P9 - Sampling Procedure for Optical Measurements in Wire Cloth Production
- Event
- SENSOR+TEST Conferences 2011
2011-06-07 - 2011-06-09
Nürnberg - Band
- Proceedings OPTO 2011
- Chapter
- OP - Poster Session
- Author(s)
- W. Hinrichs - Materialprüfanstalt (MPA) für das Bauwesen, Braunschweig (Germany)
- Pages
- 151 - 154
- DOI
- 10.5162/opto11/op9
- ISBN
- 978-3-9810993-9-3
- Price
- free
Abstract
Measurements on wire cloth products often focus on mesh sizes both in the production and in conformity assessment of the final product. Often used are transmitted bright field microscopes with collimated light. The images allow to measure the mesh size, i.e. the average distance of the wires in the products. The data processing is usually done on the basis of industry standard statistical software.
The use of such data sometimes creates problems with conformity statements which are known to be at least partly due to the calibration of the instrument. Recent research has revealed that also the sampling procedure may have a significant influence on decisions on the conformity of products.
The sampling of wire cloth consists of two steps: 1) Some small areas for measurements are chosen on a more or less arbitrary basis. 2) With aperture sizes ranging between 20 μm and 1 mm, each sample may contain between 10,000 and 25,000,000 meshes, to be measured in both warp and weft (shute) directions.
Extremely small fractions of measured items are very common in industrial production and it is easy to handle them if normal distributions can reasonably be assumed. However, the dispersion of the apertures in wire cloth are beta-distributed in the (mostly dominating) warp direction. It is therefore necessary to reconsider the measurement procedure. This has to be done with a focus on the sampling as for the definition of the distribution function a proper estimate of both a narrowest and a widest opening is indispensible.
A calculation of the producer’s and the user’s risks illustrates that the risks are usually lower when they are based on a beta-distribution. But to make use of the new distribution the producer has to learn more about his product from a measurement point of view. Data for robust estimates may be taken from different stages of the production. It is quite easy to install and run such an optical inspection system in the wire weaving industry but reliable results are only obtained when information from measurement, product and production are combined.