Sheet metal in cold climates: impact, contraction, sealing

A canopy that passed every inspection in Mumbai can fail in its first winter at 3,500 m. Not because the fabrication was wrong, but because the steel was specified at a test temperature 60 K above the temperature at which wind load hits it, and because the fixing detail assumed the panel would stay the same length all year.

Cold changes three things in a fabricated part, and they are usually handled by three different people. The metallurgy decides whether a notch, an impact and a low temperature combine into a brittle crack. The geometry decides whether a 12 m run of cladding pulls its fixings out when it contracts. The finish and the seal decide whether an enclosure still keeps water out when the gasket has hardened and the powder film has lost its flexibility.

A single number governs the first of those: the temperature at which the steel is impact tested. Everything else follows from stating the service temperature on the drawing, which most drawings do not do. The rest of the application-led fixes are collected in the applications and cases hub.

What does cold do to the steel, exactly?

It removes the plastic deformation that lets a part absorb energy. Steel is body-centred cubic, and below a transition temperature its yield strength rises while its fracture toughness falls, so a notch that would have blunted at 20 °C becomes a crack initiator at -30 °C. The transition is not a single temperature for a grade; it is a range, and the range moves with thickness, grain size, chemistry and notch severity.

Two numbers make this usable on a drawing. The first is the test temperature attached to the grade. Under EN 10025-2, a J2 quality structural steel is impact tested at -20 °C and must absorb 27 J in a Charpy test; J0 is tested at 0 °C and JR at +20 °C, with the same 27 J. Under EN 10025-3, the NL grades are tested at -50 °C. A drawing that says "S275" without a quality group has said nothing about cold performance.

The second is the thickness effect. A Charpy specimen is 10 mm square, and a 25 mm plate does not behave like three 10 mm plates stacked. Thicker sections have more constraint at the crack tip, which raises the transition temperature for the same steel. That is why structural codes step the impact requirement up as thickness increases, and why a 3 mm cover and a 20 mm plate bracket in the same assembly should not carry the same grade note.

For sheet metal at 0.8–3 mm the constraint is lower and the transition is less punishing, but the notch is often worse: a laser-cut edge with a burr, a spot weld nugget, a bend with a tight inside radius. The part that fails in service is almost never the flat parent metal.

Grade selection by test temperature

The table below is the version worth putting next to a drawing. Test temperatures are the ones attached to the grade, not a guarantee that the part is safe at that temperature under every load.

The row most often got wrong is the first. CRCA is the default material for enclosures and covers because it forms well, takes powder coating and is cheap, and none of those properties says anything about toughness. A CRCA bracket holding a damper in a cold-climate air handler is loaded dynamically, has bend radii and punched holes, and carries no impact evidence. The correct answer is a J2 or NL plate grade, or a redesign that takes the dynamic load out of the bracket.

How much does a 12 m run actually move?

Enough to shear a fixing or bow a panel, and the arithmetic is short. Steel moves at 11.7 µm per metre per kelvin, aluminium at 23.6 and stainless at about 16. Over a 60 K swing, from an installation temperature of 20 °C to a metal temperature of -40 °C, a 12 m steel run shortens by 12,000 × 11.7 × 10⁻⁶ × 60 = 8.4 mm. The same run in aluminium shortens by 17.0 mm.

If both ends are bolted to a rigid frame, that movement has to go somewhere. A 12 m steel member restrained at both ends develops 147 MPa of thermal stress at that swing, from 210,000 MPa × 11.7 × 10⁻⁶ × 60, and the aluminium member develops 98 MPa. Neither is a margin worth relying on, and the joint usually fails before the member does: the bolts go into bearing, the holes elongate, and the connection loosens in the second winter.

A thin panel behaves differently, and this is the part that surprises designers. A 1.5 mm panel spanning 1,000 mm between supports buckles at a compressive stress of about 1.7 MPa, from the plate buckling expression with simply supported edges and a k factor of 4. It cannot reach 147 MPa, so it does not develop thermal stress; it bows instead. That bow is the oil-canning you see on cladding in the afternoon sun, and it is a movement problem rather than a strength problem. The frame takes the load and the panel takes the movement.

