CONTROL CHARTS
Basic Tools for Analyzing
Of all the tools for analyzing data, the control chart is the most useful. No
other tool captures the voice of your process better. Control charts are used
to determine whether your process is operating in statistical control. Until it
Data
is, any improvement efforts are, at best, mere process tampering. Basically,
Tools for Analyzing
a control chart is a run chart (described earlier) that includes statistically
generated upper and lower control limits.
The purpose of a control chart is to detect any unwanted changes in your
Data
process. These changes will be signalled byabnormal points on the graph.
Extensive research by Dr. Shewhart indicated that by establishing upper
and lower limits at three times the standard deviation of the process (plus
and minus, respectively), 99.73% of the common cause variation would fall
within these limits.
A process is said , therefore, to be in statistical control when the process
measurements vary randomly within the control limits; that is, the variation
present in the process is consistent and predictable over time.
The upper and lower control limits are not the same as tolerance or
specification limits.
Control limits are a function of the way your process actually performs over
time. Specification, or tolerance, limits are a function of what your process
may have been designed to do and may not necessarily have any direct
relationship to the actual performance of the process.
BENEFITS OF CONTROL CHARTS
Help you recognize and understand variability and how to control it
Identify special causes of variation and changes in performance
Keep you from fixing a process that is varying randomly within control
limits; that is, no special causes are present. If you want to improve it,
you have to objectively identify and eliminate the root causes of the
process variation
Assist in the diagnosis of process problems
Determine if process improvement effects are having the desired affects
60 Coast Guard Process Improvement Guide
Basic Tools for Analyzing
How to use it:
Use the figure on the next page to determine the type of control chart you
Data
need.
Control charts use two types of data: variables data and attributes data. In
general, if you want to use variables data, you have to take measurements in
units such as length, temperature, etc. On the other hand, attributes data
requires a good/bad or go/no-go decision and counting (for example,
number of defects, percent late, etc.).
Construct the control charts
charts.. Once you have determined the type of data to
collect, follow the appropriate chart construction techniques described
elsewhere in this section.
Upper Control Limit (UCL)
X Average
X
Lower Control Limit (LCL)
X is the Type of Control
TYPICAL Chart. OL CHART
CONTR
X is a Grand Average.
Coast Guard Process Improvement Guide 61
Basic Tools for Analyzing
HOW TO DETERMINE WHICH CONTROL CHART TO US
USE
STEP ONE
ONE: Is a Control Chart the appropriate tool to use?
Data
STEP TWO: Attribute or Measurement)?
Determine type of data (Attribute
Tools for Analyzing
STEP THREE: If Measurement use:
X-R Chart (if subgroups of five or more)
Data
X Chart (if subgroups of five or less)
If Attributes (Good/Bad, Discrete Data), use one of the following charts:
If attributes are DEFECTS: Ask, Is a constant sample size used?
If YES: np - Chart If NO: p - Chart
If attributes are DEFECTIVES: Ask, Is constant sample size used?
If YES: c - Chart If NO: u - Chart
DEFECT
DEFECT: A failure to meet one part of an acceptance criteria
DEFECTIVE: A unit that fails to meet acceptance criteria due to
one or more defects
Identify and eliminate any special or assignable causes of variation. To
determine whether these causes exist, look for one or more of the seven
signals shown on the next page. The probability that any of these seven
events will occur at random is very small. Thats why theyre a signal that
something has changed in your process.
Reduce overall variability. After you have eliminated all special causes, try
to reduce the remaining variability in your process. This usually requires
fundamental changes to the process, and these changes require manage-
ment assistance. To achieve continuous process improvement, its essential to
reduce variability.
62 Coast Guard Process Improvement Guide
Basic Tools for Analyzing
Indications That Special Causes of Variation Are Present
Data
One or more Seven or more consec- Six points in a row
points outside the utive points on one side steadily increasing or
control limits. of the centerline. decreasing.
Fourteen points alter- Two out of three Fifteen points in a row
nating up and down. consecutive points in within the center third
the outer third of the of the control region.
control region.
Eight points on both sides
of the centerline with none
in the center third of the
control region.
Common cause variations within a process are, by definition, random.
Therefore, any non-randomness indicates the influence of special causes.
Points to Remember:
Upper and lower control limits are not specification limits. They have a
mathematical relationship to the process outputs. Specification limits are
based on product or customer requirements.
Be sure you have selected the correct control chart for the type of data you
are collecting.
Having a data point fall outside the control limits is only one of many
different signals that indicate a process is out of control. If all the data
points are within the control limits, be sure to check the other signals that
indicate a special cause of variation.
Coast Guard Process Improvement Guide 63