Variables control charts

In the variables control charts, it is important to maintain control over both the process mean and
process variability. The figure below illustrates the output of a production process.
In figure a) both the mean and the standard deviation are in control at their nominal values (say 0 and

0 ); consequently most of the process output falls within the specification limits.
However in figure b) the mean has shifted to a value 1 > 0 , resulting in a higher fraction of
nonconforming product

In figure c) the process standard deviation has shifted to a value 1 > 0 . This also results in higher
process fallout, even though the process mean is still at its nominal value

X and R charts
To construct charts, select at least 20 to 25 samples. Suppose that k samples are available, each
containing n observations.
Let x1 , x 2 , ..., x k be the average of each sample. Then the best estimator of , the process average, is
the grand average
=

x=

x1 + x 2 + ... + x k
k

Thus

x would be used as the center line on the

X chart

To construct the control limits, we need an estimate of the standard deviation of a mean X =

Therefore, we need an estimate for the standard deviation of the process

We may estimate from either the standard deviations or the ranges of the k samples. For the
present, we will concentrate on the range method.
If x1 , x 2 , ..., x n is a sample of size n , then the range of the sample is the difference between the largest
and the smallest observations; that is,

R = x max x min
Let R1 , R2 ,..., Rm be the ranges of the k samples. The average range is R =

R1 + R2 + + Rk
k

We will define the standard deviation of a normal distribution in terms of the range of a normal
distribution R by means of the following variable:

W =

The variable W is called the relative range. The mean of W is a value d 2 that changes as a function of
the sample size n . The table attached to this handout contains values of d 2 for different sample sizes.

R
E [W ] = E

R
d2 =

and =

R
.
d2

The estimator of standard deviation of the mean becomes

x =

R
d2 n

X control chart parameters

UCL = x +

Upper Control Limit =

3
d2 n

R = x + A2 R

Center Line

CL = x
=

Lower Control Limit =

Values for

LCL = x

3
d2 n

R = x A2 R

A2 can be found in the appendix table

Now consider the

chart. The center line will be

R.

To determine the control limits we need an

estimate of R

R = W

V [R ] = V [W ] = 2V [W ] and

R = V [R ] = 2V [W ] = V [W ]
R = V [W ] = d 3
The value of d 3 changes with the sample size n . Values for d 3 can be found in the appendix table

Since we know that =

R
R

then R = d 3
d2
d2

R control chart parameters

UCL = R + 3d 3

Upper Control Limit =

Center Line

CL = R

LCL = R 3d 3

Lower Control Limit =

Values for

R
= D4 R
d2

D3 and D4

R
= D3 R
d2

can be found in the appendix table

Example 1
Piston rings for an automotive engine are produced by a forging process. We wish to establish statistical
control of the inside diameter of the rings manufactured by this process using X and R control charts.
Twenty five samples, each of size five, have been taken when the process was perceived as in control.
The inside diameter measurement data from these samples are shown in the table above
a) Construct a preliminary control chart to check whether the process was in control or not

We find that the center line for the R chart is

25

R=

i =1

25

0.581
= 0.023
25

For samples of n = 5, we find from the appendix table that D3 = 0 and D4 = 2.11
Therefore, the parameters for the R chart are as follows
Upper Control Limit =
Center Line

Lower Control Limit =

UCL = D4 R = 2.11(0.023) = 0.049

CL = 0.023

LCL = D3 R = 0(0.023) = 0

Sample
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25

Observations
74.030
73.995
73.988
74.002
73.992
74.009
73.995
73.985
74.008
73.998
73.994
74.004
73.983
74.006
74.012
74.000
73.994
74.006
73.984
74.000
73.982
74.004
74.010
74.015
73.982

74.002
73.992
74.024
73.996
74.007
73.994
74.006
74.003
73.995
74.000
73.998
74.000
74.002
73.967
74.014
73.984
74.012
74.010
74.002
74.010
74.001
73.999
73.989
74.008
73.984

