IMChap 010

Chapter 10 - Quality Control
CHAPTER 10
QUALITY CONTROL
Teaching Notes
As a result of increased global competition, a rapidly growing number of companies of all sizes are
paying much more attention to issues involving quality and productivity. Many statistical techniques are
available to assist organizations in improving the quality of their products and services. It is important for
companies to use these techniques in the context of an overall quality system (Total Quality
Management), which requires quality awareness and careful planning and commitment to quality at all
levels of the organization. Many companies are not only utilizing these statistical techniques themselves,
but also are requiring their suppliers to meet certain standards of quality based on various statistical
measures. This chapter covers the statistical applications of quality control. Control charts are given the
primary emphasis, but other quality control topics such as process capability and inspection are discussed
also.
When covering the material in this chapter, we need to stress that through the use of control charts, the
nonrandom (special) causes of variation must be controlled before random (common) causes of variation
and process capability can be analyzed.
Reading: In the Chips at Jays

1. Jays cares about the following quality issues:
a. Appearance: Chips must be intact (not crushed into flinders), chips must not be burned or
deformed, and chips must not have dark spots.
b. Weight: Bags should be filled to exactly 14 ounces.
c. Volume: Appropriately sized chips should be placed into large bags vs. lunch bags.
d. Freshness: Chips should be fresh (not sold after eight weeks).
e. Taste: Although this was not stated in the case, it is reasonable for us to assume that Jays
cares about the taste of the chips.
2. Jays is not overdoing it with its concern for quality because they are in the “fun food
business.” Consumers buy potato chips as a treat—not as an everyday staple. Therefore, Jays
must ensure the high quality of the chips (i.e., appearance, weight, volume, freshness, and taste).
Reading: Bar Codes Might Cut Drug Errors in Hospitals

1. Bar codes are being used in hospitals to prevent drug errors (incorrect timing of dose, incorrect
dosage, and incorrect drug). It is estimated that more than 7,000 hospital patients die each year
because of drug errors.
2. The workaround problem can be solved through three policies: (a) nurses are not permitted to
make copies of patients’ bar codes, (b) nurses are not allowed to obtain drugs for more than one
patient at a time, and (c) violation of the first two policies could lead to disciplinary action, up to
and including termination.
10-1
Copyright © 2015 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill
Education.
Answers to Discussion and Review Questions

1. The elements in the control process are:
a. Define: define what is to be controlled
b. Measure: determine how measurement will be accomplished and then measure
c. Compare to standard
d. Evaluate: establish a definition of out of control, i.e., distinguish between random and
nonrandom variability
e. Take corrective action if needed: uncover the cause of nonrandom variability and correct it
f. Monitor results: ensure that corrective action is working over a period of time
2. Control charts are based on the premise that a process that is stable will reflect randomness:
statistics of samples taken from the process (e.g., means, number of defects, etc.) will conform to
a sampling distribution (usually the normal distribution) with known characteristics so that
statistical significance tests can be performed on sample statistics, and successive samples will
not reveal any nonrandom patterns.
3. Control charts are used to distinguish between random variability and nonrandom variability of a
process, i.e., to monitor process output to determine if output is random.
4. Order of observation of process output is important if patterns (e.g., trends or cycles) in the output
are to be detected.
5. a. x Chart—A control chart used to monitor process variables by focusing on the central
tendency of a process.
b. Range control charts are used to monitor process variables by focusing on the dispersion of a
process.
c. p-chart—is a control chart for attributes, used to monitor the proportion of defectives in a
process.
d. c-chart—is a control chart for attributes, used to monitor the number of defects per unit.
6. A run is a sequence of observations with a certain characteristic, followed by one or more
observations with a different characteristic. Run tests are useful in quality control because they
enable an analyst to do a better job of detecting abnormalities in a process and they provide
insights into correcting a process that is out of control.
7. All points can be within control limits, but there may be nonrandom patterns in the data, which
would suggest an out of control process. Checking for points within control limits reveals only
extreme points that fall outside the limits and does not reveal nonrandom patterns.
8. It is usually desirable to use both an up/down and a median run test on a given set of data because
the tests are sensitive to the different types of patterns that they can detect. For example, one test
(the up/down test) can be more sensitive to trend and the other test (the median test) can be more
sensitive to bias.
9. No, there is always the possibility of a Type II (beta) error, which is concluding that the process is
in control when it is really out of control. The Type II error could occur when the analyst uses
control limits that are too wide or when the process output does not follow a normal distribution.
The analyst still should plot the data and visually check for nonrandom patterns.
10. Specifications are limits on the range of variation of output that are set by design (engineering and
customers). Control limits are statistical bounds on a sampling distribution. They indicate the
extent to which summary values such as sample means or sample ranges will tend to vary solely
on a chance basis. Process variability refers to the inherent variability of a processthe extent to
which the output of a process will tend to vary due to chance. Control limits are a function of
10-2
Education.
process variability as well as sample size and confidence level. Both control limits and process
variability essentially are independent of tolerances.
11. The problem is that even when the machine is functioning as well as it can, unacceptable output
will result. Among the possible options that should be considered are:
a. Use 100 percent inspection to screen out defectives. If destructive testing is required, this
may not be feasible.
b. Attempt to convince customers to widen tolerances (perhaps by offering a lower price) or
convince engineering to perform 100 percent inspection. The problem is that engineering may
resent this suggestiondepending on how it is handled. Moreover, it may be that the
customer believes that the tolerance is necessary for proper functioning of the final product or
service.
c. Attempt to substitute a different machine (e.g., a newer one) that has the capability to meet
the tighter specifications.
d. Train workers to perform more preventive maintenance on the existing machine and to clean
that machine regularly so that it functions better.
e. Hope for a miracle.
The assumption here is that the first four options may shift the mean and/or reduce the process
variability.
12. a. This “problem” often goes undetected because there are no complaints from customers about
output falling outside specifications. However, it is quite possible to realize decreased costs
or more profits due to having fewer defective parts. In addition, less inspection might be
appropriate in this situation.
b. A “marketing approach” to this problem might be to see if the customer is willing to pay
more for output that meets tighter tolerances. If not, perhaps the job could be shifted to
another, less capable machine, freeing up this equipment for more demanding work. A third
option would be to cut back on inspection given that virtually 100 percent of the output will
be acceptable, and even a slight out of control situation will not warrant corrective action.
13. a. An optimal level of inspection is one where the cost and effort of inspection equals the
benefits derived from inspection, or the point (number of units inspected) at which the
marginal cost of inspection equals the marginal benefit from inspection (the point at which
the sum of inspection cost and the cost of passing defectives is minimized).
b. The factors that guide the decision of how much to inspect are the cost of the product or
service, the volume of output, the cost of inspection, the cost of letting undetected defects slip
through, the stability of the process, and the number and the size of inspection lots.
c. The main issues in the decision of whether to inspect on site or in a central location are the
situation (size and mobility of the units to be inspected), the inspection time required, the
costs of process interruption, the need for a quick decision, the importance of avoiding
extraneous factors affecting samples or tests, the need for specialized equipment, and the
need for a more favorable testing environment.
d. Points that are potential candidates for inspection are raw materials and purchased parts,
finished products, before a costly operation, before an irreversible process, and before a
covering process.
14. Two basic assumptions that must be satisfied to use a process capability index are:
a. The process is stable (non-random causes of variation have been identified and eliminated).
b. The process distribution is normal.
10-3
Education.
15. It is very important for managers to maintain and promote ethical behavior in dealing with quality
issues. The company’s (and manager’s) reputation is at stake, and there may be cost, liability,
legal, and safety issues. Although the risks may differ substantially for different products and
services, ethical standards should be maintained “across the board.”
16. a. Type I error: concluding that the person is not innocent (i.e., guilty) when the person is
innocent
b. Type II error: failing to detect that the person is not innocent (i.e., guilty) when the person is
not innocent
c. Type II error: failing to detect that the cookie is not clean when it is not clean
Type I error: concluding that the cookie is not clean when it is clean
d. Type II error: failing to recognize the potentially fatal aspects of the poison by assuming that
all is well when all is not well
Taking Stock
1. a. In deciding whether to use 2-sigma or 3-sigma limits, the quality control people should be
involved as well as the accounting/recordkeeping personnel because it will be critical to
determine the cost of stopping the process unnecessarily vs. the cost of not correcting a
special cause of variation. In addition, we may want to involve the customers’ quality control
personnel because they will be affected by this decision ultimately. In addition, 2-sigma
limits are narrower in range than 3-sigma limits are and, therefore, are more responsive (i.e.,
lower Type II error), but less stable (i.e., higher Type I error).
b. The quality control department should make this decision. However, input from the
production control department may be very useful to make this decision. For larger samples,
the cost to inspect those items will be greater (and the greater the lost product will be if
destructive testing is used), and the longer the process must be held up while waiting for the
results of the sampling. However, large samples lead to fewer interruptions to a process and
less time plotting and interpreting sample data. In addition, the likelihood that a sample
statistic is close to the true population value is higher for larger samples than it is for smaller
samples. The tradeoffs when taking larger samples are the cost (inspection, interruption, and
lost product) vs. the benefits (fewer interruptions to the process and greater accuracy of
sample statistics).
Smaller samples are more likely to reveal a change in the process than larger samples are
because a change is more likely to take place within the large sample, but between small
samples. The downside of taking small samples is the increased frequency to take samples, to
plot data, to interpret data, and to take corrective action if necessary. The tradeoff here
involves cost (inspecting items, plotting values, interpreting charts, and taking corrective
action) vs. the benefits (the usefulness of the data for making decisions on whether the
process is in control).
c. Increasing the capability of the process is a significant and potentially costly adventure.
Therefore, upper management, in consultation with the quality and production departments,
should determine what type of improvements to make to improve the capability of the
process. The tradeoff involves the cost of increasing capability vs. the increased profits from
reducing failure costs and the potential to increase revenues.
2. In setting the quality standards, customers definitely should be involved because they will be
using the product ultimately. In consultation with upper management, the quality and production
departments should work as a team in establishing quality standards.
3. Technology has had a profound impact on quality. Improvement in measurement systems has
improved the measurement of quality significantly. Computer technology has enabled many
10-4
Education.
companies to perform on-line, real-time statistical process control, which has enabled companies
to respond to quality problems faster. Due to technological improvements in computerized
design, products are designed better, and thus have significantly fewer quality problems. In
addition, artificial intelligence systems forewarn potential problems before they occur. Processes
using computerized technology are able to meet tighter specifications also.
Critical Thinking Exercises

