Weak Corners in PVC Hollow Profiles: Reading the Spider Weld Line

A hollow PVC profile that splits along a straight seam under load is failing at the weld line left where the melt rejoins behind the die spider.

What the failure looks like on the line

Rigid PVC profiles are produced in long runs where a small thermal drift becomes a metres-long dimensional problem before anyone sees it.

Hollow PVC window or trunking profiles show a faint seam along the extrusion direction and fail or crack along that seam during bending, screw-fixing or a corner weld test. The rest of the cross-section remains sound and dimensional results are normal.

A spider or bridge die splits the melt into separate streams that rejoin downstream of the support legs; the interface must re-entangle under enough heat, pressure and residence time. If the melt arrives cool or the land after the spider is short, the rejoined plane retains a weak, poorly knitted weld line. Raising melt temperature or extending the relaxation land improves knitting, but the seam can never be made identical to the bulk.

Buyers and suppliers usually agree on the drawing and disagree on the conditions under which the drawing was checked. On a buyer's specification this distinction matters: a part can measure inside tolerance and still fail in use, because the drawing records geometry while the complaint records behaviour. Melt temperature at die is usually the first value both sides compare, and it is also the one most often quoted without the conditions attached to it.

The process window that decides it

Four variables move this result more than the rest. The ranges below are illustrative engineering guidance for a first discussion; confirm them on your own tool, material grade and part geometry before they become contract terms.

Table 1. Starting window for a joint process trial. Values are engineering guidance, not a measured survey.

Melt temperature at die is typically held at 175-195 C. Below the window the rejoining streams do not re-entangle at the weld plane Record it together with the measurement point and the moment in the cycle; a single number without a location is not a process setting.

Melt pressure at spider is typically held at 150-300 bar. Low pressure leaves the interface under-compacted after the legs When this drifts, the symptom usually appears at one cavity, one sheet position or one lane before it appears across the whole batch.

Land length after spider is typically held at 15-30 x wall thickness. Too short a relaxation land preserves the seam as a weak plane A change here is cheap during a trial and expensive after tooling release, so treat it as a controlled variable with a documented re-approval.

Haul-off speed is typically held at 0.8-1.2 x nominal. Running fast shortens the time available for interface healing Put it on the change-control list, otherwise an unannounced material or setting substitution will be mistaken for ordinary process noise.

Reading the response curve

The left panel shows weld-line strength retention (%) as melt temperature at die (c) increases: the response rises from 42 to 91 on the illustrative 0–100 scale. The dashed line at 70 marks a proposed acceptance limit, not a universal standard. Move that line only after the test method and the sample plan have been agreed, otherwise two suppliers can both be compliant and still deliver different parts.

Read the shape rather than a single point. A curve that changes quickly near the limit is more sensitive to ordinary process noise than one that approaches it gradually, and that difference should be reflected in how tightly the setting is controlled and how often it is checked.

Where the causes actually sit

The right panel of Figure 1 splits audited failures of this type into five groups. The percentages are illustrative engineering guidance for prioritising an investigation, not a measured survey and not a prediction for your line.

Table 2. Cause split and the corresponding first check. Shares are indicative and sum to 100.

Melt too cool at rejoin accounts for roughly 35% of the illustrative split. Cut a section across the suspected seam and inspect it under polarised light. Start here before authorising a material or tooling change, because it is the fastest explanation to confirm or eliminate.

Short relaxation land accounts for roughly 23% of the illustrative split. Bend a sample to failure and note whether the crack follows the spider line. Start here before authorising a material or tooling change, because it is the fastest explanation to confirm or eliminate.

Low pressure after spider accounts for roughly 18% of the illustrative split. Record melt pressure and temperature upstream of the spider, not at the barrel only. Start here before authorising a material or tooling change, because it is the fastest explanation to confirm or eliminate.

Checks before sign-off

Run the same sequence on a retained good sample and on the disputed batch so the comparison means something.

1. Cut a section across the suspected seam and inspect it under polarised light.
2. Bend a sample to failure and note whether the crack follows the spider line.
3. Record melt pressure and temperature upstream of the spider, not at the barrel only.
4. Compare seam strength on a slow run against the standard production speed.
5. Map each spider leg to a seam position on the extrudate before adjusting the die.

Keep a retained reference sample beside every result. When a batch is disputed, the comparison that settles it is almost always between the retained part and the current one under the same method, not between either part and a number on a drawing. Record the instrument, the operator, the conditioning time and the position on the part or sheet for each reading.

Writing it into the specification and the evidence pack

Seam strength, tested by the agreed bend or corner method, shall meet the stated minimum relative to the bulk wall, with the test direction and conditioning defined in the purchase specification.

Write the requirement as a behaviour at a stated condition, not as a brand of material or a machine setting. State the test method, the conditioning, the sample plan and the rule for deciding a pass. Where a document is needed, ask the supplier for documentation that covers the actual construction and check its scope against your own market before approval; a document for the base material alone does not describe the finished part you are buying.

At first-article approval, freeze the drawing revision, the material and colour identification, the process window, the tool and cavity identification, and the inspection method. Any later change to grade, site, tool, or window should trigger a new first article rather than a paper waiver. Keep retained samples from the approved lot and from each production lot so that a future comparison is possible.

Table 3. Minimum evidence pack for a first-article decision and a later batch dispute.

All ranges in this article are illustrative engineering guidance rather than measured data. Treat them as a starting agenda for a process trial, not as a certificate for a batch that has not been run.