3D Knitting Defects: A Lot That Failed AQL 2.5 Twice, the Eight Faults Behind It, and the Fix for Each

The Lot That Failed Twice

A 4,000-piece order of E12 300D rPET totes failed its first pre-shipment inspection on a critical, failed the second on majors, and passed the third — and the eight faults behind those three inspections were almost all preventable at set-up rather than at inspection.

The first inspection found a dropped stitch at the handle root on one piece, which is a critical because it sits on the load path, plus six majors spread across barre, needle lines and an oil mark. That is seven defects against an accept number of 10, so on count alone the lot should have passed. It failed because the critical is not counted against the AQL at all: a critical defect carries zero acceptance regardless of how clean the other 199 pieces were, and that is the single most misunderstood rule in knit bag inspection.

The second inspection, run after a re-knit of the affected segment, found four majors and no critical, which is inside the accept number and should have released the lot. It did not, because three of the four majors were the same fault in the same position, and a repeating fault is treated as a systematic issue rather than a random one. That rule is not in the standard; it is in how a competent inspector reads one. A lot with three instances of one fault in one place is telling you something about the process, and releasing it on count would have shipped the problem.

How AQL 2.5 Turned 4,000 Pieces Into a Sample of 200

A lot of 4,000 pieces under General Inspection Level II draws code letter L and a sample size of 200, with an accept number of 10 and a reject number of 11 at AQL 2.5 — and under General Inspection Level I the same lot would draw only 80 pieces, which is the choice that quietly halves the sensitivity of the inspection.

The inspection level is a decision a buyer makes, usually without knowing they have made it. Level II is the default and is the right setting for a first programme with a partner whose process is not yet characterised. Level I halves the sample and is used only where a long record of clean lots justifies it. Level III increases it and is appropriate where the consequence of a miss is high, such as a bag carrying a stated load. The switching rules in ISO 2859-1 then move a programme between normal, tightened and reduced inspection based on results, and a programme that fails twice should expect to be on tightened inspection for the next several lots.

Lot size

4,000 pieces

General Inspection Level II code letter

L

Sample size

200 pieces

Accept and reject at AQL 2.5

accept 10, reject 11

Critical defects

zero acceptance, counted separately from the AQL

Observed fault rate after correction

3.1% reduced to 0.4%

Fault One and Two: Dropped Stitches and Needle Lines

A dropped stitch and a needle line look similar and have different causes: a dropped stitch is a vertical ladder left by a loop that failed to close, usually from tension below about 8 cN, while a needle line is a single vertical line of tight or loose loops caused by local resistance at one needle.

The distinction decides the fix and therefore the cost. A dropped stitch at the handle root is the critical that failed the first inspection: the load path runs through that band, and a released loop there propagates under load. The fix is a needle or tension correction plus a re-knit of the affected segment, and the prevention is a first-piece check that inspects the aperture band specifically rather than the body generally. A needle line is a major rather than a critical, because it is a surface fault and does not sit on the load path unless it happens to. The fix is to replace the needle, and the prevention is a needle change every eight working hours rather than when a needle visibly fails.

• Dropped stitch: vertical ladder, classed critical where it sits on the load path, fix at tension and needles.
• Needle line: single vertical line of uneven loops, classed major, fix by replacing the needle at that position.
• Both are vertical, so both point at a single needle rather than at the feeding system.
• Inspect the handle aperture band specifically on the first piece; that is where a critical will be found.
• Change needles on a time basis, not a failure basis: damage is invisible until it shows as a fault.

Fault Three and Four: Barre and Hole Formation

Barre is a horizontal band repeating at a fixed interval and is caused by cyclical tension variation above 1.5 cN, while a hole is a broken set of filaments around a burst loop and is caused by tension above about 25 cN on a fine yarn — one is a feeding-system fault and the other is a set-point fault.

Three of the four majors at the second inspection were barre in the same position, which is what stopped the lot being released on count. The repeating interval pointed at the cone traverse, and the underlying cause was a storage feeder left set for a heavier count from the previous run. Barre is a major rather than a critical because it affects appearance, not function, but a barre across the reinforcement band is worth treating more seriously, because the band's density is what the load path depends on. A hole is different in kind: it is a structural breach, classed critical if it is more than a few millimetres, and it usually indicates the tension setting was carried over from a heavier yarn.

Fault Five and Six: Shaping Distortion and Skew

Shaping distortion is a visible stepping or puckering where the needle selection narrows or widens the piece, and skew is a condition where courses run off square across the width — the first is a geometry problem in the file, the second a take-down problem on the bed.

