Quality Control

UDPS3143 C3 - 1
Chapter 3 : Control Charts for Attributes

Each item inspected is classified as conforming or nonconforming to specifications on that
quality characteristic. Sometimes, the term defective or nondefective is used. Quality
characteristics of this type that cannot be represented numerically are called attributes.
Sometimes, it is more convenient to deal with the number of defects or nonconformities
observed.

3.1 p-chart : The control chart for fraction nonconforming
If an item does not conform to standard on one or more of the several quality characteristics that
are examined by inspector, the item is classified as nonconforming.

Suppose the production process is operating in a stable manner,
i) the quality of each units produced are independent, and
ii) the probability that any nonconforming unit is p
0
.

If a random sample of n units is selected, let D
i
be the number of units that are nonconforming in
the sample, then
D
i
~ ( , ) b n p
n x p p C x D P
x n x
x
n
i
..., , 2 , 1 , 0 ) 1 ( ) ( = = =



The sample fraction nonconforming,
n
D
p
i
i
= .
Mean and variance of
i
p are

1 1
( ) ( ) ( ) ( )
i
i
p i i
D
E p E E D np p
n n n
= = = = =

2

2 2
1 1 (1 )
( ) ( ) ( ) ( (1 ))
i
i
p i i
D p p
Var p Var Var D np p
n n n n
o

= = = = =

Case 1 : Given standard value of p (
0
p p = )

n
p p
p LCL
p CL
n
p p
p UCL
) 1 (
3
) 1 (
3
0 0
0
0
0 0
0

=
=

+ =


The statistic sample fraction nonconforming
n
p p p ... , ,
2 1
are plotted with control limits CL,
UCL and LCL. If all the points plotted are within control limits, we conclude that the process is
in-control at the level p. Otherwise, we conclude that the fraction nonconforming has shifted to a
new level and the process is out-of-control.



Quality Control
UDPS3143 C3 - 2
Case 2 : Standard value of p unknown
p is estimated from observed data of m preliminary samples, each of size n. ( 20 > m )

Suppose there are D
i
nonconforming units in sample i,
n
D
p
i
i
= ,
Estimate of p is given by the average of individual sample fraction nonconforming,

=
=
= =
m
i
i
m
i
i
p
m mn
D
p
1
1

1
.

The trial control limits are
n
p p
p LCL
p CL
n
p p
p UCL
) 1 (
3
) 1 (
3

=
=

+ =


If all points plot inside the trial control limits and no systematic behavior is evident, then we
conclude that the process was in-control in the past, and the trial control limits are suitable for
controlling current and future production.

If one or more points fall outside the control limits, look for assignable causes. If an assignable
cause is found, discard the points and recalculate the trial control limits. If assignable cause is
not found for only one or two points, retain the points. The remaining points are checked
whether there are still any points beyond the new control limits.

This process is continued until all points show in-control status, at which the trial control limits
are adopted for future process control. If p is known and serve as a target value, determine
whether an out-of-control condition is caused by the process being out-of-control at the target p
but in-control at some other value of p.

Note : 1) If 0 < LCL , then we set 0 LCL = , since
n
p p p ... , ,
2 1
are positive values.
2) Points below LCL may not indicate an improved quality. Often they are caused
by errors, inexperienced inspectors, improper calibrated test instrument.



Example 3.1
Frozen orange juice concentrate is packed in 6-oz cardboard cans. A nonconforming can is one,
when filled, it could possibly leak either on the side seam or around the bottom joint. 30 samples
of 50 = n cans each were selected at half an hour intervals over a three-shift period in which the
machine was in continuous operation. The data is as follows.



Quality Control
UDPS3143 C3 - 3


( )
( ) ( )
( )
30
1

0.24 0.30 0.12

0.2313
30 30
1 0.2313 0.7687
3 0.2313 3 0.4102
50
1
3 0.2313 0.1789 0.0524
i
i
p
p
p p
UCL p
n
CL p
p p
LCL p
n
=
+ + +
= = =

= + = + =
=

= + = =

Samples 15 and 23 above UCL
process not in-control.
Investigate for assignable causes
Sample 15 a new batch of cardboard
Sample 23 inexperienced operator
eliminate samples 15 & 23 and new
control limits computed.