The joint detail follows from the movement.

• One fixed end, one sliding end. The fixed end carries location; the sliding end gets slotted holes with the slot axis along the run.
• Slot length. Take the hole diameter, add the calculated movement and add 1 mm for installation. An M10 bolt in an 11 mm hole on a 12 m steel run needs an 11 + 8.4 + 1 = 20 mm slot, and that slot needs a square washer so the bolt does not bear on its edge.
• Installation temperature. State it. A run installed at 45 °C on a sun-heated roof and taken to -35 °C sees 80 K, which is 11.2 mm for steel and 22.6 mm for aluminium.
• Two fixed ends on a long panel. Put the movement into a lap joint, a formed expansion feature or a gasket instead of into the panel.

The same arithmetic applies to cable trays, pipe runs and busbar supports inside an enclosure, which is why a cold-climate enclosure often needs its internal brackets slotted rather than fully located.

Why does the failure start at the weld?

Because a weld toe is a notch with a radius of a few tenths of a millimetre, and a notch raises the local stress and shifts the transition temperature upwards. The parent metal can be certified to -50 °C while the heat-affected zone beside a sharp toe behaves like a much lower grade.

Three details decide it. Toe geometry: a dressed toe with a 1 mm radius has a stress concentration near 1.5, and an undressed toe with a 0.1 mm radius sits near 4. Undercut: 0.5 mm of undercut on 3 mm plate removes a sixth of the section at the point of highest stress, which is why weld quality levels limit it by thickness rather than as one number. Stop-start points: a crater at the end of a run is a built-in crack starter, so welds in cold-service parts run out to a plate edge or onto a run-off tab.

ISO 5817 defines the quality levels that make those limits contractual. Level B is stringent, level C intermediate and level D moderate, and each sets limits on undercut, porosity and toe geometry. A cold-service bracket needs the level stated on the drawing, because "good weld" cannot be inspected and level B costs more than level D.

Two workshop notes matter at low ambient temperature. Carbon steel does not normally need preheat in thin sheet, but a bay at 2 °C is not a bay at 30 °C, and most procedures call for the joint to be warmed and kept dry before welding to avoid hydrogen-assisted cracking in the heat-affected zone. And coating follows weld inspection, because 85 µm of zinc or 70 µm of powder hides a toe crack completely.

What happens to the seal when the temperature drops?

It hardens, and a hardened seal needs more force to compress the same amount. An EPDM gasket at 70 Shore A is compliant at 30 °C and noticeably stiffer at -30 °C, with a glass transition around -50 °C; silicone stays flexible to about -60 °C, and polychloroprene stiffens well before either. The practical consequence is that a door sealing at 10 N per 100 mm of gasket in summer needs closer to 15 N per 100 mm in winter, and the latch has to supply the difference.

For a 1,200 mm door with a 3.4 m gasket, that is 340 N in summer and 510 N in winter. The door frame has to carry the winter figure without bowing, because a stile that bows 0.4 mm at mid-span unloads the gasket exactly where the cold air arrives. This is why hose-test results from a warm workshop do not predict winter performance: the test measures the seal at the temperature the gasket was designed for, not the temperature it will see.

Three changes make a cold-climate door work. A gasket with a low glass transition, EPDM or silicone rather than a general-purpose blend. A latch pattern that puts a latch within 300 mm of every corner and no more than 600 mm between latches. And a door stiffener or a return flange deep enough to hold the winter gasket load with less than 0.3 mm of bow.

Condensation is the second half of the problem. A warm enclosure in cold air has a cold inner surface, and moisture condenses on it, runs to the lowest point and freezes. Water expands about 9% on freezing, so a closed hem, a lap joint or a hollow section that holds water will be forced apart over a few seasons. Drain holes of at least 6 mm at the lowest point of every closed section, and no upward-facing open hem, prevent it.

