Process capability
Once a process is in a state of statistical control, it is important to evaluate its process capability, which
is the ability of the process to produce quality characteristics that conforms to design specifications.
Design specifications
Upper specification limit (USL). Value above which the quality characteristic will be totally inappropriate
(e.g., the piece will not function properly and must be scrapped or reworked).
Target value (nominal value). Ideal value of the quality characteristic
Lower specification limit (LSL). Value below which the quality characteristic will be totally inappropriate
(e.g., the piece will not function properly and must scrapped or reworked).
To define process capability we consider variables data, and we assume that the quality characteristic of
interest follows a normal distribution.
Assessing process capability
The process capability is assessed by means of two indexes C pk and C p .
The index C pk describes the capability of the process as it is, while C p describes the potential capability
of the process
The following are the steps to find C pk or C p :
1. Estimate the process mean . If a X control chart is used, then = X . If a chart for individual
values is used, then = X .
2. Estimate the process standard deviations by using either
R
s
or
.
d2
c4
Note that (the estimate of the process standard deviation) is not the same quantity that is
used to compute the control limits of the variables chart. For example, for the X chart, the
=
control limits are X 3 X , where X is the standard deviation of the sample mean.
3. Standardize LSL and USL by finding the corresponding Z scores
Z LSL =
LSL
ZUSL =
USL
and
4. Determine the capability indexes
C pk =
z min
, where z min = {Z LSL , ZUSL }
3
A C pk = 1 means that the minimum distance between one of the specification limits and the
estimated process mean( ) is equal to 3 .
A C pk = 1.33 means that the minimum distance between one of the specification limits and the
estimated process mean( ) is equal to 4 . This is the minimum acceptable value considered
in the United States and Japan (since the fraction of units to rework or scrap will be very low).
The potential capability index C p represents the ideal case where the estimated process mean
is right in the middle between the upper and the lower control limit.
Cp =
USL LSL USL LSL
=
UCL LCL
6
If C p < 1 , the control limits are wider than the specification limits.
If C p = 1 , the control limits are the same than the specification limits.
If C p > 1 , the control limits are narrower than the specification limits.
If C p = 2 , We reach the 6-sigma quality!!!!
The process capability C p measures the potential capability of the process, that is the greatest
capability that the process can achieve without reducing the process standard deviation
What has to happen in the process so that C p = C pk ?
What can you do to increase process capability?
How would you find C p if you only have one valid control limit?
Example 1
The design specifications for a piston rod used in an automatic transmission call for the rod length to be
between 71.4 and 72.8 mm. The process is monitored with an X chart and an S chart, using samples of
=
size n = 5 . These show the process to be in control. The values of X and s are X = 71.8 mm and
s = 0.20 mm.
a) Compute the value for C pk . Is the process capability acceptable?
b) Assume that it is possible to adjust the process mean to any desired value. To what value should
it be set to maximize the value of C pk without reducing the standard deviation? What will the
value of C pk be?
Recommended exercises.
1. The thickness specification for aluminum sheets 0.246-0.254 mm. Data from an X chart , based
=
in samples of size 6, that shows that the process is in control, yield values of X = 0.248 mm and
s = 0.002 mm.
a. Compute the value C pk for this process.
b. Is the process capability acceptable? Explain.
c. To what value should the process mean be set to maximize the value of C pk without reducing
the standard deviation? What will the value of C pk be?
d. What will the process capability then be?
2. The specification for the diameters of ball bearings is 15.40-15.60 mm. Data from an X chart ,
=
based in samples of size 8, that shows that the process is in control, yield values of X = 15.52
mm and R = 0.05 mm.
a. Compute the value for C pk for this process.
b. Is the process capability acceptable? Explain.
c. To what value should the process mean be set to maximize the value of C pk without reducing
the standard deviation? What will the value of C pk be?
d. What will the process capability then be?
Appendix table. Reproduced from: Scheaffer, Mulekar & McClave Probability and statistics for
engineers
