# Chapter 2 Exercise Solutions

Several exercises in this chapter differ from those in the 4th edition. An “*” following the
exercise number indicates that the description has changed (e.g., new values). A second
exercise number in parentheses indicates that the exercise number has changed. For
example, “2-16* (2-9)” means that exercise 2-16 was 2-9 in the 4th edition, and that the
description also differs from the 4th edition (in this case, asking for a time series plot
instead of a digidot plot). New exercises are denoted with an “☺”.
2-1*.
(a)
x = ∑ xi n = (16.05 + 16.03 +
n

i =1

(b)
n

( )
n

∑ x − ∑ xi

i =1

s=

2
i

+ 16.07 ) 12 = 16.029 oz

2

n

i =1

n −1

=

(16.052 +

+ 16.07 2 ) − (16.05 +
12 − 1

+ 16.07) 2 12

= 0.0202 oz

MTB > Stat > Basic Statistics > Display Descriptive Statistics
Descriptive Statistics: Ex2-1
Variable
Ex2-1
Variable
Ex2-1

N N*
Mean
12
0 16.029
Maximum
16.070

SE Mean
0.00583

StDev
0.0202

Minimum
16.000

Q1
16.013

Median
16.025

Q3
16.048

2-2.
(a)
x = ∑ xi n = ( 50.001 + 49.998 +
n

i =1

(b)
n

s=

( )
n

2
∑ xi − ∑ xi

i =1

i =1

n −1

+ 50.004 ) 8 = 50.002 mm

2

n

=

(50.0012 +

+ 50.0042 ) − (50.001 +
8 −1

+ 50.004) 2 8

= 0.003 mm

MTB > Stat > Basic Statistics > Display Descriptive Statistics
Descriptive Statistics: Ex2-2
Variable
Ex2-2
Variable
Ex2-2

N N*
Mean
8
0 50.002
Maximum
50.006

SE Mean
0.00122

StDev
0.00344

Minimum
49.996

Q1
49.999

Median
50.003

Q3
50.005

2-1

Chapter 2 Exercise Solutions
2-3.
(a)
x = ∑ xi n = ( 953 + 955 +
n

i =1

(b)
n

s=

( )
n

∑ x − ∑ xi

i =1

2
i

i =1

n −1

+ 959 ) 9 = 952.9 °F

2

n

=

(9532 +

+ 9592 ) − (953 +
9 −1

+ 959) 2 9

= 3.7 °F

MTB > Stat > Basic Statistics > Display Descriptive Statistics
Descriptive Statistics: Ex2-3
Variable
Ex2-3
Variable
Ex2-3

N N*
Mean
9
0 952.89
Maximum
959.00

SE Mean
1.24

StDev
3.72

Minimum
948.00

Q1
949.50

Median
953.00

Q3
956.00

2-4.
(a)
In ranked order, the data are {948, 949, 950, 951, 953, 954, 955, 957, 959}. The sample
median is the middle value.
(b)
Since the median is the value dividing the ranked sample observations in half, it remains
the same regardless of the size of the largest measurement.

2-5.
MTB > Stat > Basic Statistics > Display Descriptive Statistics
Descriptive Statistics: Ex2-5
Variable
Ex2-5
Variable
Ex2-5

N N*
Mean
8
0 121.25
Maximum
156.00

SE Mean
8.00

StDev
22.63

Minimum
96.00

Q1
102.50

Median
117.00

Q3
144.50

2-2

Chapter 2 Exercise Solutions
2-6.
(a), (d)
MTB > Stat > Basic Statistics > Display Descriptive Statistics
Descriptive Statistics: Ex2-6
Variable
Ex2-6
Variable
Ex2-6

N N*
Mean
40
0 129.98
Maximum
160.00

SE Mean
1.41

StDev
8.91

Minimum
118.00

Q1
124.00

Median
128.00

Q3
135.25

(b)
Use √n = √40 ≅ 7 bins
MTB > Graph > Histogram > Simple
Histogram of Time to Failure (Ex2-6)
20

Frequency

15

10

5

0

112

120

128

136
Hours

144

152

160

(c)
MTB > Graph > Stem-and-Leaf
Stem-and-Leaf Display: Ex2-6
Stem-and-leaf of Ex2-6
Leaf Unit = 1.0
2
11 89
5
12 011
8
12 233
17 12 444455555
19 12 67
(5) 12 88999
16 13 0111
12 13 33
10 13
10 13 677
7
13
7
14 001
4
14 22
HI 151, 160

N

= 40

2-3

Chapter 2 Exercise Solutions
2-7.
Use

n = 90 ≅ 9 bins

MTB > Graph > Histogram > Simple

Histogram of Process Yield (Ex2-7)
18
16
14

Frequency

12
10
8
6
4
2
0

84

88

92

96

Yield

2-4

Chapter 2 Exercise Solutions
2-8.
(a)
Stem-and-Leaf Plot
2
12o|68
6
13*|3134
12
13o|776978
28
14*|3133101332423404
(15)
14o|585669589889695
37
15*|3324223422112232
21
15o|568987666
12
16*|144011
6
16o|85996
1
17*|0
Stem Freq|Leaf

(b)
Use

n = 80 ≅ 9 bins

MTB > Graph > Histogram > Simple
Histogram of Viscosity Data (Ex 2-8)
20

Frequency

15

10

5

0

13

14

15
Viscosity

16

17

Note that the histogram has 10 bins. The number of bins can be changed by editing the
X scale. However, if 9 bins are specified, MINITAB generates an 8-bin histogram.
Constructing a 9-bin histogram requires manual specification of the bin cut points.
Recall that this formula is an approximation, and therefore either 8 or 10 bins should
suffice for assessing the distribution of the data.

2-5

Chapter 2 Exercise Solutions
2-8(c) continued
MTB > %hbins 12.5 17 .5 c7
Row
1
2
3
4
5
6
7
8
9
10
11

Intervals
12.25 to 12.75
12.75 to 13.25
13.25 to 13.75
13.75 to 14.25
14.25 to 14.75
14.75 to 15.25
15.25 to 15.75
15.75 to 16.25
16.25 to 16.75
16.75 to 17.25
Totals

Frequencies
1
2
7
9
16
18
12
7
4
4
80

Percents
1.25
2.50
8.75
11.25
20.00
22.50
15.00
8.75
5.00
5.00
100.00

(d)
MTB > Graph > Stem-and-Leaf
Stem-and-Leaf Display: Ex2-8
Stem-and-leaf of Ex2-8 N = 80
Leaf Unit = 0.10
2
12 68
6
13 1334
12
13 677789
28
14 0011122333333444
(15) 14 555566688889999
37
15 1122222222333344
21
15 566667889
12
16 011144
6
16 56899
1
17 0

median observation rank is (0.5)(80) + 0.5 = 40.5
x0.50 = (14.9 + 14.9)/2 = 14.9
Q1 observation rank is (0.25)(80) + 0.5 = 20.5
Q1 = (14.3 + 14.3)/2 = 14.3
Q3 observation rank is (0.75)(80) + 0.5 = 60.5
Q3 = (15.6 + 15.5)/2 = 15.55
(d)
10th percentile observation rank = (0.10)(80) + 0.5 = 8.5
x0.10 = (13.7 + 13.7)/2 = 13.7
90th percentile observation rank is (0.90)(80) + 0.5 = 72.5
x0.90 = (16.4 + 16.1)/2 = 16.25

2-6

Chapter 2 Exercise Solutions
2-9 ☺.
MTB > Graph > Probability Plot > Single
Probability Plot of Liquid Detergent (Ex2-1)
Normal
99

Mean
StDev
N
P-Value

95
90

16.03
0.02021
12
0.297
0.532

Percent

80
70
60
50
40
30
20
10
5

1

15.98

16.00

16.02
16.04
Fluid Ounces

16.06

16.08

When plotted on a normal probability plot, the data points tend to fall along a straight
line, indicating that a normal distribution adequately describes the volume of detergent.
2-10 ☺.
MTB > Graph > Probability Plot > Single
Probability Plot of Furnace Temperatures (Ex2-3)
Normal
99

Mean
StDev
N
P-Value

95
90

952.9
3.723
9
0.166
0.908

Percent

80
70
60
50
40
30
20
10
5

1

945.0

947.5

950.0 952.5
955.0
957.5
Temperature (deg F)

960.0

962.5

When plotted on a normal probability plot, the data points tend to fall along a straight
line, indicating that a normal distribution adequately describes the furnace temperatures.

2-7

Chapter 2 Exercise Solutions
2-11 ☺.
MTB > Graph > Probability Plot > Single
Probability Plot of Failure Times (Ex2-6)
Normal
99

Mean
StDev
N
P-Value

95
90

130.0
8.914
40
1.259
<0.005

Percent

80
70
60
50
40
30
20
10
5

1

110

120

130

140

150

160

Hours

When plotted on a normal probability plot, the data points do not fall along a straight line,
indicating that the normal distribution does not reasonably describe the failure times.
2-12 ☺.
MTB > Graph > Probability Plot > Single
Probability Plot of Process Yield Data (Ex2-7)
Normal
99.9

Mean
StDev
N
P-Value

99
95

Percent

90

89.48
4.158
90
0.956
0.015

80
70
60
50
40
30
20
10
5
1
0.1

80

85

90
Yield

95

100

105

When plotted on a normal probability plot, the data points do not fall along a straight line,
indicating that the normal distribution does not reasonably describe process yield.

2-8

Chapter 2 Exercise Solutions
2-13 ☺.
MTB > Graph > Probability Plot > Single

(In the dialog box, select Distribution to choose the distributions)
Probability Plot of Viscosity Data (Ex2-8)
Normal
99.9

Mean
StDev
N
P-Value

99

Percent

95
90

14.90
0.9804
80
0.249
0.740

80
70
60
50
40
30
20
10
5
1
0.1

12

13

14

15
Viscosity

16

17

18

Probability Plot of Viscosity Data (Ex2-8)
Lognormal
99.9

Loc
Scale
N
P-Value

99

Percent

95
90

2.699
0.06595
80
0.216
0.841

80
70
60
50
40
30
20
10
5
1
0.1

12

13

14

15
Viscosity

16

17

18

19

2-9

However.1 10 11 12 13 14 Viscosity 15 16 17 18 Both the normal and lognormal distributions appear to be reasonable models for the data. with no bends or curves.9 99 Shape Scale N AD P-Value 90 80 70 60 50 40 30 20 16. the plot points on the Weibull probability plot are not straight—particularly in the tails— indicating it is not a reasonable model.10 15.010 10 5 3 2 1 0. 2-10 .Chapter 2 Exercise Solutions 2-13 continued Probability Plot of Viscosity Data (Ex2-8) Weibull Percent 99. the plot points tend to fall along a straight line.032 <0.36 80 1.

137 Percent 80 70 60 50 40 30 20 10 5 1 -5000 0 5000 10000 15000 Cycles to Failure 20000 25000 Probability Plot of Cycles to Failure (Ex2-14) Lognormal 99 Loc Scale N AD P-Value 95 90 8.776 0.Chapter 2 Exercise Solutions 2-14 ☺. MTB > Graph > Probability Plot > Single (In the dialog box.549 0.521 0. select Distribution to choose the distributions) Probability Plot of Cycles to Failure (Ex2-14) Normal 99 Mean StDev N AD P-Value 95 90 8700 6157 20 0.8537 20 0.163 Percent 80 70 60 50 40 30 20 10 5 1 1000 10000 Cycles to Failure 100000 2-11 .

336 >0.464 9624 20 0.Chapter 2 Exercise Solutions 2-14 continued Probability Plot of Cycles to Failure (Ex2-14) Weibull 99 Shape Scale N AD P-Value Percent 90 80 70 60 50 40 1.250 30 20 10 5 3 2 1 1000 10000 Cycles to Failure Plotted points do not tend to fall on a straight line on any of the probability plots. though the Weibull distribution appears to best fit the data in the tails. 2-12 .

0 Concentration.873 Percent 80 70 60 50 40 30 20 10 5 1 -50 0 50 Concentration.426 <0.201 0. ppm 100.9347 1. ppm 100 150 Probability Plot of Concentration (Ex2-15) Lognormal 99 95 90 Percent 80 70 60 50 40 30 20 10 5 1 0.651 40 0.470 22.005 Loc Scale N AD P-Value 0.0 2-13 .1 1.56 40 8.0 10.Chapter 2 Exercise Solutions 2-15 ☺. select Distribution to choose the distributions) Probability Plot of Concentration (Ex2-15) Normal 99 95 90 Mean StDev N AD P-Value 9. MTB > Graph > Probability Plot > Single (In the dialog box.

6132 5.000 The lognormal distribution appears to be a reasonable model for the concentration data. 2-14 . ppm 100.782 40 0.Chapter 2 Exercise Solutions 2-15 continued Probability Plot of Concentration (Ex2-15) Weibull 99 Shape Scale N AD P-Value Percent 90 80 70 60 50 40 0.100 1.091 30 20 10 5 3 2 1 0.001 0.010 0. Plotted points on the normal and Weibull probability plots tend to fall off a straight line.637 0.000 10.000 Concentration.

MTB > Graph > Time Series Plot > Single (or Stat > Time Series > Time Series Plot) Time Series Plot of Viscosity Data (Ex2-8) 17 Ex2-8 16 15 14 13 12 1 8 16 24 32 40 48 56 Time Order of Collection 64 72 80 From visual examination. MTB > Graph > Time Series Plot > Single (or Stat > Time Series > Time Series Plot) Time Series Plot of Yield Data (Ex2-7) 100 Ex2-7 95 90 85 1 9 18 27 36 45 54 63 Time Order of Collection 72 81 90 Time may be an important source of variability. indicating that time is not an important source of variability. there are no trends.Chapter 2 Exercise Solutions 2-16* (2-9). shifts or obvious patterns in the data. as evidenced by potentially cyclic behavior. 2-17* (2-10). 2-15 .

2-19 (2-11). MTB > Stat > Basic Statistics > Display Descriptive Statistics Descriptive Statistics: Ex2-7 Variable Ex2-7 Variable Ex2-7 N N* Mean 90 0 89. The order in which the data were collected may be an important source of variability.438 StDev 4.000 SE Mean 0.100 Median 89.158 Minimum 82. MTB > Graph > Time Series Plot > Single (or Stat > Time Series > Time Series Plot) Time Series Plot of Concentration Data (Ex2-15) 140 120 100 Ex2-15 80 60 40 20 0 4 8 12 16 20 24 28 Time Order of Collection 32 36 40 Although most of the readings are between 0 and 20. 35). there are two unusually large readings (9.600 Q1 86.Chapter 2 Exercise Solutions 2-18 ☺.125 2-16 . as well as occasional “spikes” around 20.476 Maximum 98.250 Q3 93.

6.06 Fluid Ounces 16. The data appear to be fairly well scattered. for example.04 16. 4.Chapter 2 Exercise Solutions 2-20 (2-12). The stem-and-leaf plot suggests that certain values may occur more frequently than others.00 2-17 . 2-21 (2-13). MTB > Graph > Boxplot > Simple Boxplot of Detergent Data (Ex2-1) 16. MTB > Graph > Stem-and-Leaf Stem-and-Leaf Display: Ex2-7 Stem-and-leaf of Ex2-7 Leaf Unit = 0.07 16.02 16.01 16.03 16. and 7.10 2 82 69 6 83 0167 14 84 01112569 20 85 011144 30 86 1114444667 38 87 33335667 43 88 22368 (6) 89 114667 41 90 0011345666 31 91 1247 27 92 144 24 93 11227 19 94 11133467 11 95 1236 7 96 1348 3 97 38 1 98 0 N = 90 Neither the stem-and-leaf plot nor the frequency histogram reveals much about an underlying distribution or a central tendency in the data. those ending in 1.05 16.

x = 10 2 / 36. 10. x = 9 3 / 36. x: {the sum of two up dice faces} sample space: {2.. 3. MTB > Graph > Boxplot > Simple Boxplot of Bearing Bore Diameters (Ex2-2) 50.. 4.1} = 1 × 1 + 1 × 1 + 1 × 1 = 3 6 6 6 6 6 6 36 . 12} Pr{x = 2} = Pr{1.9975 49. otherwise 2-24 (2-16). x = ∑ xi p( xi ) = 2 (1 36 ) + 3 ( 2 36 ) + 11 i =1 n ∑ xi p( xi ) − ⎡⎢ ∑ xi p ( xi ) ⎤⎥ ⎣ i =1 ⎦ S = i =1 n −1 n + 12 (1 36 ) = 7 2 n = 5.1} = 1 × 1 + 1 × 1 = 2 6 6 6 6 36 Pr{x = 4} = Pr{1.0000 49. 2} + Pr{2. 8. x = 2 2 / 36.38 10 2-18 .0050 mm 50. x = 8 4 / 36. 6.92 − 7 2 11 = 0.0075 50. 2} + Pr{3. x = 11 1/ 36. 7. 11. 9.1} = 1 × 1 = 1 6 6 36 Pr{x = 3} = Pr{1. 5. x = 6 6 / 36.Chapter 2 Exercise Solutions 2-22 (2-14). x = 12 0. ( ) ( ( ) ) ( ) ( ) ⎧1/ 36.9950 2-23 (2-15). x = 7 p ( x) = ⎨ ⎩5 / 36.3} + Pr{2. x = 4 4 / 36.0025 50. x = 3 3 / 36. x = 5 5 / 36.

05 2-19 . ∫ f ( x)dx must equal unity.5 + 5k ) =1 3 10k = 0. (a) Pr{x = 1} = p(1) = e −0. x = 1 (1 + 2k ) / 3.0100.Chapter 2 Exercise Solutions 2-25 (2-17).02.5 + 5k ) / 3.01 (0.0100 Pr{x ≥ 1} = 1 − Pr{x = 0} = 1 − p(0) = 1 − 0! Cutting the rate at which defects occur reduces the probability of one or more defects by approximately one-half.0198 0! (c) This is a Poisson distribution with parameter λ = 0.02)1 = 0.02 (0. from 0.02 (0. use F ( x) = ∑ p( xi ) = 1 i =1 (1 + 3k ) + (1 + 2k ) + (0.9900 = 0.01. µ = 1/λ = 1 (Eqn.9802 = 0. x = 3 0. −∞ ∞ −x −x ∞ ∫ ke dx = [−ke ]0 = − k[0 − 1] = k ⇒ 1 0 This is an exponential distribution with parameter λ=1. 2-32) σ2 = 1/λ2 = 1 (Eqn.5 k = 0. x ~ POI(0. +∞ For f(x) to be a probability distribution.01)0 = 1 − 0.0196 1! (b) Pr{x ≥ 1} = 1 − Pr{x = 0} = 1 − p(0) = 1 − e −0. This is a Poisson distribution with parameter λ = 0. e −0. 2-33) 2-27 (2-19). otherwise (a) ∞ To solve for k.02). x ~ POI(0.02)0 = 1 − 0. ⎧(1 + 3k ) / 3. 2-26 (2-18).0198 to 0. x = 2 p ( x) = ⎨ ⎩(0.01).

383. 0 < r < 1. x = 1 ⎪ 3 ⎪ 1.750.383) + 22 (0. p ( x) = kr x .… ∞ F ( x) = ∑ kr x = 1 by definition i =0 k ⎡⎣1 (1 − r ) ⎤⎦ = 1 k = 1− r 2-29 (2-21).05) ⎤ + 2× ⎢ + 3× ⎢ ⎥ ⎥ ⎥⎦ = 1.867 3 3 3 ⎣ ⎦ ⎣ ⎦ ⎣ 3 µ = ∑ xi p( xi ) = 1× ⎢ i =1 3 σ 2 = ∑ xi2 p( xi ) − µ 2 = 12 (0.1 + 0.75 = 1. (a) This is an exponential distribution with parameter λ = 0.1 ⎪ F ( x) = ⎨ = 0. (b) Mfg. 2.867 2 = 0. x = 3 ⎪ 3 ⎩ 2-28 (2-20).125: Pr{x ≤ 1} = F (1) = 1 − e −0.250) − 1.05) ⎤ ⎡1 + 2(0.1. 2-20 .10/calculator.90/calculator.15 + 1.000. cost = \$50/calculator Sale profit = \$25/calculator Net profit = \$[-50(1 + 0.125(1) = 0. x = 0.15 + 1. x = 2 3 ⎪ ⎪1.5 + 5(0.05) ⎤ ⎡ 0. The effect of warranty replacements is to decrease profit by \$5.118 Approximately 11.367) + 32 (0.Chapter 2 Exercise Solutions 2-27 continued (b) ⎡1 + 3(0.8% will fail during the first year.118) + 75]/calculator = \$19.615 i =1 (c) 1.15 ⎧ = 0.

12 Pr{x < 12} = F (12) = ∫ −∞ 12 f ( x) dx = ∫ 4( x − 11.921 ⎝4⎠ 2-21 . The process is stopped if x ≥ 1.75 = 11.78 = 0. and the process will be stopped to look for a cause.01 and n = 25.02)0 (1 − 0. 4 ⎛ 50 ⎞ Pr{ pˆ ≤ 0. ⎛ 25 ⎞ Pr{x ≥ 1} = 1 − Pr{x < 1} = 1 − Pr{x = 0} = 1 − ⎜ ⎟ (0.777821 2-32* (2-24).02) 49 + ⎝0⎠ ⎝1⎠ ⎛ 50 ⎞ + ⎜ ⎟ (0.01)0 (1 − 0.36 = 0. TRUE) (2) MTB > Calc > Probability Distributions > Binomial Cumulative Distribution Function Binomial with n = 25 and p = 0.04} = Pr{x ≤ 2} = ∑ ⎜ ⎟(0. ⎛ 25 ⎞ Pr{x ≥ 1} = 1 − Pr{x < 1} = 1 − Pr{x = 0} = 1 − ⎜ ⎟ (0.75 = 0.875 − 11.125 12 11.02) x (1 − 0. This exercise may also be solved using Excel or MINITAB: (1) Excel Function BINOMDIST(x. p. This is a somewhat difficult operating situation.04) 25 = 1 − 0.64 ⎝0⎠ 2-33* (2-25).75 2-31* (2-23).22 ⎝0⎠ This decision rule means that 22% of the samples will have one or more nonconforming units.02)(50− x ) x =0 ⎝ x ⎠ ⎛ 50 ⎞ ⎛ 50 ⎞ = ⎜ ⎟ (0. x ~ BIN(25.Chapter 2 Exercise Solutions 2-30 (2-22).02) 46 = 0. n.02 and n = 50.04) Stop process if x ≥ 1.02)1 (1 − 0.01) 25 = 1 − 0.04)0 (1 − 0.02) 4 (1 − 0. This is a binomial distribution with parameter p = 0.02)50 + ⎜ ⎟ (0.75 4 x2 2 12 − 47 x 11.75) dx = 11.01 x P( X <= x ) 0 0. 0. This is a binomial distribution with parameter p = 0.

Chapter 2 Exercise Solutions 2-34* (2-26).018 k=3 1 − Pr{x ≤ n(kσ + p )} = 1 − Pr{x ≤ 100(3(0.01)} = 1 − Pr{x ≤ 4} 4 ⎛ 100 ⎞ ⎡ ⎤ ⎛100 ⎞ (0.01) x (0.99) ⎥ ⎝ 1 ⎠ ⎝ 2 ⎠ ⎣⎝ 0 ⎠ ⎦ = 1 − [0.0100) + 0.99) + ⎜ ⎟ (0.99)97 ⎥ = 1− ∑ ⎜ ⎟ ⎟ x =0 ⎝ x ⎠ ⎝ 3 ⎠ ⎣ ⎦ = 1 − [0.01)} = 1 − Pr{x ≤ 3} 3 ⎛ 100 ⎞ ⎡ ⎤ ⎛100 ⎞ (0.01)3 (0.0100 Pr{ pˆ > kσ + p} = 1 − Pr{ pˆ ≤ kσ + p} = 1 − Pr{x ≤ n(kσ + p)} k=1 1 − Pr{x ≤ n(kσ + p )} = 1 − Pr{x ≤ 100(1(0.079 k=2 1 − Pr{x ≤ n(kσ + p )} = 1 − Pr{x ≤ 100(2(0.003 2-22 .921] = 0.01 and n = 100.01)} = 1 − Pr{x ≤ 2} 2 ⎛ 100 ⎞ 100 − x x = 1− ∑ ⎜ ⎟(0.982 + ⎜ (0.01) (0.01(1 − 0.0100) + 0.99)96 ⎥ = 1− ∑ ⎜ ⎟ ⎟ x =0 ⎝ x ⎠ ⎝ 4 ⎠ ⎣ ⎦ = 1 − [0.982] = 0.992] = 0.01) 4 (0.921 + ⎜ (0.01) x (0.99)100− x = 1 − ⎢0.01) x =0 ⎝ x ⎠ ⎡⎛100 ⎞ ⎤ ⎛ 100 ⎞ ⎛100 ⎞ 0 100 1 99 2 98 = 1 − ⎢⎜ ⎟ (0.01) (0. σ = 0. This is a binomial distribution with parameter p = 0.01) (1 − 0.0100) + 0.99) + ⎜ ⎟ (0.01) 100 = 0.01) (0.99)100− x = 1 − ⎢0.

p = D/N = 2/25 = 0.1. 2-23 . This is a hypergeometric distribution with N = 25 and n = 5. without replacement. (c) For N = 150.1.659 ⎝0⎠ This approximation. The binomial approximation is not satisfactory in this case.033 ≤ 0. violates the rule-ofthumb that n/N = 5/25 = 0. 649) Pr{Acceptance} = p (0) = ⎝ ⎠ ⎝ = = 0. M = 2.08)5 = 0.Chapter 2 Exercise Solutions 2-35* (2-27). N) (2) MTB > Calc > Probability Distributions > Hypergeometric Cumulative Distribution Function Hypergeometric with N = 25.08 and n = 5. so the binomial approximation would be a satisfactory approximation the hypergeometric in this case.130) ⎛ 25 ⎞ ⎜ ⎟ ⎝5⎠ This exercise may also be solved using Excel or MINITAB: (1) Excel Function HYPGEOMDIST(x. (a) Given D = 2 and x = 0: ⎛ 2 ⎞ ⎛ 25 − 2 ⎞ ⎜ ⎟⎜ ⎟ 0 5 − 0 ⎠ (1)(33.633 (53.633333 (b) For the binomial approximation to the hypergeometric.20 be less than the suggested 0. ⎛5⎞ Pr{acceptance} = p (0) = ⎜ ⎟ (0. n/N = 5/150 = 0. n. D.08)0 (1 − 0. and n = 5 x P( X <= x ) 0 0. though close to the exact solution for x = 0.

433 ⎛ 30 ⎞ ⎜ ⎟ ⎝5⎠ 2-24 . ⎛ 3 ⎞⎛ 30 − 3 ⎞ ⎜ ⎟⎜ ⎟ 1 ⎠⎝ 5 − 1 ⎠ (3)(17.Chapter 2 Exercise Solutions 2-35 continued (d) Find n to satisfy Pr{x ≥ 1 | D ≥ 5} ≥ 0.506) ⎛ 30 ⎞ ⎜ ⎟ ⎝5⎠ ⎛ 3 ⎞ ⎛ 27 ⎞ ⎜ ⎟⎜ ⎟ 0 5 Pr{x ≥ 1} = 1 − Pr{x = 0} = 1 − p(0) = 1 − ⎝ ⎠ ⎝ ⎠ = 1 − 0.369 (142. n = 5.057 p(0) = ⎝ ⎠ ⎝ ⎠ = (3. 756) = 0.95.05. 2-36 (2-28). 268. 400) ⎛ 25 ⎞ ⎜ ⎟ ⎝ 11 ⎠ Let sample size n = 11. and D = 3.567 = 0.960) = 0.038 p(0) = ⎝ ⎠ ⎝ ⎠ = (4. This is a hypergeometric distribution with N = 30. 760) ⎛ 25 ⎞ ⎜ ⎟ ⎝ 10 ⎠ try n = 11 ⎛ 5 ⎞ ⎛ 20 ⎞ ⎜ ⎟⎜ ⎟ 0 11 (1)(167. or equivalently Pr{x = 0 | D = 5} < 0. 457. ⎛ 5 ⎞ ⎛ 25 − 5 ⎞ ⎛ 5 ⎞ ⎛ 20 ⎞ ⎜ ⎟⎜ ⎟ ⎜ ⎟⎜ ⎟ 0⎠⎝ n − 0 ⎠ ⎝ 0⎠⎝ n ⎠ ⎝ = p(0) = ⎛ 25 ⎞ ⎛ 25 ⎞ ⎜ ⎟ ⎜ ⎟ ⎝n⎠ ⎝n⎠ try n = 10 ⎛ 5 ⎞ ⎛ 20 ⎞ ⎜ ⎟⎜ ⎟ 0 10 (1)(184.550) ⎝ Pr{x = 1} = p (1) = = = 0.

364 ⎝0⎠ Pr{x ≥ 2} = 1 − Pr{x ≤ 1} = 1 − [Pr{x = 0} + Pr{x = 1}] = 1 − p(0) − p(1) ⎛ 50 ⎞ = 1 − 0.1 x P( X <= x ) 0 0.372 = 0.904837 2-39 (2-31).01 errors/bill. n = 50 pages.01)1 = 0. x.Chapter 2 Exercise Solutions 2-37 (2-29).00001 (0.99999 = 0.020)1 (1 − 0. Checking n/N = 50/500 = 0.905 = 0.020.364 − ⎜ ⎟ (0. e −0. TRUE) (2) MTB > Calc > Probability Distributions > Poisson Cumulative Distribution Function Poisson with mean = 0. and D = 10 errors.1.020)50−0 = (1)(1)(0.00001 stones/bottle. This is a Poisson distribution with λ = 0.020)0 (1 − 0.0099 Pr{x = 1} = p (1) = 1 2-25 .01 (0. ⎛ 50 ⎞ Pr{x = 0} = p(0) = ⎜ ⎟ (0. Pr{x ≥ 1} = 1 − Pr{x = 0} = 1 − e −0.095 Pr{x ≥ 1} = 1 − Pr{x = 0} = 1 − p(0) = 1 − 0! This exercise may also be solved using Excel or MINITAB: (1) Excel Function POISSON(λ.364) = 0. e −0.364 − 0.1 defects/unit.1 (0.00001)0 = 1 − 0.1)0 = 1 − 0.00001 0! 2-40 (2-32).1 ≤ 0. with p = D/N = 10/500 = 0. This is a hypergeometric distribution with N = 500 pages. This is a Poisson distribution with λ = 0. the binomial distribution can be used to approximate the hypergeometric.020)50−1 = 1 − 0. This is a Poisson distribution with λ = 0.264 ⎝1⎠ 2-38 (2-30).

… ∞ d ∞ 1 µ = ∑ t ⎡⎣ p(1 − p)t −1 ⎤⎦ = p ⎡ ∑ q t ⎤ = ⎢ ⎥ t =1 dq ⎣ t =1 ⎦ p 2-42 (2-34). σ.0110.945 = 0.? 48 − 40 ⎫ ⎧ Pr{x > 48} = 1 − Pr{x ≤ 48} = 1 − Pr ⎨ z ≤ ⎬ = 1 − Pr{z ≤ 1. t = 1. ⎛ 51 − 1⎞ 3 51−3 Pr{x = 51} = p(51) = ⎜ = (1225)(0.9948 = 0.010 0.012 0.000) × (0.01) 3 1 − ⎝ ⎠ Pr{x > 51} = Pr{r = 0} + Pr{r = 1} + Pr{r = 2} ⎛ 50 ⎞ ⎛ 50 ⎞ ⎛ 50 ⎞ = ⎜ ⎟ 0.9950 + ⎜ ⎟ 0.159 5 ⎭ ⎩ So.055 So. 2-26 .01.? 35 − 40 ⎫ ⎧ Pr{x ≤ 35} = Pr ⎨ z ≤ ⎬ = Pr{z ≤ −1} = Φ (−1) = 0. µ. This exercise may also be solved using Excel or MINITAB: (1) Excel Function NORMDIST(X.9949 ⎜ ⎟ 0.617290) = 0.159) = 7950. n = 50. is (50. x ~ N (40.000) × (0. 2.0008 ⎟ (0.01) (1 − 0. and x = 1 + (5000/100) = 51. This is a Pascal distribution with Pr{defective weld} = p = 0. Pr(t ) = p(1 − p)t −1 .6} 5 ⎭ ⎩ = 1 − Φ (1.9862 ⎝0⎠ ⎝1⎠ ⎝2⎠ 2-43* (2-35).000001)(0.000 How many fail the minimum specification.3. the number that fail the minimum specification are (50. LSL = 35 lb. TRUE) (2) MTB > Calc > Probability Distributions > Normal Cumulative Distribution Function Normal with mean = 40 and standard deviation = 5 x P( X <= x ) 35 0. 52). r = 3 welds.158655 How many exceed 48 lb.Chapter 2 Exercise Solutions 2-41 (2-33).6) = 1 − 0. the number that exceed 48 lb.055) = 2750.

Desire Pr{Conformance} = 1 / 1000 = 0.9979 = 0.0005) = −3. USL = 5. is currently centered between the specification limits (target = 5 V).95 − 5 = = 0. so σ = LSL − µ 4.001.02 ⎠ 2-45* (2-37).00621 = 0. The process. Shifting the process mean in either direction would increase the number of nonconformities produced. Assume that the process remains centered between the specification limits at 5 V.5) = 0. x ~ N ( µ .0152 to have at least 999 of 1000 conform to specification. x ~ N(5.9991 32 − µ = −1.05 V Pr{Conformance} = Pr{LSL ≤ x ≤ USL} = Pr{x ≤ USL} − Pr{x ≤ LSL} ⎛ 5.0021 2-27 . Find µ such that Pr{x < 32} = 0.0005 z = Φ −1 (0. LSL = 4.95 − 5 ⎞ = Φ⎜ ⎟−Φ⎜ ⎟ = Φ (2.99379 − 0. Need Pr{x ≤ LSL} = 0.015 z −3.95 V. 352) Pr{x > 1000} = 1 − Pr{x ≤ 1000} 1000 − 900 ⎫ ⎧ = 1 − Pr ⎨ x ≤ ⎬ 35 ⎩ ⎭ = 1 − Φ (2.001 / 2 = 0.29 Process variance must be reduced to 0. 2-46 (2-38). Φ ( z ) = 0.0228) = −1. 0.0228.05 − 5 ⎞ ⎛ 4. 42 ).8571) = 1 − 0.98758 ⎝ 0. x ~ N(900.0 2-47 (2-39). Φ −1 (0.9991 4 µ = −4(−1.29 z= LSL − µ σ . with mean 5 V.Chapter 2 Exercise Solutions 2-44* (2-36).5) − Φ (−2.022).02 ⎠ ⎝ 0.9991) + 32 = 40.0005.

000 1.0124) + 5 (0. then choose process 1 2-28 .Chapter 2 Exercise Solutions 2-48 (2-40).5) = 1 − 0.0062 = 0.500 ⎫ ⎧ ⎧ = 1 − Pr ⎨ z2 ≤ ⎬ + Pr ⎨ z2 ≤ ⎬ 500 500 ⎩ ⎭ ⎩ ⎭ = 1 − Φ (5) + Φ (−5) = 1 − 1.500 ⎫ ⎧ ⎧ = 1 − Pr ⎨ z1 ≤ ⎬ + Pr ⎨ z1 ≤ ⎬ 1.0620.000 h sales = \$10/unit.0000) + 5 (0. σ22 = 5002).500 ⎫ 5. 000 ⎩ ⎭ ⎩ ⎭ = 1 − Φ (2. LSL = 5. x ~ N(5000.5) + Φ (−2. defect = \$5/unit. 000 − 7. USL = 10.005 Φ −1 (0. x1 ~ N(7500.000 h. 000 − 7. Find LSL such that Pr{x < LSL} = 0.0000 + 0.500 ⎫ 5.005) = −2.9380 – c1 For Process 2 proportion defective = p2 = 1 − Pr{LSL ≤ x2 ≤ USL} = 1 − Pr{x2 ≤ USL} + Pr{x2 ≤ LSL} 10. x2 ~ N(7500. profit = \$10 × Pr{good} + \$5 × Pr{bad} – c For Process 1 proportion defective = p1 = 1 − Pr{LSL ≤ x1 ≤ USL} = 1 − Pr{x1 ≤ USL} + Pr{x1 ≤ LSL} 10.5758) + 5000 = 4871 2-49 (2-41). 000 − 7.0000 profit for process 2 = 10 (1 – 0. σ12 = 10002).9938 + 0. 000 − 7.0124 profit for process 1 = 10 (1 – 0.5758 50 LSL = 50(−2.0000 = 0.5758 LSL − 5000 = −2.0000) – c2 = 10 – c2 If c2 > c1 + 0.0124) – c1 = 9. 502).

2 ⎣ ⎝ C0 + C2 ⎠ ⎦ 2-29 .Chapter 2 Exercise Solutions 2-50 (2-42). Proportion less than lower specification: 6−µ⎫ ⎧ pl = Pr{x < 6} = Pr ⎨ z ≤ ⎬ = Φ (6 − µ ) 1 ⎭ ⎩ Proportion greater than upper specification: 8−µ ⎫ ⎧ pu = Pr{x > 8} = 1 − Pr{x ≤ 8} = 1 − Pr ⎨ z ≤ ⎬ = 1 − Φ (8 − µ ) 1 ⎭ ⎩ Profit = +C0 pwithin − C1 pl − C2 pu = C0 [Φ (8 − µ ) − Φ (6 − µ )] − C1[Φ (6 − µ )] − C2 [1 − Φ (8 − µ )] = (C0 + C2 )[Φ (8 − µ )] − (C0 + C1 )[Φ (6 − µ )] − C2 d d ⎡ 8− µ 1 ⎤ [Φ (8 − µ )] = exp(−t 2 / 2)dt ⎥ ∫ ⎢ dµ d µ ⎣ −∞ 2π ⎦ Set s = 8 – µ and use chain rule d d ⎡s 1 1 ⎤ ds =− [Φ (8 − µ )] = ⎢ ∫ exp(−t 2 / 2)dt ⎥ exp −1/ 2 × (8 − µ ) 2 dµ ds ⎣ −∞ 2π 2π ⎦ dµ ( ) d (Profit) ⎡ 1 ⎤ ⎡ 1 ⎤ = −(C0 + C2 ) ⎢ exp −1/ 2 × (8 − µ ) 2 ⎥ + (C0 + C1 ) ⎢ exp −1/ 2 × (6 − µ ) 2 ⎥ dµ ⎣ 2π ⎦ ⎣ 2π ⎦ ( ) ( ) Setting equal to zero 2 C0 + C1 exp −1/ 2 × (8 − µ ) = = exp(2µ − 14) C0 + C2 exp −1/ 2 × (8 − µ ) 2 ( ( So µ = ) ) ⎤ 1 ⎡ ⎛ C0 + C1 ⎞ ⎢ln ⎜ ⎟ + 14 ⎥ maximizes the expected profit.

1.... n ⎝ x⎠ ∞ n ⎡⎛ n ⎞ ⎤ n −1 µ = E ( x) = ∑ xi p ( xi ) = ∑ x ⎢⎜ ⎟ p x (1 − p ) n − x ⎥ = n ⎡⎣ p + (1 − p ) ⎤⎦ p = np i =1 x =0 ⎣⎝ x ⎠ ⎦ σ 2 = E[( x − µ ) 2 ] = E ( x 2 ) − [ E ( x)]2 ∞ n ⎡⎛ n ⎞ n− x ⎤ E ( x 2 ) = ∑ xi2 p ( xi ) = ∑ x 2 ⎢⎜ ⎟ p x (1 − p ) ⎥ =np + (np ) 2 − np 2 i =1 x =0 ⎣⎝ x ⎠ ⎦ σ 2 = ⎡⎣ np + (np) 2 − np 2 ⎤⎦ − [ np ] = np(1 − p) 2 2-52 (2-44). p ( x) = ⎜ ⎟ p x (1 − p) n − x .. x = 0.Chapter 2 Exercise Solutions 2-51 (2-43). ⎛n⎞ For the binomial distribution. x = 0.1. p( x) = ∞ ∞ ⎛ e− λ λ x ⎝ x! µ = E[ x] = ∑ xi p( xi ) = ∑ x ⎜ i =1 x =0 e− λ λ x . For the Poisson distribution.… x! ⎞ − λ ∞ λ ( x −1) =e− λ λ eλ = λ ⎟=e λ∑ x = 0 ( x − 1)! ⎠ ( ) σ 2 = E[( x − µ ) 2 ] = E ( x 2 ) − [ E ( x)]2 ⎛ e− λ λ x ⎞ 2 E ( x ) = ∑ x p( xi ) = ∑ x ⎜ ⎟=λ +λ i =1 x =0 ⎝ x! ⎠ 2 ∞ ∞ 2 i 2 σ 2 = (λ 2 + λ ) − [ λ ] = λ 2 2-30 .

setting u = x 2 and dv = λ exp(−λ x) uv − ∫ vdu = ⎡⎣ x exp(−λ x) ⎤⎦ 2 +∞ 0 +∞ + 2 ∫ x exp(−λ x)dx = (0 − 0) + 0 σ2 = 2 λ 2 − 1 λ 2 = 2 λ2 1 λ2 2-31 . f ( x ) = λ e − λ x . For the exponential distribution.Chapter 2 Exercise Solutions 2-53 (2-45). setting u = x and dv = λ exp(−λ x) +∞ +∞ 1 1 uv − ∫ vdu = ⎡⎣ − x exp ( −λ x ) ⎤⎦ 0 + ∫ exp ( −λ x ) dx = 0 + = λ 0 λ For the variance: ⎛1⎞ σ = E[( x − µ ) ] = E ( x ) − [ E ( x) ] = E ( x ) − ⎜ ⎟ ⎝λ⎠ 2 2 2 2 +∞ +∞ −∞ 0 2 2 E ( x 2 ) = ∫ x 2 f ( x)dx = ∫ x 2 λ exp(−λ x)dx Integrate by parts. x ≥ 0 For the mean: +∞ +∞ 0 0 µ = ∫ xf ( x)dx = ∫ x ( λ e − λ x )dx Integrate by parts.

25 + 1.25101) Z -6.025 = 1.000 3-2.25.00000] = 0 (c) ⎞ ≤ µ ≤ x + Z ⎛σ ⎞ x − Zα / 2 ⎛⎜ σ ⎟ ⎟ α /2 ⎜ n⎠ n⎠ ⎝ ⎝ ( 8.Chapter 3 Exercise Solutions 3-1. H1: µ > 125. 3-1 . 8. α = 0. H1: µ ≠ 8. x − µ0 8.2535 cm. α = 0.2535 − 8.78 P 0.25 Z0 = = = 6.2535 The assumed standard deviation = 0. σ = 2 psi (a) µ0 = 125.002 15 ) 8. and conclude that the mean tensile strength exceeds 125 psi.05/2 = Z0. Reject H0 if |Z0| > Zα/2. x = 8.2535 vs not = 8.25.645 Reject H0: µ = 125. n = 15.25000 0. (b) P-value = 2[1 − Φ(Z0)] = 2[1 − Φ(6.78 σ n 0.96 Reject H0: µ = 8.25 vs.249 ≤ µ ≤ 8. x − µ0 127 − 125 Z0 = = = 2.25 cm.24899.05 = 1. σ = 0. and conclude that the mean bearing ID is not equal to 8.828 σ n 2 8 Zα = Z0.002 cm (a) µ0 = 8.05 Test H0: µ = 8.251 MTB > Stat > Basic Statistics > 1-Sample Z > Summarized data One-Sample Z Test of mu = 8.96 0.00052 (8.05 Test H0: µ = 125 vs.96 0.002 15 Zα/2 = Z0. n = 8.78)] = 2[1 − 1. Reject H0 if Z0 > Zα.25 − 1.002 ) ( 15 ≤ µ ≤ 8.25.002 N Mean SE Mean 95% CI 15 8. x = 127 psi.

833 Reject H0: µ = 25.6248 SE Mean 0.05 Test H0: µ = 25 vs.00234 (c) In strength tests.645 ( 2 8 ) ≤ µ x − Zα σ 125.5138 95% Lower Bound 25. x ~ N(µ.0581 T 1.62.Chapter 3 Exercise Solutions 3-2 continued (b) P-value = 1 − Φ(Z0) = 1 − Φ(2. we usually are interested in whether some minimum requirement is met. n = 10 (a) x = 26. x − µ0 26. (d) ( n) ≤ µ 127 − 1.828) = 1 − 0. n−1 = t0.95 P 0. and conclude that the mean life exceeds 25 h.62 10 tα. H1: µ > 25. Reject H0 if t0 > tα. MTB > Stat > Basic Statistics > 1-Sample t > Samples in columns One-Sample T: Ex3-3 Test of mu = 25 vs > 25 Variable Ex3-3 N 10 Mean 26.707 125. α = 0.0.952 S n 1. A one-sided hypothesis test lets us do this.8 ≤ µ MTB > Stat > Basic Statistics > 1-Sample Z > Summarized data One-Sample Z Test of mu = 125 vs > 125 The assumed standard deviation = 2 95% Lower N Mean SE Mean Bound Z 8 127.05.99766 = 0. σ).837 2. µ0 = 25. not simply that the mean does not equal the hypothesized value. 10−1 = 1.0000 StDev 1. s = 1.0 − 25 t0 = = = 1.002 3-3.042 3-2 .000 0.83 P 0.

833 1.9419) T 1. Hours 30 32 The plotted points fall approximately along a straight line.0000 1.94 MTB > Stat > Basic Statistics > 1-Sample t > Samples in columns One-Sample T: Ex3-3 Test of mu = 25 vs not = 25 Variable N Mean StDev Ex3-3 10 26.95% CI 99 Mean StDev N AD P-Value 95 90 26 1.10 x − tα / 2.0 − 1.833 1.083 (c) MTB > Graph > Probability Plot > Single Probability Plot of Battery Service Life (Ex3-3) Normal .114 0.0581.Chapter 3 Exercise Solutions 3-3 continued (b) α = 0.986 Percent 80 70 60 50 40 30 20 10 5 1 20 22 24 26 28 Lifetime.62 n ≤ µ ≤ x + tα / 2. so the assumption that battery life is normally distributed is appropriate.95 P 0.n −1 S ( 26.5138 90% CI (25.62 10 ) 25.0 + 1.6248 SE Mean 0. 3-3 . 26.n −1 S ) n ( 10 ≤ µ ≤ 26.625 10 0.06 ≤ µ ≤ 26.

089 S n 0. α = 0.3934.4 vs not = 13.05.3990) T -3.4 Variable N Mean StDev SE Mean Ex3-5 10 13. n = 10.05 Test H0: µ = 13.025.62 25.39618 − 13. H1: µ ≠ 13. Reject H0 if |t0| > tα/2.39618 − 3.00391 10 tα/2.4 vs.00391 95% CI (13. s = 0.Chapter 3 Exercise Solutions 3-4.62 h.013 3-4 . x − µ0 13.3922. σ).09 P 0. x = 26.0039 0. n−1 = t0. x = 13.2498 0.3962 0.4 vs not = 13. 3-5.39618 × 1000 Å.0012 (b) α = 0.4.01 ( x − tα / 2.833 ( ) 10 ) ≤ µ n ≤µ x − tα .2498 ( 0. 9 = 2.4 × 1000 Å.00391 µ0 = 13.013 n ≤ µ ≤ x + tα / 2. s = 1.4 t0 = = = −3. tα.262 Reject H0: µ = 13.n −1 S ( 13. n = 10.n −1 S 10 ) 13. x ~ N(µ. MTB > Stat > Basic Statistics > 1-Sample t > Samples in columns One-Sample T: Ex3-5 Test of mu = 13.9 = 1. (a) x ~ N(µ.0 h. α = 0.0 − 1.05.0012 99% CI (13.3962 0.4 Variable N Mean StDev SE Mean Ex3-5 10 13.39216 ≤ µ ≤ 13. 13.40020 MTB > Stat > Basic Statistics > 1-Sample t > Samples in columns One-Sample T: Ex3-5 Test of mu = 13.4 × 1000 Å.06 ≤ µ The manufacturer might be interested in a lower confidence interval on mean battery life when establishing a warranty policy. and conclude that the mean thickness differs from 13. n−1 = t0.0039 0.4002) T -3. 13.39618 + 3.833 1.n −1 S ( 26.4. σ).00391 P 0.09 ) ( n) 10 ) ≤ µ ≤ 13.

385 13. MTB > Stat > Basic Statistics > 1-Sample t > Samples in columns One-Sample T: Ex3-6 Test of mu = 12 vs > 12 Variable Ex3-6 N 10 Mean 12.076 3-5 .57 P 0.01 n = 10. so the assumption that battery life is normally distributed is not appropriate.9880 T 1. H1: µ > 12. µ0 = 12.003909 10 0. x = 12.005. n−1 = t0.40 0.395 13.015.015 − 12 t0 = = = 1.5655 S n 0.237 0. 9 = 3. instead of a straight line.0303 SE Mean 0. and conclude that there is not enough evidence that the mean fill volume exceeds 12 oz. 3-6. α = 0.711 Percent 80 70 60 50 40 30 20 10 5 1 13. x − µ0 12. x1000 Angstroms 13. (a) x ~ N(µ.400 13.250 Do not reject H0: µ = 12.030 Test H0: µ = 12 vs.380 13.95% CI 99 Mean StDev N AD P-Value 95 90 13. Reject H0 if t0 > tα. s = 0.0096 99% Lower Bound 11.Chapter 3 Exercise Solutions 3-5 continued (c) MTB > Graph > Probability Plot > Single Probability Plot of Photoresist Thickness (Ex3-5) Normal .0303 10 tα/2.390 13.0150 StDev 0.405 Thickness. σ).410 The plotted points form a reverse-“S” shape.

262 x − tα / 2.05 Fill Volume.05.274 0. ounces 12.90 11. find n ( ( 2 ⎡ Zα / 2 σ ⎣ 2 ⎡1.262 ( S 10 ) ≤ µ ≤ 12.62 ( S 10 ) 11.57 P 0. 3-7.015 + 2.00 12.152 (c) MTB > Graph > Probability Plot > Single Probability Plot of Fill Volume (Ex3-6) Normal .9600.9600 4 ⎣ ) ) n ⎤ = total width ⎦ n ⎤ =1 ⎦ n = 246 3-6 . Zα/2 = Z0. α = 0. 12.993 ≤ µ ≤ 12.025 = 1.9933.10 12.0367) T 1.95 12.03028 10 0.Chapter 3 Exercise Solutions 3-6 continued (b) α = 0.025.0096 95% CI (11.05 tα/2. total confidence interval width = 1 lb.n −1 S ( ) ( n) 12.95% CI 99 Mean StDev N AD P-Value 95 90 12.02 0. σ = 4 lb. 9 = 2.037 MTB > Stat > Basic Statistics > 1-Sample t > Samples in columns One-Sample T: Ex3-6 Test of mu = 12 vs not = 12 Variable N Mean StDev Ex3-6 10 12.15 The plotted points fall approximately along a straight line. so the assumption that fill volume is normally distributed is appropriate.582 Percent 80 70 60 50 40 30 20 10 5 1 11.015 − 2. n−1 = t0.0150 0.0303 SE Mean 0.n −1 S n ≤ µ ≤ x + tα / 2.

0001 N Mean SE Mean 95% CI 25 0.131 Reject H0: µ = 12. n−1 = t0.025 = 1.504600 0.Chapter 3 Exercise Solutions 3-8.0001 n ≤ µ ≤ x + Zα / 2 σ 25 ) 0. MTB > Stat > Basic Statistics > 1-Sample t > Samples in columns One-Sample T: Ex3-9 Test of mu = 12 vs not = 12 Variable N Mean StDev Ex3-9 16 10.2498 95% CI (9.504561.00 P 0.999 V (a) µ0 = 12.50456 ≤ µ ≤ 0. MTB > Stat > Basic Statistics > 1-Sample Z > Summarized data One-Sample Z Test of mu = 0. s = 0.000 (b) P-value = 2[1 − Φ(Z0)] = 2[1 − Φ(105)] = 2[1 − 1] = 0 (c) ( x − Zα / 2 σ ( 0.259 V. µ0 = 0. and conclude that the mean output voltage differs from 12V.50464 3-9.999 16 tα/2.960 0.5025. n = 16. x ~ N(µ.96 Reject H0: µ = 0.5025 The assumed standard deviation = 0.05 n = 25.025.05 Test H0: µ = 12 vs. x − µ0 10.7270. H1: µ ≠ 0.97 P 0.5025 = 105 Z0 = σ n 0.259 − 12 t0 = = = −6. x = 10. (a) x ~ N(µ.2594 0.7917) T -6. σ = 0.05/2 = Z0.5025. 10.5046 − 0.971 S n 0. σ). x −µ 0 = 0.504639) Z 105. 15 = 2. α = 0. H1: µ ≠ 12.5025.9990 SE Mean 0.5046 + 1. Reject H0 if |Z0| > Zα/2.000 3-7 .5025 vs not = 0.000020 (0. Reject H0 if |t0| > tα/2. 0.960 ( 0. x = 0.5025 vs.5025.0001 ) ( n) 25 ) ≤ µ ≤ 0.0001 25 Zα/2 = Z0.5046 − 1.0001 in Test H0: µ = 0. α = 0.5046 in. σ). and conclude that the mean rod diameter differs from 0.

α = 0. and conclude that there is insufficient evidence that the variance differs from 1. cannot be rejected.738 ≤ σ ≤ 1.05 Test H0: σ 2 = 1 vs.488 6. n−1 = χ20.999 (16 − 1)0. n −1 (16 − 1)0.9992 2 ≤σ ≤ 27.16−1 = 6.9992 = 14.727 ≤ µ ≤ 10. (16 − 1)0.131 0.970 1 σ 02 χ2α/2. Reject H0 if χ 20 > χ 2α/2.16−1 = 27. 3-8 . the null hypothesis.391 2 0.975.259 − 2. σ 02 = 1.792 (c) σ 02 = 1.131( 0.999 ) ( n) 16 ) ≤ µ ≤ 10.488 χ21−α/2.999 n ≤ µ ≤ x + tα / 2. n-1. χ 02 = (n − 1) S 2 = (d) (n − 1) S 2 χα2 / 2.025.n −1 S ( 10.Chapter 3 Exercise Solutions 3-9 continued (b) ( x − tα / 2.262 Do not reject H0: σ 2 = 1. n-1 or χ20 < χ 21-α/2. H0: σ 2 = 1.259 + 2.546 Since the 95% confidence interval on σ contains the hypothesized value.262 2 0.n −1 S 16 ) 9.n −1 ≤σ2 ≤ (n − 1) S 2 χ2 1−α / 2. n−1 = χ20. H1: σ 2 ≠ 1.545 ≤ σ ≤ 2.

Chapter 3 Exercise Solutions 3-9 (d) continued MTB > Stat > Basic Statistics > Graphical Summary Summary for Output Voltage (Ex3-9) A nderson-D arling N ormality Test 8 9 10 11 A -S quared P -V alue 0.05.15 σ2 ≤ (n − 1) S 2 χ12−α .50 10.767 M ean S tD ev V ariance S kew ness Kurtosis N 10.2609 α = 0.738 1.727 10.n −1 = χ 0.140 11.492793 16 M inimum 1st Q uartile M edian 3rd Q uartile M aximum 12 8.062 σ ≤ 1.998 0.9992 σ ≤ 7.75 11.546 Mean Median 9.00 10.23 0.50 9.2609 2 σ ≤ 2.999 0.945 95% C onfidence Interv al for S tD ev 95% Confidence Intervals 0.000 95% C onfidence Interv al for M ean 9.25 10.533 10.430 10.95.792 95% C onfidence Interv al for M edian 9.00 (e) 2 = 7.259 0.436 2 3-9 .370 9.n −1 (16 − 1)0.75 10. χ12−α .150 12.116487 -0.

682 2 2 0. x2 = 2.682)] = 2[1 − 1.0152 20 σ 1 n1 + σ 2 n2 Zα/2 = Z0.025 = 1. 3-10.26 0. x1 = 2.04 l. (b) P-value = 2[1 − Φ(Z0)] = 2[1 − Φ(−7. n2 = 20.0102 25 + 0.767 Percent 80 70 60 50 40 30 20 10 5 1 7 8 9 10 11 Output Voltage 12 13 14 From visual examination of the plot. ( x1 − x2 ) − ∆ 0 (2. the assumption of a normal distribution for output voltage seems appropriate.010 l. (a) α = 0.96 Reject H0: µ1 – µ2 = 0.9990 16 0. ∆ 0 = 0 Test H0: µ1 – µ2 = 0 versus H0: µ1 – µ2 ≠ 0.Chapter 3 Exercise Solutions 3-9 continued (f) MTB > Graph > Probability Plot > Single Probability Plot of Output Voltage (Ex3-9) Normal .015 l. σ1 = 0. Reject H0 if Z0 > Zα/2 or Z0 < –Zα/2.04 − 2.07) − 0 Z0 = = = −7.05/2 = Z0.00000] = 0 3-10 . and conclude that there is a difference in mean net contents between machine 1 and machine 2.95% CI 99 Mean StDev N AD P-Value 95 90 10. σ2 = 0.96 −Zα/2 = −1.230 0.07 l. n1 = 25.05.

1604(0.1604 Sp = (n1 − 1) S12 + (n2 − 1) S22 (7 − 1)0. S1 = 0.376) + 2.07) + 1.mu (Ex3-11T2) Estimate for difference: 0.Chapter 3 Exercise Solutions 3-10 continued (c) ( x1 − x2 ) − Zα / 2 2 (2.9600 0.1252 = = 0.07) − 1.127969.376.025. x1 = 1. 0.010 σ 12 25 n1 + σ 22 + 0.383 − 1.006607 95% CI for difference: (-0. the readings will be the same. Ex3-11T2 Two-sample T for Ex3-11T1 vs Ex3-11T2 N Mean StDev SE Mean Ex3-11T1 7 1.125 0. tα/2.120) 1 7 + 1 8 ≤ ( µ1 − µ 2 ) ≤ (1.115 0.044 Difference = mu (Ex3-11T1) . (a) MTB > Stat > Basic Statistics > 2-Sample t > Samples in different columns Two-Sample T-Test and CI: Ex3-11T1.127 ≤ ( µ1 − µ 2 ) ≤ 0.383 − 1. and conclude that there is not sufficient evidence of a difference between measurements obtained by the two technicians.125 α = 0.038 ≤ ( µ1 − µ 2 ) ≤ −0.04 − 2.015 2 20 The confidence interval for the difference does not contain zero.1604(0.9600 0.115. n2 = 8. S2 = 0.022 25 + 0. (b) The practical implication of this test is that it does not matter which technician measures parts.120) 1 7 + 1 8 −0.n + n 1 2 −2 S p 1 n1 + 1 n2 ≤ ( µ1 − µ 2 ) ≤ ( x1 − x2 ) + tα / 2. x2 = 1.120 n1 + n2 − 2 7+8−2 ( x1 − x2 ) − tα / 2. We can conclude that there is no difference in measurements obtained by the two technicians. (c) n1 = 7. We can conclude that the machines do not fill to the same volume.05. we would have been concerned that the technicians obtained different measurements.141183) T-Test of difference = 0 (vs not =): T-Value = 0. If the null hypothesis had been rejected.1152 + (8 − 1)0. 13 = 2.04 − 2.010 20 −0.383 0. 3-11 .917 DF = 13 Do not reject H0: µ1 – µ2 = 0.141 The confidence interval for the difference contains zero.043 Ex3-11T2 8 1.376) − 2.11 Both use Pooled StDev = 0.n + n 1 2 −2 S p 1 n1 + 1 n2 (1. 3-11.1204 P-Value = 0.015 2 n2 σ 12 ≤ ( µ1 − µ 2 ) ≤ ( x1 − x2 ) + Zα / 2 n1 + σ 22 n2 2 ≤ ( µ1 − µ 2 ) ≤ (2.376 0. n1+n2−2 = t0.383. and an investigation should be undertaken to understand why.

and an investigation should be undertaken to understand why.5 1.4 Data 1.6.05/ 2. Ex3-11T2 F-Test Test Statistic P-Value Ex3-11T1 0.15 0.n −1.n 1 2 −1 2 or F0 < F1−α / 2.8−1 = F0.920 0.025.n 1 2 −1 = F1−0.Chapter 3 Exercise Solutions 3-11 continued (d) α = 0.2 1.176 MTB > Stat > Basic Statistics > 2 Variances > Summarized data Test for Equal Variances for Ex3-11T1.n −1.7 −1.n −1.10 0. and conclude that there is no difference in variability of measurements obtained by the two technicians.7 = 0. 3-12 .125 = 0.854 Lev ene's Test Test Statistic P-Value Ex3-11T2 0.n −1 = F0.6. If the null hypothesis is rejected.7 = 5.119 2 1 1 2 2 2 2 F1−α / 2.8464 Fα / 2.05 Test H 0 : σ 12 = σ 22 versus H1 : σ 12 ≠ σ 22 .3 1.6 Do not reject H0.25 95% Bonferroni Confidence Intervals for StDevs 0.7 −1.115 0.05/ 2.20 0.n −1. we would have been concerned about the difference in measurement variability between the technicians.30 Ex3-11T1 Ex3-11T2 1.85 0. Reject H 0 if F0 > Fα / 2.01 0.n −1. 1 2 F0 = S S = 0.8−1 = F0.975.

05 F1−α / 2. n1 −1 = F0.1252 0.n −1 ≤ ≤ 2 1−α / 2.1954) (5. Ex3-11T2) Normal .975.6955) ≤ ≤ 0.025.n −1 = χ 0.n 2 −1.1148 7 0. 3-13 . n1 −1 2 1 S2 S σ σ 12 0.6 1.n −1.95% CI 99 Variable Ex3-11T1 Ex3-11T2 95 90 Mean 1.376.383 1.0128.4 Data 1.n −1 ≤σ2 ≤ (n − 1) S 2 χ12−α / 2.1252 (8 − 1)0.975.025.6899 α = 0. n2 −1.1152 0.6 = 5. x2 = 1.065 (g) MTB > Graph > Probability Plot > Multiple Probability Plot of Surface Finish by Technician (Ex3-11T1.821 σ 22 (f) n2 = 8.693 70 60 50 40 30 20 10 5 1 1.1152 (0. S2 = 0.6955 S12 σ S F Fα / 2.943 0.n −1 (8 − 1)0.6899 2 0.1252 σ 22 0.1249 8 0.165 ≤ σ 12 ≤ 4. Fα / 2.n 2 −1.8 The normality assumption seems reasonable for these readings.142 0.235 0.6 = 0.7 = 1.2 1.n2 −1 = χ 0.1954.Chapter 3 Exercise Solutions 3-11 continued (e) α = 0.376 Percent 80 StDev N AD P 0. χ12−α / 2.0128 1.7 2 (n − 1) S 2 χα2 / 2. n1 −1 2 2 1 1 2 2 2 2 = F0.7.0 1.1252 2 ≤σ ≤ 16.7.125 2 2 = 16.007 ≤ σ ≤ 0.05. χα2 / 2.

From Eqn. x2 = 149.n + n 1 2 −2 S p 1 n1 + 1 n2 (147. and conclude that there is no difference between the quenching processes.22 10 + 10 − 2 n1 + n2 − 2 ( x1 − x2 ) − tα / 2.025.1009(5.4) + 2. S2 = 5.22) 1 10 + 1 10 ≤ ( µ1 − µ 2 ) ≤ (147.n + n 1 2 −2 S p 1 n1 + 1 n2 ≤ ( µ1 − µ 2 ) ≤ ( x1 − x2 ) + tα / 2.4) − 2.97 1.4.77 Both use Pooled StDev = 5. 3. x1 = 147.1009 Sp = (n1 − 1) S12 + (n2 − 1) S 22 (10 − 1)4.05. the test statistic is 2 2 2 + S n S n x − x 1 1 2 2 1 2 with degrees of freedom ν = −2 t0* = 2 2 2 2 2 S1 n1 + S 22 n2 S1 n1 S 2 n2 + ( n1 + 1) ( n2 + 1) ( ( ) ) ( ) A 100(1-α)% confidence interval on the difference in means would be: ( x1 − x2 ) − tα / 2.2217 P-Value = 0.462 = = 5.ν S12 n1 + S 22 n2 ≤ ( µ1 − µ2 ) ≤ ( x1 − x2 ) + tα / 2. 18 = 2.7 ≤ ( µ1 − µ 2 ) ≤ 3.6 Ex3-13OQ 10 149.6 − 149.70615.46 1. n1+n2−2 = t0.10615) T-Test of difference = 0 (vs not =): T-Value = -0.1 3-14 . for σ 12 ≠ σ 22 and both unknown.46 (a) Assume σ 12 = σ 22 MTB > Stat > Basic Statistics > 2-Sample t > Samples in different columns Two-Sample T-Test and CI: Ex3-13SQ.97 2 + (10 − 1)5.1009(5.60 4. (b) α = 0.22) 1 10 + 1 10 −6.6.ν S12 n1 + S 22 n2 3-13. Ex3-13OQ Two-sample T for Ex3-13SQ vs Ex3-13OQ N Mean StDev SE Mean Ex3-13SQ 10 147. S1 = 4.97 Oil quench: n2 = 10. 3-54 and 3-55.80000 95% CI for difference: (-6.6 − 149. tα/2. Saltwater quench: n1 = 10.40 5.Chapter 3 Exercise Solutions 3-12.451 DF = 18 Do not reject H0.7 Difference = mu (Ex3-13SQ) .mu (Ex3-13OQ) Estimate for difference: -1.

975.461 10 0.0260) σ 22 5. n2 −1.2484. 1 1 σ 22 S S22 σ 12 4.025. 3-15 .906 90 Percent 80 70 60 50 40 30 20 10 5 1 130 140 150 Hardness 160 170 The normal distribution assumptions for both the saltwater and oil quench methods seem reasonable.9.Chapter 3 Exercise Solutions 3-13 continued (c) α = 0. 2 1.95% CI 99 Variable Ex3-13SQ Ex3-13OQ 95 AD P Mean StDev N 147.9 = 4.97 2 4.n =F = 0. n1 −1 2 1 2 α / 2.21 ≤ σ 12 ≤ 3. the assumption of equal variances seems reasonable.2484) (4.462 0. n1 −1 = F0.n 0. Fα / 2.218 0.779 149.6 4.97 2 ≤ ≤ (0.4 5.0260 S12 σ 12 S ≤ ≤ F F − α n − n − 1 / 2.9 2 −1.05 F1−α / 2.9. n1 −1 2 2 −1.34 σ 22 Since the confidence interval includes the ratio of 1. (d) MTB > Graph > Probability Plot > Multiple Probability Plot of Quench Hardness (Ex3-13SQ.971 10 0.169 0. Ex3-13OQ) Normal .462 5.

5) − 20 = = −0.645 pˆ − Zα / 2 pˆ (1 − pˆ ) n ≤ p ≤ pˆ + Zα / 2 pˆ (1 − pˆ ) n 0.1 vs p not = 0.10. pˆ = x/n = 18/200 = 0.09 − 1.10.47 P-Value 0.96 Do not reject H0.10) Zα/2 = Z0.5) − np0 (18 + 0.645 0. 0.09) 200 ≤ p ≤ 0.Chapter 3 Exercise Solutions 3-14.10/2 = Z0.645 0.10.3536 np0 (1 − p0 ) 20(1 − 0.10. not an approximation. Reject H0 if |Z0| > Zα/2.10) = 20 Since (x = 18) < (np0 = 20).050338. P-value = 2[1 − Φ|Z0|] = 2[1 − Φ|−0.10 versus H1: p ≠ 0. x = 18. n = 200.7236 MTB > Stat > Basic Statistics > 1 Proportion > Summarized data Test and CI for One Proportion Test of p = 0. α = 0.05.6382] = 0.057 ≤ p ≤ 0.09(1 − 0.025 = 1.05 = 1. np0 = 200(0.09(1 − 0. Test H0: p = 0.123 3-16 .05/2 = Z0.09) 200 0.637 Note that MINITAB uses an exact method. use the normal approximation to the binomial for x < np0. Zα/2 = Z0.1 Sample X N Sample p 95% CI 1 18 200 0. (b) α = 0.090000 (0. and conclude that the sample process fraction nonconforming does not differ from 0. Z0 = ( x + 0.09 + 1.3536|] = 2[1 − 0.129662) Z-Value -0.09 (a) p0 = 0.

0387 np0 (1 − p0 ) 40(1 − 0.13 + 1.08 versus H1: p ≠ 0. Test H0: p = 0. pˆ = x/n = 65/500 = 0.159478) Z-Value 4.05.13) 500 p ≤ 0. ( x − 0.08 Sample X N Sample p 95% CI 1 65 500 0. (b) P-value = 2[1 − Φ|Z0|] = 2[1 − Φ|4.12 P-Value 0.05/2 = Z0.99997] = 0.130 (a) p0 = 0.08.08.96 Reject H0. 0. Reject H0 if |Z0| > Zα/2.08) Zα/2 = Z0. α = 0.000 Note that MINITAB uses an exact method.130000 (0. MTB > Stat > Basic Statistics > 1 Proportion > Summarized data Test and CI for One Proportion Test of p = 0.645 0. Zα = Z0.00006 (c) α = 0.Chapter 3 Exercise Solutions 3-15.08) = 40 Since (x = 65) > (np0 = 40). np0 = 500(0.5) − 40 Z0 = = = 4. x = 65. not an approximation.645 p ≤ pˆ + Zα pˆ (1 − pˆ ) n p ≤ 0.025 = 1.05 = 1. and conclude the sample process fraction nonconforming differs from 0.05.08.100522.0387|] = 2[1 − 0.13(1 − 0. use the normal approximation to the binomial for x > np0.08 vs p not = 0.5) − np0 (65 − 0.155 3-17 . n = 500.

0580079.77 P-Value = 0.06(1 − 0.025 = 1.645 0.05 − 0.067) + 1. Conclude there is no strong evidence to indicate a difference between the fraction nonconforming for the two processes.067 (b) Use α = 0.067(1 − 0.7842 −Zα/2 = −1.066667 Difference = p (1) .05) 200 + 0.05/2 = Z0.Chapter 3 Exercise Solutions 3-16.050 − 0.05(1 − 0. (a) n1 = 200. x2 = 20.052 ≤ ( p1 − p2 ) ≤ 0.067) 300 pˆ 1 (1 − pˆ 1 ) + pˆ 2 (1 − pˆ 2 ) n1 n2 ≤ ( p1 − p2 ) ≤ (0. x1 = 10. pˆ 2 = x2/n2 = 20/300 = 0.0166667 95% CI for difference: (-0.05(1 − 0.06) (1 200 + 1 300 ) = −0.05) 200 + 0.0246745) Test for difference = 0 (vs not = 0): Z = -0.05 n2 = 300.442 (c) ( pˆ 1 − pˆ 2 ) − Z α / 2 pˆ 1 (1 − pˆ 1 ) + pˆ 2 (1 − pˆ 2 ) n1 n2 ≤ ( p1 − p2 ) ≤ ( pˆ 1 − pˆ 2 ) + Z α / 2 (0.067(1 − 0.96 Zα/2 = Z0.p (2) Estimate for difference: -0.067 = 0.067) − 1. MTB > Stat > Basic Statistics > 2 Proportions > Summarized data Test and CI for Two Proportions Sample X N Sample p 1 10 200 0.96 Do not reject H0. Test H0: p1 = p2 versus H1: p1 ≠ p2.645 0.018 3-18 .05.06 n1 + n2 200 + 300 Z0 = pˆ1 − pˆ 2 pˆ (1 − pˆ ) (1 n1 + 1 n2 ) 0.05 − 0. Reject H0 if Z0 > Zα/2 or Z0 < –Zα/2 pˆ = x1 + x2 10 + 20 = = 0.050000 2 20 300 0. pˆ1 = x1/n1 = 10/200 = 0.067) 300 −0. 0.

18 = 2.85 − 8.69405 F-Test (normal distribution) Test statistic = 1.18 (a) Test H 0 : σ 12 = σ 22 versus H1 : σ 12 ≠ σ 22 .554 1 10 + 1 8 MTB > Stat > Basic Statistics > 2-Sample t > Summarized data Two-Sample T-Test and CI Sample N Mean StDev SE Mean 1 10 9. 10+8−2 = 1. tα.n2 −1 = F0.n1+n2−2 = t0. 3-19 . x1 = 9.18 = 1.7 = 4.* before: n1 = 10.55525 2.0987 < 4.88 Difference = mu (1) .83 2 8 8.85.n2 − 2 = F0. so do not reject H 0 MTB > Stat > Basic Statistics > 2 Variances > Summarized data Test for Equal Variances 95% Bonferroni confidence intervals for standard deviations Sample N Lower StDev Upper 1 10 1.2383 F0 = S12 S 22 = 6.60576 5.025.9.975.05. S 22 = 6.8232 and > 0.49 0.n1 −1.n1 −1. α = 0.746 SP = t0 = SP ( n1 − 1) S12 + ( n2 − 1) S22 n1 + n2 − 2 = (10 − 1) 6.n1 −1.70449 2.n1 −1.461 1 n1 + 1 n2 2. x2 = 8. (b) Test H0: µ1 = µ2 versus H1: µ1 > µ2.082 DF = 16 The mean impurity after installation of the new purification unit is not less than before.79 6.08 2. p-value = 0.77000 95% lower bound for difference: -0.5582 P-Value = 0.mu (2) Estimate for difference: 1.n2 −1 Fα / 2.46 Both use Pooled StDev = 2.0987 F0 = 1.79 + (8 − 1) 6.7 = 0.85 2.79 after: n2 = 8.05.922 The impurity variances before and after installation are the same.554 10 + 8 − 2 x1 − x2 9.10.9.24710 2 8 1.2383.48596 5.n2 − 2 or F0 < F1−α / 2.34856 T-Test of difference = 0 (vs >): T-Value = 1.Chapter 3 Exercise Solutions 3-17. F1−α / 2. Reject H0 if t0 > tα.n1+n2−2.08.08 = = 1. at α = 0.8232.61 0. S12 = 6.05 Reject H 0 if F0 > Fα / 2.

x1 = 175.80 3.4714) > −1.320 DF = 30 Note: For equal variances and sample sizes. test H0: µ1 − µ2 = − 5 versus H1: µ1 − µ2 < − 5. Reject H0 if Z0 < − Zα .8 − 181.8 psi.69978 T-Test of difference = -5 (vs <): T-Value = -0.50000 95% upper bound for difference: -3. x2 = 181.00 0.00 0. The P-values are close due to the sample sizes.47 Both use Pooled StDev = 3. So test a difference ∆0 = −5. P-value = Φ(Z0) = Φ(−0. n1 = 16. 3-20 .3) − (−5) = −0.0000 P-Value = 0.30 3.05 = −1. so do not reject H0.75 Difference = mu (1) .Chapter 3 Exercise Solutions 3-18.645. so H1: µ1 + 5 < µ2.3 psi. the Z-value is the same as the t-value. ∆0 = −5 −Zα = −Z0.75 2 16 181.0 psi Want to demonstrate that µ2 is greater than µ1 by at least 5 psi. σ1 = σ2 = 3.mu (2) Estimate for difference: -5.645 x x − ( 1 2 ) − ∆ 0 = (175.4714) = 0. n2 = 16. The mean strength of Design 2 does not exceed Design 1 by 5 psi.4714 Z0 = 32 16 + 32 16 σ 12 n1 + σ 22 n2 (Z0 = −0.3187 MTB > Stat > Basic Statistics > 2-Sample t > Summarized data Two-Sample T-Test and CI Sample N Mean StDev SE Mean 1 16 175.

005.001621 0.001311 0.0013112 ( ) n = −0.151167 0. MTB > Stat > Basic Statistics > Paired t > Samples in Columns Paired T-Test and CI: Ex3-19MC.8188.000417) T-Test of mean difference = 0 (vs not = 0): T-Value = -1.10 (|t0| = 1. j ) = 1 ⎡⎣( 0.001250. n1 + n2 − 2.000379 95% CI for mean difference: (-0.295 3-21 .151) + " + ( 0.10 P-Value = 0.8188 d=1 ( n∑ j =1 n xMicrometer.151583 0. 0.151 − 0.001311 12 = −1.000417 0.000241 Ex3-19VC 12 0. n1 + n2 − 2 = t0. so do not reject H0.Ex3-19VC N Mean StDev SE Mean Ex3-19MC 12 0. tα/2. Test H0: µd = 0 versus H1: µd ≠ 0.000835 0.Chapter 3 Exercise Solutions 3-19. Reject H0 if |t0| > tα/2.000417 0. There is no strong evidence to indicate that the two calipers differ in their mean measurements.22 = 2.152 ) ⎤⎦ = −0.10) < 2. j − xVernier.000417 12 ⎛ n ⎞ − d ⎜∑dj ⎟ ∑ j =1 ⎝ j =1 ⎠ S d2 = ( n − 1) n 2 j t0 = d (S d ) 2 n = 0.150 − 0.000468 Difference 12 -0. Ex3-19VC Paired T for Ex3-19MC .

