6.

1 Table of Hole Diameter Data

sample Carbon-fiber composite material

Number 1 2 3 4 5 x Ri
-
1 -30 50 10 30 8 80
20
- -
2 0 50 30 0 110
60 20
3 -50 10 20 30 20 6 80
-
4 -10 -10 30 50 8 70
20
5 20 -40 50 20 10 12 90
-
6 0 0 40 20 4 80
40
-
7 0 0 20 -10 -2 40
20
-
8 70 -30 30 0 12 100
10
-
9 0 0 20 10 2 40
20
10 10 20 30 10 50 24 40
11 40 0 20 0 20 16 40
12 30 20 30 10 40 26 30
13 30 -30 0 10 10 4 60
-
14 30 -10 50 -30 6 80
10
15 10 -10 50 40 0 18 60
-
16 0 0 30 0 4 40
10
17 20 20 30 30 -20 16 50
18 10 -20 50 30 10 16 70
19 50 -10 40 20 0 20 60
20 50 0 0 30 10 18 50
21 127
Σxi and ΣRi
8 0
10. 63.
x and R 9 5

(a) set up x and R charts on the process. Is the process in statistical control?
From the problem, it is known that n =5, therefore we use Appendix table VI that
D3=0 and D4=2.114.
LC = R =63.5

LCL = R D3 = 63.5 (0) = 0

 UCL = R D4 = 63.5 (2.114) =134.239
For x chart, it is used A2= 0.577, therefore
UCL = x +A2 R = 10.9 + 0.577 (63.5) = 47.5395

LC = x =10.9
LCL = x -A2 R = 10.9 - 0.577 (63.5) = -25.7395

R Char t of C1 ; ...; C5
140
UCL=134.3

120

100
Sample Range

80
_
R=63.5
60

40

20

0 LCL=0

1 3 5 7 9 11 13 15 17 19
Sample

6.1.1 R chart
Xbar Chart of C1 ; ...; C5
50
UCL=47.53

40

30

20
Sample Mean

_
_
10 X=10.9

-10

-20
LCL=-25.73
-30
1 3 5 7 9 11 13 15 17 19
Sample

6.1.2 x chart

From the x chart and R chart, there aren’t no indication of out-of-control condition.
Since they both exhibit control, we would conclude that the process in in control.
(b) Estimate the process standard deviation using the range method

 R 63.5
σ= = = 27.367
d2 2.326
(c) if specification are at nominal ±100, what can you say about the capability of this
process? Calculate the PCR Cp.

x in ten-thousandths of an inch = 10.9 x 10,000 = 109000 inch

USL and LSL= 109,100 and 108,900 or we can say 10.91 and 10.89 (in ten-
thousandths of inch).
The process capability:

Cp USL − LSL 10.91 − 10.89 10.91 − 10.89(10,000)
= = = = 1.218
6σ̂ 6(27.367) 6(27.367 )

 1 1
P = C x100% = x 100% = 0.821 x 100% ≈ 82%.
p 1.218
The process uses up about 82% of the specification band.
6.7
numb
er of
sampl
e x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 xi si
2. 0. 0. 1. 0.
1
5 5 2 -1 1 -1 5 5 5 -1.5 0.5 1.333333
0. 1. 1.
2
0 0 5 1 5 1 -1 1 5 -1 0.45 0.926463
-
3 1. 1.
5 1 1 -1 0 5 -1 -1 1 -1 -0.1 1.125463
-
4 0. 1. 1.
0 5 -2 0 -1 5 5 0 -2 -1.5 -0.6 1.173788
- - -
5 0. 0. 0. 0.
0 0 0 5 5 1 5 5 0 0 0 0.471405
-
6 0. 0.
1 5 0 0 0 5 -1 1 -2 1 0 0.971825
1.
7
1 -1 -1 -1 0 5 0 1 0 0 0.05 0.895979
- -
8 1. 0. 1. 0.
0 5 5 5 0 0 0 -1 5 -0.5 -0.15 0.818196
-
9 1. 1. 1. 0.
-2 5 5 5 0 0 5 1 0 1 0.2 1.183216
- - -
10 0. 3. 1. 1.
5 5 0 -1 5 5 -1 -1 1 0.5 -0.15 1.528434
- -
11 1. 1. 0. 0.
0 5 0 0 2 5 5 5 2 -1 0.3 1.206464
- -
12 0. 0. 1. 0.
0 -2 5 0 5 2 5 0 5 -1 0 1.154701
 - - -
13 0. 0. 0. 0. 1.
-1 5 5 -1 0 5 5 5 -1 -1 -0.55 0.68516
- -
14 0. 0. 1. 1.
5 1 -1 5 -2 -1 5 0 5 1.5 -0.15 1.248332
1. 1.
15
1 0 5 5 1 -1 0 1 -2 -1.5 0.15 1.270389

∑ -0.05 15.99315
x and s -0.00333 1.06621

(a) Set up x and s control charts on this process. Does the process exhibit
statistical control? If necessary, construct revised control limits.
 The parameters for s chart can be written as:
UCL = B4 s = 1.716 x 1.06621= 1.829616
Center line = s = 1.06621
LCL = B3 s = 0.284 x 1.06621 = 0.302804
 The parameters for x chart can be written as:

3s 3(1.06621)
UCL = x + = − 0.00333 + = 1.036551
c4 n 0.9727 10
Center line = x = -0.00333
3s 3(1.06621)
LCL = x − = − 0.00333 − = -1.04322
c4 n 0.9727 10

S Char t of 1 ; ...; 1 0
2.0

UCL=1.830

1.5
Sample St Dev

_
S=1.066
1.0

0.5

LCL=0.302

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Sample
 6.7.1. s chart

Xbar Char t of 1 ; ...; 1 0

1.0 UCL=1.040

Sample Mean 0.5

_
_
0.0 X=-0.003

-0.5

-1.0 LCL=-1.047

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

6.7.2 x chart

The process exhibit statistical control.