1. If the analysis of the output of a process suggests that there is an unusual occurrence, but the
result of the investigation cannot pinpoint or determine the assignable causes, the limits may be
set too tight. In addition, if measurement or recordkeeping is poor, it may be very difficult to
identify the assignable causes of variation. The manager may want to widen control limits or to
improve measurement and recordkeeping.
2. A single standard would be easier to work with, and everyone would know what the standard is.
Multiple standards might be used if the cost of errors differed significantly across products or
services. For instance, not meeting the specifications for a product such as paper clips would not
have the same degree of importance as not meeting the specifications for heart-monitoring
equipment. Also, differing standards might reflect progress in continuous improvement. As
processes are improved, the standard should be updated to reflect that, and perhaps be used as a
competitive strength in the marketplace.
3. Student answers will vary. Some possible answers follow:
Inspection: Cutting back on inspection to decrease costs would violate the Utilitarian Principle
due to the risk of increased failure costs. If a manager forced an inspector to alter inspection data,
this action would violate the Rights Principle.
Process Control: If a company did not use process control and defective units caused significant
harm to customers or to the community, this action would violate the Common Good Principle. If
a manager cut back on process control to save money, this action violate the Utilitarian Principle
due to the risk of increased failure costs.
Process Capability: If a manager knowingly lied about process capability, this action would
violate the Virtue Principle. If a manager decided not to evaluate process capability when
customers consider process capability to be important, this action would violate the Utilitarian
Principle due to the resulting loss of business.
4. Even though a repetitive operation can check and reject parts automatically, the rejecting of parts
reveals little information other than the number of parts rejected. Control charts still are needed to
detect nonrandom patterns in the process data.
Additional Experiential Learning Exercises
Sampling Demonstrations. Bowls of colored beads (e.g., 1,000 beads, 40% white, 30% green,
20% red, 5% black, 3% yellow, etc.) are available with paddles that have indentations to
facilitate obtaining samples of various sizes quickly. Use this exercise to demonstrate sampling
variability: take repeated small samples (or have a student do it) to demonstrate that different
percentages of a color appear in different samples. Then, increase the sample size, focusing on a
specific color, and have students recognize that there is less sampling variability as the sample
size increases. Afterward, explain why in process sampling, small samples are used even though
large samples are more “accurate.” One issue is the cost, time, and disruption caused by taking
large samples. Even more important is the ability to capture process changes that might occur
between small samples (e.g., taking 6 samples of n = 10 over time could reveal a trend, whereas
taking 1 sample of n = 60 would not).
Process Control Demonstrations. Obtain 30 clear plastic “zipper” bags and an ample supply
10-5
Education.
of colored beads or jelly beans.
a. Place 10-20 beads/beans in each of 10 bags, focusing, for example, on red beads/beans.
Number the bags 1 to 10. Put 0, 1, or 2 red beads/beans in each bag, along with other
colors of beads/beans. Pass the bags out to the class and ask them what kind of control
chart would be appropriate (p-chart or c-chart). Then, assume that the red beads/beans
are defects so we will construct a c-chart. Ask students to report (in the order in which
they took the samples) the number of defects that they have counted. Next, have the
students calculate 2s control limits and then observe that all sample data are within the
control limits.
b. In a second set of bags numbered 11-21, arrange it so that bag #14 has too many red
beads/beans, making it “out of control.” Pass out those bags and have the students report
the numbers of red that they find. When the student with bag #14 reports a large number
of reds, tell the class that the process would be halted while efforts were made to find
and correct the problem. Then, resume “sampling” with the process now back in
control.
c. Arrange the third set so that a trend begins to appear at about the fifth bag, but with all
results still within the limits. Point out that the process does not appear to be random;
therefore, it would be halted at the ninth or tenth sample to find and correct the problem.
Solutions
1. Given:
.0062 .0062
Specs: 24 oz. to 25 oz.
=
 = 24.5 oz. [assume  = x ] -2.5 0 +2.5 z-scale
 = .2 oz. 24 24.5 25 16
a. [refers to population]
𝑥 − 𝜇 24.5 − 24
𝑧= = = 2.50
𝜎 .2
Using Appendix B, Table A, we have:
Prob(z > +2.50) = .5000 - .4938 = .0062
Prob(z < -2.50) = .0062
.0062 + .0062 = .0124
b. [refers to sample]
95.44% = 47.72% on each side of the mean.
Using Appendix B, Table A, find the z value corresponding to .4772: z = 2.00.
If n = 16 per sample, then we have the following control limits. Round intermediate values to
a maximum of three decimals. Round Mean control limits to three decimals:
  .2 
  2.00  24.5  2.00    24.5  0.100 or 24.400 to 24.600
n  16 
10-6
Education.
2. Given:
 = 1.0 liter
 = .01 liter
n = 25

a. Control limits :   z
n
Round Mean control limits to three decimals.
97% = 48.5% (.4850) on each side of the mean (round probability to four decimals).
Using Appendix B, Table A, find the value of z that corresponds to .4850:

z = 2.17.
 .01 
Control limits: 1.0  2.17   1.0  .004
 25 
UCL = 1.0 + .004 = 1.004

LCL = 1.0 – .004 = 0.996
b.
1.007
Mean Chart
1.005 1.004
1.003
1.001 UCL
1.000
LCL
0.998 Mean
0.996
0.996
0.994
1 2 3 4 5 6
Sample mean 1 is above the UCL and Sample mean 5 is below the LCL. The process is not in
control.
10-7
Education.
3. Given:
Six samples of n = 20 observations were obtained. The sample means and ranges are shown
below:
Sample Mean Range

1 3.06 .42
2 3.15 .50
3 3.11 .41
4 3.13 .46
5 3.06 .46
6 3.09 .45
a. Using the factors in Table 10.3, determine upper and lower limits for Mean and Range charts.
Round all intermediate values for Mean and Range charts to a maximum of three decimals.
Round Mean and Range chart control limits to three decimals.
Sample Mean Range

1 3.06 .42
2 3.15 .50
3 3.11 .41
4 3.13 .46
5 3.06 .46
6 3.09 .45
Mean 3.100 0.450
From Table 10.3 with n = 20: A2 = 0.18; D3 = 0.41; D4 = 1.59.

