three pitfalls, and none of them concerns the panel at midspan. The first is the core: rigid polyurethane and mineral wool give very different thermal performance and incomparable fire behaviour, and the choice is a safety decision before it is an economic one. The second is the joints: the panel's declared performance is the panel's, not the wall's, and the difference is made by gaskets and laps, which go in before closing up because afterwards you cannot reach them. The third is the site: a cut made with an angle grinder burns the coating for metres around, and the rust spots appear a year later, all at once.
click a beat · arrow keys ← → walk the site · dashed pauses are the chemistry at work
The detail of the craftthe purlin out of plane
Checking purlin alignment and coplanarity, verifying the design span and fitting the bearing gaskets.
Run a line over the purlins before you unload the panels. A centimetre out of plane here is an open joint for twelve metres.site notebook — editorial synthesis
The declared figure and the real one
The data sheet gives the panel's U-value: the one measured at the centre, where there is only skin, core and skin. In the finished work, though, the wall is also made of joints, and every joint is a point where the two skins come close and the metal short-circuits the heat. On a panel one metre wide that joint recurs every metre: the wall's real U-value can differ appreciably from the catalogue figure, and the difference is called the joint's linear thermal transmittance.
Schematic. The joint's linear thermal transmittance is a figure the manufacturer declares for its own system: it must be requested and entered into the calculation, because using the panel U-value alone overestimates the envelope's performance.
Two cores, two worlds
This is the decision that matters more than any other in this system, and it is not made on price. Polyurethane insulates far better for a given thickness and is half the weight: where the thermal requirement is tight and the fire requirement loose, it wins. Mineral wool insulates less and weighs twice as much, but it is mineral: where the panel has to compartment, or the use imposes strict requirements, there is no comparison because there is no alternative. And the site point is that, once installed, the two are indistinguishable.
The reaction-to-fire class and any fire resistance rating are read from the marking and the documentation of the specific panel: they cannot be inferred from appearance or from the product family.
Yes, and it is one of the commonest interventions on industrial building stock, but with three mandatory checks. The first is structural: you are adding weight to a structure designed decades ago, often for lower snow loads than today's. The second concerns the existing roof: if there is an asbestos cement sheet underneath, the question is not technical but regulatory, and must be handled as such. The third is hygrothermal: a cavity forms between the old covering and the new panel, and if the old covering is not vapour-tight that cavity becomes the coldest point of the system, where indoor moisture condenses. In many cases the correct answer is to ventilate that cavity, not to seal it.
Because in many cases it is not rainwater: it is condensation. The inner skin is the coldest surface in the space, and if indoor vapour manages to reach it — typically through a joint without tape, an unsealed service penetration or a missing end closure — it condenses and drips. Telling the two apart is simple: a leak appears while it is raining, condensation appears when it is cold outside and moisture is being produced inside, even in sunshine. Confusing the two leads to redoing external seals that were never the problem.
Service life is decided by the sheet's coating, not by the core. A standard paint system on pre-coated steel in an ordinary urban environment lasts many decades; the same panel in marine air or an aggressive atmosphere needs a higher-grade system, and using the standard one means watching degradation start from the cut edges within a few years. It is also why touching up the edges is not fussiness: a site cut exposes bare steel, and corrosion always starts there before it starts in the open field.
Yes, and it is one of the system's real advantages, but with a geometric constraint: the panels interlock in sequence, so to remove one in the middle of a roof you generally have to free the joint by dismantling towards the edge, or cut the panel and close up with a made-to-measure piece and sealed laps. A clean replacement is possible where the panel is accessible from one end. It is worth keeping a few spare panels at the end of the job, because the same profile and the same colour are rarely identical years later.