Total Quality Management

Module : Process Capability Analysis

Dr Appaso M Gadade
Assistant Professor, MED
Email: appaso.gadade@thapar.edu
 Syllabus
1. Need and significance;
2. Process capability for variable data;
3. Process capability indices;
4. interpreting the indices;
5. Six Sigma Process Quality.
 References
1. Juran, J.M. and Gryna, F.M., Quality Planning & Analysis, 3rd Edition, McGraw Hill (2001).
2. Grant, E.L., Statistical Quality Control, 3rd Edition, McGraw Hill (2014).
3. Beckwith, T.G., Marangoni, R.D., and Lienhard, J.H., Mechanical Measurements, 6th Edition, Pearson
Education (2006).
 Need and Significance
 Measures that indicate the ability of the process to meet specifications;
these are, in some sense, measures of process performance.
 Process capability analysis should be conducted only when a process is in a state
of statistical control. A process is said to be in statistical control when the only
sources of variation in the system are common causes
 Process capability refers to the uniformity of the process
 The variability of critical- to- quality characteristics in the process is a measure of the uniformity of output
 Two ways of variability:
1. The natural or inherent variability in a critical- to- quality characteristics at a specified time: instantaneous
variability
2. variability in a critical- to- quality characteristics over time

 Estimate of process capability may be in the form of a probability distribution having a specified shape, center (mean),
and spread (standard deviation)
Need and Significance
 Need and Significance
1. Uniformity of output:

2. Maintained or improved quality.

3. Product and process design facilitated.

4. Assistance in vendor selection and control.

5. Reduction in total cost.

 Tolerance limit
 Natural tolerance limits, also known as process capability limits, are established or influenced by the process
itself. They represent the inherent variation in the quality characteristic of the individual items produced by a
process in control
 They are estimated based on the population of values or, more
typically, from large representative samples.

 Statistical tolerance limits are the limits of an interval that (with a given level of confidence γ) contains atleast a
specified proportion(1-α) of the population.
 These limits are found from sampling information.
 Is
Process
Stable
The Define
?

Quality Process

Improvement Select
Measures
Model
Collect &
Interpret
Data

Is Process Stable? Is No Investigate &

Process
Fix Special
Stable
Causes
Purpose: Determine the ?

stability of key Yes

measures of the
Is
product. Improve
Process
No
Process
Capable
Capability
?

Yes
Use SPC to
Maintain
Current
Process

6-8
 Is
Process
Capable Process Capable
?

A stable process that meets customer requirements.

Histogram
Control Chart

UCL

CL

LCL
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32
Run Order Lower Spec Upper Spec
Target

Capability assessments for unstable processes, may not be indicative of

how the process is actually performing.
8-9
Capability Assessment for Counting Measures Is
Process
Capable
“Order Entry Process” ?

Week # Errors
1 15
2 22 Control Chart
3 18 35
4 10 30
UCL=28.0
5 13 Number 25
6 9 of Errors
20
7 27
15 Avg=16.0
8 12
9 24 10
10 22
5 LCL=4.0
11 8
12 8 0
5 10 15 20
13 26 Week
14 16
15 20
16 10 Is this process adequate as is?
17 16
18 9 Should it be improved?
19 15
20 20
Total 320
8-10
 Is
Capability Assessment for Instrument Process
Capable
Measures ?

Histogram
Control Chart

UCL

CL

LCL
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32
Run Order Lower Spec Upper Spec
Target

Is this process adequate as is?

Should it be improved?
8-11
Measures of Process Capability Is
Process
Capable
USL - LSL
Cp = Specification Range =
?
True Process Range 6sc

Cp < 1.0  Process is not capable of meeting specs

Cp = 1.0  Process is marginally capable
Cp > 1.0  Process is capable of meeting specs
Problem: We are assuming the process has a target that is
in the center of the specification range, and that
the process is in fact centered on that target.