Shaping distortion appears when the radius is too tight for the gauge. At E12 the minimum shaping step is one needle, about 2.1 mm, so a curve of small radius shows as a faceted edge or as a pucker where too many needles are taken out in one course. The fix is in the file: increase the radius, or spread the narrowing over more courses. Skew has a different signature — the whole piece is off square rather than locally distorted — and it comes from take-down that does not follow the width profile of a shaped body. On a whole-garment piece the take-down has to vary as the body narrows and widens, and a fixed value over- tensions the narrow sections and leaves the wide ones slack.

Fault Seven and Eight: Contamination and Yarn Irregularity

Contamination covers oil marks from the bed and handling marks from downstream, and yarn irregularity covers knots, slubs and count variation between lots — and both are classed major on appearance grounds even though neither originates in the stitch structure.

Oil marks on a knitted bag almost always come from the needle bed and the cam box rather than from handling, and they usually mean a lubrication interval was missed or a seal is weeping. They are worth tracing rather than cleaning, because an oil mark that appears on several pieces in sequence indicates a machine condition that will keep producing them. Yarn irregularity is a production partner question rather than a knitting question. A slub is a thick place in the yarn that produces a visible bump in the fabric; a knot left long snags on the needle; and a count change between lots shifts GSM outside the plus or minus 2% tolerance. All three are prevented by inspecting cones at goods-in rather than discovering them at the finished piece.

The Three Set-Up Omissions Behind Five of the Eight

Five of the eight faults traced to three omissions at set-up — no first-piece check on the aperture band, needles changed on failure rather than on a schedule, and feeder settings carried over from the previous run — and all three cost nothing to fix.

That is the uncomfortable part of the story. Nothing about this lot required new equipment, a different yarn or a longer cycle. The first omission let a critical reach the sample of 200. The second produced the needle lines. The third produced the barre that stopped the second inspection from releasing the lot. A set-up sheet with four checks — programme revision, feeder setting for the count in use, needle condition, and the aperture band on the first piece — would have caught all three before a single production piece was knitted. The cost of that sheet is a few minutes per set-up; the cost of the failure was two inspection cycles, a re-knit and three weeks.

Classification: Critical, Major and Minor Under AQL 2.5

Three classes decide whether a lot is released: a critical is a defect that makes the bag unsafe or unfit for its stated load, a major affects saleability or function, and a minor is a workmanship deviation — and only majors and minors are counted against the AQL 2.5 accept number.

The classification has to be written down before production, because it is the document that settles an argument at the carton. A dropped stitch at the handle root is critical, and no number of clean pieces in the sample makes it acceptable. A barre visible at arm's length is major and counts against the 10. A loose thread that can be trimmed is minor. Where programmes go wrong is in the middle: a buyer who classifies everything as major protects against appearance faults but will reject lots that are functionally fine, and a buyer who classifies everything as minor will ship faults that come back. Naming the three classes with two examples each is usually enough.

What the Second and Third Inspections Found

After the re-knit and the set-up sheet, the fault rate fell from 3.1% to 0.4% and the third inspection found two minors and nothing else, which released the lot — and the programme moved to tightened inspection for the following three lots under the switching rules.

The switching rules are worth understanding because they are a consequence buyers rarely anticipate. Under ISO 2859-1 a programme that fails moves to tightened inspection, which means a larger sample and a lower accept number for the same AQL, and it stays there until a defined run of clean lots earns a return to normal. That is not a penalty; it is the standard's way of saying the process has not yet demonstrated stability. Commercially it means the cost of a failed inspection is not one inspection. It is a period of more expensive inspection on subsequent lots, and it is a further argument for the set-up sheet.

Programme windows sit behind all of it: MOQ 100 pcs per style and per colourway, 7 working days of sampling, 30 days of bulk, 7 production lines at about 200,000 pcs per month and FOB Xiamen as the standard term. The reference unit cost for a standard E12 tote at specification is USD 4.80. For the tension variables behind four of the eight faults, see yarn tension control and the defects it produces.

Frequently Asked Questions

People Also Ask

The Inspection Is Not the Control

An AQL plan tells you how bad a lot is; it does not make the lot better. The lot described in this article passed on the third attempt because three omissions at set-up were fixed, not because the inspection was made stricter. Set-up checks, scheduled needle changes and a feeder setting confirmed for the count actually loaded are all free, and all three are worth more than any tightening of the sampling plan.

For the tension variables behind four of the eight faults, see yarn tension control and the defects it produces. For the wider inspection framework, see knit bag QC and testing.