Chapter 18: Statistical Quality Control 1

Chapter 18 Statistical Quality Control

LEARNING OBJECTIVES Chapter 18 presents basic concepts in quality control, with a particular emphasis on statistical quality control techniques, thereby enabling you to: 1. 2. 3. 4. 5. Understand the concepts of quality, quality control, and total quality management. Understand the importance of statistical quality control in total quality management. Learn about process analysis and some process analysis tools. Learn how to construct x charts, R charts, p charts, and c charts. Understand the theory and application of acceptance sampling.

CHAPTER TEACHING STRATEGY The objective of this chapter is to present the major concepts of statistical quality control including control charts and acceptance sampling in a context of total quality management. Too many texts focus only on a few statistical quality control procedures and fail to provide the student with a managerial, decisionmaking context within which to use quality control statistical techniques. In this text, the concepts of total quality management along with some varying definitions of quality and some of the major theories in quality are presented. From this, the student can formulate a backdrop for the statistical techniques presented. Some statistics instructors argue that students are exposed to some of this material in other courses. However, the background material on quality control is relatively brief; and at the very least, students should be required to read over these pages before beginning the study of control charts.

Chapter 18: Statistical Quality Control 2

The background material helps the students understand that everyone does not agree on what a quality product is. After all, if there is no agreement on what is quality, then it is very difficult to ascertain or measure if it is being accomplished. The notion of in-process quality control helps the student understand why we generate the data that we use to construct control charts. Once the student is in the work world, it will be incumbent upon him/her to determine what measurements should be taken and monitored. A discussion on what types of measurements can be garnered in a particular business setting might be worthy of some class time. For example, if a hospital lab wants to improve quality, how would they go about it? What measurements might be useful? How about a production line of computer chips? The chapter contains "some important quality concepts". The attempt is to familiarize, if only in passing, the student with some of the more well-known quality concepts. Included in the chapter are such things as team-building, benchmarking, just-in-time, reengineering, FMEA, Six Sigma, and Poka-Yoke all of which can effect the types of measurements being taken and the statistical techniques being used. It is a disservice to send students into the business world armed with statistical techniques such as acceptance sampling and control charts but with their heads in the sand about how the techniques fit into the total quality picture. Chapter 18 contains a section on process analysis. Improving quality usually involves an investigation of the process from which the product emerges. The most obvious example of a process is a manufacturing assembly line. However, even in most service industries such insurance, banking, or healthcare there are processes. A useful class activity might be to brainstorm aboutbrainstorm about what kind of process is involved in a person buying gasoline for their car, checking in to a hospital, or purchasing a health club membership. Think about it from a companys perspective. What activities must occur in order for a person to get their car filled up? In analyzing process, we first discuss the construction of flowcharts. Flowcharting can be very beneficial in identifying activities and flows that need to be studied for quality improvement. One very important outcome of a flowchart is the identification of bottlenecks. You may find out that all applications for employment, for example, must pass across a clerks desk where they sit for several days. This backs up the system and prevents flow. Other process techniques include fishbone diagrams, Pareto analysis, and control charts. In this chapter, four types of control charts are presented. Two of the charts, the x bar chart and the R chart, deal with measurements of product attributes such as weight, length, temperature and others. The other two charts deal with whether or not items are in compliance with specifications (p chart) or the number of noncompliances per item (c chart). The c chart is less widely known and used than the other three. As part of the material on control charts, a discussion on variation is presented. Variation is one of the main concerns of quality control. A discussion on various types of variation that can occur in a business setting can be profitable in helping the student understand why particular measurements are charted and controlled.

Chapter 18: Statistical Quality Control 3

CHAPTER OUTLINE 18.1 Introduction to Quality Control What is Quality Control? Total Quality Management Some Important Quality Concepts Benchmarking Just-in-Time Systems Reengineering Failure Mode and Effects Analysis (FMEA) Poka-Yoke Six Sigma Design for Six Sigma Lean Manufacturing Team Building Process Analysis Flowcharts Pareto Analysis Cause-and-Effect (Fishbone) Diagrams Control Charts Check Sheets Histogram Scatter Chart Control Charts Variation Types of Control Charts x Chart R Charts p Charts c Charts Interpreting Control Charts Acceptance Sampling Single Sample Plan Double-Sample Plan Multiple-Sample Plan Determining Error and OC Curves

