If a process delivers output of variable or poor quality it is common for engineers to reach some estimate of which process parameter is the likely cause of the problem and to adjust that parameter. The adjustment could be to material quality or to process condition. This approach sometimes provides a totally satisfactory result, but often the improvement is unsatisfactory or further problems appear.
It is unlikely that the first few changes made in an attempt to improve a process actually target the correct variables. The variables are changed individually so that the way in which they interact is not identified, and the optimum combination is never achieved.
A solution, combining practical process experience, statistics and common sense is available. The technique often referred to as Taguchi experimentation was used initially in Japan. Through using an orthogonal array the effect of each variable can be measured in combination with other variables.
The use of statistics on measurements for adhesives has helped manufacturing companies such as Sidlaw - Colodense to improve their production quality. Sidlaw - Colodense manufacture metallised film laminates which are used in the packaging of potato crisps, biscuits, ground coffee etc.
Taguchi analysis has helped Sidlaw - Colodense to improve productivity, gain new business and allow it to meet its customer demands and specification. The packaging materials consist of a reverse printed clear polypropylene/polyester film which would carry the product description etc. and the metallised polypropylene/polyester film. The metallised film is laminated to the reverse printed layer such that the print is sandwiched between the two layers. These film products are common in the snack food industry .
The objective was to focus on the problem of variable interply bond strengths on metallised and clear polypropylene laminates and produce a minimum lamination bond strength of 2N / 25 mm (in peel)
Reverse printed clear polypropylene/polyester film carrying the product description etc. and Metallised polypropylene/polyester film
Mixed adhesive. Not specified
Surface corona treatment (untreated or boost treatment)
The bonding was carried out at the factory location.
Adhesive bonding of reverse printed clear polypropylene/polyester film which carries the product description etc. to a metallised polypropylene/polyester film. The metallised film is laminated to the reverse printed layer such that the print is sandwiched between the two layers.
Nip 40 to 50 ºC, 2 zone drying, gravure roller etch, chill. Web speed varied in study
2 zone drying 60 – 90 ºC or 70 – 110 ºC. Chill roll after gravure roller etch
An initial list of 28 variables was established. These were ranked and the highest ranking 8 variables were selected for an initial experiment as follows:
|
Factor |
Experiment |
Level 1 |
Level 2 |
|
Surface Corona Treatment |
A |
Untreated |
Boost Treatment |
|
Adhesive Mixing Ratio |
10:1 |
10:2 | |
|
Adhesive Type |
C |
Current |
Alternative |
|
Nip Temperature |
D |
40oC |
50oC |
|
Drying Temperature (2 Zones) |
E |
60 - 90oC |
70 - 110 ºC |
|
Gravure Roller Etch |
F |
130 / 25mm |
140 / 25 mm |
|
Chill Roll Temperature |
G |
Chilled |
Ambient |
|
Web Speed |
H |
130 m / min |
180 m / min |
Variables B, C and F were found to be significant to the process, with variable C being of dominant significance.
A confirmation trial was run with the significant variables set at “best” conditions.
|
Factor |
Setting |
|
Adhesive type |
New (alternative) |
|
Mix Ratio |
10 : 1 |
|
Gravure Roller Etch |
140 / 25mm |
All other factors were set at the most convenient levels.
The orthogonal array measures the effect of each variable in combination with other variables. Properly designed experiments assessing each variable at typically two levels, enable the array to separate the effect of each variable within a matrix of manageable size. Thus it offers the opportunity to identify and rank the process variables which are used to define the process conditions, and to identify the individual variables and combinations which can lead to optimum process performance. Formulas are used to covert the process variables into a final solution of their effects.
The trial response fell below specification, but the mode of failure had changed from adhesive (failure from the substrate surface) to adherend failure (the metal stripping from the polypropylene). Thus the bond had improved to exceed the substrate strength and this was reflected in an overall improvement in bond strength. A second trial with a superior substrate satisfied fully the project objectives with the mean bond strength exceeding twice the minimum specification.
The Taguchi method also encouraged co-operation between the laminator operators and the production and technical management teams at Sidlaw and allowed other areas of manufacturing to be improved.
Most importantly for Sidlaw - Colodense there now exists a process by which a major improvement in the interply strengths on polypropylene laminates can be achieved.
DTI MTS programme adhesives – Dissemination Guides ADH5CS4.DOC
NPL / Sidlaw - Colodense / ESR Technology Limited
Analysis was conducted by Sidlaw - Colodense
