Statistical Process Control (SPC) is a method that
uses statistical tools and techniques to monitor, control, and improve the
product development processes. There are several basic premises behind this
method, including the fact that there will always be some variation between
two products, even when produced using the same processes. This variation will
occur in a definite pattern, which can be used to identify process
abnormalities or impending failures. The SPC method provides a means for
measuring and analyzing variations in process capabilities, which, in turn,
can identify opportunities for process correction. This methodology is often
included as part of a larger Statistical Quality Control (SQC) initiative
(discussed in Appendix
objective of which is to improve product quality by improving the quality of the
related manufacturing processes. Successful implementation of SPC results in
increased production throughput, decreased manufacturing defects, and improved
Process and product variation is expected and can be
attributed to variations in one or more of the following: materials, equipment,
methods, environment, and personnel. The causes of variation can be categorized
into two types: (1) chance or system causes which are those built-in process
characteristics that are beyond human control and cannot be corrected and (2)
assignable or special causes which are those process anomalies that can be
detected and corrected. In order to maintain or improve product quality,
assignable causes for variation should be minimized or, ideally, eliminated from
the product manufacturing process.
The first step in the SPC method is to capture process
variation under normal, stable conditions in which all controllable sources of
variation have been eliminated. A histogram, or frequency distribution, is
used to measure and analyze process variation. When used for analyzing process
capability, a frequency distribution is a count of the number of times a
particular measurement occurs as a result of the process. A histogram (Figure
F.8) is bar graph depiction of a frequency distribution. When all assignable
causes for variation have been eliminated, the measurements will tend to be
somewhat evenly distributed around an average value. Ideally, this average value
will equal the measurement required in the applicable specifications. Standard
deviations, also referred to as sigma, describe how the measurements fit around
the average. Any variation induced by any of the factors noted above will
distort the normal distribution of the measurements.
Because frequency distributions and histograms provide
a snapshot depiction of a process, they should not be used to analyze a
continuous manufacturing process. Instead, a control chart should be used.
There are two types of control charts - variable and attribute. An average and
range chart (X-bar, R chart), shown in Figure
F.9, is a variable control chart that is used
to determine how the average output of a process compares to specified
requirements. An attribute chart is best suited when conducting "pass/fail"
types of inspections. When used properly, both types of control charts provide
an immediate visual indication of when a process is operating outside of
previously specified limits and in need of corrective action.
Once all assignable causes have been eliminated from a
process, the capability of the process is determined and expressed numerically
as the capability index or the capability ratio. Both numbers are based upon
the process tolerance, which is equal to six standard deviation (Six Sigma) of
the process distribution. Many companies are adopting the Six Sigma approach,
which converts the total defects per unit to a standard deviation as a process
and quality improvement technique. (Six Sigma is discussed in Appendix F.1.20
Once a process problem has been identified using the
tools above, other techniques, such as cause and effect diagrams and Pareto
analysis, are used to help decide what corrective actions should be taken. Cause
and effect diagrams are often used by teams as a brainstorming tool to identify
the potential sources of process variation. Pareto analyses can be used to
establish priorities for solving problems.
To achieve maximum benefits, SPC methods must be
integrated into a company's normal way of doing business. When successfully
implemented, SPC will aid in optimizing production processes and capabilities,
thereby improving the overall quality of products.
Burr, I. W. (1976). Statistical Quality Control Methods. New York: Marcel Dekker, Inc.
Doty, L. A. (1996). Statistical Process Control. Industrial Press.
Evans, J. R. (1991). Statistical Process Control for Quality Improvement: A Training Guide to Learning SPC.
Grant, E. L., & Leavenworth, R. S. (1988). Statistical Quality Control (6th ed.). New York: McGraw-Hill Book Company.
Montgomery, D. C. (1985). Introduction to Statistical Quality Control. New York: John Wiley & Sons.
Oakland, J. S. (1996). Statistical Process Control: A Really Practical Guide. Butterworth-Heinemann.
Rhyder, R. F. (1997). Manufacturing Process Design and Optimization. New York.
Smith, G. M. (1997). Statistical Process Control and Quality Improvement. Prentice Hall.
Wheeler, D. J., & Chambers, D. S. (1992). Understanding Statistical Process Control (2nd ed.). Knoxville, TN: SPC Press.