18.2

18.3

18.4

Chapter 18: Statistical Quality Control 4

KEY TERMS Acceptance Sampling After-Process Quality Control Benchmarking c Chart Cause-and-Effect Diagram Centerline Check Sheet Consumer's Risk Control Chart Design for Six Sigma Double-Sample Plan Failure Mode and Effects Analysis Fishbone Diagram Flowchart Histogram In-Process Quality Control Ishikawa Diagram Just-in-Time Inventory Systems Lean Manufacturing Lower Control Limit (LCL) Manufacturing Quality Multiple-Sample Plan Operating Characteristic (OC) Curve p Chart Pareto Analysis Pareto Chart Poka-Yoke Process Producers Risk Product Quality Quality Quality Circle Quality Control R Chart Reengineering Scatter Chart Single-Sample Plan Six Sigma Team Building Total Quality Management Transcendent Quality Upper Control Limit (UCL) User Quality Value Quality x Chart

Chapter 18: Statistical Quality Control 5

SOLUTIONS TO PROBLEMS IN CHAPTER 18 18.2 Complaint Busy Signal Too long a Wait Could not get through Get Disconnected Transferred to the Wrong Person Poor Connection Total Number 420 184 85 37 10 8 744 99.99 % of Total 56.45 24.73 11.42 4.97 1.34 1.08

18.4

x 1 = 27.00, x 2 = 24.29, x 3 = 25.29, x 4 = 27.71, x 5 = 25.86

R1 = 8, R2 = 8, R3 = 9, R4 = 7, R5 = 6
x

= 26.03

= 7.6

For x Chart: Centerline: UCL:

x

Since n = 7, A2 = 0.419 = 26.03

+ A2 R = 26.03 + (0.419)(7.6) = 29.21

Chapter 18: Statistical Quality Control 6

LCL:

- A2 R = 26.03 (0.419)(7.6) = 22.85 Since n = 7, D3 = 0.076

R

For R Chart: Centerline: UCL: LCL:

x

D4 = 1.924

= 7.6

D4 R = (1.924)(7.6) = 14.62 D3 R = (0.076)(7.6) = 0.58

Chart:

R Chart:

Chapter 18: Statistical Quality Control 7

18.5

x 1 = 4.55, x 2 = 4.10, x 3 = 4.80, x 4 = 4.70, x 5 = 4.30, x 6 = 4.73, x 7 = 4.38

R1 = 1.3, R2 = 1.0, R3 = 1.3, R4 = 0.2, R5 = 1.1, R6 = 0.8, R7 = 0.6

x

= 4.51

= 0.90

For x Chart: Centerline: UCL: LCL:

x x

Since n = 4, A2 = 0.729 = 4.51

+ A2 R = 4.51 + (0.729)(0.90) = 5.17 - A2 R = 4.51 (0.729)(0.90) = 3.85 Since n = 4, D3 = 0

R

For R Chart: Centerline: UCL: LCL:

x

D4 = 2.282

= 0.90

D4 R = (2.282)(0.90) = 2.05 D3 R = 0

Chart:

Chapter 18: Statistical Quality Control 8

R Chart:

18.6

p1 = .02, p2 = .07, p3 = .04, p4 = .03, p5 = .03 p6 = .05, p7 = .02, p8 = .00, p9 = .01, p10 = .06

p = .033 Centerline: p = .033

(. 033 )(. 967 ) = .033 + .054 = .087 100 (. 033 )(. 967 ) = .033 - .054 = .000 100

UCL: .033 + 3

LCL: pChart:

.033 - 3

Chapter 18: Statistical Quality Control 9

18.7

p1 = .025, p2 = .000, p3 = .025, p4 = .075, p5 = .05, p6 = .125, p7 = .05

p = .050 Centerline: p = .050

(. 05 )(. 95 ) = .05 + .1034 = .1534 40 (. 05 )(. 95 ) = .05 - .1034 = .000 40

UCL: .05 + 3

LCL:

.05 - 3

p Chart:

18.8

22 35

= 0.62857
c

Centerline: UCL:

= 0.62857 = 0.62857 + 3
0.62857

c +3 c

0.62857 + 2.37847 = 3.00704 LCL:

c +3 c

= 0.62857 - 3

0.62857

0.62857 2.37847 = .000

Chapter 18: Statistical Quality Control 10

c Chart:

18.9

43 32

= 1.34375
c

Centerline: UCL:

= 1.34375 = 1.34375 + 3
1.3 3 5 47

c +3 c

1.34375 + 3.47761 = 4.82136 LCL:

c +3 c

= 1.34375 - 3

1.3 3 5 47

1.34375 - 3.47761 = 0.000 c Chart:

18.10 a.) Six or more consecutive points are decreasing. Two of three consecutive points are in the outer one-third (near LCL). Four out of five points are in the outer two-thirds (near LCL).