05 (a) Test H0: σ2 = 1 versus H1: σ2 < 1. s = 11. Reject H0 if χ20 < χ21-α. χ21-α/2.19 3-22 .Chapter 3 Exercise Solutions 3-20.5.14 ≤ σ ≤ 2.19 = 8. (a) The alternative hypothesis H1: µ > 150 is preferable to H1: µ < 150 we desire a true mean weld strength greater than 150 psi. s = 1.91 1.7. x = 153.025.4389) < 1. The standard deviation of the fill volume is not less than 1ml. so do not reject H0.254 T 1. n-1 = χ20.7291.75 χ20 = 42.52 32. x = 752.85 ≤ σ 2 ≤ (20 − 1)1. t0 = ( x − µ ) S n = (153.05 Test H0: µ = 150 versus H1: µ > 150.19 = 1. χ21-α. µ > 150. n −1 = t0.44 P 0. n −1. H0 must be rejected in favor of the alternative H1.7 − 150 ) 11. Reject H0 if t0 > tα.80 1.1170.n −1 ≤ σ 2 ≤ (n − 1) S 2 χ12−α / 2. tα.19 = 10.6 ml.91. (b) n = 20.05. α = 0.75 > 10. n = 20.571 95% Lower Bound 149. (b) χ2α/2.7291. n-1.4389 ( ) ( ) (t0 = 1. There is insufficient evidence to indicate that the mean strength is greater than 150 psi.52 8.85. n-1 = χ20.1170 χ 02 = ⎡⎣(n − 1) S 2 ⎤⎦ σ 02 = ⎡⎣(20 − 1)1.19 = 32. so do not reject H0. n-1 = χ20.52 ⎤⎦ 1 = 42. MTB > Stat > Basic Statistics > 1-Sample t > Summarized data One-Sample T Test of mu = 150 vs > 150 N 20 Mean 153.5.700 StDev 11.5 20 = 1.975.500 SE Mean 2.30 ≤ σ 2 ≤ 4.95.n −1 (20 − 1)1. In order to achieve this result. (n − 1) S 2 χα2 / 2.083 3-21. α = 0.

172 Percent 80 70 60 50 40 30 20 10 5 1 748 750 752 754 Fill Volume.2 (c) MTB > Graph > Probability Plot > SIngle Probability Plot of Pinot Gris Fill Volume (Ex3-21) Normal .37 0.95% CI 99 Mean StDev N AD P-Value 95 90 752.51 0.17 2.00 95% C onfidence Interv al for S tD ev 95% Confidence Intervals 1. 3-23 . ml 756 758 The plotted points do not fall approximately along a straight line.00 753.281321 0.4 752.191843 20 M inimum 1st Q uartile M edian 3rd Q uartile M aximum 756 750.55 1.54 2.8 753.172 M ean S tD ev V ariance S kew ness Kurtosis N 752.00 751.Chapter 3 Exercise Solutions 3-21 (b) continued MTB > Stat > Basic Statistics > Graphical Summary Summary for Pinot Gris Fill Volume.00 756.00 95% C onfidence Interv al for M ean 751.538 20 0.511 0.00 753.25 Mean Median 752.0 752.83 753.27 95% C onfidence Interv al for M edian 752.6 1. so the assumption that battery life is normally distributed is not appropriate.25 753. ml (Ex3-21) A nderson-D arling N ormality Test 749 750 751 752 753 754 755 A -S quared P -V alue 0.

σ2 = 9.05 Assumed standard deviation = 3 Sample Target Difference Size Power Actual Power 5 4 0. β.05. α = 0.Chapter 3 Exercise Solutions 3-22.0000 = 0. From Eqn. What n is needed such that the Type II error. µ0 = 15.3733) − Φ (−5. µ1 = 20. is less than or equal to 0.10? δ = µ1 − µ2 = 20 − 15 = 5 d = δ σ = 5 9 = 1.05. the operating characteristic curve for two-sided at α = 0. Test H0: µ = 15 versus H1: µ ≠ 15.9 0.0. So. 3-46. n = 4. then Φ − Zα / 2 − δ n σ is likely to be small compared with β.96 − 5 4 3 ( ) ( ) ( ) ( ) = Φ (−1. ( β ≈ Φ Zα / 2 − δ n σ ( ) Φ ( β ) ≈ Φ −1 Zα / 2 − δ n σ ) − Z β ≈ Zα / 2 − δ n σ n ≈ ⎡⎣( Zα / 2 + Z β )σ δ ⎤⎦ 2 3-24 . Let µ1 = µ0 + δ.96 − 5 4 3 − Φ −1. β = Φ Zα / 2 − δ n σ − Φ − Zα / 2 − δ n σ ( ) ( ) ( ) If δ > 0.0848 − 0.6667 From Figure 3-7. Check: β = Φ Zα / 2 − δ n σ − Φ − Zα / 2 − δ n σ = Φ 1.915181 3-23.2933) = 0.0848 MTB > Stat > Power and Sample Size > 1-Sample Z Power and Sample Size 1-Sample Z Test Testing mean = null (versus not = null) Calculating power for mean = null + difference Alpha = 0.

x1 − x2 Maximize: Z 0 = Subject to: n1 + n2 = N . an equivalent statement is Minimize: L = σ 12 n1 + σ 22 n2 = σ 12 n1 + σ 22 N − n1 σ 22 ⎞ dL ⎡ −1 2 dL ⎛ σ 12 −1 + n1 σ 1 + ( N − n1 ) σ 22 ⎤ ⎜ ⎟= ⎦ dn1 ⎝ n1 N − n1 ⎠ dn1 ⎣ = −1n1−2σ 12 + (−1)(−1) ( N − n1 ) σ 22 = 0 −2 =− σ 12 n12 + σ 22 ( N − n1 ) 2 =0 n1 σ 1 = n2 σ 2 Allocate N between n1 and n2 according to the ratio of the standard deviations. reject H 0 if Z 0 > Zα / 2 3-25 . Assume µ1 = 2µ2 and let Q = ( x1 − x2 ) . σ 1 n1 + σ 2 n2 2 2 Since ( x1 − x2 ) is fixed. x2 . x1 independent of x2 . x1 . n2 . E (Q) = E ( x1 − 2 x2 ) = µ1 − 2µ2 = 0 var(Q) = var( x1 − 2 x2 ) = var( x1 ) + var(2 x2 ) = var( x1 ) + 22 var( x2 ) = Z0 = var( x1 ) var( x2 ) +4 n1 n2 Q−0 x1 − 2 x2 = SD(Q) σ 12 n1 + 4 σ 22 n2 And. Given x ~ N . n1 . 3-25.Chapter 3 Exercise Solutions 3-24.

n = 1000. x = 3.15 versus H1: λ ≠ 0. Zα/2 = Z0. at α = 0.15 100 = −1.01.3162) < 1. x2.3162 (Z0 = 0.5758. Zα/2 = Z0.5758 Z 0 = ( x − λ0 ) λ0 n = ( 0. x = x/n = ~ N(λ. (a) Wish to test H0: λ = λ0 versus H1: λ ≠ λ0. Zα = Z0.Chapter 3 Exercise Solutions 3-26. so do not reject H0.96.110 Test H0: λ = 0.005 = 2. if n is large. 3-28. x ~ Poi(λ).6 Test H0: λ = 0. x = x/N = 11/100 = 0.15. x = x/N = 3/5 = 0.15 ) 0.5 5 = 0.96 Z 0 = ( x − λ0 ) λ0 n = ( 0. at α = 0. so reject H0. Z 0 = ( x − λ ) λ0 / n . 3-27. …. x = 688.5.110 − 0. Reject H0: λ = λ0 if |Z0| > Zα/2 (b) x ~ Poi(λ). n = 100.0328 (|Z0| = 1. xn. at α = 0.8663) > 1. λ/n).645. x = 11. n = 5.5 ) 0. n ∑x i ~ POI( nλ ) . x ~ Poi(λ). i =1 Using the normal approximation to the Poisson.5 versus H1: λ > 0.025 = 1. Select random sample of n observations x1.645 Z 0 = ( x − λ0 ) λ0 n = ( 0. 3-26 .688 Test H0: λ = 1 versus H1: λ ≠ 1.05 = 1.688 − 1) 1 1000 = −9.05.05.6 − 0.0328) < 2. so do not reject H0. Each xi ~ POI(λ).8663 (|Z0| = 9. Reject H0 if |Z0| > Zα/2. x = x/N = 688/1000 = 0. Reject H0 if |Z0| > Zα. Reject H0 if Z0 > Zα.

05.7132 DF SS MS F P 2 3. (b) MTB > Stat > ANOVA > One-Way > Graphs.7132 (F0.3167 0. However.7600 (---------*----------) 160 6 4. 3-27 .0 4.80 Pooled StDev = 0.60 4.15 = 3.2.5231 (----------*---------) 200 6 3.4167 0. the P-value is just slightly greater than 0.5 3.34% R-Sq(adj) = 23. (a) MTB > Stat > ANOVA > One-Way One-way ANOVA: Ex3-29Obs versus Ex3-29Flow Source Ex3-29Flow Error Total S = 0.630 0. Boxplots of data MTB > Graph > Boxplot > One Y. With Groups Boxplot of Etch Uniformity by C2F6 Flow Etch Uniformity (%) 5.0 3.05.5 4.9333 0.Chapter 3 Exercise Solutions 3-29.5 125 160 C2F6 Flow (SCCM) 200 Gas flow rate of 125 SCCM gives smallest mean percentage uniformity.509 17 11.648 1. so there is some evidence that gas flow rate affects the etch uniformity.05.20 4.824 3.59 0. so flow rate does not affect etch uniformity at a significance level α = 0.0 2.6823) > (F0 = 3.8214 (----------*---------) -----+---------+---------+---------+---3.053 15 7.32% Individual 95% CIs For Mean Based on Pooled StDev Level N Mean StDev -----+---------+---------+---------+---125 6 3.59).278 R-Sq = 32.00 3.

Normal plot of residuals Normal Probability Plot of the Residuals (response is Etch Uniformity (Ex3-29Obs)) 99 95 90 Percent 80 70 60 50 40 30 20 10 5 1 -2 -1 0 Residual 1 2 The normality assumption is reasonable.5 -1.Chapter 3 Exercise Solutions 3-29 continued (c) MTB > Stat > ANOVA > One-Way > Graphs.0 4. 3-28 .0 3.0 0.4 Residuals are satisfactory.6 3.2 3.4 3.0 -0.5 Residual 1.8 Fitted Value 4. (d) MTB > Stat > ANOVA > One-Way > Graphs. Residuals versus fits Residuals Versus the Fitted Values (response is Etch Uniformity (Ex3-29Obs)) 1.2 4.5 0.

say. The statistically significant difference between the mean uniformities can be seen by centering the t distribution between. Flow Rate 125 160 200 scale factor = Mean Etch Uniformity 3.Chapter 3 Exercise Solutions 3-30.9 (1 6 0 ) 4 .3% 4.5087 6 = 0.2 4 .8 M e a n E t c h U n if o r m it y The graph does not indicate a large difference between the mean etch uniformity of the three different flow rates.9% MSE n = 0.3 S c a le d t D is t r ib u t io n (2 0 0 ) (1 2 5 ) 3 .6 3 .3 3 . 125 and 200. 3-29 . and noting that 160 would fall beyond the tail of the curve.5 4 .4% 3.0 3 .

3-30 .09% Individual 95% CIs For Mean Based on Pooled StDev Level N Mean StDev ----+---------+---------+---------+----10 3 1500.0 (-----------*----------) ----+---------+---------+---------+----1440 1520 1600 1680 Pooled StDev = 71.7 (-----------*-----------) 20 3 1606. (b) Boxplot of Compressive Strength by Rodding Level 1750 Compressive Strength 1700 1650 1600 1550 1500 1450 1400 10 15 20 25 Rodding Level Level 25 exhibits considerably less variability than the other three levels.Chapter 3 Exercise Solutions 3-31.0 52.0 10.214 8 40933 5117 11 69567 R-Sq = 41.05. (a) MTB > Stat > ANOVA > One-Way > Graphs> Boxplots of data.9 (-----------*-----------) 25 3 1500.5 No difference due to rodding level at α = 0.16% R-Sq(adj) = 19.0 (-----------*----------) 15 3 1586. Normal plot of residuals One-way ANOVA: Ex3-31Str versus Ex3-31Rod Source Ex3-31Rod Error Total S = 71.87 0.7 107.7 77.53 DF SS MS F P 3 28633 9544 1.

3-31 .Chapter 3 Exercise Solutions 3-31 continued (c) Normal Probability Plot of the Residuals (response is Compressive Strength (Ex3-31Str)) 99 95 90 Percent 80 70 60 50 40 30 20 10 5 1 -150 -100 -50 0 Residual 50 100 150 The normal distribution assumption for compressive strength is reasonable.

Rodding Level 10 15 20 25 scale factor = Mean Compressive Strength 1500 1587 1607 1500 MSE n = 5117 3 = 41 S c a le d t D is tr ib u tio n (10 . 25 ) (1 5 ) (2 0 ) 1418 1459 1500 1541 1582 1623 1664 M e a n C o m p re s s iv e S tr e n g th There is no difference due to rodding level.Chapter 3 Exercise Solutions 3-32. 3-32 .

583 0.8 -0.6 -0.6 0.2 0.700 0.097 0.333 0.50 41. (b) Normal Probability Plot of the Residuals (response is Baked Density (Ex3-33Den)) 99 95 90 Percent 80 70 60 50 40 30 20 10 5 1 -0.89% R-Sq(adj) = 5.2 0.497 (----------*----------) --------+---------+---------+---------+41.00 Pooled StDev = 0.0 Residual 0.194 (----------*----------) 550 6 41.8 Normality assumption is reasonable. Normal plot of residuals One-way ANOVA: Ex3-33Den versus Ex3-33T Source DF Ex3-33T 3 Error 20 Total 23 S = 0.57% Individual 95% CIs For Mean Based on Pooled StDev Level N Mean StDev --------+---------+---------+---------+500 6 41.152 1.Chapter 3 Exercise Solutions 3-33. (a) MTB > Stat > ANOVA > One-Way > Graphs> Boxplots of data.105 2.450 0.339 (----------*----------) 575 6 41.258 2.324 Temperature level does not significantly affect mean baked anode density.4 -0.45 0.553 R-Sq = 17. 3-33 .25 41.75 42.457 0.4 0.3238 SS MS F P 0.141 (----------*----------) 525 6 41.

00 -0. Recommend 500°C for smallest variability.25 0.50 41. 3-34. 500°C (see Boxplot).00 40.50 500 525 550 Firing Temperature.75 41.00 Baked Desnity 41.50 Residual 0.25 -0.75 500 510 520 530 540 550 Temperature (deg C) 560 570 580 As firing temperature increases.50 -0. MTB > Stat > ANOVA > One-Way > Graphs> Residuals versus the Variables Residuals Versus Firing Temperature (Ex3-33T) (response is Baked Density (Ex3-33Den)) 0.Chapter 3 Exercise Solutions 3-33 continued (c) Boxplot of Baked Density by Firing Temperature 42. deg C 575 Since statistically there is no evidence to indicate that the means are different. More uniform anodes are produced at lower temperatures. which probably also incurs the smallest cost (lowest temperature).25 41.75 40. select the temperature with the smallest variance. 3-34 . so does variability.

(a) MTB > Stat > ANOVA > One-Way > Graphs> Boxplots of data One-way ANOVA: Ex3-35Rad versus Ex3-35Dia Source Ex3-35Dia Error Total S = 2.40 1.304 (----*---) 1.000 1.51 0.711 Orifice size does affect mean % radon release. % (Ex3-35Rad) 85 80 75 70 65 60 0.51 4 77.711 DF SS MS F P 5 1133.38 226.309 (---*---) 0.99 and 1.826 (---*---) 1.000 3.35 23 1265.85 0. Boxplot of Radon Released by Orifice Diameter Radon Released.4 orifice diameters.02 Orifice Diameter 1. 3-35 .750 2.000 2.37 0.0 Pooled StDev = 2.0 84.754 (---*---) ----+---------+---------+---------+----63.05.71 4 75.0 70.63 R-Sq = 89.02 4 71.25 7.99 4 62. at α = 0.0 77.750 2.71 1.Chapter 3 Exercise Solutions 3-35.559 (---*---) 1.65% Individual 95% CIs For Mean Based on Pooled StDev Level N Mean StDev ----+---------+---------+---------+----0.750 3.000 18 132.37 4 82.55% R-Sq(adj) = 86.68 30.99 Smallest % radon released at 1.40 4 65.062 (---*---) 0.

02 1.5 -5. Residuals Versus the Fitted Values (response is Radon Released (Ex3-35Rad)) 5.0 0.0 Residuals violate the normality distribution.99 The assumption of equal variance at each factor level appears to be violated.0 Residual 2. Residuals versus the Variables Normal Probability Plot of the Residuals (response is Radon Released ( Ex3-35Rad)) 99 95 90 Percent 80 70 60 50 40 30 20 10 5 1 -5.02. Residuals versus fits.0 Residual 2.71 1. 1.40.0 -2.5 -5.99).Chapter 3 Exercise Solutions 3-35 continued (b) MTB > Stat > ANOVA > One-Way > Graphs> Normal plot of residuals. with larger variances at the larger diameters (1.5 0. 3-36 .0 60 65 70 75 Fitted Value--Radon Released 80 85 Variability in residuals does not appear to depend on the magnitude of predicted (or fitted) values.40 Orifice Diameter (Ex3-35Dia) 1. 1.37 0.5 5.0 Residual 2.0 -2.51 0. Residuals Versus Orifice Diameter (Ex3-35Dia) (response is Radon Released (Ex3-35Rad)) 5.5 0.5 0.0 -2.

4366 (------*------) 4 3 1.5 6.0 4.3067 1. Boxplot of Uniformity by Wafer Position Film Thickness Uniformity (Ex3-36Un) 6 5 4 3 2 1 1 (b) σˆτ2 = 2 3 Wafer Position (Ex3-36Pos) 4 MSfactor − MSE 5.53% Individual 95% CIs For Mean Based on Pooled StDev Level N Mean StDev --------+---------+---------+---------+1 3 4. and 4.437 S = 0. 3.0 Pooled StDev = 0.407 8.008 Error 8 5.0484 3-37 .Chapter 3 Exercise Solutions 3-36.29 0.66% R-Sq(adj) = 66.3962 n 12 (c) σˆ 2 = MSE = 0.3570 (------*------) --------+---------+---------+---------+1.5 3.217 0.8076 R-Sq = 75.4636 (------*------) 2 3 1.6522 2 σˆ uniformity = σˆτ2 + σˆ 2 = 0.6522 = 1.3167 0. (a) MTB > Stat > ANOVA > One-Way > Graphs.3962 + 0.9267 0.6522 = = 0.7733 0.3853 (------*------) 3 3 1.8076 There is a statistically significant difference in wafer position. 1 is different from 2.652 Total 11 21.220 5. Boxplots of data One-way ANOVA: Ex3-36Un versus Ex3-36Pos Source DF SS MS F P Ex3-36Pos 3 16.4066 − 0.

5 1.5 0.5 0.5 -1. Residuals versus the Variables Normal Probability Plot of the Residuals (response is Uniformity (Ex3-36Un)) 99 95 90 Percent 80 70 60 50 40 30 20 10 5 1 -2 -1 0 Residual 1 2 Normality assumption is probably not unreasonable.5 3.5 1.0 2.0 1.5 The variability in residuals does appear to depend on the magnitude of predicted (or fitted) values.5 -1.0 4. 3-38 .0 -0. Residuals versus fits. Residuals Versus the Fitted Values (response is Uniformity ( Ex3-36Un)) 1.5 2.0 -0.0 Residual 0.Chapter 3 Exercise Solutions 3-36 continued (d) MTB > Stat > ANOVA > One-Way > Graphs> Normal plot of residuals. but there are two very unusual observations – the outliers at either end of the plot – therefore model adequacy is questionable.5 Fitted Value--Film Thickness Uniformity 4.0 3.5 1 2 3 Wafer Position (Ex3-36Pos) 4 Both outlier residuals are from wafer position 1. Residuals Versus Wafer Position (Ex3-36Pos) (response is Film Thickness Uniformity (Ex3-36Un)) 1.0 Residual 0.0 -1.0 -1.5 1.

Chapter 4 Exercise Solutions
Several exercises in this chapter differ from those in the 4th edition. An “*” following the
exercise number indicates that the description has changed. New exercises are denoted
with an “☺”. A second exercise number in parentheses indicates that the exercise
number has changed.
4-1.
“Chance” or “common” causes of variability represent the inherent, natural variability of
a process - its background noise. Variation resulting from “assignable” or “special”
causes represents generally large, unsatisfactory disturbances to the usual process
performance. Assignable cause variation can usually be traced, perhaps to a change in
material, equipment, or operator method.
A Shewhart control chart can be used to monitor a process and to identify occurrences of
assignable causes. There is a high probability that an assignable cause has occurred when
a plot point is outside the chart's control limits. By promptly identifying these
occurrences and acting to permanently remove their causes from the process, we can
reduce process variability in the long run.

4-2.
The control chart is mathematically equivalent to a series of statistical hypothesis tests. If
a plot point is within control limits, say for the average x , the null hypothesis that the
mean is some value is not rejected. However, if the plot point is outside the control
limits, then the hypothesis that the process mean is at some level is rejected. A control
chart shows, graphically, the results of many sequential hypothesis tests.

NOTE TO INSTRUCTOR FROM THE AUTHOR (D.C. Montgomery):
There has been some debate as to whether a control chart is really equivalent to
hypothesis testing. Deming (see Out of the Crisis, MIT Center for Advanced
Engineering Study, Cambridge, MA, pp. 369) writes that:
“Some books teach that use of a control chart is test of hypothesis: the process is
in control, or it is not. Such errors may derail self-study”.
Deming also warns against using statistical theory to study control chart behavior (falsealarm probability, OC-curves, average run lengths, and normal curve probabilities.
Wheeler (see “Shewhart’s Charts: Myths, Facts, and Competitors”, ASQC Quality
Congress Transactions (1992), Milwaukee, WI, pp. 533–538) also shares some of these
concerns:
“While one may mathematically model the control chart, and while such a model
may be useful in comparing different statistical procedures on a theoretical basis,
these models do not justify any procedure in practice, and their exact
probabilities, risks, and power curves do not actually apply in practice.”

4-1

Chapter 4 Exercise Solutions
4-2 continued
On the other hand, Shewhart, the inventor of the control chart, did not share these views
in total. From Shewhart (Statistical Method from the Viewpoint of Quality Control
(1939), U.S. Department of Agriculture Graduate School, Washington DC, p. 40, 46):
“As a background for the development of the operation of statistical
control, the formal mathematical theory of testing a statistical hypothesis
is of outstanding importance, but it would seem that we must continually
keep in mind the fundamental difference between the formal theory of
testing a statistical hypothesis and the empirical theory of testing a
hypothesis employed in the operation of statistical control. In the latter,
one must also test the hypothesis that the sample of data was obtained
under conditions that may be considered random. …
The mathematical theory of distribution characterizing the formal and
mathematical concept of a state of statistical control constitutes an
unlimited storehouse of helpful suggestions from which practical criteria
of control must be chosen, and the general theory of testing statistical
hypotheses must serve as a background to guide the choice of methods of
making a running quality report that will give the maximum service as
time goes on.”
Thus Shewhart does not discount the role of hypothesis testing and other aspects of
statistical theory. However, as we have noted in the text, the purposes of the control
chart are more general than those of hypothesis tests. The real value of a control chart is
monitoring stability over time. Also, from Shewhart’s 1939 book, (p. 36):
“The control limits as most often used in my own work have been set so that after
a state of statistical control has been reached, one will look for assignable causes
when they are not present not more than approximately three times in 1000
samples, when the distribution of the statistic used in the criterion is normal.”
Clearly, Shewhart understood the value of statistical theory in assessing control chart
performance.
My view is that the proper application of statistical theory to control charts can provide
useful information about how the charts will perform. This, in turn, will guide decisions
about what methods to use in practice. If you are going to apply a control chart
procedure to a process with unknown characteristics, it is prudent to know how it will
work in a more idealized setting. In general, before recommending a procedure for use in
practice, it should be demonstrated that there is some underlying model for which it
performs well. The study by Champ and Woodall (1987), cited in the text, that shows the
ARL performance of various sensitizing rules for control charts is a good example. This
is the basis of the recommendation against the routine use of these rules to enhance the
ability of the Shewhart chart to detect small process shifts.

4-2

Chapter 4 Exercise Solutions
4-3.
Relative to the control chart, the type I error represents the probability of concluding the
process is out of control when it isn't, meaning a plot point is outside the control limits
when in fact the process is still in control. In process operation, high frequencies of false
(and increased variability), and lack of credibility for SPC methods.
The type II error represents the probability of concluding the process is in control, when
actually it is not; this results from a plot point within the control limits even though the
process mean has shifted out of control. The effect on process operations of failing to
detect an out-of-control shift would be an increase in non-conforming product and
associated costs.

4-4.
The statement that a process is in a state of statistical control means that assignable or
special causes of variation have been removed; characteristic parameters like the mean,
standard deviation, and probability distribution are constant; and process behavior is
predictable. One implication is that any improvement in process capability (i.e., in terms
of non-conforming product) will require a change in material, equipment, method, etc.

4-5.
No. The fact that a process operates in a state of statistical control does not mean that
nearly all product meets specifications. It simply means that process behavior (mean and
variation) is statistically predictable. We may very well predict that, say, 50% of the
product will not meet specification limits! Capability is the term, which refers to the
ability to meet product specifications, and a process must be in control in order to
calculate capability.

4-6.
The logic behind the use of 3-sigma limits on Shewhart control charts is that they give
good results in practice. Narrower limits will result in more investigations for assignable
causes, and perhaps more false alarms. Wider limits will result in fewer investigations,
but perhaps fewer process shifts will be promptly identified.
Sometimes probability limits are used - particularly when the underlying distribution of
the plotted statistic is known. If the underlying distribution is unknown, care should be
exercised in selecting the width of the control limits. Historically, however, 3-sigma
limits have been very successful in practice.

4-3

Chapter 4 Exercise Solutions
4-7.
Warning limits on control charts are limits that are inside the control limits. When
warning limits are used, control limits are referred to as action limits. Warning limits,
say at 2-sigma, can be used to increase chart sensitivity and to signal process changes
more quickly than the 3-sigma action limits. The Western Electric rule, which addresses
this type of shift is to consider a process to be out of control if 2 of 3 plot points are
between 2 sigma and 3 sigma of the chart centerline.

4-8.
The concept of a rational subgroup is used to maximize the chance for detecting variation
between subgroups. Subgroup samples can be structured to identify process shifts. If it
is expected that a process will shift and stay at the new level until a corrective action,
then sampling consecutive (or nearly) units maximizes the variability between subgroups
and minimizes the variability within a subgroup. This maximizes the probability of
detecting a shift.

4-9.
I would want assignable causes to occur between subgroups and would prefer to select
samples as close to consecutive as possible. In most SPC applications, process changes
will not be self-correcting, but will require action to return the process to its usual
performance level. The probability of detecting a change (and therefore initiating a
corrective action) will be maximized by taking observations in a sample as close together
as possible.

4-10.
This sampling strategy will very likely underestimate the size of the true process
variability. Similar raw materials and operating conditions will tend to make any fivepiece sample alike, while variability caused by changes in batches or equipment may
remain undetected. An out-of-control signal on the R chart will be interpreted to be the
result of differences between cavities. Because true process variability will be
underestimated, there will likely be more false alarms on the x chart than there should
be.

4-4

Chapter 4 Exercise Solutions
4-11.
(a)
No.
(b)
The problem is that the process may shift to an out-of-control state and back to an incontrol state in less than one-half hour. Each subgroup should be a random sample of all
parts produced in the last 2½ hours.

4-12.
No. The problem is that with a slow, prolonged trend upwards, the sample average will
tend to be the value of the 3rd sample --- the highs and lows will average out. Assume
that the trend must last 2½ hours in order for a shift of detectable size to occur. Then a
better sampling scheme would be to simply select 5 consecutive parts every 2½ hours.

4-13.
No. If time order of the data is not preserved, it will be impossible to separate the
presence of assignable causes from underlying process variability.
4-14.
An operating characteristic curve for a control chart illustrates the tradeoffs between
sample size n and the process shift that is to be detected. Generally, larger sample sizes
are needed to increase the probability of detecting small changes to the process. If a large
shift is to be detected, then smaller sample sizes can be used.

4-15.
The costs of sampling, excessive defective units, and searches for assignable causes
impact selection of the control chart parameters of sample size n, sampling frequency h,
and control limit width. The larger n and h, the larger will be the cost of sampling. This
sampling cost must be weighed against the cost of producing non-conforming product.

4-16.
Type I and II error probabilities contain information on statistical performance; an ARL
results from their selection. ARL is more meaningful in the sense of the operations
information that is conveyed and could be considered a measure of the process
performance of the sampling plan.

4-5

Chapter 4 Exercise Solutions
4-17.
Evidence of runs, trends or cycles? NO. There are no runs of 5 points or cycles. So, we
can say that the plot point pattern appears to be random.

4-18.
Evidence of runs, trends or cycles? YES, there is one "low - high - low - high" pattern
(Samples 13 – 17), which might be part of a cycle. So, we can say that the pattern does
not appear random.

4-19.
Evidence of runs, trends or cycles? YES, there is a "low - high - low - high - low" wave
(all samples), which might be a cycle. So, we can say that the pattern does not appear
random.

4-20.
Three points exceed the 2-sigma warning limits - points #3, 11, and 20.

4-21.
Check:
• Any point outside the 3-sigma control limits? NO.
• 2 of 3 beyond 2 sigma of centerline? NO.
• 4 of 5 at 1 sigma or beyond of centerline? YES. Points #17, 18, 19, and 20 are
outside the lower 1-sigma area.
• 8 consecutive points on one side of centerline? NO.
One out-of-control criteria is satisfied.

4-22.
Four points exceed the 2-sigma warning limits - points #6, 12, 16, and 18.

4-23.
Check:
• Any point outside the 3-sigma control limits? NO. (Point #12 is within the lower
3-sigma control limit.)
• 2 of 3 beyond 2 sigma of centerline? YES, points #16, 17, and 18.
• 4 of 5 at 1 sigma or beyond of centerline? YES, points #5, 6, 7, 8, and 9.
• 8 consecutive points on one side of centerline? NO.
Two out-of-control criteria are satisfied.

4-6

Chapter 4 Exercise Solutions
4-24.
The pattern in Figure (a) matches the control chart in Figure (2).
The pattern in Figure (b) matches the control chart in Figure (4).
The pattern in Figure (c) matches the control chart in Figure (5).
The pattern in Figure (d) matches the control chart in Figure (1).
The pattern in Figure (e) matches the control chart in Figure (3).

4-25 (4-30).
Many possible solutions.
MTB > Stat > Quality Tools > Cause-and-Effect

Cause-and-Effect Diagram for Late Arrival
Driv e

Family

"Turtle"

Children/School

Route

Put out pet

Accident

Children/Homework

Fix breakfast

A rriv e late to
O ffice

Errands

Fix lunch
Eat breakfast

Carpool

Dress

Gas

Shower
Get up late

Activ ities

Coffee

Stops

4-7

Chapter 4 Exercise Solutions
4-26 (4-31).
Many possible solutions.
MTB > Stat > Quality Tools > Cause-and-Effect

Cause-and-Effect Diagram for Car Accident
Driv er

Car

A sleep

Tires

Drunk
Brakes
M isjudgment
S uspension

Talking on cell phone
D istracted

S teering

S tate of Repair

Raining

Blocked

P oor v isibility

Windy

W eather

Out-of-contr
ol car strikes
tree

Icy /snow -cov ered

4-8

Chapter 4 Exercise Solutions
4-27 (4-32).
Many possible solutions.
MTB > Stat > Quality Tools > Cause-and-Effect

Cause-and-Effect Diagram for Damaged Glassware
Glassware

Glassware
Pack aging

Deliv ery Serv ice
Handling

Crushed

Strength flaw

Dropped

Severe transport vibration

Dropped

Crushed

Internal Handling

Weak box

Broken at start

Glassware
Damaged

Droppped

Carelessly packed

Manufacturer Handling

4-9

MTB > Stat > Quality Tools > Cause-and-Effect Cause-and-Effect Diagram for Coffee-making Process Personnel Method Machine Worn-out C offee drinkers C leanliness E spresso drinkers Brew temperature Insufficient training A ge of brew Brew method C onsistently Bad C offee Ty pe of filter C offee grind A mount of w ater C offee roast C offee beans Water temperature Env ironment A mount of beans Measurement Water source Material 4-10 . Many possible solutions.Chapter 4 Exercise Solutions 4-28☺.

Snooze No Check time Awake Yes 6:30am ? Get out of bed Arrive at work … 4-31☺.8 95.1 g in iss s as Cl e tM No 37 26. Many possible solutions.7 r he Ot 0 6 4.1 55.9 g in et C m om itm t en 19 13. Yellow is nonvalue-added activity.0 To reduce total count of defects.8 81. 4-11 . The next step would be to determine the causes of “Being Rude” and to work on eliminating those causes. 1 2 0 2 10 11 4 2 4 6 3 1 9 2 3 4 0 9 2 2 Month/Day 5 6 7 1 0 1 7 10 11 1 0 1 7 9 4 TOTAL 18 21 15 13 16 19 … … … … … 31 TOTAL 1 6 9 76 7 19 2 37 17 19 138 Pareto Chart of Personal Opportunities for Improvement 140 100 120 Count 80 60 60 40 40 20 20 0 Defect g in Be Count Percent Cum % Percent 80 100 de Ru M 76 55. Example of a check sheet to collect data on personal opportunities for improvement. Defect Overeating Being Rude Not meeting commitments Missing class Etc. including defect categories and counts. “Being Rude” represents the greatest opportunity to make an improvement.3 100. beginning and end of process are shown below.Chapter 4 Exercise Solutions 4-29☺. green is value-added activity. Many possible solutions.

922093 m = 50 ⎛ 50 ⎞ 0 50 ⎟ (0.0027)0 (1 − 0.0526 ⎝0⎠ α1 = 1 − Pr{0 of 20 beyond} = 1 − ⎜ Cumulative Distribution Function Binomial with n = 20 and p = 0.947363 m = 30 ⎛ 30 ⎞ 0 30 ⎟ (0.0134 ⎝0⎠ MTB > Calc > Probability Distributions > Binomial.873556 Although the probability that a single point plots beyond the control limits is 0.0027) = 0. the probability that at least one of the points is beyond the limits also increases.0027 x P( X <= x ) 0 0.0027.973326 m = 20 ⎛ 20 ⎞ 0 20 ⎟ (0.0027 x P( X <= x ) 0 0.Chapter 4 Exercise Solutions 4-32☺.986573 m = 10 ⎛10 ⎞ 0 10 ⎟ (0.0027 x P( X <= x ) 0 0.0027) = 0.0027) (1 − 0. Cumulative Probability Cumulative Distribution Function Binomial with n = 5 and p = 0.0027) (1 − 0.0027 x P( X <= x ) 0 0.0027 x P( X <= x ) 0 0.9866 = 0.0267 ⎝0⎠ α1 = 1 − Pr{0 of 10 beyond} = 1 − ⎜ Cumulative Distribution Function Binomial with n = 10 and p = 0.9733 = 0. 4-12 . m=5 α1 = Pr{at least 1 out-of-control} = Pr{1 of 5 beyond} + Pr{2 of 5 beyond} + " + Pr{5 of 5 beyond} ⎛5⎞ = 1 − Pr{0 of 5 beyond} = 1 − ⎜ ⎟ (0.1025 ⎝0⎠ α1 = 1 − Pr{0 of 50 beyond} = 1 − ⎜ Cumulative Distribution Function Binomial with n = 50 and p = 0.0027) (1 − 0.0779 0 ⎝ ⎠ α1 = 1 − Pr{0 of 30 beyond} = 1 − ⎜ Cumulative Distribution Function Binomial with n = 30 and p = 0.0027) = 0.0027)5 = 1 − 0. as the number of samples increases (m).0027) (1 − 0.0027) = 1 − 0.

Chapter 4 Exercise Solutions 4-33☺. 4-13 . the points used to estimate these sample statistics are not independent—they do not reflect a random sample from a population. In fact. When the process mean µ and variance σ2 are unknown. However. they must be estimated by sample means x and standard deviations s. sampling frequencies are often designed to increase the likelihood of detecting a special or assignable cause. The lack of independence in the sample statistics will affect the estimates of the process population parameters.

71) = 9.0 12 10 15 37. 5-1.71 m 24 UCL x = x + A2 R = 34. A2 = 0.00 + 0. a run of n consecutive points on one side of the center line is defined as 9 points.29 UCL R = D4 R = 2.0 12 39. This can be changed under Tools > Options > Control Charts and Quality Tools > Define Tests.71 LCL = 31.2 x= 1 2 = = 34.2 + " + 34. The MINITAB convention for determining whether a point is out of control is: (1) if a plot point is within the control limits.00 m 24 R + R2 + " + Rm 3 + 4 + " + 2 R= 1 = = 4. it is out of control. An “*” indicates that the description has changed. In particular.71) = 36.0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Sample No.Chapter 5 Exercise Solutions Notes: 1. or (2) if a plot point is on or beyond the limits.0 LCL = 0 0 27.114.71) = 0. A second exercise number in parentheses indicates that the exercise number has changed. (a) for n = 5.0 2 29.0 6 33.96 UCL = 36.00 LCL x = x − A2 R = 34. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Sample No.72 CL x = x = 34. D4 = 2.0 UCL = 9.29 4 31.00 R x-bar 35. MINITAB defines some sensitizing rules for control charts differently than the standard rules. MINITAB uses pooled standard deviation to estimate standard deviation for control chart limits and capability estimates.71 LCL R = D3 R = 0(4. 4.5 + 34.115(4.577. D3 = 0 x + x + " + xm 34.96 CL R = R = 4. 3.00 − 0.72 8 CL = 34.71) = 31. it is in control. not 8. Several exercises in this chapter differ from those in the 4th edition.0 CL = 4. 5-1 .577(4.00 R chart for Bearing ID (all samples in calculations) X-bar Chart for Bearing ID (all samples in calculations) 41.577(4. 2. New exercises are denoted with an “☺”. This can be changed in dialog boxes or under Tools>Options>Control Charts and Quality Tools>Estimating Standard Deviation.

15 excluded) R chart for Bearing ID (samples 12.65 ⎞ ⎡ ⎛ 40 − 33. LCL R = 0.06 UCL R = 9.0 12 10 UCL = 9.25. The new process parameter estimates are: x = 33.29) = 0 + 1 − 0. 15 excluded) 41. σˆ x = R / d 2 = 4.65.99950 = 0.52.0 6 33.0 4 2 27. the two samples can be excluded from the control limit calculations.CL R = 4.25 8 CL = 33.CL x = 33.93 ⎠ ⎦ = Φ (−7.65 R x-bar 35.5 / 2. 15.0 UCL =36.Chapter 5 Exercise Solutions 5-1 (a) continued The process is not in statistical control.0 CL = 4.93 ⎠ ⎣ ⎝ 1.326 = 1.0 12 39. (b) pˆ = Pr{x < LSL} + Pr{x > USL} = Pr{x < 20} + Pr{x > 40} = Pr{x < 20} + [1 − Pr{x < 40}] ⎛ 20 − 33.00050 5-2 . x is beyond the upper control limit for both Sample No.52 15 37. Assuming an assignable cause is found for these two out-of-control points.65. R = 4.07) + 1 − Φ (3. 0 LCL = 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Sample No.0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Sample No.0 LCL = 31.00 x-bar Chart for Bearing ID (samples 12.5.50 31.93 UCL x = 36.06 29. LCL x = 31. 12 and Sample No.65 ⎞ ⎤ = Φ⎜ ⎟ + ⎢1 − Φ ⎜ ⎟⎥ ⎝ 1.5.

Overall Performance PPM < LSL 0.33. x = 10.25 4 0 LC L=0 2 4 6 8 10 Sample 12 14 16 18 20 The process is in statistical control with no out-of-control signals. Actual specs are 350 ± 5 V.23 4.26 12 8 _ R=6.32500 80 3.00 -28 Exp.00 PPM > USL 0. or cycles.33 10.035) MTB > Stat > Quality Tools > Capability Analysis > Normal Process Capability Analysis of Ex5-2V LSL LSL Target USL Within Ov erall Process Data -50.23 * 42 Exp.36 4.00 5-3 .00 PPM Total 0. (a) MTB > Stat > Control Charts > Variables Charts for Subgroups > Xbar-R Xbar-R Chart of Ex5-2V Sample M ean 15. so the process is capable.88 12.12282 CPU Cpk CCpk 4.5 LC L=5.44 4.00 PPM Total 0. Within Performance PPM < LSL 0.Chapter 5 Exercise Solutions 5-2.77 5. With xi = (observed voltage on unit i – 350) × 10: USLT = +50.00000 10. runs.00 PPM Total 0.5 _ _ X=10. (b) n = 4.62 50.49 CPL 6.03545 3.0 7.035 .059 = 3.00 PPM > USL 0. R = 6.25 / 2. LSLT = –50 USL − LSL +50 − (−50) Cˆ P = = = 5.0 U C L=14.00 -14 0 14 28 5.49 Overall Capability Pp PPL PPU Ppk Cpm -42 Observed Performance PPM < LSL 0.36 5.00 PPM > USL 0.25. σˆ X = R / d 2 = 6.00000 * USL Sample Mean Sample N StDev(Within) StDev(Overall) Potential (Within) Capability Cp 5.49 . 6σˆ 6(3.0 2 4 6 8 10 Sample 12 14 16 18 20 16 Sample Range U C L=14.34 6. trends.

trends.113 80 0.704 0.1 0 5 10 Ex5-2V 15 20 A normal probability plot of the transformed output voltage shows the distribution is close to normal.73 2 4 6 8 10 Sample 12 14 16 18 20 150 Sample Range U C L=134.9 0 -20 LC L=-25. runs.064 Percent 80 70 60 50 40 30 20 10 5 1 0.5 50 LC L=0 0 2 4 6 8 10 Sample 12 14 16 18 20 The process is in statistical control with no out-of-control signals. 5-4 . (a) MTB > Stat > Control Charts > Variables Charts for Subgroups > Xbar-R Xbar-R Chart of Ex5-3Dia U C L=47.Chapter 5 Exercise Solutions 5-2 continued (c) MTB > Stat > Basic Statistics > Normality Test Probability Plot of Ex5-2V Normal 99.3 100 _ R=63.9 Mean StDev N AD P-Value 99 95 90 10. 5-3. or cycles.53 Sample M ean 40 20 _ _ X=10.33 3.

90000 100 27.00 -60 Exp.00 PPM > USL 0.5 / 2.79 PPM Total 574. Overall Performance PPM < LSL 5.3) MTB > Stat > Quality Tools > Capability Analysis > Normal Process Capability Analysis of Ex5-3Dia LSL USL Within Ov erall Process Data LSL Target USL Sample Mean Sample N StDev(Within) StDev(Overall) -100.09 -30 0 30 60 1.46 1.326 = 27.32 1.09 1.00000 * 100.67 PPM Total 219. Cˆ P = 6σˆ x 6(27.00 PPM Total 0.22 CPL CPU Cpk CCpk 25.35 1.81 PPM > USL 213.30009 Potential (Within) Capability Cp 1.09 1.29384 1.Chapter 5 Exercise Solutions 5-3 continued (b) σˆ x = R / d 2 = 63.3 (c) USL = +100.17 1. LSL = –100 USL − LSL +100 − (−100) = = 1. so the process is capable.22 .48 5-5 . Within Performance PPM < LSL 24.17 * 90 Exp.22 Overall Capability Pp PPL PPU Ppk Cpm -90 Observed Performance PPM < LSL 0.30 PPM > USL 549.00000 10.

0018 0.0625 0.0024 U C L=0.00092 0.0006 0. at points: 15 If graph is updated with new data.0620 LC L=0.00 standard deviations from center line.Chapter 5 Exercise Solutions 5-4. 2 out of 3 points more than 2 standard deviations from center line (on one side of CL).0000 LC L=0 2 4 6 8 10 12 14 Sample 16 18 20 22 24 Test Results for Xbar Chart of Ex5-4Th TEST 1.0635 _ _ X=0.0012 _ R=0.0630 0.062011 1 2 4 6 8 10 12 14 Sample 16 18 20 22 24 1 Sample Range 0.0640 U C L=0. the results above may no longer be correct.063893 0. One Test Failed * WARNING * * point more than 3. One point more than 3. Test Failed at points: 22 TEST 5. Test Failed at points: 22 Test Results for R Chart of Ex5-4Th TEST 1.00 standard deviations from center line. 5-6 .002368 0.062952 0. (a) MTB > Stat > Control Charts > Variables Charts for Subgroups > Xbar-R Xbar-R Chart of Thickness (Ex5-4Th) Sample M ean 0.

000823 /1.064408] 5-7 .000486) = [0. remove the out-of-control points (samples 15.000823 0.0630 0.0006 0. Assuming assignable causes are found.06295 ± 3(0. the new control limits are: Xbar-R Chart of Thickness (Ex5-4Th) (Samples 15.0620 1 2 4 6 8 10 12 14 Sample 16 18 20 22 24 1 0. 22.0012 _ R=0.063787 0.0000 LCL=0 2 4 6 8 10 12 14 Sample 16 18 20 22 24 (b) σˆ x = R / d 2 = 0.0625 LCL=0.0640 1 Sample Mean UCL=0.Chapter 5 Exercise Solutions 5-4 continued The process is out-of-control.062104 0.693 = 0. 0. failing tests on both the x and the R charts. With the revised limits.062945 0. Removing all three samples from calculation. 22) and recalculate control limits.0635 _ _ X=0. 14 removed from control limits calculations) 0.0018 0.0024 Sample Range UCL=0.000486 (c) Natural tolerance limits are: x ± 3σˆ x = 0.002118 0.061492. sample 14 is also out-of-control on the x chart.

70 PPM Total 2157. Overall Performance PPM < LSL 3419.0015 − (−0.33 PPM > USL 1814.06450 Sample Mean 0. 15.07 Cpk 0.55 PPM Total 5233.0628 0.0624 0.028 Cˆ P = 6σˆ x 6(0.90 * 0.000486) MTB > Stat > Quality Tools > Capability Analysis > Normal Process Capability Analysis of Thickness (Ex5-4Th_w/o) (Estimated without Samples 14.0632 0.00053 Potential (Within) Capability Cp 1.Chapter 5 Exercise Solutions 5-4 continued (d) Assuming that printed circuit board thickness is normally distributed. Within Performance PPM < LSL 1467.97 0.03 Overall Capability Pp PPL PPU Ppk Cpm 0.99 CCpk 1.61 PPM > USL 689.0644 Observed Performance PPM < LSL 15151.00 PPM Total 15151.03 CPL 0. and 22 from the process capability estimation: USL − LSL +0.88 5-8 .06295 Sample N 66 StDev(Within) 0.0616 0.52 PPM > USL 0.52 Exp.99 CPU 1.31 Exp.06150 Target * USL 0.0640 0.90 0.0620 0.94 0.00049 StDev(Overall) 0. 22) LSL USL Within Ov erall Process Data LSL 0.0636 0. 15.0015) = = 1. and excluding samples 14.

990 1 2 3 4 5 6 7 8 Sample 9 10 11 12 13 14 15 Sample Range 6.0 U C L=5.0 1 2 3 4 5 6 7 8 Sample 9 10 11 12 13 14 15 The process is in statistical control. with no out-of-control signals.302 1 2 3 4 5 6 7 8 Sample 9 10 11 12 13 14 15 The process is in statistical control. with no out-of-control signals.0 -0.5 _ S =1. trends. or cycles.0 1.5 _ R=3. trends.5 LC L=0.5 _ _ X=-0. Estimate” select Sbar as method to estimate standard deviation.0 LC L=-1.830 1.2 3. Xbar-S Chart of Fill Volume (Ex5-5Vol) U C L=1.983 0.066 1.0 0.037 Sample M ean 1.003 0. runs. (b) MTB > Stat > Control Charts > Variables Charts for Subgroups > Xbar-R (Ex5-5Vol) Xbar-R Chart of Fill Volume (Ex5-5Vol) Sample M ean 1.714 0.5 -1.0 Sample StDev U C L=1. or cycles.5 LC L=0.Chapter 5 Exercise Solutions 5-5.686 4.0 0. (a) MTB > Stat > Control Charts > Variables Charts for Subgroups > Xbar-S (Ex5-5Vol) Under “Options.043 1 2 3 4 5 6 7 8 Sample 9 10 11 12 13 14 15 2.003 0.0 LC L=-0.0 U C L=0.5 _ _ X=-0.0 -0. 5-9 . There is no difference in interpretation from the x − s chart. runs.5 -1.

Otherwise there are no runs. To construct an s2 control chart. and “By Variables” is the sample ID column (Ex5-5Sample).0027 (not tabulated in textbook). If the limits had been calculated using α = 0. or cycles.0662 = 1.066 (15 − 1) ( 31. trends.010 / 2.136 2 2 UCL = s 2 (n − 1) χα2 / 2.542 LCL = s 2 (n − 1) χ12−(α / 2). and then copying & pasting to an empty worksheet column (results in Ex5-5Variance).0 1 2 3 4 5 6 7 8 9 Sample 10 11 12 13 14 15 Sample 5 signals out of control below the lower control limit.066.0 0.330 MINITAB’s control chart options do not include an s2 or variance chart.136 1. Results are displayed in the session window.33 0. sample 5 would be within the limits.32 ) = 2. In “Statistics” select “Variance”. s = 1. To obtain sample standard deviations: Stat > Basic Statistics > Store Descriptive Statistics.n −1 = 1.07 ) = 0.Chapter 5 Exercise Solutions 5-5 continued (c) Let α = 0.542 2.010 / 2).010.0662 (15 − 1) ( χ 0. CL = s 2 = 1. Copy results from the session window by holding down the keyboard “Alt” key. and there would be no difference in interpretation from either the x − s or the x−R chart. 5-10 .0 1.15 −1 ) = 1. Graph > Time Series Plot > Simple Control limits can be added using: Time/Scale > Reference Lines > Y positions Control Chart for Ex5-5Variance UCL = 2.n −1 = 1. n = 15.5 LCL = 0. first calculate the sample standard deviations and then create a time series plot.15−1 ) = 1.0662 (15 − 1) ( χ12−(0. selecting only the variance column.0662 (15 − 1) ( 4.5 s^2 (Variance) 2. “Variables” is column with sample data (Ex5-5Vol).5 CL = 1.

or cycles.6 Sample M ean U C L=16.004 Sample Range 1. (a) MTB > Stat > Control Charts > Variables Charts for Subgroups > Xbar-R Xbar-R Chart of Net Weight (Ex5-6Wt) 16.00 0.9940 2 4 6 8 10 Sample 12 14 16 18 20 U C L=1.326 = 0. x = 16. (b) n = 5.00 LC L=0 2 4 6 8 10 Sample 12 14 16 18 20 The process is in statistical control with no out-of-control signals.50 0. runs.4 _ _ X=16. σˆ x = R / d 2 = 0.475 / 2.Chapter 5 Exercise Solutions 5-6.475.204 5-11 .25 0.268.75 _ R=0.2 16. trends.268 16.0 LC L=15. R = 0.5420 16.475 0.

1 15.6 MTB > Graph > Probability Plot > Single (Ex5-6Wt) Probability Plot of Net Weight (Ex5-6Wt) Normal .Chapter 5 Exercise Solutions 5-6 continued (c) MTB > Graph > Histogram > Single (Ex5-6Wt) Histogram of Net Weight (Ex5-6Wt) 20 Frequency 15 10 5 0 15. Points on the normal probability plot generally fall along a straight line.00 Visual examination indicates that fill weights approximate a normal distribution .50 15.4 16.00 16.9 Mean StDev N AD P-Value 99 Percent 95 90 16.0 16.8 16.the histogram has one mode.2 Ex5-6Wt 16.25 Ex5-6Wt 16. 5-12 .2014 100 1.50 16.75 17.257 <0.27 0.95% CI 99.005 80 70 60 50 40 30 20 10 5 1 0.75 16. and is approximately symmetrical with a bell shape.

00 16.96 (e) ⎛ 15.82 .268 ⎞ pˆ lower = Pr{x < LSL} = Pr{x < 15.96 5-13 .41 PPM Total 19902.5 − (−0.0027 0.20196 Potential (Within) Capability Cp 0.71 0.7 − 16.00 PPM > USL 0.204 ⎝ ⎠ The MINITAB process capability analysis also reports Exp. "Overall" Performance PPM < LSL 2458.82 CPL 0.70000 Sample Mean 16.82 Overall Capability Pp PPL PPU Ppk Cpm 15. MTB > Stat > Quality Tools > Capability Analysis > Normal Under “Estimate” select Rbar as method to estimate standard deviation.70000 Target * USL 16.23 PPM > USL 16215.6 Exp.73 PPM Total 18673.94 0.26800 Sample N 100 StDev(Within) 0.61 16.8 Observed Performance PPM < LSL 0.78) = 0. Overall Performance PPM < LSL 2458.71 * 16. Within Performance PPM < LSL 2706.20421 StDev(Overall) 0.5) = = 0.73 PPM Total 18673.0 Exp.93 CPU 0. so the process is not capable of meeting Cˆ P 6σˆ x 6(0.23 PPM > USL 16215. Process Capability Analysis of Net Weight (Ex5-6Wt) LSL USL Within Ov erall Process Data LSL 15.4 0.7} = Φ ⎜ ⎟ = Φ (−2.204) specifications.83 0.Chapter 5 Exercise Solutions 5-6 continued (d) USL − LSL +0.20 PPM > USL 17196.71 Cpk 0.71 CCpk 0.00 PPM Total 0.2 16.

0 2 4 6 8 10 Sample 12 14 16 18 20 U C L=6.5 _ _ X=10. 5-14 . trends. or cycles.0 U C L=14.11 2 4 6 8 10 Sample 12 14 16 18 20 60 Sample StDev U C L=52.5 0.5 _ S =2.23 15 LC L=0 0 2 4 6 8 10 Sample 12 14 16 18 20 The process is in statistical control with no out-of-control signals. or cycles.92 5.91 Sample M ean 40 20 _ _ X=10.703 3.125 Sample StDev 6.0 7. MTB > Stat > Control Charts > Variables Charts for Subgroups > Xbar-S (Ex5-2Vl) Xbar-S Chart of Output Voltage (Ex5-2V) Sample M ean 15.5 LC L=5.0 LC L=0 2 4 6 8 10 Sample 12 14 16 18 20 The process is in statistical control with no out-of-control signals.9 0 -20 LC L=-25.0 1.71 45 30 _ S =25. runs. MTB > Stat > Control Charts > Variables Charts for Subgroups > Xbar-S (Ex5-3Dia) Xbar-S Chart of Deviations from Nominal Diameter (Ex5-3Dia) U C L=46. 5-8.Chapter 5 Exercise Solutions 5-7. runs.73 12.0 4. trends.33 10.

(a) MTB > Stat > Control Charts > Variables Charts for Subgroups > Xbar-R (Ex5-9ID) Xbar-R Chart of Inner Diameter (Ex5-9ID) U C L=74.99 LC L=73. 5-15 . trends. (b) The control limits on the x charts in Example 5-3 were calculated using S to estimate σ.98777 2 4 6 8 10 12 14 Sample 16 18 20 22 24 U C L=0.Chapter 5 Exercise Solutions 5-9☺.036 _ R=0.048 0. or cycles.01458 Sample M ean 74.000 LC L=0 2 4 6 8 10 12 14 Sample 16 18 20 22 24 The process is in statistical control with no out-of-control signals.012 0.024 0. in this exercise R was used to estimate σ.04914 Sample Range 0. They will not always be the same.02324 0. the x control limits based on S will be slightly different than limits based on R. and in general.01 _ _ X=74.00 73. runs.00118 74.

009991 ⎝ ⎠ ⎣ ⎝ ⎠⎦ = Φ (−5.00 Exp.010 74.66 Exp.05}] ⎛ 73.63 CCpk 1.28 PPM > USL 0. so the process is not capable of meeting USL − LSL 74.05 − 73. Overall Performance PPM < LSL 0. Process Capability Analysis of Inner Diameter (Ex5-9ID) LSL USL Within Ov erall Process Data LSL 73.965 73.Chapter 5 Exercise Solutions 5-9 continued (c) σˆ x = R / d 2 = 0.89 PPM Total 1.025 74.95 = = 1. Within Performance PPM < LSL 0.67 CPL 1.980 73.59 1.995 74.668 Cˆ P = 6σˆ x 6(0.01022 Potential (Within) Capability Cp 1.15 PPM > USL 0.16 pˆ = Pr{x < LSL} + Pr{x > USL} = Pr{x < 73.950 73. MTB > Stat > Quality Tools > Capability Analysis > Normal Under “Estimate” select Rbar as method to estimate standard deviation.009991 .123) + 1 − Φ (4.51 PPM Total 0.05000 Sample Mean 74.63 1.886) = 0 +1−1 =0 5-16 .009991) specifications.95000 Target * USL 74.00118 ⎞ ⎡ ⎛ 74.63 Cpk 1.009991 0.040 Observed Performance PPM < LSL 0.05} = Pr{x < 73.00999 StDev(Overall) 0.67 Overall Capability Pp PPL PPU Ppk Cpm 1.71 CPU 1.00118 Sample N 125 StDev(Within) 0.02324 / 2.59 * 73.95 − 74.05 − 74.95} + [1 − Pr{x < 74.00118 ⎞ ⎤ = Φ⎜ ⎟ + ⎢1 − Φ ⎜ ⎟⎥ 0.67 1.00 PPM Total 0.326 = 0.00 PPM > USL 0.95} + Pr{x > 74.

37.012 0. 38.Chapter 5 Exercise Solutions 5-10☺.01 0.99 LC L=73.02 1 5 5 _ _ X=74. 38.04914 0.00 73.98777 4 8 12 16 20 Sample 24 28 32 36 40 U C L=0. 39.01458 74. an assignable cause has likely occurred. Since this point and the three subsequent points plot above the upper control limit. 39. 2 out of 3 points more than 2 standard deviations from center line (on one side of CL).00 standard deviations from center line. until the x -value from the 37th sample is plotted.000 LC L=0 4 8 12 16 20 Sample 24 28 32 36 40 Test Results for Xbar Chart of Ex5-10ID TEST 1. 4 out of 5 points more than 1 standard deviation from center line (on one side of CL).00118 74.048 Sample Range U C L=74. Test Failed at points: 37. One point more than 3. 39 TEST 5. 40 TEST 6.036 _ R=0. MTB > Stat > Control Charts > Variables Charts for Subgroups > Xbar-R (Ex5-10ID) Xbar-R Chart of Inner Diameter (Ex5-10ID) 1 1 Sample M ean 74.02324 0. increasing the process mean. 40 The control charts indicate that the process is in control. 5-17 . Test Failed at points: 35. Test Failed at points: 38.024 0.

so the process is not capable. µ = 80 in-lb.949.76 5-12* (5-10).69 LCL S = B5σ x = 0. σˆ x ≈ 1.433) = 1 − 0.534) = [35.276 centerline x = µ = 80 UCL x = µ + Aσ x = 80 + 0.483(4) = 38.9727(10) = 9.932 x= i =1 = LCL x = x − A2 R = 40 − 0. ∑ xi = 2000.276(10) = 2. R = i =1 = =4 m m 50 50 UCL x = x + A2 R = 40 + 0.669(10) = 16.669.49 LCL x = µ − Aσ x = 80 − 0.LSL +5.264.579) (d) ⎛ 36 − 40 ⎞ pˆ scrap = Pr{x < LSL} = Pr{x < 36} = Φ ⎜ ⎟ = Φ (−2. ⎝ 1. 5-18 . reduce variability such that the natural process tolerance limits are closer to.949(10) = 89.579 ⎠ ⎛ 47 − 40 ⎞ pˆ rework = Pr{x > USL} = 1 − Pr{x < USL} = 1 − Φ ⎜ ⎟ = 1 − Φ (4.0) = = 1.999995 = 0.Chapter 5 Exercise Solutions 5-11 (5-9). B6 = 1.0005%. Cˆ P = 6σˆ x 6(1.004(4) = 8. Second. ∑ Ri = 200.57%.736] (c) USL .253 . 50 50 i =1 i =1 n = 6 items/sample. not 40. 44. B5 = 0.000005 ⎝ 1.51 centerline = c4σ x = 0.533) = 0.056 .483(4) = 41. (e) First. say. to reduce scrap and rework costs. and A = 0. n = 10. or 0.068 UCL R = D4 R = 2.0 − (−5. m = 50 samples (a) 50 ∑ xi 50 ∑ Ri 2000 200 = 40.0057 .727 S UCL S = B6σ x = 1. σ x = 10 in-lb.579 ⎠ or 0. center the process at 41.016 LCL R = D3 R = 0(4) = 0 (b) natural tolerance limits: x ± 3σˆ x = x ± 3 ( R / d 2 ) = 40 ± 3(4 / 2.949(10) = 70.

Chapter 5 Exercise Solutions

5-13* (5-11).
50

50

i =1

i =1

n = 4 items/subgroup; ∑ xi = 1000; ∑ Si = 72; m = 50 subgroups
(a)
50

x=

∑ xi

i =1

m

=

1000
= 20
50

50

∑ Si

72
= 1.44
50
m
UCL x = x + A3 S = 20 + 1.628(1.44) = 22.34
S=

i =1

=

LCL x = x − A3 S = 20 − 1.628(1.44) = 17.66
UCL S = B4 S = 2.266(1.44) = 3.26
LCL S = B3 S = 0(1.44) = 0
(b)
⎛S ⎞
⎛ 1.44 ⎞
natural process tolerance limits: x ± 3σˆ x = x ± 3 ⎜ ⎟ = 20 ± 3 ⎜
⎟ = [15.3, 24.7]
⎝ 0.9213 ⎠
⎝ c4 ⎠

(c)

USL - LSL
+4.0 − (−4.0)
=
= 0.85 , so the process is not capable.
Cˆ P =
6σˆ x
6(1.44 / 0.9213)

(d)
⎛ 23 − 20 ⎞
pˆ rework = Pr{x > USL} = 1 − Pr{x ≤ USL} = 1 − Φ ⎜
⎟ = 1 − Φ (1.919) = 1 − 0.9725 = 0.0275
⎝ 1.44 / 0.9213 ⎠
or 2.75%.
⎛ 15 − 20 ⎞
pˆ scrap = Pr{x < LSL} = Φ ⎜
⎟ = Φ (−3.199) = 0.00069 , or 0.069%
⎝ 1.44 / 0.9213 ⎠
Total = 2.88% + 0.069% = 2.949%
(e)
⎛ 23 − 19 ⎞
pˆ rework = 1 − Φ ⎜
⎟ = 1 − Φ (2.56) = 1 − 0.99477 = 0.00523 , or 0.523%
⎝ 1.44 / 0.9213 ⎠
⎛ 15 − 19 ⎞
pˆ scrap = Φ ⎜
⎟ = Φ (−2.56) = 0.00523 , or 0.523%
⎝ 1.44 / 0.9213 ⎠
Total = 0.523% + 0.523% = 1.046%
Centering the process would reduce rework, but increase scrap. A cost analysis is needed
to make the final decision. An alternative would be to work to improve the process by
reducing variability.

5-19

Chapter 5 Exercise Solutions

5-14 (5-12).
(a)
MTB > Stat > Control Charts > Variables Charts for Subgroups > Xbar-R
Xbar-R Chart of Critical Dimension (Ex5-14ax1, ..., Ex5-14ax5)
U C L=154.45
Sample M ean

150
140
_
_
X=130.88

130
120
110

LC L=107.31
2

4

6

8

10
Sample

12

14

16

18

20

U C L=86.40

Sample Range

80
60

_
R=40.86

40
20
0

LC L=0
2

4

6

8

10
Sample

12

14

16

18

20

The process is in statistical control with no out-of-control signals, runs, trends, or cycles.
(b)
MTB > Stat > Control Charts > Variables Charts for Subgroups > Xbar-R

Under “Options, Estimate” use subgroups 1:20 to calculate control limits.
Xbar-R Chart of Critical Dimension (Ex5-14bx1, ..., Ex5-14bx5)
1

Sample M ean

180

1

1

1

1
1
1

1
1

1

160

U C L=154.45

140

_
_
X=130.88

120
LC L=107.31

100
3

6

9

12

15
Sample

18

21

24

27

30

U C L=86.40

Sample Range

80
60

_
R=40.86

40
20
0

LC L=0
3

6

9

12

15
Sample

18

21

24

27

30

Starting at Sample #21, the process average has shifted to above the UCL = 154.45.

5-20

Chapter 5 Exercise Solutions

5-14 continued
(c)
MTB > Stat > Control Charts > Variables Charts for Subgroups > Xbar-R

Under “Options, Estimate” use subgroups 1:20 to calculate control limits.
Xbar-R Chart of Critical Dimension (Ex5-14cx1, ..., Ex5-14cx5)
1

Sample M ean

180

1

1

1

1
1
1

1

1

160

1

U C L=154.45

140

_
_
X=130.88

120

2
6

2

5

LC L=107.31

1

100

4

8

12

16

20
Sample

24

28

32

36

40

U C L=86.40

Sample Range

80
60

_
R=40.86

40
20
0

LC L=0

4

8

12

16

20
Sample

24

28

32

36

40

The adjustment overcompensated for the upward shift. The process average is now
between x and the LCL, with a run of ten points below the centerline, and one sample
(#36) below the LCL.

5-21

Chapter 5 Exercise Solutions

5-15* (5-13).
(a)
MTB > Stat > Control Charts > Variables Charts for Subgroups > Xbar-R

Xbar-R Chart of Strength Test (Ex5-15aSt)

Sample M ean

85.0

U C L=84.58

82.5
_
_
X=79.53

80.0
77.5
75.0

LC L=74.49

2

4

6

8

10
Sample

12

14

16

18

20

Sample Range

20

U C L=18.49

15
_
R=8.75

10
5
0

LC L=0

2

4

6

8

10
Sample

12

14

16

18

20

Yes, the process is in control—though we should watch for a possible cyclic pattern in
the averages.

5-22

Chapter 5 Exercise Solutions

5-15 continued
(b)
MTB > Stat > Control Charts > Variables Charts for Subgroups > Xbar-R

Under “Options, Estimate” use subgroups 1:20 to calculate control limits.
Xbar-R Chart of Strength Test (Ex5-15bSt)
1

Sample M ean

85.0

8

82.5

8

U C L=84.58

8

_
_
X=79.53

80.0
77.5
75.0

LC L=74.49
1

3

6

9

12

15

18
Sample

21

24

27

30

30

33

1
1

Sample Range

1

1
1
1

1

1

20

U C L=18.49
2

_
R=8.75

10

0

LC L=0
3

6

9

12

15

18
Sample

21

24

27

30

33

Test Results for R Chart of Ex5-15bSt
TEST
Test
TEST
Test

1. One point more than 3.00 standard deviations from center line.
Failed at points: 25, 26, 27, 31, 33, 34, 35
2. 9 points in a row on same side of center line.
Failed at points: 32, 33, 34, 35

A strongly cyclic pattern in the averages is now evident, but more importantly, there are
several out-of-control points on the range chart.

5-23

Chapter 5 Exercise Solutions

5-16 (5-14).
(a)
MTB > Stat > Control Charts > Variables Charts for Subgroups > Xbar-S
Xbar-S Chart of Strength Test (Ex5-15aSt)
Original Data

Sample Mean

85.0

UCL=84.64

82.5
_
_
X=79.53

80.0
77.5
75.0

LCL=74.43
2

4

6

8

10
Sample

12

14

16

18

20

Sample StDev

8

UCL=7.468

6
_
S=3.575

4
2
0

LCL=0
2

4

6

8

10
Sample

12

14

16

18

20

Under “Options, Estimate” use subgroups 1:20 to calculate control limits.
Xbar-S Chart of Strength Test (Ex5-15bSt)
Original plus New Data
1

Sample Mean

85.0

UCL=84.64

82.5
_
_
X=79.53

80.0
77.5
75.0

LCL=74.43
1

3

6

9

12

15

18
Sample

21

24

27

30

1

33

1
1

Sample StDev

10.0

1

1

7.5

1

1

1

1

UCL=7.47

5.0

_
S=3.57

2.5
LCL=0

0.0
3

6

9

12

15

18
Sample

21

24

27

30

33

Test Results for Xbar Chart of Ex5-15bSt
TEST 1. One point more than 3.00 standard deviations from center line.
Test Failed at points: 24, 31, 34

Test Results for S Chart of Ex5-15bSt
TEST 1. One point more than 3.00 standard deviations from center line.
Test Failed at points: 22, 25, 26, 27, 31, 33, 34, 35

5-24

Chapter 5 Exercise Solutions

5-16 continued
(b)
Yes, the s chart detects the change in process variability more quickly than the R chart
did, at sample #22 versus sample #24.

5-17 (5-15).
nold = 5; xold = 34.00; Rold = 4.7
(a)
for nnew = 3
⎡ d 2(new ) ⎤
⎡ 1.693 ⎤
UCL x = xold + A2(new ) ⎢
(4.7) = 37.50
⎥ Rold = 34 + 1.023 ⎢
⎣ 2.326 ⎥⎦
⎣⎢ d 2(old) ⎦⎥
⎡ d 2(new ) ⎤
⎡ 1.693 ⎤
LCL x = xold − A2(new ) ⎢
⎥ Rold = 34 − 1.023 ⎢
⎥ (4.7) = 30.50
⎣ 2.326 ⎦
⎢⎣ d 2(old) ⎥⎦
⎡ d 2(new ) ⎤
⎡ 1.693 ⎤
UCL R = D4(new) ⎢
⎥ Rold = 2.574 ⎢
⎥ (4.7) = 8.81
⎣ 2.326 ⎦
⎢⎣ d 2(old) ⎥⎦
⎡d

⎡ 1.693 ⎤
CL R = Rnew = ⎢ 2(new ) ⎥ Rold = ⎢
(4.7) = 3.42
⎣ 2.326 ⎥⎦
⎣⎢ d 2(old) ⎦⎥
⎡d

⎡ 1.693 ⎤
LCL R = D3(new) ⎢ 2(new ) ⎥ Rold = 0 ⎢
(4.7) = 0
⎣ 2.326 ⎥⎦
⎢⎣ d 2(old) ⎥⎦
(b)
The x control limits for n = 5 are “tighter” (31.29, 36.72) than those for n = 3 (30.50,
37.50). This means a 2σ shift in the mean would be detected more quickly with a sample
size of n = 5.

5-25

Chapter 5 Exercise Solutions

5-17 continued
(c)
for n = 8
⎡ d 2(new ) ⎤
⎡ 2.847 ⎤
UCL x = xold + A2(new ) ⎢
⎥ Rold = 34 + 0.373 ⎢
⎥ (4.7) = 36.15
⎣ 2.326 ⎦
⎣⎢ d 2(old) ⎦⎥
⎡ d 2(new ) ⎤
⎡ 2.847 ⎤
LCL x = xold − A2(new ) ⎢
(4.7) = 31.85
⎥ Rold = 34 − 0.373 ⎢
⎣ 2.326 ⎥⎦
⎢⎣ d 2(old) ⎥⎦
⎡ d 2(new ) ⎤
⎡ 2.847 ⎤
UCL R = D4(new) ⎢
(4.7) = 10.72
⎥ Rold = 1.864 ⎢
⎣ 2.326 ⎥⎦
⎣⎢ d 2(old) ⎦⎥
⎡d

⎡ 2.847 ⎤
CL R = Rnew = ⎢ 2(new ) ⎥ Rold = ⎢
(4.7) = 5.75
⎣ 2.326 ⎥⎦
⎢⎣ d 2(old) ⎥⎦
⎡d

⎡ 2.847 ⎤
LCL R = D3(new) ⎢ 2(new ) ⎥ Rold = 0.136 ⎢
⎥ (4.7) = 0.78
⎣ 2.326 ⎦
⎢⎣ d 2(old) ⎥⎦
(d)
The x control limits for n = 8 are even "tighter" (31.85, 36.15), increasing the ability of
the chart to quickly detect the 2σ shift in process mean.
5-18☺.
nold = 5, xold = 74.001, Rold = 0.023, nnew = 3
⎡ d 2(new ) ⎤
⎡ 1.693 ⎤
UCL x = xold + A2(new ) ⎢
⎥ Rold = 74.001 + 1.023 ⎢
⎥ (0.023) = 74.018
⎣ 2.326 ⎦
⎣⎢ d 2(old) ⎦⎥
⎡ d 2(new ) ⎤
⎡ 1.693 ⎤
LCL x = xold − A2(new ) ⎢
(0.023) = 73.984
⎥ Rold = 74.001 − 1.023 ⎢
⎣ 2.326 ⎥⎦
⎢⎣ d 2(old) ⎥⎦
⎡ d 2(new ) ⎤
⎡ 1.693 ⎤
UCL R = D4(new) ⎢
(0.023) = 0.043
⎥ Rold = 2.574 ⎢
⎣ 2.326 ⎥⎦
⎣⎢ d 2(old) ⎥⎦
⎡d

⎡ 1.693 ⎤
CL R = Rnew = ⎢ 2(new ) ⎥ Rold = ⎢
(0.023) = 0.017
⎣ 2.326 ⎥⎦
⎢⎣ d 2(old) ⎥⎦
⎡d

⎡ 1.693 ⎤
LCL R = D3(new) ⎢ 2(new ) ⎥ Rold = 0 ⎢
(0.023) = 0
⎣ 2.326 ⎥⎦
⎣⎢ d 2(old) ⎦⎥

5-26

Chapter 5 Exercise Solutions

5-19 (5-16).
35

35

i =1

i =1

n = 7; ∑ xi = 7805; ∑ Ri = 1200; m = 35 samples
(a)
35

x=

∑ xi

i =1

m

=

7805
= 223
35

35

∑ Ri

1200
= 34.29
m
35
UCL x = x + A2 R = 223 + 0.419(34.29) = 237.37
R=

i =1

=

LCL x = x − A2 R = 223 − 0.419(34.29) = 208.63
UCL R = D4 R = 1.924(34.29) = 65.97
LCL R = D3 R = 0.076(34.29) = 2.61

(b)
µˆ = x = 223; σˆ x = R / d 2 = 34.29 / 2.704 = 12.68
(c)

USL − LSL +35 − (−35)
=
= 0.92 , the process is not capable of meeting
Cˆ P =
6σˆ x
6(12.68)
specifications.

pˆ = Pr{x > USL} + Pr{x < LSL} = 1 − Pr{x < USL} + Pr{x < LSL} = 1 − Pr{x ≤ 255} + Pr{x ≤ 185}
⎛ 255 − 223 ⎞
⎛ 185 − 223 ⎞
= 1− Φ ⎜
⎟+ Φ⎜
⎟ = 1 − Φ (2.52) + Φ (−3.00) = 1 − 0.99413 + 0.00135 = 0.0072
⎝ 12.68 ⎠
⎝ 12.68 ⎠
(d)
The process mean should be located at the nominal dimension, 220, to minimize nonconforming units.
⎛ 255 − 220 ⎞
⎛ 185 − 220 ⎞
pˆ = 1 − Φ ⎜
⎟+Φ⎜
⎟ = 1 − Φ (2.76) + Φ (−2.76) = 1 − 0.99711 + 0.00289 = 0.00578
⎝ 12.68 ⎠
⎝ 12.68 ⎠

5-27

36) = 0 (b) σˆ x = R / d 2 = 0.87) = 1 − 0.36) = 26.50 ⎞ = 1− Φ ⎜ ⎟+Φ⎜ ⎟ = 1 − Φ (2.99506 + 0.36 / 2.36) = 0. 25 25 i =1 i =1 n = 5.90 − 26.58) + Φ (−3.326 = 0.23) 0.90 − 26.99938 + 0.155 0.50 + 0.114(0.36) = 26.50 ⎞ ⎛ 25.155 ⎝ ⎠ ⎝ ⎠ = 0.00005 0.36 m 25 UCL x = x + A2 R = 26.577(0.00499 (c) ⎛ 26.90 − 26.40 ⎞ ⎛ 25.50. ∑ Ri = 9.29 UCL R = D4 R = 2.00 = 0.76 LCL R = D3 R = 0(0.00124 5-28 .00062 = 0.23) + Φ (−3.155 0.155 pˆ = Pr{x > USL} + Pr{x < LSL} = 1 − Pr{x ≤ USL} + Pr{x < LSL} ⎛ 26. m = 25 samples (a) 25 x= ∑ xi i =1 m = 662.40 ⎞ pˆ = 1 − Φ ⎜ ⎟+Φ⎜ ⎟ = 1 − Φ (3.00.50 25 25 ∑ Ri 9.577(0.155 ⎝ ⎠ ⎝ ⎠ = 1 − 0. ∑ xi = 662.50 − 0.71 R= i =1 = LCL x = x − A2 R = 26.50 = 26.90 − 26.Chapter 5 Exercise Solutions 5-20 (5-17).