Distance from process average

Cpk = to closest specification limit = min (USL - x , x - LSL)
1 True Process Range 3sc
2
Note: a negative result is possible if the process average is outside specifications

Cpk < 1.0  Process is not capable of meeting specs

Cpk = 1.0  Process is marginally capable
Cpk > 1.0  Process is capable of meeting specs
Benefits: • Optimal values are attained by running exactly
between specs.
• Can (must) be used for 1-sided specifications
Warning: Capability assessments for unstable processes, may not be
indicative of how the process is actually performing. 8-12
 Is
Process
Capable
?
Process Capability

LSL USL

How much material is out of spec?

 In the short term?
 In the long term?

8-13
 Is
Process
Process Capability Ratios Capable
?

LSL USL

Voice of The Process

Voice of The Customer

8-14
 Is
Process
Capable
?
Process Capability - The Strategy
Centering –The Process Is On Target

Spread – Reduce The Variation

Defects Defects

LSL USL

8-15
Process Capability Ratios Is
Process
Capable
?

2 Key Metrics for Measuring Capability

USL  LSL
Cp 
6sc

X- LSL USL- X
Cpk  Min( , )
3sc 3sc

8-16
 Is
Process

Process Capability Ratios - Concept Capable

?

USL  LSL
Cp 
6sc

Total Tolerance
Cp 
Process Spread

8-17
Process Capability – Example 1 Is
Process
Capable
LSL USL ?

27 Inches 33 Inches

u = 30 Inches
s c= 1

USL  LSL
Cp  Sigma Level = 3Cp
6sc

Cp = __________ Sigma Level = __________

8-18
Process Capability – Example 2 Is
Process
Capable
LSL USL ?

29 Inches 31 Inches

u = 30 Inches
sc= 1

USL  LSL
Cp  Sigma Level = 3Cp
6sc

Cp = __________ Sigma Level = __________

8-19
Process Capability – Example 3 Is
Process
Capable
LSL USL ?

28 Inches 32 Inches

u = 30 Inches
s c = .333

USL  LSL
Cp  Sigma Level = 3Cp
6sc

Cp = __________ Sigma Level = __________

8-20
Process Capability – Example 4 Is
Process
Capable
?
LSL USL
28 Inches 32 Inches

u = 33 Inches
s c = .333

USL  LSL Sigma Level = 3Cp

Cp 
6sc

Cp = __________ Sigma Level = __________

8-21
Process Capability Ratios Is
Process
Capable
?

X- LSL USL- X
Cpk  Min( , )
3sc 3sc

X- LSL USL- X
CpL  CpU 
3sc 3sc

A metric to take into account process shift

8-22
Process Capability – Example 5 Is
Process
Capable
?
LSL USL
28 Inches 32 Inches

u = 33 Inches
s c = .333

CpL = ___________ CpU = ___________

Cpk = ____________

Recall Cp=2.0 8-23

CP & CPK Measure Short-term
I Chart for C1
Is
Process
Capable

Capability 39 ?

3.0SL=37.36

Individual Value
34
Is The Process
In Control ?
X=30.60
29
Is It Producing
Defects ?

24 -3.0SL=23.84

0 5 10 15 20 25 30 35
Observation Number

A Short-term Capability study covers a relatively

short period of time (days, weeks) generally
consisting of 30 to 50 data points. The actual
number depends on the subject under study.

8-24
Long Term Performance
I Chart for C3
Is
Process
Capable
?
50
3.0SL=47.12

Individual Value
40
Short term
Is The Process Capability
In Control ?
X=33.80
30
Is It Producing
Defects ?

20 -3.0SL=20.49

0 50 100
Observation Number

A long-term capability study covers a relatively long

period of time (weeks, months) generally consisting of
100-200 data points. Again, the actual amount depends
on the subject under study.

8-25
A Further Look at Capability Is
Process
Capable
?

