Surface Scratches: From the Sizing Sleeve to the Packing Table

A scratch is not a die line

The first distinction is between a mark made by the die and a mark made after it. A die line is a copy of an irregularity on the die land: it is continuous for the whole run, it has a fixed position on the face, and it does not change with line speed. A scratch is a contact mark: it is usually intermittent, it can change position, and it is often visible only on part of the length.

The second distinction is between a scratch and a handling mark. A scratch has a direction and a depth; a handling mark is a scuff with no clear direction. Scratches come from something moving relative to the profile at speed, which means the line. Scuffs come from the profile moving against something at low speed, which means packing and transport.

The third is between a mark in the surface and a mark through it. A shallow scratch on a glossy white profile is a cosmetic defect. A scratch deep enough to break the surface of a capped or co-extruded profile exposes the substrate and becomes a weathering and colour problem, which is a different severity and a different conversation.

Getting these three distinctions right at the start saves the most common wasted effort in this area, which is polishing a die that was never the cause. A die that produces a clean surface at low speed and a scratched one at production speed is telling you the mark is downstream.

Belt pressure has a threshold, not a trend

The chart on the left plots coil rejection against haul-off belt pressure. At 0.8 bar rejection is 1.2 percent, at 1.5 bar 2.0, at 2.5 bar 5.0, at 3.5 bar 11.0 and at 4.5 bar 19.0. The shaded zone from 2.5 bar upward is where scratches dominate the reject list.

The important feature is that the curve is not smooth. Between 0.8 and 1.5 bar the increase is small, and it is mostly the normal consequence of gripping harder. Past 2.5 bar it turns up sharply, because the belt has started to slip and recover rather than grip continuously, and every slip is a rub.

That is why the fix is not always to reduce pressure. A belt that is too loose also slips, and a slipping belt scratches more than a tight one. The target is the lowest pressure at which the belt does not slip, which is found by testing at the lowest setting and increasing until the profile stops wandering.

Belt condition matters as much as pressure. A belt with compound embedded in it, or with a hardened, glazed surface, scratches at any pressure. Belts are consumables and should be replaced on condition, not on failure, and the condition to look for is a shiny surface and embedded white residue.

The six contact points, in order

The right hand panel gives the distribution: sizing sleeve entry 28 percent, haul-off belt 24, cooling conveyor roller 18, saw infeed 14, packing table 10. Together the first two are more than half, which is where the inspection should start.

The sizing sleeve entry is the highest because it is where the profile is hottest and softest and where the clearance is smallest. A burr on the entry radius, a chip in the sleeve, or a piece of flash caught between the profile and the sleeve will mark every metre. The mark from a sleeve entry is continuous and has a fixed position, which distinguishes it from a belt mark.

The cooling conveyor roller is third because the profile is still warm when it passes over it. A roller that has stopped turning, or one with a flat spot, rubs rather than rolls. The mark is periodic, with a spacing equal to the roller circumference, which is a diagnosis in itself.

The saw infeed and the packing table are the low-speed contacts, and they produce scuffs rather than scratches. They matter on a glossy profile and they matter more in summer, when the profile is softer when it reaches the saw.

Tracing a scratch in twenty minutes

The method is to stop the line with a length of scratched profile still on it, and look at where the mark begins and ends. A mark that exists at the tank exit but not at the die has a cause between them. A mark that appears only after the haul-off has a cause downstream. This single observation usually halves the search.

The second step is to mark the profile with a felt pen at the position of the scratch and run it slowly by hand through each unit in turn. A scratch that appears after a specific unit has found its cause. It takes twenty minutes and it is far more reliable than reasoning about which unit is most likely.

The third step is to check the position around the section. A scratch on the top face comes from something above; a scratch on a corner comes from a guide or a roller edge; a scratch on the inside of a hollow profile comes from a calibration mandrel or an internal air plug, which is a different set of causes entirely.

Table 1 — Reading the pattern of a mark to find the unit responsible.

Prevention is cheaper than detection

The prevention list is short. Guides and rollers should be a material that is softer than the profile and should be free to rotate. Contact points should be the minimum number needed to control the profile, because every additional contact is an additional scratch risk. And the line should be laid out so that the profile is not dragged over anything between units.

The second item is cleanliness. A scratch is often caused by a particle: a chip of PVC, a piece of grit, a hardened drip of compound. A wipe-down of the contact points at every start-up is a two minute job that removes a fault class, and it is the first thing to be dropped when the line is busy.

The third is the packing specification. A profile stacked on a bare stillage, or strapped directly against a steel edge, will be marked in transit no matter how good the line is. A liner, a soft strap and a limit on stack height are cheap, and they are the difference between a claim and no claim.

The fourth is the record. A shop that knows which unit caused the last three scratch complaints can prioritise maintenance. A shop that treats each complaint as new will keep polishing the die and keep receiving the same complaint.

Table 2 — A contact point checklist, cheapest fix first.

What a scratch actually costs

The direct cost is the rejected length, and on a continuous profile that is usually small. The indirect cost is larger: a customer who finds scratches inspects more, and an inspecting customer finds everything else as well. The relationship damage from a recurring cosmetic defect is out of proportion to the defect.

There is also a sorting cost that is easy to underestimate. A coil with an intermittent scratch cannot be scrapped as a whole if most of it is good, so it is unwound, inspected and re-coiled, which is manual work at a rate far above the cost of making it in the first place.

Which is why the threshold in the belt curve matters commercially. Going from 3.5 bar to 2.5 bar on a line that is rejecting 11 percent of coils for scratches is worth more than the throughput lost by running a slightly looser belt. The trade is not close.