(b) Set up an R chart, and compare with the s chart in part (a)
number of
x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 R
sample
2. 0. 0. 1. 0.
1
5 5 2 -1 1 -1 5 5 5 -1.5 4
0. 1. 1.
2
0 0 5 1 5 1 -1 1 5 -1 2.5
-
3 1. 1.
5 1 1 -1 0 5 -1 -1 1 -1 3
-
4 0. 1. 1.
0 5 -2 0 -1 5 5 0 -2 -1.5 3.5
- - -
5 0. 0. 0. 0.
0 0 0 5 5 1 5 5 0 0 1.5
-
6 0. 0.
1 5 0 0 0 5 -1 1 -2 1 3
1.
7
1 -1 -1 -1 0 5 0 1 0 0 2.5
- -
8 1. 0. 1. 0.
0 5 5 5 0 0 0 -1 5 -0.5 3
-
9 1. 1. 1. 0.
-2 5 5 5 0 0 5 1 0 1 3.5
10 - 3. 0 -1 - - -1 -1 1 0.5 5
0. 5 1. 1.
 5 5 5
- -
11 1. 1. 0. 0.
0 5 0 0 2 5 5 5 2 -1 3.5
- -
12 0. 0. 1. 0.
0 -2 5 0 5 2 5 0 5 -1 4
- - -
13 0. 0. 0. 0. 1.
-1 5 5 -1 0 5 5 5 -1 -1 2
- -
14 0. 0. 1. 1.
5 1 -1 5 -2 -1 5 0 5 1.5 3.5
1. 1.
15
1 0 5 5 1 -1 0 1 -2 -1.5 3.5

∑R i 48
R 3.2

n =10, therefore we use Appendix table VI that D3=0.223 and D4=1.777. R =3.2

LCL = R D3 = 3.2 (0.223) = 0.7136

UCL = R D4 = 3.2 (1.777) = 5.6864

R Char t of 1 ; ...; 1 0
6
UCL=5.686

4
Sample Range

_
R=3.2
3

1
LCL=0.714

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Sample

6.7.3 R chart
6.13

sampl
e
x1 x2 x3 x4 x5 x R
numb
er
1 83 81.2 78.7 75.7 77 79.1 7.3
2 88.6 78.3 78.8 71 84.2 80.2 17.6
3 85.7 75.8 84.3 75.2 81 80.4 10.5
4 80.8 74.4 82.5 74.1 75.7 77.5 8.4
 5 83.4 78.4 82.6 78.2 78.9 80.3 5.2
6 75.3 79.9 87.3 89.7 81.8 82.8 14.4
7 74.5 78 80.8 73.4 79.7 77.3 7.4
8 79.2 84.4 81.5 86 74.5 81.1 11.5
9 80.5 86.2 76.2 64.1 80.2 77.4 22.1
10 75.7 75.2 71.1 82.1 74.3 75.7 11
11 80 81.5 78.4 73.8 78.1 78.4 7.7
12 80.6 81.8 79.3 73.8 81.7 79.4 8
13 82.7 81.3 79.1 82 79.5 80.9 3.6
14 79.2 74.9 78.6 77.7 75.3 77.1 4.3
15 85.5 82.1 82.8 73.4 71.7 79.1 13.8
16 78.8 79.6 80.2 79.1 80.8 79.7 2
17 82.1 78.2 75.5 78.2 82.1 79.2 6.6
18 84.5 76.9 83.5 81.2 79.2 81.1 7.6
19 79 77.8 81.2 84.4 81.6 80.8 6.6
20 84.5 73.1 78.6 78.7 80.6 79.1 11.4
1586. 187.0
∑x i and ∑Ri 7
x and R 79.3 9.4

(a) Establish x and R control charts for compressive strength using these data. Is
the process in statistical control?
 For R chart (use D3 = 0 and D4= 2.114 from Apendix table VI because n=5):

UCL = R D4 = 9.4 (2.114) =19.8716

Center line = R = 9.4

LCL = R D3 = 9.4 (0) = 0

 For x chart (use A2 =0.577 from Apendix table VI because n=5):

UCL = x +A2 R = 79.3+ 0.577 (9.4) = 84.7568
Center line = x =79.3
LCL = x -A2 R = 79.3 - 0.577 (9.4) = 73.9092
Using minitab we can see the the R Chart and x chart as below:
 R Char t of C1 ; ...; C5
25
1

20 UCL=19.77

Sample Range
15

_
10
R=9.35

0 LCL=0

1 3 5 7 9 11 13 15 17 19
Sample

6.13.1 R Chart for 20 subgroups

Xbar Char t of C1 ; ...; C5

85.0 UCL=84.97

82.5
Sample Mean

80.0 _
_
X=79.33

77.5

75.0

LCL=73.69

1 3 5 7 9 11 13 15 17 19
Sa mple

6.13.2 x Chart for 20 subgroups

From the R Chart shown above, there is an out-of-control condition observed.

The value which is out-of-control from UCL is R= 22.1 whereas UCL is 19.8716.
however, the x chart exhibit control.

(b) After establishing the control charts in part (a), 15 new subgroups were
collected and the compressive strength are shown in table 6E.8. Plot the x and
R values against the control units from part (a) and draw conclusions.

sampl
e
numb
er x1 x2 x3 x4 x5 xbar R
1 83 81.2 78.7 75.7 77 79.12 7.3
2 88.6 78.3 78.8 71 84.2 80.18 17.6
3 85.7 75.8 84.3 75.2 81 80.4 10.5
4 80.8 74.4 82.5 74.1 75.7 77.5 8.4
5 83.4 78.4 82.6 78.2 78.9 80.3 5.2
 6 75.3 79.9 87.3 89.7 81.8 82.8 14.4
7 74.5 78 80.8 73.4 79.7 77.28 7.4
8 79.2 84.4 81.5 86 74.5 81.12 11.5
9 80.5 86.2 76.2 64.1 80.2 77.44 22.1
10 75.7 75.2 71.1 82.1 74.3 75.68 11
11 80 81.5 78.4 73.8 78.1 78.36 7.7
12 80.6 81.8 79.3 73.8 81.7 79.44 8
13 82.7 81.3 79.1 82 79.5 80.92 3.6
14 79.2 74.9 78.6 77.7 75.3 77.14 4.3
15 85.5 82.1 82.8 73.4 71.7 79.1 13.8
16 78.8 79.6 80.2 79.1 80.8 79.7 2
17 82.1 78.2 75.5 78.2 82.1 79.22 6.6
18 84.5 76.9 83.5 81.2 79.2 81.06 7.6
19 79 77.8 81.2 84.4 81.6 80.8 6.6
20 84.5 73.1 78.6 78.7 80.6 79.1 11.4
21 68.9 81.5 78.2 80.8 81.5 78.18 12.6
22 69.8 68.6 80.4 84.3 83.9 77.4 15.7
23 78.5 85.2 78.4 80.3 81.7 80.82 6.8
24 76.9 86.1 86.9 94.4 83.9 85.64 17.5
25 93.6 81.6 87.8 79.6 71 82.72 22.6
26 65.5 86.8 72.4 82.6 71.4 75.74 21.3
27 78.1 65.7 83.7 93.7 93.4 82.92 28
28 74.9 72.6 81.6 87.2 72.7 77.8 14.6
29 78.1 77.1 67 75.7 76.8 74.94 11.1
30 78.7 85.4 77.7 90.7 76.7 81.84 14
31 85 60.2 68.5 71.1 82.4 73.44 24.8
32 86.4 79.2 79.8 86 75.4 81.36 11
33 78.5 99 78.3 71.4 81.8 81.8 27.6
34 68.8 62 82 77.5 76.1 73.28 20
35 83 83.7 73.1 82.2 95.3 83.46 22.2
∑x i and ∑Ri 2778 456.8
79.371 13.051
x and R 43 43