Compare the estimates of the process deviations

from the short-term and long-term data

Descriptive Statistics

Variable N Mean StdDev

short term 30 30.6 2.23
long term 180 33.8 4.44

What is the difference between the short-term and the

long-term data?
What implication does this have in doing capability
studies?

8-26
The Dynamic Process Is
Process
Capable
?

LSL USL

Short-Term
Capability

Long-Term
Capability

Over time, a process tends to shift by approximately 1.5s

8-27
Measures of Process Performance Is
Process
Specification Range = USL - LSL Capable
Pp = ?
True Process Range 6ss

P p < 1.0  Process Performance is not meeting specs

P p = 1.0  Process Performance is marginally meeting specs
P p > 1.0  Process Performance is meeting specs

Problem: We are assuming the process has a target that is

in the center of the specification range, and that
the process is in fact centered on that target.

Distance from process average

to closest specification limit min (USL - x , x - LSL)
P pk = 1 =
True Process Range 3ss
2

Note: a negative result is possible if the process average is outside specifications

Ppk < 1.0  Process Performance is not meeting specs

Ppk = 1.0  Process Performance is marginally meeting specs
Ppk > 1.0  Process Performance is meeting specs

Benefits: • Optimal values are attained by running exactly

between specs.
• Can (must) be used for 1-sided specifications 8-28
Performance vs. Capability Is
Process
Capable
?

Days
CO2Sales Outstanding
Levels for 55 Timefor 55 Days
Points

14

13

CO2-Shrt
DSO
12

11

10

Index 10 20 30 40 50

These data show that the process, if well controlled can

perform much better than it currently is

8-29
 Is
Process
Capability vs. Performance Capable
?

Days
CO2Sales Outstanding
Levels for 55 Timefor 55 Days
Points

14

13

CO2-Shrt
DSO
12

11

10

Index 10 20 30 40 50

Capability: Only Process Performance:

random or short Total Variation including
term variability shifts and drifts
(Cp & Cpk) (Pp & Ppk)
8-30
Process Performance Ratios Is
Process
Capable
?

USL  LSL
P p 
6ss

X- LSL USL- X
P pk  Min( , )
3ss 3ss

The P-family of indices are computationally the same as

the C-family of ‘capability’ indices, but use the observed
long-term standard deviation.
8-31
How Does Process Capability Relate to Is
Process
Six Sigma? A goal of Six Sigma is to have: Capable
?

Cp = 2.0
Cpk = 1.5
Cpk is 1.5 to allow for a 1.5 sigma shift from target.
If the six-sigma goal is achieved, the 1.5sc drift can
occur with a low risk of producing material out of
specification.
Example: The target viscosity for a plastic is 0.76.
sc from the process is 0.01. The lower and upper
spec from the process are 0.70 and 0.82, respectively.
0.82 - 0.70 0.12
Cp    2.0
6x(0.01) 0.06
0.82 - 0.775 0.045
Cpk (target  1.5sc )    1.5
3(0.01) 0.03
LSL USL

0.70 0.71 0.72 0.73 0.74 0.75 0.76 0.77 0.78 0.79 0.80 0.81 0.82
T-1.5sc T T+1.5sc 8-32
Six Sigma Goals (Defect Level) Is
Process
Capable
?

The Quality goal of six sigma corresponds to 3.4ppm

defect level.

For attributes data, the ppm level is calculated by:

Number of Defects
x 1,000,000
Defect Opportunit ies

For variables data, the ppm level is calculated using

the normal table.

8-33
Minitab: Process Capability Analysis Is
Process
Capable
?

