Attributes - Statistical Process Control

On the other hand, some processes have outputs with attributes to be controlled where counting the number of defects is more appropriate than measuring them. The outputs can be segregated into two categories (eg pass/fail or non-defective/defective). For example, in bottle production, the design requirement may stipulate a maximum allowable number of air bubbles in the glass. Bottles which have no more than this number of bubbles would be counted as non-defective items; bottles with more than this number would be counted as defective items. Returning to the example of controlling the diameters of discs, it might be decided, perhaps for reasons of economy, that it is more appropriate to gauge the diameters rather than measure them. This figure shows how this can be achieved by use of a simple go/not-go gap gauge. Conforming discs, as shown in this figure, will be small enough to enter the 'go' gap, but too large to enter the 'not go' gap. As a result the output will be segregated into countable 'pass' and 'reject' groups.

Much of what follows concerns the construction and use of statistical process control charts for both categories of output. The main treatment focuses on two complementary charts, sample mean and sample range charts, for the control of variables. This is followed by accounts of three charts for the control of attributes: proportion defective, number defective and number of defects.