When a Metal Roof Moves Wrong

Start by deciding whether it is thermal at all

Thermal movement gets blamed for a lot of metal roof complaints that are actually installation defects, and it also gets missed on a lot of complaints that really are thermal. One question separates the two almost every time:

Go and look twice, at two different panel temperatures, before you take anything apart. It is the cheapest diagnostic step available and it decides which half of this page you are on.

Waviness in the panel flats that comes and goes

Likely cause: restrained expansion, showing up as oil canning.

When a panel is heated and cannot grow, the growth turns into compressive stress instead. A wide thin sheet has very little compressive capacity, so it buckles into visible waviness long before it reaches any stress that would break a component. That is oil canning, and when it is thermal it is worst in strong afternoon sun and largely gone by morning.

What to check, in order:

1. Is there a second fixed point? A pipe boot screwed tight to a pan, a snow guard fastened through the flat, a trim screwed to both the panel and the wall, a solar rail clamped to the seam and bolted to a rail that is bolted to the structure. Any of these turns a floating run into a restrained one between two anchors.
2. Are the clips actually sliding? Scribe a mark and check it across a temperature swing. A clip that was installed bottomed out will not move.
3. Is the clip binding? MCA notes that a clip interface with excessive frictional resistance can exacerbate oil canning, and that clips must have a precise non-binding fit with the panel. A crushed clip, an over-driven fastener, or a clip seated out of square all add friction.
4. Is the substrate flat? If the waviness is in the same place at every temperature, it is not thermal restraint — it is deck irregularity, coil stress from the forming, or a striation-free flat panel doing what flat panels do.

Screw holes on a through-fastened roof have gone oval

Likely cause: the panel has been moving against the fastener for years, exactly as designed to not.

An exposed-fastener roof has no sliding clip. The panel is screwed to every purlin, so it pushes directly against each screw shank. BRS describes the sequence from engineering studies: the purlins roll in the direction of the applied force until the panel-to-fastener interface can no longer sustain the bearing load, and then the panel slots by tearing and piling up material around the shank.

By the time the holes are visibly oval the sealing washer has usually lost its grip, and the fix is not simply a bigger screw in the same hole.

• Short term: oversize repair fasteners with a larger bonded washer that can bridge the enlarged hole, and butyl tape or a compatible sealant under the washer.
• Diagnostic: measure the run. Screw-down systems carry a published maximum run length for exactly this reason, and a roof that exceeded it was always going to get here.
• Long term: if the run is over the manufacturer’s limit, no fastener detail fixes it permanently. A retrofit to a clipped system, or an expansion step in the run, is the real answer.

Ticking, popping or banging noises

Likely cause: stick-slip at a clip or a bearing surface.

The classic pattern is noise concentrated around sunrise and sunset, or immediately after a rain shower hits a hot roof — MCA describes a 100 °F panel temperature change within minutes in exactly that situation. The panel wants to move continuously; friction holds it until the stress builds enough to break free, and it releases in a jump. That jump is the noise.

Where to look:

• Clip-to-panel interface. MCA lists minimising frictional resistance as an aim of a good clip design specifically because friction exacerbates thermal noise. A binding or crushed clip is the most common source.
• Panel-to-substrate contact. Panels bearing directly on a metal purlin or a rough deck squeak as they slide. A slip sheet or the specified bearing plate belongs there.
• Trim laps. Long trim pieces sliding against each other at a lap make a sharper sound than a clip does.
• Fasteners in a slotted hole. Once a screw-down hole has worked, the screw itself ticks as the panel slips past it.

A clip is bottomed out, or the panel will not move at all

Likely cause: the installation temperature was never accounted for.

This is the most preventable failure in the whole subject. A clip set on a hot deck at the end of its travel has room to contract and none to expand, or the reverse on a cold morning. From then on it is a fixed point, in a place nobody designed, and the panel between it and the real fixed point is restrained.

The arithmetic that would have caught it is two lines:

The clip needs travel greater than the larger of those, not half the total range. MCA notes that some two-piece clips carry an integral centring device precisely to prevent an installation where the travelling hook is bottomed out at the time of installation, which “prevents travel limitations and in-service jamming”.

If you find bottomed-out clips on an installed roof, the remedies are limited and none of them are cheap: unseam and reset the affected section at a mid-range temperature, change to a longer-travel clip, or accept the restraint and add an expansion joint upstream so the restrained length is short enough not to matter. Work out which before you start opening seams — the thermal expansion calculator gives you the restrained length each option leaves you with.

Panels have crept downslope

Likely cause: the fixed point is not fixed.

If the panels have moved as a group toward the eave, the anchorage has slipped. MCA calls the extreme version of this the panels shifting or “clutching” off the roof, and notes that fixity is needed both to control the direction of movement and to prevent it.

Check the cinch plate and the fastener count at the pinned end. That connection has to resist gravity load down the slope and the expansion and contraction force, and those combine rather than cancel. A fixed point detailed for gravity alone will walk.

There is a second version worth ruling out: a roof that was never pinned anywhere, where the panels were expected to hold by friction in the clips. That works until the first genuinely hot day.

Leaks that only appear seasonally

Likely cause: a detail that is being cycled rather than a hole.

A leak that appears in winter and not in summer, or vice versa, is usually a joint that opens at one end of the temperature range and closes at the other. Candidates, roughly in order of how often they turn out to be it:

1. End laps. The lap has to stay sealed while the panels either side of it strain against each other. A lap without tape sealant, a backup plate and enough fasteners opens under cycling. Butyl tape survives this; a curing sealant eventually cracks.
2. Penetrations. A boot sealed to a hole cut tight around the pipe has nowhere to go, and it tears or pulls loose at one end of the cycle.
3. Ridge and transitions. The ridge takes the movement of both slopes. If the end dams or closures cannot flex, the connection is being worked open and closed every day.
4. Longitudinal trim laps. Eave trim and gutters expand along their own length. A continuous run fixed at both ends distorts, and the distortion shows up as a leak somewhere along it.
5. Tie-ins to adjacent structure. Anything rigid between the roof and something that does not move with it.

Fasteners backing out

Likely cause: cyclic loading from restrained movement.

MBCI lists fasteners backing out and holes slotting together as the consequence of not adequately compensating for movement. The mechanism is fatigue rather than a single overload: the fastener is loaded and unloaded tens of thousands of times a year, and MCA’s figure of hundreds of thousands of thermal cycles over the life of a roof is the scale that makes it happen.

Replacing the fasteners without finding the restraint gets you a repeat in a few years. Work back to what is being restrained: a clip that will not slide, a trim screwed to two things that move differently, or a run that is simply longer than the system allows.

Everything checks out and the customer is still unhappy

Oil canning is the usual case. A flat metal panel shows any deviation from planarity because it is a large specular surface at a low angle, and some degree of it is inherent to the product — it is not a structural defect and no manufacturer warrants against it. If the waviness is the same at every temperature, the causes are coil stress, forming, deck irregularity, or over-tight fastening, not thermal movement.

What helps in that conversation is evidence: a scribe-mark test showing the panels are free to move, the movement arithmetic for the run, and a photograph of the same panel at two temperatures. That distinguishes a roof that is restrained from a roof that is merely shiny, and only the first one is worth opening up.

Related calculators and guides