What does cold do to the coating?

It makes a good film brittle. A polyester powder film that passes a falling-weight impact and a mandrel bend at 23 °C has been tested at one temperature only, and both ISO 6272 for impact and ISO 1519 for mandrel bend are room-temperature methods unless a temperature is stated. At -30 °C the same film chips at the impact site and cracks along a bend that was previously compliant.

There are two ways to deal with that. The first is to keep coating out of the impact path: put the film on a face that does not receive stone chips, fork tines or the edge of a stacked panel. The second is to test the actual system cold, on a sample panel of the same substrate and film thickness, and record the result. Neither is expensive; assuming a 23 °C pass covers -30 °C is what costs money.

Galvanising behaves differently and is often the better answer for cold, wet service. Zinc stays ductile at low temperature and provides cathodic protection at a scratched edge, which powder cannot. Its limit is mechanical: hot-dip galvanising at 85 µm cracks if the part is bent afterwards, so the bend has to happen before the dip, and the part needs the vent and drain design that lets the zinc out of every closed section.

What Ganesh runs for cold-service parts

The parts that go into cold service are not exotic, and they run on the same equipment as everything else. Profiles are cut on the 12 kW fibre laser with its 40 ft × 16 ft bed, up to 20 mm, with 30–40 mm plate work on the iron worker. Forming is on the 200 T, 5-axis brake with a 4,450 mm bed to 10 mm, or on the mechanical presses from 20 T to 300 T when the volume carries a tool.

Welding is three CLOOS robotic MIG cells plus more than 30 manual MIG, spot and SPM stations, with fixtures designed and built in-house — which matters here because a fixture holds the joint gap that keeps undercut and toe geometry inside the ISO 5817 level you asked for. Coating runs through the seven-tank pretreatment, the 16 ft × 16 ft shot blast chamber where scale or rust needs removing, and the 18 ft × 14 ft powder booth, with DFT and cross-hatch adhesion checked on the line.

Range is 0.8 mm to 40 mm, ±0.5 mm typical on formed sheet metal and tighter by agreement, with first article inspection and PPAP documentation available and PDF, DXF, DWG and STEP accepted. The plants hold ISO 9001:2015 and ISO 14001:2015, and the same documentation route serves customers including Indian Railways, Mahindra and TKE. What we do not do is qualify a design for cryogenic service or run Charpy tests in-house: impact evidence comes from the mill certificate for the parent metal or from a third-party laboratory, and a weld procedure that needs impact testing is qualified outside.

What to put on the drawing

• The minimum metal service temperature in °C, and the installation temperature the joint is set at.
• The grade with its quality group, for example S275J2 to EN 10025-2, rather than the strength alone.
• The impact test temperature required, and whether the mill certificate must report it.
• The weld quality level to ISO 5817, plus any toe dressing requirement.
• The fixing arrangement at each end of every long run, with the slot length and slot axis.
• The gasket material, its compression and the latch pattern, where the part is a door.
• The coating system, film thickness and faces that must stay bare.
• Drain hole positions and diameters for every closed section.

The welded enclosures guide covers the door and frame detail that carries the gasket load, and the limits of a laser-cut edge at thick section are set out in laser cutting mild steel.

When cold is not the real problem

Three cases where the cold-climate discussion hides the actual cause. A part that cracks in the first cold snap, in a workshop that was 20 °C when it was made, has usually been cracked since it was made: a tight bend radius, a burr on a laser edge, or a weld toe that was never inspected. Cold only finishes the job.

A part that rusts rather than cracks has a coating problem, not a temperature problem; the low temperature simply supplies the condensation cycle that drives it. And a part that leaks has a gasket or a door-stiffness problem that was present in summer and only became visible when the seal hardened.

For parts that genuinely see -30 °C and below under load, the answer is a tested grade, a detail without sharp notches, a joint that lets the run move, and a seal that stays compliant. Send the service temperature and the fixing arrangement with the drawing to request a quotation and the grade and slot detail come back with the price.