 For R chart (use D3 = 0 and D4= 2.114 from Apendix table VI because n=5):

UCL = R D4 = 13.05143 (2.114) = 27.60

Center line = R = 13.05143

LCL = R D3 = 13.05143 (0) = 0

 For x chart (use A2 =0.577 from Apendix table VI because n=5):

UCL = x +A2 R = 79.37143 + 0.577 (13.05143) = 86.90
Center line = x =79.37143
LCL = x -A2 R = 79.37143 - 0.577 (13.05143) = 71.84
Using minitab we can see the the R Chart and x chart as below:

R Char t of 1 ; ...; 5
30
1 1
UCL=27.60
25

20
Sample Range

15 _
R=13.05

10

0 LCL=0

1 4 7 10 13 16 19 22 25 28 31 34
Sample

6.13.3 R Chart for 35 subgroups

Xbar Char t of 1 ; ...; 5

88
UCL=86.90
86

84

82
Sample Mean

_
_
80
X=79.37
78

76

74

72 LCL=71.84

70
1 4 7 10 13 16 19 22 25 28 31 34
Sample

6.13.4 x Chart for 35 subgroups

Conclusions:
6.19 Control charts for x and R are maintained for an important quality
characteristic. The sample size is n= 7; x and R are computed for each sample.
After 35 samples, we have found that:

35 35

∑ xi =7805 and
i =1
∑R
i =1
i =1200

(a) Set up x and R charts using these data

Control limits for the R Chart (since n=7, the value of D3 is 0.076 and D4 is
1.924 ):

UCL = D4 R = 1.924 x 34.286 =65.966

35

Center line = R = ∑R
i =1
i
=
1200
= 34.286
35
35

LCL = D3 R = 0.076 x 34.286 =2.605

7805
Control limits for the x Chart (since n=7, the value of A2=0.419 and x is =
35
223):

UCL = x +A2 R = 223 + 0.419 (34.286) =237.3658

Center line = x =223

LCL = x -A2 R = 223 - 0.419 (34.286) = 208.6342

(b) Assuming that both charts exhibit control, estimate the process mean and
standard deviation

The process mean = µ ≈ x = 223

 R 34.286
The estimator of σ is σ = = = 12.679
d2 2.704

(c) If the quality characteristic is normally distributed and if the spesifications are
220± 35, can the process meet the specifications? Estimate the fraction
nonconforming.

USL = 220 +35 = 255 and LSL = 220-35=185

P= P{x<185} + P{x>255}
  185 − 223   255 − 223 
= Φ  +1- Φ  
 12.679   12.679 

= Φ(-2.996) +1 – Φ(2.5237)

= 0.001368 + 1 - 0.994194 = 0.007174

It is about 0.7174 percent [7174 parts per millon(ppm)] of the products

produced will be aoutside of the specifications.

(d) Assuming the variance to remain constant, state where the process mean
should be located to minimize the fraction nonconforming. What would be the
value of fraction nonconforming under these conditions?

6.25 Suppose that the following the construction of the x and R control charts in
exercise 6.23, the process engineers decided to change the subgroup size to n =
2. Table 6E.11 contains 10 new subgroups of tickness data. Plot this data on the
control charts from exercise 6.23 (a) based on the new subgroup size. Is the
process in statistical control?

subgrou
p x1 x2 x3 x4 x bar R
1 459 449 435 450 448.25 24
2 443 440 442 442 441.75 3
3 457 444 449 444 448.5 13
4 469 463 453 438 455.75 31
5 443 457 445 454 449.75 14
6 444 456 456 457 453.25 13
7 445 449 450 445 447.25 5
8 446 455 449 452 450.5 9
9 444 452 457 440 448.25 17
10 432 463 463 443 450.25 31
11 445 452 453 438 447 15
12 456 457 436 457 451.5 21
13 459 445 441 447 448 18
14 441 465 438 450 448.5 27
15 460 453 457 438 452 22
16 453 444 451 435 445.75 18
17 451 460 450 457 454.5 10
18 422 431 437 429 429.75 15
19 444 446 448 467 451.25 23
20 450 450 454 454 452 4
8973.7
∑x i and ∑Ri 5 333
x and R 448.68 16.65
 75

From exercise 6.23, we know that

n old =4 , Rold = 16.65

and from Appendix Table VI we have

d2 (old) =2.059 d2 (new)= 1.128,

therefore, the new control limits on the x chart are found :

 d 2 new 
UCL = x + A2   Rold = 448.69 + 1.88(0.548)16.65= 465.84
 d 2 old 

 d 2 new 
LCL = x − A2   Rold = 448.69 - 1.88(0.548)16.65= 431.54
 d 2 old 

For the R chart, the new parameters are :

 d 2 (new) 
UCL = D4   Rold = 3.267 (0.548)16.65 = 29.808
 d 2 (old ) 

 d 2 (new) 
CL = Rnew =   Rold = 9.1242
 d 2 (old ) 

 d 2 (new) 
LCL = max {0, D3   Rold }= 0
 d 2 (old ) 

From here, we can make the chart to se whether the process in statistical
control or not.

6.31.Specifications on a cigar lighter detent are 0.3220 and 0.3200 in. Samples of size
5 are taken every 45 min with the results shown in Table 6E.13 (measured as
deviations from 0.3210 in 0.0001 in).

(a) Set up an R chart and examine the process for statistical control.

sample
numbe x1 x2 x3 x4 x5 x bar R
r
1 1 9 6 9 6 6.2 8
2 9 4 3 0 3 3.8 9
3 0 9 0 3 2 2.8 9
 4 1 1 0 2 1 1 2
5 -3 0 -1 0 -4 -1.6 4
6 -7 2 0 0 2 -0.6 9
7 -3 -1 -1 0 -2 -1.4 3
8 0 -2 -3 -3 -2 -2 3
9 2 0 -1 -3 -1 -0.6 5
10 0 2 -1 -1 2 0.4 3
11 -3 -2 -1 -1 2 -1 5
12 -16 2 0 -4 -1 -3.8 18
13 -6 -3 0 0 -8 -3.4 8
14 -3 -5 5 0 5 0.4 10
15 -1 -1 -1 -2 -1 -1.2 1
∑x i and ∑R i -1 97
-
0.0666 6.4666
x and R 7 67
Control limits for the R Chart (since n=5, the value of D3 is 0 and D4 is 2.114 ):

UCL = D4 R = 2.114 x 6.47 =13.67

15

Center line = R = ∑R
i =1
i
= 6.47
15

LCL = D3 R = 0 x 6.47 = 0

R Char t of C1 ; ...; C5
20
1

15
UCL=13.67
Sample Range

10

_
R=6.47
5

0 LCL=0

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

From the picture, there is a sample that is out of control (sample 12). Then, we
can eliminate the extrim data and recalculate the parameter.
(c) What parameters would you recommend for an R chart for on-line control?