𝑥̿̅ = 3.100 𝑅̅ = 0.450
Mean Chart:
Upper Control Limit (UCL) = 𝑥̿ + 𝐴2 𝑅̅ = 3.100 + 0.18(0.450) = 3.10 + 0.081 = 3.181
Lower Control Limit (LCL) = 𝑥̿ − 𝐴2 𝑅̅ = 3.100 − 0.18(0.450) = 3.10 − 0.081 = 3.019
Range Chart:
Upper Control Limit (UCL) = 𝐷4 𝑅̅ = 1.59(0.450) = 0.716
Lower Control Limit (LCL) = 𝐷3 𝑅̅ = 0.41(0.450) = 0.185
10-8
Education.
b.
3.200 Mean Chart

3.180 3.181
3.160
3.140
UCL
3.120
LCL
3.100 3.100
Mean
3.080
3.060
3.040
3.020 3.019
3.000
1 2 3 4 5 6
0.750
Range Chart 0.716
0.650
0.550
UCL
0.450 0.45
0.450
LCL
0.350 Range
0.250
0.185
0.150
1 2 3 4 5 6
All sample means are within the control limits on the Mean chart.
All sample ranges are within the control limits on the Range chart.
The process is in control.
10-9
Education.
4. Given:
Samples of five observations have been taken and are shown below:
Sample
1 2 3 4 5 6
79.2 80.5 79.6 78.9 80.5 79.7
78.8 78.7 79.6 79.4 79.6 80.6
80.0 81.0 80.4 79.7 80.4 80.5
78.4 80.4 80.3 79.4 80.8 80.0
81.0 80.1 80.8 80.6 78.8 81.1
Determine upper and lower control limits. Round all intermediate values to a maximum of three
decimals. Round Mean and Range chart control limits to three decimals.
Samples
1 2 3 4 5 6
79.2 80.5 79.6 78.9 80.5 79.7
78.8 78.7 79.6 79.4 79.6 80.6
80.0 81.0 80.4 79.7 80.4 80.5
78.4 80.4 80.3 79.4 80.8 80.0
81.0 80.1 80.8 80.6 78.8 81.1
Mean 79.48 80.14 80.14 79.60 80.02 80.38
Range 2.6 2.3 1.2 1.7 2.0 1.4
From Table 10.3 with n = 5: A2 = 0.58; D3 = 0; D4 = 2.11.

𝑥̿̅ = 79.960 𝑅̅ = 1.867
Mean Chart:
Range Chart:
Lower Control Limit (LCL) = 𝐷3 𝑅̅ = 0(1.867) = 0.000
10-10
Education.
81.500
Mean Chart
81.000 81.043
80.500
80.000 79.960 UCL

LCL
79.500 Mean
79.000
78.877
78.500
1 2 3 4 5 6
4.500
Range Chart
4.000 3.939
3.500
3.000
2.500
UCL
2.000
1.867 LCL
1.500 Range
1.000
0.500
0.000 0.000
1 2 3 4 5 6
All sample means are within the control limits on the Mean chart.
All sample ranges are within the control limits on the Range chart.
The process is in control.
10-11
Education.
5. Given:
Four samples of 200 credit card statements each were taken and the number with errors recorded
as shown below:
Sample 1 2 3 4
# with errors 4 2 5 9
a. Determine the fraction (proportion) defective for each sample. Round all values to a
maximum of four decimals.
n = 200 per sample
Sample 1 2 3 4
# with errors 4 2 5 9
Prop. defective
(# with errors/200) .020 .010 .025 .045
b. If the true fraction (proportion) defective is unknown, estimate the fraction defective. Round
all values to a maximum of four decimals.
(.020 + .010 + .025 + .045)/4 = .025

Or:
Total number of defects = 20
total number of defectives 20
p   .025
total number of observatio ns 4(200)
c. Estimate the mean and the standard distribution of the sampling distribution. Round all
values to a maximum of four decimals.
Mean = .025
p(1  p) .025(1  .025)
Std. dev.    .011
n 200
d. Determine control limits for an alpha risk of .03. Round all p-Chart control limits to four
decimals.
An alpha risk of .03 infers that there is .03/2 = .015 in each tail.
Using Appendix B, Table A, find the z value that corresponds to .5000 – .0150 = .4850:
z = 2.17.
UCL: .025 + 2.17(.011) = .025 + .0239 = .0489
LCL: .025 – 2.17(.011) = .025 – .0239 = .0011
10-12
Education.
e. Determine the alpha risk provided by control limits of .047 and .003.
We know the following: .025 + z(.011) = .047

Solve for z:
z(.011) = .047 – .025
z(.011) = .022
z = .022/.011
z = 2.00 (round to two decimals)
Using Appendix B, Table A, find the probability that corresponds to z = 2.00: .4772
.5000 – .4772 = .0228 in each tail
2 * .0228 = .0456 (alpha risk).
f. Using the control limits from part e, determine if the process is in control.
.0510
p-Chart .0470
.0410
.0310
.0250 UCL
.0210 LCL
Prop.
.0110
.0030
.0010
1 2 3 4
All points fall within the limits. The process is in control.
g. Assume that the long-run fraction (proportion) defective = 2% = .02. Determine the mean and
standard deviation of the sampling distribution. Round all intermediate values to a maximum
of four decimals. Round all p-Chart control limits to four decimals.
Mean = .02
p(1  p) .02(1  .02)

Std. dev.    .0099
n 200
10-13
Education.
h. Construct a control chart for the process, assuming a fraction (proportion) defective of 2%
(.02) using two-sigma control limits. Round all intermediate values to a maximum of four
decimals. Round all p-Chart control limits to four decimals.
p(1  p) .02(1  .02)

p2  .02  2  .02  .0198
n 200
UCL = .0398
LCL = .0002
.0500
p-Chart
.0400 .0398
.0300
UCL
.0200 .0200 LCL
Prop.
.0100
.0000 .0002
1 2 3 4
Sample 4 proportion defective is above the upper control limit. The process is not in control.
6. Given:
Samples of n = 200 were taken. The fraction (proportion) defective for 13 samples are given
below. Construct two-sigma control limits. Round all intermediate values to a maximum of four
decimals. Round all p-Chart control limits to four decimals.
Sample 1 2 3 4 5 6 7 8 9 10 11 12 13
Retests 1 2 2 0 2 1 2 0 2 7 3 2 1
Sample 1 2 3 4 5 6 7 8 9 10 11 12 13
Retests 1 2 2 0 2 1 2 0 2 7 3 2 1
Prop. .005 .010 .010 .000 .010 .005 .010 .000 .010 .035 .015 .010 .005

p   .0096
10-14
Education.
p(1  p) .0096(1  .0096)