Process Capability Sixpack of X

I C har t C apability H istogr am
LSL USL
UCL=103.03
100 S pecifications

Individual Value
LS L 75
Is it in 90
_
X=89.66 U S L 105
control ?
80
LCL=76.30
1 3 5 7 9 11 13 15 17 19 21 23 25 76 80 84 88 92 96 100 104

M oving Range C har t Nor mal P r ob P lot

1
UCL=16.42
A D: 0.301, P : 0.554
16
Moving Range

Is it in 8 Is it
__
control ? MR=5.03
Normal ?
0 LCL=0
1 3 5 7 9 11 13 15 17 19 21 23 25 80 90 100

Last 2 5 O bser vations C apability P lot

100 Within Within O v erall
S tDev 4.45479 S tDev 4.66887
Values

90 Cp 1.12 Pp 1.07
O v erall
C pk 1.1 P pk 1.05 How does
C pm *
80 S pecs the
5 10 15 20 25
Observation process
variation
compare to
the
spec limits ?
8-34
 Is
WARNING!!! Process
Capable
?

Statistical Assumptions Made In Capability Studies

1. Data Comes From a Stable Process

 If not, work towards getting the process in
control
 Don’t despair, you can still make some
assumptions about your process in the mean
time

2. Data are Normally Distributed

 If not, transform it (ask the instructor)

If Items #1 and #2 aren’t met, results will be misleading

8-35
 Is

Minitab: Process Capability Analysis Process

Capable
?

 Minitab has many tools that

will help you in this area. We
will find Process Capability
under:
Stat>Quality Tools>
Capability Sixpack>
Normal

Open Line1.MTW in the

Mintab Datasets Folder
and go to the Capability
Sixpack

8-36
 Is

Minitab: Process Capability Analysis Process

Capable
?

Enter the information shown below and select OK

8-37
 Is
Minitab: Process Capability Analysis Process
Capable
?

Your output should look like this:

Process Capability Sixpack of X

I C har t C apability H istogr am
LSL USL
UCL=103.03
100 S pecifications

Individual Value
LS L 72
_
90 X=89.66 U S L 108

80
LCL=76.30
1 3 5 7 9 11 13 15 17 19 21 23 25 72 78 84 90 96 102 108

M oving Range C har t Nor mal P r ob P lot

1
UCL=16.42
A D: 0.301, P : 0.554
16
Moving Range

8 __
MR=5.03

0 LCL=0
1 3 5 7 9 11 13 15 17 19 21 23 25 80 90 100

Last 2 5 O bser vations C apability P lot

100 Within Within O v erall
S tDev 4.45479 S tDev 4.66887
Values

90 Cp 1.35 Pp 1.29
O v erall
C pk 1.32 P pk 1.26
C pm *
80 S pecs
5 10 15 20 25
Observation

8-38
Example - Capability Study Results Is
Process
Capable
?

 Notice that both Cp / Cpk and Pp / Ppk values are given

 The data shows that the Capability of the process (Cpk)

is 1.32 and the Entitlement (Cp) is 1.35
 This shows that the process is on target. Why?

 The data also shows the Performance of the process

(Ppk) is reasonably capable with a value of 1.26

 This indicates the process is in control. Why?

 But how good should a process get… Next Slide?

8-39
Process Performance and Capability Is
Process
Capable
?

 The Ppk can closely approach the Cp when

 The Customer specifications truly reflect customer
requirements
 The process in under statistical control
 The data approximate the normal distribution
 The process average is very close to target.

 The Cp is like a benchmark or entitlement

 Sigma capability is driven primarily by random error
(Common Cause Sources of Variation)

 We would like Ppk to be very close to a Cpk of 1.5

 This would be a process that is performing at a Six-Sigma
level

8-40
Test Method Capability Is
Process
Capable
?
Variation Variation can be broken down into two types, process and test. These
Types two types are listed and described below:
Variation Type Component(s)
Process The variation due to the:
• process itself
• raw materials
• Sampling ?