The parameters recommended in this case are:

UCL = D4 R new= 2.114 x5.643 =11.929

15

Center line = R new= ∑R

i =1
i − R12
= 5.643
14

LCL = D3 R new= 0 x5.643 = 0

The chart is given below:

R Char t of C1 ; ...; C5

12 UCL=11.93

10

8
Sample Range

6 _
R=5.64

0 LCL=0

1 2 3 4 5 6 7 8 9 10 11 12 13 14
Sample

The R chart above indicate that there’s no out-of-control sample. Therefore,

since the R chart exhibit control, we would conclude that the parameters we
would reccomend are UCL = 11.929, Center line = 5.643 and LCL = 0

(d) Estimate the standard deviation of the process.

R 5.643
σˆ = = = 2.426
d 2 2.326

(e) What is the process capability?

USL − LSL 0.322 − 0.320

Cp= = = 1.374
6σ 6(2.426)(0.0001)
 1 1
P= x100% = x100% = 72.78%
Cp 1.374

6.37 Thirty samples each of size 7 have been collected to establish control over a
30 30
process. The following data were collected: ∑ xi = 2700 and
i =1
∑R
i =1
i = 120

(a) calculate trial control limits for the two charts

Trial control limit for x chart:

UCL = x +A2 R = 90 +0.419(4)=91.676

2700
CL = x = =90
30

LCL = x -A2 R = 90 - 0.419(4)=88.324

Trial control limit for R chart:

UCL = D4 R = 1.92 4(4)=7.696

120
CL= R = =4
30

LCL = = D3 R =0.076(4) = 0.304

(b) On the assumption that the R chart is in control, estimate the process
standard deviation.

 R 4
σ= = = 1.479
d 2 2.704

(c) Suppose an s chart were desired. What would be the appropriate control limits
and center line?

Suppose that σ is given above

UCL = B6 σ =1.806 (1.479)= 2.671

CL = σ = 1.479

LCL =B5 σ = 0.113 (1.479)= 0.167

6.43. x and R charts with n=4 and the control parameters below both exhibit control.

x Chart R Chart
 UCL = 815 UCL=46.98
Center Line = 800 Center Line=20.59
LCL= 785 LCL=0
What is the probability that a shift in the process mean to 790 will be detected in
the first sample following the shift?

First, calculate the probability of not detecting this shift that is:

β = P{ LCL ≤ x ≤ UCL µ = µ1 = µ 0 + kσ }

̴
Since x N(µ, σ 2 /n), then :

( ) (
β = Φ L−k n −Φ −L−k n )
We suppose that L=3 (the usual three-sigma limits), k=(µ1-µ0)/σ with σ can be
R 20.59
estimated by σˆ = = =10 →k=(790-800)/10= -1
d 2 2.059

then we have:

( ) (
β = Φ 3 − (−1) 4 − Φ − 3 − (−1) 4 )
= Φ (5)- Φ (-1)

= 1- 0.158655 = 0.841344

Last, the probability that such a shift will be detected in the first subsequent is
1-β= 1- 0.841344 = 0.158655

6.49 The following x and s charts based on n=4 have shown statistical control:

x Chart S Chart
UCL = 710 UCL=18.08
Center Line = 700 Center Line=7.979
LCL= 690 LCL=0
(a) Estimate the process parameters µ and σ

For µ, the estimate parameter is x = 700

s 7.979
For σ, the estimate parameter is σˆ = = =8.66
c 4 0.9213

(b) If the specifications are at 705 ± 15, and the process output is normally
distributed, estimate the fraction nonconforming.
  USL − LSL 720 − 690
Cp = = = 0.577
6σ 6(8.66)

  1   1 
P =   100% =  100% =1.733x100% =173.3%. the process uses up
 Cp   0.577 
more than 100% of the tolerance band. In this case the process is very yield-
sensitive, and a large number of nonconforming units will be produced.

(c) For the x Chart, find the probability of type I error, assuming σ is constant

(d) Suppose the process mean shifts to 693 and the standard deviation
simultaneously shifts to 12. Find the probability of detecting this shift on the x
Chart on the first subsequent sample.

(e) For the shift of part(d), find the average run length.

6.55. Michelson actually made 100 measurements on the velocity of light in five trials
of 20 observations each. The second set of 20 measurements is shown in table
6E.20.

(a) plot these new measurement on the control charts constructed in exercise
6.56. are these new measurements in statistical control? Give a practical
interpretation of the control charts.

(b) is there evidence that the variability in the measurement has decreased
between trial 1 and trial 2?

6.61.The vane heights for 20 of the castings from Fig. 6.25 are shown in table 6E.23.
Construct the “between/within “ control charts for these process data using a
range chart to monitor the within-castings vane height. Compare these to the
control charts shown in fig.6.27

7.1 The data in table 7E.1 represent the results of inspecting all units of a personal
computer produced for the past ten days. Does the process appear to be in
control?

fraction
non
da units non- p
conformi 1- p stdev LCL UCL
y inspected conformi
ng units
ng
-
1 0.0 0.0265 0.0196 0.1396
80 4 0.05 6 0.94 52 6 56
-
2 0.0 0.0226 0.0079 0.1279
110 7 0.064 6 0.94 43 3 3
 0.0 0.0250 -
3
90 5 0.056 6 0.94 33 0.0151 0.1351
-
4 0.0 0.0274 0.0222 0.1422
75 8 0.107 6 0.94 23 7 68
-
5 0.0 0.0208 0.0024 0.1224
130 6 0.046 6 0.94 29 9 87
-
6 0.0 0.0216 0.0050 0.1250
120 6 0.05 6 0.94 79 4 38
-
7 0.0 0.0283 0.0251 0.1451
70 4 0.057 6 0.94 85 6 55
-
8 0.0 0.0212 0.0037 0.1237
125 5 0.04 6 0.94 41 2 24
-
9 0.0 0.0231 0.0095 0.1295
105 8 0.076 6 0.94 76 3 29
0.0 0.0243 - 0.1330
10
95 7 0.074 6 0.94 66 0.0131 97
Total : Total: 0.1323
1000 Total: 60 0.62 98
Because UCL and LCL can’t be negative, then the value of LCL = 0. From the
calculation above, we can se from the p-chart below that there’s no indication of
out-of-control condition, therefore the process appear to be in control.
 P Char t of nonconfor mi ng uni t
0.16

0.14
UCL=0.1331
0.12

0.10

Pr oport ion
0.08
_
0.06 P=0.06

0.04

0.02

0.00 LCL=0

1 2 3 4 5 6 7 8 9 10
Sample
Tests performed with unequal sample sizes

7.7. A control chart indicates that the current process fraction nonconforming is 0.02.
If fifty items are inspected each day, what is the probability of defecting a shift in
the fraction nonconforming to 0.04 on the first day after the shift? By the end of
the third day following the shift?