Control Limits = p  2.00  .0096  2.00  .0096  .0138
n 200
UCL = .0234
LCL = -.0042 = 0 (cannot be negative)
.0400
p-Chart
.0350
.0300
.0250
.0234
.0200 UCL
LCL
.0150
Prop.
.0100 .0096
.0050
.0000 .0000
1 2 3 4 5 6 7 8 9 10 11 12 13
Sample 10 proportion defective is above the UCL. The process is not in control.
7. Given:
The postmaster of a small town has recorded the number of complaints per day. Determine three-
sigma control limits using the data below. Round all intermediate values to a maximum of three
decimals. Round all c-Chart control limits to three decimals.
Day 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Comp. 4 10 14 8 9 6 5 12 13 7 6 4 2 10
110
c  7.857
14
Control limits: c  3 c  7.857  3 7.857  7.857  8.409
UCL = 16.266
LCL = -0.552 = 0 (cannot be negative)
10-15
Education.
20.000
c-Chart
16.266
15.000
10.000 UCL
LCL
7.857
Comp.
5.000
0.000 0.000
1 2 3 4 5 6 7 8 9 10 11 12 13
All points are within the limits. The process is in control.
8. Given:
Given the following data on number of defects per spool of cable, determine three-sigma control
limits. Round all intermediate values to a maximum of three decimals. Round all c-Chart control
limits to three decimals.
Cable 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Defects 2 3 1 0 1 3 2 0 2 1 3 1 2 0
21
c  1.5
14
Control limits: c  3 c  1.5  3 1.5  1.5  3.674
UCL = 5.174
10-16
Education.
6.000 c-Chart
5.000
4.000
3.000 UCL
LCL
Defects
2.000
1.000
0.000
1 2 3 4 5 6 7 8 9 10 11 12 13 14
9. Given:
A telephone company took 16 samples of 100 calls as shown below. Determine 95% control
limits. Round all intermediate values to a maximum of four decimals. Round all p-Chart control
limits to four decimals.
Sample 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Errors 5 3 5 7 4 6 8 4 5 9 3 4 5 6 6 7

95% control limits = .9500/2 = .4750 in each tail. Using Appendix B, Table A, we find that z =
1.96.

p   .0544
p(1  p) .0544(1  .0544)
Control limits are p  z  .0544  1.96  .0544  .0445
n 100
UCL = .0989
LCL = .0099
Sample 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Prop. .05 .03 .05 .07 .04 .06 .08 .04 .05 .09 .03 .04 .05 .06 .06 .07
10-17
Education.
.1500
p-Chart
.1000 .0989
UCL
LCL
.0500 .0544 Prop.
.0099
.0000
1 2 3 4 5 6 7 8 9 10 11 12 13
10. Given:
A tool wears at the rate of .004 centimeters per piece, i.e., the metal shaft will increase by .004
centimeters per piece. The process sigma = .02 centimeters. Specifications are set at 15.0 to 15.2
centimeters. A three-sigma cushion is set. Determine the number of shafts that can be processed
before tool replacement becomes necessary. Round all values to a maximum of three decimals.
The key is that the mean starts low and then drifts up by .004 centimeters per piece.
1) Observe that the process mean starts at three-sigma above the lower specification limit:
15.0 + 3(.02) = 15.06
2) Determine what the ending process mean will be when it is three-sigma below the upper
specification limit:
15.2 – 3(.02) = 15.14
3) Determine how many pieces can be produced before the mean shifts from its starting point up
to its ending point:
15.14 − 15.06
= 20 𝑝𝑖𝑒𝑐𝑒𝑠
. 004
10-18
Education.
11. Given:
Specifications are 78 to 81 minutes for the computer upgrades from Problem 4. Estimate the % of
process output that can be expected to fall within 78 to 81 minutes. The data from Problem 4 are
repeated below.
Sample
1 2 3 4 5 6
79.2 80.5 79.6 78.9 80.5 79.7
78.8 78.7 79.6 79.4 79.6 80.6
80.0 81.0 80.4 79.7 80.4 80.5
78.4 80.4 80.3 79.4 80.8 80.0
81.0 80.1 80.8 80.6 78.8 81.1
Looking at the data above, we see that there are 30 total points, and 1 point (3.33% of the total) is
outside the specifications limits.
The proportion within the specification limits is 29/30 = .9667 or 96.67% (round to two
decimals). This corresponds to .9667 falling within the limits. .9667/2 = .4834. Using Appendix
B, Table A, we see that .4834 corresponds to z = 2.13.
Because z = 2.13 < z = 3.00, we assume that specifications are being met.
12. Given:
We have the following sample means for a process below (n = 14 units per sample). The process
standard deviation is .146.
Sample Mean Sample Mean Sample Mean Sample Mean

1 3.86 11 3.88 21 3.84 31 3.88
2 3.90 12 3.86 22 3.82 32 3.76
3 3.83 13 3.88 23 3.89 33 3.83
4 3.81 14 3.81 24 3.86 34 3.77
5 3.84 15 3.83 25 3.88 35 3.86
6 3.83 16 3.86 26 3.90 36 3.80
7 3.87 17 3.82 27 3.81 37 3.84
8 3.88 18 3.86 28 3.86 38 3.79
9 3.84 19 3.84 29 3.98 39 3.85
10 3.80 20 3.87 30 3.96
a. Round intermediate values to a maximum of three decimals. Round Mean control limits to
three decimals.
x
 x  150 .15  3.85
39 39
 0.146
Control limits are x  3  .385  3  3.85  .117
n 14
10-19
Education.
UCL = 3.85 + .117 = 3.967

LCL = 3.85 – .117 = 3.733
4.000 Mean Chart

3.967
3.950
3.900
3.850 3.850 UCL

LCL
Mean
3.800
3.750
3.733
3.700
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39
Sample 29 mean is outside the limits. The process is not in control.
b. Analyze the data using a median run test and an up/down run test.
Median is 3.85. The original data are transformed below to A/B and U/D.
Sample A/B Mean U/D Sample A/B Mean U/D
1 A 3.86  21 B 3.84 D
2 A 3.90 U 22 B 3.82 D
3 B 3.83 D 23 A 3.89 U
4 B 3.81 D 24 A 3.86 D
5 B 3.84 U 25 A 3.88 U
6 B 3.83 D 26 A 3.90 U
7 A 3.87 U 27 B 3.81 D
8 A 3.88 U 28 A 3.86 U
9 B 3.84 D 29 A 3.98 U
10 B 3.80 D 30 A 3.96 D
11 A 3.88 U 31 A 3.88 D
12 A 3.86 D 32 B 3.76 D
13 A 3.88 U 33 B 3.83 U
14 B 3.81 D 34 B 3.77 D
15 B 3.83 U 35 A 3.86 U
16 A 3.86 U 36 B 3.80 D
17 B 3.82 D 37 B 3.84 U
18 A 3.86 U 38 B 3.79 D
19 B 3.84 D 39 [B] 3.85 U
20 A 3.87 U
10-20
Education.
Median Test:
Sample 39 was tied with the median. As shown below, the expected number of runs = 20.5.
The expected number of runs will be greater than the observed number or runs (either 18 or
19 depending on how we label Sample 39). If we label Sample 39 as [B], this will increase
the difference between the observed number of runs and the expected number of runs.
Therefore, Sample 39 was labeled as [B] to maximize the ztest statistic.
Observed number of runs = 18.
Expected number of runs (N = total number of observations = 39):

𝑁 39
𝐸(𝑟)𝑚𝑒𝑑 = 2
+1= 2
+ 1 = 20.5 (round to 1 decimal)
Standard deviation:
𝑁−1 39−1
𝜎𝑚𝑒𝑑 = √ 4
=√ 4
= 3.08 (round to two decimals)
𝑂𝑏𝑠𝑒𝑟𝑣𝑒𝑑−𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 18−20.5
𝑧𝑡𝑒𝑠𝑡 = 𝑆𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝐷𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛
= 3.08
= -0.81 (round to two decimals)
This test statistic is within ± 2. Conclusion: Random.
Up/Down Test:
2𝑁−1 2(39)−1
𝐸(𝑟)𝑢/𝑑 = = = 25.7 (round to one decimal)
3 3
Standard deviation:
16𝑁−29 16(39)−29
𝜎𝑢/𝑑 = √ 90
=√ 90
= 2.57
= +1.28 (round to two decimals)
This test statistic is within ± 2. Conclusion: Random.
Overall Conclusion: The process is random. Neither test detected nonrandomness.
10-21
Education.
13. a. First control chart is marked with A/B and U/D:
A A B B A B A B B A B A A B A A A B B B A B A B A B
 