Test All other types of variation,

including:
• Sampling ?
• Sample preparation
• Between analysts
• Within analysts
• Test Equipment
• Environment
• Any other variation not part of
the process
NOTE: Test variation may not be just variation due only to test, it
encompasses all sources other than process. 8-41
 Is
Process
Total Variability Components Capable
?

s 2  s process  s test
2 2
total
Where: s2  s c2 (Assumes process is stable)
total
s 2  s S2 (For Unstable Processes)
total
s test
2 Can be estimated three ways:
1) Standard Reference Material – Repeated Sampling
2) Half-Blind Technique
3) Gauge R&R

s 2process (Estimated by subtracting test from total variability)

8-42
Repeated Sampling (SRM) Technique Is
Process
Capable
?

Advantages
Long-term estimate of test
variability
Easy to administrate
Can be used to assess accuracy
(known value)
Can be used to test experimental
bias (known value)
Can be used as a SPC monitor

Disadvantages
Only one level of the process is
studied
People get used to what value
"should be"

8-43
Half-Blind Technique Is
Process
Capable
?

Advantage
Long-term estimate of test
variability

Multiple levels of the

process are represented by
using different samples

Disadvantage
Hard to administrate
Cannot be used to test
accuracy

8-44
Gauge R&R Technique Is
Process
Capable
?
Advantages
Quick initial estimate of test
variability
Can be used to assess accuracy
(known value)
Can be used to test experimental
bias (known value)
Analyst portion can be estimated
Multiple process levels studied

Disadvantages
Hard to administrate

Short-term estimate of test

variability
Detailed experimental plan
needed.
Analysis more difficult
8-45
Measures of Test Method Capability Is
Process
Capable
?

Eastman has traditionally used:

 s 2

 test  100% Percent of variation due to test method
stotal 

Six Sigma uses the following measures:

s 
 test
 s

 100% Percent Gauge R&R

 total 

 5.15 *s 
 test  100% Precision to Tolerance Ratio
 Tolerance 
 
8-46
How Good is Good? … The Debate is On Is
Process
Capable
?

Eastman has traditionally used:

Percent Test Variability: Goal is 20 to 30%

Six Sigma uses the following measures:

Percent Gauge R&R: Less Than 30%

P/T Rate: Best Case 10%, Acceptable 30%

Problem: 30% Gauge R&R translates to 9% Percent

Test Variability (Unrealistic for a lot of tests in
Chemical Industry)
Note: Initial analysis is showing that processes can be monitored with SPC techniques even
when the percent test variability is as high as 50%! 8-47
Resources Allocation? Is
Process
Capable

Two-Dimensional Guidelines ?

Cp=0.5
100

90

Improving process variation will help, Cp=1.0

Poor 80
Test is primary
but test will still need to be improved
limitation on further
(BOTH)
process capability
70
Improvement
(TEST)
60
Cp=1.5
Work on
P/T Ratio

Focus on process variation first,

50 Test may become limiting after both process
Marginal

and test <6

some level of process variation l of
Improvement (BOTH) L eve Cp=2.0
40 ma
(PROCESS) < S ig
4.5
30
Focus on Process,
20 Test is Adequate
Adequate

(PROCESS) Sigma Level > 6

10

0
0 10 20 30 40 50 60 70 80 90 100
% Gage R&R
Adequate Marginal Poor 8-48
Review Is
Process
Capable
?
•Indices to measure processes capability
Cp – Goal is 2, Short Term Sigma used in Estimate

Cpk – Goal is 1.5, Short Term Sigma used in Estimate

Pp – Approach Cp, Long Term Sigma used in Estimate

Ppk – Approach Cpk, Long Term Sigma used in Estimate

Note: All of the above are equal for stable processes on target

•Indices to measure test method capability

Percent Test Variation – Goal is 20 to 30%

Percent Gauge R&R – Less than 30% (Percent Test < 9%)
50% may be more reasonable

P/T Ratio – Best Case 10%, Acceptable at 30%

8-49
 Dr Appaso M Gadade
Assistant Professor, MED
Email: appaso.gadade@thapar.edu