Cracking in Thermoforming: At the Corner, at the Trim, or Weeks Later

When the crack appears is the diagnosis

A crack is a material that has been asked to carry more strain than it can bear. Where it appears and when tells you which strain. A crack that opens as the part leaves the tool is a demoulding problem. A crack at the trimmed edge the next day is a trimming problem. A crack that appears three weeks later in a warehouse is a stress relaxation problem.

The three are often treated as one, and the fixes are then applied at random. Adding heat helps the demoulding crack, does nothing for the trim crack, and makes the delayed crack worse by increasing the residual stress that drives it.

The pattern on the part narrows it further. A crack that starts at a corner radius and runs along the wall is a strain concentration at the radius. A crack that runs along a trimmed edge is a trim quality problem. A crack that appears at the thinnest point of a deep draw is a draw ratio problem.

The chart on the right is from a production line making clear PET trays. Corner radius accounts for 41 percent of cracks, trimmed edge for 27, and the thinnest draw point for 16. Together they are 84 percent of the problem, and each has a different fix.

Cracking at a corner radius

A corner is where the sheet has to turn through a radius while it is being drawn. The material on the outside of the turn is stretched more than the material on the inside, and at a tight radius the outside surface can be stretched past its elongation at break. The crack starts there and propagates along the wall.

The risk rises sharply as the radius falls. A corner radius of 3 mm on a 1 mm sheet is a ratio of 3, which is usually safe; a radius of 1 mm is a ratio of 1, which is marginal; a sharp corner is a ratio near zero, which will crack. The design rule is a minimum radius of one and a half to two times the sheet thickness, and more on a deep draw.

On an existing tool, the fix is to reduce the strain at the corner rather than to change the radius. More heat locally, a slower draw, a plug that pre-stretches the corner, or a slightly thicker sheet all help. Local heat is the cheapest and the most effective, applied with a hot air jet aimed at the corner before the vacuum comes on.

The corner is also where the wall is thinnest, which compounds the problem. Thinning concentrates at the corner because the material there has to cover more surface. A corner that is 40 percent thinner than the nominal wall has 40 percent less material to carry the strain, and it will crack first.

Cracking at a trimmed edge

A trimmed edge is a cut through the material, and a cut leaves a stress concentration. If the trim is clean, the concentration is small and the part survives. If the trim leaves a burr, a notch or a micro-crack, the concentration is large and the part fails at that point.

The quality of the trim depends on the method. A steel rule die cuts by shear and leaves a slightly ragged edge on thick sheet. A CNC router leaves a clean edge but can leave a burr on the exit side. A hot wire seals the edge and leaves a bead, which is strong but visually different.

The most common trim problem is trimming too cold. A part trimmed before it has cooled fully cuts cleanly at the moment of trimming and then the edge relaxes, opening a micro-crack. Trimming after the part has stabilised, or annealing before trimming, removes the problem.

The second most common is trimming into a region that is already strained. A trim line that runs across a corner radius cuts through the most highly oriented material in the part, and the edge becomes a crack starter. Moving the trim line 5 mm away from the radius is often all that is needed.

The crack that appears weeks later

A delayed crack is the most expensive because it appears after the parts have left the plant. It is caused by residual stress slowly relieving itself, and it is accelerated by heat, by UV and by any chemical that attacks the polymer. A part that was fine in the warehouse fails on a pallet in the sun.

The residual stress comes from the forming process, and it is highest where the material was stretched the most and cooled the fastest. A part formed at a low temperature, drawn deeply and demoulded hot carries the most stress and is the most likely to fail late.

The test that predicts it is a heat ageing test. Hold a sample at 60 to 70 C for 48 hours and inspect. A part that cracks in that test will crack in service within a few months; a part that survives it is usually sound. The test is cheap and it is the only way to catch a delayed failure before the customer does.

The fix is to reduce the residual stress at the source: a hotter sheet, a slower draw, a warmer mould and a longer cooling time. Annealing after forming also works, at the cost of a second operation and a small dimensional change.

Demould temperature sets the delayed risk

The left hand chart is the most useful number on this page. Parts demoulded at 60 C produced 44 percent cracking within 30 days. At 80 C it was 17 percent, at 90 C it was 9 percent, at 100 C it was 4 percent and at 110 C it was 2 percent. The curve falls steeply and then flattens, and the flat part is where a production line should sit.

The mechanism is that a part released hot is still shrinking when it leaves the tool, and the shrinkage happens against the constraint of the part's own geometry. That creates a stress that stays in the part until something relieves it, and the relief is a crack.

Cooling the part further on the tool costs cycle time, and the cost is real: a part that needs to drop from 110 to 100 C might add five seconds to a thirty second cycle. Against that, the difference between 44 percent and 4 percent cracking is the difference between a viable product and an unviable one.

The measurement is a surface temperature reading on the part as it leaves the tool, taken with an infrared gun at a fixed point. It takes ten seconds and it turns demould temperature from a feeling into a setting.

Material and sheet effects

Some materials are inherently more crack prone. Polystyrene is brittle and cracks readily at a corner; PET is tougher but stress cracks under chemical attack; PP is tough but has poor low temperature impact. The material choice sets the baseline risk, and the process has to work within it.

Sheet quality matters as much as the polymer. A sheet with internal stress from a fast extrusion will crack more than a well annealed one. A sheet with a scratch or a gel particle has a built-in crack starter. If cracks always appear at the same place in the sheet rather than the same place on the part, the sheet is the cause.

Recycled content is a factor that is often overlooked. A sheet with 30 percent regrind has a lower elongation at break than a virgin sheet, and a corner that was marginal becomes a corner that cracks. If the regrind content changed and the cracking started, that is the explanation.

Table 1 — Reading the crack to find the cause

Preventing it before it happens

The first piece of a new tool should be heat aged before the run starts. Forty-eight hours at 60 C costs nothing and tells you whether the design and the process will produce parts that survive. A tool that fails this test has a design problem, and finding it at first article is far cheaper than finding it at the customer.

The second control is the demould temperature, measured and recorded for every run. A single number, taken with an infrared gun, predicts the delayed crack risk better than any other process variable. It should be on the setup sheet next to the heater time.

The third is a trim inspection. A trimmed edge under a magnifier should be clean, with no burr and no whitening. Whitening at the edge is a sign that the material yielded during trimming, and it is the precursor of a crack.

Table 2 — A control set for crack-free production

The one control that is most often missing is the heat ageing test, because it takes two days and nobody has two days at the start of a project. The compromise is to run it in parallel with the first production batch rather than before it, and to hold the batch until the result is in. That is a two day delay once, against a field failure that costs the customer relationship.