74.019
74.001
74.021
73.993
74.015
73.997
73.994
73.993
74.009
73.990
73.994
74.007
73.998
73.994
73.998
74.005
73.986
74.018
74.003
74.013
74.015
73.990
73.990
73.993
73.995

xi
73.992
74.011
74.005
74.015
73.989
73.985
74.000
74.015
74.005
74.007
73.995
74.000
73.997
74.000
73.999
73.998
74.005
74.003
74.005
74.020
74.005
74.006
74.009
74.000
74.017

74.008
74.004
74.002
74.009
74.014
73.993
74.005
73.988
74.004
73.995
73.990
73.996
74.012
73.984
74.007
73.996
74.007
74.000
73.997
74.003
73.996
74.009
74.014
74.010
74.013

74.010
74.001
74.008
74.003
74.003
73.996
74.000
73.997
74.004
73.998
73.994
74.001
73.998
73.990
74.006
73.997
74.001
74.007
73.998
74.009
74.000
74.002
74.002
74.005
73.998
= 1850.028

0.038
0.019
0.036
0.022
0.026
0.024
0.012
0.030
0.014
0.017
0.008
0.011
0.029
0.039
0.016
0.021
0.026
0.018
0.021
0.020
0.033
0.019
0.025
0.022
0.035
0.581

x = 74.001

R = 0.023

And we may now construct the x chart. The estimate for the grand average is
25

x=

x
i =1

25

1850.028
= 74.001
25

For samples of n = 5, we find from the appendix table that A2 = 0.58 .

Therefore, the parameters for the X chart are as follows

Ri

UCL = x + A2 R = 74.001 (0.58)(0.023) = 74.014

Upper Control Limit =

Center Line

CL = x
=

Lower Control Limit =

LCL = x A2 R = 74.001 (0.58)(0.023) = 73.988

The preliminary X and R charts are plotted on MINITAB:

Xbar-R Chart of 1, ..., 5

U C L=74.01461
Sample M ean

74.01
_
_
X=74.00115

74.00

73.99

LC L=73.98769
1

11

13
Sample

15

17

19

21

23

25

U C L=0.04934

Sample Range

0.048
0.036

_
R=0.02334

0.024
0.012
0.000

LC L=0
1

11

13
Sample

15

17

19

21

23

25

i) In class exercise: How would you get this graph from MINITAB?

When interpreting up X and R control charts, it is convenient to begin with the R chart. Because the
control limits on the X chart depend on the process variability, unless process variability is in control,
these limits will not have much meaning. The graph above verifies that the process is in control and that
any point should be removed.

X and S charts

In the X and S control charts, the sample standard deviation S is used to estimate the process
standard deviation
The X and S control charts are preferred over the X and R charts in the following two scenarios:
1. The sample size n is moderately large ( n > 10 or n > 12 ). The range is not an efficient
indicator of the process variability for moderate to large samples.
2. The sample size n is variable.

Construction of X and S control charts,

=

The grand average

x would be used as the center line on the

X chart

To construct the control limits, we need an estimate of the standard deviation of a mean X =
Therefore, we need an estimate for the standard deviation of the process .
First, note that
but

[ ]

E s2 = 2

E [s ]

So we cannot use direct values of the sample standard deviation as estimators of .

However, the following is true:

1/ 2

2
E [s ] =

n 1

( 2)

[(n 1) / 2]

= c4

and

s=

E [s ]
c4

Where c 4 varies with the sample size n . Values for c 4 are given in the appendix table

Suppose that k preliminary samples, each of size n are available, and let S i be the standard deviation of
the ith sample. The average of the k standard deviations is
k