From the problem above it is known that p1 = 0.02 , n=50, p 2 = 0.04

n
P{D=x} =   p (1 − p )
x n− x

x 

7.13.

(a) set up a control chart for the number nonconforming in samples of n=100

sampl sample
e number fraction of
numb sampl nonconformi nonconformin
er e size ng g
1 100 10 0.1
2 100 15 0.15
3 100 31 0.31
4 100 18 0.18
5 100 24 0.24
6 100 12 0.12
7 100 23 0.23
8 100 15 0.15
9 100 8 0.08
10 100 8 0.08
Total: 164 p = 0.164
  The np Control Chart:

p = 0.164, n=100

= n p + 3 np (1 − p ) = 100(0.164) +3 100 x0.164(1 − 0.164) = 27.508

UCL
~28

Center line = n p = 100 x 0.164=16.4

LCL= n p - 3 np (1 − p ) = 100(0.164) -3 100 x0.164(1 − 0.164) = 5.292 ~ 5

(b) For the chart established in part (a), what is the probability of detecting a
shift in the process fraction nonconforming to 0.30 on the first sample
after the shift has occured?

δ= the magnitude of the process shift = 0.3 – 0.164 =0.136

L= 3 (if it is assumed that three-sigma limits are used)

n = 100

7.19 A fraction nonconforming control chart has center line 0.01, UCL=0.0399,
LCL=0, and n=100. If three-sigma limits are used, find the smallest sample size
that would yield a positive lower control limit.

(1 − p ) L2
The sample size must be n >
p

n>
(1 − 0.01) (3) 2 = 891
0.01

Thus, if n ≥ 892 units, the control chart will have a positive lower control limit.

7.25 A fraction nonconforming control chart is to be established with a center line of

0.01 and two-sigma control limits.

(a) How large should the sample size be if the lower control limit is to be nonzero?

(1 − p ) L2
The sample size must be n >
p

n>
(1 − 0.01) (2) 2 = 396
0.01

Thus, if n ≥ 397 units, the control chart will have a positive lower control limit.
 (b) how large should the sample size be if we wish the probability of detecting a
shift to 0.04 to be 0.5?

2 2
p (1 − p ) L  3 
δ =L → n=   p (1 − p ) =   0.04(0.96) = 1.63 ≈2
n δ   0.46 

7.31.Consider an np chart with k-sigma control limits. Derive a general formula for
determining the minimum sample size to ensure that the chart has a positive
lower control limit.

Based on the LCL for fraction nonconforming, the minimum sample size to ensure
that the chart has a positive lower control limit can be obtained:

p (1 − p )
LCL = p - L >0
n

n>
(1 − p ) L2
, therefore the LCL for np chart should be:
p

LCL = np - L np (1 − p ) > 0

n>
(1 − p ) L2
p

7.37 Consider the data in Exercise 7.35. Suppose a new inspection unit is defined as
2500 m of wire.

Based on exercise 7.35 we can get the information below:

sampl
e number of number of
numb nonconformit sample nonconformit
er ies number ies
1 1 12 6
2 1 13 9
3 3 14 11
4 7 15 15
5 8 16 8
6 10 17 3
7 5 18 6
8 13 19 7
9 0 20 4
10 19 21 9
11 24 22 20
Total: Total : 189
 22

189
c= = 8.59
22

C Char t of C1
25 1

1
20 1

UCL=17.38
Sample Count

15

10 _
C=8.59

0 LCL=0

1 3 5 7 9 11 13 15 17 19 21
Sample

(a) What are the center line and control limits for a control chart for monitoring
future production based on the total number of nonconformities in the new
inspection unit?

The center line = c = 8.59

UCL = c +3 c = 8.59 +3 8.59 =17.38

LCL = c - 3 c = 8.59 -3 8.59 = - 0.2, because these calculations yield a

negative value for the LCL, then set LCL=0.

(b) What are the center line and control limits for a control chart for average
nonconformities per unit used to monitor future production?

sampl
number of
e sample averag
nonconformiti
numb size e
es
er
0.000
1 1 2500 4
0.000
2 1 2500 4
0.001
3 3 2500 2
0.002
4 7 2500 8
 0.003
5 8 2500 2
6 10 2500 0.004
7 5 2500 0.002
0.005
8 13 2500 2
9 0 2500 0
0.007
10 19 2500 6
0.009
11 24 2500 6
0.002
12 6 2500 4
0.003
13 9 2500 6
0.004
14 11 2500 4
15 15 2500 0.006
0.003
16 8 2500 2
0.001
17 3 2500 2
0.002
18 6 2500 4
0.002
19 7 2500 8
0.001
20 4 2500 6
0.003
21 9 2500 6
22 20 2500 0.008
0.075
189 6
20

u =∑
u i 0.0756
i =1 = =0.0034, therefore, the parameters for the
22
22
control chart:

The center line = u =0.0034

u 0.0034
UCL = u +3 = 0.0034 +3 =0.0069
n 2500

u 0.0034
LCL = u -3 = 0.0034 -3 ≈0
n 2500
7.43. Find 0.900 and 0.100 probability limits for a c chart when the process average is
equal to sixteen nonconformities.

7.49. A textile mill wishes to establish a control procedure on flaws in towels it

manufactures. Using an inspection unit of 50 units, past inspection data show that
100 previous inspection units had 850 total flaws. What type of control chart is
appropriate? Design the control chart such that it has two-sided probability
control limits of α=0.06, approximately. Give the center line and control limits.

The appropriate control chart for this case is control charts for nonconformities
because each product contain more than one nonconformities. We can use the u
Chart

From the problem above, it is known that:

n=50, then the past inspection data shows 100 sample number, number of
nonconformities= 850, α=0.06→β= 1-0.06 = 0.94

850
u= =8.5
100

β = P{ x < UCL u} − P{ x ≤ LCL u}

7.55. A production line assembles electric clocks. The average number of

nonconformities per clock is estimated to be 0.75. The quality engineer wishes to
establish a c chart for this operation, using an inspection unit of six clocks. Find
the three-sigma limits for this chart.

It is known that c =0.75

Inspection unit= 6

UCL = c +3 c = 0.75+3 0.75 = 3.348

LCL = c - 3 c = 0.75-3 0.75 =-1.848, because these calculations yield a

negative value for the LCL, then set LCL=0.

7.61.(a) Set up a c chart for the total number of errors. Is the process in control?