0.998) 5-29 .0 + 1.998 and 6σˆ x = 6(3.5) = 17. Pr{detect} = 1 − Pr{not detect} = 1 − [Pr{LCL ≤ x ≤ UCL}] = 1 − [Pr{x ≤ UCL} − Pr{x ≤ LCL}] ⎡ ⎛ UCL − µ ⎞ ⎛ LCL − µ x x new ⎟ − Φ ⎜ new = 1 − ⎢Φ ⎜ ⎢ ⎜ ⎟ ⎜ n n σ σ x x ⎠ ⎝ ⎣ ⎝ = 1 − Φ (7) + Φ (1) = 1 − 1 + 0.427(1. m = 50 samples (a) σˆ x = S / c4 = 1.0 − 1.9400 = 1. 2(10) = 20. So.6 5 1.5) = 22. x = 20.326 = 3.089(1.99 is larger than the width of the tolerance band.84134 = 0. n = 5. not all of the output will meet specification.8338 6σˆ x 6(3.20) − Φ (−5. LSL=90 σˆ x = R / d 2 = 9.84134 ⎞⎤ ⎡ ⎛ ⎞ ⎛ ⎞⎤ ⎟ ⎥ = 1 − ⎢Φ ⎜ 209 − 188 ⎟ − Φ ⎜ 191 − 188 ⎟ ⎥ ⎟⎥ ⎢⎣ ⎝ 6 4 ⎠ ⎝ 6 4 ⎠ ⎥⎦ ⎠⎦ 5-23 (5-20).427(1.5) = 3.79) 1. x = 104. even if the mean is located at the nominal dimension.86 UCL S = B4 S = 2.13 LCL S = B3 S = 0(1. n = 5. R = 9.6 5 ⎝ ⎠ ⎝ ⎠ = 0. X ~ N .14 − 22 ⎞ ⎛ 17.30 / 2. S = 1. 100.14 LCL x = x − A3 S = 20.5) = 0 (c) Pr{in control} = Pr{LCL ≤ x ≤ UCL} = Pr{x ≤ UCL} − Pr{x ≤ LCL} ⎛ 22.998) = 23. USL − LSL +10 − (−10) Cˆ P = = = 0.86 − 22 ⎞ = Φ⎜ ⎟ −Φ⎜ ⎟ = Φ (0.57926 5-22 (5-19).Chapter 5 Exercise Solutions 5-21 (5-18).60 (b) UCL x = x + A3 S = 20.57926 − 0 = 0.5 / 0.5.30. USL=110.

606(2.128(2.30 LCL x = µ − Aσ x = 10 − 2.7979(2.686(2.Chapter 5 Exercise Solutions 5-24* (5-21).121(2.5) = 0 5-30 .52 LCL S = B5σ = 0(2.5) = 6.70 (b) centerline R = d 2σ x = 1.22 LCL R = D1σ = 0(2. n = 2.99 UCL S = B6σ = 2.5) = 15.5) = 2. σ = 2.82 UCL R = D2σ = 3. µ = 10.121(2. These are standard values.5) = 9. x (a) centerline x = µ = 10 UCL x = µ + Aσ x = 10 + 2.5 .5) = 0 (c) centerline S = c4σ x = 0.5) = 1.5) = 4.

56) = 9.63 LCL x = x − A2 R = 20 − 0.96 (c) USL − LSL +5 − (−5) = = 0. so the process is not capable of meeting Cˆ P = 6σˆ x 6(1.326 = 1.114(4.05938 − 0 = 0.85 .577(4.96) specifications.Chapter 5 Exercise Solutions 5-25 (5-22).37 − 24 ⎞ = Φ ⎜⎜ ⎟⎟ − Φ ⎜⎜ ⎟⎟ = Φ ⎜ ⎟−Φ⎜ ⎟ ⎝ 1. R = 4.63 − 24 ⎞ ⎛ 17.56) = 0 (b) σˆ x = R / d 2 = 4.96 5 ⎠ ⎝ σˆ x n ⎠ ⎝ σˆ x n ⎠ = Φ (−1.56) = 22. n = 5.05938 5-31 .577(4.56) = 17.56) = 0.56.56) + Φ (−7. x = 20.64 LCL R = D3 R = 0(4.37 UCL R = D4 R = 2.96 5 ⎠ ⎝ 1. m = 25 samples (a) UCL x = x + A2 R = 20 + 0.56 / 2. (d) Pr{not detect} = Pr{LCL ≤ x ≤ UCL} = Pr{x ≤ UCL} − Pr{x ≤ LCL} ⎛ UCL x − µnew ⎞ ⎛ LCL x − µnew ⎞ ⎛ 22.

so these limits can be used for future production. 5-32 .Chapter 5 Exercise Solutions 5-26☺.82 450 _ _ X=448. Excluding subgroup 18 from control limits calculations: Xbar-R Chart of TiW Thickness (Ex5-26Th) Excluding subgroup 18 from calculations UCL=461. Test Failed at points: 18 The process is out of control on the x chart at subgroup 18.69 440 LC L=436.56 430 1 2 4 6 8 10 Sample 12 14 16 18 20 Sample Range 40 U C L=37.49 430 1 2 4 6 8 10 Sample 12 14 16 18 20 Sample Range 40 UCL=38. One point more than 3.98 30 20 _ R=16.00 standard deviations from center line. Test Failed at points: 18 No additional subgroups are beyond the control limits.00 standard deviations from center line.74 10 0 LCL=0 2 4 6 8 10 Sample 12 14 16 18 20 Test Results for Xbar Chart of Ex5-26Th TEST 1.88 Sample Mean 460 _ _ X=449. One point more than 3.65 10 0 LC L=0 2 4 6 8 10 Sample 12 14 16 18 20 Test Results for Xbar Chart of Ex5-26Th TEST 1.18 30 20 _ R=16. MTB > Stat > Control Charts > Variables Charts for Subgroups > Xbar-R Sample M ean Xbar-R Chart of TiW Thickness (Ex5-26Th) 460 U C L=460.68 450 440 LCL=437.

5-33 .7 9.269 0.74 / 2.111 80 0.059 = 8.Chapter 5 Exercise Solutions 5-26 continued (b) Excluding subgroup 18: x = 449.1 420 430 440 450 Ex5-26Th 460 470 480 A normal probability plot of the TiW thickness measurements shows the distribution is close to normal.68 σˆ x = R / d 2 = 16.9 Mean StDev N AD P-Value 99 Percent 95 90 448.13 (c) MTB > Stat > Basic Statistics > Normality Test Probability Plot of TiW Thickness (Ex5-26Th) Normal 99.672 80 70 60 50 40 30 20 10 5 1 0.

14 1. Overall Performance PPM < LSL 848.00 PPM > USL 0.18 CPU 1. or in other words.13944 Overall Capability Pp PPL PPU Ppk Cpm 420 Observed Performance PPM < LSL 0.24.23 . σ x is estimated using R .08645 9.30 440 450 460 470 1.24 CPL 1.05 * 480 Exp.92 248.09 1.68750 Sample N 80 StDev(Within) StDev(Overall) Potential (Within) Capability Cp 1. LSL = –30 USL − LSL +30 − (−30) = = 1.00 PPM Total 0. Within Performance PPM < LSL 194.17 PPM Total 1154.13) MTB > Stat > Quality Tools > Capability Analysis > Normal Process Capability Analysis of TiW Thickness (Ex5-26Th) LSL USL W ithin O v erall Process Data LSL 420.24 8.01 PPM > USL 306. Cˆ P = 6σˆ x 6(8. 5-34 . Cp = 1. This is the same result as the manual calculation. is estimated from the within-subgroup variation.18 The Potential (Within) Capability.38 PPM > USL PPM Total 53.29 Cpk 1.00000 Sample Mean 448. so the process is capable.Chapter 5 Exercise Solutions 5-26 continued (d) USL = +30.00 430 Exp.00000 Target * USL 480.18 CCpk 1.05 1.

MTB > Stat > Control Charts > Variables Charts for Subgroups > Xbar-R Xbar-R Chart of TiW Thickness (Ex5-27Th) Using previous limits with 10 new subgroups UCL=461. Test Failed at points: 18 The process continues to be in a state of statistical control.00 standard deviations from center line.74 10 0 LCL=0 3 6 9 12 15 Sample 18 21 24 27 30 Test Results for Xbar Chart of Ex5-27Th TEST 1.88 Sample Mean 460 _ _ X=449. 5-35 .68 450 440 LCL=437.Chapter 5 Exercise Solutions 5-27☺. One point more than 3.18 30 20 _ R=16.49 430 1 3 6 9 12 15 Sample 18 21 24 27 30 Sample Range 40 UCL=38.

43 430 1 2 3 4 5 6 7 8 9 10 Sample UCL=29.128 ⎤ LCL R = D3(new) ⎢ ⎥ Rold = 0 ⎢ ⎥ (16.96 Sample Range 30 20 10 _ R=9.92 ⎥ Rold = 449. Parameters. xold 449.128 ⎤ UCL x = xold + A2(new ) ⎢ (16.74) = 9.059 ⎥⎦ ⎣⎢ d 2(old) ⎦⎥ ⎡ d 2(new ) ⎤ ⎡ 1.68 − 1.13 MTB > Stat > Control Charts > Variables Charts for Subgroups > Xbar-R Select Xbar-R options. Limits derived from N=4 subgroups Sample Mean 470 UCL=466.267 ⎢ ⎣ 2. nnew = 2 ⎡ d 2(new ) ⎤ ⎡ 1.880 ⎢ ⎣ 2.Chapter 5 Exercise Solutions 5-28☺.128 ⎤ LCL x = xold − A2(new ) ⎢ (16.96 ⎥ Rold = 3.74) = 432.74) = 29. 5-36 .880 ⎢ ⎣ 2. Xbar-R Chart of TiW Thickness (Ex5-28Th) New subgroups with N=2.128 ⎤ UCL R = D4(new) ⎢ (16.44 ⎥ Rold = 449.74) = 466.17 0 LCL=0 1 2 3 4 5 6 7 8 9 10 Sample The process remains in statistical control.74) = 0 ⎣ 2. nold = 4.128 = 8. R old = 16.059 ⎦ ⎢⎣ d 2(old) ⎥⎦ σˆ new = Rnew d 2(new ) = 9.17 ⎣ 2.68 450 440 LCL=432. and enter new parameter values.059 ⎥⎦ ⎢⎣ d 2(old) ⎥⎦ ⎡ d 2(new ) ⎤ ⎡ 1.68.059 ⎦ ⎣⎢ d 2(old) ⎦⎥ ⎡ d 2(new ) ⎤ ⎡ 1.68 + 1.128 ⎤ CL R = Rnew = ⎢ ⎥ Rold = ⎢ ⎥ (16.74.17 1.93 460 _ _ X=449.059 ⎥⎦ ⎣⎢ d 2(old) ⎦⎥ ⎡ d 2(new ) ⎤ ⎡ 1.

68 450 440 LCL=437.00 standard deviations from center line.44 15 10 _ S=7. The process is out of control on the x chart at subgroup 18.15 430 1 3 6 9 12 15 Sample 18 21 24 27 30 20 Sample StDev UCL=17. One point more than 3. After finding assignable cause. 5-37 . Test Failed at points: 18 No additional subgroups are beyond the control limits.44 15 10 _ S=7.Chapter 5 Exercise Solutions 5-29☺. with prior limits UCL=462. Xbar-S Chart of Thickness (Ex5-27Th) 10 subgroups of new data.15 430 1 2 4 6 8 10 Sample 12 14 16 18 20 20 Sample StDev UCL=17.22 Sample Mean 460 _ _ X=449.70 5 0 LCL=0 3 6 9 12 15 Sample 18 21 24 27 30 The process remains in statistical control.22 Sample Mean 460 _ _ X=449.70 5 0 LCL=0 2 4 6 8 10 Sample 12 14 16 18 20 Xbar-S Chart of Ex5-26Th Test Results for Xbar Chart of Ex5-26Th TEST 1. so these limits can be used for future production.68 450 440 LCL=437. exclude subgroup 18 from control limits calculations: MTB > Stat > Control Charts > Variables Charts for Subgroups > Xbar-S Xbar-S Chart of Thickness (Ex5-26Th) Excluding subgroup 18 from calculations UCL=462.

25) − Φ (−1.004(5) = 10. (c) ⎛ 202.04006 = 0.534 = 1.95994 β − risk = Pr{not detect} = Φ ⎜ 5-38 . 30 30 i =1 i =1 n = 6.75) = 1 − 0. ∑ Ri = 150.42 − 199 ⎞ ⎛ 197.Chapter 5 Exercise Solutions 5-30 (5-23).85 Cˆ p = 6σˆ x 6(1.97 USL − LSL +5 − (−5) = = 0.483(5) = 202. Even though the process is centered at nominal.59 UCL R = D4 R = 2. the variation is large relative to the tolerance. m = 30 samples (a) 30 x= ∑ xi = i =1 m 6000 = 200 30 30 ∑ Ri 150 =5 m 30 UCL x = x + A2 R = 200 + 0.02 LCL R = D3 R = 0(5) = 0 (b) σˆ x = R / d 2 = 5 / 2. ∑ xi = 6000.97 6 ⎠ ⎝ 1.97) The process is not capable of meeting specification.483(5) = 197.97 6 ⎠ = Φ (4.42 R= i =1 = LCL x = x − A2 R = 200 − 0.59 − 199 ⎞ ⎟−Φ⎜ ⎟ ⎝ 1.

935 UCL R = D4 R = 2.66) − Φ (−0.771 UCL R = D4 R = 2. R = 3.33 Pr{detecting shift on 1st sample} = 1 − Pr{not detecting shift on 1st sample} = 1− β ( ) ( ) = 1 − ⎡Φ ( 3 − (−1.37 ] = 0.95 UCL x = x + A2 R = 104.577(3.577(3.95) = 8.065 LCL x = x − A2 R = 104 − 0. R = 3.326 = 1.539 (c) UNTL = x + 3σˆ x = 104 + 3(1. n = 4.579) = 0 (b) Without sample #4.34) ] = 1 − [1 − 0.62 LNTL = x − 3σˆ x = 104 − 3(1.05.579) = 7.579) = 101.566 LCL R = D3 R = 0(3. L = 3.38 5-39 .577(3.577(3.579 UCL x = x + A2 R = 104 + 0.579) = 106.Chapter 5 Exercise Solutions 5-31 (5-24).350 LCL R = D3 R = 0(3.33) 4 ) ⎤ ⎣ ⎦ = 1 − ⎡Φ L − k n − Φ − L − k n ⎤ ⎣ ⎦ = 1 − [ Φ (5.579 / 2.33) 4 ) − Φ ( −3 − (−1. µ0 = 100.05 − 0.329 LCL x = x − A2 R = 104.539) = 108. excluding #4 and recalculating: x = 104.114(3. µ1 = 92 k = ( µ1 − µ0 ) σ = ( 92 − 100 ) 6 = −1. σ = 6.95) = 101. (a) x = 104.539) = 99.05 + 0.95) = 106.37 5-32 (5-25). So.95) = 0 Sample #4 is out of control on the Range chart. σˆ x = R / d 2 = 3.114(3.

26 30 m ∑ Ri 144 = 4.0000 = 0. ⎛ 107 − 103 ⎞ ⎛ 99 − 103 ⎞ pˆ = 1 − Φ ⎜ ⎟+ Φ⎜ ⎟ = 1 − Φ (5. if σˆ x = 0.8) = 0 (b) σˆ x = R / d 2 = 4.Chapter 5 Exercise Solutions 5-32 continued (d) ⎛ 107 − 104 ⎞ ⎛ 99 − 104 ⎞ pˆ = 1 − Φ ⎜ ⎟+ Φ⎜ ⎟ = 1 − Φ (1.0000 ⎝ 0.9744 + 0.064 ⎠ 5-40 . then almost 100% of parts will be within specification.9953 + 0.03 R= i =1 = LCL x = x − A2 R = 20.997) = 1 − 1.667 .95) + Φ (−3.60) = 1 − 0.147 LCL R = D 3 R = 0(4.667 ⎠ 5-33 (5-26).8) = 23.26 − 0.326 = 2.8 / 2.997) + Φ (−5.064) = 0.26 + 0. ∑ xi = 607.0047 = 0.539 ⎠ ⎝ 1.539 ⎠ (e) To reduce the fraction nonconforming.8) = 17. m = 30 i =1 (a) m x= ∑ xi i =1 m = 607.26 ⎞ pˆ = Pr{x < LSL} = Φ ⎜ ⎟ = Φ (−2. i =1 30 ∑ Ri = 144.25) = 1 − 0.539 ⎠ ⎝ 1. ⎛ 107 − 103 ⎞ ⎛ 99 − 103 ⎞ pˆ = 1 − Φ ⎜ ⎟+ Φ⎜ ⎟ = 1 − Φ (2.577(4.539 ⎠ Next work on reducing the variability.114(4.8.8) = 10.0000 + 0.60) + Φ (−2.8 = 20.064 ⎛ 16 − 20. 30 n = 5.577(4.0006 = 0.8 30 m UCL x = x + A2 R = 20.0195 ⎝ 2.0094 ⎝ 1.49 UCL R = D 4 R = 2. first center the process at nominal.0262 ⎝ 1.667 ⎠ ⎝ 0.

One point more than 3.sample #12 exceeds the upper control limit on the Range chart. (a) MTB > Stat > Control Charts > Variables Charts for Subgroups > R Under “Options. Test Failed at points: 12 Process is not in statistical control -. 5-41 .Chapter 5 Exercise Solutions 5-34 (5-27).67 10 _ R=6. R Chart of Detent (Ex5-34Det) 20 1 15 Sample Range UCL=13. Estimate” select Rbar as method to estimate standard deviation.47 5 0 LCL=0 1 2 3 4 5 6 7 8 9 Sample 10 11 12 13 14 15 Test Results for R Chart of Ex5-34Det TEST 1.00 standard deviations from center line.

64 5 0 LCL=0 1 2 3 4 5 6 7 8 9 Sample 10 11 12 13 14 15 Test Results for R Chart of Ex5-34Det TEST 1.93 10 _ R=5.326 = 2.38 6σˆ x 6(2. Without sample #12: USL − LSL 0.42 (d) Assume the cigar lighter detent is normally distributed.64 / 2.3200 Cˆ P = = = 1.3220 − 0.Chapter 5 Exercise Solutions 5-34 continued (b) Excluding Sample Number 12: MTB > Stat > Control Charts > Variables Charts for Subgroups > R Under “Options. One point more than 3.00 standard deviations from center line. Estimate” omit subgroup 12 and select Rbar.0001) 5-42 .42 × 0. Test Failed at points: 12 (c) Without sample #12: σˆ x = R / d 2 = 5. R Chart of Detent (Ex5-34Det) Sample 12 Excluded from Calculations 20 1 Sample Range 15 UCL=11.

0 1 1 1 1 1 UCL=3. Failed at points: 1. 2. 24 Test Results for R Chart of Ex5-35Det TEST Test TEST Test 1. One point more than 3.2 LCL=-3.00 standard deviations from center line. 2 out of 3 points more than 2 standard deviations from center line (on one side of CL). Failed at points: 24. 17. 19. 23 2.45 5 2. 13. Test Failed at points: 15. Failed at points: 24. One point more than 3. Failed at points: 12 2. 20. 25 5-43 . 13. 9 points in a row on same side of center line. 16. 20. 12. 4 out of 5 points more than 1 standard deviation from center line (on one side of CL). 18. 9 points in a row on same side of center line. 13-15 1 1 Sample Mean 5.05 1 12 14 Sample 16 18 20 22 24 1 15 UCL=11. 18.5 6 6 2 2 0.00 standard deviations from center line. Xbar-R Chart of Ex5-35Det Limits based on Samples 1-11.64 LCL=0 2 4 6 8 10 12 14 Sample 16 18 20 22 24 Test Results for Xbar Chart of Ex5-35Det TEST Test TEST Test TEST 1. 3. Test Failed at points: 2. 20 TEST 6.Chapter 5 Exercise Solutions 5-35 (5-28). Estimate” use subgroups 1:11 and 13:15. 22. 23.93 2 10 2 5 0 _ R=5. MTB > Stat > Control Charts > Variables Charts for Subgroups > R Under “Options. and select Rbar.0 2 4 6 8 10 20 Sample Range _ _ X=0. 25 5. 17.0 6 -2.5 1 -5.

0 2 4 6 8 10 20 Sample Range UCL=1.52 _ R=4. While the limits on the above charts may be used to monitor future production. 2. 5-44 . and 13 are out of control on the x chart. 3.5 5 6 6 2 5 2 -2.66 0. Note that samples 1. excluding 1.0 LCL=-3. No additional samples are out of control on the x chart.26 1 12 14 Sample 16 18 20 22 24 1 15 1 1 10 1 1 2 2 UCL=9.5 1 -5. the fact that 6 of 15 samples were out of control and eliminated from calculations is an early indication of process instability. 2.94 _ _ X=-0. Removing sample 3 gives Xbar-R Chart of Ex5-35Det Limits based on first 15 samples. the process appears to be unstable. 12.5 5 0 LCL=0 2 4 6 8 10 12 14 Sample 16 18 20 22 24 Sample 14 is now out of control on the R chart.0 1 1 1 1 1 1 2. (a) Given the large number of points after sample 15 beyond both the x and R control limits on the charts above. 12 and 13 1 1 Sample Mean 5. If these samples are removed and the limits recalculated. sample 3 is also out of control on the x chart.Chapter 5 Exercise Solutions 5-35 continued We are trying to establish trial control limits from the first 15 samples to monitor future production.

5 _ _ X=1.0 2 4 6 8 10 Sample Range 20 LCL=-3. 16.00 standard deviations from center line.55 Sample Mean 5.0 -2. 12.48 5 LCL=0 0 2 4 6 8 10 12 14 Sample 16 18 20 22 24 With Test 1 only: Test Results for Xbar Chart of Ex5-35Det TEST 1.Chapter 5 Exercise Solutions 5-35 continued (b) Xbar-R Chart of Detent (Ex5-35Det) All Samples in Calculations 1 1 1 UCL=5.0 2.00 standard deviations from center line.5 1 -5.23 0. 17 Test Results for R Chart of Ex5-35Det TEST 1. Test Failed at points: 1. One point more than 3. One point more than 3.82 15 10 _ R=7.08 1 12 14 Sample 16 18 20 22 24 1 UCL=15. 13. Test Failed at points: 12 5-45 .

0 2 4 6 8 10 Sample Range 20 1 12 14 Sample 16 18 20 22 24 1 15 UCL=14.96 1 -5. 13. 12. One point more than 3.48 10 _ R=6.00 standard deviations from center line. Test Failed at points: 12 5-46 . 20 Test Results for R Chart of Ex5-35Det TEST 1.85 5 0 LCL=0 2 4 6 8 10 12 14 Sample 16 18 20 22 24 With Test 1 only: Test Results for Xbar Chart of Ex5-35Det TEST 1. 16. 16. and 17 from calculations: Xbar-R Chart of Detent (Ex5-35Det) Samples 1. 13.Chapter 5 Exercise Solutions 5-35 (b) continued Removing samples 1.0 UCL=4. 17 excluded from calculations 1 1 1 1 Sample Mean 5. 12. 17.0 -2.5 _ _ X=0. One point more than 3.94 2.00 standard deviations from center line. 16. Test Failed at points: 1.99 0. 13.5 LCL=-2. 12.

Chapter 5 Exercise Solutions 5-35 continued Sample 20 is now also out of control.5 1 -5.66 2. 17. 16. 20 Test Results for R Chart of Ex5-35Det TEST 1.24 10 _ R=6. 17. 16.11 1 12 14 Sample 16 18 20 22 24 1 15 UCL=14. with runs of points both above and below the centerline. and that the sources of variation need to be identified and removed. 12.74 5 LCL=0 0 2 4 6 8 10 12 14 Sample 16 18 20 22 24 With Test 1 only: Test Results for Xbar Chart of Ex5-35Det TEST 1. 18. 13.00 standard deviations from center line. Test Failed at points: 12 Sample 18 is now out-of-control. Test Failed at points: 1.0 1 1 1 UCL=4. One point more than 3.0 -2. Xbar-R Chart of Ex5-35Det Samples 1. 13. 5-47 . One point more than 3.0 2 4 6 8 10 Sample Range 20 LCL=-3.78 0. for a total 7 of the 25 samples.00 standard deviations from center line.5 _ _ X=0. 12. 20 excluded from calculations 1 1 Sample Mean 5. Removing sample 20 from calculations. This suggests that the process is inherently unstable.

i y 2 2 (b) Want Pr{(x − y) < 0.i = 6.09 − z ⎞ ⎜ ⎟ = −2.300 σˆ x = Rx / d 2 = ∑ Rx . Let z = x − y.09 − z ⎞ Φ⎜ ⎟ = 0.326 = 0.500 ⎛ 0.09 − z ⎞ Φ −1 ⎜ ⎟ = Φ (0.5121 ⎝ 0.006) ⎝ 0. 30 30 i =1 i =1 n = 6.870.0 30 m σˆ x = R / d 2 = 45. Then σˆ z = σˆ x2 + σˆ y2 = 0. ∑ Rx .346 5-37 (5-30).0) = 0 (b) m ∑ xi 12.006.608 / 20) / 2.5121(0.978 ( = R / d = (∑ R 20 ) d = (18.0 30 m UCL R = D4 R = 2.18 R= i =1 = LCL R = D3 R = 0(45.978 /10) / 2. ∑ Ri = 1350.09} = 0. m y = 10. (a) 20 10 i =1 i =1 n = 5.004(45.i = 18.870 = 429.326 = 0. m = 30 (a) m ∑ Ri 1350 = 45.0) = 90.758 µˆ = x = i =1 = 5-48 .42 + 0.534 = 17. mx = 20.006 ⎝ σˆ z ⎠ ⎛ 0.09 = 1.400 / m ) d = (6. ∑ xi = 12.500) + 0. ∑ Ry .500 ⎠ z = +2.608.32 = 0.500 ⎠ ⎛ 0.Chapter 5 Exercise Solutions 5-36 (5-29).i / mx i =1 σˆ y 10 y 2 i =1 y .0 / 2.

5-38 (5-31).40 = 400 USL − LSL 480 − 400 = = 0. ∑ xi = 12.0516 = 0.9979 + 0.9213 = 14.63) = 1 − 0.0 30 m σˆ x = S / c4 = 13. LSL = 440 .87) + Φ (−1.870 = 429. m = 30 (a) m ∑ Si 410 = 13.834 µˆ = x = i =1 = 5-49 .Chapter 5 Exercise Solutions 5-37 continued (c) USL = 440 + 40 = 480.751 Cˆ p 6σˆ x 6(17. 30 30 i =1 i =1 n = 4. ∑ Si = 410.667 / 0.667 30 m UCL S = B4 S = 2.870.0537 ⎝ 17.969 S= i =1 = LCL S = B3 S = 0(13.667) = 30.758 ⎠ (d) To minimize fraction nonconforming the mean should be located at the nominal dimension (440) for a constant variance.758) ⎛ 480 − 429 ⎞ ⎛ 400 − 429 ⎞ pˆ = 1 − Φ ⎜ ⎟+Φ⎜ ⎟ = 1 − Φ (2.667) = 0 (b) m ∑ xi 12.758 ⎠ ⎝ 17.266(13.

807 ( = µ − k (σ UCL x = µ + kσ x = µ + k σ x LCL x = µ − kσ x x ) ( 4 ) = 111. LCL x = 96.50-197. (a) n = 4. 5-50 .9213) 6 ( S c4 ) The process is not capable of meeting specifications.772 n = 100 + 2.Chapter 5 Exercise Solutions 5-39 (5-32).992 1 − β 1 − Pr{not detect} 1 − 0. n = 5.6658 1 − [Pr{not detected}]3 = 1 − (0. µ = 98.000 0. 1 1 1 ARL1 = = = = 2. k = 3.6658 5-42 (5-35). µ = 100.56) ⎟ −Φ⎜ ⎟ = Φ⎜ σx σx ⎝ ⎠ ⎝ ⎠ ⎝ 8 5 ⎠ ⎝ 8 5 ⎠ = 0.0025 = 2.807 ( 8 4 ) = 88.807 8 5-40 (5-33).7678 6σˆ x 6 (1. centerline x = 100.005 / 2 = Z 0.2877 = 0. USL − LSL USL − LSL 202.9535 − 0. σ x = 8 ( = µ − 2 (σ UCL x = µ + 2σ x = µ + 2 σ x LCL x = µ − 2σ x x ) ( 4 ) = 108 n ) = 100 − 2 ( 8 4 ) = 92 n = 100 + 2 8 (b) k = Zα / 2 = Z 0. UCL x = 104. σ x = 8 Pr{out-of-control signal by at least 3rd plot point} = 1 − Pr{not detected by 3rd sample} = 1 − [Pr{not detected}]3 Pr{not detected} = Pr{LCL x ≤ x ≤ UCL x } = Pr{x ≤ UCL x } − Pr{x ≤ LCL x } ⎛ 104 − 98 ⎞ ⎛ 96 − 98 ⎞ ⎛ UCL x − µ ⎞ ⎛ LCL x − µ ⎞ = Φ⎜ ⎟−Φ⎜ ⎟ = Φ (1.50 Cˆ P = = = = 0.6658)3 = 0.7049 5-41 (5-34).68) − Φ (−0.228 n ) = 100 − 2.

5.0 1.9974.96 (10 4 ) = 190.75.05/ 2 = Z 0.0 2.0 k 5-51 .6000 0. 0.5 2.0}.0 0. L = 3 1.2266.25. LSL = 600 − 20 = 580 (a) USL − LSL USL − LSL 620 − 580 Cˆ P = = = = 1. L = 3.0.Chapter 5 Exercise Solutions 5-43 (5-36).0000} Operating Characteristic Curve for n = 9.9332.5.9213(10) = 9. 0. 0. σ x = 10 (a) centerline S = c4σ = 0.0013.970 ) 6 ( R d2 ) Process is capable of meeting specifications. n = 9. 0.0668. 0.111 6σˆ x 6 (17.7734. 3.0000 beta 0.82 / 2. 0.0.025 = 1. 0. 1. 0.8 1.0000.5 3.2 n = 200 + 1.2000 1.088(10) = 20. USL = 600 + 20 = 620.96 10 5-44 (5-37). µ = 200.8000 0. 0. 0.5.5 0.3 1.0000 0. 1.4000 0. 2. n = 4. 2.5.96 ( = µ − k (σ UCL x = µ + kσ x = µ + k σ x LCL x = µ − kσ x x ) ( 4 ) = 209.8 n ) = 200 − 1. (b) n = 9. β = {0. 1. β = Φ L − k n − Φ − L − k n ( ) ( ) for k = {0.213 UCL S = B6σ x = 2.2000 0.88 LCL S = B5σ x = 0(10) = 0 (b) k = Zα / 2 = Z 0.

59 / 2.419) = 0. 30 30 i =1 i =1 n = 7.324 UCL R = D 4 R = 1.304 (b) σˆ x = R / d 2 = 4 / 2. ∑ Ri = 120.479 (c) S = c4σˆ x = 0.696 LCL R = D 3 R = 0.0000 = 0.118(1.676 x= i =1 = LCL x = x − A2 R = 90 − 0.479) = 1.576 (12 9 ) = 589.671 LCL S = 0. n = 9.7 n = 600 + 2.419(4) = 88.076(4) = 0.005 = 2.3 n ) = 600 − 2.924(4) = 7.419) = 2.059 = 10 Pr{detect shift on 1st sample} = Pr{x < LCL} + Pr{x > UCL} = Pr{x < LCL} + 1 − Pr{x ≤ UCL} ⎛ 785 − 790 ⎞ ⎛ 815 − 790 ⎞ ⎛ LCL − µnew ⎞ ⎛ UCL − µnew ⎞ = Φ⎜ ⎟ +1− Φ ⎜ ⎟ ⎟ +1− Φ ⎜ ⎟ = Φ⎜ σx σx ⎝ ⎠ ⎝ ⎠ ⎝ 10 4 ⎠ ⎝ 10 4 ⎠ = Φ (−1) + 1 − Φ (5) = 0.576 12 5-47 (5-40). σ x = 12. α =0.882(1.1587 + 1 − 1.419(4) = 91. ∑ xi = 2700.576 ( = µ − k (σ UCL x = µ + kσ x = µ + k σ x LCL x = µ − kσ x x ) ( 9 ) = 610.01 k = Zα / 2 = Z 0.01/ 2 = Z 0. R = i =1 = =4 m m 30 30 UCL x = x + A2 R = 90 + 0.9594(1.704 = 1.419 UCL S = 1. m = 30 (a) m ∑ xi m ∑ Ri 2700 120 = 90. σˆ x = R / d 2 = 20. µ = 600.167 5-46 (5-39).Chapter 5 Exercise Solutions 5-45 (5-38).1587 5-52 .

5000 = 0.424 5-53 .1587 5-49 (5-42).Chapter 5 Exercise Solutions 5-48 (5-41).30 ARL1 = 1 − β 1 − Pr{not detect} Pr{detect} 0.0000 − 0.576 ( UCL x = x + kσ x = x + k σˆ x ( LCL x = 360 − 2.576 3 ) 9 = 362.970 = 3.5000 ⎝ 3 9 ⎠ ⎝ 3 9 ⎠ (d) α = 0.576 3 ) ) ( n = 360 + 2.005 = 2. (c) µnew = 357 ⎛ UCL − µnew Pr{not detect on 1st sample} = Pr{LCL ≤ x ≤ UCL} = Φ ⎜⎜ ⎝ σˆ x n ⎞ ⎛ LCL − µnew ⎟⎟ − Φ ⎜⎜ ⎠ ⎝ σˆ x n ⎞ ⎟⎟ ⎠ ⎛ 363 − 357 ⎞ ⎛ 357 − 357 ⎞ = Φ⎜ ⎟ −Φ⎜ ⎟ = Φ (6) − Φ (0) = 1. (a) σˆ x = R / d 2 = 8.667 6σˆ x 6(3) The process is not capable of producing all items within specification. k = Zα / 2 = Z 0.576 9 = 357.91/ 2. 1 1 1 1 = = = = 6.01/ 2 = Z 0.0026 ⎝ 3 9 ⎠ ⎝ 3 9 ⎠ (b) USL − LSL +6 − (−6) Cˆ P = = = 0.0013 + 1 − 0.000 ⎛ LCL − x ⎞ ⎛ UCL − x ⎞ ⎟ +1− Φ ⎜ ⎟ ⎝ σx ⎠ ⎝ σx ⎠ α = Pr{x < LCL} + Pr{x > UCL} = Φ ⎜ ⎛ 357 − 360 ⎞ ⎛ 363 − 360 ⎞ = Φ⎜ ⎟ +1− Φ ⎜ ⎟ = Φ (−3) + 1 − Φ (3) = 0.01.9987 = 0.

351 LCL S = B3 S = 0(3. (a) σˆ x = R / d 2 = 8.8944 = 0. try moving the center of the process from its current mean of 620 closer to the nominal dimension of 610.9772 + 1 − 1.059 = 4.0000 5-54 .266(3.25) = 0.25) + 1 − Φ (1.9772 (f) Pr{detect by 3rd sample} = 1 − Pr{not detect by 3rd sample} = 1 − (Pr{not detect})3 = 1 − (1 − 0.236 / 2.1056 (d) To reduce the fraction nonconforming.Chapter 5 Exercise Solutions 5-50 (5-43).000 (b) S = c4σˆ x = 0.685) = 8.9213(4) = 3. Also consider reducing the process variability.0000 = 0.9772)3 = 1.685) = 0 (c) ⎛ LSL − x ⎞ ⎛ USL − x ⎞ pˆ = Pr{x < LSL} + Pr{x > USL} = Φ ⎜ ⎟ +1− Φ ⎜ ⎟ ⎝ σˆ x ⎠ ⎝ σˆ x ⎠ ⎛ 595 − 620 ⎞ ⎛ 625 − 620 ⎞ = Φ⎜ ⎟ +1− Φ ⎜ ⎟ 4 4 ⎝ ⎠ ⎝ ⎠ = Φ (−6.0000 + 1 − 0.865 UCL S = B4 S = 2. (e) Pr{detect on 1st sample} = Pr{x < LCL} + Pr{x > UCL} ⎛ LCL − µnew ⎞ ⎛ UCL − µnew ⎞ = Φ⎜ ⎟ +1− Φ ⎜ ⎟ σx σx ⎝ ⎠ ⎝ ⎠ ⎛ 614 − 610 ⎞ ⎛ 626 − 610 ⎞ = Φ⎜ ⎟ +1− Φ ⎜ ⎟ ⎝ 4 4 ⎠ ⎝ 4 4 ⎠ = Φ (2) + 1 − Φ (8) = 0.

41) + 1 − Φ (8.9515 = 1.738 / 0.00.9920 (e) Pr{detect by 3rd sample} = 1 − Pr{not detect by 3rd sample} = 1 − (Pr{not detect})3 = 1 − (1 − 0.827) = 711.0503 + 1 − 0.8 − 702 ⎞ ⎛ 708.31) = 0.642) = 0.52 (c) pˆ = Pr{x < LSL} + Pr{x > USL} ⎛ LSL − x ⎞ ⎛ USL − x ⎞ = Φ⎜ ⎟ +1− Φ ⎜ ⎟ ⎝ σˆ x ⎠ ⎝ σˆ x ⎠ ⎛ 703 − 706 ⎞ ⎛ 709 − 706 ⎞ = Φ⎜ ⎟ +1− Φ ⎜ ⎟ ⎝ 1.1006 (d) Pr{detect on 1st sample} = Pr{x < LCL} + Pr{x > UCL} ⎛ LCL − µnew ⎞ ⎛ UCL − µnew ⎞ = Φ⎜ ⎟ +1− Φ ⎜ ⎟ σx σx ⎝ ⎠ ⎝ ⎠ ⎛ 703.9920 + 1 − 1.48 LNTL = 706 − 3(1.827 6 ⎠ ⎝ 1.Chapter 5 Exercise Solutions 5-51 (5-44).9920)3 = 1. (a) µˆ = x = 706.827 ⎠ ⎝ 1.642) + 1 − Φ (1.0000 = 0.9497 = 0.2 − 702 ⎞ = Φ⎜ ⎟ +1− Φ ⎜ ⎟ ⎝ 1. σˆ x = S / c4 = 1.827 ⎠ = Φ (−1.827) = 700.827 (b) UNTL = x + 3σˆ x = 706 + 3(1.0000 5-55 .827 6 ⎠ = Φ (2.

5) + 1 − Φ (2.83) = 0.new ⎝ ⎠ ⎝ ⎠ ⎛ 690 − 693 ⎞ ⎛ 710 − 693 ⎞ = Φ⎜ ⎟ +1− Φ ⎜ ⎟ ⎝ 12 4 ⎠ ⎝ 12 4 ⎠ = Φ (−0.3108 (e) ARL1 = 1 1 1 1 = = = = 3.15) + 1 − Φ (2.1251 + 1 − 0.3108 5-56 .661 (b) pˆ = Pr{x < LSL} + Pr{x > USL} ⎛ LSL − x ⎞ ⎛ USL − x ⎞ = Φ⎜ ⎟ +1− Φ ⎜ ⎟ ⎝ σˆ x ⎠ ⎝ σˆ x ⎠ ⎛ 690 − 700 ⎞ ⎛ 720 − 700 ⎞ = Φ⎜ ⎟ +1− Φ ⎜ ⎟ ⎝ 8.0104 + 1 − 0.9896 = 0. (a) µˆ = x = 700. σˆ x = S / c4 = 7.661 ⎠ ⎝ 8.661 4 ⎠ ⎝ 8.9896 = 0.0208 (d) Pr{detect on 1st sample} = Pr{x < LCL} + Pr{x > UCL} ⎛ LCL − µnew ⎞ ⎛ UCL − µnew ⎞ = Φ⎜ +1− Φ ⎜ ⎟ ⎟⎟ ⎜ σ ⎟ ⎜ σ x .31) = 0.661 ⎠ = Φ (−1.661 4 ⎠ = Φ (−2.31) = 0.979 / 0.3085 + 1 − 0.9977 = 0.1355 (c) α = Pr{x < LCL} + Pr{x > UCL} ⎛ LCL − x ⎞ ⎛ UCL − x ⎞ = Φ⎜ ⎟ +1− Φ ⎜ ⎟ ⎝ σx ⎠ ⎝ σx ⎠ ⎛ 690 − 700 ⎞ ⎛ 710 − 700 ⎞ = Φ⎜ ⎟ +1− Φ ⎜ ⎟ ⎝ 8.9213 = 8.31) + 1 − Φ (2.Chapter 5 Exercise Solutions 5-52 (5-45).22 1 − β 1 − Pr{not detect} Pr{detect} 0.new x .

021055 ⎠ 5-57 .1052 16.9964) = 0.08 16.05 LC L=16.08 U C L=0.02044 25 0.050 16.100 Ex5-53Wt 16. %underfilled = 100% × Pr{x < 16 oz} ⎛ 16 − 16.00003% ⎝ 0.075 16. x = 16. σˆ x = 0.11 16. MTB > Stat > Control Charts > Variables Charts for Individuals > I-MR I-MR Chart of Weight (Ex5-53Wt) Individual Value 16.11 0.00 LC L=0 2 4 6 8 10 12 14 O bser vation 16 18 20 22 24 There may be a “sawtooth” pattern developing on the Individuals chart.14 _ X=16.342 Percent 80 70 60 50 40 30 20 10 5 1 16.02375 0.125 16.02375 MTB > Stat > Basic Statistics > Normality Test Probability Plot of Weight (Ex5-53Wt) Normal 99 Mean StDev N AD P-Value 95 90 16.021055.04 __ M R=0.397 0.02 0.150 Visual examination of the normal probability indicates that the assumption of normally distributed coffee can weights is valid.1052. MR2 = 0.Chapter 5 Exercise Solutions 5-53 (5-46).17 U C L=16.06 0.0420 2 4 6 8 10 12 14 O bser vation 16 18 20 22 24 M oving Range 0.1052 ⎞ = 100% × Φ ⎜ ⎟ = 100% × Φ (−4.1684 16.07760 0.

5 5.Chapter 5 Exercise Solutions 5-54(5-47).50 M oving Range 10.0 __ M R=3.712 15 0.0 7.465 0. σˆ x = 2.27 50 45 LC L=44.27 2.82 Individual Value 60 55 _ X=53. MR2 = 3.2667.72 1 2 3 4 5 6 7 8 9 O bser vation 10 11 12 13 14 15 U C L=10.84954.217 Percent 80 70 60 50 40 30 20 10 5 1 46 48 50 52 54 Ex5-54Har 56 58 60 Although the observations at the tails are not very close to the straight line. the p-value is greater than 0.21429 MTB > Stat > Basic Statistics > Normality Test Probability Plot of Hardness (Ex5-54Har) Normal 99 Mean StDev N AD P-Value 95 90 53. MTB > Stat > Control Charts > Variables Charts for Individuals > I-MR I-MR Chart of Hardness (Ex5-54Har) U C L=61.05. indicating that it may be reasonable to assume that hardness is normally distributed.5 LC L=0 0. 5-58 .0 1 2 3 4 5 6 7 8 9 O bser vation 10 11 12 13 14 15 x = 53.21 2.

346.511 Percent 80 70 60 50 40 30 20 10 5 1 2600 2700 2800 2900 3000 Ex5-55Vis 3100 3200 3300 Viscosity measurements do appear to follow a normal distribution.9 3200 3000 _ X=2928. MR2 = 148.158 5-59 .Chapter 5 Exercise Solutions 5-55 (5-48).1 M oving Range 480 360 240 __ M R=148.9 2 4 6 8 10 12 O bser vation 14 16 18 20 U C L=484. or other patterns.2 120 LC L=0 0 2 4 6 8 10 12 O bser vation 14 16 18 20 The process appears to be in statistical control. trends.9 2800 2600 LC L=2534. (c) µˆ = x = 2928.9. (a) MTB > Stat > Basic Statistics > Normality Test Probability Plot of Viscosity (Ex5-55Vis) Normal 99 Mean StDev N AD P-Value 95 90 2929 129.0 20 0. (b) MTB > Stat > Control Charts > Variables Charts for Individuals > I-MR I-MR Chart of Viscosity (Ex5-55Vis) Individual V alue 3400 U C L=3322. with no out-of-control points. σˆ x = 131.319 0. runs.

Chapter 5 Exercise Solutions 5-56 (5-49).9 2 4 6 8 10 12 14 Observation 16 18 20 22 24 UCL=484.9 3000 2800 2600 LCL=2534. 5-60 .9 3200 _ X=2928. However all of the new points on the I chart are above the center line. MTB > Stat > Control Charts > Variables Charts for Individuals > I-MR I-MR Chart of Viscosity (Ex5-56Vis) With five next measurements 3400 Individual Value UCL=3322. indicating that a shift in the mean may have occurred.2 120 LCL=0 0 2 4 6 8 10 12 14 Observation 16 18 20 22 24 All points are inside the control limits.1 Moving Range 480 360 240 __ MR=148.

534 30 0.338 0. 5-61 .480 Percent 80 70 60 50 40 30 20 10 5 1 40 45 50 Ex5-57aTh 55 60 The normality assumption is reasonable.75 5 0 LC L=0 3 6 9 12 15 18 O bser vation 21 24 27 30 The process is in statistical control. MTB > Stat > Basic Statistics > Normality Test Probability Plot of Oxide Thickness (Ex5-57aTh) Normal 99 Mean StDev N AD P-Value 95 90 49.79 15 10 __ M R=5.85 50 40 LC L=34. (a) MTB > Stat > Control Charts > Variables Charts for Individuals > I-MR I-MR Chart of Oxide Thickness (Ex5-57aTh) Individual V alue U C L=65.55 30 3 6 9 12 15 18 O bser vation 21 24 27 30 M oving Range 20 U C L=18.Chapter 5 Exercise Solutions 5-57 (5-50).85 4.14 60 _ X=49.

2 out of 3 points more than 2 standard deviations from center line (on one side of CL). 40 5. 37.79 15 10 __ MR=5. Test Failed at points: 34. 39. However. 39. 5-62 .55 30 4 8 12 16 20 Observation 24 28 32 36 40 Moving Range 20 UCL=18. 39.85 50 40 LCL=34. 9 points in a row on same side of center line.14 _ X=49. One point more than 3. 40 We have turned on some of the sensitizing rules in MINITAB to illustrate their use.00 standard deviations from center line. 4 out of 5 points more than 1 standard deviation from center line (on one side of CL). Test Failed at points: 35. and several 2 of 3 beyond 2 sigma on the x chart. Failed at points: 38. it is clear that the process is out of control during this period of operation. even without use of the sensitizing rules.Chapter 5 Exercise Solutions 5-57 continued (b) MTB > Stat > Control Charts > Variables Charts for Individuals > I-MR I-MR Chart of Oxide Thickness (Ex5-57bTh) With 10 new measurements and some sensitizing rules 70 1 Individual Value 5 60 6 6 2 2 UCL=65.75 5 0 LCL=0 4 8 12 16 20 Observation 24 28 32 36 40 Test Results for I Chart of Ex5-57bTh TEST Test TEST Test TEST 1. There is a run above the centerline. several 4 of 5 beyond 1 sigma. Failed at points: 38 2. 38. 40 TEST 6.

55 30 1 6 12 18 24 30 Observation 36 48 54 60 1 20 Moving Range 42 UCL=18.85 50 40 LCL=34. with Sensitizing Rules On 70 1 Individual Value 5 60 6 6 2 UCL=65.75 5 0 LCL=0 1 6 12 18 24 30 Observation 36 42 48 54 60 The process has been returned to a state of statistical control.Chapter 5 Exercise Solutions 5-57 continued (c) MTB > Stat > Control Charts > Variables Charts for Individuals > I-MR I-MR Chart of Oxide Thickness (Ex5-57cTh) 10 + 20 New Measurements.79 15 10 __ MR=5.14 2 _ X=49. 5-63 .

with two Western Electric rule violations.00 standard deviations from center line.Chapter 5 Exercise Solutions 5-58 (5-51).88 5 100 80 _ X=73. (b) MTB > Stat > Control Charts > Variables Charts for Individuals > I-MR I-MR Chart of Concentration (Ex5-58C) Individual V alue U C L=104. Test Failed at points: 17 The process is not in control. Test Failed at points: 11 Test Results for MR Chart of Ex5-58C TEST 1. (a) The normality assumption is a little bothersome for the concentration data.73 60 LC L=42.26 30 20 __ M R=11. in particular due to the curve of the larger values and three distant values.59 40 3 6 9 12 15 18 O bser vation 24 27 30 1 40 M oving Range 21 U C L=38.71 10 0 LC L=0 3 6 9 12 15 18 O bser vation 21 24 27 30 Test Results for I Chart of Ex5-58C TEST 5. 5-64 . 2 out of 3 points more than 2 standard deviations from center line (on one side of CL). One point more than 3.

again due to the curve of the larger values and three distant values.9 4.288 0.3 4.2 4.0 4.327 Percent 80 70 60 50 40 30 20 10 5 1 3. 5-65 .1567 30 0.7 The normality assumption is still troubling for the natural log of concentration.1 4.6 4.5 4.Chapter 5 Exercise Solutions 5-58 continued (c) MTB > Stat > Basic Statistics > Normality Test Probability Plot of ln(Concentration) (Ex5-58lnC) Normal 99 Mean StDev N AD P-Value 95 90 4.408 0.4 Ex5-58lnC 4.

8689 3 6 9 12 15 18 O bser vation 21 24 27 30 1 U C L=0.2884 4.24 __ M R=0. 5-66 .8 U C L=4.12 0.00 LC L=0 3 6 9 12 15 18 O bser vation 21 24 27 30 Test Results for I Chart of Ex5-58lnC TEST 5. Test Failed at points: 11 Test Results for MR Chart of Ex5-58lnC TEST 1. with the same to Western Electric Rules violations. One point more than 3.7079 5 4.2 4. There does not appear to be much difference between the two control charts (actual and natural log). 2 out of 3 points more than 2 standard deviations from center line (on one side of CL). Test Failed at points: 17 The process is still not in control.48 0.00 standard deviations from center line.Chapter 5 Exercise Solutions 5-58 continued (d) MTB > Stat > Control Charts > Variables Charts for Individuals > I-MR I-MR Chart of ln(Concentration) (Ex5-58lnC) Individual Value 4.1577 0.6 4.4 _ X=4.5154 M oving Range 0.0 LC L=3.36 0.

6 100 0 LC L=0 2 4 6 8 10 12 O bser vation 14 16 18 20 I-MR Chart of Ex5-59Vel Test Results for MR Chart of Ex5-59Vel TEST 1.7 300 200 __ M R=122. For now. consider the process to be in a state of statistical process control. One point more than 3.9 500 2 4 6 8 10 12 O bser vation 14 16 18 20 1 M oving Range 400 U C L=400. There may also be an early indication of less variability in the later measurements.067 Percent 80 70 60 50 40 30 20 10 5 1 600 700 800 900 Ex5-59Vel 1000 1100 1200 Velocity of light measurements are approximately normally distributed. Since neither of these measurements seems unusual.1 1000 _ X=909 750 LC L=582. MTB > Stat > Control Charts > Variables Charts for Individuals > I-MR I-MR Chart of Velocity of Light (Ex5-59Vel) Individual V alue 1250 U C L=1235.00 standard deviations from center line. MTB > Stat > Basic Statistics > Normality Test Probability Plot of Velocity of Light (Ex5-59Vel) Normal 99 Mean StDev N AD P-Value 95 90 909 104. 5-67 .Chapter 5 Exercise Solutions 5-59☺. use all data for control limits calculations.9 20 0. Test Failed at points: 8 The out-of-control signal on the moving range chart indicates a significantly large difference between successive measurements (7 and 8).672 0.

00 standard deviations from center line. One point more than 3. 40 Test Results for MR Chart of Ex5-60Vel TEST Test TEST Test 1. 5-68 .Chapter 5 Exercise Solutions 5-60☺.9 500 4 8 12 16 20 Observation 24 28 32 36 40 1 UCL=400. Estimate to specify which subgroups to use in calculations I-MR Chart of Velocity of Light (Ex5-60Vel) New measurements with old limits Individual Value 1250 UCL=1235.7 Moving Range 400 300 200 100 2 __ MR=122. which is reflected in the number of observations below the centerline of the moving range chart. 9 points in a row on same side of center line. Test Failed at points: 36. Failed at points: 36. 37.1 1000 _ X=909 2 2 2 750 2 2 LCL=582. Failed at points: 8 2. (a) MTB > Stat > Control Charts > Variables Charts for Individuals > I-MR Select I-MR Options. There is a distinct downward trend in measurements. meaning the method is producing gradually smaller measurements. 39. 38. however the method of measuring is producing varying results—this is a chart of the measurement process. 9 points in a row on same side of center line.6 2 0 LCL=0 4 8 12 16 20 Observation 24 28 32 36 40 I-MR Chart of Ex5-60Vel Test Results for I Chart of Ex5-60Vel TEST 2. 37 The velocity of light in air is not changing. (b) Early measurements exhibit more variability than the later measurements.

3493 30 0.546 30 1.shaped curve to the plot points on a normal probability plot. as well as the Anderson-Darling test p-value. so a compressive transform such as natural log or square-root may be appropriate.2 3.07 5.626 0.158 <0.0 2.6 The distribution of the natural-log transformed uniformity measurements is approximately normally distributed. as evidenced by the “S”.4 2.093 Percent 80 70 60 50 40 30 20 10 5 1 2. (a) MTB > Stat > Basic Statistics > Normality Test Probability Plot of Uniformity Determinations (Ex5-61Un) Normal 99 Mean StDev N AD P-Value 95 90 15.005 Percent 80 70 60 50 40 30 20 10 5 1 0 5 10 15 20 Ex5-61Un 25 30 35 The data are not normally distributed. Probability Plot of ln(Uniformity) (Ex5-61lnUn) Normal 99 Mean StDev N AD P-Value 95 90 2.653 0. 5-69 .Chapter 5 Exercise Solutions 5-61☺. The data are skewed right.8 Ex5-61lnUn 3.

0 _ X=2.00 0.75 0.720 1.5 3.50 __ M R=0. 5-70 .00 LC L=0 3 6 9 12 15 18 O bser vation 21 24 27 30 The etching process appears to be in statistical control.351 0.0 LC L=1.586 Individual V alue 3.653 2.25 0.5 2.Chapter 5 Exercise Solutions 5-61 continued (b) MTB > Stat > Control Charts > Variables Charts for Individuals > I-MR I-MR Chart of ln (Uniformity) (Ex5-61lnUn) U C L=3.5 3 6 9 12 15 18 O bser vation 21 24 27 30 U C L=1.146 M oving Range 1.

824 0.01847 20 1.005 Percent 80 70 60 50 40 30 20 10 5 1 0.87 Purity is not normally distributed. 5-71 .81 0.85 0.86 0.174 <0.83 0.79 0.78 0.82 0.84 Ex5-62Pur 0.Chapter 5 Exercise Solutions 5-62 (5-52).80 0. (a) MTB > Stat > Basic Statistics > Normality Test Probability Plot of Batch Purity (Ex5-62Pur) Normal 99 Mean StDev N AD P-Value 95 90 0.

86459 5 0. 20 The process is not in statistical control. Test Failed at points: 11. (c) all data: µˆ = 0. σˆ x = 0.824 0.84 _ X=0.000 LC L=0 2 4 6 8 10 12 O bser vation 14 16 18 20 Test Results for I Chart of Ex5-62Pur TEST 1. 4 out of 5 points more than 1 standard deviation from center line (on one side of CL).824 .00 standard deviations from center line. 2 out of 3 points more than 2 standard deviations from center line (on one side of CL).80 LC L=0.78341 0. Test Failed at points: 19 TEST 6.82 6 0.8216 . One point more than 3.Chapter 5 Exercise Solutions 5-62 continued (b) MTB > Stat > Control Charts > Variables Charts for Individuals > I-MR I-MR Chart of Purity (Ex5-62Pur) 1 Individual V alue U C L=0.86 6 0.0133 5-72 .01526 0.024 __ M R=0.04987 M oving Range 0.012 0.036 0. Test Failed at points: 18 TEST 5.048 0. σˆ x = 0.78 2 4 6 8 10 12 O bser vation 14 16 18 20 U C L=0.0135 without sample 18: µˆ = 0.

08 16.08 U C L=0.05 LC L=16.1052 16.Chapter 5 Exercise Solutions 5-63 (5-53). control limits for both are essentially the same.14 _ X=16.1681 16. 5-73 . MTB > Stat > Control Charts > Variables Charts for Individuals > I-MR Select “Estimate” to change the method of estimating sigma I-MR Chart of Can Weight (Ex5-53Wt) Individual V alue 16.11 16.06 0.07726 0.02365 0.00 LC L=0 2 4 6 8 10 12 14 O bser vation 16 18 20 22 24 There is no difference between this chart and the one in Exercise 5-53.0423 2 4 6 8 10 12 14 O bser vation 16 18 20 22 24 M oving Range 0.02 0.17 U C L=16.04 __ M R=0.

27 50 LC L=45.5 5.66 7.96 2. with no practical difference for this set of observations.13 Individual V alue 60 55 _ X=53.5 0.41 45 1 2 3 4 5 6 7 8 9 O bser vation 10 11 12 13 14 15 M oving Range 10. 5-74 .0 __ M R=2.0 U C L=9.0 LC L=0 1 2 3 4 5 6 7 8 9 O bser vation 10 11 12 13 14 15 The median moving range method gives slightly tighter control limits for both the Individual and Moving Range charts.Chapter 5 Exercise Solutions 5-64 (5-54). MTB > Stat > Control Charts > Variables Charts for Individuals > I-MR Select “Estimate” to change the method of estimating sigma I-MR Chart of Hardness-Coded (Ex5-54Har) U C L=61.

1 2 4 6 8 10 12 O bser vation 14 16 18 20 U C L=502. MTB > Stat > Control Charts > Variables Charts for Individuals > I-MR Select “Estimate” to change the method of estimating sigma I-MR Chart of Polymer Viscosity (Ex5-55Vis) Individual Value 3400 U C L=3337. 5-75 .7 120 0 LC L=0 2 4 6 8 10 12 O bser vation 14 16 18 20 The median moving range method gives slightly wider control limits for both the Individual and Moving Range charts.7 3200 3000 _ X=2928. with no practical meaning for this set of observations.9 2800 2600 LC L=2520.2 M oving Range 480 360 240 __ M R=153.Chapter 5 Exercise Solutions 5-65 (5-55).

00 standard deviations from center line. Test Failed at points: 34. 39. MTB > Stat > Control Charts > Variables Charts for Individuals > I-MR (a) I-MR Chart of Oxide Thickness (Ex5-57cTh) All 60 Observations--Average Moving Range Method 70 1 Individual Value 5 60 2 6 UCL=66. the out-of-control signal on the MR chart for observation 41 is reflected by the shift between observations 40 and 41 on the Individuals chart.05 50 40 LCL=36. Test Failed at points: 41 Recall that observations on the Moving Range chart are correlated with those on the Individuals chart—that is. One point more than 3. 2 out of 3 points more than 2 standard deviations from center line (on one side of CL). 40 TEST 6.Chapter 5 Exercise Solutions 5-66 (5-56).03 1 6 12 18 24 30 Observation 36 48 54 60 1 20 Moving Range 42 UCL=18. 39. 37. Remove observation 38 and recalculate control limits. 40 5.07 2 6 _ X=51. Failed at points: 38 2. 4 out of 5 points more than 1 standard deviation from center line (on one side of CL). 9 points in a row on same side of center line. Test Failed at points: 35.46 15 10 __ MR=5.00 standard deviations from center line. 40 Test Results for MR Chart of Ex5-57cTh TEST 1. 39. Failed at points: 38. 38. 5-76 . One point more than 3.65 5 0 LCL=0 1 6 12 18 24 30 Observation 36 42 48 54 60 Test Results for I Chart of Ex5-57cTh TEST Test TEST Test TEST 1.

37.00 standard deviations from center line. 40 5.22 15 10 5 __ MR=5. 4 out of 5 points more than 1 standard deviation from center line (on one side of CL). One point more than 3.Chapter 5 Exercise Solutions 5-66 (a) continued Excluding observation 38 from calculations: I-MR Chart of Oxide Thickness (Ex5-57cTh) Less Observation 38 -. 2 out of 3 points more than 2 standard deviations from center line (on one side of CL). One point more than 3. Test Failed at points: 41 5-77 . 40 Test Results for MR Chart of Ex5-57cTh TEST 1.94 1 6 12 18 24 30 Observation 36 42 48 54 60 1 20 Moving Range UCL=18. 38. 39. Failed at points: 38 2. Test Failed at points: 35.58 0 LCL=0 1 6 12 18 24 30 Observation 36 42 48 54 60 Test Results for I Chart of Ex5-57cTh TEST Test TEST Test TEST 1. Failed at points: 38.Average Moving Range Method 70 1 Individual Value 5 60 2 6 UCL=65. 39.60 2 6 _ X=50. 40 TEST 6. 9 points in a row on same side of center line. 39.00 standard deviations from center line.77 50 40 LCL=35. Test Failed at points: 34.

Test Failed at points: 41 5-78 .27 1 6 12 18 24 30 Observation 36 42 48 54 60 1 20 Moving Range UCL=18.00 standard deviations from center line. Test Failed at points: 34. 39.00 standard deviations from center line.16 15 10 5 __ MR=5. 40 5. One point more than 3. 9 points in a row on same side of center line.Chapter 5 Exercise Solutions 5-66 continued (b) I-MR Chart of Oxide Thickness (Ex5-57cTh) All 60 Observations -. One point more than 3.Median Moving Range Method 70 1 Individual Value 5 60 2 6 UCL=65. 39. 39. 2 out of 3 points more than 2 standard deviations from center line (on one side of CL).05 50 40 LCL=36. 37. Failed at points: 38 2. 40 Test Results for MR Chart of Ex5-57cTh TEST 1.56 0 LCL=0 1 6 12 18 24 30 Observation 36 42 48 54 60 Test Results for I Chart of Ex5-57cTh TEST Test TEST Test TEST 1. Failed at points: 38. 4 out of 5 points more than 1 standard deviation from center line (on one side of CL). 40 TEST 6. 38.83 2 6 _ X=51. Test Failed at points: 35.

9 points in a row on same side of center line. 39.93 1 6 12 18 24 30 Observation 36 42 48 54 60 1 20 Moving Range UCL=17. 39.00 standard deviations from center line. 4 out of 5 points more than 1 standard deviation from center line (on one side of CL). 2 out of 3 points more than 2 standard deviations from center line (on one side of CL).77 50 40 LCL=36. 40 Test Results for MR Chart of Ex5-57cTh TEST 1. 40 TEST 6. One point more than 3. 40 5. Test Failed at points: 41 (c) The control limits estimated by the median moving range are tighter and detect the shift in process level at an earlier sample. 33.61 6 _ X=50. 37.Chapter 5 Exercise Solutions 5-66 (b) continued Excluding observation 38 from calculations: I-MR Chart of Oxide Thickness (Ex5-57cTh) Excluding Observation 38 from Calculations -. 5-79 . 38 2. Test Failed at points: 35. Failed at points: 38. Failed at points: 33.Median Moving Range Method 70 1 1 Individual Value 5 60 2 6 2 UCL=64.00 standard deviations from center line. One point more than 3. Test Failed at points: 34. 38.20 0 LCL=0 1 6 12 18 24 30 Observation 36 42 48 54 60 Test Results for I Chart of Ex5-57cTh TEST Test TEST Test TEST 1.00 15 10 5 __ MR=5. 39.

Chapter 5 Exercise Solutions 5-67 (5-57).157 (b) MTB > Stat > Basic Statistics > Descriptive Statistics Descriptive Statistics: Ex5-67Meas Variable Ex5-67Meas Total Count 25 Mean 10.079 2 4 6 8 10 12 14 O bser vation 16 18 20 22 24 U C L=4.549 10 8 LC L=7.262 M oving Range 4 3 2 __ M R=1. (a) MTB > Stat > Control Charts > Variables Charts for Individuals > I-MR I-MR Chart of Measurements (Ex5-67Meas) Individual V alue 14 U C L=14.305 1 0 LC L=0 2 4 6 8 10 12 14 O bser vation 16 18 20 22 24 σˆ x = R / d 2 = 1.018 12 _ X=10.342 / 0.549 StDev 1.342 Median 10.682 5-80 .7979 = 1.630 σˆ x = S / c4 = 1.305 /1.128 = 1.

there are fewer observations to estimate the standard deviation.693 = 1.598 / 2.435 (e) As the span of the moving range is increased.283 1 LCL=0 0 2 4 6 8 10 12 14 Observation 16 18 20 22 24 σˆ x = R / d 2 = 1.735 = 1.128 = 1.961 12 _ X=10. For this example.137 (d) Average MR3 Chart: σˆ x = R / d 2 = 2.283 /1. 5-81 .059 = 1.210 Average MR4 Chart: σˆ x = R / d 2 = 2.689 = 1.262 Average MR19 Chart: σˆ x = R / d 2 = 5.049 /1.137 2 4 6 8 10 12 14 Observation 16 18 20 22 24 UCL=4. This tends to be true for unstable processes. σ gets larger as the span increases.36 / 3.549 10 8 LCL=7.192 Moving Range 4 3 2 __ MR=1. and the estimate becomes less reliable.406 Average MR20 Chart: σˆ x = R / d 2 = 5.186 / 3.Chapter 5 Exercise Solutions 5-67 continued (c) MTB > Stat > Control Charts > Variables Charts for Individuals > I-MR I-MR Chart of Measurements (Ex5-67Meas) Median Moving Range Method--Span = 2 Individual Value 14 UCL=13.

07074 _ R=0... Ex5-68v5 Standard Deviations Between 0. Ex5-68v5) Subgroup Mean UCL=5.04 0.08 UCL=0.00 LCL=0 2 4 6 8 10 Ex5-68Cast 12 14 16 18 20 I-MR-R/S Standard Deviations of Ex5-68v1. MTB > Stat > Control Charts > Variables Charts for Subgroups > I-MR-R/S (Between/Within) Select “I-MR-R/S Options.7566 5.0143831 Between/Within 0.03345 0..6563 MR of Subgroup Mean 2 4 6 8 10 12 14 16 18 20 UCL=0. . 5-82 .70 LCL=5. Estimate” and choose R-bar method to estimate standard deviation I-MR-R (Between/Within) Chart of Vane Heights (Ex5-68v1.10 0.0377 0. For this example.Chapter 5 Exercise Solutions 5-68 (5-58).00 LCL=0 2 4 6 8 10 12 14 16 18 20 Sample Range 0.. the R chart tells the same story—same data pattern and no out-of-control points. When compared to the s chart for all data.0358361 The Individuals and Moving Range charts for the subgroup means are identical. . the control schemes are identical.82 _ X=5.0328230 Within 0.76 5.1233 0..8569 5.05 __ MR=0..

Ex5-69d5 Test Results for Xbar Chart of Ex5-69d1. 9. plot points on the x chart bounce below and above the control limits. .. between castings or within a casting.70 1 1 2 4 6 8 10 Sample 12 14 16 18 20 U C L=0. (a) MTB > Stat > Control Charts > Variables Charts for Subgroups > Xbar-R Xbar-R Chart of Casting Diameter (Ex5-69d1. 5-83 . we might expect that the diameter of a single casting will not change much with location..82 1 Sample M ean 1 11.03 0.09 _ R=0. 2 out of 3 points more than 2 standard deviations from center line (on one side of CL). Since these are high precision castings. 7. If no assignable cause can be found for these out-of-control points. Ex5-69d5 TEST 1..7226 11.Chapter 5 Exercise Solutions 5-69 (5-59)..06 0.00 standard deviations from center line. This will lead to a understanding of the greatest source of variability. we may want to consider treating the averages as an Individual value and graphing “between/within” range charts. Test Failed at points: 7 (b) Though the R chart is in control. Test Failed at points: 5.1292 Sample Range 0. 17 TEST 5.76 _ _ X=11.7931 11. Ex5-69d5) 1 11.00 LC L=0 2 4 6 8 10 Sample 12 14 16 18 20 Xbar-R Chart of Ex5-69d1.. .. 13.73 LC L=11. One point more than 3..7579 11.0611 0..79 U C L=11. ..12 0.

7 LCL=11..9 Subgroup Mean UCL=11. (e) The “within” chart is the usual R chart (n > 1).. damage.0262640 Between/Within 0. Though the nature of this process leads us to believe that the diameter at any location on a single casting does not change much.. in the wax mold.6472 MR of Subgroup Mean 2 4 6 8 10 12 14 16 18 20 UCL=0. Ex5-69d5 Standard Deviations Between 0.0437327 (d) We are taking several diameter measurements on a single precision casting...Chapter 5 Exercise Solutions 5-69 continued (c) MTB > Stat > Control Charts > Variables Charts for Subgroups > I-MR-R/S (Between/Within) Select “I-MR-R/S Options.8 _ X=11.7579 11.00 2 4 6 8 10 12 14 16 18 20 Sample Range UCL=0.05 0.0416 0.05 LCL=0 0. Ex5-69d5 I-MR-R/S Standard Deviations of Ex5-69d1. ...0611 0..1292 0.10 __ MR=0. etc. . It describes the measurement variability within a sample (variability in diameter of a single casting). Ex5-69d5 11.1360 0.. .0349679 Within 0.8685 11. 5-84 . Estimate” and choose R-bar method to estimate standard deviation MR-R/S (Between/Within) Chart of Casting Diameter (Ex5-69d1..00 LCL=0 2 4 6 8 10 Sample 12 14 16 18 20 I-MR-R/S (Between/Within) Chart of Ex5-69d1.10 _ R=0. we should continue to monitor “within” to look for wear.

needs to be monitored. If it is necessary to separately monitor the variation at each location. then only one location. (d) Within-wafer variability can still be monitored with randomly selected test sites. within-wafer variability could be monitored with a standard X − R control chart. However. 5-85 . a multivariate process control technique could be used.Chapter 5 Exercise Solutions 5-70 (5-60). (a) Both total process variability and the overall process average could be estimated from a single measurement on one wafer from each lot. with one X − R chart. Then a chart for individual x and moving range would provide information on lot-to-lot variability. Individuals X and Moving Range charts should be used for process monitoring. then either five X − R charts or some multivariate technique is needed. If the positions are essentially identical. (b) Assuming that each wafer is processed separately. The Minitab “between/within” control charts do this in three graphs: (1) wafer mean ( xww ) is an “individual value”. Alternatively. no information will be obtained about the pattern of variability within a wafer. (c) Both between-wafer and total process variability could be estimated from measurements at one point on five consecutive wafers. (e) The simplest scheme would be to randomly select one wafer from each lot and treat the average of all measurements on that wafer as one observation. (2) moving range is the difference between successive wafers. The data from each wafer could also be used to monitor between-wafer variability by maintaining an individuals X and moving range chart for each of the five fixed positions. and (3) sample range is the difference within a wafer ( Rww ) .

074 − 3(0.05 2.9 Mean StDev N AD P-Value 99 Percent 95 90 2.074 0.333 <0. s = 0.074.005 80 70 60 50 40 30 20 10 5 1 0.1 1. the plot points do fall along a straight line.95 2.90 1.939 5-86 .04515) = 1. The natural tolerance limits (± 3 sigma above and below mean) are: x = 2.25 Although the p-value is very small. The wafer critical dimension is approximately normally distributed.04515 UNTL = x + 3s = 2. (a) MTB > Stat > Basic Statistics > Normality Test Probability Plot of Critical Dimensions (Ex5-71All) Normal 99.10 Ex5-71All 2.00 2.04515 200 1.Chapter 5 Exercise Solutions 5-71 (5-61).15 2.209 LNTL = x − 3s = 2.04515) = 2.074 + 3(0. with many repeated values.20 2.

00 LCL=0 4 8 12 16 20 24 28 32 Sample (Lot Number-Wafer Order) 36 40 The Range chart is in control..14 Sample Range 0. construct an R chart for each sample of 5 wafer positions (two wafers per lot number).07 0. .16 UCL=0.04 0.12 0. indicating that within-wafer variability is also in control.10 0. MTB > Stat > Control Charts > Variables Charts for Subgroups > R R Chart of Critical Dimension Within Wafer (Ex5-71p1.1480 0..Chapter 5 Exercise Solutions 5-71 continued (b) To evaluate within-wafer variability.08 _ R=0.02 0. Ex5-71p5) 0. 5-87 .06 0.. for a total of 40 subgroups.

. set up Individuals and Moving Range charts where the x statistic is the average wafer measurement and the moving range is calculated between two wafer averages.00 LCL=1. indicating that between-wafer variability is also in-control.1603 2.16 UCL=0. 5-88 .00 4 8 12 16 20 24 28 32 36 40 Sample Range 0. Estimate” and choose R-bar method to estimate standard deviation I-MR-R/S (Between/Within) Chart of Crit Dim (Ex5-71p1.0735 6 6 2.1066 0.0395043 Both “between” control charts (Individuals and Moving Range) are in control.05 __ MR=0. .00 LCL=0 4 8 12 16 20 Sample 24 28 32 36 40 I-MR-R/S Standard Deviations of Ex5-71p1. The “within” chart (Range) is not required to evaluate variability between wafers...07 0.0326 LCL=0 0.08 0..10 0.0300946 Between/Within 0. .9868 MR of Subgroup Mean 4 8 12 16 20 24 28 32 36 40 UCL=0. Ex5-71p5) Subgroup Mean Variability between wafers 2.16 UCL=2. Ex5-71p5 Standard Deviations Between 0..0255911 Within 0.Chapter 5 Exercise Solutions 5-71 continued (c) To evaluate variability between wafers. MTB > Stat > Control Charts > Variables Charts for Subgroups > I-MR-R/S (Between/Within) Select “I-MR-R/S Options.1480 _ R=0.08 _ X=2..

5-89 .0735 2.0459 0. MTB > Stat > Control Charts > Variables Charts for Subgroups > I-MR-R/S (Between/Within) I-MR-R/S (Between/Within) Chart of Ex5-71All Subgroup Mean Lot-to-Lot Variability 2.00 LCL=0 2 4 6 8 10 12 14 16 18 20 Sample Range UCL=0.1 _ X=2.05 LCL=0.2 UCL=2.0503081 All three control charts are in control. (2) moving range between lot averages.9515 2 4 6 8 10 12 14 16 18 20 MR of Subgroup Mean 0.096 0.0214 2 4 6 8 10 Ex5-7Lot All 12 14 16 18 20 I-MR-R/S Standard Deviations of Ex5-71All Standard Deviations Between 0.15 _ R=0.1956 2. three charts are needed: (1) lot average.10 0. and (3) range within a lot—the Minitab “between/within” control charts.0394733 Within 0.16 UCL=0.Chapter 5 Exercise Solutions 5-71 continued (d) To evaluate lot-to-lot variability.1500 0.0 LCL=1.08 __ MR=0. indicating that the lot-to-lot variability is also in-control.1706 0.0311891 Between/Within 0.

