Lecture

2
Statistics in Textile Engineering
By M. Naveed Akhtar, UMT
PH. +92 321 6682395 E-Mail mnabpk@gmail.com
Copyright 2013 M. Naveed Akhtar, UMT 1-1

SAMPLING IN TEXTILES
MA-310 Statistical Methods for Textile Engineers
By

M. NAVEED AKHTAR
Contact: 0306-7122490 E-Mail: naveed.akhtar@umt.edu.pk
Copyright 2013 M. Naveed Akhtar, UMT Email: mnabpk@gmail.com

Reference Books:
1. Physical testing of Textiles by B. P. Saville 2. Principles of Textile Testing by J. E. Booth

POPULATION AND SAMPLE

POPULATION
The whole bulk of the material available for testing is termed the population In Textiles population may be fibre bales, loose fibre mass, laps, sliver, roving, yarn bobbins, yarn cones, fabric or garments

POPULATION AND SAMPLE

SAMPLE A relatively small number of individual members which is selected to represent the whole population is termed sample

OBJECTIVE OF SAMPLING
The aim of sampling is to produce an unbiased sample in which the proportions of, for example, the different fibre lengths in the sample are the same as those in the bulk

UNITS OF SAMPLE
The units for the sample will be the same as the population The units may be numbers or weight measure in textiles such as grams, kilograms etc

IMPORTANT TERMS
Consignment: Quantity of material delivered at the same time Test Lot or Batch: All containers of textile material of one defined type and quality, delivered to one customer according to one dispatch note. It is equivalent to statistical population Laboratory Sample: A sample derived from the test lot by random sampling for being tested in the laboratory

IMPORTANT TERMS
Test Specimen: Specimen actually derived from the Laboratory Sample for individual measurement. Container or Case: A shipping unit identified on the dispatch note, eg, carton, box, bale etc. It may or may not contain packages. Package: Elementary units of material present in each container of the consignment which can be unwound.

FIBRE SAMPLING
Zoning Zoning is a method that is used for selecting samples from raw cotton or wool or other loose fibre where the properties may vary considerably from place to place (ie heterogeneous). At least 40 small samples of fibres (approx 50 gram each) are taken randomly from differently widely spaced places of the whole lot.

FIBRE SAMPLING (zoning)

Each sample is divided into two halves- only one half is retained at random and is again divided into two halves. The procedure is repeated until n/x fibres remain in the sample where n = total number of fibres required in sample and x = number of original samples taken

FIBRE SAMPLING (zoning)

Fibres from all the samples are then united together to get final test sample of correct size containing n fibres

FIBRE SAMPLING (zoning)

FIBRE SAMPLING
Core Sampling A tube with a sharpened tip is forced into the bale of cotton or raw wool and a core of fibres is withdrawn The tubes are 600mm long so as to penetrate halfway into the bale A detachable cutting tip with internal diameter slightly smaller than that of tube is used 14 to 18mm diameter cores used for varying sample size All cores combined to get required sample size

FIBRE SAMPLING
Fibre Sampling from Sliver, Roving and Yarn Problems of length and extent bias faced in such sampling Length Bias: That is longer fibres will have more chance of being selected Extent: It is the distance parallel to the strand axis through which a fibre extends Extent Bias: The chance of a fibre being selected from a strand is proportional to its extent

FIBRE SAMPLING (extent bias)

FIBRE SAMPLING (sliver, roving)

Fibre extent instead of fibre length determines chance of selection Length bias must be avoided is testing fibre length, but also has effect while testing fibre fineness, strength etc. To avoid bias- prepare a numerical sample Prepare a length biased sample so that its bias can be taken care of during calculation

FIBRE SAMPLING
Numerical Sample The percentage by numbers of fibres is each length group should be the same in the sample as it is in the bulk A and B represent two planes. Solid circles show all those fibres whose left ends lay between A and B If all fibres left to A are removed- all those fibres marked with solid circles will be selected Similarly by drawing another plane right to B and repeating the activity- similar samples are prepared

FIBRE SAMPLING (numerical)

FIBRE SAMPLING
Length Biased Sample The percentage of fibres in any length group is proportional to the product of length and the percentage of fibres of that length in the bulk Removal of one such sample changes the composition of remaining bulk- as sample removed contains higher proportion of longer fibres

