Internal Stress in Extruded Profiles: Annealing, Quenching and Machining Distortion

The profile that measures right and fails later

A customer orders a PVC profile with a straightness tolerance of 2 mm per metre. It arrives measuring 0.8 mm. Three weeks later, in the customer's warehouse, it is measuring 3.5 mm and the assembly line rejects it. Nothing was done to it. The material did it by itself.

That is internal stress. When a profile cools, the outer surface freezes first while the core is still molten. The core then tries to shrink but is constrained by the frozen skin. The result is a locked-in stress field: compression on the surface, tension in the core. It is invisible, it is not a defect in the usual sense, and it only reveals itself when something releases it — cutting, machining, heating, or simply time.

Where the stress comes from

Four sources contribute, and they are worth separating because each has a different fix.

Table 1 — The four sources of residual stress. Most real profiles have all four, and the dominant one decides which fix works.

Differential cooling is the one people think of first and it is usually the largest. The others matter more than expected on thin-wall profiles, where draw-down can be 3:1 or higher and the frozen orientation is significant.

How to know you have a problem

Stress is not visible, so it has to be inferred. Four practical tests, in order of how much they cost.

Table 2 — Stress detection methods. The cut-and-watch test is free and catches most problems; the polarised light method tells you where the stress is concentrated.

The cut-and-watch test is the one worth institutionalising. Take a 300 mm sample from the line every shift, cut it, lay it on a flat surface at room temperature, and measure the bow at 1 hour, 8 hours and 24 hours. A profile that moves more than 1 mm over 24 hours has enough stress to cause problems downstream. Record the numbers — the trend matters more than any single reading.

Cooling rate: the main lever

Slower cooling reduces the temperature difference between skin and core, and therefore reduces the stress. That is the whole principle. The difficulty is that slower cooling costs line speed, and line speed is money.

Table 3 — Cooling strategies and their stress penalty. The graded tank is the standard compromise: hot at the entry where the wall is thickest, cold at the exit.

A graded tank runs 45–60 °C in the first section, 30–40 °C in the middle and 15–25 °C at the end. The profile enters warm water, which slows the skin freezing enough that the core can catch up, and then cools progressively. The result is a much flatter stress profile through the wall.

The cost is tank length. A graded tank needs 30–50% more length than a single-temperature tank for the same output, and the water heating is an operating cost. On a high-volume commodity profile it may not pay. On a technical profile with a straightness tolerance, it usually does.

Annealing: the step most suppliers skip

Annealing relaxes residual stress after the profile is made. It is a separate operation — an oven, a controlled ramp, a controlled cool — and it is the only way to get stress down to a level that survives machining.

Table 4 — Annealing schedules. The temperature must stay below the point where the profile distorts under its own weight — for PVC that is the limiting constraint, not the stress relaxation kinetics.

The cooling rate is as important as the soak. Cooling at 20 °C/h re-introduces most of the stress you just removed, because the same differential-freezing mechanism operates in reverse. A proper anneal cools at 3–8 °C/h, which on a batch oven means overnight.

PVC is the awkward case because its annealing window is narrow: hot enough to relax stress, cool enough not to distort. 60–70 °C works for rigid PVC; above 75 °C a long profile will sag in the middle of the oven.

Why machining is where it shows up

A profile with locked-in stress can sit in a warehouse for a year without visible change. Then someone machines a groove into it, and it moves.

Machining removes material, which removes the stressed layer on one side only. The remaining stress field is now unbalanced, and the part bends towards the cut. The amount of movement depends on how much material was removed and how deep the stressed layer went.

Table 5 — Operations that release stress. The general rule is that any operation removing material from one side only will bend the part.

The practical consequence for design: if a profile needs a machined feature, design it so the removal is symmetric, or specify annealing in the purchase order. A supplier who has never been asked for annealed profile will not offer it.

Specifying stress control

Table 6 — How to write stress requirements. The 24-hour condition is the important one; a same-day straightness check tells you almost nothing.

The single most useful clause is the 24-hour straightness measurement. It costs the supplier a day of warehouse space and it catches the failure mode that causes the most field complaints. Without it, a profile that is straight on the line and bowed in the container passes every inspection.