04 0.0585) = 0.06 0.16 1 0. For these solutions. New exercises are denoted with an “☺”.1289 n 100 LCL p = p − 3 p (1 − p ) 0. MINITAB defines some sensitizing rules for control charts differently than the standard rules.0704 ⇒ 0 n 100 MTB > Stat > Control Charts > Attributes Charts > P P Chart of Nonconforming Assemblies (Ex6-1Num) 0.08 _ P=0. This can be changed under Tools > Options > Control Charts and Quality Tools > Define Tests. not 8.0585 0.0585 20(100) UCL p = p + 3 p (1 − p ) 0.0585 + 3 = 0.0585 − 3 = 0.Chapter 6 Exercise Solutions Notes: 1.0585(1 − 0. a run of n consecutive points on one side of the center line is defined as 9 points. p = i =1 ∑ Di i =1 mn = 117 = 0.0585) = 0. In particular.10 0. m m n = 100. m = 20.0585 − 0. 2. 6-1. 3. If a plot point is within the control limits.14 UCL=0.0585(1 − 0. Test Failed at points: 12 6-1 .12 0.02 0. we follow the MINITAB convention for determining whether a point is out of control. If a plot point is on or beyond the control limits. it is considered to be in control.1289 Proportion 0. Also fewer special cause tests are available for attributes control charts. One point more than 3.00 LCL=0 2 4 6 8 10 12 Sample 14 16 18 20 Test Results for P Chart of Ex6-1Num TEST 1. it is considered to be out of control.00 standard deviations from center line. ∑ Di = 117.

0537 − 0.14 UCL=0.0676 ⇒ 0 100 MTB > Stat > Control Charts > Attributes Charts > P P Chart of Nonconforming Assemblies (Ex6-1Num) Sample 12 removed from calculations 0.10 0.Chapter 6 Exercise Solutions 6-1 continued Sample 12 is out-of-control. m = 19.16 1 0.04 0.00 LCL=0 2 4 6 8 10 12 Sample 14 16 18 20 Test Results for P Chart of Ex6-1Num TEST 1.1213 Proportion 0.0537) = 0. so remove from control limit calculation: m m n = 100. ∑ Di = 102. Test Failed at points: 12 6-2 .0537) = 0.0537 0.0537 19(100) UCL p = 0. p = i =1 ∑ Di i =1 mn = 102 = 0. One point more than 3.0537 − 3 0.08 _ P=0.02 0.0537(1 − 0.0537(1 − 0.12 0.06 0.1213 100 LCL p = 0.0537 + 3 0.00 standard deviations from center line.

Test Failed at points: 9. One point more than 3.0597 n 150 LCL p = p − 3 p (1 − p ) 0.0230(1 − 0.0230) = 0.0367 ⇒ 0 n 150 MTB > Stat > Control Charts > Attributes Charts > P P Chart of Nonconforming Switches (Ex6-2Num) 1 0.0230(1 − 0.Chapter 6 Exercise Solutions 6-2.00 standard deviations from center line. m = 20. 17 6-3 . m m n = 150. ∑ Di = 69.00 2 4 6 8 10 12 Sample 14 16 18 20 Test Results for P Chart of Ex6-2Num TEST 1.0230 + 3 = 0.06 UCL=0.10 0.0597 0.023 0.08 Proportion 1 0. p = ∑ Di i =1 mn i =1 = 69 = 0.0230 − 3 = 0.0230 − 0.0230 20(150) UCL p = p + 3 p (1 − p ) 0.0230) = 0.02 LCL=0 0.04 _ P=0.

17 6-4 .0473 0.10 0.0163 0. One point more than 3. Test Failed at points: 1.00 standard deviations from center line.00 LCL=0 2 4 6 8 10 12 Sample 14 16 18 20 Test Results for P Chart of Ex6-2Num TEST 1. 9.06 1 UCL=0.02 _ P=0.Chapter 6 Exercise Solutions 6-2 continued Re-calculate control limits without samples 9 and 17: MTB > Stat > Control Charts > Attributes Charts > P P Chart of Nonconforming Switches (Ex6-2Num) Samples 9 and 17 excluded from calculations 1 0.04 0.08 Proportion 1 0.

02 _ P=0. 9.0289 ⇒ 0 150 MTB > Stat > Control Charts > Attributes Charts > P P Chart of Nonconforming Switches (Ex6-2Num) Samples 1.0141) = 0.0141 − 3 0.08 Proportion 1 0.0141 − 0.0141(1 − 0. 9.0141 0. m = 17. ∑ Di = 36.04 0.10 0. Test Failed at points: 1. One point more than 3. 17 excluded from calculations 1 0.00 standard deviations from center line.0430 0.0141 + 3 0. p = i =1 ∑ Di i =1 mn = 36 = 0. 17 6-5 .06 1 UCL=0.00 LCL=0 2 4 6 8 10 12 Sample 14 16 18 20 Test Results for P Chart of Ex6-2Num TEST 1.0141 17(150) UCL p = 0.Chapter 6 Exercise Solutions 6-2 continued Also remove sample 1 from control limits calculation: m m n = 150.0430 150 LCL p = 0.0141) = 0.0141(1 − 0.

12 0.1331 Proportion 0.06) 80 = 0.06(1 − 0.16 0.04 0. p − 3 p (1 − p ) ni } As an example.1397 LCL1 = p − 3 p (1 − p ) n1 = 0.00 LCL=0 1 2 3 4 5 6 Sample 7 8 9 10 Tests performed with unequal sample sizes The process appears to be in statistical control.06 0. NOTE: There is an error in the table in the textbook.06 + 3 0.06) 80 = 0. ∑ ni = 1000. The Fraction Nonconforming for Day 5 should be 0.0797 ⇒ 0 MTB > Stat > Control Charts > Attributes Charts > P P Chart of Nonconforming Units (Ex6-3Num) 0.06 − 3 0.10 0. ∑ Di = 60.Chapter 6 Exercise Solutions 6-3. for n = 80: UCL1 = p + 3 p (1 − p ) n1 = 0.046. p = ∑ Di ∑ ni = 60 1000 = 0.06(1 − 0.06 − 0.06 UCLi = p + 3 p (1 − p ) ni and LCLi = max{0. 6-6 .06 0.02 0.14 UCL=0.08 _ P=0. m m m m i =1 i =1 i =1 i =1 m = 10.

(a) m m i =1 i =1 n = 150.0167 − 0.0167 − 3 0.0167(1 − 0.0314 ⇒ 0 MTB > Stat > Control Charts > Attributes Charts > P P Chart of Nonconforming Units (Ex6-4Num) 0. m = 20.Chapter 6 Exercise Solutions 6-4.9 Select n = 530. (1 − p) 2 n> L p (1 − 0. ∑ Di = 50.01667 0. p = ∑ Di mn = 50 20(150) = 0.0167 UCL = p + 3 p (1 − p ) n = 0.0167) 150 = 0.0480 LCL = p − 3 p (1 − p ) n = 0.00 2 4 6 8 10 12 Sample 14 16 18 20 The process appears to be in statistical control. (b) Using Equation 6-12.02 _ P=0.0167) 2 (3) > 0.0167 > 529.05 UCL=0.01 LCL=0 0.03 0. 6-7 .04 0.0167 + 3 0.0167(1 − 0.0167) 150 = 0.04802 Proportion 0.

3. 11. 15. One point more than 3.1228) 2500 = 0.Chapter 6 Exercise Solutions 6-5.200 1 1 0.150 UCL=0.1228(1 − 0. 20 (b) So many subgroups are out of control (11 of 20) that the data should not be used to establish control limits for future production.1228(1 − 0. Test Failed at points: 1. (a) UCL = p + 3 p (1 − p ) n = 0. Instead. 19.125 _ P=0.1228 − 3 0.1228 0.075 1 1 0.1031 1 1 1 0. 12.1031 MTB > Stat > Control Charts > Attributes Charts > P P Chart of Nonconforming Belts (Ex6-5Num) 0.1228 + 3 0. the process should be investigated for causes of the wild swings in p. 2. 5.1228) 2500 = 0. 16.1425 LCL = p − 3 p (1 − p ) n = 0.1425 0.050 1 2 4 6 8 10 12 Sample 14 16 18 20 Test Results for P Chart of Ex6-5Num TEST 1. 17. 6-8 .175 1 1 1 Proportion 0.100 LCL=0.00 standard deviations from center line.

008) = 9.98 8 6 __ NP=4 4 2 LCL=0 0 1 2 3 4 5 6 Ex6-6Day 7 8 9 10 Test Results for NP Chart of Ex6-6Num TEST 1.976 ⇒ 0 MTB > Stat > Control Charts > Attributes Charts > NP NP Chart of Number of Nonconforming Units (Ex6-6Num) 1 12 Sample Count 10 UCL=9.008) = 4 − 5.00 standard deviations from center line. Test Failed at points: 6 6-9 . One point more than 3.Chapter 6 Exercise Solutions 6-6. UCL = np + 3 np (1 − p ) = 4 + 3 4(1 − 0.976 LCL = np − 3 np (1 − p ) = 4 − 3 4(1 − 0.

One point more than 3.11 2 0 LCL=0 1 2 3 4 5 6 Ex6-6Day 7 8 9 10 Test Results for NP Chart of Ex6-6Num TEST 1. Test Failed at points: 6 Recommend using control limits from second chart (calculated less sample 6). 6-10 .39 8 6 4 __ NP=3.Chapter 6 Exercise Solutions 6.6 continued Recalculate control limits without sample 6: NP Chart of Number of Nonconforming Units (Ex6-6Num) Day 6 excluded from control limits calculations 1 12 Sample Count 10 UCL=8.00 standard deviations from center line.

0044(1 − 0.0044 − 3 0. ( pˆ 6 = 0.0044 − 0. The data from the shipment do not indicate statistical control.02) 50 = 0.0044 + 3 0.02.1 and n = 50 is "large".278)3 = 0.857 + 0.0044) 250 = 0.02) 50 = 0. Pr{detected by 3rd sample} = 1 – Pr{detected after 3rd} = 1 – (1 – 0. n = 250. From the 6th sample. use the Poisson approximation to the binomial with λ = npnew = 50(0.0044 i =1 10 UCL = p + 3 p (1 − p ) n = 0.0794 LCL = p − 3 p (1 − p ) n = 0. the UCL.0170 10 m = 10.04 < 0.2) + POI(0. p = UCL = p − 3 p (1 − p ) n = 0.624 6-8.020) > 0.02(1 − 0.Chapter 6 Exercise Solutions 6-7.0594 ⇒ 0 Since pnew = 0. 0.02(1 − 0.135 = 0.278 where POI(⋅) is the cumulative Poisson distribution.0794) | 2} + Pr{D ≤ 50(0) | 2} = 1 – POI(3.02 + 3 0. 6-11 .0170.2) = 1 – 0.0440 = 0.0440.02 − 0. p = 0.02 − 3 0. n = 50 UCL = p + 3 p (1 − p ) n = 0.00.0126 ⇒ 0 No.04) = 2.0044(1 − 0. ∑ pˆ i = 0.0044) 250 = 0. Pr{detect|shift} = 1 – Pr{not detect|shift} =1–β = 1 – [Pr{D < nUCL | λ} – Pr{D ≤ nLCL | λ}] = 1 – Pr{D < 50(0.

10 + 3 0.212 Pr{D ≤ 13|p} 0.22 0.598077 0.1125 ⇒ 0 β = Pr{D < nUCL | p} − Pr{D ≤ nLCL | p} = Pr{D < 64(0.996172 0.215 0.503553 Pr{D ≤ 0|p} 0.598077 0.10) 64 = 0.480098 0.10) 64 = 0.000000 0.10 0.05 0.6) | p} − Pr{D ≤ 0 | p} p 0.44154 0.10 − 0.000000 0.10(1 − 0.10 − 3 0. (1 − p ) 2 n> L p (1 − 0.001179 0.20 0.480098 0.519279 0. p = 0.10) 2 (3) > 0.2125 LCL = p − 3 p (1 − p ) n = 0.503553 Assuming L = 3 sigma control limits.000000 β 0.000000 0.44154 0.000000 0.2125) | p} − Pr{D ≤ 64(0) | p} = Pr{D < 13.962475 0.Chapter 6 Exercise Solutions 6-9. n = 64 UCL = p + 3 p (1 − p ) n = 0.999999 0.21 0.10.519279 0.10 > 81 6-12 .037524 0.10(1 − 0.994993 0.

2) ⎟⎟ ⎝ ⎠ ⎝ ⎠ = 1 − Φ (1.030 Pr{detect by at least 3rd} = 1 – Pr{detected after 3rd} = 1 – (1 – 0.970 + 0.030)3 = 0.875) + Φ (−3.000 = 0. (1 − p ) 2 n> L p (1 − 0.16) = 5. Pr{detect shift on 1st sample} = 1 − β = 1 − [Pr{D < UCL | p} − Pr{D ≤ LCL | p}] ⎛ UCL + 1/ 2 − np ⎞ ⎛ LCL − 1/ 2 − np ⎞ = 1− Φ ⎜ +Φ⎜ ⎟ ⎟ ⎜ ⎜ np (1 − p) ⎟⎠ np (1 − p ) ⎟⎠ ⎝ ⎝ ⎛ 27 + 0.5 − 20 ⎞ = 1− Φ ⎜ + Φ⎜ ⎟ ⎜ 20(1 − 0. np = 16. p = 16 100 = 0. desire Pr{detect} = 0.10 δ =p new 2 2 ⎛k⎞ ⎛ 3 ⎞ n = ⎜ ⎟ p(1 − p) = ⎜ ⎟ (0. 6-11. p = 0.25 So.00 (a) npnew = 20.16 > 47. assume k = 3 sigma control limits new − p = 0.00 LCL = np − 3 np (1 − p ) = 16 − 3 16(1 − 0.875) = 1 − 0.50. n = 100. p = 0.10 ⎠ 6-13 . n = 48 is the minimum sample size for a positive LCL.10) = 81 ⎝δ ⎠ ⎝ 0.10.20.Chapter 6 Exercise Solutions 6-10. so use normal approximation to binomial distribution.16) = 27.10 = 0.16 UCL = np + 3 np(1 − p ) = 16 + 3 16(1 − 0.0 > 15.2) ⎟ ⎜ 20(1 − 0.0.5 − 20 ⎞ ⎛ 5 − 0.20 − 0.16) 2 (3) > 0.0873 (b) Assuming L = 3 sigma control limits.10)(1 − 0.

37) = 0.08) 100 ⎟ ⎜ 0. so use the normal approximation to the binomial. Pr{type I error} = α = Pr{D < LCL | p} + Pr{D > UCL | p} = Pr{D < LCL | p} + [1 – Pr{D ≤ UCL | p}] = Pr{D < 0 | 8} + [1 – Pr{D ≤ 16 | 8}] = 0 + [1 – POI(16. p = 0.1 and n =100 is large.08(1 − 0.161.996] = 0.14 LCL = np − 3 np (1 − p ) = 8 − 3 8(1 − 0.08(1 − 0.08) 100 ⎟⎟ ⎝ ⎠ ⎝ ⎠ = Φ (−1.20 = Φ⎜ −Φ⎜ ⎟ ⎜ 0.07494)4 = 0. n = 100. LCL = 0 (a) np = 100(0.Chapter 6 Exercise Solutions 6-12. UCL = 0. so use Poisson approximation to the binomial.8)] = 0 + [1 – 0.07494 − 0 = 0.1388 ⇒ 0 (b) p = 0.20 0 − 0.004 where POI(⋅) is the cumulative Poisson distribution. (c) npnew = 100(0.080) = 8 − 8.080) = 16.99997 6-14 .080 < 0.20) = 20 > 15.07494 (d) Pr{detect shift by at most 4th sample} = 1 – Pr{not detect by 4th} = 1 – (0. Pr{type II error} = β = Pr{ pˆ < UCL | pnew } − Pr{ pˆ ≤ LCL | pnew } ⎛ UCL − pnew ⎞ ⎛ LCL − pnew ⎞ = Φ⎜ −Φ⎜ ⎟ ⎜ p (1 − p ) n ⎟ ⎜ p (1 − p ) n ⎟⎟ ⎝ ⎠ ⎝ ⎠ ⎛ ⎞ ⎛ ⎞ 0.080) = 8 UCL = np + 3 np(1 − p ) = 8 + 3 8(1 − 0.161 − 0.08.44) − Φ (−7.

(a) p = 0.533) + Φ (−4.533) = 1 − 0.1(1 − 0.297) = 0.07 − 3 0.032 − 0.07(1 − 0.07(1 − 0.297)(0.07) 400 = 0.297 + (1 – 0.20) = 400 ⎝δ ⎠ ⎝ 0.20)(1 − 0.032 (b) npnew = 400(0.10) = > 40.20 and L = 3 sigma control limits (1 − p ) 2 n> L p (1 − 0.20 > 36 For Pr{detect} = 0.1) 400 ⎟⎟ ⎝ ⎠ ⎝ ⎠ = 1 − Φ (0.1(1 − 0. Pr{detect on 1st sample} = 1 − Pr{not detect on 1st sample} = 1− β = 1 − [Pr{ pˆ < UCL | p} − Pr{ pˆ ≤ LCL | p}] ⎛ UCL − p ⎞ ⎛ LCL − p ⎞ = 1− Φ ⎜ +Φ⎜ ⎟ ⎜ p (1 − p) n ⎟ ⎜ p(1 − p ) n ⎟⎟ ⎝ ⎠ ⎝ ⎠ ⎛ ⎛ 0.07) 400 = 0. δ =pnew − p = 0.000 = 0.506 6-14.20 = 0.1 ⎞ 0.26. k = 3 sigma control limits.26 − 0.1 ⎞ = 1− Φ ⎜ + Φ⎜ ⎟ ⎜ 0. n = 400 UCL = p + 3 p(1 − p) n = 0.Chapter 6 Exercise Solutions 6-13.07. p = 0.108 − 0. so use the normal approximation to the binomial.20) 2 (3) > 0.06 2 2 ⎛k⎞ ⎛ 3 ⎞ n = ⎜ ⎟ p(1 − p) = ⎜ ⎟ (0.06 ⎠ 6-15 .07 + 3 0.50 after a shift to pnew = 0.108 LCL = p − 3 p(1 − p ) n = 0.297 (c) Pr{detect on 1st or 2nd sample} = Pr{detect on 1st} + Pr{not on 1st}×Pr{detect on 2nd} = 0.703 + 0.1) 400 ⎟ ⎜ 0.

51 Sample Count 25 20 __ NP=16. n = 100.4(1 − 0.4 + 3 16.Chapter 6 Exercise Solutions 6-15.164) = 27. Test Failed at points: 3 6-16 .164.4 UCL = np + 3 np (1 − p ) = 16.4(1 − 0.51 LCL = np − 3 np (1 − p ) = 16. i =1 10 p = ∑ Di i =1 ( mn ) = 164 [10(100)] = 0. 10 ∑ Di = 164.164) = 5.4 − 3 16. One point more than 3.4 15 10 LCL=5. np = 16.00 standard deviations from center line.29 5 1 2 3 4 5 6 Sample 7 8 9 10 Test Results for NP Chart of Ex6-15Num TEST 1. (a) m = 10.292 MTB > Stat > Control Charts > Attributes Charts > NP NP Chart of Number Nonconforming (Ex6-15Num) 1 30 UCL=27.

13 1 2 3 4 5 6 Sample 7 8 9 10 Test Results for NP Chart of Ex6-15Num TEST 1.78 15 10 5 LCL=4. Test Failed at points: 3 6-17 . One point more than 3.42 Sample Count 25 20 __ NP=14.00 standard deviations from center line.Chapter 6 Exercise Solutions 6-15 continued Recalculate control limits less sample 3: NP Chart of Number Nonconforming (Ex6-15Num) Sample 3 excluded from calculations 1 30 UCL=25.

the normal approximation to the binomial can be used.03) 200 = 0.0662 LCL p = p − 3 p (1 − p ) n = 0.03 + 3 0.30 is not too far from 0. Pr{detect by at least 4th} = 1 – Pr{detect after 4th} = 1 – (1 – 0.10 and n is large.16) + POI(0. Pr{detect on 1st sample | p} = 1 − Pr{not detect | p} = 1− β = 1 − [Pr{ pˆ < UCL | p} − Pr{ pˆ ≤ LCL | p}] = 1 − Pr{D < nUCL | np} + Pr{D ≤ nLCL | np} = 1 − Pr{D < 200(0.187) + (0.30.3) ⎟⎟ 30(1 − 0. Pr{detect on 1st} = 1 − Pr{not detect on 1st} = 1− β = 1 − [Pr{D < UCL | p} − Pr{D ≤ LCL | p}] ⎛ UCL + 1/ 2 − np ⎞ ⎛ LCL − 1/ 2 − np ⎞ = 1− Φ ⎜ + Φ⎜ ⎟ ⎟ ⎜ ⎜ np (1 − p ) ⎟⎠ np (1 − p ) ⎟⎠ ⎝ ⎝ ⎛ 25. Since (pnew = 0.13 − 0.03) 200 = 0.000) = 0.16) = 1 − 0.42 + 0.03(1 − 0. use the Poisson approximation to the binomial.8903) + Φ (−5. Since p = 0.813 6-16.08)} + Pr{D ≤ 200(0) | 200(0.50.9943 6-18 .03(1 − 0.2745 + 0. and n = 100 > 10.7255 where POI(⋅) is the cumulative Poisson distribution.5 − 30 ⎞ ⎛ 4.03 − 3 0.03 − 0.Chapter 6 Exercise Solutions 6-15 continued (b) pnew = 0.0662) | 200(0.08) < 0.0362 ⇒ 0 (b) pnew = 0.3) ⎟⎠ ⎝ ⎝ ⎠ = 1 − Φ (−0.7255)4 = 0.7544) = 1 − (0.5 − 30 ⎞ = 1− Φ ⎜ +Φ⎜ ⎟ ⎜ ⎜ 30(1 − 0.08)} = 1 − POI(13.000 = 0. (a) UCL p = p + 3 p (1 − p ) n = 0.08.

000 = 0.5 − 60 ⎞ = 1− Φ ⎜ + Φ⎜ ⎟ ⎜ 60(1 − 0. np = 400(0.Chapter 6 Exercise Solutions 6-17.10.417 + 0.39) = 1 − 0.10) = 40 = np + 3 np (1 − p ) = 40 + 3 40(1 − 0.15) = 60 > 15.15) ⎟⎟ ⎝ ⎠ ⎝ ⎠ = 1 − Φ (−0.583 6-19 . so use the normal approximation to the binomial.5 − 60 ⎞ ⎛ 22 − 0.210) + Φ (−5. (a) m p = ∑ Di i =1 UCL LCL np np ( mn ) = 1200 [30(400)] = 0.10) = 22 (b) npnew = 400 (0. Pr{detect on 1st sample | p} = 1 − Pr{not detect on 1st sample | p} = 1− β = 1 − [Pr{D < UCL | np} − Pr{D ≤ LCL | np}] ⎛ UCL + 1/ 2 − np ⎞ ⎛ LCL − 1/ 2 − np ⎞ = 1− Φ ⎜ + Φ⎜ ⎟ ⎟ ⎜ ⎜ np (1 − p ) ⎟⎠ np (1 − p ) ⎟⎠ ⎝ ⎝ ⎛ 58 + 0.10) = 58 = np − 3 np (1 − p ) = 40 − 3 40(1 − 0.15) ⎟ ⎜ 60(1 − 0.

λ = npnew = 100(0. not 6-18.996 = 0.20) = 20. This is a continuation of Exercise 6-17.10) = 10. 1 – β = 0.1 ⎠ ⎝ UCL − p ⎠ 2 (b) Using the Poisson approximation to the binomial. 20) − POI(1.004 where POI(⋅) is the cumulative Poisson distribution.381 where POI(⋅) is the cumulative Poisson distribution.1) ⎜ ⎟ = 100 ⎝ 0.000 + 1 − 0. (a) UCL = p + 3 p (1 − p ) n 2 ⎛ ⎞ 3 3 ⎛ ⎞ n = p (1 − p ) ⎜ ⎟ = 0.381 ARL1 = 1/(1 –β) = 1/(1 – 0. from 6-18(c).1(1 − 0.19 − 0.583 ARL1 = 1/(1 –β) = 1/(0.381) = 1. Pr{type I error} = Pr{ pˆ < LCL | p} + Pr{ pˆ > UCL | p} = Pr{D < nLCL | λ} + 1 − Pr{D ≤ nUCL | λ} = Pr{D < 100(0. NOTE: There is an error in the textbook. Using the Poisson approximation to the binomial.715 ≅ 2 6-20. β = 0. (c) pnew = 0.19) |10} = POI(0.Chapter 6 Exercise Solutions 6-18.616 ≅ 2 6-20 .381 − 0. Pr{type II error} = β = Pr{D < nUCL | λ} − Pr{D ≤ nLCL | λ} = Pr{D < 100(0.01) |10} + 1 − Pr{D ≤ 100(0. 6-19.583) = 1. λ = np = 100(0.19) | 20} − Pr{D ≤ 100(0.20.01) | 20} = POI(18.10) = 0.000 = 0. from 6-17(b).10) + 1 − POI(19. 20) = 0.

6-21 . UCLi] [0.0221) 0.0221 and control limits are at p ± 3 p (1 − p ) / ni [LCLi.01 LCL=0 0.0221(1 − 0. preferred in many cases.02213 0. 0.0221) ni . 0.0216 / ni .02 0.00 2 4 6 8 10 12 Sample 14 16 18 20 Tests performed with unequal sample sizes Process is in statistical control. In those cases. 0. 0.Chapter 6 Exercise Solutions 6-21. and to plot Z i = ( pˆ i − 0.0662] [0.07 0. would be to construct standardized control limits with control limits at ± 3.04 0. The second approach. (a) For a p chart with variable sample size: p = ∑ i Di ∑ i ni = 83 / 3750 = 0.0581] [0.05005 0.05 UCL=0.0221 ± 3 0.06 Proportion 0. The first approach would be to plot pˆ i and use constant limits unless there is a different size sample or a plot point near a control limit. (b) There are two approaches for controlling future production.0533] [0. calculate the exact control limits by p ± 3 p (1 − p ) / ni = 0.0500] ni 100 150 200 250 MTB > Stat > Control Charts > Attributes Charts > P P Chart of Second Visit Required (Ex6-21Sec) 0.03 _ P=0.

00 LCL=0 2 4 6 8 10 12 Sample 14 16 18 20 Process is in statistical control.06 UCL=0. however MINITAB accepts only integer values for n.04 0. MTB > Stat > Basic Statistics > Display Descriptive Statistics Descriptive Statistics: Ex6-21Req Variable Ex6-21Req N 20 Mean 187.02 0. and carefully examine points near the control limits. 6-22 .5 Average sample size is 187.05 Proportion 0.03 _ P=0.01 0. Use a sample size of n = 187.02219 0.05451 0. MTB > Stat > Control Charts > Attributes Charts > P P Chart of Second Visit Required (Ex6-21Sec) Limits based on average sample size (n=187) 0.5.Chapter 6 Exercise Solutions 6-22.

0216 / ni MTB > Stat > Control Charts > Variables Charts for Individuals > Individuals I Chart of Standardized Second Visit Data (Ex6-23zi) UCL=3 3 Individual Value 2 1 _ X=0 0 -1 -2 -3 LCL=-3 2 4 6 8 10 12 Observation 14 16 18 20 Process is in statistical control.0221) 0. zi = ( pˆ i − p ) p (1 − p ) ni = ( pˆ i − 0. 6-23 .Chapter 6 Exercise Solutions 6-23.

UCL250 = 0.06 0.0533.02 0. LCL = 0 UCL100 = 0.0581.03 0. UCL200 = 0.05 0.0221. UCL150 = 0.04 0.07 Variable Ex6-24pi Ex6-24n100 Ex6-24n150 Ex6-24n200 Ex6-24n250 Ex6-24C L Ex6-24LCL 0.0662. CL = 0.Chapter 6 Exercise Solutions 6-24.00 2 4 6 8 10 12 Week 14 16 18 20 6-24 .0500 MTB > Graph > Time Series Plot > Multiple Control Chart of Second Visit Data with Limits for Various Sample Sizes (Ex6-24pi) Proportion of Second Visits Required 0.01 0.

n = 100 ⎛ 1− p ⎞ 2 n>⎜ ⎟L ⎝ p ⎠ ⎛ 1 − 0. p = CL = 0. LCL = 0.05) = 7. p = CL = 0. 6-27.0191 (a) 0.0399.64 (c) n = 400 is large and p = 0.8349 np (1 − p ) = 20 − 2.05) 400 = 0.03 on 1st sample} = 1 − Pr{not detect} = 1− β = 1 − [Pr{D < UCL | λ} − Pr{D ≤ LCL | λ}] = 1 − Pr{D < 32. LCL = 0.0809 = 0.0109) L = 2.8349 20(1 − 0.0000 + 0.1.05 + L(0. use Poisson approximation to binomial.05) = 32.8349 np(1 − p) = 20 + 2. Pr{detect shift to 0.05) = 20 UCL = np + 2.64 |12} = 1 − POI(32.36 |12} + Pr{D ≤ 7. The np chart is inappropriate for varying sample sizes because the centerline (process center) would change with each ni.05 + L 0.05(1 − 0. n = 400. 6-25 .0500.0809.01.12) = 1 − 1.05 < 0.12) + POI(7.01 ⎠ > 891 ≥ 892 6-26.Chapter 6 Exercise Solutions 6-25.36 LCL = np − 2.0895 where POI(·) is the cumulative Poisson distribution.01 ⎞ 2 >⎜ ⎟3 ⎝ 0.8349 (b) CL = np = 400(0.8349 20(1 − 0.0895 = 0. UCL = 0. UCL = 0.

Chapter 6 Exercise Solutions 6-28.03 ⎠ 6-26 .00) + Φ (−8. p = 0.05(1 − 0.15) 400 ⎟ ⎝ ⎠ ⎝ ⎠ = 1 − Φ (−3.15) 400 ⎟ ⎜ 0.01. Pr{detect on 1st sample after shift} = 1 − Pr{not detect} = 1− β = 1 − [Pr{ pˆ < UCL | p} − Pr{ pˆ ≤ LCL | p}] ⎛ UCL − p ⎞ ⎛ LCL − p ⎞ = 1− Φ ⎜ +Φ⎜ ⎟ ⎜ p (1 − p ) n ⎟ ⎜ p (1 − p ) n ⎟⎟ ⎝ ⎠ ⎝ ⎠ ⎞ ⎛ ⎞ ⎛ 0.15 0.01)(1 − 0. L = 2 (a) ⎛ 1− p ⎞ 2 n>⎜ ⎟L ⎝ p ⎠ ⎛ 1 − 0.00135 + 0.01 ⎞ 2 >⎜ ⎟2 ⎝ 0.24 (b) p = 15.19) = 1 − 0.0500 + L 0. (a) UCL = p + L p (1 − p ) n 0.01 = 0.05) 400 L = 4.01 ⎠ > 396 ≥ 397 (b) δ = 0. λ = np = 400(0.15(1 − 0.0038 − 0.0962 = 0.000 = 0. use normal approximation to binomial.99865 6-29.15(1 − 0.03 2 2 ⎛L⎞ ⎛ 2 ⎞ n = ⎜ ⎟ p(1 − p) = ⎜ ⎟ (0.15) = 60 > 15.04 – 0.0962 − 0.01) = 44 ⎝δ ⎠ ⎝ 0.15 = 1− Φ ⎜ +Φ⎜ ⎟ ⎟ ⎜ 0.

(a) Pr{type I error} = Pr{ pˆ < LCL | p} + Pr{ pˆ > UCL | p} = Pr{D < nLCL | np} + 1 − Pr{D ≤ nUCL | np} = Pr{D < 100(0. (b) Pr{type II error} =β = Pr{D < nUCL | np} − Pr{D ≤ nLCL | np} = Pr{D < 100(0.Chapter 6 Exercise Solutions 6-30.075) |100(0.742 where POI(⋅) is the cumulative Poisson distribution.002 = 0.04)} + 1 − Pr{D ≤ 100(0.018 + 1 − 0.04)} = POI(0.0050) |100(0. 6-27 .070 where POI(⋅) is the cumulative Poisson distribution. 4) + 1 − POI(7.005) |100(0.948 = 0. 6) − POI(0.744 − 0.075) |100(0.06)} − Pr{D ≤ 100(0. 4) = 0. 6) = 0.06) = POI(7.

070: ARL0 = 1/α = 1/0.0008 0.9881 0.Chapter 6 Exercise Solutions 6-30 continued (c) β = Pr{D < nUCL | np} − Pr{D ≤ nLCL | np} = Pr{D < 100(0.3 p (d) from part (a).5|np} 1.0067 0.4530 0.0000 0.5|np} 1.0180 0.25 np 0 0.0000 0.3239 0.15 0.3935 0.8666 0. α = 0.04 0.15 0.9306 0.03 0.0000 0 0.5 |100 p} Excel : workbook Chap06.8636 0.07 0.7440 0.5.0000 0.3238 0.29 ≅ 15 from part (b).5 1 2 3 4 5 6 7 8 9 10 12.02 0.0001 0.0000 1.5000 0.070 = 14.0000 0.0000 0.0000 0.2 0. UCL=7.0698 0.5 1.0742: ARL1 = 1/(1 –β) = 1/(1 – 0. LCL=0.4526 0.1 0.5987 0.6000 0.125 0.9489 0.0698 0.861 ≅ 4 6-28 .06 0.0180 0.1 0. CL=4.6065 0.8599 0.2 0.2000 0.01 0.7000 Beta 0.05 0.0009 0.0000 0.0008 0.0025 0.9000 0.0498 0.3679 0.6321 0.25 0. β = 0.005 0.1000 0.5 15 20 25 Pr{D<7.4000 0.05 0.0000 beta 0.0000 1.742) = 3.9383 0.0000 OC Curve for n=100.9989 0.0750) |100 p} − Pr{D ≤ 100(0.08 0.0000 Pr{D<=0.xls : worksheet Ex6-30 p 0 0.0003 0.1353 0.3000 0.5 |100 p} − Pr{D ≤ 0.0050) |100 p} = Pr{D < 7.09 0.7415 0.8000 0.5978 0.0000 0.2202 0.2202 0.0183 0.

03 _ P=0.02(1 − 0. p = 0.05 0.062 LCL = p − 3 p (1 − p ) n = 0.07 UCL=0. p = 0.02 0.038 and σˆ p = 0.02) 100 = 0.01 0.06 0. LCL = np − k np (1 − p ) > 0 np > k np (1 − p ) ⎛ 1− p ⎞ n > k2 ⎜ ⎟ ⎝ p ⎠ 6-29 .02 − 3 0.02 (a) UCL = p + 3 p (1 − p ) n = 0.08 0.0191 6-32.04 0.02(1 − 0.02 0.00 standard deviations from center line.02 + 3 0. Test Failed at points: 4 Sample 4 exceeds the upper control limit.02) 100 ⇒ 0 (b) MTB > Stat > Control Charts > Attributes Charts > P P Chart of Number Nonconforming (Ex6-31Num) 0.062 Proportion 0. n = 100.09 1 0.00 LCL=0 1 2 3 4 5 6 Sample 7 8 9 10 Test Results for P Chart of Ex6-31Num TEST 1.Chapter 6 Exercise Solutions 6-31. One point more than 3.

505(1 − 0. ∑ D = 50. p = 0.204 7 Sample Count 6 5 4 3 __ NP=2.213 LCL = np − 3 np (1 − p ) = 2.Chapter 6 Exercise Solutions 6-33.5 2 1 LCL=0 0 2 4 6 8 10 12 Sample 14 16 18 20 The process is in control.0167) = 2.0167) = 7. m = 20. 6-30 .505 UCL = np + 3 np (1 − p ) = 2. n = 150.505 + 3 2.0167 CL = np = 150(0. results are the same as for the p chart.505 − 4.708 ⇒ 0 MTB > Stat > Control Charts > Attributes Charts > NP NP Chart of Numer of Nonconforming Units (Ex6-4Num) 8 UCL=7.

23 LCL = np − 3 np (1 − p ) = 307 − 3 307(1 − 0.77 MTB > Stat > Control Charts > Attributes Charts > NP NP Chart of Number of Nonconforming Belts (Ex6-5Num) 500 1 1 1 400 Sample Count 1 1 UCL=356. 3.3 __ NP=307. 11. 16.1228) = 356.Chapter 6 Exercise Solutions 6-34.8 200 1 1 1 1 1 1 100 2 4 6 8 10 12 Sample 14 16 18 20 Test Results for NP Chart of Ex6-5Num TEST 1. 20 Like the p control chart. 2. 15. One point more than 3. 12. many subgroups are out of control (11 of 20).1228) = 307 UCL = np + 3 np (1 − p ) = 307 + 3 307(1 − 0.00 standard deviations from center line. 19. 6-31 . CL = np = 2500(0. indicating that this data should not be used to establish control limits for future production. 5.1 300 LCL=257. 17.1228) = 257. Test Failed at points: 1.

6-32 .414 -3 1 2 3 4 5 6 Observation 7 8 9 10 The process is in control.0564 / ni MTB > Stat > Control Charts > Variables Charts for Individuals > Individuals I Chart of Standardized Fraction Nonconforming (Ex6-35zi) 3 UCL=2.06 zi = ( pˆ i − 0.06(1 − 0.06) / ni = ( pˆ i − 0.06) 0.040 0 -1 -2 LCL=-2. p = 0.Chapter 6 Exercise Solutions 6-35.06) 0. results are the same as for the p chart.494 Individual Value 2 1 _ X=0.

Test Failed at points: 13 No.969 6 5 4 3 _ C=2.36 = 6. The plate process does not seem to be in statistical control.36 UCL = c + 3 c = 2.Chapter 6 Exercise Solutions 6-36. CL = c = 2.00 standard deviations from center line. 6-33 .97 LCL = c − 3 c = 2.36 2 1 0 LCL=0 2 4 6 8 10 12 14 Sample 16 18 20 22 24 Test Results for C Chart of Ex6-36Num TEST 1.36 + 3 2. One point more than 3.36 ⇒ 0 MTB > Stat > Control Charts > Attributes Charts > C C Chart of Number of Nonconformities on Plate (Ex6-36Num) 9 1 8 Sample Count 7 UCL=6.36 − 3 2.

1. 1.8 _ U=0.7007 / ni ni 18 20 21 22 24 [LCLi.2487] [0.1527.2 1.Chapter 6 Exercise Solutions 6-37.0 0. 1.7007 − 3 0.249 Sample Count Per Unit 1.2622] [0.7007 / ni LCLi = u − 3 u ni = 0. 1.6 0.4 UCL=1.2 LCL=0.701 0.2926] [0.7007 + 3 0. 1.2133] MTB > Stat > Control Charts > Attributes Charts > U U Chart of Imperfections in Paper Rolls (Ex6-37Imp) 1.1392.2361] [0. CL = u = 0.0 2 4 6 8 10 12 Sample 14 16 18 20 Tests performed with unequal sample sizes 6-34 .7007 UCLi = u + 3 u ni = 0. UCLi] [0.1088.4 0.1653.1881.153 0.

7007. however MINITAB accepts only integer values for n.0 2 4 6 8 10 12 Sample 14 16 18 20 6-35 .550 Average sample size is 20.55 UCL = u + 3 u n = 0.289 Sample Count Per Unit 1.4 UCL=1.2 LCL=0. and carefully examine points near the control limits.0 0.4 0.7007 / 20. Use a sample size of n = 20.151 0.55 = 1. CL = u = 0.72 0.6 0. n = 20.55 = 0.55.Chapter 6 Exercise Solutions 6-38.8 _ U=0.2 1.1467 MTB > Stat > Basic Statistics > Display Descriptive Statistics Descriptive Statistics: Ex6-37Rol Variable Ex6-37Rol N 20 Mean 20. MTB > Stat > Control Charts > Attributes Charts > U U Chart of Imperfections in Paper Rolls (Ex6-37Imp) with average sample size n=20 1.2547 LCL = u − 3 u n = 0.7007 + 3 0.7007 − 3 0.7007 / 20.

7007) 0.898 Individual Value 1 _ X=-0.Chapter 6 Exercise Solutions 6-39.7007 / ni MTB > Stat > Control Charts > Variables Charts for Individuals > Individuals I Chart of Standardized Paper Roll Imperfections (Ex6-39zi) 2 UCL=1. zi = (ui − u ) u ni = (ui − 0.906 -2 2 4 6 8 10 12 Observation 14 16 18 20 6-36 .004 0 -1 LCL=-1.

174 5 Sample Count 4 3 2 _ C=1. Use these limits to control future production. 6-37 .5 + 3 1.5 = 5.17 LCL ⇒ 0 MTB > Stat > Control Charts > Attributes Charts > C C Chart of Cassette Deck Nonconformities (Ex6-40Num) UCL=5. c chart based on # of nonconformities per cassette deck CL = c = 1.5 1 0 LCL=0 2 4 6 8 10 Sample 12 14 16 18 Process is in statistical control.5 UCL = c + 3 c = 1.Chapter 6 Exercise Solutions 6-40.

59 5 0 LCL=0 2 4 6 8 10 12 Sample 14 16 18 20 22 Test Results for C Chart of Ex6-41Num TEST 1. LCL = c − 3 c = 8.59 = 17. 22 6-38 .59.38 15 10 _ C=8. One point more than 3. Test Failed at points: 10.59 ⇒ 0 MTB > Stat > Control Charts > Attributes Charts > C C Chart of Number of Nonconformities (Ex6-41Num) per 1000 meters telephone cable 25 1 20 1 1 Sample Count UCL=17. CL = c = 8.Chapter 6 Exercise Solutions 6-41. UCL = c + 3 c = 8. 11.59 + 3 8.384.59 − 3 8.00 standard deviations from center line.

17 5 LCL=0 0 2 4 6 8 10 12 Sample 14 16 18 20 22 Test Results for C Chart of Ex6-41Num TEST 1. three subgroups exceed the UCL. 15. 11. LCL ⇒ 0 C Chart of Number of Nonconformities (Ex6-41Num) Samples 10.17 = 13. 11. CL = c = 6. then re-calculate the control limits. 22 excluded from calculations 25 1 Sample Count 20 1 1 1 15 UCL=13. 22 6-39 . UCL = c + 3 c = 6. One point more than 3.00 standard deviations from center line. 11 and 22.Chapter 6 Exercise Solutions 6-41 continued Process is not in statistical control. Subgroup 15 will then be out of control and should also be excluded. 15.17 + 3 6. Test Failed at points: 10.62.17.62 10 _ C=6. Exclude subgroups 10.

43 = 3.17) = 15.Chapter 6 Exercise Solutions 6-42.42 /1 = 27. and since n = 1.20 LCL = u − 3 u n = 15.00 CL = u = total inspection units (18 / 4) UCL = u + 3 u n = 6 + 3 6 1 = 13. A u chart of average nonconformities per inspection unit is appropriate. is the average number of nonconformities found in 2500m.42 + 3 15.5) = 6 UCL = nc + 3 nc = 6 + 3 6 = 13. A c chart of the total number of nonconformities per inspection unit is appropriate. (a) The new inspection unit is n = 4 cassette decks. (a) The new inspection unit is n = 2500/1000 = 2. total nonconformities 27 = = 6. is the total number of nonconformities found while inspecting a sample 2500m in length.65 The plot point. CL = nc = 4(1. A c chart of the total number of nonconformities per inspection unit is appropriate. u . c . this is the same as the total number of nonconformities.43 UCL = nc + 3 nc = 15.64 The plot point. CL = nc = 2. A u chart of average nonconformities per inspection unit is appropriate. total nonconformities 111 = = 15.35 LCL = u − 3 u n = 6 − 3 6 1 ⇒ 0 6-43.43 + 3 15. (b) The sample is n =1 new inspection units.35 LCL = nc − 3 nc = 6 − 3 6 ⇒ 0 (b) The sample is n =1 new inspection units.5 of the old unit.42 CL = u = total inspection units (18 × 1000) / 2500 UCL = u + 3 u n = 15.42 − 3 15.43 = 27.43 − 3 15.42 /1 = 3.21 LCL = nc − 3 nc = 15. 6-40 .5(6.

396 Sample Count Per Unit 1.422 4 = −0. However. (If this was a c chart for total nonconformities in the sample.Chapter 6 Exercise Solutions 6-44. (c) The new sample is n = 8/4 = 2 inspection units.8 0.396 LCL = u − 3 u n = 0.422 + 3 0.422 UCL = u + 3 u n = 0.2 0.4 UCL=1.4 0. the same control limits may be used for future production with the new sample size.0 0.) 6-41 .75 16 = 0.0 LCL=0 2 4 6 8 10 Sample 12 14 16 (b) The process is in statistical control. since this chart was established for average nonconformities per unit. (a) A u chart of average number of nonconformities per unit is appropriate.422 − 3 0.2 1.422 0.422 4 = 1. the control limits would need revision. with n = 4 transmissions in each inspection.6 _ U=0. CL = u = ∑ ui m = ( ∑ xi / n ) m = (27 / 4) 16 = 6.211 ⇒ 0 MTB > Stat > Control Charts > Attributes Charts > U U Chart of Manual Transmission Subassemblies (Ex6-44Num) 1.

(a) CL = c = 9 UCL = c + 3 c = 9 + 3 9 = 18 LCL = c − 3 c = 9 − 3 9 = 0 (b) c = 16. UCL = 21 Pr{x ≤ 10 | c = 16} = 0. Use the cumulative Poisson tables. n = 4 CL = u = c / n = 4 / 4 = 1 UCL = u + 3 u n = 1 + 3 1/ 4 = 2. n = 4 CL = u = c / n = 16 / 4 = 4 UCL = u + 3 u n = 4 + 3 4 / 4 = 7 LCL = u − 3 u n = 4 − 3 4 / 4 = 1 6-42 .0774. (a) CL = c = 4 UCL = c + 3 c = 4 + 3 4 = 10 LCL = c − 3 c = 4 − 3 4 ⇒ 0 (b) c = 4.9108.Chapter 6 Exercise Solutions 6-45. c = 16 Pr{x ≤ 21| c = 16} = 0.5 LCL = u − 3 u n = 1 − 3 1/ 4 ⇒ 0 6-46. LCL = 10 6-47.

054 6 3 = 3.4 10 = 0. u chart with u = 6.976 for the c chart.019 0. As a comparison.020 u n = 6 − 2.96 7.20 6-50.96 7.4 + 1.784 6-43 . the normal distribution gives UCL = c + z0.020.00.00 LCL = c − z0. the normal distribution gives: UCL = u + z0. As a comparison.952 UCL = x/n = 20/10 = 2.70.016 LCL = u + z0.062 0. c = u × n = 18.95 u n = 1.923 0. From the cumulative Poisson tables: x 9 10 26 27 Pr{D ≤ x | c = 18} 0.975 c = 7. Find limits such that Pr{D ≤ UCL} = 0.983 UCL = x/n = 27/3 = 9.4(10) = 14: x 7 8 19 20 Pr{D ≤ x | c = 14} 0.030 0. UCL = 13 and LCL = 2.645 1.032 0.055 0.4 − 1.645 1.954 0. and LCL = x/n = 9/3 = 3.6 = 2.980 u n = 6 + 2.0 and n = 3.6} 0.6 − 1.054 6 3 = 8.025 c = 7. As a comparison. and LCL = x/n = 7/10 = 0.905 LCL = u + z0.05 u n = 1.6 = 13.4 10 = 2.972 0. the normal distribution gives: UCL = u + z0. Using the cumulative Poisson distribution with c = u n = 1.Chapter 6 Exercise Solutions 6-48.095 6-49. Using the cumulative Poisson distribution: x 2 3 12 13 Pr{D ≤ x | c = 7.015 0.6 + 1.980 and Pr{D < LCL} = 0.

0. So set UCL = 6.0} = 0. Pr{x ≤ 6 | c = 2.995. LCLi = 7 − 3 7 / ni MTB > Stat > Control Charts > Attributes Charts > U U Chart of Total Number of Imperfections (Ex6-51Imp) 16 UCL=14.Chapter 6 Exercise Solutions 6-51. 6-52.005) = 0. (b) Pr{two consecutive out-of-control points} = (0. UCLi = 7 + 3 7 / ni .005)(0.00003 6-44 . (a) From the cumulative Poisson table.94 Sample Count Per Unit 14 12 10 8 _ U=7 6 4 2 0 LCL=0 1 2 3 4 5 6 Ex6-51Day 7 8 9 10 Tests performed with unequal sample sizes The process is in statistical control. u chart with control limits based on each sample size: u = 7.

5}] = POI(0. From the cumulative Poisson distribution: x Pr{D ≤ x | c = 8. c = 850 /100 = 8.88 8.78 |10. CL = c = ∑ D m = 924 /176 = 5.000 + [1 − 0.5 UCL = nc + 3 nc = 10.22 LCL = nc − 3 nc = 10.5 − 1.030 13 0.88 8.5 − 3 10.5 = 0.03 c = 8.5 .5 + 3 10.10.97 c = 8.25 + 3 5.973 LCL = 3 and UCL = 13.949 14 0. the normal distribution gives UCL = c + z0.5} + [1 − Pr{D ≤ 20. A c chart with one inspection unit equal to 50 manufacturing units is appropriate.5 + 1.98 LCL = c + z0.78 (c) Pr{type I error} = Pr{D < LCL | c} + Pr{D > UCL | c} = Pr{D < 0.003 6-45 .5 = 13.5 = 3.10.5 = 20. (b) Let new inspection unit n = 2 water heaters CL = nc = 2(5.5} 3 0.22 |10.Chapter 6 Exercise Solutions 6-53.5) ] = 0. (a) Plot the number of nonconformities per water heater on a c chart.02 6-54.25 = 12.25) = 10.12 LCL ⇒ 0 Plot the results after inspection of each water heater.5) + [1 − POI(20. For comparison.25 UCL = c + 3 c = 5. approximately 8/day.997 ] = 0.

72 | 0.0 average number of nonconformities/unit.98 0. 6-46 . when in fact it is.Chapter 6 Exercise Solutions 6-55.00 An UCL = 9 will give a probability of 0.533}] = 0 + [1 − POI(2. ∑ Di = 16 in 30 refrigerators.983 = 0.mtw Ex6-55X 0 1 2 3 4 5 6 7 8 9 10 11 Ex6-55alpha 0.99.89 0.017 where POI(⋅) is the cumulative Poisson distribution.99 of concluding the process is in control.43 0.95 0. 6-56.533 ± 3 0. 0. Use a c chart for nonconformities with an inspection unit n = 1 refrigerator.24 0.533} + [1 − Pr{D ≤ 2. u = 4.533 = [0. Desire α = 0.723] (b) α = Pr{D < LCL | c} + Pr{D > UCL | c} = Pr{D < 0 | 0.79 0.00 1.02 0. Use the cumulative Poisson distribution to determine the UCL: MTB : worksheet Chap06. 2.63 0.533 (a) 3-sigma limits are c ± 3 c = 0. c = 16 / 30 = 0.533) ] = 1 − 0.99 1.09 0.

533 = [0. (d) ARL1 = 1 1 = = 1. (d) ARL1 = 1 1 = = 2.1.993] (b) α = Pr{D < LCL | c } + Pr{D > UCL | c } = Pr{D < 0 | 0.533 (a) c ± 2 c = 0.271 6-47 . c = 0.533 + 2 0.1004 where POI(⋅) is the cumulative Poisson distribution. 2) − POI(0.5414 where POI(⋅) is the cumulative Poisson distribution.8996 = 0.72 | 2.993 | 0.135 = 0.533} + [1 − Pr{D ≤ 1. 2) = 0.372 ≈ 2 1 − β 1 − 0. 0.541 6-57. 2) − POI(0.533}] = 0 + [1 − POI(1.0} = POI(2.533)] = 1 − 0.1353 = 0.993 | 2} − Pr{D ≤ 0 | 2} = POI(1.406 − 0. 2) = 0.0} − Pr{D ≤ 0 | 2. (c) β = Pr{D < UCL | c} − Pr{D ≤ LCL | c} = Pr{D < 1.271 where POI(⋅) is the cumulative Poisson distribution.Chapter 6 Exercise Solutions 6-56 continued (c) β = Pr{not detecting shift} = Pr{D < UCL | c} − Pr{D ≤ LCL | c} = Pr{D < 2.6767 − 0.18 ≈ 2 1 − β 1 − 0.

1 inspection unit = 6 clocks.Chapter 6 Exercise Solutions 6-58. 1 inspection unit = 10 radios.75(6) = 4.5 UCL = c + 3 c = 4.5 + 3 4. u = average # nonconformities/calculator = 2 (a) c chart with c = u × n = 2(2) = 4 nonconformities/inspection unit CL = c = 4 UCL = c + k c = 4 + 3 4 = 10 LCL = c − k c = 4 − 3 4 ⇒ 0 (b) Type I error = α = Pr{D < LCL | c } + Pr{D > UCL | c } = Pr{D < 0 | 4} + [1 − Pr{D ≤ 10 | 4}] = 0 + [1 − POI(10.5(10) = 5 UCL = c + 3 c = 5 + 3 5 = 11. u = 0.86 LCL ⇒ 0 6-61.003 where POI(⋅) is the cumulative Poisson distribution.708 LCL ⇒ 0 6-59. L: sigma control limits nc − L nc > 0 nc > L nc n > L2 c 6-48 .997 = 0. 4)] = 1 − 0. 6-60.75 nonconformities/clock CL = c = u × n = 0. u = 0. c: nonconformities per unit.5 = 10.5 average nonconformities/radio CL = c = u × n = 0.

(b) Probability Plot of Transformed "Days-between-Homicides" (Ex6-62t27) Normal .005 Percent 80 70 60 50 40 30 20 10 5 1 -30 -20 -10 0 10 20 Ex6-62Bet 30 40 50 There is a huge curve in the plot points. indicating that the normal distribution assumption is not reasonable. 6-49 .806 0.760 Percent 80 70 60 50 40 30 20 10 5 1 0 1 2 Ex6-62t27 3 4 The 0.25 12.5635 28 0. (a) MTB > Graphs > Probability Plot > Single Probability Plot of Days-Between-Homicides (Ex6-62Bet) Normal .04 28 1.95% CI 99 Mean StDev N AD P-Value 95 90 12.95% CI 99 Mean StDev N AD P-Value 95 90 1.572 <0.2777th root transformation makes the data more closely resemble a sample from a normal distribution.238 0.Chapter 6 Exercise Solutions 6-62.

95% CI 99 Mean StDev N AD P-Value 95 90 1.0 3.0 1.25th root transformation makes the data more closely resemble a sample from a normal distribution.246 0.807 Percent 80 70 60 50 40 30 20 10 5 1 0.0 Ex6-62t25 2.0 0.5 2.5 UCL=3.5 LCL=0.5 2.223 0.806 1.Chapter 6 Exercise Solutions 6-62 continued (c) Probability Plot of Transformed "Days-betwee-Homicides" (Ex6-62t25) Normal .5 The 0.0 _ X=1.5 1.4789 28 0.695 0. (d) MTB > Stat > Control Charts > Variables Charts for Individuals > Individuals I Chart of Transformed Homicide Data (0.366 3.0 3 6 9 12 15 18 Observation 21 24 27 6-50 .5 1.5 3.2777 root) (Ex6-62t27) 3.0 0. It is not very different from the transformed data in (b).0 Individual Value 2.

0 _ X=1. (f) The “process” is stable. etc. One could track the number of orders filled incorrectly (wrong parts. etc. which affects the rate at which homicides occur. mean and LCL. Consider a product distribution center (or any warehouse) with processes for filling and shipping orders. (hopefully caught internally with a verification step). etc.0 Individual Value 2.. packaged incorrectly (wrong material. meaning that the days-between-homicides is approximately constant.5 1. the mean time between may get longer (or shorter) with plot points above the upper (or below the lower) control limit.25 root) (Ex6-62t25) UCL=3. wrong package labeling).025 3.0 0.365 0.695 1. policy. law.Chapter 6 Exercise Solutions 6-62 continued (e) I Chart of Transformed Homicide Data (0.0 3 6 9 12 15 18 Observation 21 24 27 Both Individuals charts are similar. If a change is made. invoiced incorrectly. workforce. too few/many parts. There are endless possibilities for collection of attributes data from nonmanufacturing processes. Or consider an accounting firm—errors in statements.5 LCL=0. There is no difference in interpretation.).5 2. 6-51 . wrong part labeling. 6-63. with an identical pattern of points relative to the UCL. say in population. errors in tax preparation.

it may be hard to determine with sufficient accuracy—both the “start” and the “end”. Test Failed at points: 15 The rate of monthly CAT scans is out of control. if you’re looking for data on time-between-events that must be obtained from external sources (for example. Also. the conditions of use and the definition of “failure” may not be consistently applied. One point more than 3. There are ways to address these difficulties. Collection of “start” time data may be facilitated by serializing or date coding product. MTB > Stat > Control Charts > Attributes Charts > U U Chart of CAT Scans (Ex6-65NSCANB) Sample Count Per Unit 40 1 35 UCL=35. The “process characteristic” to be controlled is the rate of CAT scans. A u chart which monitors the average number of CAT scans per NYRSB is appropriate.86 20 15 LCL=15. The variable NYRSB can be thought of as an “inspection unit”. it can usually be obtained reliably and consistently. If time-between-events data (say failure time) is being sought for internally generated data. time-to-field failures). However.Chapter 6 Exercise Solutions 6-64. 6-52 .94 30 25 _ U=25. 6-65☺.00 standard deviations from center line. representing an identical “area of opportunity” for each “sample”.77 94 94 94 94 94 94 94 94 94 94 94 94 95 95 95 M EB A R PR A Y UN JUL UG EP C T OV EC A N EB A R A J F J S F D A J A M N O M M Ex6-65MON Tests performed with unequal sample sizes Test Results for U Chart of Ex6-65NSCANB TEST 1.

0 2300 2200 LCL=2129. representing an identical “area of opportunity” for each “sample”. A u chart which monitors the average number of office visits per NYRSB is appropriate. The variable NYRSE can be thought of as an “inspection unit”.5 2400 _ U=2303.5 2100 JAN94 FEB94 MAR94 APR94 MAY94 Ex6-66aMON JUN94 JUL94 AUG94 Tests performed with unequal sample sizes The chart is in statistical control 6-53 .Chapter 6 Exercise Solutions 6-66☺. (a) MTB > Stat > Control Charts > Attributes Charts > U U Chart of Number of Office Visits (Ex6-66aNVIS) Phase 1 Sample Count Per Unit 2500 UCL=2476. The “process characteristic” to be controlled is the rate of office visits.

0 2200 2100 LCL=2141.00 standard deviations from center line. 6-54 . 11.Chapter 6 Exercise Solutions 6-66 continued (b) U Chart of Number of Office Visits (Ex6-66NVIS) Phase 1 Limits 1 2800 Sample Count Per Unit 1 2700 2600 2500 1 1 1 1 1 UCL=2465. 12. Test Failed at points: 9. 15 The phase 2 data appears to have shifted up from phase 1.0 2400 2300 _ U=2303. 14. One point more than 3. 13. 10. The 2nd phase is not in statistical control relative to the 1st phase.0 4 4 4 4 4 4 4 4 5 5 4 5 4 4 4 N9 EB9 A R9 PR9 A Y9 UN9 UL9 UG9 EP9 CT 9 V 9 EC9 A N9 EB9 AR9 A J F M J J F M J S A M A NO D O Ex6-66MON Tests performed with unequal sample sizes Test Results for U Chart of Ex6-66NVIS TEST 1.

Chapter 6 Exercise Solutions 6-66 continued (c) U Chart of Number of Office Visits (Ex6-66NVIS) Phase 2 Sample Count Per Unit 2800 UCL=2796. 6-55 .6 2400 9 10 11 12 Sample 13 14 15 Tests performed with unequal sample sizes The Phase 2 data.5 2600 2500 LCL=2450.5 2700 _ U=2623. separated from the Phase 1 data. are in statistical control.

010) ( ) Cˆ pk = min Cˆ pl . A second exercise number in parentheses indicates that the exercise number has changed.010 SL = 74.035] USL − LSL 74.10 Cˆ pu = 3σˆ 3(1. samples 12 and 15 are out of control. R = 0. and the new process parameters are used in the process capability analysis.001 − 73.65 = = 1. An “*” indicates that the description has changed.5. R = 4.93 USL = 40.Chapter 7 Exercise Solutions Note: Several exercises in this chapter differ from those in the 4th edition.326 = 0.13 Cˆ pu = 3σˆ 3(0. 7-1.65. µˆ = x = 33.001 = = 1. New exercises are denoted with an “☺”.17 Cˆ p 6σˆ 6(0.36 Cˆ pl = 3σˆ 3(1.965. Cˆ pu = 1. LSL = 20 USL − LSL 40 − 20 = = 1. n = 5.035 − 74.010) µˆ − LSL 74.20 Cˆ pl = 3σˆ 3(0.93) USL − µˆ 40 − 33. µˆ = x = 74.023.965 = = 1. Cˆ pu = 1.73 Cˆ p 6σˆ 6(1.035 − 73. In Exercise 5-1.10 7-1 .035 = [73.93) µˆ − LSL 33.65 − 20 = = 2.13 7-2.010) USL − µˆ 74.965 = = 1.001.023 2.93) ( ) Cˆ pk = min Cˆ pl . σˆ = R d 2 = 1. 74.000 ± 0. σˆ = R d 2 = 0.

04) The process produces product that uses approximately 18% of the total specification band.Chapter 7 Exercise Solutions 7-3. with an estimated percent defective much less than 1 ppb. However.54 pk = = 0.48 Cˆ pm = = = 1.375 = = −3.375. σˆ x = R d 2 = 6.48 Cˆ p = 6σˆ x 6(3.4128 S 3. LSL x = [(350 − 5) − 350] × 10 = −50 xi = (obsi − 350) ×10 USL x − LSL x 50 − (−50) = = 5.34 pk pu pl This is an extremely capable process.04 Cˆ p 5.375 − 0 = = 3.43 1 + 3.25. Rx = 6.34 Cˆ pu = 3σˆ x 3(3.059 = 3.62 Cˆ pl = 3σˆ x 3(3. indicating that the process is not centered and is not achieving potential capability. Cˆ ) = 4.412 1 + ξˆ 2 7-2 . ξˆ = µˆ − T 10. this PCR does not tell where the mean is located within the specification band. T − x 0 − 10. USL x − µˆ 50 − 10.25 2.04 Cˆ pkm = Cˆ 1. Note that the Cpk is less than Cp.04 USL x = [(350 + 5) − 350] ×10 = 50. the mean µ lies within approximately the middle fourth of the specification band.54 2 1+V 1 + (−3.41 σˆ 3.4128) 2 V= Since Cpm is greater than 4/3.04) Cˆ = min(Cˆ .375 = = 4.04) µˆ − LSL x 10.375 − (−50) = = 6. µˆ = x = 10.

00273) Cˆ = min(Cˆ .00109 Cˆ pu = = = 1.09 3σˆ 3(0.01 Cˆ p = = = 1.00109 − 0 = = 0.399) 2 V= Since Cpm is greater than 1.00109.33 T − x 0 − 0. USL − µˆ 0.01) Cˆ pl = = = 1.22 p ˆ C pm = = = 1. the mean µ lies within approximately the middle third of the specification band.09 pk = = 1. ξˆ = µˆ − T 0. failing to meet the minimally acceptable definition of capable Cpk ≥ 1.399 S 0. n = 5.00273 Cˆ pkm = Cˆ 1.09 pk pl pu This process is not considered capable.00109 = = −0.Chapter 7 Exercise Solutions 7-4.01 2 2 ˆ 1 0.399 + 1+ ξ 7-3 . tolerances: 0 ± 0. Cˆ ) = 1.35 3σˆ 3(0. σˆ x = 0.00273) µˆ − LSL 0.22 6σˆ 6(0. R = 0.399 σˆ 0.00273 .00109 − (−0.01 USL − LSL 0.00273 Cˆ 1.00273) The process produces product that uses approximately 82% of the total specification band.01 + 0.13 1+V 2 1 + (−0.01 − 0.00635. x = 0.

49 3σˆ x 3(1.117 ⎭ ⎣ 1.48 3σˆ x 3(1.999996] = 0.000000] = 0.117) USL x − µˆ (95 + 10) − 100 Actual: Cˆ pu = = = 1. s = 1.117) (b) µˆ − LSL x 100 − (95 − 10) Cˆ pl = = = 4.429) + [1 − Φ (4.476) ] = 0.000000 + [1 − 1.953) + [1 − Φ (8.117 ⎭ ⎣ 1.117 (a) USL − LSL (95 + 10) − (95 − 10) = = 2.117) Cˆ = min(Cˆ . µˆ = x = 100. Cˆ ) = 1. σˆ x = s c4 = 1.000000 7-4 .05.117 ⎭⎥⎦ ⎩ ⎩ = Φ (−13.05 0.0000 + [1 − 0.953) ] = 0.9400 = 1.98 Potential: Cˆ p = 6σˆ 6(1.117 ⎭⎥⎦ ⎩ ⎩ = Φ (−8.000004 85 − 95 ⎫ ⎡ ⎧ ⎧ 105 − 95 ⎫⎤ pˆ Potential = Pr ⎨ z < ⎬ + ⎢1 − Pr ⎨ z ≤ ⎬ 1.Chapter 7 Exercise Solutions 7-5.49 pk pl pu (c) pˆ Actual = Pr{x < LSL} + Pr{x > USL} = Pr{x < LSL} + [1 − Pr{x ≤ USL}] LSL − µˆ ⎫ ⎡ USL − µˆ ⎫⎤ ⎧ ⎧ = Pr ⎨ z < ⎬ + ⎢1 − Pr ⎨ z ≤ ⎬⎥ σˆ ⎭ ⎣ σˆ ⎩ ⎩ ⎭⎦ 85 − 100 ⎫ ⎡ ⎧ ⎧ 105 − 100 ⎫⎤ = Pr ⎨ z < ⎬ + ⎢1 − Pr ⎨ z ≤ ⎬ 1.

2941) ] = 0. R = 3.70) Cˆ = min(Cˆ .0000026 7-5 . µˆ = x = 199.5. (b) USL − µˆ 208 − 199 Cˆ pu = = = 1.70 ⎭ ⎩ = 2 × 0.57 Potential: Cˆ p = 6σˆ 6(1. Cˆ ) = 1. the fraction nonconforming would be: ⎧ 192 − 200 ⎫ pˆ Potential = 2 × Pr ⎨ z < ⎬ 1.70) The process produces product that uses approximately 64% of the total specification band.37 pk pl pu (c) The current fraction nonconforming is: pˆ Actual = Pr{x < LSL} + Pr{x > USL} = Pr{x < LSL} + [1 − Pr{x ≤ USL}] LSL − µˆ ⎫ ⎡ USL − µˆ ⎫ ⎤ ⎧ ⎧ = Pr ⎨ z < ⎬ + ⎢1 − Pr ⎨ z ≤ ⎬⎥ σˆ ⎭ ⎣ σˆ ⎩ ⎩ ⎭⎦ 208 − 199 ⎫⎤ ⎧ 192 − 199 ⎫ ⎡ ⎧ = Pr ⎨ z < ⎬ + ⎢1 − Pr ⎨ z ≤ ⎬ 1.70) µˆ − LSL 199 − 192 = = 1. n = 4. σˆ x = R d 2 = 3.5 2. LSL = 200 – 8 = 192 (a) USL − LSL 208 − 192 = = 1.1176) + [1 − Φ (5.37 Actual: Cˆ pl = 3σˆ 3(1.70 ⎭ ⎣ 1.76 3σˆ 3(1.70 ⎭⎥⎦ ⎩ ⎩ = Φ (−4.059 = 1.0000191 + [1 − 1] = 0.70 USL = 200 + 8 = 208.0000191 If the process mean could be centered at the specification target.Chapter 7 Exercise Solutions 7-6☺.0000013 = 0.

12094) + [1 − Φ (2.216 ⎭⎥⎦ ⎩ ⎩ = Φ (−2.216) (b) USL − µˆ 45 − 39. 2 (d) The current fraction nonconforming is: pˆ Actual = Pr{x < LSL} + Pr{x > USL} = Pr{x < LSL} + [1 − Pr{x ≤ USL}] LSL − µˆ ⎫ ⎡ USL − µˆ ⎫⎤ ⎧ ⎧ = Pr ⎨ z < ⎬ + ⎢1 − Pr ⎨ z ≤ ⎬⎥ σˆ ⎭ ⎣ σˆ ⎩ ⎩ ⎭⎦ 35 − 39.39170)] = 0. n = 2. µˆ = x = 39.7 − 40 = = −0.0169634 + [1 − 0. Cpk.216 ⎭ ⎣ 2.025348 7-6 .991615] = 0.71 pk pl pu (c) V= Cˆ pm x − T 39.216) Cˆ = min(Cˆ .71 Actual: Cˆ pl = 3σˆ 3(2. Cˆ ) = 0.71 = 0.216) µˆ − LSL 39.70 1+V 1 + (−0.7. LSL = 40 – 5 = 35 (a) USL − LSL 45 − 35 = = 0.7 − 35 = = 0.75 = = = 0.7 ⎫⎤ ⎧ ⎧ = Pr ⎨ z < ⎬ + ⎢1 − Pr ⎨ z ≤ ⎬ 2.75 Potential: Cˆ p = 6σˆ 6(2.Chapter 7 Exercise Solutions 7-7☺. R = 2.5.80 3σˆ 3(2.135) 2 The closeness of estimates for Cp.7 ⎫ ⎡ 45 − 39.5 1. σˆ x = R d 2 = 2. Cpm.74 2 1+V 1 + (−0.216 USL = 40 + 5 = 45.135) 2 Cˆ pkm = Cˆ pk = 0.7 Cˆ pu = = = 0. and Cpkm indicate that the process mean is very close to the specification target.128 = 2.216 Cˆ p 0.135 s 2.

S = 2.13) USL − µˆ 80 + 8 − 75 = = 2.000172 7-7 .47 3σˆ 3(2.47 pk pl pu (c) Let µˆ = 80 pˆ Potential = Pr{x < LSL} + Pr{x > USL} LSL − µˆ ⎫ USL − µˆ ⎫ ⎧ ⎧ = Pr ⎨ z < ⎬ + 1 − Pr ⎨ z ≤ ⎬ σˆ ⎭ σˆ ⎩ ⎩ ⎭ 72 − 80 ⎫ 88 − 80 ⎫ ⎧ ⎧ = Pr ⎨ z < ⎬ + 1 − Pr ⎨ z ≤ ⎬ 2.000086 + 1 − 0.26} = 2 × 0.13 ⎭ 2.Chapter 7 Exercise Solutions 7-7 (d) continued If the process mean could be centered at the specification target. Cˆ ) = 0.756) = 0.756) + 1 − Φ (3. µˆ = 75.02382 7-8 (7-6).13) Cˆ = min(Cˆ .13 (a) USL − LSL 2(8) Potential: Cˆ p = = = 1.13 ⎭ ⎩ ⎩ = Φ (−3. σˆ = Sˆ c4 = 2 0.216 ⎭ ⎩ = 2 × Pr{z < −2.9400 = 2.01191 = 0.999914 = 0. the fraction nonconforming would be: 35 − 40 ⎫ ⎧ pˆ Potential = 2 × Pr ⎨ z < ⎬ 2.03 Actual: Cˆ pu = 3σˆ 3(2.25 6σˆ 6(2.13) (b) µˆ − LSL 75 − (80 − 8) Cˆ pl = = = 0.

133 6(1. sB = 1.064) µˆ − LSL x 105 − (100 − 10) Cˆ pl = x = = 4.191 Cˆ p 1.045 3σˆ x 3(3.064 Cˆ p = USL − LSL 6σˆ = (100 + 10) − (100 − 10) = 3.191 ⎭ ⎩ ⎩ = Φ (−3.9400 = 3. Assume n = 5 Process A µˆ = x A = 100.191) µˆ − LSL x 100 − (100 − 10) Cˆ pl = = = 1.191 USL − LSL (100 + 10) − (100 − 10) = = 1.13) = 0.13) + 1 − Φ (3.191) Cˆ = min(Cˆ . Cˆ ) = 1.566 Cˆ pu = 3σˆ x 3(1.064) USL x − µˆ x (100 + 10) − 105 = = 1.699 3σˆ x 3(1.99913 = 0.064) Cˆ pk = min(Cˆ pl .00087 + 1 − 0.045 2 1+V 1 + (0) 2 pˆ = Pr{x < LSL} + Pr{x > USL} = Pr{x < LSL} + 1 − Pr{x ≤ USL} LSL − µˆ ⎫ USL − µˆ ⎫ ⎧ ⎧ = Pr ⎨ z < ⎬ + 1 − Pr ⎨ z ≤ ⎬ σˆ ⎭ σˆ ⎩ ⎩ ⎭ 90 − 100 ⎫ ⎧ ⎧ 110 − 100 ⎫ = Pr ⎨ z < ⎬ + 1 − Pr ⎨ z ≤ ⎬ 3.9400 = 1. σˆ A = s A c4 = 3 0. σˆ B = sB c4 = 1 0.045 3σˆ x 3(3.00174 Process B µˆ = xB = 105. Cˆ pu ) = 1. s A = 3.045 = = = 1.191 ⎭ 3.045 Cˆ p = 6σˆ 6(3.191) USL x − µˆ (100 + 10) − 100 Cˆ pu = = = 1.045 pk pl V= Cˆ pm pu x − T 100 − 100 = =0 s 3.Chapter 7 Exercise Solutions 7-9 (7-7).566 7-8 .

000001 Prefer to use Process B with estimated process fallout of 0. For the parts Cˆ pk . 7-10 (7-8).000000 + 1 − 0.133 Cˆ pm = = = 0.191) 2 = 14.652 1+V 2 1 + (−4.699) 2 90 − 105 ⎫ ⎧ ⎧ 110 − 105 ⎫ pˆ = Pr ⎨ z < ⎬ + 1 − Pr ⎨ z ≤ ⎬ 1.699 s 1.098) + 1 − Φ (4.064) 2 = 4. A = 1. σˆ A = 20σˆ 2 = 20(3.064 Cˆ p 3. B indicates that more parts from Process B are within ( ) ( ) specification than from Process A. σˆ B = 20σˆ 2 = 20(1.045 < 1.064 ⎭ 1. 7-9 .999999 = 0.064 ⎭ ⎩ ⎩ = Φ (−14.271 Process B: µˆ B = 20(105) = 2100.000001 instead of Process A with estimated fallout 0.566 = Cˆ pk . Process A: µˆ A = 20(100) = 2000.699) = 0.Chapter 7 Exercise Solutions 7-9 continued x − T 100 − 105 V= = = −4.758 Process B will result in fewer defective assemblies.001726.

9975 ⎫ ⎧ ⎧ pˆ = Pr ⎨ z < ⎬ = Pr ⎨ z < ⎬ = Φ (−0.9968 0.323 0.950 0. p84 = 1. MTB > Stat > Basic Statistics > Normality Test Probability Plot of 1-kg Containers (Ex7-9Wt) Normal 99 Mean StDev N AD P-Value 95 90 0.9975 − 0.985 kg µˆ − LSL 0.0225) = 0.025 1.02167 15 0.0225 6σˆ = 6(0.0200 − 0.556) = 0.0200 σˆ = p84 − p50 = 1.9975.985 − 0.0225 σˆ ⎭ ⎩ ⎩ ⎭ 7-10 .975 1.19 C pl = 3σˆ 3(0. LSL = 0.Chapter 7 Exercise Solutions 7-11 (7-9).1350 7-12☺.050 A normal probability plot of the 1-kg container weights shows the distribution is close to normal.000 Ex7-9Wt 1.289105 0.9975 = 0. x ≈ p50 = 0.492 Percent 80 70 60 50 40 30 20 10 5 1 0.985 = = 0.0225) LSL − µˆ ⎫ 0.

009242 25 0.98 19.00 0.99986 p84 = 20.99 20.) Probability Plot of Disk Height (Ex7-13Ht) Normal 99 95 90 84 Percent 80 Mean StDev N AD P-Value 20.00919 6σˆ = 6(0.00919) = 0.174 70 60 50 40 30 50 19.00905 − 19. MTB > Stat > Basic Statistics > Normality Test (Add percentile lines at Y values 50 and 84 to estimate µ and σ.02 A normal probability plot of computer disk heights shows the distribution is close to normal.Chapter 7 Exercise Solutions 7-13☺.00905 σˆ = p84 − p50 = 20.00 Disk Height.00905 20 20.01 20.99986 = 0.05514 7-11 . x ≈ p50 = 19. mm 20.515 0.99986 10 5 1 19.