  
   
  
   
    
  

D D D U D U D D U D U U D U U D D D U U D U D U D
Median Test:

𝑁 26
+1= 2
+ 1 = 14.0 (round to one decimal)
Standard deviation:
𝑁−1 26−1
=√ 4
= 2.50
This test statistic is within ± 1.96. Conclusion: Random.
Up/Down Test:
Expected number of runs (N = number of observations = 26):
2𝑁−1 2(26)−1
𝐸(𝑟)𝑢/𝑑 = 3
= 3
= 17.0 (round to one decimal)
Standard deviation:
16𝑁−29 16(26)−29
𝜎𝑢/𝑑 = √ 90
=√ 90
𝑧𝑡𝑒𝑠𝑡 = = = 0.00 (round to two decimals)
𝑆𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝐷𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛 2.07
10-22
Education.
b. Second control chart is marked with A/B and U/D:
A A A A B A A B B B B B A B B B A A A A B B B B B B
  
  
    
     
 
   
  
U D U D U D D U D U D U D U D U D U D D U D U D D
Median Test:

𝑁 26
𝐸(𝑟)𝑚𝑒𝑑 = +1= + 1 = 14.0 (round to one decimal)
2 2
Standard deviation:
𝑁−1 26−1
𝜎𝑚𝑒𝑑 = √ =√ = 2.50 (round to two decimals)
4 4
𝑧𝑡𝑒𝑠𝑡 = = = −2.40 (round to two decimals)
This test statistic is not within ± 1.96. Conclusion: Nonrandom.
Up/Down Test:
Expected number of runs (N = number of observations = 26):
2𝑁−1 2(26)−1
𝐸(𝑟)𝑢/𝑑 = 3
= 3
Standard deviation:
16𝑁−29 16(26)−29
𝜎𝑢/𝑑 = √ =√ = 2.07 (round to two decimals)
90 90
= 2.07

Overall Conclusion: Nonrandom variations are present in the data.
10-23
Education.
14. a. Given:
Test z-score
Median +1.37
Up/Down +1.05
Because neither z-score exceeds +2.00, the process output is probably random.
b. N = 20. An analyst counted 14 runs above/below the median and 8 up/down runs.
Median Test:

𝑁 20
2 2
Standard deviation:
𝑁−1 20−1
=√ 4
= 2.18
Up/Down Test:
Expected number of runs (round to 1 decimal) (N = number of observations = 20):
2𝑁−1 2(20)−1
𝐸(𝑟)𝑢/𝑑 = 3
= 3
Standard deviation:
16𝑁−29 16(20)−29
𝜎𝑢/𝑑 = √ 90
=√ 90
= 1.80
= −2.78 (round to two decimals)
10-24
Education.
c. Given:
Data from Chapter 10, Problem 8:
Cable 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Defects 2 3 1 0 1 3 2 0 2 1 3 1 2 0
21
c  1.5
14
Median is 1.5: A = Above, B = Below, U = Up, D = Down.
Sample: 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Median: A A B B B A A B A B A B A B
Data: 2 3 1 0 1 3 2 0 2 1 3 1 2 0
Up/down: – U D D U U D D U D U D U D
Median Test:

𝑁 14
+1= 2
Standard deviation:
𝑁−1 14−1
=√ 4
= 1.80
Up/Down Test:
Observed number of runs = 10

2𝑁−1 2(14)−1
𝐸(𝑟)𝑢/𝑑 = 3
= 3
Standard deviation:
16𝑁−29 16(14)−29
𝜎𝑢/𝑑 = √ 90
=√ 90
10-25
Education.
= 1.47

d. Data from Chapter 10, Problem 7:

Median is 7.5.
Day: 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Median: B A A A A B B A A B B B B A
Data: 4 10 14 8 9 6 5 12 13 7 6 4 2 10
Up/down – U U D U D D U U D D D D U
Median Test:

𝑁 14
2 2
Standard deviation:
𝑁−1 14−1
4 4
= 1.80
Up/Down Test:

2𝑁−1 2(14)−1
3 3
Standard deviation:
16𝑁−29 16(14)−29
𝜎𝑢/𝑑 = √ 90
=√ 90
𝑧𝑡𝑒𝑠𝑡 = = = −1.36 (round to two decimals)

10-26
Education.
15. Given:
Expense voucher values are provided below. The data already have been converted to A/B and
U/D. Assume a median of 31.
Day Amount Day Amount Day Amount

1 B 27.69  21 B 28.60 U 41 B 26.76 D
2 B 28.13 U 22 B 20.02 D 42 B 30.51 U
3 A 33.02 U 23 B 26.67 U 43 B 29.35 D
4 B 30.31 D 24 A 36.40 U 44 B 24.09 D
5 A 31.59 U 25 A 32.07 D 45 B 22.45 D
6 A 33.64 U 26 A 44.10 U 46 B 25.16 U
7 A 34.73 U 27 A 41.44 D 47 B 26.11 U
8 A 35.09 U 28 B 29.62 D 48 B 29.84 U
9 A 33.39 D 29 B 30.12 U 49 A 31.75 U
10 A 32.51 D 30 B 26.39 D 50 B 29.14 D
11 B 27.98 D 31 A 40.54 U 51 A 37.78 U
12 A 31.25 U 32 A 36.31 D 52 A 34.16 D
13 A 33.98 U 33 B 27.14 D 53 A 38.28 U
14 B 25.56 D 34 B 30.38 U 54 B 29.49 D
15 B 24.46 D 35 A 31.96 U 55 B 30.81 U
16 B 29.65 U 36 A 32.03 U 56 B 30.60 D
17 A 31.08 U 37 A 34.40 U 57 A 34.46 U
18 A 33.03 U 38 B 25.67 D 58 A 35.10 U
19 B 29.10 D 39 A 35.80 U 59 A 31.76 D
20 B 25.19 D 40 A 32.23 D 60 A 34.90 U
Median Test:

𝑁 60
+1= 2
Standard deviation:
𝑁−1 60−1
4 4
= 3.84
10-27
Education.
Up/Down Test:

2𝑁−1 2(60)−1
𝐸(𝑟)𝑢/𝑑 = 3
= 3
Standard deviation:
16𝑁−29 16(60)−29
𝜎𝑢/𝑑 = √ 90
=√ 90
= 3.22
16. Given:
Inside diameter of successive parts decreases at an average rate of .001 centimeters per part.
The process sigma = .05 centimeters. Specifications are set at 3.0 to 3.5 centimeters. A three-
sigma cushion is set. Determine the number of parts that can be processed before tool
replacement becomes necessary. Round all values to a maximum of two decimals.
The key is that the mean starts high and then drifts down by .001 centimeters per part.
1) Observe that the process mean starts three-sigma below the upper specification limit:
Starting process mean = 3.5 – 3(.05) = 3.35
2) Determine what the ending process mean will be when it is three-sigma above the lower
specification limit:
Ending process mean = 3.0 + 3(.05) = 3.15
3) Determine how many pieces can be produced before the mean shifts from its starting point
down to its ending point:
3.35 − 3.15
= 200 𝑝𝑎𝑟𝑡𝑠
. 001
10-28
Education.
17. Given:
Refer back to Solved Problem 2.
Specifications are 9.65 to 10.35 minutes. Estimate the % of process output that can be expected to
fall within 9.65 to 10.35 minutes. The data from Solved Problem 2 are repeated below.
Sample 1 Sample 2 Sample 3 Sample 4 Sample 5

10.2 10.3 9.7 9.9 9.8
9.9 9.8 9.9 10.3 10.2
9.8 9.9 9.9 10.1 10.3
10.1 10.4 10.1 10.5 9.7
Looking at the data above, we see that there are 20 total points, and 2 points (10% of the total) are
outside the specifications limits.
The proportion within the specification limits is 18/20 = .9000 or 90.00% (round to two
decimals). This corresponds to .9000 falling within the limits. .9000/2 = .4500.
Using Appendix B, Table A, we see that .4500 falls midway between .4495 (z = 1.64) and .4505
(z = 1.65).
Because z = 1.64 & z = 1.65 < z = 3.00, we assume that specifications are not being met.
The manager may want to look at shifting the mean or reducing the sigma of the process.
18. Given:
A process has three steps. The scrap rate is 10% at the Step 1, 6% at Step 2, and 6% at Step 3.
a. If desired daily output is 450 units, determine the number of units that must be started:
Let x be the number of units started initially at Step 1.

