The panel is the easy part

A cold room is quoted as a stack of insulated panels, and the panel rarely fails. What fails is the joint between two panels, the fixing that passes through one, or the sealant asked to do a job the gasket could not. By the time a freezer at minus 25 C shows ice on the outside of a wall, the insulation has been wet for two winters.

The panel is a sandwich: two steel skins bonded to a rigid foam core, cam-locked edge to edge. The engineering lives in what the sandwich does not solve — heat flow at the edges, vapour moving towards the cold side, and steel that is 3 mm shorter per 3 m at minus 25 C than at plus 30 C.

Fifty-seven kelvin through 100 mm

Take a store at minus 25 C in a 32 C room. The wall carries 57 K, and a 100 mm polyurethane core at 0.022 W/mK gives a U-value of 0.22 W/m2K before surface films. A 100 m2 envelope therefore passes about 1,254 W. At 150 mm the same wall passes 836 W, so the extra 50 mm saves roughly 420 W, or 3,600 kWh a year.

What the U-value does not describe is the edge. A panel joint has a cam lock, a gasket and a line of sealant, and the steel skins run continuously across all three. The core is interrupted at every joint, which is why the perimeter of each panel runs colder than its middle.

The dew point is decided by the fixing

At 30 C and 70 per cent relative humidity the dew point is 23.9 C. A 100 mm panel at U 0.22 keeps its warm face at about 28.5 C, so the visible surface stays dry. The problem is anything that bridges the core. A 10 mm steel tie through 100 mm of foam conducts about 0.039 W/K, or 2.2 W across 57 K, and four of them in a 3.6 m2 panel add 9 W to the 45 W the panel passes.

The heat is not the issue; the temperature at the cold end of that tie is, because it sits inside the insulation where the vapour condenses. Once the foam around a tie is wet, its conductivity rises, the cold spot grows, and the panel is in a loop it cannot leave. Hangers and brackets belong on an independent frame.

Vapour moves one way only

The vapour pressure inside a freezer at minus 25 C is about 0.063 kPa. Outside, at 30 C and 70 per cent, it is 2.97 kPa, roughly 47 times higher, and the gradient points inward all year. Polyurethane at 100 mm has a diffusion resistance around 100 times that of air, an equivalent air layer of 10 m, which is a real barrier provided it is continuous.

It is never continuous at a penetration. A drain, a cable, a sensor pocket or a door frame breaks the line, and the vapour finds it. The barrier belongs on the warm side, sealed to the panel face and dressed around every service; the cold-side skin is a second line of defence.

0.6 mm of stainless, and which stainless

A 0.6 mm skin weighs 4.8 kg per square metre in 304, light enough to handle and thick enough to take a washdown hose without oil canning. The 304 and 304L coil is offered from 0.3 to 22 mm in widths from 40 to 1500 mm, and the food-industry applications on the 304 grade page are why it is the default skin for a room cleaned daily.

The cheaper alternative is 201 stainless: 16 to 18 per cent chromium, 3.5 to 5.5 per cent nickel and 5.5 to 7.5 per cent manganese, with 0.15 per cent carbon against 0.07 in 304. It reaches 685 MPa tensile and 292 MPa yield: not a weak material, only a less corrosion-resistant one. The page lists freezers and cold rooms among its applications, which is fair for a dry wall.

It stops being fair where the skin is welded, scrubbed with a chlorinated sanitiser, or used as a floor. For those, 316L coil with 2 to 3 per cent molybdenum and 0.03 per cent carbon is the safer order, and the 316L data puts the molybdenum range in the same place.

Welding a skin thinner than the electrode

A 0.6 mm skin is welded at 30 to 50 A with matching filler and a back purge, with heat input low enough that the sheet does not dish between stiffeners. Carbon decides whether that weld survives cleaning: 201 at 0.15 per cent will sensitise in the heat-affected zone, 304L at 0.03 per cent will not. The 201 coil is stocked in 0.5, 0.6, 0.8 and 1.0 mm, which covers the skin gauges, but the grade choice is a welding decision as much as a price one.

Corners are the other weak point. A coved corner formed from a bent radius is cleanable; a square corner with a bead of sealant is a crevice that holds food debris and chloride.

Three millimetres on a three-metre panel

Stainless expands at about 17.2 microns per metre per kelvin. Over the 55 K between a warm installation day and a running freezer, a 3 m panel changes length by 2.8 mm, and a 6 m wall by 5.7 mm. The cam lock and gasket absorb some of that; the perimeter sealant absorbs the rest. A panel restrained at both ends with a rigid bead will tear the bead or bow the skin.

The same movement explains why the joint is a mechanical design and not a gap. A joint that is tight on installation day is loose at temperature, and one sealed cold is compressed when the room returns to ambient.

The floor and the door

The floor is where insulation meets ground water. A 100 mm floor panel under a freezer needs a heater mat at 20 to 30 W per square metre, because the ground beneath sits above freezing and will heave if it freezes. On a 100 m2 room that is 2 to 3 kW of continuous load.

The door is the other opening. A 0.9 by 2.0 m leaf is 1.8 m2 of warm-air interface, and every opening drags moist air in. A frame heater keeps the gasket above dew point, a square stainless tube makes a frame that does not rot at the base, and a 304 drain pipe takes the melt water away.

What the enquiry should carry

Give the room temperature, the ambient design condition and the panel thickness; the skin grade and gauge with the cleaning regime; the joint type and vapour barrier detail; the floor heater load and door size. Then say what the outside looks like, because a plant room wall in Z275 galvanized coil, a roof in PPGL and a screen in electro-galvanized sheet are three corrosion duties on one building.

None of it is expensive to state. All of it is expensive to discover after the first winter, when ice on the outside of a panel is the only drawing that shows where the vapour went.