07) = 24.340 70 60 50 40 30 50 20 1 13. MTB > Stat > Basic Statistics > Normality Test (Add percentile lines at Y values 50 and 84 to estimate µ and σ.42 7-12 .2 p84 = 17.27 σˆ = p84 − p50 = 17.27 10 10 15 20 Reimbursement Cycle Time.) Probability Plot of Cycle Time (Ex7-14CT) Normal 99 95 90 84 Percent 80 Mean StDev N AD P-Value 13.07 6σˆ = 6(4.2 5 5 17. Days 25 A normal probability plot of reimbursement cycle times shows the distribution is close to normal.Chapter 7 Exercise Solutions 7-14☺.097 30 0.401 0.2 4.2 = 4.27 − 13. x ≈ p50 = 13.

2 6σˆ = 6(12.98 10 90 100 110 Response Time.463 0.58 C pu = 3σˆ 3(12.78 ⎫ pˆ = Pr ⎨ z > ⎬ = 1 − Pr ⎨ z < ⎬ = 1 − Pr ⎨ z < ⎬ σˆ σˆ 12.98 − 98.243 70 60 50 40 30 50 20 1 98.739) = 1 − 0.) Probability Plot of Response Time (Ex7-15Resp) Normal 99 95 90 84 Percent 80 Mean StDev N AD P-Value 98.2 ⎩ ⎭ ⎩ ⎭ ⎩ ⎭ = 1 − Φ (1.958983 = 0.041017 7-13 .78 5 70 80 110.78 p84 = 110. (a) x ≈ p50 = 98.Chapter 7 Exercise Solutions 7-15☺.98 σˆ = p84 − p50 = 110.78 = = 0. minutes 120 130 A normal probability plot of response times shows the distribution is close to normal.78 = 12.2) = 73.78 12. MTB > Stat > Basic Statistics > Normality Test (Add percentile lines at Y values 50 and 84 to estimate µ and σ.2) USL − µˆ ⎫ USL − µˆ ⎫ ⎧ ⎧ ⎧ 120 − 98.27 40 0.2 (b) USL = 2 hrs = 120 mins USL − µˆ 120 − 98.

) Probability Plot of Hardness Data (Ex5-59Har) Normal 99 95 90 84 Percent 80 Mean StDev N AD P-Value 53. x ≈ p50 = 53.14 7-14 .27 2.465 0.217 70 60 50 40 30 50 20 1 53.27 p84 = 55.96 σˆ = p84 − p50 = 55. MTB > Stat > Basic Statistics > Normality Test (Add percentile lines at Y values 50 and 84 to estimate µ and σ.Chapter 7 Exercise Solutions 7-16 (7-10).27 5 46 48 50 52 54 Hardness 55.96 − 53.712 15 0.69) = 16.96 10 56 58 60 A normal probability plot of hardness data shows the distribution is close to normal.69 6σˆ = 6(2.27 = 2.

587 Percent 80 70 60 50 40 30 20 10 5 1 1000 1500 2000 Ex7-17FT 2500 3000 The plot shows that the data is not normally distributed. so it is not appropriate to estimate capability.272 0. MTB > Stat > Basic Statistics > Normality Test Probability Plot of Failure Times (Ex7-17FT) Normal 99 Mean StDev N AD P-Value 95 90 1919 507.1 10 0. 7-15 .Chapter 7 Exercise Solutions 7-17 (7-11).

11 Cˆ p = 6σˆ 6(1.33 ⎞ χ = 49 ⎜ ⎟ = 37.36 χ12−α / 2. n = 25.49 χ2 Cˆ p 1−α .36 ≤ C p ≤ 1. n −1 0.n −1 = χ 0. USL = 85.24 = 39.11 12.n −1 ≤ C p n −1 1. 7-19 (7-13). 1.11 25 − 1 25 − 1 0.42 This confidence interval is wide enough that the process may either be capable (ppm = 27) or far from it (ppm ≈ 16.52 p 1 − α = 0.88 α = 0.9303 = 1.9303 χ12−α .975.5 (a) USL − LSL 85 − 75 = = 1. S = 1.33 2 ⎛ 1.33 at a 95% confidence level.49 49 = 1.95.24 = 12.80 ≤ C p ≤ 1.40 2 χα2/ 2. LSL = 75.025.5) (b) α = 0.12 2 1−α .52 χ12−α .Chapter 7 Exercise Solutions 7-18 (7-12).n −1 = χ 0.n −1 χα2/ 2.52 33.05 χ2 = χ2 1−α / 2.395).26 ≤ C p 49 The company cannot demonstrate that the PCR exceeds 1.40 39.95 2 = 33. n = 50 Cˆ = 1.52 ⎝ 1.52 ⎠ 1 − α = 0.n −1 ˆ ˆ Cp ≤ Cp ≤ Cp n −1 n −1 1.49 7-16 .

96 ⎥ ≤ C pk ≤ 0. USL = 100.6) Cˆ pk = min(Cˆ pl .63 3σˆ x 3(1. Cˆ pu ) = 0.96 ⎥ 2 2 9(30)(0.8313 7-17 . n = 30.6.960 ⎡ 1 1 ⎤ ⎥ ≤ C pk ≤ Cˆ pk + Cˆ pk ⎢1 − zα / 2 2 2(n − 1) ⎥ 9nCˆ pk ⎢⎣ ⎦ ⎡ 1 1 ⎤ ⎢1 + zα / 2 ⎥ + 2 2(n − 1) ⎥ 9nCˆ pk ⎢⎣ ⎦ ⎡ ⎤ ⎡ ⎤ 1 1 1 1 + + 0.025 = 1. LSL = 90 (a) USL x − µˆ x 100 − 97 Cˆ pu = = = 0.63 ⎢1 − 1.63) 2(30 − 1) ⎦ 9(30)(0.6) µˆ − LSL x 97 − 90 Cˆ pl = x = = 1. S = 1.Chapter 7 Exercise Solutions 7-20 (7-14).63) 2(30 − 1) ⎦ ⎣ ⎣ 0.05 zα / 2 = z0. x = 97.63 ⎢1 + 1.46 3σˆ x 3(1.4287 ≤ C pk ≤ 0.63 (b) α = 0.

42) 2(50 − 1) ⎦ ⎣ ⎣ 0.63 ⎢1 − 1. nominal = 2225.79 The approximation yields a narrower confidence interval. zα / 2 = 1. 7-23 (7-17). 7-20.96 ⎥ 2(30 − 1) ⎦ 2(30 − 1) ⎦ ⎣ ⎣ 0.96 ⎥ ≤ C pk ≤ 0. n = 50 (a) USL x − µˆ x 2350 − 2275 = = 0. but it is not too far off. σ OI = 0. n = 30 ⎡ ⎡ 1 ⎤ 1 ⎤ Cˆ pk ⎢1 − zα / 2 ⎥ ≤ C pk ≤ Cˆ pk ⎢1 + zα / 2 ⎥ 2(n − 1) ⎦ 2(n − 1) ⎦ ⎣ ⎣ ⎡ ⎤ ⎡ ⎤ 1 1 0.960 ⎡ 1 1 ⎤ ⎥ ≤ C pk ≤ Cˆ pk Cˆ pk ⎢1 − zα / 2 + 2 2(n − 1) ⎥ 9nCˆ pk ⎢⎣ ⎦ ⎡ 1 1 ⎤ ⎢1 + zα / 2 ⎥ + 2 2(n − 1) ⎥ 9nCˆ pk ⎢⎣ ⎦ ⎡ ⎤ ⎡ ⎤ 1 1 1 1 0.5443 7-22 (7-16).63 ⎢1+1. USL = 2350. from Ex.47 ≤ C pk ≤ 0. σˆ I = 3.63.42) 2(50 − 1) ⎦ 9(50)(0.Chapter 7 Exercise Solutions 7-21 (7-15).42 ⎢1 − 1.96. LSL = 2100.96 + + ⎥ ≤ C pk ≤ 0.42 (b) α = 0.42 ⎢1 + 1.97 Cˆ pl = x 3σˆ x 3(60) Cˆ pk = min(Cˆ pl . Cˆ pu ) = 0.42 Cˆ pu = 3σˆ x 3(60) µˆ − LSL x 2275 − 2100 = = 0. σˆ Total = 5 2 2 2 σˆ Total = σˆ Meas + σˆ Process 2 2 σˆ Process = σˆ Total − σˆ Meas = 52 − 32 = 4 7-18 . Cˆ pk = 0.025 = 1.96 ⎥ 2 2 9(50)(0.2957 ≤ C pk ≤ 0.05. x = 2275. zα / 2 = z0. s = 60.

One point more than 3. 15.482 MTB > Stat > Control Charts > Variables Charts for Subgroups > X-bar R Xbar-R Chart of Part Measurements (Ex7-24All) 30 1 Sample M ean U C L=27. and the x chart has a few out-of-control parts.8 20 1 15 2 4 6 8 1 1 10 Sample 12 14 16 18 LC L=16.Chapter 7 Exercise Solutions 7-24 (7-18). x = 21.07 25 _ _ X=21.482) P = ×100 = ×100 = 49.00 standard deviations from center line. 20 The R chart is in control. (a) n = 2.15 Sample Range 8 6 4 _ R=2. σˆ Gauge = 2. R = 2. Test Failed at points: 8.8 2 0 LC L=0 2 4 6 8 10 Sample 12 14 16 18 20 Test Results for Xbar Chart of Ex7-24All TEST 1.53 20 U C L=9. The new gauge is more repeatable than the old one. (b) specs: 25 ± 15 6σˆ Gauge 6(2.8.8.6% T USL − LSL 2(15) 7-19 . 12.

R = 2.359 × 100 = 62.2.693 = 1.553 Sample M ean 100. σˆ Gauge = R d 2 = 2.717 − 1. This indicates that the operator is not having difficulty making consistent measurements.5% 4.00 standard deviations from center line.0 U C L=5.872 σˆ Product = 1.847 1 1 2 3 4 5 6 7 8 9 10 Sample Sample Range 6. 3 The x chart has a couple out-of-control points.5 0. and the R chart is in control.0 LC L=0 1 2 3 4 5 6 7 8 9 10 Sample Test Results for Xbar Chart of Ex7-25All TEST 1.2 97.695 (c) σˆ Gauge σˆ Total × 100 = 1. One point more than 3.272 T USL − LSL 115 − 85 7-20 .359 2 σˆ Total = 4. LSL = 100 – 15 = 85 6σˆ Gauge P 6(1. (b) x = 98.717 2 2 2 σˆ Product = σˆ Total − σˆ Gauge = 4.Chapter 7 Exercise Solutions 7-25 (7-19).0 1 U C L=100.5 99.3592 = 2.5 96.5 3.0 LC L=95.3 1.359) = = = 0.717 (d) USL = 100 + 15 = 115.3 1. MTB > Stat > Control Charts > Variables Charts for Subgroups > X-bar R Xbar-R Chart of Part Mesaurements (Ex7-25All) 102.0 _ R=2. Test Failed at points: 2.921 4.0 _ _ X=98.3.

70.151 = 1.151 (b) 2 2 2 ˆ2 ˆ2 σˆ Measurement Error = σ Repeatability + σ Reproducibility = 1.87.693 σˆ Repeatability = R d 2 = 2.191) = ×100 = × 100 = 35.00 1.181 + 0.693 = 1.17 n = 2 operators d 2 = 1. (a) Excel : workbook Chap07.30 R = 2.xls : worksheet Ex7-26 x1 = 50.00 n = 3 repeat measurements d 2 = 1.Chapter 7 Exercise Solutions 7-26 (7-20).128 = 0.418 σˆ Measurement Error = 1. x2 = 49.128 σˆ Reproducibility = Rx d 2 = 0.03.191 (c) specs: 50 ± 10 6σˆ Gauge P 6(1.7% T USL − LSL 60 − 40 7-21 . R1 = 1. R2 = 2.17 1.181 Rx = 0.

82 1 15.5 3.0 1 2 3 4 5 6 7 8 Sample 9 10 11 12 13 14 15 Test Results for R Chart of Ex7-27All TEST 1.5 LC L=17.154 (b) MTB > Stat > Control Charts > Variables Charts for Subgroups > X-bar R Xbar-R Chart of Part Measurements (Ex7-27All) 1 1 Sample M ean 25. 7-22 .0 U C L=23. Test Failed at points: 11.00 standard deviations from center line.0 11 12 1 1 13 14 15 U C L=5.Chapter 7 Exercise Solutions 7-27 (7-21).359 Gauge capability: 6σˆ = 8.5 LC L=0 0.010 4.0 _ R=1.0 1 1 2 3 4 5 6 7 8 Sample 9 10 Sample Range 6. One point more than 3.533 1.58 22.5 _ _ X=20.128 = 1.0 17.7 20. 12 Out-of-control points on R chart indicate operator difficulty with using gage.533 1. (a) σˆ Gauge = R d 2 = 1.

0106 0. For the reduced model: ANOVA: Ex7-28Reading versus Ex7-28Part.10).72 P 0. Levels 20 3 3. 15. 13.0149 -0. 1.500 0.74% Variance component 10.1399 0. MTB > Stat > ANOVA > Balanced ANOVA In Results. 10. 20 Ex7-28Op Values 1.0149 = pn 20(2) MSP − MSP×O 62.65 1.173 0. we can fit a reduced model without that term.000 0. 4. 16.861 R-Sq(adj) = 90.391 − 0.995825 1 2 3 4 9.992 = = −0.050 0. 8. 2. 17.391 Ex7-28Op 2 2. 19. Ex7-28Op Factor Ex7-28Part Ex7-28Op Type random random Factor Ex7-28Part 18.712 Error 60 59.425 62. Ex7-28Op … 1 2 3 Source Ex7-28Part Ex7-28Op Error Variance component 10. Since the Part × Operator term is not significant (α = 0.2798 0. 3 Analysis of Variance for Ex7-28Reading Source DF SS MS Ex7-28Part 19 1185.992 MSP×O − MSE 0. 12.308 Ex7-28Part*Ex7-28Op 38 27.9917 Error term 3 3 4 Expected Mean Square for Each Term (using unrestricted model) (4) + 2 (3) + 6 (1) (4) + 2 (3) + 40 (2) (4) + 2 (3) (4) 2 σˆ Repeatability = MS Error = 0. 2.712 2 σˆ Part = = 10.617 1. select “Display expected mean squares and variance components” ANOVA: Ex7-28Reading versus Ex7-28Part. 14.712 2 σˆ Operator = = 0.1400 ⇒ 0 n 2 MSO − MSP×O 1. 5.Chapter 7 Exercise Solutions 7-28☺. 6.2513 0. 7. 11.8832 Error term 3 3 Expected Mean Square for Each Term (using unrestricted model) (3) + 6 (1) (3) + 40 (2) (3) 7-23 . Note the Part × Operator variance component is negative.592 S = 0.33% Source Ex7-28Part Ex7-28Op Ex7-28Part*Ex7-28Op Error F 87.2798 = on 3(2) 2 σˆ Part×Operator = The manual calculations match the MINITAB results.84 0. R-Sq = 95.308 − 0.712 − 0.992 Total 119 1274.

94541 5.0305 Number of Distinct Categories = 4 %Study Var (%SV) 28.20176 19.232 Gage R&R Source Total Gage R&R Repeatability Reproducibility Ex7-28Op Part-To-Part Total Variation VarComp 0.8832 0.00 7-24 . repeatability.8938 0.05 59.0106 10.3908 1.1451 %Contribution (of VarComp) 8. and total gauge variability may also be found using: MTB > Stat > Quality Tools > Gage Study > Gage R&R Study (Crossed) Gage R&R Study .9917 F 87.4814 P 0.10310 0.7178 P 0.0106 0.3083 0.861 Two-Way ANOVA Table Without Interaction Source Ex7-28Part Ex7-28Op Repeatability Total DF 19 2 98 119 SS 1185.9454 n = = 0.93977 5.8832 = 0.59 MS 62.Chapter 7 Exercise Solutions (a) 2 2 σˆ Reproducibility = σˆ Operator = 0.43 2.09 3.32 28.10.000 0. reproducibility.8380 0.6724 Repeatability 0.09 95.00 Study Var Source StdDev (SD) (6 * SD) Total Gage R&R 0.6447 1.8832 F 70.33842 20.6470 1.15 3.59 MS 62.0106 + 0.8938 σˆ Gauge = 0.98 100.1050 /T = P USL-LSL 60 − 6 This gauge is borderline capable since the estimate of P/T ratio just exceeds 0.62 27.55 1274.000 0.6186 Ex7-28Op 0.10 91.10 0.173 0.92 0.2106 Total Variation 3.9454 (c) 6 × σˆ Gauge 6 × 0.3083 0.91 100.6386 Reproducibility 0.ANOVA Method Two-Way ANOVA Table With Interaction Source Ex7-28Part Ex7-28Op Ex7-28Part * Ex7-28Op Repeatability Total DF 19 2 38 60 119 SS 1185.7118 0.50 1274.62 86.43 2.0106 2 2 σˆ Repeatability = σˆ Error = 0.6186 Part-To-Part 3.02 7.3908 1. Estimates of variance components.2513 11.10310 0.8832 (b) 2 2 2 σˆ Gauge = σˆ Reproducibility + σˆ Repeatability = 0.

15 20 LCL=0 0 1 2 Ex7-28Op Xbar Chart by Ex7-28Op 1 2 3 Ex7-28Op * Ex7-28Part Interaction 3 30 30 25 UCL=24.23 Average Sample Mean 8 Ex7-28Op 1 2 3 25 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 1 5 16 17 18 19 20 Ex7-28Part 7-25 .55 _ _ X=22.757 30 25 2 _ R=1.39 20 LCL=20.Chapter 7 Exercise Solutions 7-28 continued Visual representations of variability and stability are also provided: Gage R&R (ANOVA) for Ex7-28Reading Reported by : Tolerance: M isc: G age name: Date of study : Components of Variation Ex7-28Reading by Ex7-28Part 100 % Contribution 30 Percent % Study Var 25 50 20 0 Gage R&R Repeat Reprod 1 Part-to-Part 2 3 4 5 R Chart by Ex7-28Op Sample Range 4 1 2 6 7 9 10 11 12 13 14 15 16 17 18 19 20 Ex7-28Part Ex7-28Reading by Ex7-28Op 3 UCL=3.

94 DR 1 − ρˆ P 1 − 0. µ = µ1 + µ2 + µ3 = 100 + 75 + 75 = 250 σ = σ 12 + σ 22 + σ 33 = 42 + 42 + 22 = 6 Pr{x > 262} = 1 − Pr{x ≤ 262} 262 − µ ⎫ ⎧ = 1 − Pr ⎨ z ≤ ⎬ σ ⎭ ⎩ 262 − 250 ⎫ ⎧ = 1 − Pr ⎨ z ≤ ⎬ 6 ⎩ ⎭ = 1 − Φ (2. but larger than a value of two (or less) that indicates inadequate gauge capability.2513. exceeding the minimum recommendation of four.9198 m = 1 + ρˆ P = 1 + 0. This is near the AIAG-recommended value of five levels or more. By this measure.9772 = 0.9198) = = 4.9198 2 σˆ Total 11. σˆ Total = 11.1451 n= SNR 2 ρˆ P 2(0. (Also note that the MINITAB Gage R&R output indicates “Number of Distinct Categories = 4”.79 1 − ρˆ P 1 − 0.0228 7-26 .9198 = 23.94. this is also the number of distinct categories of parts that the gauge is able to distinguish) DR = 23. 2 2 σˆ Part = 10.2513 = = 0.Chapter 7 Exercise Solutions 7-29☺. 7-30 (7-22).000) = 1 − 0.79 indicates that fewer than five distinct levels can be reliably obtained from the measurements.9198 SNR = 4.1451 ρˆ P = 2 σˆ Part 10. the gauge is capable.

0)(4.01) + 3.42 = 0.04252 7-27 .12 ) + 32 (0.01) = 0.4 σ y2 = σ 12 + σ 22 = 0.08(10)(20)(3) = 48 2 σˆ Weight ≅ d 2 ⎡⎣ µˆW2 σˆ L2σˆT2 + µˆ L2σˆW2 σˆT2 + µˆT2σˆW2 σˆ L2 ⎤⎦ = 0.32 ).01)(0.0(3.01)(0.9 − 0.02 (0.9 µ y = µ1 − µ2 = 20 − 19.2743 + 1 − 0.22 )(0.25 ⎠ = Φ (−0.25 ⎠ ⎝ 0.8413 = 0.6 = 0.9} ⎛ 0.9} = Pr{ y < 0. Weight = d × W × L × T ≅ d [ µW µ L µT + (W − µW ) µ L µT + ( L − µ L ) µW µT + (T − µT ) µW µ L ] µˆ Weight ≅ d [ µW µ L µT ] = 0.02 (0.50 Pr{Nonconformities} = Pr{ y < LSL} + Pr{ y > USL} = Pr{ y < 0.01)(0.4 ⎞ = Φ⎜ ⎟ +1− Φ ⎜ ⎟ ⎝ 0.1} + Pr{ y > 0.0061 7-33 (7-25).1 or y = x1 − x2 > 0.02 (0.12 ) + 202 (0.2 2 )(0.4 ⎞ ⎛ 0. x2 ~ N (19. Volume = L × H × W ≅ µ L µ H µW + ( L − µ L ) µ H µW + ( H − µ H ) µ L µW + (W − µW ) µ L µ H µˆ Volume ≅ µ L µ H µW = 6.082 ⎡⎣10 2 (0.1 − 0.6.6) + 1 − Φ (1.4330 7-32 (7-24).32 ) ⎤⎦ = 0.25 σ y = 0.Chapter 7 Exercise Solutions 7-31 (7-23).32 )(0.42 ) Nonconformities will occur if y = x1 − x2 < 0.32 + 0.1} + 1 − Pr{ y ≤ 0.01) + 4.0) = 72.00) = 0. 0.00181 σˆ Weight ≅ 0.0 2 σ Volume ≅ µ L2σ H2 σ W2 + µ H2 σ L2σ W2 + µW2 σ L2σ H2 = 6. 0. x1 ~ N (20.

1026 − (3.1282 2⎟ 26 ⎜⎝ 3 2 2 ⎣ 26 ⎦ ⎠ 4 4 4 ⎞ 1 ⎛5 2 ⎡1 ⎤ E ( x 2 ) = ∫ x 2 f ( x)dx = ∫ x 2 ⎢ (5 x − 2) ⎥ dx = ⎜ x 4 − x3 ⎟ = 10. I = E ( R1 + R2 ) µI ≅ µE (µR + µR ) 1 σ I2 ≅ σ 2 µE σ 2 + σ R2 2 ( R (µR + µR ) (µR + µR ) 1 2 E + 2 1 1 2 ) 2 7-28 .05(3.05 x)2 ⎤ ⎥ =⎢ ∂x ⎢⎣ ⎥⎦ 2 σ x2 µx = 2(3 + 0.1026 26 ⎜⎝ 4 2 3 2 ⎟⎠ 2 ⎣ 26 ⎦ s = (3 + 0.1001 7-35 (7-27).1282) ] (0.9629 2 ⎡ ∂ g ( x) ⎤ σ ≅⎢ ⎣ ∂ x ⎥⎦ 2 2 2 s σ x2 µx ⎡ ∂ (3 + 0.05 x) 2 and f ( x) = σ x2 = E ( x 2 ) − [ E ( x)] = 10.05µ x )(0.3170) = 0.3170 2 µ s ≅ g ( x) = [3 + 0.1282)] = 9.05( µ x )] = [3 + 0. 1 (5 x − 2).Chapter 7 Exercise Solutions 7-34 (7-26). 2 ≤ x ≤ 4 26 4 4 4⎞ 1 ⎛5 3 ⎡1 ⎤ E ( x) = µ x = ∫ xf ( x)dx = ∫ x ⎢ (5 x − 2) ⎥ dx = ⎜ x − x 2 ⎟ = 3.05(3.05)σ x2 = 2 [3 + 0.1282) 2 = 0.05)(0.

512) − 0.4 1 ⎛ 2 − 0. x2 ~ N ( µ2 .09} = 0.4002 ).006 ⎞ = 1− Φ ⎜ ⎟ ⎝ 0.000005 σ y = 0.010.09 − µ y = −2.512 0.346 7-37 (7-29).004.01 γ = 0. 0.4 = 5.006) σ y ⎭⎪ ⎩⎪ 0.3002 = 0.01 ⎠ 4 2 n≅ 7-29 .3002 ) µ y = µ1 − µ2 σ y = σ 12 + σ 22 = 0.0012 ) Interference occurs if y = ID – OD < 0 µ y = µID − µOD = 2.006 2 2 σ y2 = σ ID + σ OD = 0.9964 7-38 (7-30).0022 ) and OD ~ N (2.80 2 χ12−γ .5(−2.989 1 ⎛ 2 − α ⎞ χ1−γ .683) = 1 − 0. ID ~ N (2.Chapter 7 Exercise Solutions 7-36 (7-28).5 µ y = −[0.4 = χ 0.5 Pr{ y < 0.0012 = 0.0022 + 0. x1 ~ N ( µ1 .002236 Pr{positive clearance} = 1 − Pr{interference} = 1 − Pr{ y < 0} ⎛ 0 − 0.989 +⎜ = +⎜ = 299 ⎟ ⎟ 2 ⎝ α ⎠ 4 2 ⎝ 0.004 = 0.006 ⎧⎪ 0.010 − 2. 0. 0.20. α = 0.4002 + 0.09 − µ y ⎫⎪ −1 Pr ⎨ z < ⎬ = Φ (0.0036 = 0. 0.000005 ⎠ = 1 − Φ (−2.01 ⎞ 5.09] = 1.

γ = 0.355(10) = 323. x ~ N (300.90 . one-sided From Appendix VIII: K = 2. α = 0.05 ⎞ 7.08 7-42 (7-34).90 2 χ12−γ . α = 0.4 = 7.12 ).95 . x ~ N (85. α = 0.779 1 ⎛ 2 − α ⎞ χ1−γ .95 . one-sided From Appendix VIII: K = 2.05 ⎠ 4 2 n≅ After the data are collected.Chapter 7 Exercise Solutions 7-39 (7-31).05 γ = 0. α = 0.838 x − KS = 85 − 1. n = 10. γ = 0.162 7-41 (7-33). n = 20. γ = 0.838(1) = 83.102 ). 7-30 . one-sided From Appendix VIII: K = 1.10.208(10) = 372.05. x ~ N (350. a natural tolerance interval would be the smallest to largest observations.779 +⎜ = +⎜ = 77 ⎟ ⎟ 2 ⎝ α ⎠ 4 2 ⎝ 0. n = 25.4 1 ⎛ 2 − 0.102 ).10.208 UTL = x + KS = 350 + 2.4 = χ 0.355 UTL = x + KS = 300 + 2.10.55 7-40 (7-32).

while in part (b) the interval refers to a parameter of a distribution (an unknown constant).95) n= = = 59 log(1 − α ) log(1 − 0. part (b) is a confidence interval on mean thickness.39 = 2. 7-31 . x ~ N 0. In part (a).n −1 = t0.1263.0003) = [0.05) The largest observation would be the nonparametric upper tolerance limit. and two-sided From Appendix VII: K = 2.0003 ) 40 = [0.1265] Part (a) is a tolerance interval on individual thickness observations.1257.05. the interval relates to individual observations (random variables).95 log(1 − γ ) log(1 − 0.Chapter 7 Exercise Solutions 7-43 (7-35).1264 ± 2.n −1 S ( n = 0.023 0.05. α = 0.95.445 TI on x : x ± KS = 0.05.025. 7-44 (7-36). 0.445(0. 0.1264 ± 2. γ = 0. 0.00032 ( ) (a) α = 0.023 CI on x : x ± tα / 2. γ = 0. tα / 2.1264.1271] (b) α = 0.

5. The point at which the assignable cause occurred can be determined by counting the number of increasing plot points. 8-1 . New exercises are denoted with an “☺”. h = 5 1250 Cumulative Sum 1000 750 500 250 UCL=125 0 0 LCL=-125 2 4 6 8 10 12 Sample 14 16 18 20 The process signals out of control at observation 10.4345 No. A number in parentheses gives the exercise number from the 4th edition. 8-1. The estimate used for σ is much smaller than that from the data. (b) σˆ = MR2 d 2 = 38. µ0 = 1050.Chapter 8 Exercise Solutions Several exercises in this chapter differ from those in the 4th edition. δ = 1σ.8421/1. The assignable cause occurred after observation 10 – 3 = 7. std dev = 25. σ = 25. An “*” following the exercise number indicates that the description has changed. H = 5σ = 5(25) = 125 (a) MTB > Stat > Control Charts > Time-Weighted Charts > CUSUM CUSUM Chart of Molecular Weight (Ex8-1mole) target = 1050. k = 0. K = (δ/2)σ = (1/2)25 = 12.5.128 = 34.

The assignable cause occurred after observation 10 – 3 = 7. h = 5 50 Cumulative Sum 40 30 20 10 UCL=5 0 0 LCL=-5 2 4 6 8 10 12 Sample 14 16 18 20 The process signals out of control at observation 10. k = 0. MTB > Stat > Control Charts > Time-Weighted Charts > CUSUM CUSUM Chart of Standardized Molecular Weight (Ex8-2std) target = 1050.5.Chapter 8 Exercise Solutions 8-2. 8-2 . std dev = 25.

5 MTB > Stat > Control Charts > Time-Weighted Charts > CUSUM CUSUM Chart of Molecular Weight (Ex8-1mole) FIR=H/2 = 62. in std dev units = 62.5.5 std dev units) 1250 Cumulative Sum 1000 750 500 250 UCL=125 0 0 LCL=-125 2 4 6 8 10 12 Ex8-1Obs 14 16 18 20 For example.5.5] = 45 Using the tabular CUSUM. σ = 25. 8-3 . the same as the CUSUM without a FIR feature. K = 12.Chapter 8 Exercise Solutions 8-3.5 FIR = H/2 = 62. xi − ( µ0 − K ) + C0+ ⎤⎦ = max [ 0. (a) µ0 = 1050. the process signals out of control at observation 10. H/2 = 125/2 = 62.5. h = 5. C1+ = max ⎡⎣0.5) + 62.5 (or 2.1045 − (1050 + 12.5/25 = 2. k = 0.

5SL=0 2 4 6 8 10 12 Observation 14 16 18 20 Using 3. However there does appear to be a trend up from observations 6 through 12—this is the situation detected by the cumulative sum. there are no out-of-control signals.Chapter 8 Exercise Solutions 8-3 continued (b) MTB > Stat > Control Charts > Variables Charts for Individuals > I-MR I-MR Chart of Molecular Weight (Ex8-1mole) with 3.5SL=995.6 120 80 __ MR=38.5-sigma limits +3.8 40 0 -3.8 2 4 6 8 10 12 Observation 14 16 18 20 160 Moving Range +3.5SL=1236.5σ limits on the Individuals chart.8 Individual Value 1200 _ X=1116.3 1100 1000 -3. 8-4 .5SL=141.

05.05 is probably not reasonable.77 + 1.77 0.128 = 0.48 8-5 .5 exp[−2(−0.2385 -0.1 -0.05.5)(5.5 = −0.0 0 -0.77 (0.02.2 LCL=-0.0027 = 371.5 = −0.166 = 4. In Exercise 8-4: µ0 = 8.5) 2 1 1 1 2 = + = = 0. ∆ + = δ * − k = 0 − 0.02.128 = 0.05) = 0. so σ = 0. h = 4.3 UCL=0.2385 Cumulative Sum 0.0166 . h = 4.964 2(−0. and h=4.964 ARL+0 = ARL−0 ≅ ARL0 = 1/ 0.02.2385 (a) MTB > Stat > Control Charts > Time-Weighted Charts > CUSUM CUSUM Chart of Can Weight (Ex8-4can) target = 8. H = hσ = 4. σ = 0. σ = 0.936)] + 2(−0.3 2 4 6 8 10 12 14 Sample 16 18 20 22 24 There are no out-of-control signals. b = h + 1. (b) σˆ = MR2 1.5)(5.0186957 /1.166 = 5. ∆ − = −δ * − k = −0 − 0.936) − 1 = 742.Chapter 8 Exercise Solutions 8-4. k=1/2.2 0.936 δ * = 0.0027 + − ARL0 ARL0 ARL0 742.5. µ0 = 8. k = 1/ 2.1 0. k = 0.77.5.77.

4 -0. h = 8.176) − 1 = 741.25.02.4005 0.2 -0.1 0 0.6771 ARL+0 = ARL−0 ≅ ARL0 = 1 0.01 (0. 8-6 . h = 8. H = hσ = 8.166 = 9.25.01 + 1.02.6771 2(−0.25.25)(9.05.25 = −0.01.5 2 4 6 8 10 12 14 Sample 16 18 20 22 24 There are no out-of-control signals. h = 8.3 Cumulative Sum 0.25.05.25 exp[−2(−0.25)(9.4005 MTB > Stat > Control Charts > Time-Weighted Charts > CUSUM CUSUM Chart of Can Weight (Ex8-4can) target = 8.0 -0. k = 0.176 δ * = 0.25 = −0. k = 0. µ0 = 8.Chapter 8 Exercise Solutions 8-5.0027 + − ARL0 ARL0 ARL0 741.2 0.02.1 -0. ∆ − = −δ * − k = −0 − 0. k = 0. ∆ + = δ * − k = 0 − 0.3 LCL=-0.176)] + 2(−0. σ = 0.0027 = 370. σ = 0.84 The theoretical performance of these two CUSUM schemes is the same for Exercises 8-4 and 8-5.25) 2 1 1 1 2 = + = = 0.166 = 8. b = h + 1.4 0.01 UCL=0. In Exercise 8-5: µ0 = 8.4005 -0.01.05) = 0.

2 LCL=-0.5. FIR in # of standard deviations = h/2 = 4. σ = 0. target = 8.3 UCL=0.Chapter 8 Exercise Solutions 8-6. H = h σ = 4. µ0 = 8. Process was out of control at process start-up.0 0 -0.05.2385 -0.77/2 = 2.1 -0.1 0. h = 4.385 MTB > Stat > Control Charts > Time-Weighted Charts > CUSUM CUSUM Chart of Ex8-4can FIR = 2.05) = 0. k = 0. 8-7 . k = 1/2.2 0.77 0.77 (0.2385 FIR = H/2.77.385 std dev.00.00.2385 Cumulative Sum 0. h = 4.3 2 4 6 8 10 12 14 Sample 16 18 20 22 24 The process signals out of control at observation 20.

8 1 8 16 24 32 40 48 Sample 56 64 72 80 Test Results for CUSUM Chart of Ex8-7temp TEST.7215 /1. h = 5 75 UCL=60. One point beyond control limits. k = 1/ 2.128 = 12. (a) σˆ = MR2 d 2 = 13. h = 5 MTB > Stat > Control Charts > Time-Weighted Charts > CUSUM CUSUM Chart of Temperature Readings (Ex8-7temp) target = 950.8 Cumulative Sum 50 25 0 0 -25 -50 LCL=-60. k = 0. 8-8 .16. σˆ = 12. The assignable cause occurred after observation 12 – 10 = 2.5.Chapter 8 Exercise Solutions 8-7.16 (b) µ0 = 950. Test Failed at points: 12. 13 The process signals out of control at observation 12.

629 (from a Moving Range chart with CL = 6. 31. 30. Test Failed at points: 12.Chapter 8 Exercise Solutions 8-8.1 0 -100 -200 3 6 9 12 15 18 Sample 21 24 27 30 Test Results for CUSUM Chart of Ex8-8con TEST. 24. 22. 14. 29. (a) σˆ = MR2 d 2 = 6. 16. 18. 23.1 0 LCL=-28. h = 5 400 Cumulative Sum 300 200 100 UCL=28. 25.128 = 5. h = 5 MTB > Stat > Control Charts > Time-Weighted Charts > CUSUM CUSUM Chart of Bath Concentrations (Ex8-8con) target = 175. k = 1/ 2. 19. 28. 13. σˆ = 5. 32 The process signals out of control on the lower side at sample 3 and on the upper side at sample 12. k = 1/2. 21.35) (b) µ0 = 175.629. One point beyond control limits. 8-9 .629. 26. 17. 27. 20.35 /1. 15. std dev = 5. Assignable causes occurred after startup (sample 3 – 3 = 0) and after sample 8 (12 – 4).

to give longer in-control ARLs with shorter out-of-control ARLs. should be balanced by a larger control limit. 18 The process signals out of control on the lower side at sample 2 and on the upper side at sample 16.7 0 0 Cumulative Sum LCL=-47.01 MTB > Stat > Control Charts > Time-Weighted Charts > CUSUM CUSUM Chart of Viscosity Measurements (Ex8-9vis) target = 3200.25.Chapter 8 Exercise Solutions 8-9. Assignable causes occurred after startup (sample 2 – 2) and after sample 9 (16 – 7).71/1.25. 8-10 . h = 8. k = 0. (a) σˆ = MR2 d 2 = 6.949.7 -100 -200 -300 -400 -500 4 8 12 16 20 Sample 24 28 32 36 Test Results for CUSUM Chart of Ex8-9vis TEST.01 UCL=47.01.128 = 5. k = 0. Test Failed at points: 16. std dev = 5. σˆ = 5. 17.25. h = 8. One point beyond control limits.949. h = 8.71) (b) µ0 = 3200. (c) Selecting a smaller shift to detect.949 (from a Moving Range chart with CL = 6. k = 0.

h = 4. 42.4 4 8 12 16 20 24 Sample 28 32 36 40 44 Test Results for CUSUM Chart of Exm5-1x1...2504 MTB > Stat > Control Charts > Time-Weighted Charts > CUSUM CUSUM Chart of Flow Width Data (Exm5-1x1. 8-11 . µ0 = 1. Exm5-1x5) target = 1. Exm5-1x5 TEST. and remains above the upper limit.0626 δ = 1.2 LCL=-0.6 0. 41. ..8 0.0 Cumulative Sum 0. σ = 0. k = 0.50.0 -0.5. n = 5.14 5 = 0. 43. This CUSUM detects the shift in process mean earlier.0313. 44. at sample 40 versus sample 43. σ x = σ n = 0. h = 4 1. Test Failed at points: 40. K = kσ x = 0.250 0.50.14..5. 45 The CUSUM chart signals out of control at sample 40. The x -R chart shown in Figure 5-4 signals out of control at sample 43. H = hσ x = 0. k = δ 2 = 0.14. std dev = 0.Chapter 8 Exercise Solutions 8-10*..250 -0. .2 1. One point beyond control limits.4 UCL=0..2 0 0.

36 1050 0 -2.12 2.15 2 0.49 1008 -1.05 1169 4.349 ( ) Excel file: workbook Chap08.22 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 The process is out of control after observation 10 – 3 = 7.498 3 OOC 7 8.40 1128 3.822 0.61 1146 3. Vi = | yi | − 0.52 -0.68 1.2 -1.21 1095 1.52 5.52 4.N.54 12 OOC 7 36.738 2 5.04 3.26 1167 4.848 2 0 0 0.84 3.26 1139 3.94 3 0.55 7 OOC 7 21.44 5 OOC 7 14.13 1125 3 2.2 -1.57 1 1.07 1055 0.65 4 0 0 0 0 1.9 10 OOC 7 30.36 1087 1.90 1151 4.29 1064 0.Chapter 8 Exercise Solutions 8-11.5 5 Obs.56 -0.049 4 OOC 7 11.48 1.61 1163 4.84 3.84 one-sided upper cusum Si+ N+ OOC? When? 0 0 0 0 0 0 0 0 0 0.xls : worksheet Ex8-11 mu0 = sigma = delta = k= h= 1050 25 1 sigma 0.8 1.35 6 OOC 7 16.86 1 0.OOC? When? 0 0.74 1188 5. Process variability is increasing.76 3.56 3.88 13 OOC 7 one-sided lower cusum Si.66 9 OOC 7 26.07 1037 -0.989 1 1. i No FIR xi 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 yi vi 1045 -0.68 3.631 1 2.56 8 OOC 7 23.78 11 OOC 7 33.71 1238 7. 8-12 .50 1125 3 2.38 1146 3.52 3.

0000 -2.68 7.47 13 OOC 0 192 3.62 17.65 23.62 3. 8-13 .9540 1.27 23.2633 3.46 31 19 OOC 0 182 1.4869 2.56 10.04 18 OOC 0 185 1.32 28.6646 2.822 0.5528 3.18 15.55 16 OOC 0 190 2.OOC? When? 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1. Vi = | yi | − 0.5987 1.6294 (from Exercise 8-8) 1 sigma 0.65 20.7304 3.45 9 OOC 0 195 3.8 7 OOC 0 162 -2.68 6 OOC 0 158 -3.6646 2.56 11 OOC 0 184 1.36 20. Process variability has been increasing since start-up.349 ( ) Excel file : workbook Chap08.74 15 OOC 0 197 3.37 17 OOC 0 189 2.2435 0.946 2 150 -4.05 23 10 OOC 0 179 0.3093 2.00 19.9540 1.0199 2.129 3 OOC 0 151 -4.32 12 OOC 0 175 0.59 14 OOC 0 186 1.06 22.7764 1.5 5 one-sided upper cusum xi yi vi Si+ N+ OOC? When? 0 160 -2.N.0199 2.0199 2.4410 3.31 13 5 OOC 0 154 -3.86 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 … The process was last in control at period 2 – 2 = 0.xls : worksheet Ex8-12 mu0 = sigma = delta = k= h= i No FIR 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 175 5.84 31.7106 0.3 8 OOC 0 186 1.822 1 158 -3.32 1.Chapter 8 Exercise Solutions 8-12.31 26.16 30.9081 3.62 22.9081 3.34 20 OOC 0 one-sided lower cusum Si.19 4 OOC 0 153 -3.

3) 2 exp[−2(−0.3)(9.2)(9.005 = 215.166 = 8 + 1. 767 2(−0.040 = 25.Chapter 8 Exercise Solutions 8-13.7)(9.166 exp[−2(0.166 = 9.23 Out of control ARL Performance: δ * = 0.556 ARL0 = 1/ 0.7 b = h + 1.5 − 0.2 = −0.023 2(0.5 ∆ + = δ * − k = 0.02 8-14 .7) 2 1 1 1 1 1 = + = + = 0.5 − 0.166 exp[−2(−0.2 = 0.2) 2 1 1 1 2 = + = = 0.166)] + 2(0.166 = 8 + 1.166) − 1 ARL+1 = = 25.005 + − ARL0 ARL0 ARL0 430.556 2(−0.166)] + 2(−0.166 = 9. 767 ARL1 = 1/ 0.2 ∆ − = −δ * − k = −0 − 0. Standardized.166) − 1 ARL−1 = = 381.2 and h = 8 In control ARL performance: δ* = 0 ∆ + = δ * − k = 0 − 0.3)(9.166)] + 2(−0. two-sided cusum with k = 0.166) − 1 ARL+0 = ARL−0 ≅ = 430.3 ∆ − = −δ * − k = −0.040 + − ARL1 ARL1 ARL1 25.2)(9.023 381.2 b = h + 1.2 = −0.7)(9.2 = −0.

5. The upper CUSUM is used to detect upward shifts in the level of the process.Chapter 8 Exercise Solutions 8-14.762 MTB > Stat > Control Charts > Time-Weighted Charts > CUSUM CUSUM Chart of Viscosity Measurements (Ex8-9vis) target = 3150 1200 Cumulative Sum 1000 800 600 400 200 UCL=30 0 LCL=-30 0 4 8 12 16 20 Sample 24 28 32 36 MINITAB displays both the upper and lower sides of a CUSUM chart on the same graph. 8-15 .95238. H = hs = 5(5. The assignable cause occurred at start-up (2 – 2). there is no option to display a single-sided chart. s = 5. µ0 = 3150.5 (5. The process signals out of control on the upper side at sample 2. h = 5 K = ks = 0.95238) = 2.95238) = 29. k = 0.976.

35 H = hσˆ = 5(108. of either a trend/drift or a shit in measurements.6) µ0 = 734. σˆ = MR2 d 2 = 122.5. k = 0.5 4000 Cumulative Sum 3000 2000 1000 UCL=544 0 0 LCL=-544 4 8 12 16 20 24 Sample 28 32 36 40 The Individuals I-MR chart. displayed a distinct downward trend in measurements. 8-16 . The CUSUM chart reflects a consistent run above the target value 734.5 MTB > Stat > Control Charts > Time-Weighted Charts > CUSUM CUSUM Chart of Light Velocity (Ex5-60Vel) target = 734. There is a distinct signal on both charts.6 /1.7) = 543.5.7) = 54. starting at about sample 18. The outof-control signals should lead us to investigate and determine the assignable cause.5(108.128 = 108. h = 5 K = kσˆ = 0. from virtually the first sample.Chapter 8 Exercise Solutions 8-15☺.7 (from a Moving Range chart with CL = 122. with a centerline at x = 909 .5.

5.8 750 _ _ X=734.7 MTB > Stat > Control Charts > Time-Weighted Charts > EWMA EWMA Chart of Light Velocity (Ex5-60Vel) lambda = 0. L = 2. CL = µ0 = 734.7. λ = 0.1. and also indicates the slight downward trend starting at about sample 22.1.7SL=801. L = 2. σ = 108.5 700 -2.7 900 EWMA 850 800 +2.5.2 4 8 12 16 20 24 Sample 28 32 36 40 The EWMA chart reflects the consistent trend above the target value.7SL=667.Chapter 8 Exercise Solutions 8-16☺. 734. 8-17 .

L = 3 limits = µ0 ± Lσ λ (2 − λ ) = 10 ± 3(1) 0.1) = [9.69] (b) λ = 0. 8-18 (8-16).5] As λ increases. L = 2.1 (2 − 0.7 1140 1120 EWMA 1100 1080 +2.1.11. λ = 0.11] (c) λ = 0.4 _ _ X=1050 1060 1040 -2.51 MTB > Stat > Control Charts > Time-Weighted Charts > EWMA EWMA Chart of Molecular Weight (Ex8-1mole) lambda = 0. UCL = 1065.Chapter 8 Exercise Solutions 8-17 (8-15). LCL = 1034.5. L = 3 limits = µ0 ± Lσ λ (2 − λ ) = 10 ± 3(1) 0. 8-18 .4. the same as the CUSUM chart.31. σ = 25.1.1. CL = µ0 = 1050. the width of the control limits also increases.7.7SL=1034.10. L = 3 limits = µ0 ± Lσ λ (2 − λ ) = 10 ± 3(1) 0. (a) λ = 0.2 (2 − 0.2) = [9.6 2 4 6 8 10 12 Sample 14 16 18 20 Process exceeds upper control limit at sample 10.4) = [8.49.2.7SL=1065. L = 2.4 (2 − 0.

02 8.9700 7. LCL = 7.07.0700 8.2.Chapter 8 Exercise Solutions 8-19 (8-17).02 8.04 _ _ X=8.96 2 4 6 8 10 12 14 Sample 16 18 20 22 24 The process is in control.05.98 LCL=7.97 MTB > Stat > Control Charts > Time-Weighted Charts > EWMA EWMA Chart of Can Weight (Ex8-4can) lambda = 0.00 7.2. Assume σ = 0.08 UCL=8. L = 3.06 EWMA 8. L = 3 8. λ = 0. CL = µ0 = 8.02. 8-19 . UCL = 8.

99 -2.05087 8.99 MTB > Stat > Control Charts > Time-Weighted Charts > EWMA EWMA Chart of Can Weight (Ex8-4can) lambda = 0.03 _ _ X=8. CL = µ0 = 8. λ = 0.1. L = 2. LCL = 7.05 8.7 +2.00 7.04 EWMA 8.7SL=7.05. Assume σ = 0. There is not much difference between the control charts.05.02 8.1.Chapter 8 Exercise Solutions 8-20 (8-18). L = 2.7SL=8.98913 2 4 6 8 10 12 14 Sample 16 18 20 22 24 The process is in control. 8-20 .02.01 8.02 8.7. UCL = 8.

8-21 . LCL = 942.7 960 +2. UCL = 957. 12 and 13. L = 2.7SL=957.1. Test Failed at points: 12.7. L = 2. but not flagged on chart).7SL=942.53 EWMA 955 _ _ X=950 950 945 -2.47 1 8 16 24 32 40 48 Sample 56 64 72 80 Test Results for EWMA Chart of Ex8-7temp TEST. MTB > Stat > Control Charts > Time-Weighted Charts > EWMA EWMA Chart of Temperature Readings (Ex8-7temp) lambda = 0.16 .1.53. λ = 0. CL = µ0 = 950. One point beyond control limits. 13 Process is out of control at samples 8 (beyond upper limit.47. σˆ = 12.Chapter 8 Exercise Solutions 8-21 (8-19).

λ = 0.76. as compared to the chart in the Exercise 21 (with the smaller λ) which signaled out of control at earlier samples.16 .24. Test Failed at points: 70 With the larger λ. UCL = 968. σˆ = 12. L = 3 970 UCL=968.4. LCL = 931. MTB > Stat > Control Charts > Time-Weighted Charts > EWMA EWMA Chart of Temperature Readings ( Ex8-7temp) lambda = 0.Chapter 8 Exercise Solutions 8-22 (8-20).76 930 1 8 16 24 32 40 48 Sample 56 64 72 80 Test Results for EWMA Chart of Ex8-7temp TEST. 8-22 .24 EWMA 960 _ _ X=950 950 940 LCL=931. CL = µ0 = 950. One point beyond control limits.4. the process is out of control at observation 70. L = 3.

L = 2. CL = µ0 = 175.05.634 .6 190 EWMA 185 180 +2.6.6SL=172.Chapter 8 Exercise Solutions 8-23 (8-21). 8-23 .05. σˆ = 5.6SL=177. MTB > Stat > Control Charts > Time-Weighted Charts > EWMA EWMA Chart of Bath Concentrations (Ex8-8con) lambda = 0. The process average of µˆ = 183.70 170 3 6 9 12 15 18 Sample 21 24 27 30 Process is out of control.594 is too far from the process target of µ0 = 175 for the process variability.30 _ _ X=175 175 -2. LCL = 172.30. The data is grouped into three increasing levels. λ = 0.70. L = 2. UCL = 177.

25 12.b. 8-24 .7 20.5 10 r .9 26 .7SL=18.0 7.ecJa F e F e M M a A M Ma Ma J Ju Ju Ju Ju O N N D D S S Ex6-62Day In Exercise 6-62.7 MTB > Stat > Control Charts > Time-Weighted Charts > EWMA EWMA Chart of Homicide Data (Ex6-62Bet) lambda = 0.ep.1. L = 2. a properly designed EWMA chart is very robust to the assumption of normally distributed data.0 +2.and 0. L = 2. As mentioned in the text (Section 8.1 t.0 20 23 25 r .7 16 16 22 25 l -6 l.Oc Oc O c ct.2777th.5 10.5 -2.ec.4 t.n.y .n .4-3).Ju Ju Ju Ju l.23 5.b.1. Individuals control charts of 0.7 24 28 .n.ov o v .9 22 24 t.27 17. λ = 0. The EWMA chart also does not signal out of control.7SL=6.p ay y .2 25 28 29 n.un n.8 l.5 EWMA 15.25th-root transformed data showed no out-of-control signals.Sep ep .0 _ _ X=12.8 19 .4 .Chapter 8 Exercise Solutions 8-24☺.a r .

95 (from Exercise 8-9).1.Chapter 8 Exercise Solutions 8-25 (8-22).7SL=3203.95. µ0 = 3200. lambda=0. sigma=5. L = 2.32 EWMA 3195 3190 3185 3180 3175 4 8 12 16 20 24 Ex5-60Meas 28 32 36 The process is out of control from the first sample.1.68 _ _ X=3200 3200 -2. σˆ = 5.7SL=3196. λ = 0.7 3205 +2. 8-25 . L=2.7 MTB > Stat > Control Charts > Time-Weighted Charts > EWMA EWMA Chart of Viscosity (Ex8-9vis) Target=3200.

13. 12. 17. µ0 = 1050. Test Failed at points: 10. 14. UCL = 1080. One point beyond control limits. CL = 1050. σ = 25.Chapter 8 Exercise Solutions 8-26 (8-23). 11.6. target value = 1050. 20 Process is out of control at observation 10.4 1000 2 4 6 8 10 12 Sample 14 16 18 20 Test Results for Moving Average Chart of Ex8-1mole TEST. LCL = 1019. 16. w = 6. 8-26 .4 MTB > Stat > Control Charts > Time-Weighted Charts > Moving Average Moving Average Chart of Molecular Weight (Ex8-1mole) w = 6. 15. 18.6 _ _ X=1050 1050 LCL=1019. std dev = 25 1200 Moving Average 1150 1100 UCL=1080. 19. the same result as for Exercise 8-1.

Chapter 8 Exercise Solutions

8-27 (24).
w = 5, µ0 = 8.02, σ = 0.05, CL = 8.02, UCL = 8.087, LCL = 7.953
MTB > Stat > Control Charts > Time-Weighted Charts > Moving Average

Moving Average Chart of Can Weight (Ex8-4can)
w = 5, process target = 8.02, std dev = 0.05
8.20

Moving Average

8.15
8.10

UCL=8.0871

8.05

_
_
X=8.02

8.00

LCL=7.9529

7.95
7.90

2

4

6

8

10

12 14
Sample

16

18

20

22

24

The process is in control, the same result as for Exercise 8-4.

8-27

Chapter 8 Exercise Solutions

8-28☺.
w=5
MTB > Stat > Control Charts > Time-Weighted Charts > Moving Average

Moving Average Chart of Homicide Data (Ex6-62Bet)
w = 5, target and std dev estimated from data
UCL=25.31

25

Moving Average

20
15

_
_
X=12.25

10
5
0

LCL=-0.81
3

6

9

12

15
Sample

18

21

24

27

Because these plot points are an average of five observations, the nonnormality of the
individual observations should be of less concern. The approximate normality of the
averages is a consequence of the Central Limit Theorem.

8-28

Chapter 8 Exercise Solutions

8-29 (8-25).
Assume that t is so large that the starting value Z 0 = x has no effect.

⎡ ∞

E ( Z t ) = E[λ xt + (1 − λ )( Z t −1 )] = E ⎢λ ∑ (1 − λ ) j xt − j ⎥ = λ ∑ (1 − λ ) j E ( xt − j )
j =0
⎣ j =0

Since E ( xt − j ) = µ and λ ∑ (1-λ ) j = 1 , E ( Z t ) = µ
j =0

8-30 (8-26).
⎡ ∞

var( Z t ) = var ⎢ λ ∑ (1 − λ ) j xt − j ⎥
⎣ j =0

= ⎢λ 2 ∑ (1 − λ ) 2 j ⎥ ⎡⎣ var( xt − j ) ⎤⎦
⎣ j =0

λ ⎛σ 2 ⎞
=

2−λ ⎝ n ⎠

8-31 (8-27).
For the EWMA chart, the steady-state control limits are x ± 3σ

λ
.
(2 − λ )n

⎛ 2 ⎞

⎝ w + 1 ⎠ = x ± 3σ 1 = x ± 3σ ,
Substituting λ = 2/(w + 1), x ± 3σ
2 ⎞
wn

wn
⎜2−
⎟n
w +1 ⎠

which are the same as the limits for the MA chart.

8-32 (8-28).

1 w−1
w −1
. In the
∑ j=
w j =0
2
EWMA, the weight given to a sample mean j periods ago is λ(1 - λ)j , so the average age

1− λ
. By equating average ages:
is λ ∑ (1 − λ ) j j =

The average age of the data in a w-period moving average is

j =0

λ

1− λ

w −1
λ
2
2
λ=
w +1
=

8-29

Chapter 8 Exercise Solutions

8-33 (8-29).
For n > 1, Control limits = µ0 ±

3 ⎛ σ ⎞

⎟ = µ0 ±
w⎝ n⎠
wn

8-34 (8-30).
x chart: CL = 10, UCL = 16, LCL = 4
UCL = CL + kσ x
16 = 10 − kσ x

kσ x = 6
EWMA chart:
UCL = CL + lσ λ [(2 − λ )n]
= CL + l σ

n 0.1 (2 − 0.1) = 10 + 6(0.2294) = 11.3765

LCL = 10 − 6(0.2294) = 8.6236

8-35 (8-31).
λ = 0.4
For EWMA, steady-state limits are ± Lσ λ (2 − λ )
For Shewhart, steady-state limits are ± kσ

kσ = Lσ λ (2 − λ )
k = L 0.4 (2 − 0.4)
k = 0.5 L

8-30

Chapter 8 Exercise Solutions

8-36 (8-32).
The two alternatives to plot a CUSUM chart with transformed data are:
1. Transform the data, target (if given), and standard deviation (if given), then use
these results in the CUSUM Chart dialog box, or
2. Transform the target (if given) and standard deviation (if given), then use the
Box-Cox tab under CUSUM Options to transform the data.
The solution below uses alternative #2.
From Example 6-6, transform time-between-failures (Y) data to approximately normal
distribution with X = Y 0.2777.
TY = 700, TX = 700 0.2777 = 6.167, k = 0.5, h = 5
MTB > Stat > Control Charts > Time-Weighted Charts > CUSUM

CUSUM Chart of Transformed Failure Data (Ex8-37trans)
X = Y^0.277, target - 6.167, k = 0.5, h = 5
UCL=10.46

Cumulative Sum

10

5

0

0

-5

-10

LCL=-10.46
2

4

6

8

10
12
Sample

14

16

18

20

A one-sided lower CUSUM is needed to detect an increase in failure rate, or equivalently
a decrease in the time-between-failures. Evaluate the lower CUSUM on the MINITAB
chart to assess stability.
The process is in control.

8-31

Chapter 8 Exercise Solutions

8-37 (8-33).

µ0 = 700, h = 5, k = 0.5, estimate σ using the average moving range
MTB > Stat > Control Charts > Time-Weighted Charts > CUSUM,
also CUSUM options > Estimate > Average Moving Range

CUSUM Chart of Valve Failure Data (Ex8-37fail)
Target=700, h=5, k=0.5
4000
UCL=3530
3000

Cumulative Sum

2000
1000
0

0
-1000
-2000
-3000

LCL=-3530

-4000
2

4

6

8

10
12
Ex8-37No

14

16

18

20

A one-sided lower CUSUM is needed to detect an increase in failure rate. Evaluate the
lower CUSUM on the MINITAB chart to assess stability.
The process is in control.
Though the data are not normal, the CUSUM works fairly well for monitoring the
process; this chart is very similar to the one constructed with the transformed data.

8-32

Chapter 8 Exercise Solutions

8-38 (8-34).
µ0 = TX = 700 0.2777 = 6.167, λ = 0.1, L = 2.7
MTB > Stat > Control Charts > Time-Weighted Charts > EWMA

EWMA Chart of Transformed Failure Data (Ex8-37trans)
X = Y^0.2777, target = 6.167, lambda = 0.1, L = 2.7
7.5

+2.7SL=7.453

7.0

EWMA

6.5

_
_
X=6.167

6.0
5.5
5.0

-2.7SL=4.881
2

4

6

8

10
12
Sample

14

16

18

20

Valve failure times are in control.

8-39 (8-35).
The standard (two-sided) EWMA can be modified to form a one-sided statistic in much
the same way a CUSUM can be made into a one-sided statistic. The standard (two-sided)
EWMA is
zi = λ xi + (1 − λ ) zi−1
Assume that the center line is at µ0. Then a one-sided upper EWMA is
zi+ = max ⎡⎣ µ0 , λ xi + (1 − λ ) zi −1 ⎤⎦ ,
and the one-sided lower EWMA is
zi− = min ⎡⎣ µ0 , λ xi + (1 − λ ) zi −1 ⎤⎦ .

8-33

Chapter 9 Exercise Solutions
Note: Many of the exercises in this chapter were solved using Microsoft Excel 2002, not
MINITAB. The solutions, with formulas, charts, etc., are in Chap09.xls.
9-1.
σˆ A = 2.530, nA = 15, µˆ A = 101.40
σˆ B = 2.297, nB = 9, µˆ B = 60.444
σˆ C = 1.815, nC = 18, µˆ C = 75.333

σˆ D = 1.875, nD = 18, µˆ D = 50.111
Standard deviations are approximately the same, so the DNOM chart can be used.
R = 3.8, σˆ = 2.245, n = 3
x chart: CL = 0.55, UCL = 4.44, LCL = −3.34
R chart: CL = 3.8, UCL = D4 R = 2.574 (3.8) = 9.78, LCL = 0
Stat > Control Charts > Variables Charts for Subgroups > Xbar-R Chart

Xbar-R Chart of Measurements (Ex9-1Xi)
U C L=4.438

Sample M ean

4
2

_
_
X=0.55

0
-2

LC L=-3.338

-4
2

4

6

8

10
Sample

12

14

16

18

20

Sample Range

10.0

U C L=9.78

7.5
5.0

_
R=3.8

2.5
0.0

LC L=0
2

4

6

8

10
Sample

12

14

16

18

20

Process is in control, with no samples beyond the control limits or unusual plot patterns.

9-1

Chapter 9 Exercise Solutions
9-2.
Since the standard deviations are not the same, use a standardized x and R charts.
Calculations for standardized values are in:
Excel : workbook Chap09.xls : worksheet : Ex9-2.

n = 4, D3 = 0, D4 = 2.282, A2 = 0.729; RA = 19.3, RB = 44.8, RC = 278.2
Graph > Time Series Plot > Simple
Control Chart of Standardized Xbar (Ex9-2Xsi)
1.5

1.0

Ex9-2Xsi

+A2 = 0.729

0.5

0

0.0

-0.5
-A2 = -0.729

-1.0
Ex9-2Samp
Ex9-2Part

2
A

4
A

6
A

8
B

10
B

12
C

14
C

16
C

18
C

20
C

Control Chart of Standardized R (Ex9-2Rsi)
2.5
D4 = 2.282

Ex9-2Rsi

2.0

1.5

1.006

1.0

0.5

0.0

Ex9-2Samp
Ex9-2Part

D3 = 0

2
A

4
A

6
A

8
B

10
B

12
C

14
C

16
C

18
C

20
C

Process is out of control at Sample 16 on the x chart.

9-2

c1130 = 64. c1385 = 26. c8611 = 50.63 4.67. 9-4. Note: In the textbook. In a short production run situation. a standardized CUSUM could be used to detect smaller deviations from the target value. Set up a standardized c chart for defect counts.25.67 50. j − µ0. 9-3 .25 64.63. j ) / σ j (where j represents the part type) would be used to calculate each plot statistic. Stat > Basic Statistics > Display Descriptive Statistics Descriptive Statistics: Rx9-4Def Rx9-4Def 1055 1130 1261 1385 4610 8611 13. with CL = 0.13 Stat > Control Charts > Variables Charts for Individuals > Individuals I Chart of Standardized Total # of Defects (Ex9-4Zi) 3 UCL=3 Individual Value 2 1 _ X=0 0 -1 -2 -3 LCL=-3 4 8 12 16 20 24 Observation 28 32 36 40 Process is in control.13 c1055 = 13. The chart would be designed so that δ. LCL = −3. c1261 = 12.67. The plot statistic is Z i = ( ci − c ) c. UCL = +3. c4610 = 4.67 26. The standardized variable ( yi . the 4th part on Day 246 should be “1385” not “1395”. in standard deviation units.00 12. is the same for each part type.Chapter 9 Exercise Solutions 9-3.00.

574 55.00 59.00 45.Chapter 9 Exercise Solutions 9-5.xls : Worksheet Ex9-5 Grand Avg = Avg R = s= n= A2 = D3 = D4 = Xbar UCL = Xbar LCL = R UCL = R LCL = 52. not just a specific one. so the process is out-of-control.988 2.379 50.596 6. Excel : Workbook Chap09.00 Ex9-5Xbmin 53.00 Ex9-5XbUCL 51. There are many values of x max and x min that are outside the control limits.023 0 2.00 57. 9-4 .000 Group Xbar Control Chart 61.00 Ex9-5XbLCL 49. The assignable cause affects all heads.00 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Sam ple Group Range Control Chart 8 Range 6 Ex9-5Rmax Ex9-5RUCL 4 Ex9-5RLCL 2 0 1 3 5 7 9 11 13 15 17 19 Sam ple There is no situation where one single head gives the maximum or minimum value of x six times in a row.00 47.338 4 heads 3 units 1.00 Ex9-5Xbmax Xbar 55.017 0.

xls : Worksheet Ex9-6 Group Control Chart for Xbar 65 60 Xbar Ex9-5Xbmax Ex9-5Xbmin 55 Ex9-5XbUCL Ex9-5XbLCL 50 45 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 Sample Group Control Chart for Range 7 6 Range 5 Ex9-5Rmax 4 Ex9-5RUCL 3 Ex9-5RLCL 2 1 0 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 Sample The last four samples from Head 4 are the maximum of all heads. 9-5 . Excel : Workbook Chap09.Chapter 9 Exercise Solutions 9-6. a process change may have caused output of this head to be different from the others.

158 4 heads 2 units 1.Chapter 9 Exercise Solutions 9-7. 9-6 . 70 Individ. Obs. 6 5 Ex9-7aMRmax 4 Ex9-7aMRUCL 3 Ex9-7aMRLCL 2 1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Sample See the discussion in Exercise 9-5.988 2. 65 Ex9-7aXmax 60 Ex9-7aXmin 55 Ex9-7aXbUCL 50 Ex9-7aXbLCL 45 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Sample Group Control Chart for Moving Range 8 7 Individ.xls : Worksheet Ex9-7A Grand Avg = Avg MR = s= n= d2 = D3 = D4 = Xbar UCL = Xbar LCL = R UCL = R LCL = 52. (a) Excel : Workbook Chap09.727 47.248 7.128 0 3.267 58.000 Group Control Chart for Individual Obs.050 0. Obs.

xls : Worksheet Ex9-7b Grand Avg = Avg MR = s= n= d2 = D3 = D4 = Xbar UCL = Xbar LCL = R UCL = R LCL = 52.248 7.727 47. indicating a potential process change.000 Group Control Chart for Individual Obs. 65 Individ. 9-7 .267 58.128 0 3.158 4 heads 2 units 1. 60 Ex9-7bXmax Ex9-7bXmin 55 Ex9-7bXbUCL Ex9-7bXbLCL 50 45 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 Sample Group Control Chart for Moving Range 8 7 MR 6 5 Ex9-7bMRmax 4 Ex9-7bMRUCL 3 Ex9-7bMRLCL 2 1 0 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 Sample The last four samples from Head 4 remain the maximum of all heads. Obs.050 0.988 2.Chapter 9 Exercise Solutions 9-7 continued (b) Excel : Workbook Chap09.

Chapter 9 Exercise Solutions 9-7 continued (c) Stat > Control Charts > Variables Charts for Subgroups > Xbar-S Chart Note: Use “Sbar” as the method for estimating standard deviation.816 2 4 6 8 10 Sample 12 14 16 18 20 U C L=4.415 Sample StDev 4 3 _ S =1.988 52 50 LC L=49. . Ex9-7X4) U C L=56. 9-8 .948 2 1 0 LC L=0 2 4 6 8 10 Sample 12 14 16 18 20 Failure to recognize the multiple stream nature of the process had led to control charts that fail to identify the out-of-control conditions in this process.159 Sample M ean 56 54 _ _ X=52.. Xbar-S Chart of Head Measurements (Ex9-7X1...

6 2..0 LC L=0 3 6 9 12 15 Sample 18 21 24 27 30 Test Results for S Chart of Ex9-7X1.00 standard deviations from center line.. Ex9-7X4) U C L=56. Ex9-7X4 TEST 1.948 1. Test Failed at points: 27.415 3.2 0..816 3 6 9 12 15 Sample 18 21 24 Sample StDev 4.4 _ S =1. . .Chapter 9 Exercise Solutions 9-7 continued (d) Stat > Control Charts > Variables Charts for Subgroups > Xbar-S Chart Note: Use “Sbar” as the method for estimating standard deviation. 9-9 . Xbar-S Chart of Head Measurements (Ex9-7X1...988 52 50 LC L=49..159 Sample M ean 56 54 _ _ X=52. 29 Only the S chart gives any indication of out-of-control process. One point more than 3.8 27 1 30 1 U C L=4.

. as shown on the R chart is in control.. σˆ = R / d 2 = 0.55025. . Stat > Basic Statistics > Display Descriptive Statistics Descriptive Statistics: Ex9-8Xbar.001 LCL=0 0.000 2 4 6 8 10 12 Sample 14 16 18 20 The process variability. 9-10 .000976)] = 6.00227 / 2. Ex9-8Rdum4) 0.003 _ R=0.00227 0. Ex9-8R Variable Ex9-8Xbar Ex9-8R Mean 0.002270 n=5 x = 0.552 − 0.55025 0.548 ) [6(0.005 UCL=0.000976 n = ( USL-LSL ) 6σˆ = ( 0.Chapter 9 Exercise Solutions 9-8.004 0.00227.004800 Sample Range 0.326 = 0.83 PCR Stat > Control Charts > Variables Charts for Subgroups > R Chart R Chart of Range Values ( Ex9-8R.002 0. R = 0..

Chapter 9 Exercise Solutions

9-8 continued
(a)
3-sigma limits:
δ = 0.01, Zδ = Z 0.01 = 2.33

(
LCL = LSL + ( Z

UCL = USL − Zδ − 3
δ

−3

)
(
n ) σˆ = (0.550 − 0.020) + ( 2.33 − 3

n σˆ = (0.550 + 0.020) − 2.33 − 3

)
20 ) (0.000976) = 0.5316

20 (0.000976) = 0.5684

Graph > Time Series Plot > Simple

Note: Reference lines have been used set to the control limit values.
Control Chart of Xbar Values (Ex9-8Xbar)
0.57

UCL = 0.5684

Ex9-8Xbar

0.56

0.55

0.54

LCL = 0.5316

0.53
2

4

6

8

10
12
Ex9-8Samp

14

16

18

20

The process mean falls within the limits that define 1% fraction nonconforming.
Notice that the control chart does not have a centerline. Since this type of control scheme
allows the process mean to vary over the interval—with the assumption that the overall
process performance is not appreciably affected—a centerline is not needed.

9-11

Chapter 9 Exercise Solutions

9-8 continued
(b)
γ = 0.01, Zγ = Z 0.01 = 2.33
1 − β = 0.90, Z β = z0.10 = 1.28

(
LCL = LSL + ( Z

UCL = USL − Zγ + Z β
γ

+ Zβ

)
(
n ) σˆ = (0.550 − 0.020) + ( 2.33 + 1.28

n σˆ = (0.550 + 0.020) − 2.33 + 1.28

)
20 ) (0.000976) = 0.5326

20 (0.000976) = 0.5674

Chart control limits for part (b) are slightly narrower than for part (a).
Graph > Time Series Plot > Simple

Note: Reference lines have been used set to the control limit values.
Control Chart of Xbar Values (Ex9-8Xbar)
0.57
UCL = 0.5674

Ex9-8Xbar

0.56

0.55

0.54

LCL = 0.5326
0.53
2

4

6

8

10
12
Ex9-8Samp

14

16

18

20

The process mean falls within the limits defined by 0.90 probability of detecting a 1%
fraction nonconforming.

9-12

Chapter 9 Exercise Solutions

9-9.
(a)
3-sigma limits:
n = 5, δ = 0.001, Zδ = Z 0.001 = 3.090
USL = 40 + 8 = 48, LSL = 40 − 8 = 32

(

UCL = USL − Zδ − 3

(

)

(

n σ = 48 − 3.090 − 3

)

(

n σ = 32 + 3.090 − 3

LCL = LSL+ Zδ − 3

)

5 (2.0) = 44.503

)

5 (2.0) = 35.497

Graph > Time Series Plot > Simple

Note: Reference lines have been used set to the control limit values.
Modified Control Chart of Xbar Values (Ex9-9Xbar)
3-sigma Control Limits
45.0

UCL = 44.5

Ex9-9Xbar

42.5

40

40.0

37.5

LCL = 35.5

35.0
2

4

6

8

10
12
Ex9-9Samp

14

16

18

20

Process is out of control at sample #6.

9-13

Chapter 9 Exercise Solutions

9-9 continued
(b)
2-sigma limits:
UCL = USL − Zδ − 2

(

(

LCL = LSL+ Zδ − 2

)

(

n σ = 48 − 3.090 − 2

)

(

n σ = 32 + 3.090 − 2

)

5 (2.0) = 43.609

)

5 (2.0) = 36.391

Graph > Time Series Plot > Simple

Note: Reference lines have been used set to the control limit values.
Modified Control Chart of Xbar Values (Ex9-9Xbar)
2-sigma Control Limits
45
44

UCL = 43.61

43
Ex9-9Xbar

42
41
40

40
39
38
37

LCL = 36.39

36
2

4

6

8

10
12
Ex9-9Samp

14

16

18

20

With 3-sigma limits, sample #6 exceeds the UCL, while with 2-sigma limits both samples
#6 and #10 exceed the UCL.

9-14

Chapter 9 Exercise Solutions

9-9 continued
(c)
γ = 0.05, Zγ = Z 0.05 = 1.645
1 − β = 0.95, Z β = Z 0.05 = 1.645

(
LCL = LSL + ( Z

UCL = USL − Zγ + zβ
γ

+ zβ

5 ) (2.0) = 43.239
)
(
n ) σ = 32 + (1.645 + 1.645 5 ) (2.0) = 36.761
n σ = 48 − 1.645 + 1.645

Graph > Time Series Plot > Simple

Note: Reference lines have been used set to the control limit values.
Acceptance Control Chart of Xbar Values (Ex9-9Xbar)
45
44
UCL = 43.24

43

Ex9-9Xbar

42
41
40

40

39
38
37

LCL = 36.76

36
2

4

6

8

10
12
Ex9-9Samp

14

16

18

20

Sample #18 also signals an out-of-control condition.