#If investigations shows that during
the entire 2-hour period a temporary
inexperience worker is on duty to
collect samples 21 to 24, then we
discard all 4 samples even if only
sample 23 exceeds the control limit
(The fraction nonconforming might
not be correct).

Quality Control
UDPS3143 C3 - 4
( ) ( )
( )
( )
( )
301
0.215
28 50
1
3
0.215 0.785
0.215 0.3893
50
0.215
1
3
0.215 0.1743 0.0407
p
p p
UCL p
n
CL p
p p
LCL p
n
= =

= +
= + =
= =

=
= =






There is no obvious nonrandom patterns for the remaining 28 samples.
We conclude that the process is in-control at the level 2150 . 0 = p and the revised control limits
should be adopted for monitoring current production.

The process is operating in a stable manner and no unusual operator-controllable problems. But,
the fraction nonconforming is too high. Unlikely the work-force level can improve the process
quality. Intervention by management is required to improve performance. Engineers studied the
process and made several adjustments to machines. We collect in addition 24 samples of size
50 = n .


Process performance improves following the machine adjustments (less fraction
nonconforming).
Quality Control
UDPS3143 C3 - 5


Collect additional 24 samples of size n =
50, sample 41 below LCL but no
assignable cause retain the point.








Hypothesis Testing
To test the hypothesis that the process fraction nonconforming has decreased after machine
adjustments,

2 1 1
2 1 0
:
:
p p H
p p H
>
=

where p
1
=process fraction nonconforming before adjustments.
P
2
=process fraction nonconforming after adjustments

( )
( )
( )( ) ( )( )
1 1
2 2
1 1 2 2
1 2
0.2150 preliminary data
133
0.1108
50 24
28 50 0.2150 24 50 0.1108
0.1669
1400 1200
p p
p p
n p n p
p
n n
= =
= = =
+
+
= = =
+ +



If H
0
is true, using normal approximation, test statistic,
( )
( )
( )
1 2 1 2
1 2

0.2150 0.1108 0
7.1
1 1
1 1
0.1669 0.8331
1
1400 1200
p p p p
Z
p p
n n


= = =
| | | |
+
+
|
|
\ .
\ .

Critical region is Z
0.05
=1.645 for o =0.05
Z =7.1 >Z
0.05
, reject H
0
and conclude that there has been a significant decrease in fraction
nonconforming.







Quality Control
UDPS3143 C3 - 6
Revised control limits

After successful process adjustments
( ) ( ) ( )
( )
0.1108
1 0.1108 0.8892
3 0.1108 0.2440
50
1
3 0.1108 0.1332 0.0224 0
CL p
p p
UCL p
n
p p
LCL p
n
= =

= + = + =

= = = =


After machine adjustments, all points fall inside the revised control limits Process is in-
control at this new level.

3.2 Design of the fraction nonconforming control chart
p-chart has 3 parameters to be specified:
- Choice of sample size, n
- Frequency of sampling
- Width of control limits

Choice of sample size n
Sample size must be large enough to allow the opportunity for some nonconforming items to be
present on average.

a) It is relatively common to have 100% inspection of all process output over some
convenient period of time.

b) If p is small, choose n sufficiently large so that probability of finding at least one
nonconforming unit in the sample is high. Otherwise, it is possible that only one
nonconforming unit in a sample indicates out-of-control condition.
Example
If , 8 , 01 . 0 = = n p

( ) 0.01 0.99
0.01 3 0.1155
8
UCL = + =
Quality Control
UDPS3143 C3 - 7
If there is one nonconforming unit in the sample,

( )
1
0.125
8
i
p UCL = = >

Since for any 0 > p , there is a positive possibility of producing some defectives, it is
unreasonable in many cases to conclude that the process is out-of-control upon observing
a single nonconforming item. To avoid this pitfall, we choose the sample size n so that
> > ) 1 (
i
D P where D
i
=number of nonconforming units.

Example
95 . 0 = , 95 . 0 ) 1 ( > >
i
D P
Using Poisson approximation to Binomial, from statistical table,
np = must be exceed 3.0


3.0
3.0
300
0.01
mean np
n
= >
> =


c) Duncan suggested that sample size should be large enough so that
50 . 0 ) , , ( ~ p from shift process a detecing P o
50 . 0 ) (
1
~ + = o p p p level at control of out process P
50 . 0 ) (
1 1 1
~ + = > < o p p UCL p or LCL p P
Assuming that normal approximation to the binomial applies, we choose n so that
UCL p =
1
,
n
p p
k p p
) 1 (
+ = +o

n
p p
k
) 1 (
= o

2
(1 )
k
n p p
o
| |
=
|
\ .