Number of
record record record record record
day nonconformit
1 2 3 4 5
ies
1 8 7 1 11 17 44
2 11 1 11 2 9 34
3 1 1 8 2 5 17
4 3 2 5 1 4 15
5 3 2 13 6 5 29
 6 6 3 3 3 1 16
7 8 8 2 1 5 24
8 4 10 2 6 4 26
9 1 6 1 3 2 13
10 15 1 3 2 8 29
11 1 7 13 5 1 27
12 6 7 9 3 1 26
13 7 6 3 3 1 20
14 2 9 3 8 7 29
15 6 14 7 1 8 36
16 2 9 4 2 1 18
17 11 1 1 3 2 18
18 5 5 19 1 3 33
19 6 15 5 6 6 38
20 2 7 9 2 8 28
21 7 5 6 14 10 42
22 4 3 8 1 2 18
23 4 1 4 20 5 34
24 15 2 7 10 17 51
25 2 15 3 11 2 33
Total : 698
698
c= =27.92
25

UCL = c +3 c = 27.92 +3 27.92 = 43.77

LCL = c - 3 c = 27.92 -3 27.92 = 12.068

C Char t of C1
1
50

1
UCL=43.77
40
Sample Count

30 _
C=27.92

20

LCL=12.07
10
1 3 5 7 9 11 13 15 17 19 21 23 25
Sample
 From the c chart above, it is indicate that the process is out of control because
sample 24 is plotted outside the UCL.

(c)Set up a t chart for the total number of errors, assuming a geometric

distribution with a=1. Is the process in control?

(d) discuss the findings from parts (a) and (b). Is the poisson distribution a good
model for the customer error data? Is there evidence of this in the data?

9.1. A machine is used to fill cans with motor oil additive. A single sample can is
selected every hour and the wight of the can is obtained. Since the filling process
is automated, it has very stable variability, and long experience indicates that
σ=0.05 oz. The individual observations for 24 hours of operation are shown in
table 9E.1

(a) Assuming that the process target is 8.02 oz, set up tabular cusum for this
process. Design the cusum using the standardized values h=4.77 and k=1/2.

µ0= 8.02, n= 1, σ=0.05, h=4.77 , k=1/2

[ ]
C i+ =max 0, xi − ( µ 0 + K ) + C i+−1 , with C 0+ =0

[ ]
C i− =max 0, ( µ 0 − K ) − xi + C i−−1 , with C 0− =0

Calculation (for i=1, for the rest is ilustrated from the table):

[
C1+ =max 0, x1 − ( µ 0 + K ) + C 0+ ]
= max [ 0, 8 − ( 8.02 + 0.5) + 0]

= max [ 0, − 0.52]

=0

[
C1− =max 0, ( µ 0 − K ) − x1 + C 0− ]
= max [ 0, ( 8.02 − 0.5) − 8 + 0]

= max [ 0, − 0.48]

=0

sample a b
numbe x
r xi- Ci N 7.52- Ci N
8.52 + + xi - -
1 8 -0.52 0 0 -0.48 0 0
 8.0
2
1 -0.51 0 0 -0.49 0 0
8.0
3
2 -0.5 0 0 -0.5 0 0
8.0
4
1 -0.51 0 0 -0.49 0 0
5 8 -0.52 0 0 -0.48 0 0
8.0
6
1 -0.51 0 0 -0.49 0 0
8.0
7
6 -0.46 0 0 -0.54 0 0
8.0
8
7 -0.45 0 0 -0.55 0 0
8.0
9
1 -0.51 0 0 -0.49 0 0
8.0
10
4 -0.48 0 0 -0.52 0 0
8.0
11
2 -0.5 0 0 -0.5 0 0
8.0
12
1 -0.51 0 0 -0.49 0 0
8.0
13
5 -0.47 0 0 -0.53 0 0
8.0
14
4 -0.48 0 0 -0.52 0 0
8.0
15
3 -0.49 0 0 -0.51 0 0
8.0
16
5 -0.47 0 0 -0.53 0 0
8.0
17
6 -0.46 0 0 -0.54 0 0
8.0
18
4 -0.48 0 0 -0.52 0 0
8.0
19
5 -0.47 0 0 -0.53 0 0
8.0
20
6 -0.46 0 0 -0.54 0 0
8.0
21
4 -0.48 0 0 -0.52 0 0
8.0
22
2 -0.5 0 0 -0.5 0 0
8.0
23
3 -0.49 0 0 -0.51 0 0
8.0
24
5 -0.47 0 0 -0.53 0 0

(b) Does the value of σ=0.05 seem reasonable for this process?

If σ=0.05 , then H= 5(0.05) = 0.25 It seems to be reasonable because there’s

no out-of-control process.
9.7. Set up a tabular cusum scheme for the flow width data used in example 6.1 (see
Tables 6.1 and 6.2). when the procedure is applied to all 45 samples, does the
cusum react more quickly than the x chart to the shift in the process mean? Use
σ=0.14 in setting up the cusum, and design the procedure to quickly detect a
shift of about 1σ.

sampl wafers
e
numb xbar=
er 1 2 3 4 5 xi
1.323 1.412 1.674 1.457 1.691 1.511
1 5 8 4 3 4 9
1.431 1.359 1.607 1.466 1.610 1.495
2 4 2 5 6 9 1
1.428 1.487 1.493 1.432 1.567 1.481
3 4 1 2 4 4 7
1.502 1.635 1.384 1.283 1.550 1.471
4 8 2 1 1 7 2
1.560 1.273 1.526 1.436 1.644 1.488
5 4 5 5 3 1 2
1.595 1.545 1.357 1.328 1.419 1.449
6 5 1 4 1 8 2
1.627 1.506 1.836 1.417 1.514 1.580
7 4 4 6 7 4 5
1.430 1.663 1.606 1.551 1.534
8 1.419 3 7 7 9 3
1.388 1.727 1.535 1.517 1.368 1.507
9 4 7 5 6 8 6
1.403 1.669 1.508 1.462 1.513
10 9 7 9 7 1.522 4
1.415 1.766 1.427 1.592 1.418 1.524
11 8 7 8 8 1 2
1.582 1.335 1.577 1.390 1.755 1.528
12 1 5 7 8 9 4
1.285 1.410 1.444 1.639 1.192 1.394
13 6 6 7 8 8 7
1.495 1.403 1.589 1.645 1.496 1.526
14 1 6 3 8 9 1
1.358 1.286 1.599 1.249 1.547 1.408
15 9 3 6 7 1 3
1.574 1.530 1.517 1.183 1.866 1.534
16 7 1 1 9 2 4
1.726 1.395 1.501 1.444 1.487
17 1.368 9 7 4 9 4
1.416 1.386 1.305 1.557 1.457
18 3 4 7 1.621 3 3
1.579 1.418 1.654 1.511 1.724 1.577
19 6 5 1 6 7 7
20 1.710 1.441 1.236 1.382 1.760 1.506
 6 2 1 1 0
1.437 1.505 1.348
21 1 1 5 1.567 1.488 1.4691
1.473 1.593 1.658 1.497 1.539
22 8 6 3 3 1.472 0
1.591 1.433 1.555 1.529 1.686 1.559
23 7 3 1 5 6 2
1.639 1.524 1.570 1.556 1.568
24 9 3 5 3 1.553 8
1.579 1.366 1.373 1.688 1.526
25 7 3 1.624 2 7 4
1.448 1.545 1.453 1.430 1.620 1.499
26 3 8 8 3 6 8
1.543 1.689 1.335 1.418 1.514
27 5 9 1.583 8 7 2
1.517 1.344 1.472 1.665 1.666 1.533
28 5 6 3 7 1 2
1.545 1.093 1.407 1.503 1.526 1.415
29 4 1 2 9 4 2
1.441 1.505 1.512 1.626 1.509
30 8 9 4 1.462 3 7
1.430 1.272 1.594 1.539 1.525 1.472
31 1 5 5 7 2 4
1.498 1.450 1.617 1.583 1.496 1.529
32 1 6 4 7 2 2
1.300 1.623 1.583 1.645 1.531
33 9 1.506 1 1 4 7
1.413 1.460 1.580 1.711 1.731 1.579
34 2 3 8 1 3 3
1.381 1.313 1.495 1.489 1.459 1.427
35 7 5 3 4 6 9
1.576 1.701 1.402 1.277 1.454 1.482
36 5 4 6 3 1 4
1.493 1.437 1.513 1.480 1.529 1.491
37 6 3 9 8 3 0
1.572 1.673 1.504 1.565 1.747
38 9 8 8 1 3 1.6128
1.808 1.551 1.438 1.655 1.656
39 9 3 1.825 9 8 0
1.623 1.539 1.673 1.869 1.503 1.642
40 6 3 8 8 6 0
1.793 1.734 1.639 1.779
41 1.412 1 5 1 1 1.6716
1.737 1.566 1.780 1.550 1.625
42 2 3 1.491 9 4 2
1.597 1.739 1.683 1.667 1.797 1.697
43 1 4 2 7 4 0
1.429 1.653 1.913 1.727
44 5 6 4 2 1.437 1.6321
 1.621 1.791 1.674 1.940 1.770
45 7 1.822 5 4 4 0