FIBRE SAMPLING(Length Bias)

If A and B are two planes through the sliver then chance of a fibre crossing these lines is proportional to its length. If fibres crossing this area are selected- longer fibres will have more chance of being selected Fibres are gripped along narrow line of contact, loose fibres removed by combing both sides of contact. Such sample is also known as tuft sample Such samples used for length measurement by Fibrograph

FIBRE SAMPLING (length bias)

FIBRE SAMPLING (tuft sample)

The histograms show the mean fibre length from both the numerical sample and tuft sample

FIBRE SAMPLING

FIBRE SAMPLING
Random Draw Method Used for sampling sliver and top Sliver is parted by hand and placed on two velvet boards with the parted end near the front of the first board. The opposite end of the sliver is weighed down with a glass plate to stop it moving Discard a 2mm fringe of fibres from the parted end using a wide grip

FIBRE SAMPLING (Random Draw)

This procedure is repeated until a distance equal to that of the longest fibre in the sliver has been removed All succeeding draws will be used as sample as these will be representing all fibre lengths They represent a numerical sample where all the fibres with ends between two lines are taken as the sample All the fibres of the sample must be measured

FIBRE SAMPLING (Random Draw)

FIBRE SAMPLING
Cut Square Method Used for sampling of yarn A length of the yarn being tested is cut off and the end untwisted by hand The end is laid on a small velvet board and covered with a glass plate The untwisted end of the yarn is then cut about 5mm from the edge of the plate All the fibres that project in front of the glass plate are removed one by one with a pair of forceps and discarded

FIBRE SAMPLING (Cut Square)

All the cut fibres are removed, leaving only fibres with their natural length The glass plate is then moved back a few millimetres, exposing more fibre ends These are then removed one by one and measured When these have all been measured the plate is moved back again until a total of 50 fibres have been measured In each case once the plate has been moved all projecting fibre ends must be removed and measured

FIBRE SAMPLING (Cut Square)

The whole process is then repeated on fresh lengths of yarn chosen at random from the bulk, until sufficient fibres have been measured

FIBRE SAMPLING (Cut Square)

YARN SAMPLING
Ten packages are selected at random from the consignment If the consignment contains more than five cases, five cases are selected at random from it Test sample will consist of two packages selected at random from each case The appropriate number of tests are then carried out on each package

FABRIC SAMPLING
Fabric samples are always taken from the warp and weft separately The warp direction should be marked on each sample before it is cut out No two specimens should contain the same set of warp or weft threads Samples should not be taken from within 50mm of the selvedge

FABRIC SAMPLING

MEASUREMENT
A quantitative comparison between a predefined standard and the object being measured The actual process of measurement is always subject to errors Error is the difference between the measured value and the 'true' value Precision is the quality that characterises the ability of a measuring instrument to give the same value of the quantity measured

MEASUREMENT (Accuracy)
Precision of any measurement is obtained by making a number of identical measurements and estimating the dispersion of the results about the mean by calculating Standard Deviation or Coefficient of Variation Accuracy is nearness to the 'true value of the quantity being measured It is obtained by calibration of the measuring system against the appropriate standards

MEASUREMENT (Sensitivity)
Sensitivity is the least change in the measured quantity that will cause an observable change in the instrument reading It can be increased by amplifying the output or by using a magnifying lens to read the scale Errors will also amplify if there is no increase in accuracy of the calibration and a reduction in sources of variation

STATISTICAL TERMS

STATISTICAL TERMS

STATISTICAL TERMS

STATISTICAL TERMS
Coefficient of variation (CV): It is often used as a measure of dispersion It is the standard deviation expressed as a percentage of the mean

STATISTICAL TERMS
Standard error of the mean: It is a measure of the reliability of the mean value obtained from a sample of a particular size. It is the standard deviation of the means that would be obtained if repeated samples of the given size were measured

STATISTICAL TERMS

The standard error is used to place confidence limits on the mean that has been measured

STATISTICAL TERMS
There is 95% probability that the population mean lies within (t x standard error) of the measured mean value. For large samples or parent universe the value of t is 1.96 For smaller samples (less than 30 size) the value of t is greater and can be calculated from t-tables