Width is the dimension everybody governs. Length — specifically how much a ribbon moves in length once it leaves the mill and meets heat, humidity, tension and time — is the dimension most programmes discover only after cutting. A buyer who commissions 500,000 pre-cut lengths at a specified 100 cm receives, in a hot container or a warm warehouse, ribbon that measures 97.5 cm or 102 cm depending on the substrate, the finishing route and how the length was measured at the mill. Half a millimetre of width drift is invisible on a bow; two and a half centimetres of length drift is a scrap rate, a mis-declared carton dimension, or a finished good that fails its own packaging spec. The Smith Ribbon 258-module mill-side brand-buyer ribbon shrinkage, dimensional-relaxation and conditioning-correction interlock architecture treats ribbon length as what it actually is — a measured, conditioned, specification-governed material response — so that a cut length means the same thing at the mill, in transit, and three years later on the shelf.
Ribbon length drift is not one phenomenon but four, and they stack. Each moves the finished length, and each therefore dictates a specific control.
Every material expands when heated and contracts when cooled. Polyester has a positive thermal expansion coefficient, and a ribbon that leaves a cool, conditioned mill floor and travels through a hot container or sits in a warm warehouse changes length in the opposite direction to what the specification assumed. This is why a length measured in an air-conditioned QC room at 22 °C and 50 percent relative humidity is not the length the ribbon will present at 35 °C and 80 percent relative humidity in a distribution warehouse. Conditioning — bringing a sample to a defined temperature and humidity before measuring — is not a formality; it is what makes a length number mean anything.
Woven ribbon is under internal tension from the loom and from the winding process. Left to relax, a woven fabric does not stay at its as-woven length — it can move, and it moves most in the first hours after being taken off tension, then more slowly over days. This is relaxation, and it is distinct from thermal behaviour: relaxation is time- and tension-dependent and largely irreversible within a reasonable timeframe, whereas thermal response reverses as conditions reverse. A ribbon wound tight onto a core under high tension will relax measurably over the first 48-72 hours, and the amount depends on the substrate construction and the finishing route.
Heat setting, coating, calendering, washing and other finishing steps are where a fabric is stretched and then allowed or forced to stabilise. The net length change through finishing — the finishing shrinkage — is a property of the route, not a defect. A route that deliberately over-sets the ribbon for stability will have a different as-shipped length than one that does not. The critical buyer-facing point: the shrinkage the mill measures should be the finished, relaxed shrinkage, not the shrinkage of an intermediate or tensioned state, because that intermediate state is not what ships.
Cut length inherits everything above, then adds measurement method. A cut length measured off a roll under tension differs from one measured after relaxation. Batch-to-batch variation in relaxation means two lots cut to the same nominal length can present different finished lengths if one lot is cut fresher than the other. Where a programme depends on a specific cut length — bows, trims, wrap straps, gift-bag handles — this batching interaction is what produces the intermittent “some lengths are off” pattern that never resolves to a single root cause.
The ladder replaces “the mill controls length” with an explicit model of who decides what, at what reference condition, with what tolerance, and with what release evidence.
The brand and mill jointly fix the reference conditioning condition at which all length measurements are taken — for example 23 °C and 50 percent relative humidity for a minimum conditioning period before measurement. Every length number in the programme, incoming inspection included, is expressed at that reference. Without a named reference condition, “100 cm plus or minus 0.5 cm” is not a complete specification, because it is silently four different specifications depending on where and when it is measured.
The mill confirms to the brand which substrate and finishing route the article runs, because the shrinkage of a given article is a property of that route, not of “ribbon” in the abstract. An article that moves from an unset satin route to a heat-set satin route has a different shrinkage and a different relaxation curve. This tier exists to ensure the shrinkage figure attached to an article is attached to a specific, stable route — and that a route change triggers a re-baseline rather than a silent substitution.
The brand sets the tolerance on the two distinct quantities, separately: finished relaxed shrinkage (the as-shipped length versus the pre-relaxation length, at the reference condition) and relaxation drift over time (how much a finished, relaxed article continues to move over 24, 72 and 168 hours). These are different numbers with different failure consequences and are routinely conflated. A realistic finished relaxed shrinkage band for a woven polyester article is on the order of 1-3 percent depending on route, and a realistic post-cut relaxation drift over the first week is on the order of a fraction of a percent — but it is that drift, accumulated on a large batch, that produces intermittent failure.
The mill owns the measurement method and cadence: conditioning protocol, measurement instrument (a calibrated length gauge or extensometer, not a tape held against a table), sampling points, and the recording of both the finished relaxed figure and the drift figures. The mill reports the measurement, not only the pass/fail, so the brand can see trend rather than a binary. Method authority without method definition is where length disputes are lost, so the method itself must be specified, not assumed.
The brand owns disposition of out-of-tolerance length lots — accept within use tolerance, re-cut, accept-as-compensated, downgrade to a secondary use, or reject — and the mill cannot self-disposition a length excursion without a brand disposition note. Length failures are unusually susceptible to “close enough” self-disposition precisely because the failure is often invisible to the eye; a governed disposition tier is what stops a marginal length lot from quietly entering a finished good.
The workflow places a measurement between every pair of processes that can introduce length change, so drift is attributed to the stage that caused it.
Before production, the article’s substrate and finishing route are fixed and a baseline finished relaxed shrinkage figure is established from a production-representative sample. This baseline is the reference against which every later lot is compared and the figure against which a reorder is judged. A programme with no baseline has nothing to compare to and therefore cannot detect drift, only observe it.
The first production article is conditioned to the reference condition and then measured. This establishes the as-shipped figure for the run and confirms the route is behaving as the baseline predicted. Measuring without conditioning is the single most common methodological error in length control, because it measures the room rather than the ribbon.