S =

S
i =1

And we will use

S
as the estimator of .
c4

X control chart parameters

=

Upper Control Limit =

UCL = x +

3S
c4 n

= x + A3 S

Center Line

CL = x
=

Lower Control Limit =

Values for

LCL = x

3S
c4 n

= x A3 S

A3 can be found in the appendix table

Now consider the

chart. The center line will be

. To determine the control limits we need an

estimate of S . It has been previously shown that S = 1 c42 . Then S =

S control chart parameters

Upper Control Limit =

Center Line

Lower Control Limit =

Values for

UCL = S +

3S
1 c42 = B4 S
c4

CL = S

LCL = S

3S
1 c42 = B3 S
c4

B3 and B4 can be found in the appendix table

S
1 c 42
c4

Example 2
Construct the X and S control charts using the piston-ring inside diameter measurements (example 1)
Sample
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25

Observations
74.030
73.995
73.988
74.002
73.992
74.009
73.995
73.985
74.008
73.998
73.994
74.004
73.983
74.006
74.012
74.000
73.994
74.006
73.984
74.000
73.982
74.004
74.010
74.015
73.982

74.002
73.992
74.024
73.996
74.007
73.994
74.006
74.003
73.995
74.000
73.998
74.000
74.002
73.967
74.014
73.984
74.012
74.010
74.002
74.010
74.001
73.999
73.989
74.008
73.984

74.019
74.001
74.021
73.993
74.015
73.997
73.994
73.993
74.009
73.990
73.994
74.007
73.998
73.994
73.998
74.005
73.986
74.018
74.003
74.013
74.015
73.990
73.990
73.993
73.995

73.992
74.011
74.005
74.015
73.989
73.985
74.000
74.015
74.005
74.007
73.995
74.000
73.997
74.000
73.999
73.998
74.005
74.003
74.005
74.020
74.005
74.006
74.009
74.000
74.017

74.008
74.004
74.002
74.009
74.014
73.993
74.005
73.988
74.004
73.995
73.990
73.996
74.012
73.984
74.007
73.996
74.007
74.000
73.997
74.003
73.996
74.009
74.014
74.010
74.013

xi

Si

74.010
74.001
74.008
74.003
74.003
73.996
74.000
73.997
74.004
73.998
73.994
74.001
73.998
73.990
74.006
73.997
74.001
74.007
73.998
74.009
74.000
74.002
74.002
74.005
73.998

0.0148
0.0075
0.0147
0.0091
0.0122
0.0087
0.0055
0.0123
0.0055
0.0063
0.0029
0.0042
0.0105
0.0153
0.0073
0.0078
0.0106
0.0070
0.0085
0.0080
0.0122
0.0074
0.0119
0.0087
0.0162

= 1850.028

0.2350

x = 74.001 S = 0.0094

Note that we have assumed that the sample standard deviation is defined as

S=

(x
i =1

x)2

n 1

The grand average and the average standard deviation are

and

x=

1 25
1
xi =
(1850.028) = 74.001

25 i =1
25

S =

1 25
1
Si =
(0.2350) = 0.0094

25 i =1
25

For samples of n = 5, we find from the appendix table that A3 = 1.43 , B3 = 0 and B4 = 2.09
Consequently , the parameters for the X charts are

UCL = x + A3 S = 74.001 + (1.43 0.0094) = 74.014

CL = x = 74.001
LCL = x A3 S = 74.001 (1.43 0.0094) = 73.988
and for the S chart

UCL = B4 S = 2.09 0.0094 = 0.0196

CL = S = 0.0094
LCL = B3 S = 0 0.0094 = 0

The control charts are plotted as follows

Xbar-S Chart of 1, ..., 5

U C L=74.01461
Sample M ean

74.01
_
_
X=74.00115

74.00

73.99

LC L=73.98769
1

11

13
Sample

15

17

19

21

23

25

Sample StDev

0.020

U C L=0.01970

0.015
_
S =0.00943

0.010
0.005
0.000

LC L=0
1

11

13
Sample

15

17

19

21

23

25

ii) In class exercise: How would you get this graph from MINITAB?