With a scrap rate of 10% (.10) at Step 1, the input to Step 2 is x - .10x = .90x.
With a scrap rate of 6% (.06) at Step 2, the input to Step 3 = (1-.06)(.90x) = .94(.90x).
With a scrap rate of 6% (.06) at Step 3, the output of Step 3 = (1-.06)(.94)(.90x) =
(.94)(.94)(.90x).
Multiplying out: The output of Step 3 = .7952x (round to four decimals).
Let .7952x = 450
x = 450/.7952 = 565.90 (round up to an integer value of 566 units)
10-29
Education.
b. Assume the scrap rate is 50% of the original values. Now, the scrap rate is 5% at the Step 1,
3% at Step 2, and 3% at Step 3.
Let x be the number of units started initially at Step 1.

With a scrap rate of .05% (.05) at Step 1, the input to Step 2 is x - .05x = .95x.
With a scrap rate of 3% (.03) at Step 2, the input to Step 3 = (1-.03)(.95x) = .97(.95x).
With a scrap rate of 3% (.03) at Step 3, the output of Step 3 = (1-.03)(.97)(.95x) =
(.97)(.97)(.95x).
Multiplying out: The output of Step 3 = .8939x (round to four decimals).
Let .8939x = 450
x = 450/.8939 = 503.41 (round up to an integer value of 504 units)
Savings = 566 – 504 = 62 units.
c. Assume that scrap costs $10/unit.

Scrap (for original scrap rate) = Input – Output = 566 – 450 = 116 units
Cost of scrap = 116 * $10 = $1,160.00 (round to two decimals)
19. Given:
Refer back to the data in Example 5. Two additional observations have been taken: the first had
three defects and the second had four defects. Combine these last two with the first eighteen
observations. The original data are shown in the table below. The data already have been
converted to A/B and U/D.
Sample
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
3 2 4 5 1 2 4 1 2 1 3 4 2 4 2 1 3 1 3 4
A B A A B B A B B B A A B A B B A B A A
D U U D U U D U D U U D U D D U D U U
Determining the Median:

We must arrange the data from lowest to highest.
1 1 1 1 1 2 2 2 2 2 3 3 3 3 4 4 4 4 4 5
Median = 2.5
10-30
Education.
Median Test:

𝑁 20
2 2
Standard deviation:
𝑁−1 20−1
4 4
𝑧𝑡𝑒𝑠𝑡 = = = +0.92 (round to two decimals)
Up/Down Test:

2𝑁−1 2(20)−1
𝐸(𝑟)𝑢/𝑑 = 3
= 3
Standard deviation:
16𝑁−29 16(20)−29
𝜎𝑢/𝑑 = √ 90
=√ 90
𝑧𝑡𝑒𝑠𝑡 = = = +0.56 (round to two decimals)
10-31
Education.
20. Given:
A teller had the following service times for 20 randomly selected customers:
SAMPLE
1 2 3 4
4.5 4.6 4.5 4.7
4.2 4.5 4.6 4.6
4.2 4.4 4.4 4.8
4.3 4.7 4.4 4.5
4.3 4.3 4.6 4.9
a. Determine the mean of each sample. Round to a maximum of three decimals.
SAMPLE
1 2 3 4
4.5 4.6 4.5 4.7
4.2 4.5 4.6 4.6
4.2 4.4 4.4 4.8
4.3 4.7 4.4 4.5
4.3 4.3 4.6 4.9
Mean 4.3 4.5 4.5 4.7
b. Estimate the mean and standard deviation of the process:
=
x = (4.3 + 4.5 + 4.5 + 4.7)/4 = 4.5
We do not know the population standard deviation; therefore, we must estimate the
population standard deviation using the sample data above and calculate the sample standard
deviation (s) with N = 20:
1
𝑠=√ ∑𝑁 (𝑥 − 𝑥̅ )2 = .192 (round to three decimals)
𝑁−1 𝑖=1 𝑖
c. Estimate of mean and standard deviation of the sampling distribution:
Mean = 4.5.
𝜎 .192
𝜎𝑥̅ = = = .086 (round to three decimals)
√𝑛 √5
10-32
Education.
d. Three-sigma control limits (round to three decimals):

4.5 ± 3.00(.086) = 4.5 ± .258
UCL = 4.5 + .258 = 4.758
LCL = 4.5 – .258 = 4.242
Using Appendix B, Table A, z = + 3.00 corresponds to .4987 under the curve.
That leaves .5000 – .4987 = .0013 in the tail.
2 * .0013 = .0026 (alpha risk)
e. Alpha risk with control limits of 4.14 and 4.86:
x 4.86  4.5 .36

z    4.19
 .086 .086
The risk is very close to 0.
f. No means are outside the limits of 4.14 to 4.86.
g. Construct control charts for means and ranges (round to three decimals):
SAMPLE
1 2 3 4
4.5 4.6 4.5 4.7
4.2 4.5 4.6 4.6
4.2 4.4 4.4 4.8
4.3 4.7 4.4 4.5
4.3 4.3 4.6 4.9
Mean 4.3 4.5 4.5 4.7
Range .3 .4 .2 .4

R = (.3 + .4 + .2 + .4)/4 = 0.325 (round to three decimals)
n=5
Mean Chart: A2 = 0.58

Upper Control Limit (UCL) = 𝑥̿ + 𝐴2 𝑅̅ = 4.5 + 0.58(0.325) = 4.5 + .189 = 4.689
Lower Control Limit (LCL) = 𝑥̿ − 𝐴2 𝑅̅ = 4.5 − 0.58(0.325) = 4.5 − .189 = 4.311
Range Chart: D4 = 2.11; D3 = 0

Lower Control Limit (LCL) = 𝐷3 𝑅̅ = 0(0.325) = 0.000
10-33
Education.
4.800
Mean Chart
4.700 4.689
4.600
4.500 4.500 UCL

LCL
4.400 Mean
4.300 4.312
4.200
1 2 3 4
Sample mean 1 is below the LCL and Sample mean 4 is above the UCL.
0.800
Range Chart
0.686
0.600
0.400 UCL
0.325 LCL
Range
0.200
0.000 0.000
1 2 3 4
All sample ranges are within the limits.
h. The control limits are different for means in parts d and g because two different measures of
dispersion are being used—the standard deviation and the range.
i. If the process has a known mean of 4.4 and a known standard deviation of .18, calculate
three-sigma control limits for a Mean chart:
 0.18 
4.4  3   4.4  .241
 5 
UCL = 4.641
LCL = 4.159
10-34
Education.
4.900
Mean Chart
4.700
4.641
4.500 UCL
LCL
4.400
Range
4.300
4.159
4.100
1 2 3 4
Sample mean 4 is above the UCL.
21. Given:
Process has a mean of .04 and a standard deviation of .003. The allowable variation is from .03 to
.05.
a. Process capability (the process is centered) (round to two decimals):

specificat ion width .02 .02
Cp     1.11
process width 6(.003) .018
b. To be capable, the process capability ratio must be at least 1.33. In this instance, the index is
1.11, so the process is not capable.
22. Given:
We have the standard deviation for each of five processes and the specifications for a job on that
machine:
Standard Deviation Job Specification

