There are grey boundaries between the techniques described below and so a training course can easily overlap two or more of the methods. Basic introductions typically last two days and are sufficient for some clients. Other clients need greater depth in certain directions : this is where our flexibility and vast range of materials comes into action. If you want training in everything in full detail it will take well over a month !
Design of Experiments
We concentrate on Taguchi's methods for design of experiments since they minimise the use of statistics and concentrate instead on engineering understanding. They are key to the design of low cost, quality products and processes since they give a structured and efficient way to perform development trials.
Taguchi methods A two day session gives sufficient information for participants to start performing effective development or problem solving trials. Further training is often organised as a workshop to concentrate on actual examples. Short overviews are also available and can be adapted to the specific needs of different audiences.
Robustness is a key part of the Taguchi approach which minimises variation at low cost. A one day training course follows the two day introduction.
Tolerance design. A one day course covers the concepts and practicalities of statistical tolerancing. It shows how this method in combination with Taguchi allows performance to be improved by tightening control of a few key variables. It is relevant for both products and processes.
Failure Modes and Effects Analysis (FMEA) is a standard tool to evaluate potential problems. During a one- or two day course the participants begin to apply the method to their own products or processes. Fault Tree Analysis (FTA) and Process Decision Programme Charts (PDPC) are complementary techniques which can be more relevant.
Qualification tests, often called validation or verification trials, are used to demonstrate that new designs will work under real conditions. Similar needs are met by acceptance trials on new processes where they give high confidence that a process is stable and repeatable. Novel methods, such as Taguchi and supersaturated arrays, reduce the number of trials and reveal design errors which would not be found by traditional means.
Reliability is often regarded as either a black art or the domain of mathematical specialists. However, simple analysis methods and trials on products, supported by an understanding of customer perceptions of reliability, can be used to obtain cost effective improvements.
An introductory course explains the concepts and jargon. The direction of further training depends on the challenges which are being met and the knowledge which the participants have of design of experiments. It can show how to collect and interpret field failure data and link in to problem solving. It can outline the approaches for improving reliability and go into considerable depth where relevant. Methods for demonstrating reliability and the traditional analysis tools such as Weibull are also covered.
Many clients choose a workshop approach which combines some training with support for a specific problem.
The basic concepts for reliability are just as applicable to the machines used to make products as to the products themselves. However, there are significant links to maintenance, team work and productivity which are brought out in Total Productive Maintenance, TPM. This subject can be as large or small as you want it and customised training is essential.
Statistical process control
Statistical process control (SPC) techniques are used for problem solving and improvement, to maintain control of critical parameters, and to prevent over-reaction to minor variation.
A two day course for managers outlines the methods, shows where they can be applied and covers the problems which often arise during introduction. Lack of support from above is a frequent reason for poor implementation.
A two day course for operators shows how to maintain and interpret the charts. The course must be customised to allow for the particular types of chart which are adopted, and can be adapted for software based charting. It ties in strongly with problem solving.
Additional modules expand on the specific needs of different industries, for example :
• Short runs or small batches
• Machines which make many parts in parallel, such as injection moulding with a multi-cavity tool or
• Processes which must be controlled very tightly
• Processes which are highly capable but appear to be out of control
• Processes which do not have normal distributions
• Applications in administrative or service functions.
The aim in these modules is to help the participants to find practical solutions and not become bogged down in statistics.