Example
01 . 0 = p , process shift to 05 . 0
1
= p , using 3-sigma control limits.

( ) ( )
2
1
3
0.05, 0.05 0.01 0.04, 0.01 0.99 56
0.04
p p n o o
| |
= + = = = = =
|
\ .


Small shift (o small)
1
o
large n large
large sample size is needed to detect a small shift.

Quality Control
UDPS3143 C3 - 8
d) If p is small, choose n large enough so that 0 > LCL
0
) 1 (
>

=
n
p p
k p LCL

2
1
k
p
p
n

>

Example 05 . 0 = p with 3-sigma limits


( )
2
1 0.05
3 171
0.05
172
n
n

> =
=


Note : p-chart is based on binomial probability model; that is, the probability of occurrence of a
nonconforming unit is constant and successive units of production are independent. If
nonconforming units are clustered together and probability of occurrence of
nonconforming unit depends on previous units being nonconforming, the p-chart should
not be used.


3.3 Variable sample size
n
i
=sample size of i
th
sample

3.3.1 Variable-width control limits
Case 1 : Standard given (
0
p p = )
Control limits for of i
th
sample

i
i
n
p p
p LCL
p CL
n
p p
p UCL
) 1 (
3
) 1 (
3
0 0
0
0
0 0
0

=
=

+ =


Case 2 : No standard given
Control limits for of i
th
sample

p CL
n
p p
p p LCL UCL
i
p
=

= =
) 1 (
3 3 ,

o

where

=
=
=
m
i
i
m
i
i
n
D
p
1
1

Quality Control
UDPS3143 C3 - 9
Note : The width of the control limits is proportional to
i
n
1
.
Example 3.2
Consider the following data with variable sample size, construct a control chart with variable
width.

For point 1,
( )
( )( )
( )
1
1
234
0.096
2450
0.096
1
3
0.096 0.904
0.096 3 0.183
100
1
3
0.096 0.087 0.009
i
i
D
p
n
CL p
p p
UCL p
n
p p
LCL p
n
= = =
= =

= +
= + =

=
= =

Quality Control
UDPS3143 C3 - 10
3.3.2 Control limits based on an average sample size
The average sample size,
m
n
n
m
i
i
=
=
1

With this approach, we would have an approximate set of control limits. This assumes that the
future sample sizes will not differ greatly from those previously observed. With this, the control
limits will be constant.

However, if there is an unusually large variation in the size of a particular sample or if a point
plots near the approximate control limits, then the exact control limits for that point should be
determined.


Case 1 : Standard given (
0
p p = )

0
0 0
0
) 1 (
3 ,
p CL
n
p p
p LCL UCL
=

=


Case 2 : No standard given

p CL
n
p p
p LCL UCL
=

=
) 1 (
3 ,



Example 3.3
Recalculate the limits in Ex 3.2 with the average sample size approach.


( )
( )( )
( )
1
1
2450
98
25
0.096
1
3
0.096 0.904
0.096 3 0.185
98
1
3 0.096 0.089 0.007
n
CL
p p
UCL p
n
p p
LCL p
n
= =
=

= +
= + =

= = =





Quality Control
UDPS3143 C3 - 11
3.3.3 Standardized control limits

Plot the standardized values,
i
i
i
n p p
p p
z
) 1 (

= for given standard p.

Use p p = if no standard given.
The control limits are 3 = UCL , 0 = CL , 3 = LCL

Example 3.4
Recalculate the limits in Ex 3.2 with the standardized value approach.


Quality Control
UDPS3143 C3 - 12
The control chart for fraction nonconforming is widely used in non-manufacturing applications
of statistical process control.

Examples
1. Number of employee paychecks in error / distributed late.
2. Number of check requests not paid within standard period.
3. Number of deliveries made by supplier not on time.