Chapter 18: Statistical Quality Control 11

b.) This is a relatively healthy control chart with no obvious rule violations. c.) One point is above the UCL. Two out of three consecutive points are in the outer one-third (both near LCL and near UCL). There are six consecutive increasing points.

18.11 While there are no points outside the limits, the first chart exhibits some problems. The chart ends with 9 consecutive points below the centerline. Of these 9 consecutive points, there are at least 4 out of 5 in the outer 2/3 of the lower region. The second control chart contains no points outside the control limit. However, near the end, there are 8 consecutive points above the centerline. The p chart contains no points outside the upper control limit. Three times, the chart contains two out of three points in the outer third. However, this occurs in the lower third where the proportion of noncompliance items approaches zero and is probably not a problem to be concerned about. Overall, this seems to display a process that is in control. One concern might be the wide swings in the proportions at samples 15, 16 and 22 and 23.

18.12 For the first sample: If x1 > 4 then reject If x1 < 2 then accept If 2 < x1 < 4 then take a second sample For the second sample, c2 = 3: If x1 + x2 < c2 then accept If x1 + x2 > c2 then reject But x1 = 2 and x2 = 2 so x1 + x2 = 4 > 3

Reject the lot because x1 + x2 = 4 > c2 = 3 This is a double sample acceptance plan 18.13 n = 10 c=0 p0 = .05

Chapter 18: Statistical Quality Control 12

P(x = 0) = 10C0(.05)0(.95)10 = .5987 1 - P(x = 0) = 1 - .5987 = .4013 The producer's risk is .4013 p1 = .14 P(x = 0) = 15C0(.14)0(.86)10 = .2213 The consumer's risk is .2213

18.14 n = 12

c=1

p0 = .04

Producer's Risk = 1 - [P(x = 0) + P(x = 1)] = 1 - [12C0(.04)0(.96)12 + 12C1(.04)1(.96)11] = 1 - [.6127 + .30635] = 1 - .91905 = .08095 p1 = .15 Consumer's Risk = P(x = 0) + P(x = 1) = 12C0(.15)0(.85)12 +
12

C1(.15)1(.85)11 = .14224 + .30122 = .44346

Chapter 18: Statistical Quality Control 13

18.15 n = 8

c=0 p .01 .02 .03 .04 .06 .07 .08 .09 .10 .11 .12 .13 .14 .15

p0 = .03 Probability

p1 = .1

.9227 .8506 .7837 .7214 Producer's Risk for (p0 = .03) = .05 .6634 1 - .7837 = .2163 .6096 .5596 .5132 Consumer's Risk for (p1 = .10) = .4305 .4703 .4305 .3937 .3596 .3282 .2992 .2725

OC Chart:

Chapter 18: Statistical Quality Control 14

18.16 n = 11 p .02 .04 .06 .08 .10 .12 .14 .16 .18 .20 .22 .24

c=1

p0 = .08

p1 = .20

Probability .9805 .9308 .8618 .7819 Producer's Risk for (p0 = .08) = 1 - .7819 = .2181 .6974 .6127 .5311 .4547 Consumer's Risk for (p1 = 20) = .3221 .3849 .3221 .2667 .2186

OC Chart:

Chapter 18: Statistical Quality Control 15

18.17 Stop N (no) D K L M (yes) Stop Stop (no) (no) Start A B (yes) C E F G (yes) H(no) J Stop (yes) I

18.18

Problem 1 2 3 4 5 6 7 8 9 10

Frequency 673 29 108 379 73 564 12 402 54 202 2496

Percent of Total 26.96 1.16 4.33 15.18 2.92 22.60 0.48 16.11 2.16 8.09

Chapter 18: Statistical Quality Control 16

Pareto Chart:

18.19 Fishbone Diagram:

Chapter 18: Statistical Quality Control 17

18.20 a) n = 13, c = 1, p0 = .05, p1 = .12 Under p0 = .05, the probability of acceptance is: C0(.05)0(.95)13 + 13C1(.05)1(.95)12 = (1)(1)(.51334) + (13)(.05)(.54036) = .51334 + .35123 = .86457
13

The probability of being rejected = 1 - .86457 = .13543 Under p1 = .12, the probability of acceptance is: C0(.12)0(.88)13 + 13C1(.12)1(.88)12 = (1)(1)(.18979) + (13)(.12)(.21567) = .18979 + .33645 = .52624
13

