Warping in Thermoformed Parts: Four Causes That Look Like One

Warp is a strain that was built in

A thermoformed part does not warp because it is thin or because the material is cheap. It warps because different parts of it shrank by different amounts, and the part had no choice but to bend to accommodate the difference. The bow is the visible result of an internal strain that was frozen in during cooling.

The strain comes from a temperature difference. A surface that cools first becomes rigid while the rest of the wall is still soft and still shrinking. When the rest finally cools it wants to contract, and it cannot, because the first surface is already set. The part relieves the resulting stress by bowing, usually towards the side that cooled last.

That single mechanism has four common triggers, and they are often present together. Uneven cooling between the two faces, stretch imbalance across the part, a mould temperature gradient, and release before the part is cool enough to hold its shape. Each produces a different bow pattern, which is the key to telling them apart.

The measurement that makes the conversation concrete is bow in millimetres per metre, taken with a straight edge and a feeler gauge across the longest dimension. A part with 1 mm of bow over 300 mm is 3.3 mm per metre, which is enough to fail most assembly requirements and not enough to be obvious on a bench.

Face to face temperature difference

The single number that predicts warping best is the difference between the two faces of the sheet at the moment of demoulding. It can be measured with an infrared gun on the part as it leaves the tool, and it correlates almost linearly with the bow that develops.

The chart on the left is measured on a 1.5 mm PP tray. At a zero difference the bow is 0.2 mm per metre, which is within most specifications. At 20 C the bow is 4.9 mm per metre, and at 40 C it is 7.8. The relationship is not perfectly linear because the part also stiffens as it cools, but the trend is clear enough to use as a control.

A 20 C difference between faces is easy to create without noticing. The mould side is in contact with aluminium at 30 C; the plug side is in contact with air at 22 C. That is an 8 C difference from the tooling alone, before any cooling air is added. Add a fan on one side and it becomes 25 C.

The fix is to make the two faces cool at the same rate. That usually means slowing the cooling on the mould side rather than speeding it up on the free side, because the mould side is the one that sets the surface finish and the detail. Insulating the mould slightly, or warming it, evens the rates.

Uneven cooling: the biggest single cause

Uneven cooling across a part is different from a difference between faces. It means one region cools faster than another: the thin walls cool before the thick ribs, the edges cool before the centre, the area near the vacuum holes cools before the area between them.

Thin to thick transitions are the worst case. A part with a 0.8 mm wall and a 3 mm rib has two regions with completely different cooling times, and the rib will still be shrinking when the wall is rigid. The result is a part that bows towards the rib side, and no amount of mould temperature adjustment will fix it because the problem is the geometry.

The fixes are geometric as much as thermal. A gradual transition from wall to rib rather than a step, a shorter rib, a cored-out thick section, or a change in the vacuum hole pattern to cool the thick area faster all help. On an existing tool, local air jets directed at the thick sections are the usual retrofit.

Cooling air is often applied evenly when it should be applied selectively. A manifold with three adjustable nozzles aimed at the three thickest regions does more for warp than doubling the airflow through a single wide nozzle, and it costs less compressed air.

Stretch imbalance and orientation

When a sheet is drawn into a deep cavity, the material at the bottom of the draw is stretched less than the material at the sides. The stretched material is oriented, and oriented material shrinks more on cooling. The result is a part that pulls towards the walls, which reads as a bow across the base.

This is why deep parts warp more than shallow ones, and why parts with an asymmetric draw warp in a predictable direction. The side with the deeper draw shrinks more, and the part curves towards it. A part with a deep draw on one side and a shallow one on the other will always bow, no matter how even the cooling is.

The fix is to even out the stretch. A plug that pre-stretches the sheet before the vacuum comes on distributes the draw more evenly, and a plug with a shape matched to the cavity does it better than a flat one. On a very asymmetric part, a stepped plug or a local pre-heat can be used to bias the stretch.

Sheet orientation from the extrusion process also matters and is often invisible. A sheet that was drawn down heavily during extrusion has orientation in the machine direction, and a part formed from it will shrink more in that direction. If the warp always runs the same way relative to the sheet, that is the cause.

Mould temperature gradient

A mould that is 10 C hotter at one end than the other produces a part that cools unevenly across its length, and the part bows along that gradient. The gradient is easy to create: a water circuit that enters at one end and exits at the other, a cartridge heater in one section only, or a mould sitting in a draught.

The check takes five minutes with an infrared gun. Measure the mould surface at four points immediately after a part is removed, and compare. A spread of more than 5 C is worth correcting; more than 10 C will produce visible warp on a large part.

The fix is usually a change to the water circuit rather than a new mould. A series circuit replaced by a parallel one, a baffle added to a drilled channel, or two circuits at different temperatures all even out the gradient. On a small mould, simply insulating the back face can make a measurable difference.

Releasing too hot

A part released above its heat deflection temperature has no stiffness and will sag under its own weight before it reaches the trimming station. The bow that results is not a moulding fault, it is a handling fault, and it looks identical to the others on the bench.

The symptom is a part that is flat when it comes off the tool and bowed by the time it is stacked. If the bow depends on how the parts were stacked rather than on which cavity they came from, the problem is release temperature or the stacking method, not the tool.

The fix is to hold the part on the tool longer, or to support it after release until it is cool. A simple fixture that holds the part flat for the first thirty seconds after demoulding solves more warp problems than any process change, and it costs almost nothing.

Table 1 — Reading the bow pattern to find the cause

Fixing warp after the fact

Once the part is cool the warp is set, and it cannot be removed by reheating without changing the material properties. A part that is re-heated and flattened will relax the internal stress, but it will also shrink, and the dimensions that were in tolerance will move.

The exception is a part that is warped but not yet fully cooled. A holding fixture that constrains the part for the first minute after demoulding, while it is still above its heat deflection temperature, will flatten it permanently. This is standard practice on large parts and is often the cheapest fix available.

For parts that are already warped and cannot be reworked, the only remaining option is to sort. That is worth doing only if the bow is intermittent, because a bow that is present in every part is a process problem that sorting does not solve.

Table 2 — What works at each stage, in order of cost

The order of the table is deliberate. Every fix above the line is cheaper than every fix below it, and the ones at the top are process settings rather than equipment. A shop that reaches for a holding fixture before it has measured the face to face temperature is solving the symptom at the wrong end of the process.