3.4 The OC-curve for the p-chart and ARL
OC curve is a plot of | against process fraction nonconforming.

) ( ) ( p is mean control in process P p level at OC p OC = =
) (
0
p H Accepting P = |
) ( p limits control within fall points P =
) ( p UCL p LCL P < < =
) ( ) ( p LCL p P p UCL p P s < =
) ( ) ( p LCL
n
D
P p UCL
n
D
P s < =
) ( ) ( p nLCL D P p nUCL D P s < =
where D ~ ) , ( p n B

Example 3.5
Given 50 = n , 0303 . 0 = LCL , 3697 . 0 = UCL , compute the OC-curve for p-chart.

( )( ) ( ) ( ) ( )
( ) ( )
( ) ( )
50 0.3697 | 50 0.0303 |
18.49| 1.52|
18| 1|
P D p P D p
P D p P D p
P D p P D p
| = < s
= < s
= s s


Quality Control
UDPS3143 C3 - 13

Average run length
For any Shewhart chart,
) (
1
limtis control beyond plots points sample P
ARL =

o
1
= ARL if process is in-control

|
=
1
1
ARL if process is out-of-control

Example 3.6
Given 50 = n , 20 . 0 = p , 0303 . 0 = LCL , 3697 . 0 = UCL ,
a) What is the ARL if the process is in-control with 20 . 0 = = p p ?
b) What is the ARL if the process shifts to 30 . 0 = p ?

a) Process is in-control with 0.2 p p = =

( ) point plots in-control | 0.9973; 1 , 0.0027
1 1
370
0.0027
P p
ARL
| o | o
o
= = = =
= = =

b) If process shift to p =0.3, | =0.8594 (from table)

1 1
7
1 1 0.8594
ARL
|
= = =


It will take about 7 samples, on the average to detect this shift .

3.5 The np control chart for number of nonconforming items
For fixed n, np =number of nonconforming items in a sample of size n
Let x =the number of nonconforming item / unit in the process follows a binomial distribution
with parameter n, p.
( ) ~ , X bin n p for 0,1,2,3,..., x n = .
( ) E X np = , ( ) ( ) 1 Var X np p =
Quality Control
UDPS3143 C3 - 14
If standard given,

np CL
p np np LCL UCL
=
= ) 1 ( 3 ,


If standard unknown, estimate p using p .

Example 3.7
Refer to Ex. 3.1, calculate the control limits for a np control chart.

( ) ( ) ( )( )
( )
50, 0.2313
3 1 50 0.2313 3 50 0.2313 0.7687 20.51
11.565
3 1 11.565 8.945 2.62
n p
UCL np np p
CL np
LCL np np p
= =
= + = + =
= =
= = =

Since no. of nonconforming units is an integer,
If 20 units are nonconforming, process is in-control
If 21 units are nonconforming, process is out-of-control.


3.6 The c-chart and u-chart for number of nonconformities (defects)
A nonconforming item is a unit of product that does not satisfy one or more of the specifications.
Each specific point at which a specification is not satisfied results on a defect / nonconformity.
Consequently, a nonconforming item will contain at least one nonconformity.

The variable of interest is the number of nonconformities in a unit or the average number of
nonconformities per unit.

Since n is large and p, which is the probability for occurrence of each nonconformity, is small,
Poisson distribution is used to model the number of nonconformities.

Note : If X ~ Poisson(), then ... , 2 , 1 , 0 ,
!
) ( = = =

x
x
e
x X P
x

and = ) (X E , = ) (X Var .

3.6.1 c-chart
Suppose m samples are taken with a single unit of sampling.

Let c
i
=number of nonconformities in i
th
sample
c
i
~ Poisson(), m i ..., , 2 , 1 =
where 0 > is the mean of nonconformities for each sample.
... , 2 , 1 , 0 ,
!
) ( = =

i
i
c
i
c
c
e
c P
i

; = ) (
i
c E , = ) (
i
c Var

Quality Control
UDPS3143 C3 - 15
Case 1 : Standard given (
0
= )

0
0 0
3 ,

=
=
CL
LCL UCL


Case 2 : No standard given
Estimate using

=
= =
m
i
i
c
m
c c
1
1


c CL
c c LCL UCL
=
= 3 ,



Example 3.8
The number of nonconformities observed in 26 successive samples of 100 printed circuit boards
is given in the table below. Construct a control chart for nonconformities.