9.13.Consider the velocity of light data introduced in exercises 6.56 and 6.55. use
only the 20 obseervations in Exercise 6.56 to set up a cusum with target value
734.5. plot all 40 observations from both exercises 6.56 and 6.55 on this cusum.
What conclusions can you draw?
Cusum with target value 734.5 for 20 observations:

a b
sample
x N Ci N
number xi-735 Ci+ 734-xi
+ - -
1 850 115 115 1 -116 0 0
100
2
0 265 380 2 -266 0 0
3 740 5 385 3 -6 0 0
4 980 245 630 4 -246 0 0
5 900 165 795 5 -166 0 0
6 930 195 990 6 -196 0 0
107 132
7
0 335 5 7 -336 0 0
124 8
8
650 -85 0 8 84 4 1
143
9
930 195 5 9 -196 0 0
146 1
10
760 25 0 0 -26 0 0
157 1
11
850 115 5 1 -116 0 0
165 1
12
810 75 0 2 -76 0 0
186 1
13
950 215 5 3 -216 0 0
100 213 1
14
0 265 0 4 -266 0 0
237 1
15
980 245 5 5 -246 0 0
100 264 1
16
0 265 0 6 -266 0 0
288 1
17
980 245 5 7 -246 0 0
311 1
18
960 225 0 8 -226 0 0
325 1
19
880 145 5 9 -146 0 0
348 2
20
960 225 0 0 -226 0 0
 Plot of Cusum with target value 734.5 for all observations:

CUS UM Char t of C1

4000

3000
Cumulat iv e Sum

2000

1000

UCL=359
0 0
LCL=-359

1 5 9 13 17 21 25 29 33 37
Sample

It indicates that the process is out of control.

9.19.Rework Exercise 9.4 using an EWMA control chart with λ= 0.1 and L=2.7.
Compare your results to those obtained within the cusum.

Based on the exercise 9.4, the tabular cusum for target value=1050 and σ=25 is
described below:

sampl
a b
e
x
numb N N
er xi-1062.5 Ci+ 1037.5-xi Ci-
+ -
1 1045 -17.5 0 0 -7.5 0 0
2 1055 -7.5 0 0 -17.5 0 0
3 1037 -25.5 0 0 0.5 0.5 1
4 1064 1.5 1.5 1 -26.5 0 0
5 1095 32.5 34 2 -57.5 0 0
6 1008 -54.5 0 0 29.5 29.5 1
7 1050 -12.5 0 0 -12.5 17 2
8 1087 24.5 24.5 1 -49.5 0 0
9 1125 62.5 87 2 -87.5 0 0
10 1146 83.5 170.5 3 -108.5 0 0
11 1139 76.5 247 4 -101.5 0 0
12 169 -893.5 0 0 868.5 868.5 1
13 1151 88.5 88.5 1 -113.5 755 2
14 1128 65.5 154 2 -90.5 664.5 3
 15 1238 175.5 329.5 3 -200.5 464 4
16 1125 62.5 392 4 -87.5 376.5 5
17 1163 100.5 492.5 5 -125.5 251 6
18 1188 125.5 618 6 -150.5 100.5 7
19 1146 83.5 701.5 7 -108.5 0 0
20 1167 104.5 806 8 -129.5 0 0

The calculations for EWMA control chart with λ = 0.1 and L=2.7 are summarized
in the table below:

To ilustrate the calculation, consider the first observation is:

UCL= µ 0 + Lσ
λ
(2 − λ)
[ ]
1 − (1 − λ ) = 1050 + 2.7(25)
2i 0.1
( 2 − 0.1)
[ ]
1 − (1 − 0.1) =1056.75
2

Center line = µ0 =1050

LCL = µ 0 − Lσ
λ
(2 − λ)
[ ]
1 − (1 − λ ) =1050 − 2.7(25)
2i 0.1
[
( 2 − 0.1)
]
1 − (1 − 0.1) =1043.25
2

z1 = λx1 +(1- λ)z0 = 0.1(1045)+0.9(1050) = 1049.5

sampl
sample e
x EWMA x EWMA
number numb
er
1058.
1 11
1045 1049.5 1139 9
2 1055 1050.5 12 169 961.9
3 1037 1048.7 13 1151 1060.1
1057.
4 14
1064 1051.4 1128 8
1068.
5 15
1095 1054.5 1238 8
1057.
6 16
1008 1045.8 1125 5
7 1050 1050 17 1163 1061.3
1063.
8 18
1087 1053.7 1188 8
1059.
9 19
1125 1057.5 1146 6
10 1146 1059.6 20 1167 1061.7
 EWM A Char t of C1
1100

UCL=1083.4
1080
_
_
X=1066.3
1060

LCL=1049.2
EWMA

1040

1020

1000

1 3 5 7 9 11 13 15 17 19
Sample

9.25.Derive the variance of the exponentially weighted moving average zi

S i2 = λ ( xi − z i ) 2 + (1 − λ ) S i2−1

9.31.An EWMA control chart uses λ=0.4. How wide will the limit be on the Shewhart
control chart, expressed as a multiple of width of the steady-state EWMA limits?