The probability of being rejected = 1 - .52624 = .47376 b) n = 20, c = 2, p0 = .03 The probability of acceptance is: C0(.03)0(.97)19 + 20C1(.03)1(.97)19 + 20C2(.03)2(.97)18 = (1)(1)(.54379) + (20)(.03)(.56061) + (190)(.0009)(.57795) = .54379 + .33637 + .09883 = .97899
20

The probability of being rejected = 1 - .97899 = .02101 which is the producers risk

18.21

p1 = .06, p2 = .22, p3 = .14, p4 = .04, p5 = .10, p6 = .16, p7 = .00, p8 = .18, p9 = .02, p10 = .12

p=

52 = .104 500

Centerline:

p = .104
(. 104 )(. 896 ) 50

UCL: .104 + 3

= .104 + .130 = .234

Chapter 18: Statistical Quality Control 18

LCL:

.104 - 3

(. 104 )(. 896 ) 50

= .104 - .130 = .000

p Chart:

18.22

x 1 = 24.022, x 2 = 24.048, x 3 = 23.996, x 4 = 24.000, x 5 = 23.998, x 6 = 24.018, x 7 = 24.000, x 8 = 24.034, x 9 = 24.014, x 10 = 24.002, x 11 = 24.012, x 12 = 24.022

R1 = .06, R2 = .09, R3 = .08, R4 = .03, R5 = .05, R6 = .05, R7 = .05, R8 = .08, R9 = .03, R10 = .01, R11 = .04, R12 = .05
x

= 24.01383

= 0.05167

For x Chart: Centerline: UCL: LCL:

x

Since n = 12, A2 = .266 x = 24.01383

+ A2 R = 24.01383 + (0.266)(.05167) = 24.01383 + .01374 = 24.02757 - A2 R = 24.01383 - (0.266)(.05167) = 24.01383 - .01374 = 24.00009 Since n = 12, D3 = .284 D4 = 1.716

For R Chart:

Chapter 18: Statistical Quality Control 19

Centerline: UCL: LCL:

x

= .05167

D4 R = (1.716)(.05167) = .08866 D3 R = (.284)(.05167) = .01467

Chart:

R Chart:

Chapter 18: Statistical Quality Control 20

18.23 n = 15, c = 0, p0 = .02, p1 = .10 p .01 .02 .04 .06 .08 .10 .12 .14 Probability .8601 .7386 .5421 .3953 .2863 .2059 .1470 .1041

Producer's Risk for (p0 = .02) = 1 - .7386 = .2614 Consumer's Risk for (p1 = .10) = .2059 OC Curve:

Chapter 18: Statistical Quality Control 21

18.24

77 36

= 2.13889
c

Centerline: UCL:

= 2.13889 = 2.13889 + 3
2.13889

c +3 c

2.13889 + 4.38748 = 6.52637 LCL:

c +3 c

= 2.13889 - 3

2.13889

2.13889 4.38748 = .00000 c Chart:

18.25

x 1 = 1.2100, x 2 = 1.2050, x 3 = 1.1900, x 4 = 1.1725, x 5 = 1.2075, x 6 = 1.2025, x 7 = 1.1950, x 8 = 1.1950, x 9 = 1.1850

R1 = .04, R2 = .02, R3 = .04, R4 = .04, R5 = .06, R6 = .02, R7 = .07, R8 = .07, R9 = .06,

x

= 1.19583

= 0.04667

For x Chart: Centerline:

Since n = 4, A2 = .729 = 1.19583

Chapter 18: Statistical Quality Control 22

UCL: LCL:

+ A2 R = 1.19583 + .729(.04667) = 1.19583 + .03402 = 1.22985 x - A2 R = 1.19583 - .729(.04667) = 1.19583 - .03402 = 1.16181 Since n = 9, D3 = .184
R

For R Chart: Centerline: UCL: LCL:

x

D4 = 1.816

= .04667

D4 R = (1.816)(.04667) = .08475 D3 R = (.184)(.04667) = .00859

Chart:

R chart:

18.26

x 1 = 14.99333, x 2 = 15.00000, x 3 = 14.97833, x 4 = 14.97833,

Chapter 18: Statistical Quality Control 23

x 5 = 15.01333, x 6 = 15.00000, x 7 = 15.01667, x 8 = 14.99667,

R1 = .03, R2 = .07, R3 = .05, R4 = .05, R5 = .04, R6 = .05, R7 = .05, R8 = .06

x

= 14.99854
x

= 0.05

For

Chart:

Since n = 6, A2 = .483
x

Centerline: UCL: LCL:

x

= 14.99854

+ A2 R = 14.99854 + .483(.05) = 14.00854 + .02415 = 15.02269 - A2 R = 14.99854 - .483(.05) = 14.00854 - .02415 = 14.97439 Since n = 6, D3 = 0
R

For R Chart: Centerline: UCL: LCL:

x

D4 = 2.004

= .05

D4 R = 2.004(.05) = .1002 D3 R = 0(.05) = .0000

Chart:

R chart:

Chapter 18: Statistical Quality Control 24

18.27

p1 = .12, p2 = .04, p3 = .00, p4 = .02667, p5 = .09333, p6 = .18667, p7 = .14667, p8 = .10667, p9 = .06667, p10 = .05333, p11 = .0000, p12 = .09333

p=

70 = .07778 900

Centerline:

p = .07778

UCL: .07778 + 3

(. 07778 )(. 92222 ) = .07778 + .09278 = .17056 75 (. 07778 )(. 92222 ) = .07778 - .09278 = .00000 75

LCL:

.07778 - 3

p Chart:

Chapter 18: Statistical Quality Control 25

18.28

16 25

= 0.64
c

Centerline: UCL:

= 0.64 = 0.64 + 3
0 .6 4

c +3 c

0.64 + 2.4 = 3.04 LCL:

c +3 c

= 0.64 - 3

0 .6 4

0.64 2.4 = .00000 c Chart:

Chapter 18: Statistical Quality Control 26

18.29 n = 10 p

c=2

p0 = .10

p1 = .30

Probability .9885 .9298 .8202 .6778 .5256 .3828 .2616 .1673 .0996 .0547

.05 .10 .15 .20 .25 .30 .35 .40 .45 .50

Producer's Risk for (p0 = .10) = 1 - .9298 = .0702 Consumer's Risk for (p1 = .30) = .3828

Chapter 18: Statistical Quality Control 27

18.30

81 40

= 2.025
c

Centerline: UCL:

= 2.025 = 2.025 + 3
2.02 5

c +3 c

2.025 + 4.26907 = 6.29407 LCL:

c +3 c

= 2.025 - 3

2.0 25

2.025 4.26907 = .00000 c Chart:

18.31

p1 = .05, p2 = .00, p3 = .15, p4 = .075, p5 = .025, p6 = .025, p7 = .125, p8 = .00, p9 = .10, p10 = .075, p11 = .05, p12 = .05, p13 = .15, p14 = .025, p15 = .000

p=

36 = .06 600

Centerline:

p = .06

Chapter 18: Statistical Quality Control 28

UCL: .06 + 3

(. 06 )(. 94 ) 40 (. 06 )(. 94 ) 40

= .06 + .11265 = .17265

LCL:

.06 - 3

= .06 - .112658 = .00000

p Chart:

18.32 The process appears to be in control. Only 1 sample mean is beyond the outer limits (97% of the means are within the limits). There are no more than four means in a row on one side of the centerline. There are no more than five consecutive points decreasing or three consecutive points increasing. About 2/3 (67%) of the points are within the inner 1/3 of the confidence bands (+ 1 x ).

18.33 There are some items to be concerned about with this chart. Only one sample range is above the upper control limit. However, near the beginning of the chart there are eight sample ranges in a row below the centerline. Later in the run, there are nine sample ranges in a row above the centerline. The quality manager or operator might want to determine if there is some systematic reason why there is a string of ranges below the centerline and, perhaps more importantly, why there are a string of ranges above the centerline.

Chapter 18: Statistical Quality Control 29

18.34 This p chart reveals that two of the sixty samples (about 3%) produce proportions that are too large. Nine of the sixty samples (15%) produce proportions large enough to be greater than 1 p above the centerline. In general, this chart indicates a process that is under control.

18.35 The centerline of the c chart indicates that the process is averaging 0.74 nonconformances per part. Twenty-five of the fifty sampled items have zero nonconformances. None of the samples exceed the upper control limit for nonconformances. However, the upper control limit is 3.321 nonconformances whichnonconformances, which, in and of itself, may be too many. Indeed, three of the fifty (6%) samples actually had three nonconformances. An additional six samples (12%) had two nonconformances. One matter of concern may be that there is a run of ten samples in which nine of the samples exceed the centerline (samples 12 through 21). The question raised by this phenomenon is whether or not there is a systematic flaw in the process that produces strings of nonconforming items.