Note: The inspection unit =100 boards
26
1
516
19.85
26 26
i
i
c
c
=
= = =

Trial control limits are
3 19.85 3 19.85 33.22
19.85
3 19.85 13.37 6.48
UCL c c
CL c
LCL c c
= + = + =
= =
= = =

Sample 6 & 20 excluded,
472
19.67
24
c = =


Quality Control
UDPS3143 C3 - 16
Revised control limits are
3 19.67 3 19.67 32.97
19.67
3 19.67 13.30 6.37
UCL c c
CL c
LCL c c
= + = + =
= =
= = =

No out-of-control signal found. However,
no. of nonconformities per 100 boards is still
unacceptably high. Management action is
required.











Further analysis of nonconformities
Defect or nonconformity data provides more information than fraction nonconforming, because
usually different types of nonconformities exist.

For example, in the printed circuit board process, there are 16 different types of defects. Defect
data for 500 boards are plotted on a Pareto chart below.



Over 60% of total defects is due to 2 defect types, solder insufficiency & solder cold joints.
Improve the wave soldering process and yield will improve the products dramatically.

Quality Control
UDPS3143 C3 - 17
3.6.2 u-chart
In c-chart, one inspection unit is used (a group of 5 units of products / 10 units of products).

With m sample size of n inspection units each (i.e. 100 boards), 2 approaches to construct control
charts:
a) Redefine a new inspection unit which is equal to n times the old inspection unit.
New control limits are
c n CL
c n c n LCL UCL
=
= 3 ,

where c is the observed mean number of nonconformities in original inspection unit.

b) Use u-chart for average number of nonconformities per inspection unit.

n
c
u
i
i
= where c
i
=number of nonconformities in i
th
sample of n inspection units.

c
i
~ Poisson(n) [ is mean of nonconformities in one unit]




Case 1 : Standard given (
0
= )

0
0
) (
1
) ( ) (

= = = =
n
n
c E
n n
c
E u E
i
i
i


n n
n
c Var
n n
c
Var u Var
i
i
i
0
2
0
2
) (
1
) ( ) (

= = = =
0
0
0
3 ,

=
=
CL
n
LCL UCL


Case 2 : No standard given

u CL
n
u
u LCL UCL
=
= 3 ,
where

=
=
m
i
i
u
m
u
1
1












Quality Control
UDPS3143 C3 - 18
Example 3.9
A PC manufacturer wishes to establish a control chart for nonconformities per unit on the final
assembly line. Data on the number of nonconformities in 20 samples of 5 computers each are
shown below.












38.6
1.93
20
u = =
Control limits are
1.93
3 1.93 3 3.79
5
1.93
3 1.93 1.86 0.07
u
UCL u
n
CL u
u
LCL u
n
= + = + =
= =
= = =

The preliminary data do not exhibit out-of-control.
Adopt trial control limit for current control process.



Quality Control
UDPS3143 C3 - 19
3.7 The operating-characteristic function

3.7.1 c-chart
OC curve is a plot of | against the true mean number of defects.

) ( ) ( c LCL x P c UCL x P s < = |


where x ~ ) (c Poisson

Example 3.10
Refer to Ex. 3.8, compute the OC-curve for c-chart.


3.7.2 u-chart
OC curve is a plot of | against the true mean number of defects per unit.


( ) ( )
( ) ( )
P x UCL u P x LCL u
P c nUCL u P c nLCL u
| = < s
= < s



where c ~ ) (nu Poisson





( ) ( )
( ) ( )
33.22| 6.48|
33| 6|
P X c P X c
P X c P X c
| = < s
= s s

Quality Control
UDPS3143 C3 - 20
3.8 Procedure with variable sample size
For 100% inspection, number of units in a sample usually is not a constant.
Therefore, a control chart for nonconformities per unit (u-chart) is used.

Example 3.11
In a textile finishing plant, dyed cloth is inspected for the occurrence of defects per 50 square
meters. The data is shown below.


1 inspection unit =50 square meters
500
10
50
n = =


3 different approaches

3.8.1 Control limits for each sample

=
=
= =
m
i
i
m
i
i
n
c
units inspection of no Total
ities nonconform of no Total
u
1
1
.
.


u CL
n
u
u LCL UCL
i
=
= 3 ,

Quality Control
UDPS3143 C3 - 21


Sample 1:
153
1.42
107.5
1.42
1.42 3 1.42 1.13 2.55
10
1.42 1.13 0.29
CL u
UCL
LCL
= = =
= + = + =
= =


Control limits change as the sample size varies, but center line remains constant.
meaningful comparisons between the sample.


