10/16/2024

Determining Process Capability

Chapter 8 (sections 8.1-8.4 & 8.6)

Mahmoud Awad, Ph.D.

American University of Sharjah
INE311: Quality Engineering

Process Capability

Natural tolerance limits are defined as follows:

1
 10/16/2024

Process Capability Relative to VOC

LSL USL

Most capable

Defects Defects

Least capable

As the standard deviation increases …the defect levels increase

Stability before Capability

Thursday

Wednesday

Monday
Tuesday

2
 10/16/2024

Uses of process capability data:

3
 10/16/2024

At least
100
observatio
ns….

Chapter 8 Statistical Quality Control, 7th Edition by Douglas C. Montgomery. 8

Copyright (c) 2013 John Wiley & Sons, Inc.

4
 10/16/2024

Probability Plotting

Chapter 8 Statistical Quality Control, 7th Edition by Douglas C. Montgomery. 9

Copyright (c) 2013 John Wiley & Sons, Inc.

• The distribution may not be normal; other types of probability

plots can be useful in determining the appropriate distribution.

Statistical Quality Control, 7th Edition by Douglas C. Montgomery. 10

Copyright (c) 2013 John Wiley & Sons, Inc.

10

5
 10/16/2024

Statistical Quality Control, 7th Edition by Douglas C. Montgomery. 11

Copyright (c) 2013 John Wiley & Sons, Inc.

11

Statistical Quality Control, 7th Edition by Douglas C. Montgomery. 12

Copyright (c) 2013 John Wiley & Sons, Inc.

12

6
 10/16/2024

For the hard bake process:

Statistical Quality Control, 7th Edition by Douglas C. Montgomery. 13

Copyright (c) 2013 John Wiley & Sons, Inc.

13

One-Sided PCR

Statistical Quality Control, 7th Edition by Douglas C. Montgomery. 14

Copyright (c) 2013 John Wiley & Sons, Inc.

14

7
 10/16/2024

Interpretation of the PCR

Only Valid if process is normal, in statistical control, and centered

Statistical Quality Control, 7th Edition by Douglas C. Montgomery. 15
Copyright (c) 2013 John Wiley & Sons, Inc.

15

Assumptions for Interpretation of Numbers in

Table 8.2

• Violation of these assumptions can lead to big trouble in using the data in Table 8.2.

Statistical Quality Control, 7th Edition by Douglas C. Montgomery. 16

Copyright (c) 2013 John Wiley & Sons, Inc.

16

8
 10/16/2024

17

17

• Example 1: Two suppliers A and B are applying for a tender to supply a

part with specifications of 50 ± 12. Both suppliers provided the following
information, assuming both processes are in control, which supplier should
you select?

Supplier A Supplier B
μ= 50, σ = 2 μ= 53, σ = 2

18

18

9
 10/16/2024

A Measure of Actual Capability

19

19

Capability for Off-Center Process

• Cp does not take
process centering
into account
• It is a measure
of potential
capability, not
actual capability

20

20

10
 10/16/2024

Reasons for Poor Process Capability

Process may have

good potential
capability
How to improve capability?
21

21

More about Process centering, can we rely on Cpk only?

Example 2: Two suppliers C and D are applying for a tender to supply a part
with specifications of 50 ± 15. Both suppliers provided the following
information, assuming both processes are in control, which supplier should you
select?

Supplier Mean σ

C 50 5

D 57.5 2.5

Cpk depends inversely on σ, so a large value of Cpk does not really tell us anything
about the location within LSL, USL interval

𝑈𝑆𝐿 − 𝐿𝑆𝐿 𝑈𝑆𝐿 − 𝐿𝑆𝐿

𝐶𝑝𝑚 = =
6𝜏 6 𝜎 2 +(𝜇 − 𝑇)2 ) 22

22

11
 10/16/2024

• So what metric should we use?

– Calculate Cp first
– Calculate Cpk and compare with Cp
– If Cp=Cpk, then process is centered
– When comparing two processes, calculate Cpm
• A Cpm ≥ 1 implies that the mean lies within the middle
third of the specification band …. Prove it !!

23

23

24

24

12
 10/16/2024

25

25

26

26

13
 10/16/2024

27

27

Normality and Process Capability Ratios

• The assumption of normality is critical to the usual
interpretation of these ratios (such as Table 8.2)
• For non-normal data, options are
1. Transform non-normal data to normal
2. Modify the definitions of PCRs for general families of
distributions

𝑈𝑆𝐿 − 𝐿𝑆𝐿
𝐶𝑝 (𝑞) =
𝑥0.99865 − 𝑥0.00135

28

28

14
 10/16/2024

Option 1: Transform Data

LSL= 130 Ohms, USL = 440 Ohms

29

29

30

30

15
 10/16/2024

Option 2: Use general family of distributions

31

31

32

32

16
 10/16/2024

Distribution Analysis: Resistivity

Standard 95.0% Normal CI
Percent Percentile Error Lower Upper
0.1 95.8813 13.1362 73.3019 125.416
0.135 98.3802 13.1539 75.7004 127.855
1 119.207 13.0834 96.1349 147.817
2 128.840 12.9517 105.800 156.898
3 135.352 12.8401 112.386 163.009
4 140.466 12.7441 117.583 167.802
5 144.768 12.6599 121.965 171.834
6 148.534 12.5851 125.807 175.366
7 151.916 12.5181 129.259 178.543
8 155.009 12.4577 132.419 181.454
9 157.878 12.4031 135.347 184.159
10 160.565 12.3537 138.089 186.698
20 182.018 12.0758 159.824 207.294
30 199.244 12.1152 176.859 224.462
40 215.249 12.4672 192.150 241.126
50 231.370 13.1909 206.908 258.723
60 248.697 14.4045 222.008 278.594
70 268.675 16.3370 238.490 302.681
80 294.103 19.5121 258.242 334.944
90 333.398 25.6513 286.730 387.663 𝑈𝑆𝐿 − 𝐿𝑆𝐿
91 339.073 26.6380 290.684 395.516 𝐶𝑝 (𝑞) =
92 345.347 27.7546 295.017 404.263 𝑥0.99865 − 𝑥0.00135
93 352.379 29.0366 299.826 414.143
94 360.403 30.5366 305.258 425.510
95 369.777 32.3369 311.532 438.912
96 381.103 34.5765 319.018 455.271
97 395.503 37.5194 328.398 476.320
98 415.491 41.7673 341.189 505.975
99 449.066 49.2865 362.149 556.842
99.865 544.133 72.7532 418.693 707.156
99.9 558.315 76.4918 426.835 730.294
46

33

Process Capability Analysis using Control Charts

34

34

17
 10/16/2024

Since LSL = 200

35

35

36

36

18
 10/16/2024

37

37

38

38

19
 10/16/2024

39

39

40

40

20