Two details change the count. First, the order area usually includes a waste allowance, so the purchased area is higher than the installed area; the panel count should be based on what you buy, not what ends up on the wall. Second, a wall with many openings or a strong pattern match may need more material, which raises the panel count again. The 417 figure is the baseline for a simple 300 m\u00b2 order with no waste allowance.<\/p>\n
![\"Flexible](\"https:\/\/jmsdecor.com\/wp-content\/uploads\/2025\/02\/workshop3.jpg\")
<\/p>\n
Does \u201cbelow 1%\u201d mean area or panel count?<\/h2>\n
The two readings give similar numbers but different conversations. One percent of 300 m\u00b2 is 3 m\u00b2. One percent of 417 panels is about 4.17 panels. Four panels at 0.72 m\u00b2 each is 2.88 m\u00b2, so a four-panel defect count is roughly consistent with the 3 m\u00b2 area reading.<\/p>\n
The readings diverge when defects are concentrated. One panel with a large area of surface damage can be 0.7 m\u00b2 of defective material but only one panel; four panels with small edge chips can be four defects but a small fraction of a square metre. If the contract does not say whether the rate is measured per panel or per square metre, both sides can be right and the argument has no resolution.<\/p>\n
There is a second ambiguity: \u201cbelow 1%\u201d is an upper bound, not a prediction. A batch can be at 0%, at 0.4%, or at 0.9% and still satisfy the sentence. The buyer\u2019s protection is not the number; it is the inspection and the remedy written into the order.<\/p>\n
What counts as a defect?<\/h2>\n
A defect rate without a definition is a marketing number. For flexible stone, separate the categories before you discuss percentages.<\/p>\n
- \n
- Functional defects.<\/strong> Delamination between the stone face and the backing, a crack through the sheet, backing separation at a cut edge, thickness or weight outside the declared band, or a panel that fails the agreed bend test. These affect performance and should not be installed.<\/li>\n
- Cosmetic defects.<\/strong> Edge chips, surface marks, colour outside the approved sample range, or pattern misregistration. Some are acceptable if they fall on a cut edge or in a concealed area; others are not.<\/li>\n
- Packaging defects.<\/strong> Crushed corners, wet packs, or missing batch labels. These are acceptance items in their own right, not part of the panel defect rate.<\/li>\n<\/ul>\n
The acceptance criteria should be written per property: thickness band, weight band, minimum bend radius, mesh coverage, colour against the retained sample, edge condition, and packaging. A panel that meets every criterion is not a defect even if it looks different from its neighbour; natural variation is expected within the approved range.<\/p>\n
Does the defect rate apply to my shipment or to the whole production run?<\/h2>\n
This is the question most buyers never ask, and it changes what the number means. A factory\u2019s \u201cbelow 1%\u201d is usually measured across a production period \u2014 a month, a quarter, or a rolling average \u2014 not on the specific lot that goes into your container. A run of 10,000 m\u00b2 at 0.8% can still produce a container with a cluster of defects, because defects tend to concentrate around a shift, a pigment lot, or a machine adjustment rather than spreading evenly.<\/p>\n
Ask whether the rate is per batch or a company average, and ask for the defect count recorded for your batch. A supplier that tracks real numbers can answer; a supplier that only has a brochure figure will not. If the number is an average, treat it as context, not as an acceptance criterion \u2014 your protection is the sampling plan and the remedy, not the average.<\/p>\n
The COA describes the batch, but it does not usually report the defect distribution. That is another reason to inspect the actual panels rather than rely on a document that was written before the container was packed.<\/p>\n
Un \u00e9chantillon peut-il prouver un taux de d\u00e9fauts inf\u00e9rieur \u00e0 11 TP3T ?<\/h2>\n
Habituellement non, et l'arithm\u00e9tique en est la raison. Si vous inspectez 50 panneaux et ne trouvez aucun d\u00e9faut, une r\u00e8gle statistique standard (souvent appel\u00e9e r\u00e8gle des trois) place la borne sup\u00e9rieure de confiance \u00e0 951 TP3T \u00e0 environ 61 TP3T. Ne trouver aucun d\u00e9faut dans 50 unit\u00e9s ne prouve pas que le taux est inf\u00e9rieur \u00e0 11 TP3T ; cela montre seulement qu'il est peu probable que le taux d\u00e9passe environ 61 TP3T.<\/p>\n
Pour bornrer le taux en dessous de 11 TP3T avec z\u00e9ro d\u00e9faut au niveau de confiance de 951 TP3T, il faut inspecter environ 300 panneaux, soit la majeure partie d'une commande de 417 panneaux. L'inspection de 100 panneaux donne une borne sup\u00e9rieure d'environ 31 TP3T. Un \u00e9chantillon de 80 panneaux donne environ 3,751 TP3T. Plus l'\u00e9chantillon est petit, plus l'all\u00e9gation est faible.<\/p>\n
C'est pourquoi un plan d'\u00e9chantillonnage compte plus que le chiffre annonc\u00e9. Le plan d\u00e9finit la taille de l'\u00e9chantillon, le nombre d'acceptation et les risques accept\u00e9s par les deux parties. Le R\u00e9f\u00e9rence NIST\/SEMATECH sur l'\u00e9chantillonnage d'acceptation<\/a> explique la logique ; les chiffres adapt\u00e9s \u00e0 votre commande doivent \u00eatre convenus avec le fournisseur, et non copi\u00e9s depuis un tableau g\u00e9n\u00e9rique.<\/p>\n
Quel plan d'\u00e9chantillonnage dois-je convenir avant la production ?<\/h2>\n
Un plan praticable r\u00e9pond \u00e0 cinq questions : qu'est-ce que le lot, combien de panneaux sont \u00e9chantillonn\u00e9s, combien de d\u00e9fauts sont autoris\u00e9s, que se passe-t-il si la limite est d\u00e9pass\u00e9e, et qui inspecte. Le tableau ci-dessous montre la forme du plan ; les valeurs sont un exemple \u00e0 discuter, et non une norme recommand\u00e9e.<\/p>\n
- Cosmetic defects.<\/strong> Edge chips, surface marks, colour outside the approved sample range, or pattern misregistration. Some are acceptable if they fall on a cut edge or in a concealed area; others are not.<\/li>\n
![\"Two](\"https:\/\/jmsdecor.com\/wp-content\/uploads\/2026\/09\/check-stone-bendable-.png\")
<\/p>\n