9-15

Chapter 9 Exercise Solutions

9-10.
Design an acceptance control chart.
Accept in-control fraction nonconforming = 0.1% → δ = 0.001, Zδ = Z 0.001 = 3.090
with probability 1 − α = 0.95 → α = 0.05, Zα = Z 0.05 = 1.645
Reject at fraction nonconforming = 2% → γ = 0.02, Zγ = Z 0.02 = 2.054
with probability 1 − β = 0.90 → β = 0.10, Z β = Z 0.10 = 1.282
2

⎛ Z + Z β ⎞ ⎛ 1.645 + 1.282 ⎞ 2
n=⎜ α
=
= 7.98 ≈ 8
⎜ Z − Z ⎟⎟ ⎜⎝ 3.090 − 2.054 ⎟⎠
δ
γ

(
LCL = LSL + ( Z

UCL = USL − Zγ + Z β
γ

+ Zβ

8 ) σ = USL − 2.507σ
)
(
n ) σ = LSL + ( 2.054 + 1.282 8 ) σ = LSL + 2.507σ
n σ = USL − 2.054 + 1.282

9-16

Chapter 9 Exercise Solutions

9-11.
µ = 0, σ = 1.0, n = 5, δ = 0.00135, Zδ = Z0.00135 = 3.00
For 3-sigma limits, Zα = 3

(

UCL = USL − zδ − zα

)

(

n σ = USL − 3.000 − 3

)

5 (1.0) = USL − 1.658

⎛ USL − 1.658 − µ0 ⎞
⎛ UCL − µ0 ⎞
Pr{Accept} = Pr{x < UCL} = Φ ⎜
=
Φ

⎟ = Φ (∆ − 1.658) 5

σ
n
1.0
5

where ∆ = USL − µ0

(

(

For 2-sigma limits, Zα = 2 ⇒ Pr{Accept} = Φ (∆ − 2.106) 5

)

)

⎛ USL − µ0 ⎞
p = Pr{x > USL} = 1 − Pr{x ≤ USL} = 1 − Φ ⎜
⎟ = 1 − Φ (∆ )
σ

Excel : Workbook Chap09.xls : Worksheet Ex9-11
DELTA=USL-mu0
3.50
3.25
3.00
2.50
2.25
2.00
1.75
1.50
1.00
0.50
0.25
0.00

CumNorm(DELTA)
0.9998
0.9994
0.9987
0.9938
0.9878
0.9772
0.9599
0.9332
0.8413
0.6915
0.5987
0.5000

p
0.0002
0.0006
0.0013
0.0062
0.0122
0.0228
0.0401
0.0668
0.1587
0.3085
0.4013
0.5000

Pr(Accept@3)
1.0000
0.9998
0.9987
0.9701
0.9072
0.7778
0.5815
0.3619
0.0706
0.0048
0.0008
0.0001

Pr(Accept@2)
0.9991
0.9947
0.9772
0.8108
0.6263
0.4063
0.2130
0.0877
0.0067
0.0002
0.0000
0.0000

Operating Curves

Pr{Acceptance}

1.0000
0.8000
0.6000
0.4000
0.2000
0.0000
0.0000

0.1000

0.2000

0.3000

0.4000

0.5000

Fraction Defective, p
Pr(Accept@3)

Pr(Accept@2)

9-17

Chapter 9 Exercise Solutions

9-12.
Design a modified control chart.
n = 8, USL = 8.01, LSL = 7.99, S = 0.001
δ = 0.00135, Zδ = Z0.00135 = 3.000
For 3-sigma control limits, Zα = 3

(

UCL = USL − Zδ − Zα

(

LCL = LSL+ Zδ − Zα

)

(

n σ = 8.01 − 3.000 − 3

)

(

n σ = 7.99 + 3.000 − 3

)

8 (0.001) = 8.008

)

8 (0.001) = 7.992

9-13.
Design a modified control chart.
n = 4, USL = 70, LSL = 30, S = 4
δ = 0.01, Zδ = 2.326
1 − α = 0.995, α = 0.005, Zα = 2.576

(
LCL = LSL + ( Z

UCL = USL − Zδ − Zα
δ

− Zα

4 ) (4) = 65.848
)
(
n ) σ = (50 − 20) + ( 2.326 − 2.576 4 ) (4) = 34.152
n σ = (50 + 20) − 2.326 − 2.576

9-14.
Design a modified control chart.
n = 4, USL = 820, LSL = 780, S = 4
δ = 0.01, Zδ = 2.326
1 − α = 0.90, α = 0.10, Zα = 1.282

(
LCL = LSL + ( Z

UCL = USL − Zδ − Zα
δ

− Zα

4 ) (4) = 813.26
)
(
n ) σ = (800 − 20) + ( 2.326 − 1.282 4 ) (4) = 786.74
n σ = (800 + 20) − 2.326 − 1.282

9-18

Chapter 9 Exercise Solutions

9-15.
n = 4, R = 8.236, x = 620.00
(a)
σˆ x = R d 2 = 8.236 2.059 = 4.000
(b)
pˆ = Pr{x < LSL} + Pr{x > USL}
= Pr{x < 595} + [1 − Pr{x ≤ 625}]
⎛ 595 − 620 ⎞ ⎡
⎛ 625 − 620 ⎞ ⎤
= Φ⎜
⎟ + ⎢1 − Φ ⎜
⎟⎥
⎝ 4.000 ⎠ ⎣
⎝ 4.000 ⎠ ⎦
= 0.0000 + [1 − 0.8944]
= 0.1056
(c)
δ = 0.005, Zδ = Z 0.005 = 2.576

α = 0.01, Zα = Z 0.01 = 2.326

(
LCL = LSL + ( Z

UCL = USL − Zδ − Zα
δ

− Zα

4 ) 4 = 619.35
)
(
n ) σ = 595 + ( 2.576 − 2.326 4 ) 4 = 600.65
n σ = 625 − 2.576 − 2.326

9-19

84 48.0 -0.16 49.105969 0.8 Autocorrelation 0.4 0.Chapter 9 Exercise Solutions 9-16.6 -0.74 49. the 5th row should be “2000” not “2006”.220536 0.8 0.2 0. the 5th column.373245 0.0 Partial Autocorrelation 0.0 -0.055640 T 5.37 1.4 -0.072562 -0.48 … The decaying sine wave of the ACFs combined with a spike at lag 1 for the PACFs suggests an autoregressive process of order 1.30 0.4 -0.6 0.0 2 4 6 8 10 Lag 12 14 16 18 Autocorrelation Function: Ex9-16mole Lag 1 2 3 4 5 ACF 0. 9-20 .658253 -0.23 LBQ 33.4 0.8 -1.033132 -0. (a) Stat > Time Series > Autocorrelation Autocorrelation Function for Molecular Weight Measurements (Ex9-16mole) (with 5% significance limits for the autocorrelations) 1.70 2. AR(1).29 -0.92 0.8 -1.42 -0.29 … Stat > Time Series > Partial Autocorrelation rtial Autocorrelation Function for Molecular Weight Measurements (Ex9-16mo (with 5% significance limits for the partial autocorrelations) 1.658253 0.2 -0.6 -0.96 -0.2 0.110802 -0. Note: In the textbook.0 0.6 0.70 -0.039599 T 5.0 2 4 6 8 10 Lag 12 14 16 18 Partial Autocorrelation Function: Ex9-16mole Lag 1 2 3 4 5 PACF 0.2 -0.81 44.

9. 4 out of 5 points more than 1 standard deviation from center line (on one side of CL). 13. 15. 8. 14. 9-21 . 6 points in a row all increasing or all decreasing. Failed at points: 8. 35. Failed at points: 12. with big swings and very few observations actually near the mean.97 1. LCL = 1953 σˆ = MR d 2 = 17. UCL = 2049. One point more than 3. 13.128 = 15. 70 6. Failed at points: 12.93 Stat > Control Charts > Variables Charts for Individuals > Individuals I Chart of Molecular Weight Measurements (Ex9-16mole) 1 1 2050 UCL=2048. 10. 2 out of 3 points more than 2 standard deviations from center line (on one side of CL).9 2000 6 1975 6 2 1950 2 5 2 1 1 1 1 7 1 1 LCL=1953. 15. 40. 31. 32. Failed at points: 6. 13. 32. 36. 11.00 standard deviations from center line. 36. 14. 12. 16. 14. violating many runs tests.1 1 1 14 21 28 35 42 49 Observation 56 63 70 Test Results for I Chart of Ex9-16mole TEST Test TEST Test TEST Test TEST Test TEST Test TEST Test 1. 9 points in a row on same side of center line. 33. Failed at points: 7. 13. 12.Chapter 9 Exercise Solutions 9-16 continued (b) x chart: CL = 2001. 53 5. 34. 14. 7. 69 2. 8.7 8 6 6 6 6 6 2025 Individual Value 3 _ X=2000. Failed at points: 7. 37 8. 15 3. 12. 8 points in a row more than 1 standard deviation from center line (above and below CL). 11. 37 The process is out of control on the x chart. 35.

One point more than 3.4 0.693 4 30929.697 606. 9-22 .698 605.19 Constant 605.1 0.7 0 -25 -50 LCL=-59.494 8 30897.400 1200.4 -75 1 7 14 21 28 35 42 49 Observation 56 63 70 Test Results for I Chart of Ex9-16res TEST 1.364 256.21 7.698 605.756 3 32083.0 0.843 2 35687.1 0.1 0.000 Stat > Control Charts > Variables Charts for Individuals > Individuals I Chart of Residuals from Molecular Weight Model (Ex9-16res) 1 UCL=58.0852 8.0 50 Individual Value 25 _ X=-0.7 0.3 0.00 standard deviations from center line.998 6 30897.956 7 30897.6979 0.100 1800.693 613.Chapter 9 Exercise Solutions 9-16 continued (c) Stat > Time Series > ARIMA ARIMA Model: Ex9-16mole Estimates at each iteration Iteration SSE Parameters 0 50173. There are no other violations of special cause tests.550 900.196 2.9 0.675 650.942 1 41717.197 5 30898. Test Failed at points: 16 Observation 16 signals out of control above the upper limit.250 1500.82… P 0.196 Relative change in each estimate less than 0.02 Mean 2003.000 0.6 0.0010 Final Estimates of Parameters Type Coef SE Coef T AR 1 0.

h = 5 100 UCL=97.Chapter 9 Exercise Solutions 9-17. Stat > Control Charts > Time-Weighted Charts > CUSUM CUSUM Chart of Residuals from Molecular Weight Model (Ex9-16res) mu0 = 0. The residuals are in control. Let µ0 = 0. δ = 1 sigma.5. k = 0.9 -100 1 7 14 21 28 35 42 Sample 49 56 63 70 No observations exceed the control limit. k = 0.9 Cumulative Sum 50 0 0 -50 LCL=-97. h = 5. 9-23 .5.

Chapter 9 Exercise Solutions 9-18. L = 2.7SL=-12.7SL=11.71 0 -5 -10 -2.83 -15 1 7 14 21 28 35 42 Sample 49 56 63 70 Process is in control.1 and L = 2.42 10 EWMA 5 _ _ X=-0.7 (approximately the same as a CUSUM with k = 0. Stat > Control Charts > Time-Weighted Charts > EWMA EWMA Chart of Residuals from Molecular Weight Model (Ex9-16res) lambda = 0.1. Let λ = 0.5 and h = 5). 9-24 .7 +2.

502 1970.480 No 1916.930 1988.848 1993.000 1750.000 2050. fit an ARIMA (0.128 = 15. To find the optimal λ.1.415 2026.806 1968.9238 σˆ = MR d 2 = 17.217 2074.574 1954.000 1950.128 1940.000 1900.000 1800.1)).1) (= EWMA = IMA(1.415 2026.376 2002.084 2012.211 2.613 No 1978.004 No 1921.608 2016.533 2031.Chapter 9 Exercise Solutions 9-19.920 No 1948.731 1984.521 0.479 2017.65 Constant -0.304 2017.749 2092.0762 = 0.000 2000.244 No 1898.947 2044.000 1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70 73 Obs. No.281 2065.644 LCL OOC? 1953.xls : Worksheet Ex9-19 t xt 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 2048 2025 2017 1995 1983 1943 1940 1947 1972 1983 1935 1948 1966 1954 1970 2039 zt 2000.145 No 1996.014 1938.057 1982.654 CL UCL 2000.772 No 1907.97 1.700 No 1922.467 1946.582 1947.731 1984.780 No 1899.128 1940.722 1996.046 1946.947 2044.282 1964.384 1995.326 No 1892.057 1982. 9-25 .467 1946.524 2044.842 2033.1181 0.842 2048.014 1938.0762 0.930 λ = 1 – MA1 = 1 – 0.000 2100.806 1968. Stat > Time Series > ARIMA ARIMA Model: Ex9-16mole … Final Estimates of Parameters Type Coef SE Coef T MA 1 0.000 1850.722 1996. CL UCL LCL xt Observation 6 exceeds the upper control limit compared to one out-of-control signal at observation 16 on the Individuals control chart.046 1946.255 No 1934.212 No 1890.479 2017.040 above UCL … Xt.393 -0.574 1954.93 Excel : Workbook Chap09.269 1994.502 1970.282 1964. Molecular Weight EWMA Moving Center-Line Control Chart for Molecular Weight 2150.816 1986.859 2029.677 No 1969.929 No 1936.582 1947.665 No 1898.09 … P 0.

8 -1.86 144.46 1.158358 T 7.520417 0.4 0.8 Autocorrelation 0.04 -0.Chapter 9 Exercise Solutions 9-20 (a) Stat > Time Series > Autocorrelation Autocorrelation Function for Concentration Readings (Ex9-20conc) (with 5% significance limits for the autocorrelations) 1.6 -0.0 2 4 6 8 10 12 14 16 18 20 22 24 Lag Partial Autocorrelation Function: Ex9-20conc Lag 1 2 3 4 5 PACF 0.12 … Stat > Time Series > Partial Autocorrelation Partial Autocorrelation Function for Concentration Readings (Ex9-20conc) (with 5% significance limits for the partial autocorrelations) 1.05 2.004498 -0.10 1.2 0.390108 0.38 127.58 … The decaying sine wave of the ACFs combined with a spike at lag 1 for the PACFs suggests an autoregressive process of order 1.0 -0.0 0.4 -0.4 0.6 -0.03 150.095134 -0.635375 0.8 -1.0 2 4 6 8 10 12 14 16 18 20 22 24 Lag Autocorrelation Function: Ex9-20conc Lag 1 2 3 4 5 ACF 0.2 -0.2 0.2 -0.95 -1.77 -0.6 0. AR(1).8 0.46 4.6 0. 9-26 .746174 0.23 LBQ 57.37 3.746174 0.238198 T 7.0 Partial Autocorrelation 0.4 -0.36 99.0 -0.177336 -0.

44 The process is out of control on the x chart. 17. 21. 22. 100 TEST 5. 39. 68. 41. 94. 44. 28. 72. 96. 66. 44. 29. 89. Test Failed at points: 8. 68. 97.128 = 3. 38. 20. 65. 18. 30. 66. 43. 73. 94. with big swings and very few observations actually near the mean. 9 points in a row on same side of center line. 23. 9-27 . 19. 99 TEST 8. 40. 71.64 1.01 11 1 11 30 40 50 60 Observation 70 80 90 100 Test Results for I Chart of Ex9-20conc TEST 1. 22. 86. 2 out of 3 points more than 2 standard deviations from center line (on one side of CL). 34. 38. 42. 14. 41. Test Failed at points: 15. 88. 36. 39. 29. 88. 21.227 Stat > Control Charts > Variables Charts for Individuals > Individuals I Chart of Concentration Readings (Ex9-20conc) 215 1 1 1 1 210 5 Individual Value 5 6 205 1 1 22 52 2 11 UCL=209. 12. 37. 12. 36. 95. 37. 69. 4 out of 5 points more than 1 standard deviation from center line (on one side of CL). 69. 39. 21. 89.Chapter 9 Exercise Solutions 9-20 continued (b) σˆ = MR d 2 = 3. 94. 42. 42. 98. 43. Test Failed at points: 11. 93. 99. 34.34 1 20 _ X=200. Test Failed at points: 10. 95 TEST 2. 93. 38.68 6 200 2 195 6 62 5 190 1 1 1 10 6 2 22 2 2 2 2 2 2 2 55 2 6 66 LCL=190. 43. 36. 10. 41. 16. 86. 87. Test Failed at points: 15. violating many runs tests. 40. 41. 15. 42. 95 TEST 6. 89. 13. One point more than 3. 88. 43. 37.00 standard deviations from center line. 8 points in a row more than 1 standard deviation from center line (above and below CL). 40.

9-28 .4155 120.Chapter 9 Exercise Solutions 9-20 continued (c) Stat > Time Series > ARIMA ARIMA Model: Ex9-20conc … Final Estimates of Parameters Type Coef SE Coef T AR 1 0.1734 0.73 -15 1 10 20 30 40 50 60 Observation 70 80 90 100 Test Results for I Chart of Ex9-20res TEST 4.62 Individual Value 10 5 4 _ X=-0. however this is not unlikely for a dataset of 100 observations.7493 0.000 Stat > Control Charts > Variables Charts for Individuals > Individuals I Chart of Residuals from Concentration Model (Ex9-20res) 15 UCL=13.20 Constant 50.76 Mean 200. Consider the process to be in control.122 1.000 0.0669 11. Test Failed at points: 29 Observation 29 signals out of control for test 4.657 … P 0.05 0 -5 -10 LCL=-13. 14 points in a row alternating up and down.

0 -0.0 -0.4 -0.6 0.Chapter 9 Exercise Solutions 9-20 continued (d) Stat > Time Series > Autocorrelation Autocorrelation Function for Residuals from Concentration Model (Ex9-20res) (with 5% significance limits for the autocorrelations) 1.4 -0.8 0.8 Autocorrelation 0.0 2 4 6 8 10 12 14 Lag 16 18 20 22 24 9-29 .6 -0.2 -0.6 0.0 2 4 6 8 10 12 14 Lag 16 18 20 22 24 Stat > Time Series > Partial Autocorrelation tial Autocorrelation Function for Residuals from Concentration Model (Ex9-20r (with 5% significance limits for the partial autocorrelations) 1.8 -1.4 0.0 0.0 Partial Autocorrelation 0.2 0.4 0.2 -0.6 -0.2 0.8 -1.

133 100 0.Chapter 9 Exercise Solutions 9-20 (d) continued Stat > Basic Statistics > Normality Test Probability Plot of Residuals from Concentration Model (Ex9-20res) Normal 99.9 Mean StDev N AD P-Value 99 95 Percent 90 -0. 9-30 .1 -15 -10 -5 0 Ex9-20res 5 10 Visual examination of the ACF. PACF and normal probability plot indicates that the residuals are normal and uncorrelated.05075 4.343 80 70 60 50 40 30 20 10 5 1 0.407 0.

Stat > Control Charts > Time-Weighted Charts > CUSUM CUSUM Chart of Residuals from Concentration Model (Ex9-20res) mu0 = 0. k = 0.5. δ = 1 sigma.5. Let µ0 = 0. The residuals are in control. and the AR(1) model for concentration should be a good fit. h = 5 UCL=22.79 1 10 20 30 40 50 60 Sample 70 80 90 100 No observations exceed the control limit. k = 0.79 Cumulative Sum 20 10 0 0 -10 -20 LCL=-22. h = 5.Chapter 9 Exercise Solutions 9-21. 9-31 .

773 2 EWMA 1 _ _ X=-0.1. L = 2.7SL=-2.5 and h = 5).1 and L = 2. 9-32 .051 0 -1 -2 -2.Chapter 9 Exercise Solutions 9-22.7 (approximately the same as a CUSUM with k = 0. The residuals are in control.7 3 +2. Let λ = 0.7SL=2.874 -3 1 10 20 30 40 50 60 Sample 70 80 90 100 No observations exceed the control limit. Stat > Control Charts > Time-Weighted Charts > EWMA EWMA Chart of Residuals from Concentration Model (Ex9-20res) lambda = 0.

341 200.3034 -0.182 0 0 0 0 0 0 0 0 0 … Xt.820 203.418 200. Concentration EWMA Moving Center-Line Chart for Concentration 230 220 210 200 190 180 170 160 150 1 7 13 19 25 31 37 43 49 55 61 67 73 79 85 91 97 Obs.506 211.2945 0.243 201.0452 0.562 191.979 below LCL 181.15 … P 0.047 211.2945 = 0.1. 9-33 . No.813 198.099 209.706 sigma^ = t xt 0 1 2 3 4 5 6 7 8 9 10 204 202 201 202 197 201 198 188 195 189 zt 200.544 190.494 208.xls : Worksheet Ex9-23 lamda = 0.227 Excel : Workbook Chap09.737 190.691 212.003 0.139 193.882 λ = 1 – MA1 = 1 – 0.7055 σˆ = MR d 2 = 3.458 184.366 201.1) (= EWMA = IMA(1.825 202.825 202.418 200.924 211.813 198. 90). 56.23 CL 200.Chapter 9 Exercise Solutions 9-23. fit an ARIMA (0. To find the optimal λ.010 202.329 193.660 191.132 188.02 Constant -0.863 190.0975 3.558 188.243 201.128 = 3.863 UCL = LCL = OOC? 209.144 192. xt CL UCL = LCL = The control chart of concentration data signals out of control at three observations (8.685 192.239 198.64 1.660 191.366 201.239 198.139 193.432 3.010 202. Stat > Time Series > ARIMA ARIMA Model: Ex9-20conc … Final Estimates of Parameters Type Coef SE Coef T MA 1 0.920 208.1)).

78 -1.865899 0.0 Partial Autocorrelation 0.592580 0.31 211.6 0.6 -0.2 -0.078040 -0.865899 -0.86 196.66 4.13 2.2 0.25 133.4 -0.36 1. PACF graph suggest order 1.31… Stat > Time Series > Partial Autocorrelation Partial Autocorrelation Function for Temperature Measurements (Ex9-24temp) (with 5% significance limits for the partial autocorrelations) 1.66 -0.0 2 4 6 8 10 12 14 16 18 20 22 24 Lag Partial Autocorrelation Function: Ex9-24temp Lag 1 2 3 4 5 PACF 0.0 0.047106 -0.4 0.489422 0.6 0.373763 T 8.8 Autocorrelation 0.143236 0.8 0.0 -0.47 -1.8 -1.Chapter 9 Exercise Solutions 9-24.0 -0.0 2 4 6 8 10 12 14 16 18 20 22 24 Lag Autocorrelation Function: Ex9-24temp Lag 1 2 3 4 5 ACF 0.71 LBQ 77.13… Slow decay of ACFs with sinusoidal wave indicates autoregressive process.112785 T 8.43 0.67 3.6 -0.94 170.737994 0.2 -0. (a) Stat > Time Series > Autocorrelation Autocorrelation Function for Temperature Measurements (Ex9-24temp) (with 5% significance limits for the autocorrelations) 1.8 -1.4 0.2 0.4 -0. 9-34 .

21. The temperature measurements appear to wander over time. 26. … TEST 3. 19.00 standard deviations from center line. 16. Test Failed at points: 65. 22. 3. 4 out of 5 points more than 1 standard deviation from center line (on one side of CL). 33. 21. 23.52 6 6 5 6 66 5 52 2 2 2 1 1 1 11 1 5 11 1 LCL=491. 23. 18. 18. 32. 36. 3. 25. 21. One point more than 3. 27. 27. 6 points in a row all increasing or all decreasing.81 22 22 2 2 500 490 5 8 2 1 _ X=506. 16. 19. 24. 22. violating many of the tests for special causes. 19. 17. … TEST 6. 22. 26. 39. Test Failed at points: 2. 22.Chapter 9 Exercise Solutions 9-24 continued (b) Stat > Control Charts > Variables Charts for Individuals > Individuals I Chart of Temperaure Measurements (Ex9-24temp) 540 1 1 11 530 11 1 1 Individual Value 520 52 2 510 480 2 2 2 8 1 1 11 1 1 1 6 11 1 UCL=521. 4. 20. 19. 24. … TEST 8. 2 out of 3 points more than 2 standard deviations from center line (on one side of CL). … Process is out of control. 6. 24. 20.23 5 11 1 1 1 1 1 1 1 470 1 1 10 20 30 40 50 60 Observation 70 80 90 1 100 Test Results for I Chart of Ex9-24temp TEST 1. 9 points in a row on same side of center line. 8 points in a row more than 1 standard deviation from center line (above and below CL). 21. 18. 21. 20. 24. 18. Test Failed at points: 4. 34. 38. 22. 23. 9-35 . … TEST 2. 28. 25. 5. 24. 23. Test Failed at points: 1. 71 TEST 5. Test Failed at points: 20. Test Failed at points: 17. 23. 2. 37. 17.

000 Stat > Control Charts > Variables Charts for Individuals > Individuals I Chart of Residuals from Temperature Model (Ex9-24res) UCL=22.727 6.0480 18.000 0. One point more than 3.12 Mean 503. Test Failed at points: 71 Observation 94 signals out of control above the upper limit.67 Constant 52. and observation 71 fails Test 5.22 0 -10 -20 LCL=-21.985 … P 0. 9-36 .8960 0.23 20 5 Individual Value 10 _ X=0.7263 72.80 1 1 10 20 30 40 50 60 Observation 70 80 90 100 Test Results for I Chart of Ex9-24res TEST 1. 2 out of 3 points more than 2 standard deviations from center line (on one side of CL). The residuals do not exhibit cycles in the original temperature readings. and points are distributed between the control limits. Test Failed at points: 94 TEST 5. The chemical process is in control.Chapter 9 Exercise Solutions 9-24 continued (c) Stat > Time Series > ARIMA ARIMA Model: Ex9-24temp … Final Estimates of Parameters Type Coef SE Coef T AR 1 0.00 standard deviations from center line.3794 0.

indicating the process is in control.69 -40 1 10 20 30 40 50 60 Sample 70 80 90 100 No observations exceed the control limits.69 30 Cumulative Sum 20 10 0 0 -10 -20 -30 LCL=-36.5. MTB > Stat > Control Charts > Time-Weighted Charts > CUSUM CUSUM Chart of Residuals from Temperature Model (Ex9-24res) k = 0.Chapter 9 Exercise Solutions 9-25. 9-37 . h = 5 40 UCL=36. This is the same conclusion as applying an Individuals control chart to the model residuals. The residuals are in control.

indicating the process is in control. 9-38 .76 EWMA 2. L = 2.0 +2.1.33 -5. This is the same conclusion as applying the Individuals and CUSUM control charts to the model residuals.Chapter 9 Exercise Solutions 9-26.5 -2. The residuals are in control.0 1 10 20 30 40 50 60 Sample 70 80 90 100 No observations exceed the control limits. MTB > Stat > Control Charts > Time-Weighted Charts > EWMA EWMA Chart of Residuals from Temperature Model (Ex9-24res) lambda = 0.7SL=4.22 0.7 5.0 -2.5 _ _ X=0.7SL=-4.

520 474.xls : Worksheet Ex9-27 lambda = t xt 0 1 2 3 4 5 6 7 8 9 10 491 482 490 495 499 499 507 503 510 509 0.812 489.560 524.070 487.940 467. 94).75) Excel : Workbook Chap09.227 below LCL 476.098 UCL 506.267 509.429 494.940 514.0794 = 0.812 489.558 498. 9-39 .265 483.438 0.Chapter 9 Exercise Solutions 9-27.10 … P 0.647 498.1019 0.850 513.722 509.558 498. xt CL UCL LCL A few observations exceed the upper limit (46. To find the optimal λ.679 491.363 503.232 482.78 Constant -0.1.037 CL 5.173 0 0 0 0 0 0 0 0 0 EWMA Moving Center-Line Chart for Temperature Xt.0711 0.465 LCL 521. 58.972 506.758 OOC? 491.355 483.105 504.6784 -0. fit an ARIMA (0.972 506.137 479.128 = 5.267 509.75 1.647 498.921 sigma^ = zt 506.363 503.520 492.1)).974 494.1) (= EWMA = IMA(1.655 518. Stat > Time Series > ARIMA ARIMA Model: Ex9-24temp … Final Estimates of Parameters Type Coef SE Coef T MA 1 0.265 521. 69) and the lower limit (1.917 λ = 1 – MA1 = 1 – 0. similar to the two out-of-control signals on the Individuals control chart (71.232 482.525 498. 94).9206 σˆ = MR d 2 = 5.520 492.813 507.429 494. Temperature 570 550 530 510 490 470 450 1 7 13 19 25 31 37 43 49 55 61 67 73 79 85 91 97 Sample No.0794 0.465 509.0975 (from a Moving Range chart with CL = 5.

This would tend to produce an estimate of the process standard deviation that is too small. it is not good practice to estimate σ2 from S2. (b) S2 is still an unbiased estimator of σ2 when the data are positively autocorrelated. adjacent observations will tend to be similar. therefore making the moving ranges smaller.Chapter 9 Exercise Solutions 9-28. There is nothing in the derivation of the expected value of S2 = σ2 that depends on an assumption of independence. (a) When the data are positively autocorrelated. 9-40 . Since it is difficult to determine whether a process generating autocorrelated data – or really any process – is in control. it is generally a bad practice to use S2 to estimate σ2. (c) If assignable causes are present.

071963 T 4.54 LBQ 25.85 r1 = 0.22 -0. (a) Stat > Time Series > Autocorrelation Autocorrelation Function for Viscosity Readings (Ex9-29Vis) (with 5% significance limits for the autocorrelations) 1.2 -0.4 0.494137 -0.16 25.0 0.41 32.Chapter 9 Exercise Solutions 9-29.0 2 4 6 8 10 12 14 Lag 16 18 20 22 24 Autocorrelation Function: Ex9-29Vis Lag 1 2 3 4 5 … ACF 0. indicating a strong positive correlation at lag 1.41 -2. There is a serious problem with autocorrelation in viscosity readings.78 41.49.2 0. 9-41 .264612 -0.283150 -0.17 -2.4 -0.94 -0.6 0.8 Autocorrelation 0.6 -0.29 41.049610 -0.8 -1.0 -0.

75 TEST 7. One point more than 3. 15 points within 1 standard deviation of center line (above and below CL). violating many of the tests for special causes. 38. The viscosity measurements appear to wander over time.57 25 6 20 6 55 LCL=20. 86 TEST 6.Chapter 9 Exercise Solutions 9-29 continued (b) Stat > Control Charts > Variables Charts for Individuals > Individuals I Chart of Viscosity (Ex9-29Vis) 40 UCL=37. 9-42 . 92 TEST 5. Test Failed at points: 40. 8 points in a row more than 1 standard deviation from center line (above and below CL).11 5 Individual Value 35 6 6 7 30 _ X=28. 60. 64. 2 out of 3 points more than 2 standard deviations from center line (on one side of CL). 86. Test Failed at points: 2. Test Failed at points: 22 TEST 8. 63. Test Failed at points: 64 Process is out of control. 59.03 1 1 1 1 10 20 30 40 50 60 Observation 1 70 80 90 100 Test Results for I Chart of Ex9-29Vis TEST 1. 58.00 standard deviations from center line. Test Failed at points: 38. 4 out of 5 points more than 1 standard deviation from center line (on one side of CL).

24 10 0 0 -10 LCL=-14.5.569 MTB > Stat > Control Charts > Time-Weighted Charts > CUSUM CUSUM Chart of Viscosity (Ex9-29Vis) target = 28. k = 0. h = 5 20 Cumulative Sum UCL=14.24 -20 1 10 20 30 40 50 60 Sample 70 80 90 100 Several observations are out of control on both the lower and upper sides.569.Chapter 9 Exercise Solutions 9-29 continued (c) Let target = µ0 = 28. 9-43 .

7SL=30.7 32 31 +2.7SL=26.759 EWMA 30 _ _ X=28.15. L = 2. 9-44 . There are wide swings in the plot points and several are beyond the control limits.Chapter 9 Exercise Solutions 9-29 continued (d) MTB > Stat > Control Charts > Time-Weighted Charts > CUSUM EWMA Chart of Ex9-29Vis lambda = 0.380 26 25 1 10 20 30 40 50 60 Sample 70 80 90 100 The process is not in control.569 29 28 27 -2.

000 20. 55.785 37.1)).1579 σˆ = MR d 2 = 3.922 below LCL 9.328 2.984 18.356 25.680 33.873 39.500 26.909 29.898 33.479 29.760 29.909 29.940 18.550 Zi 28.788 22.330 19.785 31.Chapter 9 Exercise Solutions 9-29 continued (e) To find the optimal λ.21 1.367 20.330 31.452 37.000 1 7 13 19 25 31 37 43 49 55 61 67 73 79 85 91 97 Obs.000 30.000 40.05 P 0.000 10.4839 0.750 30.122 17.665 26. fit an ARIMA (0.1579 0.000 31. 85).665 26.527 26. Xi CL UCL LCL A few observations exceed the upper limit (87) and the lower limit (2.482 26.980 25.298 32.898 33.022 38. 9-45 . Viscosity 50.007 27.1579) = 1. No.441 42.1007 -1. 37. Stat > Time Series > ARIMA ARIMA Model: Ex9-29Vis … Final Estimates of Parameters Type Coef SE Coef T MA 1 -0.479 29.1) (= EWMA = IMA(1.330 31.57 Constant 0.0231 0.937 20.1.560 27.207 35.85 LCL OOC? 19.8457 (from a Moving Range chart with CL = 5.841 41.75) Excel : Workbook Chap09.144 1.243 0 0 0 0 0 0 0 0 0 EWMA Moving Center-Line Chart for Viscosity Xt.464 18.069 35.128 = 2.756 24.464 18.482 26.962 λ = 1 – MA1 = 1 – (– 0.120 0.xls : Worksheet Ex9-29 lambda = l Xi 0 1 2 3 4 5 6 7 8 9 10 … 29.025 35.158 sigma^ = CL UCL 28.298 32.030 32.527 26.000 0.

20.54 Mean 28.3278 62.5017 0. 19.Chapter 9 Exercise Solutions 9-29 continued (f) Stat > Time Series > ARIMA ARIMA Model: Ex9-29Vis … Final Estimates of Parameters Type Coef SE Coef T AR 1 0.04 0 7 7 -5 LCL=-9.7193 0. 22 The model residuals signal a potential issue with viscosity around observation 20. with a good distribution of points between the control limits and no patterns.6514 0.72 Constant 20. 21. Test Failed at points: 18. 15 points within 1 standard deviation of center line (above and below CL). Otherwise the process appears to be in control. 9-46 .68 -10 1 10 20 30 40 50 60 Observation 70 80 90 100 Test Results for I Chart of Ex9-29res TEST 7.0922 -4.000 Stat > Control Charts > Variables Charts for Individuals > Individuals I Chart of Residuals from Viscosity AR(2) Model (Ex9-29res) 10 UCL=9.000 0.0923 7.79 AR 2 -0.4349 0.60 Individual Value 5 7 7 7 _ X=-0.000 0.4581 … P 0.

a4 = \$100/hr g = 0. kopt = 2.895 E(L) = \$3. α = 0. a'3 = \$5.05hr/sample.472) − Φ (−7. h = 1. λ = 0.0705 − 0. hopt = 1. δ = 2. α = 0. 1 − β = 0.Chapter 9 Exercise Solutions 9-30. D = 2hr (a) Excel : workbook Chap09.50/sample.0 a1 = \$0. a3 = \$2.79/hr (b) n = 3.10/unit.0000 = 0.50.210.6098/hr 9-47 . a2 = \$0.027.00.0705 h λ h2 = 0.231.01/hr or 1/λ = 100hr.27 −λh λh 1− e ( ) E(L) = \$3.xls : worksheet Ex9-30a n = 5.4992 τ≅ − 2 12 α e−λh α ≅ = 0.0027 ⎛ ( µ + k σ n ) − ( µ0 + 2σ ) ⎞ ⎛ ( µ0 − k σ n ) − ( µ0 + 2σ ) ⎞ β = Φ ⎜⎜ 0 ⎟⎟ − Φ ⎜⎜ ⎟⎟ σ n σ n ⎝ ⎠ ⎝ ⎠ ( ) ( = Φ 3 − 2 5 − Φ −3 − 2 5 ) = Φ (−1. k = 3.472) = 0.

λ = 0.918 E(L) = \$4.368. kopt = 3.01392/hr 9-48 .0027 ≅ = = 0.10/unit. a'3 = \$50.0000 = 0.5 0.27 −λh λ h 0. k = 3.5.0027 α e−λh α ≅ = = 0.472) − Φ (−7. α = 0.05hr/sample.0 a1 = \$0. δ = 2. D = 2hr (a) Excel : workbook Chap09. α = 0. h = 1.00207.Chapter 9 Exercise Solutions 9-31.12/hr (b) n = 5.50/sample. 1 − β = 0. hopt = 1.98/hr (c) n = 5.080. a3 = \$25.01(1) 1− e ( ) E(L) = \$4.4992 τ≅ − 2 12 2 12 α e−λh α 0. a2 = \$0.472) = 0. β = 0. α = 0.0705 h λ h 2 1 0. a4 = \$100/hr g = 0. k = 3.2498 τ≅ − 2 12 2 12 0.54 −λh λ h 0.01(0.5) 1− e ( ) E(L) = \$4.xls : worksheet Ex9-31 n = 5.01/hr or 1/λ = 100hr.0705 h λ h 2 0.0027 ⎛ ( µ + k σ n ) − ( µ + δσ ) ⎞ ⎛ ( µ − k σ n ) − ( µ + δσ ) ⎞ 0 0 ⎟−Φ⎜ 0 ⎟ β = Φ⎜ 0 ⎜ ⎟ ⎜ ⎟ σ n σ n ⎝ ⎠ ⎝ ⎠ ( ) ( ) = Φ ( 3 − 2 5 ) − Φ ( −3 − 2 5 ) = Φ k − δ n − Φ −k − δ n = Φ (−1.01(0.0705 − 0.01(12 ) = − = 0.52 ) = − = 0. h = 0.0027.

β = 0. hopt = 2.240. ∆ = \$25 n = 5.05 hr/sample D = 1 hr (a) n = 5. D1 = 2hr V0 = \$500. α = 0.xls : worksheet Ex9-33 λ = 0.52 ) = − = 0. kopt = 2.775 h λ h 2 0.16/hr 9-33.489018.54 −λh λ h 0.5 0. k = 3.775 − 0.0 a1 = \$2/sample a2 = \$0.Chapter 9 Exercise Solutions 9-32. Excel : workbook Chap09. 1 − β = 0.0705 E(L) = \$13. h = 1.17/hr (b) n = 10. α = 0.2498 τ≅ − 2 12 2 12 0.01(0.472) − Φ (−7.8218083 E(L) = \$10. h = 0.472) = 0.0027 α e−λh α ≅ = = 0.0000 = 0.0027.0027 β = Φ k − δ n − Φ −k − δ n ( ) ( ) = Φ ( 3 − 1 5 ) − Φ ( −3 − 1 5 ) = Φ (−1.5) 1− e ( ) E(L) = \$16.01(0.xls : worksheet Ex9-32 D0 = 2hr. Excel : workbook Chap09.01/hr or 1/λ = 100hr δ = 2.025091.5.39762/hr 9-49 .50/unit a'3 = \$75 a3 = \$50 a4 = \$200/hr g = 0. α = 0. k = 3.

0015 1. It is good practice visually examine data in order to understand the type of tool wear occurring.00028 5 = 1.0035 − 3(0.00028 CL = R = 0.0025 1. σˆ = R d 2 = 0.00028 5 ) = 1. Graph > Time Series Plot > With Groups Time Series Plot of Ex9-34Xb USL = 1.00196 x chart at tool reset: CL = USL − 3σ = 1.00028 5 = 1. The plot below shows that the tool has been reset to approximately the same level as initially and the rate of tool wear is approximately the same after reset.0030 1.0015 + 3(0.00064 2.0015 1 2 3 4 5 6 7 8 9 10 11 12 Ex9-34Sample n = 5.00228 9-50 . UCL = D4 R = 2.00266 (maximum permissible average) UCL = CL + 3σ x = 1.00304 LCL = CL − 3σ x ( ) = 1.00064.00234 UCL = CL + 3σ x = 1.114(0.00272 LCL = CL − 3σ x ( ) = 1. R = 0.00234 − 3 ( 0.00064) = 0.0020 LSL = 1.Chapter 9 Exercise Solutions 9-34.00028 5 ) = 1.00266 − 3 ( 0.326 = 0.00234 + 3 0.00028) = 1.0035 Ex9-34Reset A fter Before Ex9-34Xb 1.00135. LCL = 0 x chart initial settings: CL = LSL + 3σ = 1.00064.00028) = 1.00266 + 3 0.0035 1.

9127 22.00 30.mn − m − p +1 mn − m − p + 1 2(50 + 1)(25 − 1) = F0.00 20.1690 0. xbar1 xbar2 1 58 32 2 60 33 3 50 27 4 54 31 5 63 38 6 53 30 7 42 20 8 55 31 9 46 25 3 2 5 3 -5 -3 -1 1 8 8 -2 0 -13 -10 0 1 -9 -5 0.6338 14.1850 0 diff1 diff2 matrix calc t2 = n * calc UCL = LCL = OOC? In control In control In control In control In control In control Above UCL In control In control … T^2 Control Chart for Quality Characteristics 60.0817 2.8169 0.1690 0.1850 0 14.00 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Sample Number Process is out of control at samples 7 and 14. p = 2 characteristics.0451 1.Chapter 10 Exercise Solutions Note: MINITAB’s Tsquared functionality does not use summary statistics.1850 0 14.5211 0.6901 0.00 10.0423 0.5408 13.00 50.2.1850 0 14.4254 10.00 0.00 T^2 40.1850 0 14. 10-1.xls : worksheet Ex10-1 Sample No.1850 0 14.186 LCL = 0 Excel : workbook Chap10.001.1199 50(25) − 50 − 2 + 1 = ( 2448 1199 ) (6. 10-1 .948) = 14.7042 0. Let α = 0.1268 3. Phase 2 T 2 control charts with m = 50 preliminary samples.0676 1.1268 0.1850 0 14.1268 3.1850 0 14.1850 0 14. p (m + 1)(n − 1) UCL = Fα .001. p . so many of these exercises have been solved in Excel.0282 0. n = 25 sample size.

2 3.8 11 3.1 0.4249 17.6 -0.xls : worksheet Ex10-2 Sample No.1 3.0528 0.3. p .2 0.0852 0.4249 17.9 -1 0.1887 0.7 0.4249 17.9 3.8 4.3 0.6844 0.4249 17.1271 11. xbar1 xbar2 xbar3 1 3.001.5 0.1216 0.4 3 2.0808 0.4 4 2.8 6 4 4.1 1.4249 17.8084 1.4 -0.2 -0.7 0.4249 17.5 7 3.9 15 3.5741 2.268 30(10) − 30 − 3 + 1 ⎛ 837 ⎞ =⎜ ⎟ (5.2 0.3 -0.1 4.2 0.3 0 -0.8800 0.1 0 0.00 0.4 0.2 3.00 20.4249 17.6 -1 -0.00 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Sample Number Process is out of control at sample 13.2 14 3.8 3 2.1189 1.5 5 3 3.4249 17.7 3 2 3.579) ⎝ 268 ⎠ = 17. n = 10 sample size. p = 3 characteristics.00 10.7 9 2.7 0.1 0.425 LCL = 0 Excel : workbook Chap10.8 0.7 3.1 0.4 -0.4249 17.5932 0.mn − m − p +1 mn − m − p + 1 3(30 + 1)(10 − 1) = F0.2 0.3990 0.5 -0.3719 0.Chapter 10 Exercise Solutions 10-2.00 T^2 15.4249 17.0397 10.3 2.001.2 3 8 3 3.2 -0.0684 0.2706 0.5 -0.00 5.2 0 1 1.8 4 2.0793 20. Phase 2 T 2 control limits with m = 30 preliminary samples.1399 1.3 3.2372 2.4249 17.2 1.4249 17.5 0.6 3 2.2 10 2 2. Let α = 0.0593 10.2 0.7927 1.4 0.5 -0.8692 8. 10-2 .1 -0.3122 3.4 0.4249 17.7 4 3 13 4.1880 1.2 0.6 3.8525 matrix calc t2 = n * calc UCL = LCL = OOC? 17.4249 17. p (m + 1)(n − 1) UCL = Fα .8 diff1 diff2 diff3 0.6 2.2 0 -0.6922 0.1 3 2.3966 1.4249 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 In control In control In control In control In control In control In control In control In control In control In control In control Above UCL In control In control T^2 Control Chart for Quality Characteristics 25.8 0.6574 6.5279 0.3 2.9 3 12 3.6 1.

Let α = 0.1690 0.8150 0 13. xbar1 xbar2 1 2 UCL = LCL = OOC? 4 5 6 7 8 9 10 11 12 13 14 15 60 33 50 27 54 31 63 38 53 30 42 20 55 31 46 25 50 29 49 27 57 30 58 33 75 45 55 27 3 2 5 3 -5 -3 -1 1 8 8 -2 0 -13 -10 0 1 -9 -5 -5 -1 -6 -3 2 0 3 3 20 15 0 -3 0.1690 0.8150 0 13.8150 0 13.001.2535 6.0423 0.0282 0. Since population parameters are known.8150 0 13.1268 3.0704 0. Same results as for parameters estimated from samples.0000 0.9014 2.0000 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Sample Number Process is out of control at samples 7 and 14.0676 1.7042 0. 10-3 .0761 1.0676 1.8169 0.8150 0 diff1 diff2 matrix calc t2 = n * calc 3 58 32 In control In control In control In control In control In control Above UCL In control In control In control In control In control In control Above UCL In control Phase II T^2 Control Chart 60.8150 0 13.8169 0.2028 5.0000 30.0000 T^2 40.0000 10.8150 0 13.2676 6.0000 50.8150 0 13.816 control limit: UCL = χα2 .6901 0.1127 52.8150 0 13.xls : worksheet Ex10-3 Sample No.2 Excel : workbook Chap10.5408 13.0817 2. the chi-square formula will be used for the upper 2 = 13.8150 0 13.Chapter 10 Exercise Solutions 10-3.1268 0.8150 0 13.0451 1.1268 3.8150 0 13.8150 0 13.8150 0 13.8150 0 13.9127 22.6901 0.6901 0.3380 13. Phase 2 T 2 control limits with p = 2 characteristics. p = χ 0.4254 10.6338 0.5211 0.0000 20.001.

1189 1.2660 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 In control In control In control In control In control In control In control In control In control In control In control In control Above UCL In control In control Phase II T^2 Control Chart 25.1 0.8 0.2660 16.2660 16.3966 1.4 4 2.3 0.3 0 -0.2 -0.3719 0.7 3.5 -0.0808 0.6574 6.Chapter 10 Exercise Solutions 10-4.6 2.2 0 1 1.5 0.2 -0.7 9 2.1880 1.6 -0.1887 0.5 -0.1 0 0.5 5 3 3.2 0.7 4 3 13 4.2660 16.8 3 2.2 10 2 2. xbar1 xbar2 xbar3 1 3.8084 1.7927 1.2660 16.4 -0.0852 0.6844 0.5 0. Since population parameters are known.7 3 2 3.2 14 3.8 4.0684 0.2660 16.7 0.8 diff1 diff2 diff3 0.0397 10.1 1.1271 11.266 control limit: UCL = χα2 .2660 16.2660 16.001.1 -0.2 0.6 3 2.8 4 2.9 15 3.00 10.3 -0.5279 0.3990 0.4 0. Phase 2 T 2 control limits with p = 3 characteristics.2660 16.2372 2.2 3.7 0.xls : worksheet Ex10-4 Sample No.4 3 2. Same as results for parameters estimated from samples.4 0.8525 matrix calc t2 = n * calc UCL = LCL = OOC? 16.0593 10.9 3.8800 0.00 5.0793 20. 10-4 .6 -1 -0.00 T^2 15.2706 0.3122 3.4 0.2 1.0528 0.4 -0.1 3 2.5932 0.001.2660 16.3 3.2 0.1399 1.1 0.6 1.1216 0.2660 16.3 2.8 11 3.2 0. p = χ 0.1 4.3 Excel : workbook Chap10.2 0 -0.7 0.2 3.9 3 12 3.00 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Sample Number Process is out of control at sample 13.2660 16.5 7 3.2 0. the chi-square formula will be used for the upper 2 = 16.5 -0.1 0.2 3 8 3 3.3 2.2 0.2660 16.6922 0. Let α = 0.00 0.6 3.8692 8.8 6 4 4.2660 16.8 0.2660 16.5741 2.9 -1 0.00 20.1 3.

107 3.005 (a) Phase II limits: p (m + 1)(n − 1) UCL = Fα .xls : worksheet Ex10-5 m 30 40 50 60 70 80 90 100 num denom 372 55 492 75 612 95 732 115 852 135 972 155 1092 175 1212 195 F 3.531) ⎝ 55 ⎠ = 23. Find "m" such that exact Phase II limit is within 1% of chi-square limit.1095 19.6.294 3.9447 717 718 719 720 721 722 8616 8628 8640 8652 8664 8676 3.Chapter 10 Exercise Solutions 10-5.338 3.733.548) = 18.7337 18.209 UCL 23.107 3.3184 20. (c) Quality characteristics.7324 … 1429 1431 1433 1435 1437 1439 720 preliminary samples must be taken to ensure that the exact Phase II limit is within 1% of the chi-square limit.263 3.005. Samples size. p = χ 0.005. n = 3 sample size.7332 18. 10-5 .8820 22.9650 20.107 3.240 3.005.107 3.7325 18. m = 30 preliminary samples. Excel : workbook Chap10.55 30(3) − 30 − 6 + 1 ⎛ 372 ⎞ =⎜ ⎟ (3. p = 6.882 LCL = 0 (b) 2 chi-square limit: UCL = χα2 .6 = 18.3527 21. n = 3.531 3.7331 18.5042 20.5920 20.407 3. 1.7328 18.mn − m − p +1 mn − m − p + 1 6(30 + 1)(3 − 1) = F0. p = 6 characteristics. α = 0.01(18.548 The Phase II UCL is almost 30% larger than the chi-square limit.223 3.107 3. p . α = 0.107 18.

105 3.4119 19. p . p = 6 characteristics.7330 18.1422 19.548) = 18.xls : worksheet Ex10-6 m 30 40 50 60 70 80 90 100 num 744 984 1224 1464 1704 1944 2184 2424 denom 115 155 195 235 275 315 355 395 F 3.01(18.6 = 18.548 The Phase II UCL is almost 15% larger than the chi-square limit.7324 18. α = 0. Find "m" such that exact Phase II limit is within 1% of chi-square limit.Chapter 10 Exercise Solutions 10-6.105 3.005. 10-6 .240 3.7318 … 411 preliminary samples must be taken to ensure that the exact Phase II limit is within 1% of the chi-square limit.149 UCL 21. α = 0.294 3.174 3.6685 19.115 30(5) − 30 − 6 + 1 ⎛ 744 ⎞ =⎜ ⎟ (3.189 3.005. n = 5.209 3. Samples size.3232 390 400 410 411 412 9384 9624 9864 9888 9912 1555 1595 1635 1639 1643 3.8641 19.6. Excel : workbook Chap10.7424 18.005.5692 20. p = 6.7376 18.005 (a) Phase II UCL: p (m + 1)(n − 1) UCL = Fα .5237 19. m = 30 preliminary samples.309 (b) 2 chi-square UCL: UCL = χα2 .106 3.mn − m − p +1 mn − m − p + 1 6(30 + 1)(5 − 1) = F0.105 3.105 18. 1. (c) Quality characteristics.294) ⎝ 115 ⎠ = 21.164 3.155 3.3087 20.733. p = χ 0. n = 5 sample size.

xls : worksheet Ex10-7 m 25 35 45 55 65 75 85 95 105 num 520 720 920 1120 1320 1520 1720 1920 2120 denom 41 61 81 101 121 141 161 181 201 F 3.799 2.704 2. m = 25 preliminary samples.10 = 25. n = 3.101) ⎝ 41 ⎠ = 39. 10-7 .3808 28.7953 30.Chapter 10 Exercise Solutions 10-7.742 2.005 (a) Phase II UCL: p (m + 1)(n − 1) UCL = Fα .440.mn − m − p +1 mn − m − p + 1 10(25 + 1)(3 − 1) = F0.005. n = 3 sample size.01(25.4405 25. α = 0.530 2.4940 28.8549 28.005.101 2.188) = 25. Excel : workbook Chap10.10.0154 27.41 25(3) − 25 − 10 + 1 ⎛ 520 ⎞ =⎜ ⎟ (3. p = 10 characteristics.530 2. p = χ 0. Find "m" such that exact Phase II limit is within 1% of chi-square limit.657 2. α = 0. p .629 UCL 39.188 The Phase II UCL is more than 55% larger than the chi-square limit.4394 … 988 preliminary samples must be taken to ensure that the exact Phase II limit is within 1% of the chi-square limit.3259 34.677 2.4398 25.7246 986 987 988 989 990 19740 19760 19780 19800 19820 1963 1965 1967 1969 1971 2. (c) Quality characteristics.530 2. 1.4401 25.326 (b) 2 chi-square UCL: UCL = χα2 .4024 29.4399 25.530 25.005.1991 31. Samples size.897 2.641 2.530 2. p = 10.

Find "m" such that exact Phase II limit is within 1% of chi-square limit. α = 0. p = χ 0.xls : worksheet Ex10-8 m 25 35 45 55 65 75 85 95 105 num denom F UCL 1040 91 2.4394 544 preliminary samples must be taken to ensure that the exact Phase II limit is within 1% of the chi-square limit. m = 25 preliminary samples.5335 540 541 542 543 544 545 21640 21680 21720 21760 21800 21840 2151 2155 2159 2163 2167 2171 2. α = 0.623 27.767 31.440.10.005.594 27.606 27.4141 3040 291 2.005.188 The Phase II UCL is more than 25% larger than the chi-square limit.6812 4240 411 2.91 25(5) − 25 − 10 + 1 ⎛ 1040 ⎞ =⎜ ⎟ (2.8651 3840 371 2. n = 5 sample size.005.4408 25.529 2.585 26.529 2.005 (a) Phase II UCL: p (m + 1)(n − 1) UCL = Fα . Excel : workbook Chap10. 10-8 .10 = 25.767) ⎝ 91 ⎠ = 31.689 29. p = 10 characteristics. p .1011 3440 331 2.01(25.6251 1440 131 2.529 2.8495 2640 251 2. p = 10.4399 25.625 (b) 2 chi-square UCL: UCL = χα2 .572 26.529 2.578 26.188) = 25.mn − m − p +1 mn − m − p + 1 10(25 + 1)(5 − 1) = F0.529 2. Samples size.Chapter 10 Exercise Solutions 10-8.5595 1840 171 2.648 28.4405 25. (c) Quality characteristics.4419 25. n = 5. 1.4967 2240 211 2.4413 25.529 25.

788) ⎝ 41 ⎠ = 32. Phase I UCL: p (m − 1)(n − 1) UCL = Fα .7 ⎥ ⎢ ⎥ ⎣0.mn − m − p +1 mn − m − p + 1 10(25 + 1)(3 − 1) = F0.01.277 10-9 .7 ⎥ ⎥ Σ=⎢ ⎢0.638 Phase II UCL: p (m + 1)(n − 1) UCL = Fα .7 0.788) ⎝ 41 ⎠ = 35. Assume α = 0.41 25(3) − 25 − 10 + 1 ⎛ 520 ⎞ =⎜ ⎟ (2. Excel : workbook Chap10. p .360 10-10.7 ⎤ ⎢0. p = χ 0.7 0.7 0. p . m = 25 preliminary samples.10.01.Chapter 10 Exercise Solutions 10-9.7 0. n = 3 sample size.7 1 0.7 1 0.41 25(3) − 25 − 10 + 1 ⎛ 480 ⎞ =⎜ ⎟ (2.xls : worksheet Ex10-10 (a) ⎡ 1 0.7 0.4 = 13.mn − m − p +1 mn − m − p + 1 10(25 − 1)(3 − 1) = F0.01. p = 10 quality characteristics.7 0.7 1 ⎦ (b) 2 UCL = χα2 .01.10.

7 = 15.806 Yes.635 T(1)2 = 15.7 ⎥⎥ ⎜ ⎢⎢3.313 di = T 2 − T(2i ) d1 = d 2 = d3 = d 4 = 15.313 2 2 2 2 T(1) = T(2) = T(3) = T(4) = 15.5 0 0.800 2 T(4) = 25.806 − 15.7 1 ⎦ ⎜⎝ ⎣3.689 Investigate variables 1 and 3.5⎥ ⎢ 0⎥ ⎟ ⎦ ⎝⎣ ⎦ ⎣ ⎦⎠ ⎝⎣ ⎦ ⎣ ⎦⎠ ⎣ = 15.7 1 0.7 0.01. Since T 2 = 15.313 = 0. all variables had the same shift).280) > (UCL = 13.µ (1) ) ⎛ ⎡3.µ )′ Σ −1 ( y .Chapter 10 Exercise Solutions 10-10 continued (c) T 2 = n ( y .7 1 ⎥ ⎜ ⎢3.5⎥⎥ − ⎢⎢0 ⎥⎥ ⎟ ⎢⎢ 0.300 T(3)2 = 14. no variable is identified as a relatively large contributor. First.5⎥ ⎢ 0 ⎥ ⎟ ⎢ 0. (f) 2 χ 0.590.585 2 T(2) = 21.7 = 1⎜ ⎢ ⎥ − ⎢ ⎥ ⎟ ⎢ ⎜ ⎢3.5⎤ ⎡ 0⎤ ⎞ ⎜ ⎟ 1 0.1 . an out-of-control signal is generated.5⎦ ⎣ 0 ⎦ ⎠ ⎣ 0.7 ⎤ −1 ⎛ ⎡3.479.1 = 6. (e) Since (T 2 = 28.7 ⎥ ⎜ ⎢3. d 2 = 6.01.979.277 ) .01.5⎥⎥ − ⎢⎢ 0⎥⎥ ⎟ ⎜ ⎢3.493 2 χ 0.7 ⎜⎜ ⎢ ⎥ ⎢ ⎥ ⎟⎟ ⎢ ⎝ ⎣3.5⎤ ⎡ 0 ⎤ ⎞′ ⎡ 1 ⎜⎢ ⎥ ⎢ ⎥⎟ ⎢ 3.7 ⎤ −1 ⎛ ⎡3. since all di are smaller than χ 0.5⎥⎥ ⎢⎢ 0⎥⎥ ⎟ − 0.5⎤ ⎡ 0⎤ ⎞ ⎜ ⎟ ⎜ ⎟ = 1⎜ ⎢⎢3. this information does not assist in identifying which a process variable shifted.277) . (d) −1 T(1)2 = n ( y (1) .7 0. since the standardized observations are equal (that is.7 0. an out-of-control signal is generated.µ ) ⎛ ⎡3.7 ⎥⎥ ⎜ ⎢⎢3. 10-10 .7 0.7 0.µ (1) )′ Σ (1) ( y (1) .635 2 No.7 1 0.5⎥ ⎢ 0⎥ ⎟ ⎥ ⎜⎢ ⎥ ⎢ ⎥⎟ 0. d 4 = 2.5⎥ ⎢0 ⎥ ⎟ ⎢ 0.5⎦ ⎣ 0⎦ ⎟⎠ ) > ( UCL = 13.7 0. d3 = 13. Second.5⎤ ⎡0 ⎤ ⎞′ ⎡ 1 0.806 ( 0.1 = 6. d1 = 12.694.

815) .1 = 3. Excel : workbook Chap10.8 1 ⎥⎦ (b) 2 UCL = χα2 .538) > (UCL = 7.154 > ( UCL = 7.154 Variables 2 and 3 should be investigated.094 Since an out-of-control signal was not generated in (e).815 (c) T 2 = 11.8 0. an out-of-control signal is generated.154 Yes. (f) 2 χ 0.8⎤ Σ = ⎢⎢ 0.1 = 3. d 2 = 1. d3 = 6. p = χ 0.538 T(3)2 = 4.Chapter 10 Exercise Solutions 10-11. 10-11 .05.538 2 T(2) = 5.376 T(3)2 = 5. it is not necessary to calculate the diagnostic quantities. an out-of-control signal is not generated.778.000.815 ) .05.05. d1 = 1.8 0. ( ) (d) 2 χ 0.841 T(1)2 = 5.841 T(1)2 = 11.8 1 0.444.xls : worksheet Ex10-11 (a) ⎡ 1 0.111. d3 = 2.8⎥⎥ ⎢⎣ 0. d1 = 0. (e) Since (T 2 = 6. Since T 2 = 11.000.3 = 7.000. d 2 = 8. This is confirmed since none of the di’s exceeds the UCL.043 2 T(2) = 2.

S 2 = ⎢⎢ −0.014 ⎥⎥ ⎢⎣ 5.526 ⎥⎦ ⎢⎣ 43.641 V'V 133.001 ⎦ 10-13.012 ⎥⎥ ⎢⎣ −0.071 0.305 3. S1 = ⎢ ⎥ ⎣ −0.740 67.440 −0.150 S2 2.256 0.607 21.392 1.561⎤ ⎡121.125] .780 80.104] .395 −0.950 ⎥ .012 7.720 0.395 ⎤ x′ = [15.256 43.xls : worksheet Ex10-13 m = 40 ⎡ 4.305 5.256 0.740 80.307 1.101 −0. S1 = ⎢⎢ −0. S2 = ⎢ V′V = ⎢ ⎥ ⎥ ⎣ −0.014 27.001 ⎦ ⎡121.158 10-12 .003⎤ .339 0.003 0. Excel : workbook Chap10.207 S1 3.003 −0.599 ⎥⎦ ⎡ 1.561 0.xls : worksheet Ex10-14 xbar xbar1 xbar2 10.Chapter 10 Exercise Solutions 10-12.553 −0.003 0.950 587.720 ⎤ ⎢ ⎥ V′V = ⎢ −0.553 −0.256 ⎤ ⎡ 1.001 −0.392 1.016 0.016 5.001 0. Excel : workbook Chap10.016 ⎤ x′ = [15.339 0.104 88.440 −0. Excel : workbook Chap10.016 0.xls : worksheet Ex10-12 m = 40 ⎡ 4.071 ⎦ ⎣ −0.282 3.000 ⎥⎦ 10-14.101 −0.

75 1 0.308 0.9231 -0.1 with an UCL = H = 12.75 0.17 6.308 0.308 y' Sigma-1 y = 1.75 1 0.75 0. H) pair that closely minimizes ARL1 1 1.73.xls : worksheet Ex10-15 p= mu' = Sigma = y' = 4 0 0 0 0 1 0.75 0.87 UCL = H = 12.9231 -0.Chapter 10 Exercise Solutions 10-15.9231 3.9231 -0. 10-13 .0769 -0.17.53 ARL1 = Select λ = 0.75 0.75 0.9231 3.9231 y= -0.75 0.75 0.231 delta = 1.0769 -0.5 delta = 0.109 ARL0 = 200 Sigma-1 = 3. This gives an ARL1 between 7.0769 1 1 1 1 From Table 10-3.75 1 1 1 1 1 y' Sigma-1 = 0.0769 -0.9231 -0.1 0.9231 -0. Excel : workbook Chap10.9231 3.75 0.9231 -0. select (lambda.9231 -0.308 0.73 13.9231 -0.75 0.2 lambda = 12.22 and 12.

26 16.568 -2.432 -2.26.568 -2.432 -2.432 7.568 -2. select (lambda.xls : worksheet Ex10-16 p= mu' = Sigma = y' = 4 0 0 0 0 1 0.270 0.65 Select λ = 0.03 ARLmin = 14.432 -2.Chapter 10 Exercise Solutions 10-16.9 0.9 0.9 1 0.432 7.432 7.432 -2.60 7.432 -2. H) pair delta = 1 1.432 -2.432 -2.105 0.9 1 1 1 1 1 y' Sigma-1 = 0.432 -2.040 ARL0 = 500 Sigma-1 = y= 7.9 0.9 0.270 y' Sigma-1 y = 1. 10-14 .9 0.432 -2. This gives an ARLmin near 14.9 0.9 0.568 1 1 1 1 From Table 10-4.270 0.105 with an UCL = H = 15.9 1 0.18 UCL = H = 15.270 0.081 delta = 1.9 0.60. Excel : workbook Chap10.9 0.5 lambda = 0.

3 lambda = 8.2 0.5 1.556 0.49 5.054 ARL0 = 200 Sigma-1 = 2. This gives an ARL1 between 5.111 delta = 1.5 delta = 0. H) pair that closely minimizes ARL1 1 1 1.8 0.2 with an UCL = H = 9.556 y' Sigma-1 y = 1. Excel : workbook Chap10.65.7778 -2.8 1 1 1 y' Sigma-1 = 0.08 UCL = H = 10.15 10.49 and 10.2222 -2.65 10.7778 y= 1 1 From Table 10-3.20.65 9.2222 2.xls : worksheet Ex10-17 p= mu' = Sigma = y' = 2 0 0 1 0.48 ARL1 = Select λ = 0.2 0.Chapter 10 Exercise Solutions 10-17.20 5.1 0. select (lambda.64 9. 10-15 .

12 5 6 95 _ X=91. 10 TEST 5.25 90 6 85 6 6 5 5 5 5 LCL=82. 4 out of 5 points more than 1 standard deviation from center line (on one side of CL).00 standard deviations from center line.Chapter 10 Exercise Solutions 10-18. 9. 36. 2 out of 3 points more than 2 standard deviations from center line (on one side of CL). Test Failed at points: 5. 8 points in a row more than 1 standard deviation from center line (above and below CL). Test Failed at points: 9. 38. Stat > Control Charts > Variables Charts for Individuals > Individuals I Chart of y2 Output Variable (Tab10-5y2) 110 1 Individual Value 105 1 5 100 UCL=100. Test Failed at points: 40 10-16 . 39. 103. 37. 12. 35. 11. One point more than 3. 10. 37. the y2 values for Observations 8. 10. and 10 should be 100. Test Failed at points: 8. 40 TEST 6. (a) Note: In the textbook Table 10-5. and 107. 11. 9. 40 TEST 8. 39.38 80 4 8 12 16 20 24 Observation 28 32 36 40 Test Results for I Chart of Tab10-5y2 TEST 1.

57 4 8 12 16 20 24 Observation 28 32 36 40 Test Results for I Chart of Ex10-18Res TEST 1.57 5 5 Individual Value 5.. 4 out of 5 points more than 1 standard deviation from center line (on one side of CL).5 1 1 UCL=6.9 Tab10-5x8 . 21.192 Tab10-5x4 .1 Tab10-5x7 + 10. 10-17 . 2 out of 3 points more than 2 standard deviations from center line (on one side of CL). One point more than 3.5 -5.6 Tab10-5x2 + 0.00 0. 18 TEST 5.73 Tab10-5x6 + 6. Test Failed at points: 21 Plot points on the residuals control chart are spread between the control limits and do not exhibit the downward trend of the response y2 control chart.Chapter 10 Exercise Solutions 10-18 continued (b) Stat > Regression > Regression Regression Analysis: Tab10-5y2 versus Tab10-5x1. ..0.11.5 _ X=-0. Test Failed at points: 7. Test Failed at points: 19. Tab10-5x2.0 2. 25 TEST 6.0 5 LCL=-6.435 Tab10-5x3 + 0.215 Tab10-5x9 Stat > Control Charts > Variables Charts for Individuals > Individuals I Chart of Regression Model Residuals (Ex10-18Res) 7.0 -2.666 Tab10-5x1 .2 Tab10-5x5 + 0.3.00 standard deviations from center line. The regression equation is Tab10-5y2 = 215 .

2 -0.4 0.4 -0.0 -0.2 0.8 -1.8 -1.0 1 2 3 4 5 6 7 8 9 10 Lag The decaying sine wave of ACFs for Response y2 suggests an autoregressive process.4 0.6 -0.4 -0.0 -0.6 0.0 0.6 -0. while the ACF for the residuals suggests a random process.2 -0.Chapter 10 Exercise Solutions 10-18 continued (c) Stat > Time Series > Autocorrelation Autocorrelation Function for y2 Output Variable (Tab10-5y2) (with 5% significance limits for the autocorrelations) 1.8 Autocorrelation 0. 10-18 .2 0.6 0.8 Autocorrelation 0.0 0.0 1 2 3 4 5 6 7 8 9 10 Lag Autocorrelation Function for Regression Model Residuals (Ex10-18Res) (with 5% significance limits for the autocorrelations) 1.

4 5.0171 X X 3 66.1799 X 3 67.7 18.1 0.5660 X X X X X 6 67.5 64.9 6.4378 X 2 62.83550 X X X X X 6 79..3087 1 31. x8.4 3.6387 X X X 5 67..8 64.7 8.5 1.3 3.6149 X X X X X X 7 67.9 1.5 76. a regression model of input variables x1.8 56. Response is Tab10-5y2 T T T T T T T T a a a a a a a a b b b b b b b b 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 .0 4.8 76. This solution uses MINITAB’s “Best Subsets” functionality to identify the best-fitting model with as few variables as possible.9 13.6 0.6 0.5208 X X X X 5 65.9 3.6526 X X X X 6 67..0 3.6 60.5684 X X X X X 7 67.8 34.6816 X 1 35.Chapter 10 Exercise Solutions 10-19.83914 X X X X X X 7 80..0273 X X 4 72..95522 X X X 5 79.9751 X 2 50..8 6.1 52.1 3.1 75.2 3.0 0.. x3.83020 X X X X 5 73.6147 X X X 4 64.5 0.6200 X X X X X X … T a b 1 0 5 x 9 X X X X X X X X X X X X X ****** T a b 1 0 5 x 9 X X X X X X X X X X X ****** For output variables y1 and y2.4 4..3 29.. . 10-19 .2 4..0 19.8 0.95201 X X X 4 72. and x9 maximize adjusted R2 (minimize S) and minimize Mallow’s C-p. Different approaches can be used to identify insignificant variables and reduce the number of variables in a regression model.6921 2 55.1 34.0760 X 2 55.9 4.7288 X X X 3 59.2 5.4 1.8160 X X 4 64.5 5 5 5 5 5 5 5 Mallows x x x x x x x x Vars R-Sq R-Sq(adj) C-p S 1 2 3 4 5 6 7 8 1 36.9 13.9 2.0 6. x4.1 68.0 52.2 3.5 5 5 5 5 5 5 5 Mallows x x x x x x x x Vars R-Sq R-Sq(adj) C-p S 1 2 3 4 5 6 7 8 1 43. Stat > Regression > Best Subsets Best Subsets Regression: Tab10-5y1 versus Tab10-5x1..1 3.4 60.3 69.93966 X X X X 6 79.2 4.1 5. Response is Tab10-5y1 T T T T T T T T a a a a a a a a b b b b b b b b 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 .5 71.3 1.1 4.84292 X X X X X X … Best Subsets Regression: Tab10-5y2 versus Tab10-5x1.4 3.9 60.8 7.8 61.83693 X X X X X 7 80.8 62.9 1..3 76.3 3.7 3.5 24.1 24..0 4. Tab10-5x2.9 1.7 48.9 76.1 41.7 62..8 60..0 0.3 0.8 69.9 2.6 52..5 4.4 24.1 13.6 58.9 60.2 60. Tab10-5x2.7 3.1665 X X 3 61.5 36.1 12. .

633 Tab10-5x1 + 0.6 0.3 R-Sq = 67.14 0.03 1.3441 0.47 Total 39 1303..689 F 26.000 R-Sq(adj) = 76.000 0.000 Regression Analysis: Tab10-5y2 versus Tab10-5x1.001 0. The regression equation is Tab10-5y1 = 819 + 0.0.64 Tab10-5x8 + 115 Tab10-5x9 Predictor Constant Tab10-5x1 Tab10-5x3 Tab10-5x4 Tab10-5x8 Tab10-5x9 Coef 818.000 10-20 . .5% Analysis of Variance Source DF SS Regression 5 90.7% Analysis of Variance Source DF SS Regression 5 882.03 Residual Error 34 421.75 3.830201 SE Coef 29.85 P 0.Chapter 10 Exercise Solutions 10-19 continued Stat > Regression > Regression Regression Analysis: Tab10-5y1 versus Tab10-5x1.0915 Tab10-5x4 + 2.005 0. Tab10-5x3.0.236 Tab10-5x9 Predictor Constant Tab10-5x1 Tab10-5x3 Tab10-5x4 Tab10-5x8 Tab10-5x9 Coef 244.43080 -0.50 T 1.1758 11.70 3.175 -235.075 0.41 12.4540 0.176 Tab10-5x4 + 11.03530 0.056 0. The regression equation is Tab10-5y2 = 244 .1034 3.124 Tab10-5x3 .0.431 Tab10-5x1 .2 Tab10-5x8 .7 S = 3.001 0.47 -2.454 Tab10-5x3 + 0.52081 SE Coef 123.65 R-Sq = 79.4 -0.10 5.025 R-Sq(adj) = 62..000 0.9% MS 176.6329 0.08113 0..80 0.23 P 0.81 S = 0. .001 0.51 -3.7604 23.098 0. Tab10-5x3.990 Residual Error 34 23.84 3.98 -1.198 0.40 P 0.09146 2.31 -3.434 Total 39 114.47 4.001 0.40 F 14.6367 114..35 P 0.424 T 28.5% MS 18.12396 -0.225 100.02438 0.1497 0.

11 Test Results for MR Chart of Ex10-19Res1 TEST Test TEST Test 1.105 4 8 12 16 20 24 O bser vation 28 32 36 40 1 M oving Range 3 U C L=2. One point more than 3.00 standard deviations from center line. 9 points in a row on same side of center line.Chapter 10 Exercise Solutions 10-19 continued Stat > Control Charts > Variables Charts for Individuals > Individuals I-MR Chart of y1 Regression Model Residuals (Ex10-19Res1) 1 Individual V alue 2 U C L=2.000 0 2 -1 2 -2 LC L=-2. One point more than 3.791 1 0 LC L=0 2 4 8 12 16 20 24 O bser vation 28 32 36 40 Test Results for I Chart of Ex10-19Res1 TEST Test TEST Test 1.586 2 __ M R=0. Failed at points: 11 10-21 . Failed at points: 25 2. Failed at points: 10.105 1 _ X=-0. 9 points in a row on same side of center line. Failed at points: 26 2.00 standard deviations from center line.

25 TEST 6. 18 TEST 5. One point more than 3.Chapter 10 Exercise Solutions 10-19 continued I-MR Chart of y2 Regression Model Residuals (Ex10-19Res2) Individual Value 8 1 1 5 U C L=6. observations 7 and 18 are OOC) and the subset regression model (observations 7 and 18 are OOC).52 5 4 _ X=-0. Test Failed at points: 19. 10-22 . 21. One point more than 3.02 6 4 __ M R=2. Test Failed at points: 7.45 2 0 LC L=0 4 8 12 16 20 24 O bser vation 28 32 36 40 Test Results for I Chart of Ex10-19Res2 TEST 1. there is not a significant difference between control charts for residuals from either the full regression model (Figure 10-10. Test Failed at points: 7.00 standard deviations from center line. 2 out of 3 points more than 2 standard deviations from center line (on one side of CL). Test Failed at points: 26 For response y1.52 -8 4 8 12 16 20 24 O bser vation 28 32 36 40 1 M oving Range 8 U C L=8. there is not a significant difference between control charts for residuals from either the full regression model (Exercise 10-18.00 standard deviations from center line. For response y2. no out-of-control observations) and the subset regression model (observation 25 is OOC). 21 Test Results for MR Chart of Ex10-19Res2 TEST 1. 4 out of 5 points more than 1 standard deviation from center line (on one side of CL).00 0 -4 5 LC L=-6.

7SL=0.5 _ _ X=-0.1 and L = 2.7SL=1.5 +2. However the chart for y2 residuals still indicates a problem beginning near observation 20. A potential advantage to using the EWMA control chart for residuals from a regression model is the quicker detection of small shifts in the process.0 -0.0 -2.7.347 EWMA 1.435 0.0 -0. One point beyond control limits.000 0.000 0.0 1.3 EWMA 0.0 0.Chapter 10 Exercise Solutions 10-20.3 -0. Stat > Control Charts > Time-Weighted Charts > EWMA EWMA Chart of y1 Regression Model Residuals (Ex10-19Res1) 0.1 _ _ X=-0.2 -0.2 0.5 4 8 12 16 20 24 Sample 28 32 36 40 Test Results for EWMA Chart of Ex10-19Res2 TEST.347 -1.1 -0. Test Failed at points: 21.5 +2. 10-23 .4 0.7SL=-1. 22 The EWMA control chart for residuals from the response y1 subset model has no out-ofcontrol signals.5 -1. Use λ = 0.435 4 8 12 16 20 24 Sample 28 32 36 40 EWMA Chart of y2 Regression Model Residuals (Ex10-19Res2) 2.4 -2.7SL=-0.

29514 1.257 0.444 PC2 -0.253 0.86838 10-24 .36089 0.94763 0.31493 0.29428 0.19592 0.6088 0.23100 0.50471 -0.01299 -1.82157 0.3181 1.000 Variable Ex10-21X1 Ex10-21X2 Ex10-21X3 Ex10-21X4 PC1 0.580 0.46958 0.083 0.801 PC4 0.41131 -2.17976 -1.20488 0.56081 -0.14662 Ex10-21z2 -0.19166 -0.08767 1. Ex10-21X2.330 0.70241 -0.21950 2.33164 0.13529 -1.0118 0.31445 -0.23823 -0.17849 Ex10-21z3 0.58815 1.11243 0.94470 -1.685 -0.20327 -0.19734 0.99211 -1. (a) Stat > Multivariate > Principal Components Note: To work in standardized variables in MINITAB. Ex10-21X4 Eigenanalysis of the Correlation Matrix Eigenvalue 2.11787 -1.65608 2.09129 -0.14246 -0.27825 -2.Chapter 10 Exercise Solutions 10-21. Ex10-21X3.20787 0.32286 0.0613 Proportion 0. select Correlation Matrix.03929 0.64094 1.594 0.47259 -0.99498 0. select Storage and enter columns for Scores.334 -0.30494 -0.02496 1.534 -0.87917 -2.21893 -1.42176 -0.580 0.12378 -1.794 0.015 Cumulative 0.67046 -0.286 0.061 0. Principal Component Analysis: Ex10-21X1. Note: To obtain principal component scores.832 0.607 0.104 Principal Component Scores Ex10-21z1 0.60867 -0.36528 0.387 PC3 0.985 1.10423 -0.60340 0.12420 0.62545 -0.64480 -0.83902 3.49153 0.718 -0.152 0.29168 0.

See Excel : workbook Chap10.9 0. 10-25 . Ex10-21z2.5 -2 0 2 -0.5 0.xls : worksheet Ex10-21.5 Ex1 0 -2 1 z2 all -3.0 -0.0 -0. the majority of new observations are clearly different from the original observations. Graph > Matrix Plot > Matrix of Plots with Groups Matrix Plot of Ex10-21z1all.0 Ex10-21Obs New Original 4 Ex1 0 -2 1 z1 all 0 -4 2.0 2 Ex1 0 -2 1 z3 all 0 -2 -4 0 4 -2 0 2 Although a few new points are within area defined by the original points.7 1.Chapter 10 Exercise Solutions 10-21 continued (b) Graph > Matrix Plot > Simple Matrix of Plots Matrix Plot of Ex10-21z1.7 Ex10-21z3 -0. Ex10-21z3all Principal Component Scores -3.5 2. Ex10-21z3 Principal Component Scores -2 0 2 2 0 Ex10-21z1 -2 2 0 Ex10-21z2 -2 1. Ex10-21z2all.9 (c) Note: Principal component scores for new observations were calculated in Excel.