Procedures with variable sample size

If ni is the number of observations in the ith sample, then use

x=

n x
i =1
k

i i

ni

and

i =1

2
(
)

n
S
1
i
i
i
=
1

S = k

n
m

i =1

1/ 2

The limits will be calculated as before but the constants A3 , B3 and B4 will depend on the sample size
used in each individual subgroup

Example 3. Consider the following dataset, which is a modification of the piston ring data used
previously. Note that the samples sizes vary from n = 3 to n = 5

Sample
Number

xi

Observations

74.03

74.002

74.019

73.992

73.995

73.992

74.001

73.988

74.024

74.021

74.005

74.002

73.996

73.993

73.992

74.007

74.009

74.010

0.0148

73.996

0.0046

74.002

74.008

0.0147

74.015

74.009

74.003

0.0091

74.015

73.989

74.014

74.003

0.0122

73.994

73.997

73.985

73.996

0.0099

73.995

74.006

73.994

74

73.999

0.0055

73.985

74.003

73.993

74.015

73.997

0.0123

74.008

73.995

74.009

74.005

74.004

0.0064

10

73.998

74

73.99

74.007

73.995

73.998

0.0063

11

73.994

73.998

73.994

73.995

73.99

73.994

0.0029

12

74.004

74

74.007

74

73.996

74.001

0.0042

13

73.983

74.002

73.998

73.994

0.0100

14

74.006

73.967

73.994

73.990

0.0153

15

74.012

74.014

73.998

74.008

0.0087

16

74

73.984

74.005

73.998

73.997

0.0078

17

73.994

74.012

73.986

74.005

73.999

0.0115

18

74.006

74.01

74.018

74.003

74

74.007

0.0070

19

73.984

74.002

74.003

74.005

73.997

73.998

0.0085

20

74

74.01

74.013

74.008

0.0068

21

73.982

74.001

74.015

74.005

73.996

74.000

0.0122

22

74.004

73.999

73.99

74.006

74.009

74.002

0.0074

23

74.01

73.989

73.99

74.009

74.014

74.002

0.0119

24

74.015

74.008

73.993

74

74.01

74.005

0.0087

25

73.982

73.984

73.995

74.017

74.013

73.998

0.0162

74

74.008

Si

73.988

73.984
73.996

a) Construct the preliminary X and S control charts

x=

n x
i =1
k

i i

n
i =1

5(74.010) + 3(73.996) + ... + 5(73.998) 8362.075

=
= 74.001

5 + 3 + ... + 5
88

2
(ni 1)Si

S = i =1 k

ni m
i =1

1/ 2

4(0.0148)2 + 2(0.0046 )2 + ... + 4(0.0162 )2 0.008426 1/ 2

=
=
= 0.0098
(
)
5
3
...
5
25
88
+
+
+

The parameters for the X chart are in the first point of the chart are as follows

UCL = x + A3 S = 74.001 + (1.43 0.0098) = 74.015

CL = x = 74.001
LCL = x A3 S = 74.001 (1.43 0.0098) = 73.987
and for the S chart

UCL = B4 S = 2.09 0.0098 = 0.020

CL = S = 0.0098
LCL = B3 S = 0 0.0098 = 0

The following are the complete charts obtained with MINITAB

Xbar-S Chart of 1, ..., 5
74.02

Sample Mean

UCL=74.01460
74.01
_
_
X=74.00075

74.00
73.99

LCL=73.98691

73.98
1

11

13
Sample

15

17

19

21

23

25

UCL=0.02027

Sample StDev

0.020
0.015

_
S=0.00970

0.010
0.005
0.000

LCL=0
1

11

13
Sample

15

17

19

Tests performed with unequal sample sizes

iii) In class exercise: How would you get this graph from MINITAB?