After the finishing route and after a defined relaxation period (a minimum conditioning interval post-finish), the article is measured again. The difference between the immediate and the relaxed measurement is the in-house relaxation, and it is the number that predicts how much more the article will move in the field. Capturing this difference is what converts a single length figure into a behaviour model.
Where the programme is pre-cut, the cut length is verified after cutting and after a post-cut relaxation interval, against the compensated target. This is where the batching interaction is caught: a lot cut fresher and measured sooner will present a different length than one cut earlier and measured later, and the cut-length verification makes that visible while it is still correctable.
Length is re-measured at defined intervals (24, 72, 168 hours) on retained samples, and the figures are plotted as a time series rather than logged as isolated pass/fail. A slow relaxation tail is invisible in pass/fail at any single time point but obvious as a decaying trend across the week, and the trend is what tells you whether a lot is still moving.
A reorder is measured against the retained master sample and the original route baseline, not accepted on the basis of the nominal figure. This is what makes a two-year reorder reproducible in length the way it is reproducible in width and colour, and it is the step that catches the case where a route was quietly changed between orders and the article now behaves differently even though the description is the same.
Four gates lock the conditioning requirement into the programme so that no measurement happens outside the defined reference without that fact being visible.
The model is the practical output of the architecture: it converts a measured finished relaxed shrinkage into the nominal cut length that will actually arrive at the intended finished dimension.
If an article exhibits a finished relaxed shrinkage of s (fraction) and a post-cut relaxation of c (fraction), and the intended finished cut length is L, then the nominal length to cut is:
L_nominal = L ÷ ((1 − s) × (1 − c))
Worked example: an article with a 2 percent finished relaxed shrinkage (s = 0.02) and a 0.5 percent post-cut relaxation (c = 0.005), for an intended finished length of 100 cm. The nominal cut is 100 ÷ (0.98 × 0.995) = 100 ÷ 0.9751 ≈ 102.6 cm. Cut at 102.6 cm and the ribbon relaxes to approximately the intended 100 cm in the field, rather than arriving at 98 cm because the shrinkage was never compensated. The model is deliberately simple, because the value is in applying a measured compensation consistently, not in refining the arithmetic — and because a compensation model that is too complex to apply on the floor is one that will not be applied.
Two cautions. First, s and c are article- and route-specific; applying a compensation built for a heat-set satin to an unset organza over-compensates. Second, compensation corrects the systematic component of length change; it does not correct variation, which is controlled by the tolerance and the interlock, not by the model. A programme that compensates the mean but does not govern the spread will still produce the intermittent off-length lot.
Finished relaxed shrinkage band (typical woven polyester): ungoverned 3-6 percent → governed 1-3 percent at the reference condition
Post-cut relaxation drift (168 hours): ungoverned 0.8-2.2 percent → governed 0.2-0.5 percent
Cut-length out-of-tolerance rate: ungoverned 3-6 percent of pre-cut lots → governed 0.4-1.1 percent
Field length complaints per programme-year: typically reduced by 60-80 percent
Reorder length reproducibility (vs. original): governed to within the programme tolerance rather than within a nominal figure
Roll-to-roll length variance (within a lot): reduced to the level the downstream line can absorb, by moving variance from the field into a governed in-house measurement
Because length drift presents as one symptom, the architecture decodes it into five distinguishable causes with a defined test for each.
| Phase | Timing | Content |
|---|---|---|
| Phase 1 — Reference & Baseline | Week 1-2 | Fix the reference conditioning condition for the programme; establish route baselines and finished relaxed shrinkage figures for each active article; identify which articles are pre-cut versus roll-goods |
| Phase 2 — Specification Rewrite | Week 2-4 | Add reference condition, separated finished-shrinkage and relaxation-drift tolerances, measurement method, and (for pre-cut) a compensated nominal cut length to the programme specification |
| Phase 3 — Measurement Rollout | Week 4-6 | Conditioned first-article, post-finish relaxed and post-cut verification stages live; time-series drift checks at 24/72/168 hours; instruments calibrated and labelled conditioning states recorded |
| Phase 4 — Interlock & Release | Week 6-8 | Four gates enforced; relaxation-drift release hold active; disposition authority and notes in place; retention sampling per programme |
| Phase 5 — Steady State & Reorders | Week 8 onward | Reorders measured against retained master sample and original route baseline; route changes trigger re-baseline; trend reviewed at the programme’s recurring quality cadence |
Length is the dimension nobody writes down and everybody discovers. The 258-module architecture treats it as what it is: a measured, conditioned, compensated and drift-governed material response attached to a specific route, not a number printed on a purchase order. Programmes that adopt it do not merely receive ribbon of the right length; they receive the same length at the mill, in the container, in the warehouse and on the shelf — and a pre-cut length that arrives at the dimension it was intended to arrive at, rather than short by a shrinkage nobody compensated for. In an era of pre-cut proliferation and automated downstream feeds, that is the difference between a length that is a specification and a length that is a hope.
Smith Ribbon has supplied woven and printed ribbon, bows, and decorative trims to 1,000+ brands and retailers across 50+ countries since 2004. Our 15,000 m² plant runs 200+ employees with 100,000 meters of daily capacity, certified to OEKO-TEX®, FSC®, BSCI, SEDEX, SMETA and ISO 9001.
Talk to a compliance engineer: +86 13779951780 (WeChat) | xmmsd@126.com | www.smithribbon.com
Smith Ribbon OEM since 2004 — woven ribbon, satin, grosgrain, velvet, organza, RPET, printed bows, decorative trims, and private-label accessories for global brand buyers.