(a) Stat > Multivariate > Principal Components Note: To work in standardized variables in MINITAB.000 PC9 0.513 -0.021 0.240 0.409 -0.156 -0.431 -0.3121 0.166 -0.579 0. Ex10-22x3. Note: To obtain principal component scores.323 0.050 0.267 0. (c) Graph > Matrix Plot > Simple Matrix of Plots Matrix Plot of Ex10-22z1.651 -0. ….232 0.244 -0. Ex10-22z3 Principal Component Scores -3 0 3 3 0 Ex10-22z1 -3 3 Ex10-22z2 0 -3 2 0 Ex10-22z3 -2 -3 0 3 -2 0 2 10-26 .028 0.407 0.349 0.022 0.3292 1.529 0.238 PC3 -0.380 0.105 0.022 -0. select Correlation Matrix.844 Variable Ex10-22x1 Ex10-22x2 Ex10-22x3 Ex10-22x4 Ex10-22x5 Ex10-22x6 Ex10-22x7 Ex10-22x8 Ex10-22x9 PC1 -0.048 0.018 0.508 0.471 0.357 0.997 PC8 0.074 -0.975 PC7 0.723 0. Ex10-22x9 Eigenanalysis of the Correlation Matrix Eigenvalue 3.525 -0.199 0.117 Cumulative 0.099 -0.7% of the variability is explained by the first 3 principal components.120 0.006 0.163 -0.000 0.372 0.230 0.003 1.297 0.912 PC5 0.032 -0.349 0.332 -0.124 -0.152 -0.022 0.068 0.014 0.127 -0.148 0.727 0.1407 2.448 PC2 0.089 0.262 0.202 -0.425 -0.162 0.175 0. Ex10-22x2.602 -0.108 0.135 0.204 -0.247 0.193 -0.0520 Proportion 0.465 0.0730 1.123 0. select Storage and enter columns for Scores.436 -0.419 0.262 0. Ex10-22z2.200 0.256 0.874 0.068 0.Chapter 10 Exercise Solutions 10-22.6129 0.579 -0.2542 0.322 0.1973 0.632 0.261 -0.368 0.575 0.467 -0.947 PC6 -0.662 -0.035 0.133 (b) 72.0287 0.115 PC4 -0.277 0.406 0.188 0.090 -0.035 0.145 -0. Principal Component Analysis: Ex10-22x1.089 0.

Ex10-22z3all All Principal Component Scores -4 0 4 3 Ex10-22Obs First Last 0 Ex1 0 -2 2 z1 all -3 4 0 Ex1 0 -2 2 z2 all -4 2 0 Ex1 0 -2 2 z3 all -2 -3 0 3 -2 0 2 Several points lie outside the area defined by the first 30 observations.Chapter 10 Exercise Solutions 10-22 continued (d) Note: Principal component scores for new observations were calculated in Excel. indicating that the process is not in control. See Excel : workbook Chap10. 10-27 . Graph > Matrix Plot > Matrix of Plots with Groups Matrix Plot of Ex10-22z1all. Ex10-22z2all.xls : worksheet Ex10-22.

Chapter 11 Exercise Solutions 11-1. yt : observation zt : EWMA (a) zt = λ yt + (1 − λ ) zt −1 zt = λ yt + zt −1 − λ zt −1 zt − zt −1 = λ yt + zt −1 − zt −1 − λ zt −1 zt − zt −1 = λ yt − λ zt −1 zt − zt −1 = λ ( yt − zt −1 ) (b) zt −1 − zt − 2 = λ et −1 (as a result of part (a)) zt −1 − zt − 2 + (et − et −1 ) = λ et −1 + (et − et −1 ) zt −1 + et − zt − 2 − et −1 = et − (1 − λ )et −1 yt − yt −1 = et − (1 − λ )et −1 11-11 .

975 -13.625 12.975 -13.975 -13.134 no yes no yes no no no no yes 0.127 -6.76.0 40 -8.0 L = -10 -10 Adjustment Scale -4.854 0.0 -50 1 6 11 16 21 26 31 36 41 46 Obs Orig_out Adj_out_t EWMA_t Adj_Obs_t+1 Chart with λ = 0.000 0.375 -18.993 no no no no no no 0.625 7.975 -10.375 -18.625 19. 11-12 .975 -13.xls : worksheet Ex 11-2 0 0.0 -9.975 … SS = Average = 21468 17.8 T= lambda = L= g= Obs Orig_out Orig_Nt Adj_out_t EWMA_t |EWMA_t|>L? Adj_Obs_t+1 Cum_Adj 1 2 3 4 5 6 7 8 9 10 0 16 24 29 34 24 31 26 38 29 0 16 8 5 5 -10 7 -5 12 -9 16 24 20.000 25.000 -3.984 0.0 20 -2.0 0 2.560 6.000 0.0 -18.0 -9.975 -1.969 5.690 Bounded Adjustment Chart for Ex 11-2 50 -12.625 10.0 12.975 -13.975 -1.923 10.025 -22.800 10. The chart with λ = 0.700 1.318 -7.0 0.0 -30 8.179 -6.0 0.000 0.000 5.375 -18.359 45 46 47 48 49 50 22 -9 3 12 3 12 9 -31 12 9 -9 9 8.0 -20 6.0 -9.375 -18.Chapter 11 Exercise Solutions 11-2.625 4.766 9.000 0.000 0.3 exhibits less variability and is closer to target on average.982 -8. Excel : workbook Chap11.000 0.000 0.715 -5.2 gives SS = 9780 and average deviation from target = 1.688 4.0 30 -6.000 0.975 -10.975 -0.24 6526.3 10 0.0 -40 10.375 0.000 -13.0 -9.000 11.0 L = +10 10 4.0 -10.375 -22.

5 -11. Excel : workbook Chap11.5 0. 11-13 .061 -7.288 no no no no no Adj_Obs_t+1 Cum_Adj 0 -12 0 -11 0 0 0 0 0 0 -12 -12 -23 -23 -23 -23 -23 -23 0 0 0 0 0 -11.Chapter 11 Exercise Solutions 11-3.4 exhibits less variability.5 -6.5 … SS = Average = 21468 17.8 Obs Orig_out Orig_Nt Adj_out_t EWMA_t |EWMA_t|>L? 1 2 3 4 5 6 7 8 9 10 0 16 24 29 34 24 31 26 38 29 0 16 8 5 5 -10 7 -5 12 -9 16 24 17 22 1 8 3 15 6 6.25 0.91 Bounded Adjustment Chart for Ex 11-3 80 -15 70 60 -10 50 40 30 L = +10 10 0 0 L = -10 -10 -20 Adjustment Scale -5 20 5 -30 -40 -50 10 -60 -70 15 -80 1 6 11 16 21 26 31 36 41 46 Obs Orig_out Adj_out_t EWMA_t Adj_Obs_t+1 The chart with λ = 0.264 7.037 -4.813 -3.5 -8.400 3.5 -8.5 0.958 7.022 -5.5 -11.440 6.5 -11.24 5610.xls : worksheet Ex 11-3 Target yt = lambda = L= g= 0 0.175 no yes no yes no no no no no 46 47 48 49 50 -9 3 12 3 12 -31 12 9 -9 9 -20.400 13.4 10 0.5 -11.880 0.800 12.3.440 3. but is further from target on average than for the chart with λ = 0.

6 -0.2 -1.0 … SS = Average = 21468 17.4 1.0 -14.0 20.9 -23.9 0.0 10.0 -20.8 -3.3 -22.Chapter 11 Exercise Solutions 11-4.9 5.9 -6.0 -1.3 -2.7 -6.0 -5.5 -7.xls : worksheet Ex 11-4 T= lambda = g= 0 0.3 -7.3 -13.3 1.3 -27.9 5.8 3.6 1.7 45 46 47 48 49 50 22 -9 3 12 3 12 9 -31 12 9 -9 9 11.6 -22.7 5.7 -0.8 Obs Orig_out Adj_out_t Orig_Nt Adj_Obs_t+1 Cum_Adj 1 2 3 4 5 6 7 8 9 10 0 16 24 29 34 24 31 26 38 29 0 16 8 5 5 -10 7 -5 12 -9 16.5 -4. 11-14 .0 -5.2 0. Excel : workbook Chap11.3 -2.24 5495.0 -5.3 -4.4.7 Integral Control for Ex 11-4 50 -12 -10 40 -8 30 -6 20 -2 0 0 2 -10 Adjustment Scale -4 10 4 -20 6 -30 8 -40 10 -50 12 1 6 11 16 Orig_out 21 26 Adj_out_t 31 36 41 46 Adj_Obs_t+1 The chart with process adjustment after every observation exhibits approximately the same variability and deviation from target as the chart with λ = 0.0 -9.0 20.7 -27.7 -4.0 20.0 5.7 -8.1 14.0 -19.2 -1.4 -4.

Chapter 11 Exercise Solutions 11-5.500 2.002 1.72 147.47 179.53 138.11 175.000 1.370 0.02 136.43 201.000 0 5 10 15 20 25 Vm/V1 11-15 .000 0.029 1.000 m 1.104 1.xls : worksheet Ex 11-5 t Yt 1 2 3 4 5 6 7 8 9 10 … / m => 0 16 24 29 34 24 31 26 38 29 Var_m = Var_m/Var_1 = 1 2 3 4 5 6 7 8 9 16 8 5 5 -10 7 -5 12 -9 24 13 10 -5 -3 2 7 3 29 18 0 2 -8 14 -2 34 8 7 -3 4 5 24 15 2 9 -5 31 10 14 0 26 22 5 38 13 29 147.195 1. Excel : workbook Chap11.929 1.217 0.500 1.60 151.39 162.943 Variogram for Ex 11-5 2.500 0.70 1.

325056 0.01 2.05 0.299822 0.2 -0.59 0.59 54.33 0.4 0. so the observations are correlated and there may be some mild indication of nonstationary behavior.066173 T 3.mtw : Yt Stat > Time Series > Autocorrelation Function Autocorrelation Function for Data in Table 11-1 (Yt) (with 5% significance limits for the autocorrelations) 1.12 2.345327 0.6 -0.0 -0.2 0.316478 0.28 0.164698 0.90 0.41 63.149321 0.389094 0.440855 0.03 55.68 1. The slow decline in the sample ACF also indicates the data are correlated and potentially nonstationary.39 27.75 Variogram appears to be increasing.54 1.86 63.55 48.0 1 2 3 4 5 6 7 Lag 8 9 10 11 12 13 Autocorrelation Function: Yt Lag 1 2 3 4 5 6 7 8 9 10 11 12 13 ACF 0.012158 0.0 0.85 1.440819 0.6 0.31 16.87 67.228540 0.45 1.80 41.4 -0.71 62.26 LBQ 10.8 Autocorrelation 0.58 1.44 67.334961 0. 11-16 .Chapter 11 Exercise Solutions 11-5 continued MTB : Chap11.8 -1.14 32.

0 129.8 -1.0 -5.8 162.0 15.5 -7.9 180.8 0.2 T= lambda = g= Obs.4 1.0 110.0 30.0 185.0 200.xls : worksheet Ex 11-6a 200 0.0 1 6 11 16 21 26 31 36 41 46 Obs.5 49 50 SS = Average = Variance = 11.5 Unadjusted 1.0 260.0 12.7 4.0 -0. t Orig_out Adj_out_t Adj_Obs_t+1 Significant reduction in variability with use of integral control scheme.1 177.5 208.0 -10.0 Adjustment Scale Integral Control for Ex 11-6(a) 20.8 0.7 0.7 220.510.3 -4.0 1.818.9 12.0 10.4 4.9 320.871.4 170.9 -39.0 0.2 1.Chapter 11 Exercise Solutions 11-6.1 13. (a) and (b) Excel : workbook Chap11.6 -13.4 62.0 80. 11-17 .0 187.0 63.0 230.8 -15.5 176.0 140.0 195.7 191.3 2.0 5.4 191.3 2.8 … 145.8 176.4 Adjusted 1.5 3.7 187.4 0.0 -15.1 -3.323.2 1.7 3.3 192.8 -20.1 4.0 193.0 170.0 195.0 -20.2 7.3 467.9 200.0 290.3 -6.2 160.2 0.5 192.5 24. t 1 2 3 4 5 6 7 8 9 10 Orig_out Orig_Nt Adj_out_t Adj_Obs_t+1 Cum_Adj 215.8 161.7 180.1 216.

0 198. 11-18 .4 176.888.0 260.0 -0.0 290.3 -6.0 9.5 176.0 6.0 -10.0 195.5 216.8 161.2 T= lambda = g= Obs.0 5.9 183.0 -40.5 13.6 -2.3 -4.8 7.0 140.8 -20.0 10.9 18.0 195.4 18.0 4.0 Orig_out Orig_Nt Adj_out_t Adj_Obs_t+1 Cum_Adj 215.323.1 176.0 170.5 5.8 185.0 1 6 11 16 21 26 31 36 41 46 Obs.5 -7.5 24.0 30.2 and λ = 0.0 110.0 195.4 Adjusted 1.4 175.0 -20.0 189.9 164.0 320. t Orig_out Adj_out_t Adj_Obs_t+1 Variances are similar for both integral adjustment control schemes (λ = 0.5 7.8 0.4 1.4 3.8 -0.0 2.0 7.0 -30.Chapter 11 Exercise Solutions 11-6 continued (c) Excel : workbook Chap11.0 20.1 187.4 191.3 467.0 3.0 10.7 2.5 192.0 66.0 30.2 180.9 2.9 -39. t 1.0 8.4).0 162.5 Unadjusted 1.6 -13.7 0.4 1.6 8.7 223.0 Adjustment Scale Integral Control for Ex 11-6(c) 40.5 -7.0 0.0 SS = Average = Variance = -15.871.8 193.1 3.7 … 129.6 7.995.8 1.0 80.1 21.0 230.0 200.xls : worksheet Ex 11-6c 200 0.5 50.9 208.

000 6.9 -39.1 176.0 140.5 -7 186.000 0.0 0.0 -20.3 -4.516 48.8 SS = Average = Variance = 165. Excel : workbook Chap11.600 -14.0 230.5 24.0 80.0 170.516 48.000 0.5 1.733 -1.900 -5.696 176.656 |EWMA_t|>L? Adj_Obs_t+1 Cum_Adj no no no no no no yes no no 0.0 1 6 11 16 21 26 31 36 41 46 Obs.2 145.840 191. t Orig_out Adj_out_t EWMA_t Adj_Obs_t+1 Behavior of the bounded adjustment control scheme is similar to both integral control schemes (λ = 0.5 46 47 48 49 50 -7 3.300 -4.000 0.364 1.8 -15.717 0.941 162.000 48.9 193.0 10.000 6.7 320.0 0.5 191.0 Adjustment Scale Bounded Adjustment Chart for Ex 11-7 20.016 -12.4).969 -5.459 -10.8 0 195.2 Target yt = lambda = L= g= Obs.265 182.3 -6 185.000 -0.8 -20 196 -0.0 30.870 216.233 6.3 7 11.516 178. t 1 2 3 4 5 6 7 8 9 10 Orig_out Orig_Nt Adj_out_t EWMA_t 215.233 0.412 185.253 180.4 162.516 … 122.000 0.000 0.0 -10.872 161.300 -2.516 48.632.6 -13.051 172. 11-19 .0 5.0 15.000 0.233 6.000 0.9 176.000 0.0 -15.0 0.xls : worksheet Ex 11-7 200 0.233 yes no no no no 5.699 174.323.7 216.057 -7.0 -0.000 0.000 0.9 126.000 0.2 and λ = 0.0 260.0 129.Chapter 11 Exercise Solutions 11-7.177 176.0 -5.0 290.716 181.516 48.0 200.0 110.0 0.2 133.733 -3.000 -8.702 -7.633 187.716 193.304 467.000 0.2 12 1.

680 191.0 -32.000 0.000 0.4 15 1.0 170.000 0.467 12.821 |EWMA_t|>L? Adj_Obs_t+1 Cum_Adj no no no no no no yes no no 0.784 170.958 203.467 no no no no no 0.758 188.300 -4.758 58.5 46 47 48 49 50 SS = Average = Variance = -7 3.0 -48.0 -24.000 0.323.0 16.5 -7 192.0 40.0 140.2 145.000 0.692 162.0 200.849 -10.Chapter 11 Exercise Solutions 11-8.9 -39.0 50.0 32.467 12.488 185.693 -7.0 260.000 0.326 -4. 11-110 .3 7 11.775 187.872 161.600 -23. t 1 2 3 4 5 6 7 8 9 10 Orig_out Orig_Nt Adj_out_t EWMA_t 215.000 0.7 216.958 191.300 -8.967 -0.5 191.9 176.9 199.6 -13.81 -9.758 58.0 -0.0 0.0 -40.1 176.0 0.000 0.0 1 6 11 16 21 26 31 36 41 46 Obs.000 0.000 0.467 0.773.758 58.3 -6 185.000 58.000 1.5 181.2 T= lambda = L= g= Obs.258 1.8 0 195.013 180.5 24.9 126.000 -14. t Orig_out Adj_out_t EWMA_t Adj_Obs_t+1 Behavior of both bounded adjustment control schemes are similar to each other and simlar to the integral control schemes.0 Adjustment Scale -16.000 0.900 -8.758 … 122.000 12.000 0.967 -2.256 176. Excel : workbook Chap11.758 58.083 189.000 0.2 133.513 1.8 -15.486 176.0 320.573 216.0 0.xls : worksheet Ex 11-8 200 0.8 -20 196 -1.86 Bounded Adjustment Chart for Ex 11-8 350.658 184.0 30.0 24.720 -11.0 80.0 230.3 -4.000 12.0 48.0 110.0 129.0 -8.0 290.304 467.0 8.0 0.4 162.

0 74 -3.4 54.6 -3.0 45.0 18 4.0 -11. (a) and (b) Excel : workbook Chap11.9 … SS = Average = Variance = Adjusted 108.7 64.0 3.0 26 3.Chapter 11 Exercise Solutions 11-9.2 -1.0 -13.0 Adjustment Scale Integral Control for Ex 11-9 (a) 5.0 -1.4 223.0 -4.4 0.7 22.7 0.6 -1.3 -1.7 49 50 23 26 3.6 0.8 -0.629 46.8 -3.0 0.2 51.0 58.0 53.520 44.0 -0.0 10.0 -9.2 50.xls : worksheet Ex 11-9a T= lambda = g= 50 0.8 -0.2 22.0 66 -2.51 Significant reduction in variability with use of an integral control scheme.6 -1.0 10 1 6 11 16 21 26 31 36 41 46 Obs.2 1.9 0. 90 -5.1 48.3 -1.0 42 1.0 -1.0 11.0 58 -1.8 -5.4 65.0 50 0.0 82 -4.6 Obs Orig_out Adj_out_t Orig_Nt Adj_Obs_t+1 Cum_Adj 1 2 3 4 5 6 7 8 9 10 50 58 54 45 56 56 66 55 69 56 8.7 -5.262 78.32 Unadjusted 109.0 43. t Orig_out Adj_out_t Adj_Obs_t+1 11-111 .6 55.0 34 2.0 14.4 -0.7 -0.

3 -5.0 63.0 54.5 45 -9.2 56 -13.6 56 11.4 … 49 50 SS = Average = Variance = 23 26 109. 11-112 .0 3. with a process average slightly closer to the target of 50.5 53.0 52.4 1.1 -3.3 -2.8 0.40 There is a slight reduction in variability with use of λ = 0.833 56.8 27.0 55.4 223.0 -2.51 3.5 114.819 47.0 56 0.5 1.6 Obs 1 2 3 4 5 6 7 8 9 10 Orig_out Orig_Nt Adj_out_t Adj_Obs_t+1 Cum_Adj 50 58 8.8 -8.3 -9.0 -1.5 -2.0 50.0 66 10.0 42.520 44.xls : worksheet Ex 11-9c T= lambda = g= 50 0.4 -1.4 26.0 58.0 -3.0 48.0 -0.0 46.2 55 -11. as compared to λ = 0.2.9 69 14.4.0 63.3 -6.9 -0.0 -3.4 -0.8 0.1 -0.Chapter 11 Exercise Solutions 11-9 continued (c) Excel : workbook Chap11.0 54 -4.0 -2.

0 18 8.0 26 6.0 82 -8.0 34 4.0 58 -2.0 42 2.0 1 6 11 16 21 26 31 36 41 46 Obs.Chapter 11 Exercise Solutions Integral Control for Ex 11-9(c) 90 -10.0 50 0.0 74 -6.t Orig_out Adj_out_t Adj_Obs_t+1 11-113 .0 10 10.0 66 -4.

0 82 -4.520 44.00 0.40 0.0 66 -2.00 -2.79 15.505 -5.00 no yes no yes no 0.51 SS = Average = Variance = -2.842 -4.080 45 -9 45.822 45.0 0. Excel : workbook Chap11. t 1 2 3 4 5 6 7 8 9 10 Orig_out Orig_Nt Adj_out_t EWMA_t 50 0 58 8 58 1.4 223.0 26 3.00 0.00 0.664 56 11 56.00 3.00 3.79 43.731 56 0 56. with similar means and sums of squares. 11-114 .79 17.00 0.6 Target yt = lambda = L= g= Obs.00 0.0 34 2.0 18 4.0 2 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 Obs.362 107.0 Adjustment Scale Bounded Adjustment Chart for Ex 11-10 6.79 38.xls : worksheet Ex 11-10 50 0.00 1.19 … 24 18 20 23 26 46 47 48 49 50 8 -6 2 3 3 37.00 0.00 -2.00 2.0 42 1.880 56 -13 54.00 13.0 0.0 50 0.620 121.00 -2.2 4 1.19 17.0 58 -1.00 0.00 -2.0 90 -5. but different variances (bounded adjustment variance is larger).00 0.48 31.32 0.72 98 -6.268 55 -11 53.0 74 -3.00 0.00 1.904 |EWMA_t|>L? Adj_Obs_t+1 Cum_Adj no no no no no yes no no no 0.00 0.600 54 -4 54.48 15.00 -2.0 10 5.585 66 10 66.709 -2.19 109.00 2.48 35.00 0.600 69 14 67.00 5.Chapter 11 Exercise Solutions 11-10.515 -1.00 2. t Orig_out Adj_out_t EWMA_t Adj_Obs_t+1 Nearly the same performance as the integral control scheme.00 0.

This experiment is three replicates of a factorial design in two factors—two levels of glass type and three levels of phosphor type—to investigate brightness. 12-1. select Stat > DOE > Factorial > Define Custom Factorial Design. therefore some information required by MINITAB is not included. This is how the Excel file is structured (Chap12. and one column for brightness. The remaining exercises are worked in a similar manner. the initial experimental layout must be created in MINITAB or defined by the user. After entering the data in MINITAB. the design must be defined before the results can be analyzed.xls).Chapter 12 Exercise Solutions Note: To analyze an experiment in MINITAB. The Excel data sets contain only the data given in the textbook. then for this exercise. and only the solutions are provided. Detailed MINITAB instructions are provided for Exercises 12-1 and 12-2 to define and create designs. one column for phosphor type. check “General full factorial”. Enter the data into the MINITAB worksheet using the first three columns: one column for glass type. Since the experiment layout was not created in MINITAB. The dialog box should look: 12-1 . Select the two factors (Glass Type and Phosphor Type).

Note that MINITAB added four new columns (4 through 7) to the worksheet. and click “OK” twice. MINITAB recognizes these contiguous seven columns as a designed experiment. run order. The design and data are in the MINITAB worksheet Ex12-1. For this exercise.Chapter 12 Exercise Solutions 12-1 continued Next.MTW. select “Designs”. or blocks. point type. no information is provided on standard order. so leave the selections as below. DO NOT insert or delete columns between columns 1 through 7. inserting or deleting columns will cause the design layout to become corrupt. 12-2 .

Ex12-1Phosphor Factor Ex12-1Glass Ex12-1Phosphor Type fixed fixed Levels 2 3 Values 1.44% No indication of significant interaction (P-value is greater than 0.3 466. Click on “Graphs”.000 Ex12-1Phosphor 2 933. select “Fits” and “Residuals”.08% R-Sq(adj) = 94. 2 1. The option to plot residuals versus variables is for continuous factor levels. Glass type (A) and phosphor type (B) significantly affect television tube brightness (P-values are less than 0.7 8. select “Residuals Plots : Four in one”. and their interaction. Click on “Storage”.84 0.3 133.318 Error 12 633.79 0.10). Phosphor Type. using Adjusted SS for Tests Source DF Seq SS Adj SS Adj MS F P Ex12-1Glass 1 14450.0 14450. then click on “Terms”.3 66.004 Ex12-1Glass*Ex12-1Phosphor 2 133. Click “OK”.3 52. and click “OK” twice.0 14450. click “OK”. Select the response (Brightness).0 273.8 Total 17 16150. 3 Analysis of Variance for Ex12-1Bright.0 S = 7. do not select this option. 2.26483 R-Sq = 96.26 0.3 933. verify that the selected terms are Glass Type.3 633. since the factor levels in this experiment are categorical.Chapter 12 Exercise Solutions 12-1 continued Select Stat > DOE > Factorial > Analyze Factorial Design. Residual Plots for Ex12-1Bright Normal Probability Plot of the Residuals Residuals Versus the Fitted Values 99 10 Residual Percent 90 50 0 10 -10 1 -10 0 Residual 10 220 Histogram of the Residuals 240 260 280 Fitted Value 300 Residuals Versus the Order of the Data 10 3 Residual Frequency 4 2 0 1 0 -10 -10 -5 0 5 Residual 10 15 2 4 6 8 10 12 14 Observation Order 16 18 12-3 .10). General Linear Model: Ex12-1Bright versus Ex12-1Glass.7 1.

To plot residuals versus the two factors. then click “OK” twice. If the model were re-fit with only Glass Type and Phosphor Type. histogram. Select the column with stored residuals (RESI1) as the Graph variable and select one of the factors (Glass Type or Phosphor Type) as the Categorical variable for grouping.Chapter 12 Exercise Solutions 12-1 continued Visual examination of residuals on the normal probability plot. and versus fitted values reveals no problems. Individual Value Plot of RESI1 vs Ex12-1Glass 15 RESI1 10 5 0 0 -5 -10 1 2 Ex12-1Glass Individual Value Plot of RESI1 vs Ex12-1Phosphor 15 RESI1 10 5 0 0 -5 -10 1 2 Ex12-1Phosphor 3 12-4 . Click on “Scale”. select Graph > Individual Value Plot > One Y with Groups. select the “Reference Lines” tab. the residuals should be re-examined. and enter “0” for the Y axis. The plot of residuals versus observation order is not meaningful since no order was provided with the data.

12-5 . however. Individual Value Plot of RESI1 vs Ex12-1Glass 15 RESI1 10 5 0 0 -5 -10 1 2 Ex12-1Glass Individual Value Plot of RESI1 vs Ex12-1Phosphor 15 RESI1 10 5 0 0 -5 -10 1 2 Ex12-1Phosphor 3 Variability appears to be the same for both glass types. To remove the jitter. select the graph to make it active then: Editor > Select Item > Individual Symbols and then Editor > Edit Individual Symbols > Jitter and de-select Add jitter to direction. there appears to be more variability in results with phosphor type 2.Chapter 12 Exercise Solutions 12-1 continued Note that the plot points are “jittered” about the factor levels.

Final selected combination of glass type and phosphor type depends on the desired brightness level. Select “Interaction Plot” and click on “Setup”. select the response (Brightness) and both factors (Glass Type and Phosphor Type). 12-6 .Chapter 12 Exercise Solutions 12-1 continued Select Stat > DOE > Factorial > Factorial Plots. Interaction Plot (data means) for Ex12-1Bright 310 Ex12-1Glass 1 2 300 290 Mean 280 270 260 250 240 230 220 1 2 Ex12-1Phosphor 3 The absence of a significant interaction is evident in the parallelism of the two lines. and click “OK” twice.

0 280. Ex12-1Phosphor Source DF SS MS F P Ex12-1Glass 1 14450.318 Error 12 633.0 273. To obtain results which match the output in the textbook’s Table 12.0 14450. The DOE functionality was selected to illustrate the approach that will be used for most of the remaining exercises.265 R-Sq = 96.3 52.79 0.84 0.000 Ex12-1Phosphor 1 2 3 Individual 95% CIs For Mean Based on Pooled StDev -----+---------+---------+---------+---(--*-) (--*-) -----+---------+---------+---------+---240 260 280 300 Mean 260.000 273. and complete the dialog box as below.44% Ex12-1Glass 1 2 Mean 291.3 466.333 256.Chapter 12 Exercise Solutions 12-1 continued Alternate Solution: This exercise may also be solved using MINITAB’s ANOVA functionality instead of its DOE functionality.0 S = 7. select Stat > ANOVA > Two-Way.26 0.7 8.8 Total 17 16150. Two-way ANOVA: Ex12-1Bright versus Ex12-1Glass.0 272.0 12-7 .0 264.3 66.08% R-Sq(adj) = 94.000 Ex12-1Phosphor 2 933.667 235.7 1.5.004 Interaction 2 133.667 Individual 95% CIs For Mean Based on Pooled StDev -------+---------+---------+---------+-(-------*-------) (-------*-------) (-------*-------) -------+---------+---------+---------+-256.

62 12.25 -59.40 Cutting Angle S = 49.41 -1.000 0.13 12.41 0.MTW.12 12.45 Cutting Speed*Cutting Angle -119. This solution uses the first approach. Cutting Angle.41 2.41 3.41 33.30 Cutting Speed 18. one approach to solving this exercise is to create a 23 factorial design in MINITAB. Since the standard order (Run) is provided.357 0. The design and data are in the MINITAB worksheet Ex12-2. a reduced model in Metal Hardness.009 ** 0. Select “Factors”.41 -4. Select “Designs”.41 -0. ABC) are included. C.81 8.001 ** 0. The worksheet is in run order. (a) To analyze the experiment. B. a full factorial model is not necessary.25 -5. then define a customer factorial design.Chapter 12 Exercise Solutions 12-2. Cutting Speed will also be retained to maintain a hierarchical model. Factorial Fit: Life versus Cutting Speed.89 Cutting Speed*Metal Hardness -11. highlight full factorial. 12-8 . Metal Hardness.12 12.40 Cutting Angle 71.56% Analysis of Variance for Life (coded units) Source DF Seq SS Adj SS Adj MS Main Effects 3 50317 50317 16772 2-Way Interactions 3 59741 59741 19914 3-Way Interactions 1 4830 4830 4830 Residual Error 8 19700 19700 2462 Pure Error 8 19700 19700 2463 Total 15 134588 … F 6.41 -0.25 9. Select “Terms” and verify that all terms (A. AB. Leave the design type as a 2-level factorial with default generators.10.74 Metal Hardness 84. select Stat > DOE > Factorial > Analyze Factorial Design.008 0.199 Based on ANOVA results.13 12. Another approach would be to create a worksheet containing the data.6236 R-Sq = 85.88 12. AC.98 Cutting Speed*Metal Hardness* -34.75 35. Select Stat > DOE > Factorial > Create Factorial Design. BC. and click “OK” twice. then enter the data.62 12.483 0. and change the Number of factors to “3”.199 R-Sq(adj) = 72. enter the factor names.96 P 0. to change to standard order (and ease data entry) select Stat > DOE > Display Design and choose standard order. Cutting Angle Estimated Effects and Coefficients for Life (coded units) Term Effect Coef SE Coef T Constant 413.75 -17.014 0. and Cutting Speed*Cutting Angle is more appropriate.09 1.662 0.25 42.25 -12.37 12. leave factor types as “Numeric” and factor levels as -1 and +1.36% P 0.81 Metal Hardness*Cutting Angle -24. change number of replicates to “2”. Based on P-values less than 0. and click “OK”. Both approaches would achieve the same result.020 ** 0.

015 0.25% Analysis of Variance for Life (coded units) Source DF Seq SS Adj SS Adj MS Main Effects 3 50317 50317 16772 2-Way Interactions 1 56882 56882 56882 Residual Error 11 27389 27389 2490 Lack of Fit 3 7689 7689 2563 Pure Error 8 19700 19700 2463 Total 15 134588 F 6. Choose and set-up a “Cube Plot”.0 380. Cutting Angle Estimated Effects and Coefficients for Life (coded units) Term Effect Coef SE Coef T Constant 413.85 P 0.5 1 266.47 2.13 12.0 512.47 3. B+ and C+.0 -1 1 -1 Cutting Speed Cutting Angle -1 Longest tool life is at A-.65% P 0.480 0.78 S = 49. for an average predicted life of 552.88 12.5 351.001 1.62 12.006 0.38 Cutting Angle 71.25 42. (c) From examination of the cube plot. Metal Hardness.0 1 Metal Hardness 446.425 (b) The combination that maximizes tool life is easily seen from a cube plot.008 0.47 0.8988 R-Sq = 79.5 405.25 -59. Select Stat > DOE > Factorial > Factorial Plots.13 12.47 -4. Cube Plot (data means) for Life Exercise 12-2(b) 552.25 9.000 0.88 Cutting Speed*Cutting Angle -119. we see that the low level of cutting speed and the high level of cutting angle gives good results regardless of metal hardness.12 12.73 Metal Hardness 84.04 0.001 R-Sq(adj) = 72.74 22.5.12 Cutting Speed 18.5 391. 12-9 .Chapter 12 Exercise Solutions 12-2(a) continued Factorial Fit: Life versus Cutting Speed.47 33.75 35.

Select “Terms” and verify that all terms for the reduced model (A. To find the residuals.Chapter 12 Exercise Solutions 12-3. 12-10 . Select “Graphs”. select “Storage” and choose “Residuals”. and for residuals plots choose “Normal plot” and “Residuals versus fits”. Normal Probability Plot of the Residuals (response is Life) 99 95 90 Percent 80 70 60 50 40 30 20 10 5 1 -100 -50 0 Residual 50 100 Residuals Versus the Fitted Values (response is Life) 50 Residual 25 0 -25 -50 -75 250 300 350 400 Fitted Value 450 500 550 Normal probability plot of residuals indicates that the normality assumption is reasonable. AC) are included. select Stat > DOE > Factorial > Analyze Factorial Design. Residuals versus fitted values plot shows that the equal variance assumption across the prediction range is reasonable. C. To save residuals to the worksheet. B.

The design and data are in the MINITAB worksheet Ex12-4.31 Sweetener -9.9504 -1.722 0.188 -0.500 0.MTW.9504 -0.28 1.055 0.094 0.771 0.9504 -0.89 Sweetener*Syrup to Water*Carbonation -5.30 Syrup to Water*Temperature -0.28 Carbonation*Temperature Analysis of Variance for Total Score (coded units) Source DF Seq SS Adj SS Adj MS F Main Effects 4 852.10 Carbonation*Temperature 1.629 0.10.47 P 0.13 405.094 0.9504 2.438 1. along with interactions involving both sweetener and the ratio of syrup to water.9504 -0.000 0.15 Syrup to Water*Carbonation -0.9504 -0.9504 2.9504 -1. sweetener*temperature.219 0.938 -0.781 0.469 0.9504 -0.938 1.344 0.001 0. 12-11 .9504 0.267 0. select “Terms” and verify that all terms are selected.629 * * * * * 0.Chapter 12 Exercise Solutions 12-4..379 0.69 199.36 Sweetener*Temperature -2.500 28.28 3. temperature. type of sweetener is dominant.218 From magnitude of effects. the reduced model will contain the significant terms (sweetener..025 0.47 Sweetener*Carbonation*Temperature -0.9504 1.63 852.9504 192.187 -2.69 Carbonation -2.031 0.923 0. as well as lower-order terms included in the significant terms (main effects: syrup to water.281 0.656 0.688 -1.49 Syrup to Water*Carbonation* -0. sweetener*syrup to water. carbonation.9504 -4.14 Sweetener*Carbonation 0.594 0.531 47.688 2.177 0.563 -0.000 0.9504 2.969 0.41 Temperature 3.53 47.37 2-Way Interactions 6 199.50 462.73 Sweetener*Syrup to Water*Temperature 4.77 Syrup to Water -1.49 Temperature Sweetener*Syrup to Water* 2.844 0.500 28.50 4-Way Interactions 1 47. Factorial Fit: Total Score versus Sweetener.9504 -2.9504 0. To preserve model hierarchy.218 P 0.50 462. Create a 24 factorial design in MINITAB. Syrup to Water.188 -1.062 2.07 Sweetener*Syrup to Water 4.469 0. and then enter the data. syrup to water*temperature). sweetener*syrup to water*temperature).062 -4.53 1.688 33.688 0.313 -0.531 0.10 and select terms with P-value less than 0. Use an α = 0.388 0.938 -0. Estimated Effects and Coefficients for Total Score (coded units) Term Effect Coef SE Coef T Constant 182. .031 0.64 Residual Error 16 462.344 0. Since there are two replicates of the experiment. sweetener*syrup to water*carbonation.91 Pure Error 16 462.048 0.015 0. two-factor interactions: sweetener*carbonation.91 Total 31 1967.16 7. Select Stat > DOE > Factorial > Analyze Factorial Design.687 0.125 101. syrup to water*carbonation.15 3-Way Interactions 4 405.625 213.344 0.

920 0.938 1.9244 2.094 0.594 0.040 0.313 -0.91 176. .45 Temperature 3.73 Pure Error 16 462.9244 0.37 Sweetener*Temperature -2.50 462.281 0.251 0.10 Sweetener*Syrup to Water*Carbonation -5.38 21.50 28.15 Residual Error 20 546.344 0.344 0. Estimated Effects and Coefficients for Total Score (coded units) Term Effect Coef SE Coef T Constant 182.06 391.20 Sweetener*Carbonation 0.18 Syrup to Water*Carbonation -0.81 Sweetener*Syrup to Water*Temperature 4.13 Sweetener*Syrup to Water 4.188 -2.9244 197.714 0.63 852.06 195.38 84.88 546.162 0.344 0.80 2-Way Interactions 5 176.9244 -0.688 0.000 0.062 2.969 0.486 0.781 0.53 7.062 -4.16 7.688 -1.9244 2.34 Lack of Fit 4 84.30 Syrup to Water*Temperature -0.38 1..73 Sweetener -9.563 -0.188 -0.764 0. Syrup to Water.306 0.88 27.47 P 0.001 0.54 Analysis of Variance for Total Score (coded units) Source DF Seq SS Adj SS Adj MS F Main Effects 4 852.91 Total 31 1967.90 Syrup to Water -1.9244 -0.188 -1.046 0.9244 -4.094 0.91 35.9244 -1..Chapter 12 Exercise Solutions 12-4 continued Factorial Fit: Total Score versus Sweetener.000 0.656 0.9244 -0.005 0.9244 -1.031 0.688 2.011 0.63 213.020 P 0.585 12-12 .531 0.29 3-Way Interactions 2 391.9244 -2.9244 2.71 Carbonation -2.09 0.

To find the residuals.5 0. This is not serious enough to warrant concern.0 Temperature There appears to be a slight indication of inequality of variance for sweetener and syrup ratio.0 8 Residual 8 Residual 0 0 -8 0 -8 -1.0 0.Chapter 12 Exercise Solutions 12-5.5 1.0 Carbonation 0.0 -0. Normal Probability Plot of the Residuals (response is Total Score) 99 Percent 90 50 10 1 -10 -5 0 Residual 5 10 Residuals Versus Sweetener Residuals Versus Syrup to Water (response is Total Score) (response is Total Score) 8 Residual Residual 8 0 -8 -8 -1. select Stat > DOE > Factorial > Analyze Factorial Design.5 0.0 Sweetener 0.5 0.0 -0. as well as a slight indication of an outlier.5 0.0 -1.0 -0.5 1. Select “Graphs”. Select “Terms” and verify that all terms for the reduced model are included. choose “Normal plot” of residuals and “Residuals versus variables”.0 Syrup to Water Residuals Versus Carbonation Residuals Versus Temperature (response is Total Score) (response is Total Score) 0.5 1.5 1. 12-13 .0 -1.0 -0. and then select the variables.

9244 2.046 0. From examination of the above table. factors A.30 Syrup to Water*Temperature -0.37 Sweetener*Temperature -2. .9244 -0.969 0.54 P 0.344 0.011 0. Select “Terms” and verify that all terms for the reduced model are selected.81 Sweetener*Syrup to Water*Temperature 4.094 0. Also. Factorial Fit: Total Score versus Sweetener.000 0. and ABD appear to be significant.. 12-14 .9244 -4.313 -0.688 2.040 0.71 Carbonation -2.9244 -2.563 -0.714 0.764 0.9244 2.062 2.162 0.9244 197.281 0.20 Sweetener*Carbonation 0.188 -0. and a value greater than approximately |2| would be considered significant.10 Sweetener*Syrup to Water*Carbonation -5.486 0.Chapter 12 Exercise Solutions 12-6.344 0.9244 0.020 The ratio of the coefficient estimate to the standard error is distributed as t statistic.13 Sweetener*Syrup to Water 4.45 Temperature 3. AB.251 0.9244 2.531 0.656 0. Estimated Effects and Coefficients for Total Score (coded units) Term Effect Coef SE Coef T Constant 182. ABC.18 Syrup to Water*Carbonation -0. if the confidence interval includes zero. D.9244 -1.938 1.188 -1.781 0. Select Stat > DOE > Factorial > Analyze Factorial Design.031 0.688 -1.000 0.73 Sweetener -9.188 -2.594 0.9244 -1. the factor is not significant. Syrup to Water..9244 -0.344 0.90 Syrup to Water -1.688 0.9244 -0.062 -4.094 0.920 0.

250 -0.000 Sweetener*Carbonation 1.MTW. and then enter the data.125 Syrup to Water*Carbonation* -2. Syrup to Water. Then select “Graphs”.875 Syrup to Water*Temperature -3.500 Carbonation*Temperature 1.750 Sweetener*Syrup to Water*Temperature 4.125 Carbonation 0.10 Factorial Fit: Total Score versus Sweetener.750 -0. Select Stat > DOE > Factorial > Analyze Factorial Design... Create a 24 factorial design in MINITAB.125 Sweetener*Carbonation*Temperature 0.750 0.500 2.500 Sweetener*Temperature -6. Estimated Effects and Coefficients for Total Score (coded units) Term Effect Coef Constant 183.Chapter 12 Exercise Solutions 12-7. choose the normal effects plot.250 Syrup to Water -0.5 12-15 . .250 2.875 Carbonation*Temperature … Normal Probability Plot of the Effects (response is Total Score.375 Temperature 5.750 1.000 0.750 Sweetener*Syrup to Water 4.250 0.500 -5. and set alpha to 0.000 -1.125 Syrup to Water*Carbonation -1. Since there is only one replicate of the experiment.250 Temperature Sweetener*Syrup to Water* 3.000 0.250 -3. select “Terms” and verify that all terms are selected.000 2.500 -1.10) 99 Effect Type Not Significant Significant 95 90 F actor A B C D Percent 80 70 60 50 40 N ame S w eetener S y rup to Water C arbonation Temperature 30 20 10 5 1 A -10 -5 0 Effect 5 10 Lenth's PSE = 4.625 Sweetener -10. Alpha = .500 -3. The design and data are in the MINITAB worksheet Ex12-7.500 Sweetener*Syrup to Water*Carbonation -7.

000 Sweetener -10.625 1.250 1.00 7.00 441.75 P 0.Chapter 12 Exercise Solutions 12-7 continued From visual examination of the normal probability plot of effects.59% Analysis of Variance for Total Score (coded units) Source DF Seq SS Adj SS Adj MS F Main Effects 1 441.5 0.0 187.0 There appears to be a slight indication of inequality of variance for sweetener.5 1.750 55.014 Normal Probability Plot of the Residuals (responseisTotal Score) 99 Percent 90 50 10 1 -20 -10 0 Residual 10 20 Residuals Versus the FittedValues (responseisTotal Score) Residual 10 0 -10 180.0 -0.48 0.014 S = 7.000 441.63 Total 15 1219. only factor A (sweetener) is significant.0 FittedValue Residuals Versus Sweetener (responseisTotal Score) Residual 10 0 -10 -1. as well as in the predicted values.63 Pure Error 14 778.15% R-Sq(adj) = 31.5 185. This is not serious enough to warrant concern. Re-fit and analyze the reduced model. 12-16 .93 Residual Error 14 778.82 0.0 182.5 190.45822 R-Sq = 36.865 -2.500 -5. Factorial Fit: Total Score versus Sweetener Estimated Effects and Coefficients for Total Score (coded units) Term Effect Coef SE Coef T P Constant 183.0 Sweetener 0.865 98.75 778.75 778.750 55.

A 25 design in two blocks will lose the ABCDE interaction to blocks. Day 1 a b c abc Day 2 d abd acd bcd (1) ab ac ad bc bd cd abcd Treatment combinations within a day should be run in random order.Chapter 12 Exercise Solutions 12-8. 12-9. or days. The ABCD interaction is confounded with blocks. Block 1 (1) ae ab be ac ce bc abce ad de bd abde cd acde abcd bcde Block 2 a e b abe c ace abc bce d ade abd bde acd cde bcd abcde 12-17 .

7175 Cat/React -1. Factorial Fit: Color versus Solv/React.0812 Temp*React Purity -0.1363 React Purity 4..5750 Solv/React*Temp -0.2725 -0.4562 Solv/React*React Purity -1.3650 -0.Chapter 12 Exercise Solutions 12-10.1200 0.1625 0. Select Stat > DOE > Factorial > Analyze Factorial Design.9125 -0.4187 Temp*React pH -0.0600 Cat/React*React pH 0.2300 -0. Cat/React.3513 Solv/React*Cat/React 1.1500 0.8475 12-18 . Then select “Graphs”.1062 Normal Probability Plot of the Effects (response is Color. .7700 Solv/React 1.2725 React pH -0.10) 99 Effect Ty pe Not Significant Significant D 95 90 F actor A B C D E Percent 80 70 60 50 40 N ame S olv /React C at/React Temp React P urity React pH 30 20 10 5 1 -2 -1 0 1 2 Effect 3 4 5 Lenth's PSE = 0. Alpha = ..8375 -0. and then enter the data. and set alpha to 0.4650 -0.5450 2.7325 Temp -0. select “Terms” and verify that all main effects and interaction effects are selected.4350 0.7025 -0.6150 Solv/React*React pH 0.2125 0.10.1462 Cat/React*React Purity 0. choose the normal effects plot. Since there is only one replicate of the experiment.4275 0. Estimated Effects and Coefficients for Color (coded units) Term Effect Coef Constant 2.MTW.2138 Cat/React*Temp 0. (a) Create a 25-1 factorial design in MINITAB.1825 React Purity*React pH 0.2925 0. The design and data are in the MINITAB worksheet Ex12-10.

545 2.5 1.52 2.52 2.93% Analysis of Variance for Color (coded units) Source DF Seq SS Adj SS Adj MS F Main Effects 1 82.4147 5.48 0.770 0.0 Residual plots indicate that there may be problems with both the normality and constant variance assumptions.0 React Purity 0. Re-fit and analyze the reduced model. 12-19 .Chapter 12 Exercise Solutions 12-10 (a) continued From visual examination of the normal probability plot of effects.272 0.0 -0.628 30.03 Residual Error 14 38.751 Pure Error 14 38.751 Total 15 121.000 S = 1.52 38.20% R-Sq(adj) = 65.000 React Purity 4.65876 R-Sq = 68.63 82. Factorial Fit: Color versus React Purity Estimated Effects and Coefficients for Color (coded units) Term Effect Coef SE Coef T P Constant 2.4147 6.15 P 0. only factor D (reactant purity) is significant.68 0.63 82.000 (b) Nor mal P r obability P lot of the Residuals (response is Color) 99 Percent 90 50 10 1 -4 -3 -2 -1 0 Residual 1 2 3 4 Residuals V er sus the Fitted V alues (response is Color) Residual 2 0 -2 0 1 2 3 4 5 Fitted Value Residuals Ver sus React P ur ity (response is Color) Residual 2 0 -2 -1.52 38.5 0.

425 1 0.715 1.865 -1 1 -1 React Purity -1 Solv/React 12-20 .875 4. or simply a 2-sample t-test. A cube plot in these factors shows how the design can be collapsed into a replicated 23 design.Chapter 12 Exercise Solutions 12-10 continued (c) There is only one significant factor. Looking at the original normal probability plot of effects and effect estimates. so this design collapses to a onefactor experiment. Cube Plot (data means) for Color 3. D (reactant purity).865 -2.005 5. The highest color scores are at high reactant purity.795 1 Cat/React 6. the lowest at low reactant purity.385 1. the 2nd and 3rd largest effects in absolute magnitude are A (solvent/reactant) and B (catalyst/reactant).

The aliases are: A*I = A*ACE = A*BDE = A*ABCD ⇒ A = CE = ABDE = BCD B*I = B*ACE = B*BDE = B*ABCD ⇒ B = ABCE = DE = ACD C*I = C*ACE = C*BDE = C*ABCD ⇒ C = AE = BCDE = ABD … AB*I = AB*ACE = AB*BDE = AB*ABCD ⇒ AB = BCE = ADE = CD The remaining aliases are calculated in a similar fashion. D:Time.175 -2. C:Vol.275 -0.337 E:Matl2 2.Chapter 12 Exercise Solutions 12-11. then define the experiment using Stat > DOE > Factorial > Define Custom Factorial Design.587 C:Vol 2. The design and data are in the MINITAB worksheet Ex12-11. Since there is only one replicate of the experiment. the complete defining relation is: I = ACE = BDE = ABCD. Enter the factor levels and yield data into a MINITAB worksheet.138 D:Time -0. B:Matl1. select “Terms” and verify that all main effects and twofactor interaction effects are selected.138 A:Temp*B:Matl1 1.762 B:Matl1 -5.238 A:Temp -1.913 A:Temp*D:Time -1. 12-21 .638 … Alias Structure I + A:Temp*C:Vol*E:Matl2 + B:Matl1*D:Time*E:Matl2 + A:Temp*B:Matl1*C:Vol*D:Time A:Temp + C:Vol*E:Matl2 + B:Matl1*C:Vol*D:Time + A:Temp*B:Matl1*D:Time*E:Matl2 B:Matl1 + D:Time*E:Matl2 + A:Temp*C:Vol*D:Time + A:Temp*B:Matl1*C:Vol*E:Matl2 C:Vol + A:Temp*E:Matl2 + A:Temp*B:Matl1*D:Time + B:Matl1*C:Vol*D:Time*E:Matl2 D:Time + B:Matl1*E:Matl2 + A:Temp*B:Matl1*C:Vol + A:Temp*C:Vol*D:Time*E:Matl2 E:Matl2 + A:Temp*C:Vol + B:Matl1*D:Time + A:Temp*B:Matl1*C:Vol*D:Time*E:Matl2 A:Temp*B:Matl1 + C:Vol*D:Time + A:Temp*D:Time*E:Matl2 + B:Matl1*C:Vol*E:Matl2 A:Temp*D:Time + B:Matl1*C:Vol + A:Temp*B:Matl1*E:Matl2 + C:Vol*D:Time*E:Matl2 From the Alias Structure shown in the Session Window. (a) and (b) Select Stat > DOE > Factorial > Analyze Factorial Design.275 1.275 1.525 -0.825 0. E:Matl2 Estimated Effects and Coefficients for yield (coded units) Term Effect Coef Constant 19.675 -0.MTW. Factorial Fit: yield versus A:Temp.

(d) Select Stat > DOE > Factorial > Analyze Factorial Design.913 4.9 16. The other main effects and interaction effects are calculated in the same way. C:Vol.8 + 18.9 -16.1) = ¼ (–6.9 – 16.5 16. choose the normal effects plot. and set alpha to 0.826 15.2 +23.2 15. Factorial Fit: yield versus A:Temp.739 … F * * P * * 12-22 . Since there is only one replicate of the experiment.5 + 16.2 +15.956 Residual Error 0 * * * Total 7 89.965 2-Way Interactions 2 9. select “Terms” and verify that all main effects and twofactor interaction effects are selected.8 + 18.Chapter 12 Exercise Solutions 12-11 continued (c) A B -1 1 1 -1 -1 1 -1 1 C -1 1 -1 1 -1 -1 1 1 D -1 -1 -1 1 1 1 -1 1 E -1 -1 1 -1 1 -1 1 1 1 -1 -1 -1 -1 1 1 1 yield 23.5 – 16.8 18.8 – 23. E:Matl2 … … Analysis of Variance for yield (coded units) Source DF Seq SS Adj SS Adj MS Main Effects 5 79.8 + 23.825 This are the same effect estimates provided in the MINITAB output above. Then select “Graphs”.4 – 16.2 + 15. D:Time.826 79.3) = 1.2 – 23. B:Matl1.4 16.1) = ¼ (7.4 – 16.525 [AB] = AB + BCE + ADE + CD = ¼ (+23.10.913 9.2 23.1 [A] = A + CE + BCD + ABDE = ¼ (–23.1) = –1.8 23.

Re-fit a reduced model containing only the B main effect.18 6. Select Stat > DOE > Factorial > Analyze Factorial Design.98 0.587 0.0 Effect 2.000 B:Matl1 -5. and select the “Normal plot” and “Residuals versus fits” residual plots.030 Pure Error 6 36.238 0.5 0.7375 Although none of the effects is significant at 0.0 -2.74 … 12-23 .56 53.88 0. Factorial Fit: yield versus B:Matl1 Estimated Effects and Coefficients for yield (coded units) Term Effect Coef SE Coef T P Constant 19.5 -5.18 36.18 6. main effect B (amount of material 1) is more than twice as large as the 2nd largest effect (absolute values) and falls far from a line passing through the remaining points. and pool the remaining terms to estimate error.025 … Analysis of Variance for yield (coded units) Source DF Seq SS Adj SS Adj MS F P Main Effects 1 53.030 Total 7 89.5 5.561 8.025 Residual Error 6 36. Alpha = .0 Lenth's PSE = 2.56 53.10) 99 Effect Ty pe Not Significant Significant 95 90 F actor A B C D E Percent 80 70 60 50 40 N ame A :Temp B:M atl1 C :V ol D :Time E :M atl2 30 20 B 10 5 1 -7.175 -2. Select “Terms” and select “B”. Then select “Graphs”.16 0.8682 -2.Chapter 12 Exercise Solutions 12-11 (d) continued Normal Probability Plot of the Effects (response is yield.10.8682 22.18 36.

0 Residual plots indicate a potential outlier.0 Residual 2.Chapter 12 Exercise Solutions 12-11 continued (e) Residuals Versus the Fitted Values Exercise 12-11 (e) (response is yield) 2 1 Residual 0 -1 -2 -3 -4 -5 16 17 18 19 Fitted Value 20 21 22 Normal Probability Plot of the Residuals Exercise 12-11(e) (response is yield) 99 95 90 Percent 80 70 60 50 40 30 20 10 5 1 -5.5 5.5 0. it may be necessary to make additional experimental runs 12-24 .0 -2. The run should be investigated for any issues which occurred while running the experiment. If no issues can be identified.

053 * B*D 2. choose the normal effects plot.26 0. and BC.953 -2.953 211.25 3.013 * D -7.262 C -30.75 3.00 3. BC) are significant. 12-25 .386 A*B 22.75 3.005 * B 10.67 0.50 11.00 5.87 0.00 -15.00 -10. The design and data are in the MINITAB worksheet Ex12-12. Create a 24 factorial design in MINITAB.953 -3.10) 99 95 AB 90 Percent 80 70 60 50 40 30 F actor A B C D N ame A B C D BC 20 C 10 A 5 1 Effect Ty pe Not Significant Significant -5 -4 -3 -2 -1 0 Standardized Effect 1 2 3 The main effects A and C and two two-factor interactions with B (AB.7 2.63 0.5 9.953 1. and then enter the data.85 0. Then select “Graphs”. Factorial Fit: Mole Wt versus A.151 Residual Error 5 1250 1250 250. Re-fit and analyze a reduced model containing A.953 -0. Since this is a single replicate of the experiment.50 3.953 2.79 0.50 3. AB.953 0.00 3. D Estimated Effects and Coefficients for Mole Wt (coded units) Term Effect Coef SE Coef T P Constant 837.953 0. C.MTW.014 2-Way Interactions 6 4000 4000 666.50 3. The main effect B must be kept in the model to maintain hierarchy. B. C.Chapter 12 Exercise Solutions 12-12.75 3.00 3. Alpha = . (a) Select Stat > DOE > Factorial > Analyze Factorial Design.765 C*D 7. B.953 -4.0 Total 15 15100 Normal Probability Plot of the Standardized Effects (response is Mole Wt.25 3.555 B*C -20.10.74 0.50 -18.32 0.25 3.50 -3.32 0.95 0.00 2.000 A -37.953 -0.95 0.765 A*D 5.953 0.036 * A*C -2. select “Terms” and verify that all main effects and twofactor interaction effects are selected.386 … Analysis of Variance for Mole Wt (coded units) Source DF Seq SS Adj SS Adj MS F P Main Effects 4 9850 9850 2462. and set alpha to 0.50 -1.53 0.50 1.85 0.

0 125.94 0.001 * A*B 22.172 C -30. BC”.400 246.400 3.80 0.5 9.000 2-Way Interactions 2 3625.400 1.00 -15.0 3208.3 17.Chapter 12 Exercise Solutions 12-12 continued (b) Select Stat > DOE > Factorial > Analyze Factorial Design.00 3.0 1850.0 1600.0 Lack of Fit 2 250.47 0.41 0.00 3. B.004 Residual Error 10 1850.400 -5.0 250.015 * … Analysis of Variance for Mole Wt (coded units) Source DF Seq SS Adj SS Adj MS F P Main Effects 3 9625.00 -10. AB.0 … The same terms remain significant.400 -4.00 5.0 185. Select “Terms” and select “A.0 200.400 -2.31 0.008 * B*C -20.75 3.30 0. C.0 Total 15 15100.000 A -37.0 0.50 3. Then select “Graphs”.000 * B 10.0 3625.50 11.00 3.63 0.0 1812.50 -18. and BC. and select the “Normal plot” and “Residuals versus fits” residual plots.25 3. B. C Estimated Effects and Coefficients for Mole Wt (coded units) Term Effect Coef SE Coef T P Constant 837. 12-26 . Factorial Fit: Mole Wt versus A.559 Pure Error 8 1600. AB.0 9625.51 0. A. C.34 0.

12-27 .Chapter 12 Exercise Solutions 12-12 continued (c) Normal Probability Plot of the Residuals (response is Mole Wt) 99 95 90 Percent 80 70 60 50 40 30 20 10 5 1 -30 -20 -10 0 Residual 10 20 30 Residuals Versus the Fitted Values (response is Mole Wt) 20 Residual 10 0 -10 -20 -30 790 800 810 820 830 840 Fitted Value 850 860 870 880 A “modest” outlier appears on both plots. however neither plot reveals a major problem with the normality and constant variance assumptions.

This will ensure that if both lack of fit and curvature are not significant. Also. lack of fit. the main and interaction effects are tested for significance against the correct residual error (lack of fit + curvature + pure error). Create a 24 factorial design with four center points in MINITAB. regardless of whether center points are included in the model. (a) Select Stat > DOE > Factorial > Analyze Factorial Design. To summarize MINITAB’s functionality. Assuming that lack of fit and curvature tests are not significant. all three (curvature. and pure error) should be included in the residual mean square. The design and data are in the MINITAB worksheet Ex12-13. 12-28 . DO NOT include the center points in the model (uncheck the default selection). See the dialog box below. curvature is always tested against pure error and lack of fit (if available).Chapter 12 Exercise Solutions 12-13. The inclusion/exclusion of center points in the model affects the total residual error used to test significance of effects.MTW. Select “Terms” and verify that all main effects and two-factor interactions are selected. and then enter the data.

46 0.13 0. (In MINITAB.00 9.50 11. the first test to consider is the “lack of fit” test.50 -18.57 0. the next test to consider is the “curvature” test. B.527 -0.25 9.898 A*D 5.799 B*C -20.703 A*B 22.081 B 10. which is a test of significance for the pure quadratic terms. pure error.97 0.00 9. which is a test of significance for terms not included in the model (in this exercise.527 -1.234 2-Way Interactions 6 4000 4000 666.703 … Analysis of Variance for Mole Wt (coded units) Source DF Seq SS Adj SS Adj MS F P Main Effects 4 9850 9850 2462.000 A -37. D Estimated Effects and Coefficients for Mole Wt (coded units) Term Effect Coef SE Coef T P Constant 848. If tests for both lack of fit and curvature are not significant. a better strategy is to add runs that would enable estimation of the quadratic effects. the model is not correctly specified.2 Curvature 1 8820 8820 8820.527 1.527 -1.0 16. and lack of fit (if available) and use this as the basis for testing for significant effects.898 C*D 7.00 -15.004).915 Pure Error 3 3000 3000 1000.0 Total 19 26920 … (b) The test for curvature is significant (P-value = 0.527 0.25 0.527 0.39 0.00 5.527 0.50 1.75 9. If lack of fit is not significant.50 -3.527 0. no further statistical analysis should be performed because the model is inadequate.822 Residual Error 9 13070 13070 1452.004 * Lack of Fit 5 1250 1250 250.527 -0.Chapter 12 Exercise Solutions 12-13 (a) continued When looking at results in the ANOVA table.150 D -7. then it is reasonable to pool the curvature. This approach to sequential experimentation is presented in Chapter 13.39 0.26 0. and some terms need to be added to the model.13 0.7 0.70 0.00 8. If lack of fit is significant.60 0.268 A*C -2.321 B*D 2.00 2. the three-factor and four-factor interactions). Although one could pick a “winning combination” from the experimental runs. 12-29 .) Factorial Fit: Mole Wt versus A.50 -1.527 -1.50 3. C. If this test is significant.612 C -30.05 0.50 9.52 0.5 1.75 9.25 9.25 9.00 9.75 9.52 0.00 -10.521 99.0 0. this is accomplished by not including center points in the model.18 0.

61 1.7621 0.000 2-Way Interactions 3 10610.550 0.33 Total 11 28184. Also.31 0. Then select “Graphs”.700 23.70 761.33 0.000 * D -2.500 -0.190 A*B 1.000 -1.88 0. B. choose the normal effects plot.7621 -1.0 4.13 0.0 13. including a column indicating whether a run is a center point run (1 = not center point.12 0. Select “Terms” and verify that all main effects and two-factor interactions are selected.6 5.759 C 80.6 5.250 0.7621 52. DO NOT include the center points in the model (uncheck the default selection). D Estimated Effects and Coefficients for Resist (coded units) Term Effect Coef SE Coef T P Constant 60.51 0.6 18. Then define the experiment using Stat > DOE > Factorial > Define Custom Factorial Design. Enter the factor levels and resist data into a MINITAB worksheet.000 * B -0.100 0. The design and data are in the MINITAB worksheet Ex12-15. Alpha = .30 0.6223 97.7621 47. and set alpha to 0.400 -1.0 Normal Probability Plot of the Standardized Effects (response is Resist.57 0.200 0.000 Residual Error 4 18.MTW. 0 = center point).000 0.29 0.260 … Analysis of Variance for Resist (coded units) Source DF Seq SS Adj SS Adj MS F P Main Effects 4 17555.1 3536.6 4.510 A*C 72.600 40.3 17555.65 Curvature 1 5. (a) Select Stat > DOE > Factorial > Analyze Factorial Design.433 0.83 944.72 0. C.7621 -0.000 A 47.850 0.338 Pure Error 3 13.000 * A*D -2.1 10610.400 0.10.Chapter 12 Exercise Solutions 12-15.7621 31.76 0. Factorial Fit: Resist versus A.300 0.7621 -1.3 4388.800 36.10) 99 Effect Ty pe Not Significant Significant 95 C 90 Percent 80 F actor A B C D AC 70 A 60 50 40 N ame A B C D 30 20 10 5 1 0 10 20 30 40 Standardized Effect 50 60 12-31 .

0 5.327 Pure Error 7 35.7 10599.327).8007 50.7 2066. so continue with analysis.0 41.000 A 47.85 0.8007 45. Then select “Graphs”.43 0.6537 92.60 40.46 0.3 17543. (c) Residuals Versus the Fitted Values (response is Resist) 3 2 Residual 1 0 -1 -2 -3 0 20 40 60 80 100 Fitted Value 120 140 160 180 A funnel pattern at the low value and an overall lack of consistent width suggest a problem with equal variance across the prediction range.33 0. C Estimated Effects and Coefficients for Resist (coded units) Term Effect Coef SE Coef T P Constant 60.43 0.Chapter 12 Exercise Solutions 12-15 continued Examining the normal probability plot of effects. AC”.6 1. Select Stat > DOE > Factorial > Analyze Factorial Design. and AC.0 Curvature is not significant (P-value = 0.80 36.000 * A*C 72.1 Total 11 28184.79 0.000 * … Analysis of Variance for Resist (coded units) Source DF Seq SS Adj SS Adj MS F P Main Effects 2 17543. C.89 0.6 5.1 Curvature 1 5. 12-32 . Select “Terms” and select “A.6 5.11 0.8007 29.000 * C 80.70 23. (b) Factorial Fit: Resist versus A. the main effects A and C and their twofactor interaction (AC) are significant.3 8771.30 0.40 0. C.7 10599.44 0.000 2-Way Interactions 1 10599.4 35.000 Residual Error 8 41. and select the “Normal plot” and “Residuals versus fits” residual plots.6 1710.4 5. Re-fit and analyze a reduced model containing A.

0 Residual 2.0 The normal probability plot of residuals is satisfactory.5 5. 12-33 .Chapter 12 Exercise Solutions 12-15 continued (d) Normal Probability Plot of the Residuals (response is Resist) 99 95 90 Percent 80 70 60 50 40 30 20 10 5 1 -5.5 0.0 -2. The concern with variance in the predicted resistivity indicates that a data transformation may be needed.

0 -1.65 65 .Chapter 13 Exercise Solutions Note: To analyze an experiment in MINITAB.0 − 1 ≤ x1 ≤ 1.70 70 .1 ≤ x2 ≤ 1 x2 6 = = 0. Ex13-1x1 y = 75 + 10x1 + 6x2 1. therefore some information required by MINITAB is not included. 13-1.6 13-1 . The Excel data sets contain only the data given in the textbook. (a) Graph > Contour Plot Contour Plot of Ex13-1y vs Ex13-1x2.5 1. The MINITAB instructions provided for the factorial designs in Chapter 12 are similar to those for response surface designs in this Chapter.0 (b) yˆ = 75 + 10 x1 + 6 x2 -0.75 75 .5 0.6 x1 10 ∆x1 = 1 ∆x2 = 0.80 80 .0 -0.5 0.85 > 85 Ex13-1x2 0.0 Ex13-1y < 60 60 .0 Ex13-1x1 0. the initial experimental layout must be created in MINITAB or defined by the user.5 -1.

yˆ = 50 + 2 x1 − 15 x2 + 3 x3 − 1 ≤ xi ≤ +1. (a) This design is a CCD with k = 2 and α = 1.Chapter 13 Exercise Solutions 13-2.0 13-3.0 2 ∆x3 = 2 βˆ 3 βˆ2 ∆x2 = 3 = −0. The design is not rotatable.3 select x2 with largest absolute coefficient. 2. βˆ2 = −15.20 −15 1. 13-2 . and set ∆x2 = 1.13 βˆ ∆x −15 1.5. i = 1.0 2 βˆ1 ∆x1 = = = −0.

0 = center point).2 0.3 2.0 Total 11 31084.088 0.423 7. and quadratic terms are selected.9 … Estimated Regression Coefficients for y using data in uncoded units Term Coef Constant 160.Chapter 13 Exercise Solutions 13-3 continued (b) Enter the factor levels and response data into a MINITAB worksheet.09 0.5 15.013 Interaction 1 2.9231 x1*x2 -0.875 … Analysis of Variance for y Source DF Seq SS Adj SS Adj MS F P Regression 5 30583.000 x2 3.285 -0.2 28934.017 x2*x2 15.875 Residual Error 6 501.4 6116.0 823.5 501.577 7.5 5.0 162.164 0. Select “Terms” and verify that all main effects. x2 The analysis was done using coded units.000 x1 -87.3 2. Then define the experiment using Stat > DOE > Response Surface > Define Custom Response Surface Design.3 0.0 486.461 2.000 Square 2 1647.461 -3.590 0.03 0. including a column indicating whether a run is a center point run (1 = not center point.4118 x1*x1 -10. Estimated Regression Coefficients for y Term Coef SE Coef T P Constant 160.868 4.85 0.471 x1*x1 -24.314 0. Select Stat > DOE > Response Surface > Analyze Response Surface Design.555 35.688 10.000 Linear 2 28934.03 0.7 73.5 83. The design and data are in the MINITAB worksheet Ex13-3.1 173. two-factor interactions.8682 x1 -58.704 -18.769 0.4 30583. Response Surface Regression: y versus x1.MTW.704 0.991 Pure Error 3 486.2941 x2 2.618 4.082 x1*x2 -1.5 9.6 Lack-of-Fit 3 15.18 0.441 4.8546 x2*x2 6.2 14467.273 0.7500 13-3 .0 1647.

Chapter 13 Exercise Solutions 13-3 continued (c) Stat > DOE > Response Surface > Contour/Surface Plots Contour Plot of y vs x2.49384 x2 = -0. The range for x2 agrees with the ANOVA results indicating that it is statistically insignificant (P-value = 0.471). (d) Temp = 50 x1 + 750 = 50(+1.5 and letting x2 (temperature) range from −1.6101 0. The level for temperature could be established based on other considerations. x1 1.7 13-4 .0 -0.5 to + 1.50) + 750 = 825 Time = 15 x2 + 30 = 15(−0.5 -1 0 x1 1 x1 0 x2 -1 1 From visual examination of the contour and surface plots.5 x1 = 1.0 240 180 y 120 60 1 0 -1 -1.5.5 1.5 40 50 75 100 125 150 175 200 y < > 40 50 75 100 125 150 175 200 225 225 -1. it appears that minimum purity can be achieved by setting x1 (time) = +1. x1 Surface Plot of y vs x2.22) + 30 = 26.0 x2 0.217615 y = 49. such as cost. A flag is planted at one option on the contour plot above.

64 64 .7) = 0.0 + 1.6x1 + 1.9.3x1x2 2 Ex13-4y < 56 56 .3 x1 = 0 ∂ x2 x1 = −13.3x1 x2 ) = ⇒0 ∂ x1 ∂ x1 = 1.68 68 .2x2^2 + 0.60 60 .6 − 2 x1 + 0.6 x1 + 1.0 + 1.885 x2 = [−1.6 (b) ∂ yˆ ∂ (69.4 x2 + 0.1x2 − 1x12 − 1.70 > 70 Ex13-4x2 1 0 -1 -2 -2 -1 0 Ex13-4x1 1 2 yˆ max = 70.1. Graph > Contour Plot Contour Plot of Ex13-4y vs Ex13-4x2.885)] (−2.1 − 0.3 x2 = 0 ∂ yˆ = 1. Ex13-4x1 y = 69.1x2 . x2 ≈ +0.1 − 2.2 x22 + 0.569 13-5 .9 (−15.66 66 .1x1^2 .58 58 .3(0.012 at x1 ≈ +0.62 62 .Chapter 13 Exercise Solutions 13-4.4) = 0.

43226 46.56 0.04818 1. one approach to solving this exercise is to create a two-factor response surface design in MINITAB.03706 -3.16128 0.03706 -2.012 x2*x2 -0.13355 0.377 Pure Error 4 0. A2.04309 318. (a) The design is a CCD with k = 2 and α = 1.291 … Analysis of Variance for y Source DF Seq SS Adj SS Adj MS F P Regression 5 2. The design is rotatable.0790 0. The worksheet is in run order.13355 0.00781 108. B2.142 0. (b) Since the standard order is provided.7273 x1 0.371 0.03271 0.0550 0. To analyze the experiment. then enter the data.0790 x1*x2 0. x2 The analysis was done using coded units. to change to standard order (and ease data entry) select Stat > DOE > Display Design and choose standard order.000 Linear 2 2.03228 0. highlight the design with five center points (13 runs).580 0.06678 7.19 0. Select Stat > DOE > Response Surface > Create Response Surface Design.7273 0.1249 x2*x2 -0. and enter a custom alpha value of exactly 1.01210 0. AB) is selected.2980 x2 -0.000 x1*x1 -0.4071 x1*x1 -0.703 0. The design and data are in the MINITAB worksheet Ex13-5.2980 0.000 x1 0.1249 0.MTW.00807 Total 12 2. select Stat > DOE > Response Surface > Analyze Response Surface Design.03424 -11.4.30 0.01210 0.000 Square 2 0. Estimated Regression Coefficients for y Term Coef SE Coef T P Constant 13.06499 0.0550 13-6 .132 0.291 Residual Error 7 0. Select “Terms” and verify that a full quadratic model (A.889 0.06499 0.4071 0. Leave the design type as a 2-factor.03271 0.16128 2.35 0.00928 Lack-of-Fit 3 0.000 x2 -0.54 0.070 x1*x2 0. B.4 (the rotatable design is α = 1.020 Interaction 1 0.01563 1. Select “Designs”.41421).Chapter 13 Exercise Solutions 13-5.22628 … Estimated Regression Coefficients for y using data in uncoded units Term Coef Constant 13.03228 0.01090 1. central composite design.01563 2.03424 8.01210 1. Response Surface Regression: y versus x1.

50 12.25 13.25 14.50 13.0 -0.00 12. x1 1.5 1.75 13. and Weight = 7.0 12.00 0.5 x2 Surface Plot of y vs x2.75 14.4 and setting x2 between -1 and -1. Stat > DOE > Response Surface > Contour/Surface Plots Contour Plot of y vs x2. setting x1 in a range from 0 to +1.50 13.4 will maximize viscosity.Chapter 13 Exercise Solutions 13-5 (b) continued Values of x1 and x2 maximizing the Mooney viscosity can be found from visual examination of the contour and surface plots.00 13. or using MINITAB’s Response Optimizer.5 -1.0 x1 0.0 x1 0 x2 -1 1 Stat > DOE > Response Surface > Response Optimizer In Setup.0 13.25 12.25 14. 13-7 .0 y < > 12.25 13.25 12.5 0.75 13. let Goal = maximize.00 14.0 12.50 12.00 13.5 1 0 -1 -1.5 y 13.00 12.0 -0. Lower = 10.75 14. From the plots and the optimizer. x1 0. Target = 20.

003 x2*x3 43.3844 x3*x3 -29.08 1.08 -0.0011 x2 109.363 x1*x2 66.03 21.005 0.328 0.001 Residual Error 17 98498 98498 5794 Total 26 1346735 … Estimated Regression Coefficients for y1 using data in uncoded units Term Coef Constant 327. The design is a full factorial of three factors at three levels.25 0. (a) To analyze the experiment.62 38.435 0. Select “Designs” to establish three levels for each factor.Chapter 13 Exercise Solutions 13-6.82 0. x3 The analysis was done using coded units. Select “Terms” and verify that a full quadratic model is selected.000 Linear 3 1090558 1090558 363519 62. The design and data are in the MINITAB worksheet Ex13-6.94 9.317 x2*x2 -22.01 31.000 x2 109.481 x3*x3 -29.4256 x3 131.43 17. In order to analyze this experiment using the Response Surface functionality.030 0.47 17.5833 13-8 .0056 x2*x2 -22.866 0.000 x1*x1 32.008 x1*x3 75.000 x3 131. Since the runs are listed in a patterned (but not standard) order.MTW. it must also be defined using Stat > DOE > Response Surface > Define Custom Response Surface Design.08 -0. Estimated Regression Coefficients for y1 Term Coef SE Coef T P Constant 327.38 31.4656 x1*x1 32.47 21.000 Square 3 14219 14219 4740 0.0578 x1*x2 66.97 3. x2. and then enter the data.720 0.000 x1 177.94 0.6237 x1 177. Select Stat > DOE > Factoriall > Create Factorial Design.4708 x2*x3 43.502 Interaction 3 143461 143461 47820 8. one approach to solving this exercise is to create a general full factorial design in MINITAB.453 0. Change the design type to a general full factorial design. select Stat > DOE > Response Surface > Analyze Response Surface Design.76 8. then select “Factors” to specify the actual level values.97 3.97 1.064 … Analysis of Variance for y1 Source DF Seq SS Adj SS Adj MS F P Regression 9 1248237 1248237 138693 23.06 31.94 7.099 0.0283 x1*x3 75. and select the number of factors as “3”.58 21. Response Surface Regression: y1 versus x1.74 0.935 0.983 0.94 6.00 17.