From the control limit of EWMA control chart for steady-state value, then the wide is:

λ 0.4 0.4 1
K = Lσ = = =
(2 − λ ) (2 − 0.4) 1.6 2

10.1.Discuss how you would use a cusum in the short production run solution. What
advantages would it have relative to a Shewhart chart, such as a DNOM version of
the x chart?
10.7.Reconsider the data in Exercise 10.4 and 10.6. Suppose the process
measurement are individual data values, not subgroup averages.

sampl head
e 1 2 3 4
numb xb xb xb xb
er ar R ar R ar R ar R
1 53 2 54 1 56 2 55 3
2 51 1 55 2 54 4 54 4
3 54 2 52 5 53 3 57 2
4 55 3 54 3 52 1 51 5
5 54 1 50 2 51 2 53 1
6 53 2 51 1 54 2 52 2
7 51 1 53 2 58 5 54 1
8 52 2 54 4 51 2 55 2
9 50 2 52 3 52 1 51 3
10 51 1 55 1 53 3 53 5
11 52 3 57 2 52 4 55 1
12 51 2 55 1 54 2 58 2
13 54 4 58 2 51 1 53 1
14 53 1 54 4 50 3 54 2
15 55 2 52 3 54 2 52 6
16 54 4 51 1 53 2 58 5
17 53 3 50 2 57 1 53 1
18 52 1 49 1 52 1 49 2
19 51 2 53 3 51 2 50 3
20 52 4 52 2 50 3 52 2

(a) Use observations 1-20 in Exercise 10.4 to construct appropriate group of

control charts.

• x chart:

CL= 52.988

MR 2.158
UCL = x + 3 = 52.988+ x + 3 = 58.727
d2 1.128

MR 2.158
LCL= x − 3 = 52.988- 3 = 47.248
d2 1.128

• R chart:

CL=2.158
 UCL = D4MR= 3.267x2.158= 7.05

LCL=D3MR = 0 x 2.158 = 0

s m 4 20

x=
∑
j =1
∑ xij
i =1 =
∑ ∑x
j =1 i =1
ij
=52.988
m( s ) 20(4)

s 20

MR=
∑ ∑ MR
j =1 i=2
ij
= 2.158
(m − 1) s

(b) Plot observations 21-30 from Exercise 10.6 on the charts from part (a).
Discuss your findings.

head
1 2 3 4
sample x x x x
number bar R bar R bar R bar R
21 50 3 54 1 57 2 55 5
22 51 1 53 2 54 4 54 3
23 53 2 52 4 55 3 57 1
24 54 4 54 3 53 1 56 2
25 50 2 51 1 52 2 58 4
26 51 2 55 5 54 5 54 3
27 53 1 50 2 51 4 60 1
28 54 3 51 4 54 3 61 4
29 52 2 52 1 53 2 62 3
30 52 1 53 3 50 4 60 1

(c) Using observations 1-20, construct an individual chart using the average of
the readings on all four heads as an individual measurement and an s control
chart using the individual measurements on each head. Discuss how these
charts function relative to the group control chart.

10.13. Specification on a bearing diameter are established at 8.0 ± 0.01 cm. Sample
of size n=8 are used, and a control chart for s shows statistical control, with the
best current estimate of the population standard deviation S=0.001. If the
fraction of nonconforming product that is barely acceptable is 0.135%, find the
three-sigma limits on the modified control chart for this process.

 3 
UCL=USL-  Z δ − σ
 n
 USL= 8.01 and LSL = 7.99

Z0.00135= 3

n= 8

σ= 0.001

 3 
UCL=8.01-  3 − 0.001 = 8.008061
 8

 3 
LCL= 7.99+  3 − 0.001 = 7.991939
 8

10.19. Set up a moving center line EWMA control chart for the concentration data in
Exercise 10.16. Compare it to the residuals control chart in Exercise 10.16 part
(c).

α =0.1, λ=0.7055, =3.227. observations 8, 56 and 90 exceed control limits

10.25.

(a) Discuss the use of the moving range method to estimate the process standard
deviation when the data are positively autocorrelated.

(b)Discuss the use of the sample variance s2 with positively autocorrelated data.
Specifically, if the observations at lag ρi, is s2 still an unbiased estimator for σ2?

(c) Does your answer in part (b) imply that s2 would really be a good way (in
practice) to estimate σ2 in constructing a control chart for autocorrelated data?

11.1. A product has three quality characteristics. The nominal values of these quality
characteristics and their sample covariance matrix have been determined from
the analyzis of 30 preliminary samples of size n=10 as follows:

3.0  1.40 1.02 1.05 

x = 3.5  S= 1.02 1.35 0.98
 
2.8 1.05 0.98 1.20

The sample means for each quality characteristic for 15 additional samples of size
n=10 are shown in table 11E.1. Is the process in statistical control?

sampl
e
x1 x2 x3
numb
er
1 3.1 3.7 3
 2 3.3 3.9 3.1
3 2.6 3 2.4
4 2.8 3 2.5
5 3 3.3 2.8
6 4 4.6 3.5
7 3.8 4.2 3
8 3 3.3 2.7
9 2.4 3 2.2
10 2 2.6 1.8
11 3.2 3.9 3
1
2 3.7 4 3
13 4.1 4.7 3.2
14 3.8 4 2.9
15 3.2 3.6 2.8

11.7. rework exercise 11.6, assuming that the subgroup size is n=5

(a) find the phase II control limits assuming that α=0.005.

p (m + 1)(n − 1) 6(30 + 1)(5 − 1)

UCL= F= (3.294) =21.3
nm − m − p + 1 5(30) − 30 − 6 + 1

(b) Compare the control limits from part (a) to the chi-square control limit. What is
the magnitude of the difference in the two control limits?

Chi-square control limit = χ 2 α , p = χ 2 0.005;6, =18.548

(c) how many preliminary samples would have to be taken to ensure that the
exact phase II control limit is within 1% of the chi-square control limit?

11.13. consider all 30 observations on the first two process variables in Table 11.6.
Calculate an estimate of the sample covariance matrix using both estimators S1
and S2 discussed in Section 11.3.2. Are the estimates very different? Discuss
your findings.

11.19. Consider the p=9 process variables in Table 11.5.

(a) perform a PCA on the first 30 observations. Be sure to work with the
standardized variables.

(b) how much variability is explained if only the first r=3 principal components are
retained?
(c) Construct an appropriate set of pairwise plot of the first r=3 principal
component scores.

(d) Now consider the last 10 observations. Obtain the principal component scores
and plot them on the chart in part (c). Does the process seem to be in control?