3.8.2 Control limits based on average sample size

=
=
m
i
i
n
m
n
1
1

u CL
n
u
u LCL UCL
=
= 3 ,


3.8.3 Standardized control chart
For each sample compute
i
i
i
n u
u u
z

= , and plot z
i
on a control chart with 3 = UCL , 0 = CL ,
3 = LCL .


3.9 Non-manufacturing applications of c and u-chart.
Examples
1. Errors on engineering drawings.
2. Errors on plans and documents.
3. Errors in computer software.


Quality Control
UDPS3143 C3 - 22
Additional Example:



Quality Control
UDPS3143 C3 - 23







Quality Control
UDPS3143 C3 - 24
3.10 Choices between attributes and variable control chart
Obvious choice eg. Color uses attribute

Advantages of attributes control chart:
1. Several quality characteristics can be jointly considered. Unit classified as
nonconforming if it fails to meet specification on any one of the quality characteristics.
2. Simplicity
3. Inexpensive and less time-consuming.

Advantages of variables control chart:
1. Provide more useful information about process performance.
2. Specification about process mean and variability is obtained directly.
3. Points plotting out-of-control provide information on potential cause of that out-of-
control signal.
4. Used for process capability studies.
5. Provide indication of impending trouble and allow corrective action before any defectives
are actually produced. Thus X and R charts are leading indicators of trouble and are
more powerful control tools than p-chart.
6. For a specified level of protection against process shifts, variable control charts require a
much smaller sample size than attributes control chart. (Variable-type inspection is more
expensive and time-consuming). Sample size is important especially in destructive tests.
(eg. Opening a can to measure the volume of product within or to test chemical properties
of the product.)

3.11 Guidelines for implementing control charts
1. Choosing the proper type of control charts.
2. Determining which process characteristics to control.
3. Determining where the charts should be implemented in the process.
4. Taking actions to improve processes as the result of SPC/control chart analysis.
5. Selecting data collection systems and computer software.


3.11.1 Choosing the proper type of control chart
a. X and R (or X and s) charts
Consider using measurements control charts in these situations:
1. A new process is coming on stream, or a new product is being manufactured by an
existing process.
2. The process has been in operation for some time, but it is chronically in trouble or
usable to hold the specified tolerances.
3. The process is in trouble and the control chart can be used for diagnostic purposes
(troubleshooting).
4. Destructive testing (or other expensive testing procedures) is required.
5. It is desirable to reduce acceptance-sampling or other downstream testing to a
minimum when the process can be operated in-control.
6. Attributes control charts have been used but the process is either out-of-control, or is in-
control but the yield is unacceptable.
Quality Control
UDPS3143 C3 - 25
7. There are very tight specifications, overlapping assembly tolerances or other difficult
manufacturing problems.
8. The operator must decide whether or not to adjust the process, or where a set-up must
be evaluated.
9. A change in product specifications is desired.
10. Process stability and capability must be continually demonstrated, such as in regulated
industries.

b. Attributes charts (p-charts, c-charts and u-charts)
Consider using attributes control charts in these situations:
1. Operators control the assignable causes and it is necessary to reduce process fallout.
2. The process is a complex assembly operation and product quality is measured in terms
of the occurrence of nonconformities, successful or unsuccessful product function and
so forth. (Examples include computers, office automation equipment, automobiles and
the major subsystems of these products.)
3. Process control is necessary but measurement of data cannot be obtained.
4. A historical summary of process performance is necessary. Attributes control charts,
such as p-charts, c-charts and u-charts, are very effective for summarizing information
about the process for management review.

c. Control charts for individuals
1. It is convenient or impossible to obtain more than one measurement per sample or
repeat measurements will only differ by laboratory or analysis error. Examples often
occur in chemical processes.
2. Automated testing and inspection technology allow measurement every unit produced.
In these cases, also consider the cumulative-sum control chart and the exponentially
weighted moving-average control chart.
3. The data become available very slowly and waiting for a larger sample will be
impractical or make the control procedure too slow to react to problems.

3.11.3 Action taken to improve the process.
Process improvement is the primary objective of statistical process control. The
application of control charts will give information on two key aspects of the process: (1)
statistical control and (2) capability.