TWI Knowledge Summary

Manufacturing - reducing variability

by John Weston

Variability occurs throughout an organisation and adversely effects business costs. Specifically, within manufacturing processes, variability has a significant impact on the costs associated with poor quality performance. Unfortunately, most of these costs are hidden and usually overlooked - as illustrated below.

Variability in processes is revealed in all the sources of waste. Consequently, if variability can be reduced then so can waste. For example, variability in material outside specification, impacts the seven key forms of waste:

Key Waste	Impact of variability in material
1) Transport	To and from vendor, quarantine, rework, etc
2) Waiting	Time for replacement material
3) Correction	Scrap and rework costs when the material enters the manufacturing chain Cost of managing the scrap and rework system Additional Inspection costs Cost of redesign to accommodate the different specifications
4) Process	Extra operations required to achieve quality from different specification materials
5) Inventory	Safety stock to cover future 'out of specification' deliveries
6) Over-production	Allow slack in system for poor specification material deliveries
7) Motion	Operator, inspector, engineer activities needed to deal with the above items

Variability arises from common and special causes.

- Common causes are inherent in a process and affect all products of this process.
  This variability influences capability but not predictability.

- Special causes are transient in nature and affect only some products.
  This variability influences both predictability and capability.

The variation in a process can be understood by continuously measuring performance. If a common cause variation is present, the pattern of variation will be stable (i.e. in statistical control). Conversely, if special cause variation is present, then the process is unpredictable (i.e. out of statistical control). Statistical process control (SPC) is used to measure and define process trends and suggest solutions. It is, therefore, a collection of tools for reducing waste. Examples include:

• Data collection sheets, for problem identification
• Charts and histograms, for problem identification and analysis
• Cause/effect diagrams, histograms and control charts, for problem analysis

Important SPC measures are the mean and the standard deviation ( ). The figure below plots process outputs having different mean and standard deviations values. Highly capable processes (i.e. the blue curve, which has a small deviation) are desirable. With such a process it is possible for the mean position to move in either direction (e.g. + and - ) and the process will still remain within the required tolerance limits. However, any shift in the mean position of the red curve will result in unacceptable numbers of defective components being produced. When the standard deviation is large (i.e. the violet curve) there will always be defective products.

The green curve is typical of a process that is poorly set. Under these circumstances the mean can be 're-centred' within the tolerance limits by adjusting the process parameters.

These curves illustrate the way in which statistics can be used to guide and reduce variability in manufacturing.

Also, you can use the Weldasearch literature database to supplement what you find in JoinIT.

Copyright © 2001, TWI Ltd