3189 x3*x3 16. x3 The analysis was done using coded units.202 0.89 0.3 21957.113 0.0 3408.7794 x1*x2 7.97 Linear 3 21957.09 Square 3 1805.108 12.280 x2 15.2 32650.890 22.691 x2*x3 14.198 17.65 0.192 10.404 0. select Stat > DOE > Response Surface > Analyze Response Surface Design.235 0.65 1.33 1.323 10.1978 x2*x2 -1.31 0.0 1135.826 0.815 0.1917 x1*x1 4.99 Residual Error 17 32650.82 Interaction 3 3408.81 0.65 0.7192 x1*x3 5.938 0.1083 x2*x3 14.3 7319.156 x3 29.5 1805.550 x1*x3 5.3233 x3 29.31 1.33 1.Chapter 13 Exercise Solutions 13-6 continued (b) To analyze the experiment.5278 x2 15. Select “Terms” and verify that a full quadratic model is selected.5 601.116 0.57 3.0825 F 1.483 0.719 12.33 2.361 x1*x2 7.89 0. x2.030 0.7 3018.2 1920.082 12.8896 x1 11.319 17.60 Total 26 59820.7 27170.9 … Estimated Regression Coefficients for y2 using Term Coef Constant 34.012 x1*x1 4.281 … Analysis of Variance for y2 Source DF Seq SS Adj SS Adj MS Regression 9 27170.89 -0.564 0.528 10.817 x2*x2 -1. Estimated Regression Coefficients for y2 Term Coef SE Coef T P Constant 34. Response Surface Regression: y2 versus x1.942 x3*x3 16.074 0.629 data in uncoded units 13-9 .136 x1 11.610 0.779 17.59 P 0.

0 Overlaid Contour Plot of y1.0 -1.5 -1. select the first two factors x1 and x2.0 -0.5 0.0 x1 0.Chapter 13 Exercise Solutions 13-6 continued (c) Both overlaid contour plots and the response optimizer can be used to identify settings to achieve both objectives.0 -0.0 y1 400 600 y2 0 80 0.5 x2 Hold Values x3 0 0. set low = 400 and high = 600.5 x2 Hold Values x3 -1 0. y2 1.0 -1. The goal is to minimize y2 (standard deviation) set low = 0 (the minimum of the observed results) and high = 80 (the 3rd quartile of the observed results).0 -0. Overlaid Contour Plot of y1.5 1.0 y1 400 600 y2 0 80 0.0 13-10 . y2 1.5 -1. Select “Contours” to establish the low and high contours for both y1 and y2.0 x1 0.0 -0.5 0.5 1. Stat > DOE > Response Surface > Overlaid Contour Plot After selecting the responses. Since the goal is to hold y1 (resistivity) at 500.

3 and x3 = -0. For y2.5 -1.5 1. 13-11 .5 x2 Hold Values x3 1 0. y2 1. set Goal = Minimize. Upper = 600. x2 = 0. Target = 500.0 -0.Chapter 13 Exercise Solutions 13-6 (c) continued Overlaid Contour Plot of y1.16 and standard deviation is 44.0 Stat > DOE > Response Surface > Response Optimizer In Setup.4.0.0 y1 400 600 y2 0 80 0. the predicted resistivity mean is 495.0 -1. Target = 0. Leave all Weight and Importance values at 1.5 0. Lower = 400.75. The graph below represents one possible solution. and Upper = 80.0 x1 0.0 -0. At x1 = 1. for y1 set Goal = Target.

and then define the experiment using Stat > DOE > Factorial > Define Custom Factorial Design. The design and data are in the MINITAB worksheet Ex13-7.Chapter 13 Exercise Solutions 13-7. E … Alias Structure I + A*B*C*D A + B*C*D B + A*C*D C + A*B*D D + A*B*C E + A*B*C*D*E A*B + C*D A*C + B*D A*D + B*C A*E + B*C*D*E B*E + A*C*D*E C*E + A*B*D*E D*E + A*B*C*E 13-12 . Enter the factor levels and response data into a MINITAB worksheet. A+BCD B+ACD C+ABD D+ABC E AB+CD AC+BD AD+BC AE+BCDE BE+ACDE CE+ABDE DE+ABCE ABE+CDE ACE+BDE ADE+BCE This is a resolution IV design. (a) The defining relation for this half-fraction design is I = ABCD (from examination of the plus and minus signs). B. Stat > DOE > Factorial > Analyze Factorial Design Factorial Fit: Mean versus A. but some 2-factor interactions are aliased with each other. D. C. All main effects are clear of 2-factor interactions.MTW.

11 0.98 0.01994 6.0329 -0.0912 0.01908 Lack of Fit 10 0.8090 1.91 0.1529 0.05 0.81 0. D.02021 0.000 A 0.000 B -0.1638 -0.03 0.0010 0.0637 0. E Estimated Effects and Coefficients for Height (coded units) Term Effect Coef SE Coef T P Constant 7.0198 0.501 Residual Error 32 0.000 D 0.0819 0.26 0.000 2-Way Interactions 1 0.01575 0.2387 -0.8012 0.1742 0.62707 0.02021 0.29 0.000 A*B -0.631 A*D*E -0.6256 0.0319 0.78 0.0248 0.0229 -0.001 C*E -0.000 A 0.000 B -0.01994 -4.124 B*E 0.8908 13-13 . B.0196 0.28060 14.0006 0.469 A*C 0.01994 2.02021 -0.99 0.2387 -0.6271 0.000 C -0.2421 0.37800 0.02021 -4.83846 0.02021 0.02021 377.02021 0.0298 0.01994 382.83846 1.71 0.1638 -0.62707 0.02021 2.1529 0.57 0.84 0.89078 The reduced model for mean: Factorial Fit: Height versus A.01742 0.575 A*E 0.0496 -0.2806 0. E Estimated Effects and Coefficients for Height (coded units) Term Effect Coef SE Coef T P Constant 7.2421 0.05 0.0165 0.01960 Total 47 2.0456 0.959 A*C*E 0.0765 0.554 Pure Error 32 0.0765 0.01960 Total 47 2.421 D*E 0.8012 0.0819 0.6271 0.80 0.0596 -0.76 0.07 0.1210 0.229 D 0.0296 -0.976 A*D -0.0396 0.6256 0.02021 -5.0456 0.02021 3.02021 -1.0021 0.04726 0.23 0.1194 0.01960 Pure Error 32 0.45224 23.023 3-Way Interactions 3 0.71 0.01994 -5.0098 0.000 B*E 0.58 0.41 0.62707 0.027 E -0.150 … Analysis of Variance for Height (coded units) Source DF Seq SS Adj SS Adj MS F P Main Effects 5 1.8090 0.99 0.000 Residual Error 42 0.0115 0.0148 0.76 0.1742 0.37800 0.02021 -0.48 0. B.000 … Analysis of Variance for Height (coded units) Source DF Seq SS Adj SS Adj MS F P Main Effects 4 1.0012 0. C.36769 18.02021 -0.0913 0.031 E -0.02021 5.62707 0.51 0.01994 3.02021 1.47 0.73 0.04726 0.1194 0.02021 -1.2806 0.89 0.000 2-Way Interactions 7 0.335 A*B*E 0.1210 0.Chapter 13 Exercise Solutions 13-7 continued (b) The full model for mean: Stat > DOE > Factorial > Analyze Factorial Design Factorial Fit: Height versus A. D.05400 2.

00375 -0.04375 0.003 Residual Error 8 0. E Effects and Coefficients for Range (coded units) Effect Coef 0.026806 6.13625 -0.15 -0.13403 0.077006 18.06812 0.03125 0.06125 0.01562 1 0.05 0.005 C 0.21937 0.01313 A DE 95 0.096 E -0.10 -0.02437 -0.88 0.03380 0.20) -0.01625 0.074256 17.11375 0.03062 90 A -0.50 0.02625 0.06312 99 0.008 B -0. E Estimated Effects and Coefficients for Range (coded units) Term Effect Coef SE Coef T P Constant 0.05688 (response is Range.Chapter 13 Exercise Solutions 13-7 continued (c) The full model for range: Factorial Fit: Range versus A.03380 0.01625 0.12625 -0.011 2-Way Interactions 1 0.23 0.13625 -0.81 0.01063 0.01688 60 50 0.01375 -0. C.03625 0.000 A 0.01375 -0.42 0.05688 0.21937 Normal Probability Plot of the Effects 0.01313 0.02125 -0.15 Effect 0.11375 0.07701 0.02625 0.003 A*D*E 0.01625 3.50 0. Alpha = .683 C*E -0.88 0.00687 80 0.04875 0.13875 0.31909 … 13-14 .13875 0.03062 0. C.05 0.02188 70 -0.01625 13.01625 1.00188 30 20 0.07426 0.27 0. B.13403 0. D.34 0.00812 B 10 -0. B.01625 -4.07426 0. D.03375 -0.050625 The reduced model for range: Factorial Fit: Range versus A.06937 0.19 0.06937 Percent Estimated Term Constant A B C D E A*B A*C A*D A*E B*E C*E D*E A*B*E A*C*E A*D*E Effect Ty pe Not Significant Significant F actor A B C D E N ame A B C D E Lenth's PSE = 0.443 D 0.00 0.06312 0.06125 0.01625 4.01812 40 -0.004225 Total 15 0.003 … Analysis of Variance for Range (coded units) Source DF Seq SS Adj SS Adj MS F P Main Effects 5 0.06812 5 CE -0.12625 -0.07701 0.01625 -0.01625 -3.003 3-Way Interactions 1 0.00687 0.58 0.10 0.

02185 -0.01768 90 A -0.00877 0.11744 0. B.01556 0. C.001 For both models of variability.007559 2.01997 0.56 Residual Error 8 0.06 0.007314 0.07148 -0.02 0.02 0.003 B -0.007559 15. B.00684 -0.0083496 9.001 … Analysis of Variance for StdDev (coded units) Source DF Seq SS Adj SS Adj MS F Main Effects 5 0.73 0.00468 -0.Chapter 13 Exercise Solutions 13-7 (c) continued The full model for standard deviation: Factorial Fit: StdDev versus A.20) 0.00438 CE 10 -0.00778 1 -0.06 -0.45 0.001 C 0.04 0. Factors C and E are included to keep the models hierarchical.047 E -0.73 0.07643 0.06259 0.00329 -0. D.504 D 0.03129 0.07149 -0.663 C*E -0.08 0. C.03574 0.07149 -0.007559 0.04 -0.06 0.06259 0.03822 0.01768 0.01540 0.00953 30 -0.00234 0.13 2-Way Interactions 1 0.041748 0.34 0.03574 0.020438 0.0009142 Total 15 0.54 0. E Estimated Effects and Coefficients for StdDev (coded units) Term Effect Coef SE Coef T P Constant 0.00342 0.00684 -0.01057 0.041748 0. along with factors B (heating time) and A (furnace temperature) are significant.00999 Effect 0.03129 99 -0.11744 (response is StdDev.03536 0.00342 D 80 0.092869 P 0.007559 -0.01906 0.70 0. 13-15 .03536 0.001 0.03822 Percent Estimated Term Constant A B C D E A*B A*C A*D A*E B*E C*E D*E A*B*E A*C*E A*D*E Effect Ty pe Not Significant Significant F actor A B C D E N ame A B C D E Lenth's PSE = 0.01057 0.14 0.00528 0.000 A 0.0232179 The reduced model for standard deviation: Factorial Fit: StdDev versus A.00 0.023369 0.004 0.00165 20 0.007559 5.007559 -4. D.007559 4.020438 0.03574 5 B -0. Alpha = .01093 50 40 0.0204385 22.007314 0.07643 0.07148 -0.03574 0.00528 A DE 95 0. interactions CE (transfer time × quench oil temperature) and ADE=BCE.08 -0.007559 -4.001 A*D*E 0.36 3-Way Interactions 1 0.023369 0.00770 70 60 -0. E Effects and Coefficients for StdDev (coded units) Effect Coef Normal Probability Plot of the Effects 0.0233690 25.

0 0.0 1.0 0.0 2 -1.2 Fitted Value 0.5 1 0 -1 1.00 Residuals Versus the Order of the Data 2 0 Standardized Residual Percent Normal Probability Plot of the Residuals 99 90 50 10 1 2 0 -2 -2 1 5 -1.0 0.0 (response is Height) 0 -0.5 1 Observation Order 1 -1.0 10 15 20 25 30 35 40 45 Observation Order -0.0 1.5 Residuals Versus C (response is Height) 2 -1.5 2 0 -2 1.0 0.5 1.50 7.5 0.0 -0.4 Residuals Versus the Order of the Data 1 0 -1 1 2 3 4 5 6 7 8 (response is Range) -1.75 Fitted Value 8.5 0.0 D 0.0 -1.5 0.0 E Standardized Residual Standardized Residual For range: Residual Plots for Range Histogram of the Residuals 3.0 0.Chapter 13 Exercise Solutions (d) Standardized Residual Standardized Residual For mean height: Residual Plots for Height -2 0 Standardized Residual 2 Frequency Histogram of the Residuals 10 5 0 -2.0 -0.5 1.0 0.0 B 0.5 Residuals Versus B Standardized Residual Standardized Residual -2 0.0 B 0.5 1.0 1.0 -2 -1.5 Standardized Residual Residuals Versus A Residuals Versus the Fitted Values (response is Height) 2 0 -2 7.0 0 Residuals Versus D 0 -2 -0.0 -1.5 1.0 A Residuals Versus C (response is Range) (response is Range) 0.5 0.5 Standardized Residual 0.0 E 13-16 .5 -1 Residuals Versus B -1 -1.5 1.0 D 0.0 C Residuals Versus E (response is Height) (response is Height) Standardized Residual Standardized Residual 1.5 -1.5 2 1.0 A 0.5 -1.5 0.0 -0.5 1.0 0 9 10 11 12 13 14 15 16 0 -0.0 -0.5 0.5 0.0 -1.5 1.0 -0.0 1.0 Standardized Residual 0 2 Standardized Residual 1 0 -1 Standardized Residual -2 Residuals Versus A Residuals Versus the Fitted Values Standardized Residual Standardized Residual Frequency Percent Normal Probability Plot of the Residuals 99 90 50 10 1 0.5 0.5 0.0 -0.0 1 0 -1 -1.0 -1.0 0.0 C Residuals Versus E (response is Range) (response is Range) Standardized Residual Residuals Versus D 1 0 -1 -0.5 0.5 0.0 -0.

5 -2 (response is StdDev) 0 -2 0.0 2 -1.5 0.5 -1. 13-17 . Plots of residuals versus each factor indicate that the variance may be different at different levels of factor D.0 0 Residuals Versus D -0.5 Residuals Versus C (response is StdDev) 2 -1.5 Standardized Residual 2.0 -0.0 0.0 A 0.0 B 0. A resolution V design can be generated with E = ± ABCD.0 -0.Chapter 13 Exercise Solutions 13-7 (d) continued Standardized Residual Standardized Residual For standard deviation: Residual Plots for StdDev -2 0 Standardized Residual 2 Histogram of the Residuals 4 2 0 -1.5 0.5 1.2 Fitted Value Residuals Versus the Order of the Data 2 0 -2 1 2 3 4 5 6 7 8 Standardized Residual Frequency Percent Normal Probability Plot of the Residuals 99 90 50 10 1 2 0 -2 -1.0 (response is StdDev) 0 -0. Range Plot of residuals versus predicted shows that variance is approximately constant over range of predicted values. Residuals normal probability plot indicate normality assumption is reasonable Plots of residuals versus each factor indicate that the variance may be different at different levels of factor D.0 1. (e) This is not the best 16-run design for five factors.5 0. Standard Deviation Residuals versus predicted plot and residuals normal probability plot support constant variance and normality assumptions.0 1.1 0. Plots of residuals versus each factor shows that variance is less at low level of factor E.0 2 0 -2 -1.5 -1. then none of the 2-factor interactions will be aliased with each other.0 9 10 11 12 1 3 14 15 16 -0.0 E Mean Height Plot of residuals versus predicted indicates constant variance assumption is reasonable.0 0.0 C 1.0 Residuals Versus E (response is StdDev) Standardized Residual Standardized Residual 0.0 D 1.5 2 1.0 0.5 0.0 -0.0 Residuals Versus A Residuals Versus the Fitted Values (response is StdDev) 2 0 -2 0.5 Observation Order Residuals Versus B Standardized Residual Standardized Residual -2 0.5 1.5 0. Normal probability plot of residuals support normality assumption.

and then contour plots in factors A. Free Height Variance could be calculated in the MINITAB worksheet. Overlay Plot DES IG N-E XP ERT P lo t 1.081B + 0. These contour plots could be compared with a contour plot of Mean Free Height.49 and minimum standard deviation of 0.00 A: fu rn te m p 13-18 .02 For the current factor levels.63 + 0. h 14 51 04 X -0 . so Variance of Free Height = (–0.00 0. This approach is fully described in the solution to Exercise 13-12.056 at A = –0. 49 59 4 POE( free height): 0.0 0 D: h o ld ti m e = 0 .0 0 E : o il te m p = 0 .077B)2 + σ2 Assume (following text) that σ2E = 1 and σˆ 2 = MSE = 0.99.Chapter 13 Exercise Solutions 13-8. and D could be constructed using the Graph > Contour Plot functionality.4 1 Y 0.12A – 0.0 0 -1 .15 PO E(f re e 0.50 f r ee h eight: 7. 55 0. B. 99 A ctu a l Fa cto rs C: tra n s tim e = 0 .00 -0 .046D Variance of Free Height = σ2E (–0.44 and B = 0.0 0 B: h ea t tim e 0.00 -0.12 + 0.077B)2 + 0.00 O ve rl a y Pl o t X = A: fu rn te m p Y = B: h e a t tim e f ree he igh7.50 1. Using equations (13-6) and (13-7) the mean and variance models are: Mean Free Height = 7.5 0 -1.12 + 0. and optimal settings could be identified from visual examination of both plots.02 . The overlaid contour plot below (constructed in Design-Expert) shows one solution with mean Free Height ≅ 7. Factor E is hard to control (a “noise” variable).50 0 .

B.00 -0. Assume σ D2 = σ E2 = 1 .00 A: fu rn te m p 13-19 .42 and B = 0.046)2 + σ2E (–0. 55 0. 42 POE( free height): 0.02 For the current factor levels. Overlay Plot DES IG N-E XP ERT P lo t 1. The overlaid contour plot below (constructed in Design-Expert) shows one solution with mean Free Height ≅ 7.1 52 23 3 X -0.50 f r ee h eight: 7.077B)2 + σ2 Using σˆ 2 = MSE = 0.50 0 .50 and minimum standard deviation of Free Height to be: A = – 0.12 + 0.12 + 0.50 1.0 0 E : o il te m p = 0 .16 Y 0 .00 -0 .0 0 D: h o ld ti m e = 0 .081B Variance of Free Height = σ2D (+0.Chapter 13 Exercise Solutions 13-9. and optimal settings could be identified from visual examination of both plots.99. the mean and variance are: Mean Free Height = 7.5 0 -1. and D could be constructed using the Graph > Contour Plot functionality.00 0.02 : Variance of Free Height = (0.046)2 + (–0. Free Height Variance could be calculated in the MINITAB worksheet.0 0 -1 .12A – 0. Factors D and E are noise variables. and then contour plots in factors A. Using equations (13-6) and (13-7). This approach is fully described in the solution to Exercise 13-12.077B)2 + 0.63 + 0.9 9 A ctu a l Fa cto rs C: tra n s tim e = 0 .4 95 01 PO E(f re e 0h.0 0 B: h ea t tim e 0. These contour plots could be compared with a contour plot of Mean Free Height.00 O ve rl a y Pl o t X = A: fu rn te m p Y = B: h e a t tim e f ree he igh7 .

000 x2*x2 -22. Since the runs are listed in a patterned (but not standard) order.710 -2.914 29.8750 x1*x3 -2.192 0.067 Total 19 2636.289 1. 80.132 3. Estimated Regression Coefficients for y Term Coef SE Coef T P Constant 87. 100. 90.081 3.000 0.66 13. These values are used in the Excel and MINITAB data files. 60. 87.001 0.723 23.176 1. 80. 88.000 x3*x3 2.29 2499.027 0. one approach to solving this exercise is to create a general full factorial design in MINITAB.0578 x2*x2 -7. 78.710 -3.000 0.33 92.74 405.8862 x3*x3 0. x3 The analysis was done using coded units.0509 x1*x1 -5. 70.6250 P 0.766 Lack-of-Fit 5 92. 63.513 57.689 -6.000 x1 9. and then enter the data.195 2.305 2.33 9. 75.33 18.689 5.000 0.17 329.764 0. 70.359 1.95 Square 3 1217.175 0. Note: Several y values are incorrectly listed in the textbook.205 x3 -10.710 2. x2.MTW.689 1.29 277. 68.7759 x1*x2 2. Stat > DOE > Response Surface > Analyze Response Surface Design Response Surface Regression: y versus x1.526 0. 65.764 -5.Chapter 13 Exercise Solutions 13-10.764 -8.04 Pure Error 5 45. The correct values are: 66.805 0.8279 x2 1.17 989. 83.458 7.3613 x3 -6.005 … Analysis of Variance for y Source DF Seq SS Adj SS Adj MS F Regression 9 2499.446 x1*x2 8. 82.3589 x1 5.000 x2 2.17 Linear 3 989.425 3.053 x1*x3 -7. 100. The design and data are in the MINITAB worksheet Ex13-10.74 1217.08 Residual Error 10 137.073 x2*x3 -13.000 x1*x1 -14.730 0.466 2.801 1.33 45. 91.49 Interaction 3 292.699 20.95 … Estimated Regression Coefficients for y using data in Term Coef Constant 87.069 0.227 uncoded units 13-20 .38 97.008 0. 85.6250 x2*x3 -4.794 0.38 292.305 2.66 137.356 0.

660 5.61 2490.905 0.523 2.5 0.0 -2.00 16.0 -2.000 x1 9.263 69.685 0.303 Lack-of-Fit 6 101.523 2.0 -2 -1 0 x2 1 2 -2 -1 0 x3 13-21 .067 Total 19 2636.036 0.Chapter 13 Exercise Solutions 13-10 continued Reduced model: Response Surface Regression: y versus x1.33 P 0.33 45.371 0.006 1.176 1.61 311.0 2.000 x1*x1 -14.327 Linear 3 989.00 101.048 x1*x3 -7.000 0.704 -5.660 1.5 -5.000 x1*x2 8.081 3.834 Pure Error 5 45.44 7.17 989.5 -5 1 -5 0 Residual 5 60 Histogram of the Residuals 80 Fitted Value 100 Residuals Versus the Order of the Data Residual Percent Residuals Versus x1 Residuals Versus the Fitted Values 99 Residual Frequency 5 4 2 0. Estimated Regression Coefficients for y Term Coef SE Coef T P Constant 87. x2.000 x2*x2 -22.0 -2.329 0.0 -5 0 -4 -2 0 2 Residual 4 2 4 -2 6 8 10 12 14 16 18 20 Observation Order -1 Residuals Versus x2 1 2 1 2 Residuals Versus x3 (response is y) (response is y) 5.5 2.0 0 2.40 24.994 1.000 0.38 97.131 0. x3 The analysis was done using coded units.132 3.534 Interaction 3 292.379 0.704 -8.5 Residual Residual 0 x1 0.458 Residual Error 11 146.34 146.067 x2*x3 -13.425 3.95 … F 23.289 1.86 0.647 -3.000 x2 2.647 2.07 604.660 -6.33 9.07 1209.587 0.38 292.000 0.17 329.0 5.78 45.5 0 -5.801 1.34 13.723 Square 2 1209.0 -5.195 x3 -10.229 0.647 -2.004 … Analysis of Variance for y Source DF Seq SS Adj SS Adj MS Regression 8 2490.257 Residual Plots for y Normal Probability Plot of the Residuals (response is y) 5 90 Residual 50 10 5.

0 50. 13-22 .0 80.807.0 > 1 Hold Values x2 0 70.0 60.0 100. Lower = 60.0 Hold Values x2 0 x3 2 x2 105 0 90 y 75 2 60 -1 0 -2 0 x1 -1 0 x1 2 x3 -2 1 Stat > DOE > Response Surface > Response Optimizer Goal = Maximize. x1 Surface Plot of y vs x3.0 100.0 Hold Values x3 0 0 80 y 60 2 40 -1 0 -2 0 x1 -1 0 x1 -2 1 Contour Plot of y vs x3.0 50.0 70. Upper = 120. x1 y < 70. and x3 = −1. x1 y < > Hold Values x3 0 40. x1 40. Weight = 1.0 80.682.0 1 x2 Surface Plot of y vs x2.0 60.0 70.Chapter 13 Exercise Solutions 13-10 continued Stat > DOE > Response Surface > Contour/Surface Plots Contour Plot of y vs x2.0 90. Importance = 1 One solution maximizing growth is x1 = 1. Predicted yield is approximately 108 grams.0 90.0 80.292. x2 = 0.0 80.

one approach to solving this exercise is to create a general full factorial design in MINITAB.86 1.788 Interaction 1 144.457 1.60 63.102 0.00 F 2.80 14. Since the runs are listed in a patterned (but not standard) order.58 0. x2 The analysis was done using coded units.128 0.555 0.60 139.000 x1 -1.348 2.017 13-23 .388 0.038 … Analysis of Variance for y Source DF Seq SS Adj SS Adj MS Regression 5 315.53 P 0.58 123.80 6. and then enter the data.712 1.Chapter 13 Exercise Solutions 13-11.383 0.274 x2 1.660 0. The design and data are in the MINITAB worksheet Ex13-11.02 24. Stat > DOE > Response Surface > Analyze Response Surface Design Response Surface Regression: y versus x1.119 Linear 2 48.781 1.100 19.660 -1.000 Residual Error 7 154.58 61.011 Square 2 123.00 144.209 x1*x2 6.200 2.616 0.409 x1*x1 3.000 2.09 2.60 46.186 0.878 0.058 Lack-of-Fit 3 139.02 48.085 0.970 1. Estimated Regression Coefficients for y Term Coef SE Coef T P Constant 41.40 154.40 22.076 x2*x2 2.80 3.60 315.781 2.MTW.038 12.463 1.700 Total 12 470.00 144.534 Pure Error 4 14.

0 -0.5 x2 Surface Plot of y vs x2.Chapter 13 Exercise Solutions 13-11 continued Contour Plot of y vs x2. Weight = 10. Upper = 50. to achieve predicted filtration time of 46.3415 and pressure = -0. 13-24 . Importance = 1 Recommended operating conditions are temperature = +1. x1 1.5 0 x2 -1 1 1. Weight = 1.5 50 40 -1.0785.0.0 x1 0.0 (a) Goal = Minimize. (b) Goal = Target. Upper = 55.4109 and pressure = -1. Target = 0.5 0.0 1 0 -1 x1 -1. Importance = 1 Recommended operating conditions are temperature = +1.0 60 y -0.4142. Lower = 42.0 40 45 50 55 0. Target = 46. x1 y < > 40 45 50 55 60 60 70 0. to achieve predicted filtration time of 36.7.

873 5.074 3.56 135.455 -4.22 30.1517 x1*x2 2. The design and data are in the MINITAB worksheet Ex13-12.4147 2.072 0.33 45.001 0.458 5.29 317.8333 3.116 0. Estimated Regression Coefficients for y Term Coef SE Coef T P Constant 87. x2.496 0.2894 1.092 15.000 x1 9.975 0.361 -4.38 292.209 0.3613 z -6.6250 x2*z -4.067 Total 17 2170.001 0.8750 x1*z -2. so there is a robust design problem.448 -3.115 uncoded units The coefficients for x1z and x2z (the two interactions involving the noise variable) are significant (P-values ≤ 0.003 x2*x2 -21.968 0.1250 x1*x1 -4.32 Pure Error 5 45.75 Interaction 3 292.945 Lack-of-Fit 3 90.7783 2.001 x2 2.232 0.945 x1*x2 8.105 13.94 2034.178 0.22 90.6250 P 0.38 97.804 0. z The analysis was done using coded units.28 263.873 1.256 z -6.361 -6.7192 z*z 0.448 -1.50 … Estimated Regression Coefficients for y using data in Term Coef Constant 87.003 x1*x1 -13.000 z*z 0.56 16.29 953.116 1.MTW Stat > DOE > Response Surface > Analyze Response Surface Design Response Surface Regression: y versus x1.Chapter 13 Exercise Solutions 13-12.222 0.8333 3.681 51.10).001 0.116 0.53 Square 3 953.8279 x2 1.1317 4.109 x2*z -7.3333 x1 5.1250 1.3333 1.1517 2.28 789.021 0.94 226.8013 1.084 x1*z -4.8908 x2*x2 -7.33 9.764 18.013 … Analysis of Variance for y Source DF Seq SS Adj SS Adj MS F Regression 9 2034. 13-25 .34 Linear 3 789.75 Residual Error 8 135.

289 1.0 -0.38 97.31 22.86 2034.882 2.0 13-26 .5 2.760 3.5 Residual Residual 0 x1 0.0 z 0.132 3.000 x1*x2 8.000 x1 9.20 476.62 6.008 … Analysis of Variance for y Source DF Seq SS Adj SS Adj MS Regression 8 2034.227 z -6. Estimated Regression Coefficients for y Term Coef SE Coef T P Constant 87.125 1.28 263.767 5.50 … F 16.066 x1*z -4.002 x1*x1 -13.0 -5.5 0 -5.462 0.000 x2 2.912 0.067 Total 17 2170.0 Residuals Versus x1 Residuals Versus the Fitted Values 99 -2.0 5.675 0.458 Residual Error 9 135.000 0.000 0.778 2.019 -7.20 953.092 Square 2 953.558 0.578 Pure Error 5 45.28 789.308 -1.5 0.296 0.64 15.64 135.88 17. x2.373 -4.86 254.31 90.5 -5.0 0 2.5 -5 -5 0 Residual 5 60 Histogram of the Residuals 80 Fitted Value 100 Residuals Versus the Order of the Data Residual Percent 50 1 0.357 Linear 3 789.760 3.370 0.072 Lack-of-Fit 4 90.001 x2*x2 -21.0 5 Residual 3.095 0.415 2.019 -4.0 -2 -1 0 x2 1 2 -1.33 45.308 -3.46 31.088 x2*z -7.0 -2.172 Residual Plots for y Normal Probability Plot of the Residuals Residual 10 5.801 1.361 1.49 0.38 292.5 1.Chapter 13 Exercise Solutions 13-12 continued Reduced model: Response Surface Regression: y versus x1.47 P 0.5 0.208 0.602 Interaction 3 292.0 2.5 0.767 1.548 0.0 -2.013 2.0 -5 -4 -2 0 2 Residual 4 2 4 6 8 10 12 14 16 Observation Order -2 18 -1 Residuals Versus x2 1 2 Residuals Versus z (response is y) (response is y) 5.000 0.541 56.0 Frequency (response is y) 5 90 1.33 9. z The analysis was done using coded units.

293782 y = 90.0 60.36x2 – 4.11 and x2 = – 0.0 1 x2 x2 -1 0 x1 Contour Plot of y vs x2.12 > 12 0 40.31 (mean yield of about 90 with a standard deviation between 6 and 8).63x1 – 4.13 – 2.63x2)2 + σˆ 2 = (–6.36x2 – 4.0 70.0 50.z)]} < 6 6 8 8 .69x22 + (–6.0 50.0 60.0 40. x1 1 30.z)]} vs x2.Chapter 13 Exercise Solutions 13-12 continued yPred = 87.0 80.0 70. There are other combinations that would work.13 – 2.63x1 – 4. x1 y < > -1 -1 Contour Plot of sqrt{Vz(y(x.0 70.36 + 5.86x12 – 7.708407 x2 = -0.63x1 – 4.0 60.0 80.36 + 5.109756 x2 = -0.10 10 .0 80.0 70.0 Hold Values z 1 0 -1 -1 0 x1 1 -1 0 x1 1 Examination of contour plots for Free Height show that heights greater than 90 are achieved with z = –1.0 60.83x1 + 1. (Refer to MINITAB worksheet Ex13-12.1692 50. assume σz2 = 1: Variance of Yield = σz2 (–6.63x2)2 + 15. set z = 0: Mean Yield = 87.308367 y = 90.072 This equation can be added to the worksheet and used in a contour plot with x1 and x2.0 1 x2 50.0 90.0 Hold Values z -1 0 x1 = -0.13 – 2.0 60.0 40.0 50.0 60.69x22 For the variance model. 13-27 . x1 sqrt{Vz(y(x.0 1 -1 Contour Plot of y vs x2.) Contour Plot of y vs x2.0 80.63x2)z For the mean yield model. x1 x2 x1 = -0.0 90.86x12 – 7.0 80.MTW.0 50.0 80.2054 0 x1 40.0198 -1 x1 = 0.0 Hold Values z 0 0 1 1 y < > y < > 30.393500 x2 = 0.555312 y = 90. Comparison with the contour plot for variability shows that growth greater than 90 with minimum variability is achieved at approximately x1 = – 0.83x1 + 1.0 70.0 70.

13-28 . and i =1 i =1 i =1 u =1 i< j =2 i =1 ∂zi r ∑ r r k r r ⎡ k r ⎤ V [ y (x. z ) = ∑ γ i zi + ∑ ∑ δ ij xi z j . i =1 i =1 j =1 i< j =2 r k r ∂h(x. z ) = ∑ γ i + ∑ ∑ δ ui xu + ∑ ∑ λij ( zi + z j ) + 2∑ θ i zi . then i =1 i =1 j =1 ( V [ y (x. then i =1 i =1 j =1 r ∂h(x. and i =1 i =1 i i =1 u =1 i< j =2 ∂zi r then ∑ r ⎡ k r ⎤ V [ y (x. z ) ] = σ ∑ γ i + ∑ δ ui xu 2 z r r i =1 k u =1 k ) +σ 2 r 2 r If h(x. z ) r = ∑ γ + ∑ ∑ δ ui xu + ∑ ∑ λij ( zi + z j ) . r k k ∂h(x. z ) = γ i + ∑ δ ui xu .Chapter 13 Exercise Solutions 13-13. z ) = ∑ γ i zi + ∑ ∑ δ ij xi z j + ∑ ∑ λij zi z j + ∑ θi zi2 . 13-14. z ) ] = V ∑ ⎢γ i + ∑ δ ui xu + ∑ λij ( zi + z j ) ⎥zi + σ 2 i =1 ⎣ u =1 j >i ⎦ There will be additional terms in the variance expression arising from the third term inside the square brackets. and u =1 ∂zi r If h(x. r k r r r i< j =2 i =1 If h(x. z ) ] = V ∑ ⎢γ i + ∑ δ ui xu + ∑ λij ( zi + z j ) + 2θi zi2 ⎥zi + σ 2 i =1 ⎣ u =1 j >i ⎦ There will be additional terms in the variance expression arising from the last two terms inside the square brackets. z ) = ∑ γ i zi + ∑ ∑ δ ij xi z j + ∑ ∑ λij zi z j .

26986 0. p 0.02863 Pr{d<=c} 0.16434 0.000 0. c=1 1.04428 0.20 0.93910 0.00896 0.81840 0.00515 f(d=1) 0.21807 0.99881 0.007 0.95190 0.00 Pr{acceptance} 0.86051 0.95121 0.005 0. 14-1.19556 0.97387 0.09066 0.01547 0.70382 0.20 1.009 0.xls.03379 Type-B OC Curve for n=50.37160 0.020 0.003 0.55528 0.006 0.05324 0.04533 0.010 0.74015 0.91056 0.100 f(d=0) 0.20249 0.90475 0.36417 0.02656 0. The BINOMDIST.Chapter 14 Exercise Solutions Note: Many of the exercises in this chapter are easily solved with spreadsheet application software.40048 0.24807 0.07694 0.12989 0.040 0.98998 0. HYPGEOMDIST.73577 0.001 0.100 0.080 0.140 p 14-1 .60 0.050 0.060 0.08271 0. and graphing functions in Microsoft® Excel were used for these solutions.33721 0.19000 0.12947 0.04761 0.090 0.96353 0.12649 0.27943 0.40 0.004 0.27060 0. Solutions are in the Excel workbook Chap14.77831 0.020 0.030 0.98274 0.80 0.008 0.060 0.60501 0.63633 0.30556 0.22339 0.00 0.99540 0.14467 0.28895 0.09994 0.080 0.92528 0.66924 0.040 0.120 0.002 0.070 0.06725 0.

00003 0.81857 0.000 0.00162 0.008 0.04144 0.26899 0.66978 0.120 0.30441 0.11826 0.040 0.00194 0.00030 0.080 0.99985 0.41978 0.00205 0.Chapter 14 Exercise Solutions 14-2.18486 0.00000 Pr{d<=c} 0.100 0.050 0.60577 0.80 0.99225 0.92063 0.23214 0.030 0.00000 f(d=2) 0.00388 0.36120 0.07029 0.00531 0.00208 0.97730 0.003 0.001 0.060 0.00476 0.004 0.020 0.002 0.13262 0.49536 0.00071 0.00 0.020 0.100 0.00895 0.01978 0.09057 0. c=2 1.00000 Type-B OC Curve for n=100.060 0.16404 0.00008 0.009 0.22281 0.01627 0.01687 0.60 0.05347 0.07572 0.27065 0.33068 0.08118 0.20 0.14419 0.93796 0.36773 0. p 0.40492 0.140 p 14-2 .96641 0.27341 0.01312 0.90479 0.20 1.36603 0.14498 0.00449 0.00024 0.04755 0.95327 0.200 f(d=0) 0.40 0.34920 0.040 0.16531 0.090 0.12185 0.010 0.14707 0.007 0.98590 0.36973 0.54782 0.03116 0.67669 0.006 0.05661 0.070 0.00079 0.02579 0.00000 f(d=1) 0.74048 0.00 Pr{acceptance} 0.99888 0.00592 0.44789 0.22515 0.080 0.01127 0.09880 0.03319 0.99649 0.005 0.

020 0.040 0.600 0.10 (b) Type-B OC Curve for N=5000.000 Pr{acceptance} 0. either is appropriate.10133. or β ≅ 0. or α ≅ 0.007) = 0.00 Pr{acceptance} 0.05 Pa (p = 0.10133.075) = 0.060 0.140 p Pa (p = 0.200 0. or α ≅ 0.100 0.040 0.20 0.060 0.80 0. n=50.9521.05 Pa (d = 375) = 0.000 0.080 0. (a) Type-A OC Curve for N=5000. c=1 1.40 0. or β ≅ 0. c=1 1.020 0.120 0.00 0.9521.000 0.60 0.140 p Pa (d = 35) = 0.800 0.120 0. the difference between the two curves is small.100 0.10 (c) Based on values for α and β. 14-3 .080 0.Chapter 14 Exercise Solutions 14-3.000 0. n=50.400 0.

05 = 0. p2 = 0. p1 = 0.12238.05. p1 = 0.05 = 0.10 From the binomial nomograph. the sampling plan is n = 80 and c = 7. resulting in actual α = 0. p2 = 0.01 = 0.10 From the binomial nomograph.02. p1 = 0.99.Chapter 14 Exercise Solutions 14-4.15.04786 and β = 0.01.95. 14-4 . 1 − α = 1 − 0.06. β = 0. 14-5. 1 − α = 1 − 0.95.10. p2 = 0. select n = 35 and c = 1. β = 0. 14-6. 1 − α = 1 − 0.10 From the binomial nomograph. β = 0. select a sampling plan of n = 300 and c = 12.

the consumer is protected from a LTPD = 0.01175 0.11479 0. but pays for the high probability of rejecting acceptable lots like those with p = 0. 14-5 . LTPD = 0.0000 0.00046 and Pa(N = 10.025.0000 0.0000 0.000.000.4170 0.0300 0.00600 -0.00938 0.00046 0.0010 0.0003 1. while for N = 10.0064 0.00000 5000 500 0.00967 0.0001 1.025) = 0.7060 0.0010 0.0050 0.0040 0.0200 10 N2 = n1 = pmax = cmax = binomial Pr{reject} Pr{d<=20} 0.99850 0.05 by Pa(N = 5.00006 -0.0600 0.99994 0.4409 0.00000 0.0500 0.0100 0.0080 0.00000 0.00000 0. Pa(p = 0.0250 0.00000 0.0027 0.9667 0.0000 1.0000 difference 0.0000 0.99999 0.00030 0.0400 0.99903 0.Chapter 14 Exercise Solutions 14-7.00000 0.00000 1.00000 0.99359 0.00000 0.00001 0.0070 0.00263 -0.00000 -0.11183 0.00000 Different sample sizes offer different levels of protection.0030 0.03328 0.00000.00000 1.99991 0.00046 0.0015 0.00000 0. For N = 5.0000 1.0000 1.0200 0.0000 0.00027 -0.98676 0.0700 N1 = n1 = pmax = cmax = binomial Pr{d<=10} 1.02395 0.00000 1.00095 -0.025) = 0.9903 0.0000 0.00000 1.0020 0.55910 0.294.0000 0.9995 0.99959 0.0090 0.00000 10000 1000 0.0200 20 Pr{reject} 0.00000 0. Also.9997 1.00001 -0.000) = 0.0000 1.08151 0. Pa(p = 0.8852 0.05 p 0.0001 0.18221 0.0000 1.00001 0.0004 0.99999 0.0132 0.99972 0.0000 1.0060 0.99729 0.29404 0.0000 0.000) = 0.8178 0.58304 0.182.0000 0.

2446 0.0700 0.3884 0.0080 0.1168 0.0602 0.9986 0.0900 0.98157 0.4689 0.01) = 0.00000 1. Pa(p = 0.00099 0.1000 0.3500 N1 = n1 = pmax = cmax = binomial Pr{d<=0} 0.8814 0.4654 0.0006 0.9246 0.0305 0.9426 0.7179 0.01 0 N2 = n1 = pmax = cmax = binomial Pr{reject} Pr{d<=1} 0.2204 0.0050 0.10543 0.0545 0.7961 0.1308 0.86924 0.1897 0.0010 0.81033 0.00292 0.73230 0.77958 0.4301 0.95886 0.0676 0.0000 1.7569 0.0070 0.75536 0.28210 0. and Pa(p = 0. at p = 0.03815 0.3000 0.24312 0.4949 for N = 5000.00002 0.53457 0.9618 0. p 0.0030 0.9868 0.88316 0.9990 0.01.61157 0.0008 0.65622 0.75558 0.98767 0.0000 1.0002 0.0062 0.99382 0.1439 0.0184 0.05374 0.0100 0.2000 0.0600 0.0040 0.93982 0.5052 0.00000 5000 71 0.9463 0.01315 0.00000 1.1702 0.80432 0.Chapter 14 Exercise Solutions 14-8.0004 0.2959 0.9938 0.9724 0.11858 0.0300 0.0889 0.07541 0.0000 This plan offers vastly different protections at various levels of defectives. depending on the lot size.00136 0.93236 0.70407 0.00000 1000 32 0.20391 0.0245 0.1957 0.97550 0.00622 0.6110 0. For example.8531 0.0800 0.0090 0.02758 0.0500 0.49484 0.01 1 Pr{reject} 0. 14-6 .0123 0.0020 0.01) = 0.0139 0.9994 1.0060 0.3438 0.0000 0.0400 0.7323 for N = 1000.05741 0.00063 0.2677 0.14688 0.9971 0.91107 0.53107 0.38898 0.96946 0.82981 0.94552 0.8946 0.2444 0.0276 0.0000 0.56992 0.0200 0.85608 0.98610 0.97238 0.0411 0.

15 0.05 0.0234 ATI Curve for n=35.010 0.10 0.00 0.00 0.25 0.020 AOQ 0. N = 2.05 0. n = 35.000 0.015 0.25 p AOQ Curve for n=35.Chapter 14 Exercise Solutions 14-9.20 0.025 0. c=1 0.005 0. c=1 2500 2000 ATI 1500 1000 500 0 0.15 0. c = 1.20 0.30 p 14-7 .10 0.000 ATI = n + (1 − Pa )( N − n) = 35 + (1 − Pa )(2000 − 35) = 2000 − 1965 Pa Pa p( N − n) N = (1965 2000 ) Pa p AOQ = AOQL = 0.

070 0.015 0.080 0.0000 2999 0.001 0.0072 588 0.030 0.004 0.035 0.0053 328 0.0009 2948 0.00 Pr{accept} 0.0087 AOQL 1265 0.060 0.01815 0.0034 2712 0.009 0. n = 150.0022 2834 0.0048 2517 0.0077 693 0.0061 404 0.070 0.002 0.60884 0.03292 0. c=2 1.91092 0.84615 0.090 0.00009 0.00142 0.10098 0.006 0.0067 491 0.010 0.80948 0.008 0.040 0.60 0.88019 0.050 0.42093 0.Chapter 14 Exercise Solutions 14-10.0065 2221 0.00 0.100 Pa=Pr{d<=2} 0.025 0.93769 0.98927 0.0080 1800 0.99646 0.020 0.010 0.050 0.005 0.0009 151 0.007 0.00002 AOQ ATI 0.80 0.040 0.0000 3000 (a) OC Curve for n=150.080 p 14-8 .16932 0.05840 0.030 0.0000 3000 0.00523 0. c = 2 p 0.0003 2985 0.95991 0.0037 215 0.000 0.0014 2906 0.0046 264 0.0019 160 0.0028 181 0.00036 0.003 0. N = 3000.99951 0.045 0.20 0.27341 0.40 0.97716 0.0001 2996 0.060 0.020 0.

c=2 AOQL ≅ 0.060 0.080 p (c) ATI Curve for n=150.0100 0.030 0.040 0.060 0.0000 0.020 0.020 0.0030 0.050 0.0050 0.030 0. c=2 3500 3000 2500 ATI 2000 1500 1000 500 0 0.0040 0.0070 AOQ 0.0087 0.000 0.070 0.070 0.010 0.Chapter 14 Exercise Solutions 14-10 continued (b) AOQ Curve for n=150.0090 0.0060 0.0010 0.0080 0.050 0.080 p 14-9 .010 0.000 0.040 0.0020 0.

1030 0. c=2 1.01382 0.2000 0.45947 0.68921 0.81080 0.20 0.0020 0.18920 0.60 0.00000 OC Curve for n=50.0800 0.0050 0.83946 0.09255 0. Pa 0.0030 0.00109 0.2000 Fraction defective.0010 0.89632 0.99474 0.98618 0.99871 0.00048 0.0200 0.1000 0.3000 0.0300 0.88827 0.1000 0.92157 0.1600 0.00 Probabilty of Acceptance.00343 0.0700 0.0900 0.1050 0.90745 0.99985 0.1040 0.0400 0.99998 0.0000 0.99242 0.00526 0.00206 0.0200 0.10764 0. n = 50.1400 0.99657 0.67671 0.89236 0.0800 0.58375 0.07843 0.0090 0.16054 0.77403 0.00002 0. (a) N = 5000.01043 0.0070 0.1200 0.0500 0.1010 0.31079 0.0600 0.00758 0.09614 0.0080 0.32329 0.22597 0.1020 0.99794 0.0040 0.0600 0.80 0.54053 0.0100 0.98957 0.00129 0.09985 0. p 14-10 .00 0.40 0.99952 0.00015 0. c = 2 p Pa=Pr{d<=1} Pr{reject} 0.11173 0.0060 0.1800 0.10368 0.41625 0.90015 0.00000 1.90386 0.0400 0.Chapter 14 Exercise Solutions 14-11.99891 0.

0400 0.0200 0. (d) From the nomograph.0900 0.98019 0.12158 0.18209 0.0050 0.81131 0.23424 0.45621 0.29011 0.83459 0.11340 0.81791 0.15164 0.94168 0.0500 0.55800 0.05832 0.85160 0.96075 0.88638 ≈ 0. with acceptance number c = 0.00000 Pr{reject} 0.98847 0.5000 Pa=Pr{d<=0} 0.0600 0.90461 0.1030 will be rejected about 90% of the time.0040 0.84836 0.00000 14-11 .2000 0.18869 0. not the sampling plan.0070 0.66761 0.35849 0.00080 0.Chapter 14 Exercise Solutions 14-11 continued (b) p = 0.3000 0.88660 0. quality improvement begins with the manufacturing process control.0090 0.87842 0.99996 1. The OC curve for this zero-defects plan is much steeper. will be extremely hard on the vendor because the Pa is low even if the lot fraction defective is low.86893 0.70989 0.76576 0.99920 0.0060 0.0800 0.64151 0.01153 0.92297 0.0700 0.0100 0. yielding Pa = 1 – 0.14840 0. select n = 20.1000 0.0300 0.90.0080 0.33239 0.00004 0.16541 0.0030 0.11372 = 0. (c) A zero-defects sampling plan.44200 0.4000 0. p 0.09539 0.07703 0.01981 0.0010 0.54379 0.0020 0. Generally.03925 0.13107 0.

00 Probability of acceptance.00206 ATIc =0 = n + (1 − Pa )( N − n) = 20 + (0.80 0.005. 14-12 . c = 0} = 0. Pa 0.09539 Pr{reject | p = 0.20 0.1500 0.2000 0.40 0.0500 0.005. p (e) Pr{reject | p = 0.2500 Fraction defective.60 0.0000 0.00 0.09539)(5000 − 20) = 495 ATIc = 2 = 50 + (0. c=0 1. c = 2} = 0.1000 0.00206)(5000 − 50) = 60 The c = 2 plan is preferred because the c = 0 plan will reject good lots 10% of the time.Chapter 14 Exercise Solutions 14-11 (d) continued OC Curve for n=20.

0098 0.0013 0.075 0.2770 0.0111 0.0001 0.0011 0.035 0.0522 0.0152 0.00 0.040 0.0000 0.0030 0.5405 0.0001 0.0002 0.0000 0.1635 0. n2=100.0112 0.7452 0.0000 0.0567 0.0001 0.0513 0.0000 0.5974 0.0024 0.6078 0.100 0.0001 0.0161 0.0001 0.7740 0.140 0.0444 0.000 0.0029 0.9661 0.0001 0.080 0.0000 0.0000 0.005 0.0793 0.0020 0.0521 0.3848 0.2658 0.0708 0.020 0.0000 0.0120 0.5207 0.0000 0.20 0.0791 0.0190 0.0000 0.065 0.0019 0.0004 0.0842 0. n2 = 100.9982 0.0001 0.60 0.0238 0.4162 0.0001 0.0002 0.0514 0.0051 0.6770 0.0221 0.0002 0.8706 0.7342 0.4763 0.0000 0.025 0.0001 0.0000 0.040 0.0820 0.0138 0.070 0.0002 0.0193 0.0000 0.0002 0.0004 0.0001 0.0129 0.0011 0.0339 0.0098 0.0009 0.0209 0.0000 0.060 0.080 0.0442 0.5838 0.0221 0.9737 0.8928 0.80 Pr 0.0142 Primary and Supplementary OC Curves for n1=50.1892 0.020 0.055 0.0000 0.0001 0.0767 0.120 0.0212 0.0072 0.0844 0.115 0.3108 0.0784 0.0019 0.1128 0.0165 0.d2<=3} Pr{d1=4.d3<=2} Pr{d1=5.0339 0.0000 0.8108 0.0006 0.0000 0.9414 0.0000 0.140 0.6390 0.0818 0.0021 0.0000 0.0015 0.0035 0.8395 0.090 0.0073 0. c2=6 1.1294 0.0690 0.0029 0.0000 0.0004 0.0001 0.0120 0.2333 0.0001 0.0330 0.9979 0.6767 0.0627 0.0008 0.0000 0.3274 0. c1 = 2.0707 0.2260 0.1605 0.2548 0.0000 0.0019 0.0002 0.010 0.0000 0.160 p.3233 0. c2 = 6 P d1 = PaI PrI 0.9487 0.1117 0.6892 0.0011 0.d2<=1} Pr{d1=6.045 0.0025 0.9216 0.0000 0.0000 0.0166 0.9237 0.3922 0.0238 0.40 0.0001 0.Chapter 14 Exercise Solutions 14-12.0692 0.100 0.8296 0.9373 0.0331 0.120 0.150 0.030 0.0142 0.3610 0.7560 0.00 0.0000 PaII Pa 0.9999 0.130 0. proportion defective Pr{accept on 1st sample} Pr{reject on 1st sample} Pr{accept lot} 14-13 .110 0.0028 0.9858 3 4 5 6 Pr{d1=3. c1=2.0629 0.0002 0.060 0.0073 0.0763 0.0011 0.4595 0.0000 0.d2=0} 0.050 0.9862 0.0000 0. n1 = 50.8883 0.4494 0.0000 0.0000 0.0023 0.0568 0.9779 0.5237 0.0004 0.0029 0.

h2 = 1.159 2.9389 + 1.820 -0.95.850 0.120 2.0414.009 1.9389 + 0.404 … 2.040 2.080 2.0397n n 1 2 3 4 5 … 20 21 22 23 24 25 … 45 46 47 48 49 50 XA -0.104 -0. p2 = 0.285 1.2054 = = 0.049 XR 1. Pa = h2 1. β = 0. (a) p1 = 0.0397.1 − α = 1 − 0.899 -0. Rej = 4.113 3. Pa = 1 − α = 0.5621 h1 + h2 0.025 0.10 14-14 .015 0.10 k = 1.969 1.780 -0.193 Acc n/a n/a n/a n/a n/a … n/a n/a n/a n/a 0 0 … 0 0 0 0 1 1 Rej 2 2 2 2 2 … 2 3 3 3 3 3 … 3 4 4 4 4 4 The sampling plan is n = 49.929 0.2054 p2 = 0.10.199 … 2. h1 = 0. Pa = β = 0.05 = 0.034 3.364 1.95 p = s = 0.000 2.740 … -0. s = 0.994 3.10.01.9389. Acc = 1.01.Chapter 14 Exercise Solutions 14-13.2054. (b) Three points on the OC curve are: p1 = 0. X R = 1.144 -0.890 0.0397 n.2054 + 0.153 3.245 1.055 … 0.074 3.064 -0.859 -0.0397 X A = −0.325 1.

670 … 2.1 − α = 1 − 0.991 2. Pa = β = 0.15.95.659 2. Pa = h2 1.978 -0.0660. (b) p1 = 0.507 4.375 4.923 2.472 1.02.474 0.0436 + 0.0660n.572 4.912 -0.606 … 1.0660 X A = −1.057 2.3399 = = 0.925 1.3399.05 = 0.0660n n 1 2 3 4 5 … 20 21 22 23 24 25 … 45 46 47 48 49 50 XA -0.0436.10 k = 0.95 p = s = 0. s = 0.538 1. p2 = 0.780 -0.02.441 4. Acc = 1 and Rej = 4.15. β = 0.255 XR 1.276 0. Pa = α = 0.9369.342 0.189 2. h2 = 1.10 14-15 .0436 + 1.989 … 4.408 0.725 2.Chapter 14 Exercise Solutions 14-14.846 -0.406 1.791 2.638 Acc n/a n/a n/a n/a n/a … n/a n/a n/a n/a 0 0 … 0 0 0 0 1 1 Rej 2 2 2 2 2 … 2 3 3 3 3 3 … 3 4 4 4 4 4 The sampling plan is n = 49.123 2. h1 = 1.714 … 0.309 4.3399 p2 = 0.3399 + 0.604 1.5622 h1 + h2 1.857 2. (a) p1 = 0.540 0. X R = 1.

n = 20. Ac = 1. Re = 4 Tightened sampling plan: n = 125.Chapter 14 Exercise Solutions 14-15. Ac = 0. General inspection level II. n = 50. Re = 2 14-18. Sample size code letter = L Normal: up to letter K. AQL = 1% General level II Sample size code letter = K Normal sampling plan: n = 125. Ac = 2. Re = 1 Tightened: n = 200. Ac = 1. Re = 4 14-17. n = 50. AQL = 1% General level I Normal sampling plan: Sample size code letter = H. Re = 2 Reduced sampling plan: Sample size code letter = H. Re = 2 Tightened sampling plan: Sample size code letter = J. Re = 3 Reduced sampling plan: n = 50. AQL = 0. AOQ = [ Pa × p × ( N − n)] [ N − Pa × (np) − (1 − Pa ) × ( Np) ] 14-16. N = 10. Ac = 0. n = 80.10%. N = 3000. N = 3000. Re = 1 14-16 . Ac = 1. Ac = 0. Re = 1 Reduced: up to letter K. Ac = 0. Ac = 3. n = 125.000.

3038 0.0400 0.9667 0.0030 0.9922 0.0050 0.0003 0.0200 0.1000 normal 200 3 Pa=Pr{d<=3} 0.7309 0. AQL = 0. Ac = 1.5230 0. Ac = 3.9989 0.9469 0.20 0.0500 0.9813 0.0060 0. (a) N = 5000.0125 0.0593 0. II. Sample size code letter = L Normal sampling plan: n = 200. Ac = 2. Re = 4 (b) N = 5000 n= c= p 0.8916 0.8800 0.0600 0.0001 0.1000 p.0700 0.0433 0.0395 0.0200 0.0004 0.40 0.0800 0.0022 OC Curves for N=2000.9886 0. Re = 3 Reduced sampling plan: n = 80.0101 0.9992 0.7838 0.9202 0.8922 0. proportion defective Normal Tightened Reduced 14-17 .65% 1.0090 0.0000 0.0040 0.0000 0.0100 0.0018 0.1472 0.9588 0. Re = 4 Tightened sampling plan: n = 200.1654 0.00 Pa.9220 0.0010 0.9999 0.00 0.8580 0.0400 0.8377 0.0900 0.65%.0000 reduced 80 1 Pa=Pr{d<=1} 0.8340 0.0000 0.6767 0.0070 0.0211 0.0861 0. AQL=0.4315 0.8092 0.0600 0.0001 0.0047 0.0023 0.9529 0.9163 0.9771 0.0800 0.9967 0.60 0.0300 0.2351 0. Pr{acceptance} 0.8653 0.0000 0.0080 0.9970 0.Chapter 14 Exercise Solutions 14-19.0000 tightened 200 2 Pa=Pr{d<=2} 0.9756 0. General level II.0020 0.80 0.9389 0.0090 0.9911 0.

016 0.004 0.004 0.010 0.007 0.c=2.1320 0.018 0.0031 ATI 511 605 767 962 1160 1341 1497 1564 1624 1724 1801 1858 1900 1930 1952 1967 1977 1985 1990 1993 1995 AOQ 0.008 0.0321 0.020 0.020 p.2% p 0.0046 0.0018 0. N = 2000.009 0. p = 0.002 0. AOQL=0.006 0.010 0.015 0.018 0.2886 0.8165 0. LTPD = 1%.004 0.008 0.0003 0.0464 0. Pr{acceptance} 2500 ATI 1500 1000 0.0005 0.014 0. without replacement.0008 0. n=490.020 D = N*p 2 4 6 8 10 12 14 15 16 18 20 22 24 26 28 30 32 34 36 38 40 Pa 0.0001 0.00 0.9235 0.0002 0. c = 2.0942 0.002 0.018 0.0220 0.005 0.9864 0.012 0.001 0.0068 0.012 p.21% ATI Curve for N=2000.016 0.0014 0.014 0. proportion defective 0.60 0. Note that this solution uses the cumulative binomial distribution in a spreadsheet formulation.006 0.00 2000 0.0010 0.0006 0.n=490.008 0.80 Pa.0102 0. A more precise solution would use the hypergeometric distribution to represent this sampling plan of n = 490 from N = 2000.0002 0.40 0. AOQL = 0.0664 0.5564 0.000 0.016 0.000 0.0021 0.006 0. proportion defective The AOQL is 0.0020 0.0000 OC Curve for N=2000.0001 0.011 0.0016 0.3330 0.6875 0. 14-18 .1827 0.2489 0.0015 0.017 0.002 0.010 0.019 0.0150 0.4361 0.003 0.0001 0.25% n = 490.Chapter 14 Exercise Solutions 14-20.20 500 0 0.0007 0.21%.0021 AOQL 0.0012 0.014 0.008 0.013 0.0018 0.012 0. c=2 1.

160 Fraction defective.00000 0. n=65.20000 0. p 0. (c) LTPD = 10.200 0 0. 001 Pa = Binom(3. average process fallout = p = 0. the average inspection required will be 82 units. 65.080 0. c=3 60.100 0.060 0.00000 50.40000 20.000 ATI Probability of acceptance. (b) ATI Curve for N=50.180 0. 001 − 65) = 82 On average.250 p let N = 50.50% defective (a) Minimum sampling plan that meets the quality requirements is 50. Pa OC Curve for Dodge-Romig.000 10.80000 40.3% 14-19 .200 0.20000 0.99967)(50. c=3 1. if the vendor’s process operates close to process average.100 0. c = 3.Chapter 14 Exercise Solutions 14-21.050 0.120 0. Dodge-Romig single sampling. AOQL = 3%.040 0.60000 0.000 0.150 0.99967 ATI = n + (1 − Pa )( N − n) = 65 + (1 − 0.020 0.140 0.000 0.000 1.000 0.005) = 0.001. 0. n=65.001 ≤ N ≤ 100.000.000 30.000 0. n = 65.

AOQL = 3%.9958 ATI = n + (1 − Pa )( N − n) = 65 + (1 − 0.9998 ATI = n + (1 − Pa )( N − n) = 46 + (1 − 0. LTPD = 10. 0.3% (b) c 3 d =0 d =0 Pa = ∑ binomial(n.25% n = 46.9998)(8000 − 46) ≈ 48 14-20 . LTPD = 11. p ≤ 1% n = 65. 0.0025) = 0.9958)(8000 − 65) ≈ 98 (c) N = 8000.01) = 0. p) = ∑ b(65.6% c 2 d =0 d =0 Pa = ∑ binomial(n.Chapter 14 Exercise Solutions 14-22. AOQL = 3%. c = 2. (a) N = 8000. c = 3. p ≤ 0. p ) = ∑ b(46.

000 0.180 0. Accept the lot if ⎡⎣ Z LSL = ( x − LSL ) S ⎤⎦ ≥ 1.900 1-alpha 0.800 0.7 ⎣ ⎦ Accept the lot.050 0. S = 1.xls : worksheet Ex15-1 From the variables nomograph at n = 35 and k = 1. the sampling plan is n = 35.10 = 0. (c) Excel workbook Chap15. k = 1.120 0. LSL = 0.005 0. β = 0.945 0.Chapter 15 Exercise Solutions 15-1. (b) x = 0.100 0.600 Pr{accept} p1 p Pr{accept} 0.70 ) 1.001 0.02.73.020 0.150 0.025 0. p1 = 0.05 × 10−2 ( ) ⎡ Z LSL = ( 0.010 0.988 0.060 0.050 beta 0. 1 – α = 1 – 0.70 g/cm3.000 0.38 (from nomograph) 15-1 .100 0.05 × 10−2 = 2.190 0.070 0.8571⎤ ≥ 1.7 .200 p Pa{p = 0.030 0.020 0.05 (a) From the variables nomograph.7: p2 OC Curve for n=25.190 0.90.10.820 0. k=1.7.040 0.000 0.7 1.400 0.730 0.200 0.560 0.160 0.400 0. Calculate x and S.080 0.040 0.140 0. p2 = 0.73 − 0.05} ≈ 0.016 0.

σ = 5 p1 = 0.05 = 0. kreduced = 1.51) can be obtained from the full set of tables in MIL-STD-414 using Table B-3.com/college/montgomery 15-5. No convenient mathematical method exists to do this.3. 15-4.1 − α = 1 − 0. ktightened = 1.10 From variables nomograph. .5%. AQL = 1. standard deviation unknown Assume single specification limit . LSL = 150. we know n and k are related. Inspection level II. because both Pa and pm are functions of n and k. knormal = 1. Now.Chapter 15 Exercise Solution 15-2. and generally involves some trial-and-error search. for a specified process average. Accept the lot if ⎡⎣ Z LSL = ( x − 150 ) S ⎤⎦ ≥ 2. for a given AOQL = Pa pm (N – n) / N (where pm is the value of p that maximizes AOQ). Under MIL STD 105E. Sample size code letter = L: n Ac Re Normal 200 7 8 Tightened 200 5 6 Reduced 80 3 6 The MIL STD 414 sample sizes are considerably smaller than those for MIL STD 105E. n and k will define Pa. For example.005. Then a k can be found to satisfy the AOQL equation. Repeat until the n and k that minimize ATI are found. Suppose n is arbitrarily specified.wiley.3 15-3. Calculate x and S. β = 0.95. and special Romig tables are usually employed. Finally.93 A reduced sampling (nreduced = 20. The equations do not change: AOQ = Pa p (N – n) / N and ATI = n + (1 – Pa) (N – n). n = 120 and k = 2. The table required to do this is available on the Montgomery SQC website: www.80.Form 1. ATI is found from the above equation. p2 = 0.02. N = 7000. Inspection level IV From Table 15-1 (A-2): Sample size code letter = M From Table 15-2 (B-1): n = 50. The design of a variables plan in rectifying inspection is somewhat different from the attribute plan design.

inspection level II.Chapter 15 Exercise Solution 15-6.1 − α = 1 − 0.10 x A − x2 = Φ(β ) σ n x A − 0. σ = 0.1527 .15.15 Tightened sampling: n = 7.05 x A − x1 = Φ (1 − α ) σ n x A − 0.145.00.005 n x2 = 0.00 Tightened sampling: n = 100. k = 2.000 ⎤⎦ > 2. 15-8. β = 0. so accept the lot. k = 1. AQL = 1%. k = 2. N = 100. k = 1.000 Assume inspection level IV. x = 255.15 = +1.145 = −1. AQL = 4% Sample size code letter = E Assume single specification limit Normal sampling: n = 7. LSL = 225psi. S = 10 ⎡⎣ Z LSL = ( x − LSL ) S = ( 255 − 225 ) 10 = 3. sample size code letter = O Normal sampling: n = 100.95 = 0.645 0.33 15-7. N = 500.005 g/cm3 x1 = 0.282 0.14 Assume normal sampling is in effect.005 n n ≈ 9 and the target x A = 0.

10 From the nomograph.00 (k = 1.8. (b) Note: The tables from MIL-STD-414 required to complete this part of the exercise are available on the Montgomery SQC website: www. M = 4. (c) 1 – α = 0. Ac = 5.08 From nomograph (for attributes): n = 60.90 = 0.93) Tightened: n = 50.01.10.08 (a) 1 – α = 0. Re = 5 The sample sizes required are much larger than for the other plans. Ac = 2. .000 Assume inspection level II: sample size code letter = L Normal: n = 200. M = 1. p2 = 8% = 0. Re = 4 Reduced: n = 80. p2 = 0. (d) AQL = 1%. β = 1 – 0. σ unknown Double specification limit. c = 2 The sample size is slightly larger than required for the variables plan (a).09 (k = 1. assume inspection level IV From Table A-2: sample size code letter = M From Table A-3: Normal: n = 50.95. M = 1.08) Reduced: n = 20. Variables sampling would be more efficient if σ were known.Chapter 15 Exercise Solution 15-9.01.69) σ known allows smaller sample sizes than σ unknown. p1 = 0. β = 0.com/college/montgomery AQL = 1%. N = 5. Ac = 3.71 (k = 2.wiley.10ppm. Re = 6 Tightened: n = 200.95. σ = 0. p1 = 1% = 0. N = 5000. the sampling plan is n = 30 and k = 1. target = 3ppm.

20 1.Chapter 15 Exercise Solution 15-10.06 0. chain sampling plans with c = 0 have slightly less steep OC curves.16 0.18 0.02 0.20 0.xls : worksheet Ex15-10 OC Curves for Various Plans with n=25. Excel workbook Chap15.04 0.80 0.60 0.14 0. c=0 1.08 0.2 p single I=1 I=2 I=5 I=7 Compared to single sampling with c = 0.00 0 0.12 0. .40 0.00 Pa 0.1 0.

0080 0.12 Fraction Defective.06 0.2000 0.6227 0.1482 0. average process fallout = 0.6 0.7987 0.2267 0.0 Probability of Acceptance.14 0.08 0.9685 0.9379 0.0050 0.0300 0. c=0 1. c = 0 Excel workbook Chap15.000.10% = 0.0917 0.7250 0.0020 0.04 0.0400 0.9083 0.9019 0.0200 0.0070 0.0343 0.4761 0.2512 0.1937 0.Chapter 15 Exercise Solution 15-11. N = 30. n = 32.8063 0.0000 1. Pa 0.2708 0.0489 0.2750 0.18 0.16 0.9306 0.5239 0.3000 Pa Pr{reject} 0.8 0.1381 0.xls : worksheet Ex15-11 (a) p 0.0030 0.8619 0.0800 0.0600 0.9992 0.20 .0040 0.0621 0.7488 0.8518 0.0 0.4 0.00 0.7292 0.9657 0.0700 0.02 0.10 0.2 0.8796 0.0315 0.0900 0.1752 0.0000 OC Chart for n=32.0090 0.8248 0.0008 0.9511 0.3773 0.0694 0.0060 0.7733 0.001.1204 0.0981 0. p 0.0010 0.1000 0.0100 0.2013 0.0500 0.

there is little change in performance by increasing i. (d) Pa = 0. n) + P(1.Chapter 15 Exercise Solution 15-11 continued (b) ATI = n + (1 − Pa )( N − n) = 32 + (1 − 0.9958.001 Pa = P(0.9685)(30000 − 32) = 976 (c) Chain-sampling: n = 32. n)]i P(0.9685 P(1. n) = P(1.0310 Pa = 0. but the average number inspected is much smaller.9967)(30000 − 32) = 131 Compared to conventional sampling.9685)3 = 0. n)[ P(0.32) = 0.9685 + (0. c = 0. p = 0. the Pa for chain sampling is slightly larger.9967 ATI = 32 + (1 − 0.9958)(30000 − 32) = 158 .32) = 0.0310)(0. i = 3. ATI = 32 + (1 − 0. n) = P (0.

60 0.4116 0.9606 0.7164 0.0256 0.0900 0.00 .0200 0.00 0.0000 OC Curve for ChSP-1 n=4.c=0 1.8423 0.7732 0. c = 0.4000 0.9000 0.2713 0.6000 0.0016 0.30 0.0625 0.0800 0.8756 0.0756 0.9999 0.4) 0.0300 0.0010 0.9500 P(0.1000 0.4377 0.80 0.1715 0.2401 0.8000 0.2458 0.40 0.0081 0.0000 P(1.0256 0.80 0.9613 0.2252 0.5000 0.Chapter 15 Exercise Solution 15-12.6857 0.xls : worksheet Ex15-1 p 0.0388 0.0005 Pa 0.1296 0.00 0.0500 0.0256 0.0016 0.2000 0.30 0.20 0.0040 0.2500 0.4096 0.1993 0.70 0.9072 0.70 Pa 0.9960 0.0001 0.8145 0.3000 0.4) 0.2492 0.7000 0.1536 0.50 p 0.0081 0.9815 0.7481 0.00 0.9950 0.8080 0.2916 0.0700 0.10 0.0753 0.8853 0.0001 0.10 0.40 0.0100 0.7807 0.1095 0.50 0.3456 0.0036 0.60 0.9224 0.6561 0.90 0. i = 3 Excel workbook Chap15. n = 4.4096 0.1304 0.7385 0.0600 0.20 0.0626 0.90 1.

v = 1333.5035E+02 2000.3333 0.0100 0.7586E+48 1250.3395E+06 14285.0251E+06 22.8816E+02 500.0000 0.8648E+02 133.6667 0.88584 + 0.8863 1.6788E+02 800.7266 1.8608 1.0250 0.0000 1.7161E+04 28571.0800 0. AFI = 0.0200 0.5714 0.0401E+03 66666.2222 0.4383E+07 11111.0000 6.88584) 4 = 0.4066E+03 3333.1429 0.0255E+24 2500.6712 2.0000 0.000.5729E+18 3333.2222 0.7032 1.0000 0.8192E+07 10000.6667 0.3984E+04 44.4085E+11 250.1088 1.0000 9.4286 0.530. 6)[ P(0.2867E+08 333.0040 0.6795E+03 2857.0000 0.5912E+04 25000.Chapter 15 Exercise Solution 15-13.4676E+06 25.0000 0.0277 4.0025 0.0000 2.0318E+21 125. Three different CSP-1 plans with AOQL = 0.6667 0.7143 0.4400 3.5930E+04 1111.2451E+09 285. 6)]4 = 0.0000 3.3333 0.0000 3.0000 0.2128E+12 222.3333 0. p = 0.9316 8. v = 66.2222 0.5127 4.7143 0.6298 2.3909E+02 285.3934 4. Pa = 0.4040 4.0666.5602E+03 50.1279 4.5444 2.0238E+04 1250.5714 0.3333 0.0003 3.4444 0.0350 0.0010E+04 40000.0070 0.0000 0.6195E+36 1666.0000 0.3.8328 1.0000 1. 6) + P(1.8953 2.0000 3.2131E+05 16666.3333 0.8885 1.9985 1.0179E+30 2000.0001 4.0000 0.3373E+02 10000.0000 0.0000 3.0135 7.0015 9.5471E+07 20. Pa = 0.5536 2.6667 0.1848E+03 4000.3262E+13 5000. f = 1/10 and i = 550: u = 855.0600 0.0030 0.5553E+02 6666. f = 1/10 and i = 550 3.5523. f = ½ and i = 140: u = 155.0000 2.4329E+03 2222.0852E+18 142.10847(0.8863 3.0000 8.0002 2.0000 0.02} Pa = P(0.4619E+07 400.6331E+04 33333.3333 0.5056E+04 1000.0000 0.0384 2. AFI = 0. Pa = 0.0037E+03 5000.0010 0.0000 0.9429 p 0.0652 2.0000 1.0000 2.0000 0.0000 0.1000 f = 1/2 and i = 140 u v Pa 1.0000 0.0000 0.0106 4.9739 4.5984E+02 250.0000 0.0700 0.0000 1. v = 6666.7.9502E+03 2000.2222 0.0000 0.2222 0.7715 1.9562E+02 444.0000 0. f = ½ and i = 140 2.1111 0.0060 0.6175E+02 1000.0346E+02 400.3972E+03 1666.0000 0.8266 1.0000 1.7431E+02 666.5592E+02 1333.3439E+10 6666.0000 0.7410E+23 111.0500 0.4676E+26 100.1111 0.9301 1.0020 0.0000 0.2151 7. accept.0000 0.0008 4. f = 1/100 and i = 1302: u = 4040.0000 0.0000 1.5492E+42 1428.0381 1.0029 1.5714 0.0839E+02 200.3638E+05 20000.0000 0.2765E+03 50000.0300 0.6921E+05 28.0000 0.3467E+06 500.0000 0.0000 3.0032 2.0561 2.1111 0.8571 0.5731E+13 200.4286 0.8675E+04 22222.0000 0.0001 1.0066 1.7692E+60 1000.2202 1.0450 0. N = 500.8284E+02 222. If c = 1.0150 0.4774 2.0000 0.7157E+05 666.1111 0.7998 1.0400 0.0024 3.0035E+16 166.6667 0.6790E+03 100000.95264 15-14.6355E+04 33.2222 0. Need to find Pa{p = 0.3333 0.2804E+15 4000.9429 6.0000 1.3333 0.3526 7. n = 6 If c = 0.0000 2.0015 and q = 1 – p = 0.0000 0.6667 0.1766 1.0000 0.0045 0.1116 1.0000 5.6667 0.15% = 0.8106E+02 571.4121E+27 2222.0000 f = 1/100 and i = 1302 u v Pa 2.3371E+03 66.0000 0. Average process fallout.2037E+02 333.2024.3652E+54 1111.4904 2.1429 0.4444 0.0000 3.9936E+21 2857.6944 2.0000 6.198% would be: 1.0000 0.7143 0.1604E+03 57.0162E+03 2500.0000 0.0000 1.0000 .1429 0.9590E+02 100.6020 2.6667 0.2749 6.3449E+03 80.0050 0.915.0000 0.0090 0.7594 1.0080 0.666.6667 0. AFI = 0.7927 1.0000 0.0035 0. accept if i = 4.6619E+03 1428.3521E+06 12500.6266E+04 40.0000 2.8953 1.0015 0.0900 0.0000 0.6628 2.6667 0. f = 1/100 and i = 1302 15-15.7.0000 f = 1/10 and i = 550 u v Pa 7.0000 0.

CSP-1 with AOQL = 1. . Plan A: AFI = 0.90% Plan A: f = 1/5 and i = 38 Plan B: f = 1/25 and i = 86 15-17.0375} = 0.4925 Prefer Plan B over Plan A since it has a lower Pa at the unacceptable level of p.5165 and Pa{p = 0.6043 Plan B: AFI = 0.Chapter 15 Exercise Solution 15-16.5272 and Pa{p = 0.0375} = 0.