21

23

25

Control charts with individual values

These are charts where the sample size used for process monitoring is n = 1 .
For these charts, it is possible to use the moving of two successive observations as the basis of
estimating the process variability. The moving range is defined as

MRi = xi xi 1
Where

xi is the value for observation i

xi 1 is the value for observation i 1

Let k be the number of samples (observations) used to build the control chart. We can estimate process
variability by the average of the k 1 moving range values
k

MR =

MR
i =2

k 1

R control chart parameters (using MR instead of R )

Upper Control Limit =

Center Line

Lower Control Limit =

The values for

UCL = MR + 3d 3

MR
= D4 MR = 3.267 MR
d2

CL = MR

LCL = MR 3d 3

D3 and D4

MR
= D3 MR = 0.000MR
d2

are chosen considering that the moving range is calculated with two values

X control chart parameters

k

x=

x
i =1

Upper Control Limit =

Center Line

Lower Control Limit =

The value for

3
d2

UCL = x +

3
d2 n

MR = x +

3
MR = x + 2.66 MR
d2

MR = x

3
MR = x 2.66 MR
d2

CL = x

LCL = x

3
d2 n

is chosen considering that the moving range is calculated with two values

Example 4. The viscosity of aircraft primer paint is an important quality. The product is produced in
batches, and because each batch takes several hours to produce, the production rate is too slow to
allow sample sizes greater than one. The viscosity of 15 batches is shown in the following table

Batch
number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15

Viscosity
33.75
33.05
34
33.81
33.46
34.02
33.68
33.27
33.49
33.2
33.62
33
33.54
33.12
33.84

a) Calculate a preliminary chart for the individual values

R control chart parameters (using MR instead of R )

MR

MR =

i =2

k 1

Upper Control Limit =

Center Line

Lower Control Limit =

33.05 33.75 + 34 33.05 + ... + 33.84 33.12

15 1

= 0.48

UCL = 3.267( MR ) = 3.267 * (0.48) = 1.568

CL = 0.48

LCL = 0.000( MR ) = 0.000(0.48) = 0.000

X control chart parameters

x=

x
i =1

33.75 + 33.08 + ... + 33.84

= 33.52
15

Upper Control Limit =

Center Line

Lower Control Limit =

UCL = x + 2.66 MR = 33.52 + 2.66(0.48) = 34.8

CL = 33.52

LCL = x 2.66 MR = 33.52 2.66(0.48) = 32.24

And the chart in MINITAB is as follows

I-MR Chart of Viscosity

U C L=34.802
Individual V alue

34.5
34.0
_
X=33.523

33.5
33.0
32.5

LC L=32.245
1

7
8
9
O bser vation

10

11

12

13

14

15

M oving Range

1.6

U C L=1.571

1.2
0.8
__
M R=0.481

0.4
0.0

LC L=0
1

7
8
9
O bser vation

10

11

12

13

14

15

iv) In class exercise: How would you get this graph from MINITAB?

b) The table below contains data on aircraft primer paint viscosity for batches 16-30. Plot the point
in the control chart. What do you observe?
Batch
number
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30

Viscosity
33.5
33.25
33.4
33.27
34.65
34.8
34.55
35
34.75
34.5
34.7
34.29
34.61
34.49
35.03

I-MR Chart of Viscosity

Individual V alue

35

U C L=34.802

34

_
X=33.523

33
LC L=32.245

32
1

10

13

16
O bser vation

19

22

25

28

M oving Range

1.6

U C L=1.571

1.2
0.8
__
M R=0.481

0.4
0.0

LC L=0
1

10

13

16
O bser vation

19

22

25

28

v) In class exercise: How would you get this graph from MINITAB?

An upward shift in mean viscosity has occurred around batch 20 to 21. Note that the moving range chart
also reacts to this level shift with a single large spike at sample 20. This spike on the moving-range chart
is sometimes helpful in identifying exactly where a process shift in the process mean has occurred.

Note: Caution is advised with the interpretation of the individual value charts when trends are present.
If there are trends, the moving range might be underestimating the standard deviation and the control
limits might be too narrow. This will results in an increase in false alarms.

Appendix table. Reproduced from: Scheaffer, Mulekar & McClave Probability and statistics for
engineers