Process (in.) (in.)
001 0.02 0.05
002 0.04 0.07
003 0.10 0.18
004 0.05 0.15
005 0.01 0.04
Assuming that the process mean is centered for each process, we calculate Cp for each process:
specificat ion width
Cp  (round to two decimals)
process width
Note: We must determine the specification width and the process width for each process to
calculate Cp. Specification width = 2 * Job Specification given in the table above.
Process width = 6 * standard deviation.
10-35
Education.
Standard
Deviation
Process (in.) Process Width Specification Width Cp Capable?
001 0.02 6 * 0.02 = 0.12 2 * 0.05 = 0.10 0.10/0.12 = 0.83 No
002 0.04 6 * 0.04 = 0.24 2 * 0.07 = 0.14 0.14/0.24 = 0.58 No
003 0.10 6 * 0.10 = 0.60 2 * 0.18 = 0.36 0.36/0.60 = 0.60 No
004 0.05 6 * 0.05 = 0.30 2 * 0.15 = 0.30 0.30/0.30 = 1.00 No
005 0.01 6 * 0.01 = 0.06 2 * 0.04 = 0.08 0.08/0.06 = 1.33 Yes
23. Given:
We have the information of four machines below. The specification width is .48 mm. What other
piece of information do we need to make a recommendation?
Standard Deviation
Machine Cost per unit ($) (mm.)
A 20 0.059
B 12 0.060
C 11 0.063
D 10 0.061
Assuming that the process mean is centered for each machine, we calculate Cp for each machine:
specificat ion width
Cp  (round to two decimals)
process width
Note: We must determine the process width for each machine to calculate Cp. Specification width
is given already. Process width = 6 * standard deviation.
Standard
Machine Deviation Process Width Specification Width Cp Capable?
A 0.059 6 * 0.059 = 0.35 0.48 0.48/0.35 = 1.37 Yes
B 0.060 6 * 0.060 = 0.36 0.48 0.48/0.36 = 1.33 Yes
C 0.063 6 * 0.063 = 0.38 0.48 0.48/0.38 = 1.26 No
D 0.061 6 * 0.061 = 0.37 0.48 0.48/0.37 = 1.30 No
We can narrow the choice to Machines A and B because they are the only ones with a capability
ratio of at least 1.33. We would need to know if the slight additional capability of Machine A is
worth an extra cost of $8 per unit ($20 - $12).
10-36
Education.
24. Given:
We have the mean, standard deviation, lower spec, and upper spec for three processes shown
below. Each of the processes means is not centered.
Standard Lower Upper

Process Mean Deviation Spec Spec
H 15.0 0.32 14.1 16.0
K 33.0 1.00 30.0 36.5
T 18.5 0.40 16.5 20.1
Let USL = Upper Specification Limit, LSL = Lower Specification Limit,

X = Process mean,  = Process standard deviation
For Process H:
X  LSL 15.0  14.1

  0.94
3 (3)(0.32)
USL  X 16.0  15.0
  1.04
3 (3)(0.32)
C pk  min 0.94, 1.04  0.94
Assuming the minimum acceptable C pk is 1.33, because 0.94 < 1.33, this process is not capable.
For Process K:
X  LSL 33.0  30.0

  1.00
3 (3)(1.00)
USL  X 36.5  33.0
  1.17
3 (3)(1.00)
C pk  min{ 1.00, 1.17}  1.00
Assuming the minimum acceptable C pk is 1.33, because 1.00 < 1.33, the process is not capable.
For Process T:
X  LSL 18.5  16.5

  1.67
3 (3)(0.40)
USL  X 20.1  18.5
  1.33
3 (3)(0.40)
C pk  min{ 1.67, 1.33}  1.33
Assuming the minimum acceptable C pk is 1.33, because 1.33 ≥ 1.33, the process is capable.
10-37
Education.
25. Given:
X = Process mean,  = Process standard deviation.
USL = 90 minutes, LSL = 50 minutes,

X 1 = 74 minutes,  1 = 4.0 minutes
X 2 = 72 minutes,  2 = 5.1 minutes
Neither firm’s mean repair time is centered so we must use Cpk.
For the first repair firm:
X  LSL 74  50
  2.00
3 (3)( 4.0)
USL  X 90  74
  1.33
3 (3)( 4.0)
C pk  min{ 2.00, 1.33}  1.33
Because 1.33 ≥ 1.33, Firm 1 is capable.
For the second repair firm:
X  LSL 72  50
  1.44
3 (3)(5.1)
USL  X 90  72
  1.18
3 (3)(5.1)
C pk  min{ 1.44,1.18}  1.18
Because 1.18 < 1.33, Firm 2 is not capable.
Choose Firm 1 because it is capable (assuming that process capability is the only criterion).
10-38
Education.
26. Given:
X = Process mean,  = Process standard deviation.
USL = 30 minutes, LSL = 45 minutes,

X Armand = 38 minutes,  Armand = 3.0 minutes
X Jerry = 37 minutes,  Jerry = 2.5 minutes
X Melissa = 37.5 minutes,  Melissa = 1.8 minutes
a. Determine capability of each participant:
For Armand:
X  LSL 38  30
  0.89
3 (3)(3.0)
USL  X 45  38
  0.78
3 (3)(3.0)
C pk  min{ 0.89, 0.78}  0.78
Because 0.78 < 1.33, Armand is not capable.
For Jerry:
X  LSL 37  30
  0.93
3 (3)( 2.5)
USL  X 45  37
  1.07
3 (3)( 2.5)
C pk  min{ 0.93, 1.07}  0.93
Because 0.93 < 1.33, Jerry is not capable.
For Melissa:
Because USL  X  X  LSL  7.5 , the process is centered, therefore we will use Cp to
measure process capability.
USL  LSL 45  30
Cp    1.39
6 (6)(1.8)
Because 1.39 ≥ 1.33, Melissa is capable.
b. No, the value of Cpk can never exceed Cp. When the process is centered exactly, Cpk will
equal Cp. When the process is not centered, Cpk always will be less than Cp.
10-39
Education.
27. Given:
a. Specifications for the 12-ounce chocolate bar are 330 grams to 350 grams. Average fill is 340
grams. What is the largest standard deviation (in grams) possible for the machine where it
still could be considered capable?
The process is centered so we can use Cp.
USL  LSL 350  330

Cp  
6 6
Let Cp = 1.33 and solve for σ.
350 − 330
1.33 =
6𝜎
1.33(6σ) = 20
(1.33)6σ = 20
7.98σ = 20
σ = 20/7.98
σ = 2.506 (round to three decimals)
b. The machine that fills the bar models for the 1-ounce bars has a standard deviation of .80
grams per bar. The filling machine is set to deliver an average of 1.01 ounces per bar,
Specifications for the six-bar box are 160 to 180 grams. Hint: The variance for the box is
equal to six times the bar variance.
Round to a maximum of three decimals:

1 ounce = 28.33 grams.
Bar variance = (.80)2 = 0.64
Box variance = 6 * 0.64 = 3.84
Box standard deviation = √3.84 = 1.96
Average box weight = 6 *1.01 = 6.06 ounces * 28.33 = 171.68 grams for the six-bar box.
X  LSL 171.68  160
  1.99
3 (3)(1.96)
USL  X 180  171.68
  1.41
3 (3)(1.96)
C pk  min{ 1.99,1.41}  1.41
Because 1.41 ≥ 1.33, this process is capable.
10-40
Education.
c. The lowest setting in ounces for the filling machine in terms of the six-bar box that will
provide capability:
The lowest setting implies that the mean for the six-bar box is less than average (closer to the
LSL). Set that part of the Cpk equation = 1.33:
X  LSL X  160
  1.33
3 (3)(1.96)
𝑋̅ − 160 = 1.33(3)(1.96)
𝑋̅ = 160 + 7.82 = 167.82 grams per six-bar box.
167.82 grams per box /6 bars per box= 27.97 grams per bar.
27.97 grams per bar / 28.33 grams per ounce = 0.987 ounces per bar.
28. Given:
All points are within the control limits. Next, check for nonrandomness.
N = 22. Count the number of A/B and U/D on the chart in the text.
Median Test:

𝑁 22
2 2
Standard deviation:
𝑁−1 22−1
=√ 4
= 2.29
10-41
Education.
Up/Down Test:

2𝑁−1 2(22)−1
𝐸(𝑟)𝑢/𝑑 = 3
= 3
Standard deviation:
16𝑁−29 16(22)−29
𝜎𝑢/𝑑 = √ 90
=√ 90
= 1.89
29. Given:
Observation 1 2 3 4 5 6 7 8 9 10 11 12
No. of errors 1 0 3 2 0 1 3 2 1 0 2 3
Step 1: A c-chart is appropriate.
The mean of the data is c = 18/12 = 1.50.

Control limits:
c  3 c  1.50  3 1.50  1.50  3.674
UCL = 5.174
6.000
c-Chart
5.000 5.174
4.000
3.000 UCL
LCL
Errors
2.000
1.500
1.000
0.000 0.000
1 2 3 4 5 6 7 8 9 10 11 12
10-42
Education.
All points are within the control limits.
Step 2: Conduct run tests:
Median Test:

𝑁 12
2 2
Standard deviation:
𝑁−1 12−1
=√ 4
= 1.66
Up/Down Test:

2𝑁−1 2(12)−1
3 3
Standard deviation:
16𝑁−29 16(12)−29
𝜎𝑢/𝑑 = √ 90
=√ 90
= 1.35
So far, the process appears to be random. However, there is obvious cycling in the data in the
control chart based on repeating patterns of runs in the Median Test. Therefore, the
process output is nonrandom.
10-43
Education.
Case: Toys Inc.

A consultant must consider the long-term implications of decisions suggested by management.
1. Cutting cost in design and product development may not be beneficial to the company in the long
run.
2. The trade-in and repair program, while appeasing customers in the short run, may be too costly
and will not correct the root cause of the problem.
3. Because the company thrives on its reputation of high quality products, it needs to continue to
design products of high quality that fulfill the needs of the marketplace. Manufacturing needs to
place greater emphasis on preventive quality management/control rather than inspecting already
completed parts. The company may want to consider investing more in R&D.
4. If implemented well, this strategy will enable the company to become more competitive in the
long run.
Case: Tiger Tools

1. For the first data set R = 0.873. From Table 10–2, for n = 20, A2 = 0.18. Using the hint, the
estimated standard deviation is .234:
 AR
A2 R  3 . Rearranging terms, we have 2 n 
n 3
Solving, we obtain  
(0.18)(0.873)
3
20  .234 
When determining process capability, we do not know the USL and LSL. We know the
specification width only; therefore, we will use Cp.
1.44
The process capability is  1.03 . Because this is less than 1.33, the process is not
(6)(.234 )
capable.
10-44
Education.
2. First set of data (n = 20 per sample):

From Table 10.3 with n = 20: A2 = 0.18.
𝑥̿̅ = 44.999 𝑅̅ = 0.873
Mean Chart:
45.200 Mean Chart: First Set

45.150 45.156
45.100
45.050
45.000 44.999 UCL

LCL
44.950 Mean
44.900
44.850 44.842
44.800
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
10-45
Education.
Second set of data (n = 5 per sample):

From Table 10.3 with n = 5: A2 = 0.58.
𝑥̿̅ = 44.996 𝑅̅ = 0.401
Mean Chart:
45.300
Mean Chart: Second Set
45.229
45.200
45.100
45.000 UCL
44.996
LCL
Mean
44.900
44.800
44.763
44.700
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
The second set of samples (sample size = n = 5) seems to be cycling. The first set of samples
(sample size = n = 20) did not reveal the changes that were occurring because large samples
combine the results of several different process means. By taking smaller samples (second set),
the pattern was easier to discern.
3. If the problem with cycling could be removed, the true process standard deviation probably
would be much smaller than the apparent process standard deviation. For the second set of
samples, 𝑅̅= 0.401. Performing the same calculation as for the first set, we obtain an estimate of
the process standard deviation of:
A2 R (0.58)(0.401)
 n 5  .173
3 3
1.44
The potential process capability is  1.39.
6(.173)
Because this is more than 1.33, the process is capable.
4. Small samples tend to be less reliable than large samples (the standard deviation of the sampling
distribution of means decreases as the sample size increases). Also, a manager must weigh the
cost of inspecting each item and the cost of taking a sample. If the cost to obtain a sample is high,
but the cost to inspect an item is low, larger samples might be the better choice.
10-46
Education.

IMChap 010

Uploaded by

Copyright:

Available Formats

IMChap 010

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

IMChap 010

Uploaded by

Copyright:

Available Formats

Chapter 10 - Quality Control

Reading: In the Chips at Jays

Reading: Bar Codes Might Cut Drug Errors in Hospitals

Answers to Discussion and Review Questions

Critical Thinking Exercises

Additional Experiential Learning Exercises

of colored beads or jelly beans.

Using Appendix B, Table A, find the value of z that corresponds to .4850:

UCL = 1.0 + .004 = 1.004

Sample Mean Range

Sample Mean Range

From Table 10.3 with n = 20: A2 = 0.18; D3 = 0.41; D4 = 1.59.

3.200 Mean Chart

From Table 10.3 with n = 5: A2 = 0.58; D3 = 0; D4 = 2.11.

80.000 79.960 UCL

(.020 + .010 + .025 + .045)/4 = .025

We know the following: .025 + z(.011) = .047

All points fall within the limits. The process is in control.

p(1  p) .02(1  .02)

p(1  p) .02(1  .02)

Total number of defects = 25

p(1  p) .0096(1  .0096)

All points are within the limits. The process is in control.

All points are within the limits. The process is in control.

Total number of defects = 87

total number of defectives 87

All points are within the limits. The process is in control.

15.0 + 3(.02) = 15.06

15.2 – 3(.02) = 15.14

Sample Mean Sample Mean Sample Mean Sample Mean

UCL = 3.85 + .117 = 3.967

4.000 Mean Chart

3.850 3.850 UCL

Sample 29 mean is outside the limits. The process is not in control.

Observed number of runs = 18.

Expected number of runs (N = total number of observations = 39):

This test statistic is within ± 2. Conclusion: Random.

Observed number of runs = 29.

Expected number of runs (N = total number of observations = 39):

This test statistic is within ± 2. Conclusion: Random.

Overall Conclusion: The process is random. Neither test detected nonrandomness.

13. a. First control chart is marked with A/B and U/D:

Observed number of runs = 18.

Expected number of runs (N = total number of observations = 26):

This test statistic is within ± 1.96. Conclusion: Random.

Observed number of runs = 17.

Expected number of runs (N = number of observations = 26):

This test statistic is within ± 1.96. Conclusion: Random.

Overall Conclusion: The process is random. Neither test detected nonrandomness.

b. Second control chart is marked with A/B and U/D:

Observed number of runs = 8.

This test statistic is not within ± 1.96. Conclusion: Nonrandom.

Observed number of runs = 22.

Expected number of runs (N = number of observations = 26):

This test statistic is not within ± 1.96. Conclusion: Nonrandom.

Observed number of runs = 14.

This test statistic is within ± 2.00. Conclusion: Random.

Observed number of runs = 8.

Expected number of runs (round to 1 decimal) (N = number of observations = 20):