First Edition Dated September, 2000 34
Cpk or comparable measures of process capability should be recalculated
whenever there is a significant process change. Recomputing Cpk should
accompany the recalculation of control limits for the statistical control charts.
Recalculating Cpk when there has been no substantial change in the process
(e.g., a large change in variation or a process shift), and no change in the
engineering specifications, will yield limited benefit and may drive unneeded
actions and false security.
Section 4.3.3 If the key characteristic is not capable, the
supplier shall identify and control sources of variation in the
processes that are correlated with the key characteristic and
take corrective action. D1-9000-1 contains information that is
useful in diagnosing sources of variation as well as providing
guidance in techniques to reduce measurement error and
process variation.
In some cases, it may be impossible or prohibitively
expensive to meet the requirements of 4.3.1 through 4.3.3. As
such, the customer must approve any exceptions to
paragraphs 4.3.1 through 4.3.3.
In addition to reduction of special causes of variation, the identification and
removal or control of sources of variation inherent in the process (common
cause variation) can help lead to a higher quality process.
Measurement Systems Analysis
One of the most common sources of apparent process variation lies in the
use of poor measurement systems, which include the devices used, fixtures,
tooling or shop aids, and the training of the people doing the measurement.
Poor measurement systems reduce the ability to demonstrate control or
capability and make investigation into the sources of variation difficult.
Therefore, investigations into measurement system variation are usually a
good place to start. The repeatability and reproducibility of a gage should, in
fact, be known prior to taking measurements on key features of products or
processes.
Identifying Sources of Variation
Sources of common cause variation affecting the key characteristic can be
found by investigating all of the processes that are relevant to the production
of a key characteristic. Variation within these “relevant processes” is
influenced by one or more variables and their settings. These variables and
