Process Capability Study

Stable process in
chart of averages
(centering)
Normal data
(P> 0.05)
Stable process in
Range chart
(dispersion)

Process capable
(Cpk and Ppk>
1.67)

© Copyright by Ricardo Espinosa. 1

Quality Core Tools

Process Capability Study

Rev k

Ing. Ricardo Espinosa Álvarez, CQE, CSQP, CRE, CSSBB, CMQ/OE, CPIM, CSCP
Black Belt Certificado por ASQ
Especialista Certificado en Core Tools por AIAG
Auditor Líder Certificado ISO9001:2015 por Exemplar Global
Auditor Certificado de Proveedores IATF 16949:2016 por AIAG
Auditor Certificado VDA 6.3
Email: [email protected]
www.kaizensigma.com
LinkedIn: Ricardo Espinosa
2

1
 Process Capability Study Methodology

© Copyright by Ricardo Espinosa. 3

Process Capability Study

Application of the Process Capability study
• The purpose of a Process Capability study, is to determine if the process is capable of consistently
meeting the customer's specifications, and to take corrective actions if necessary.

Requirements of a Process Capability study

• Stable process.
• There are no special causes. It is verified with the control chart of averages and ranges, averages
and sigmas or I-MR. This applies to indices Cp and Cpk. Stability is to ensure that the process
remains centered on the nominal product specification or process objective and that the process
variation remains within control limits. Indices Pp and Ppk they do not require the process to be
stable.
• Individual data normality.
• The individual data is normally distributed. Normality is a requirement for estimating defects in
parts per million ppm.
• Capable process.
• Cpk and Ppk meet customer requirements (1.67 for the automotive sector and 1.33 for the non-
automotive sector).

© Copyright by Ricardo Espinosa. 4

2
 Process Capability Study
The requirements of any Capability Study include the following:
• Choose the product quality characteristic to be analyzed and its specifications. The priority is the
special quality and safety characteristics of the product and those product characteristics that require
improvement in Six Sigma projects. These characteristics must be in the PFMEA, control plan and work
instructions.
• Verify the correct operation of the manufacturing equipment. Carry out the necessary preventive and
corrective maintenance.
• Validate work instructions and staff skills. For the set-up and operation of the manufacturing
equipment.
• Validate the measurement system. High variation in the measurement system can increase process
variation and will produce an unreliable estimate of process capability. Ensure the calibration of the
measurement equipment and make the necessary analysis of resolution, type 1, linearity, accuracy and
GR&R.

© Copyright by Ricardo Espinosa. 5

Process Capability Study

Procedure for conducting a Process Capability study.
• Equipment set up.
• Adjust the manufacturing equipment so that the average readings match the nominal specification
of the product characteristic.
• Collect information.
• Measure 125 consecutive pieces and analyze with Xbar-R chart with size 5 subgroups. Number the
pieces.
• If there are several sources of variation (equipment, molds, cavities), carry out a study for each of
them.
• If the process is low volume or the measurement is destructive, measure 30 consecutive pieces
and analyze with I-MR chart with subgroups of size 1.
• Do not adjust the process once the measurements have started (if adjusted, start the
measurements again and discard measurements taken prior to adjustment.). This is to evaluate
the natural variation of the process.
• Keep a record of the changes observed during the study (tooling changes, machine stoppages, tool
sharpening, Operator changes, measurement system calibrations, etc.) to analyze potential causes of
variation and eliminate them.

© Copyright by Ricardo Espinosa. 6

3
 Process Capability Study
Procedure for conducting a Process Capability study.
• Check the process stability.
• Ensure that the process is under statistical control. This is accomplished through control charts. In
the event that the process is out of control, the special cause variation must be identified and
eliminated through a root cause analysis.
• Verify that the individual data is normally distributed.
• Chart the individual data on a histogram and assess their normality with the Anderson-Darling test.
• If the data is not normal, eliminate special causes or use the non-normal data distribution that
corresponds to the data (Weibull, lognormal, exponential, etc.).
• Do not transform data.
• Determine process capability.
• In case the process is not capable, make improvements to the process.

© Copyright by Ricardo Espinosa. 7

Process Capability Study

Process Capability improvements:
• Center the process on the nominal specification or objective of the product characteristic. Generally
this involves adjusting the process parameters (pressure, voltage, speed, etc.).
• Eliminate variation due to special causes. Special causes are identified by the control chart and are
generally due to significant changes in the process (adjustments, equipment failures, etc.).
• Reduce variation due to common causes. Common causes occur in stable processes. To reduce this
type of variation, regression analysis or design of experiments are generally used to determine the
optimal levels of the input variables (pressure, velocity, temperature, etc.).
• Repeat the Capability study. Once the improvements have been made to have a stable, capable and
normal process, repeat the Capability study. Maintain fixed the control limits of the characteristic to be
controlled and implement the control chart in real time.
• Check the set-up and operation of the equipment. Implement control charts in the special
characteristics of the process (Temperature, pressure, etc.)

© Copyright by Ricardo Espinosa. 8

4
 Process Capability Study
Potential causes for non normal data:
• Errors in measurement or data entry
• Significant changes in the manufacturing process (Special causes)
• Machine failures
• Over-adjustments
• Lack of standardization in set up and operation of equipment.
• Data mixed (Different machines, molds, cavities, etc.)
• Incomplete data
• Nature of manufacturing process (GD&T characteristics, time related variables, etc.)
• Bad resolution of measurement system

© Copyright by Ricardo Espinosa. 9

Process Capability Study

Potential causes for non normal data:
• Errors in measurement or data entry
• Significant changes in the manufacturing process (Special causes)
• Machine failures
• Over-adjustments
• Lack of standardization in set up and operation of equipment.
Outlier

© Copyright by Ricardo Espinosa. 10

10

5
 Process Capability Study
Potential causes for non normal data:
• Bad resolution of measurement system

© Copyright by Ricardo Espinosa. 11

11

Process Capability Study

Potential causes for non normal data:
• Data mixed (Different machines, molds, cavities, etc.)

© Copyright by Ricardo Espinosa. 12

12

6
 Process Capability Study
Potential causes for non normal data:
• Incomplete data

© Copyright by Ricardo Espinosa. 13

13

Process Capability Study

• Potential causes for non normal data:
• Nature of manufacturing process (Characteristics from product GD&T)

© Copyright by Ricardo Espinosa. 14

14

7
 Process Capability Study
Potential causes for non normal data:
• Nature of manufacturing process (Time related variables)

© Copyright by Ricardo Espinosa. 15

15

Internal And Total Variation

© Copyright by Ricardo Espinosa. 16

16

8
 Internal And Total Variation
Total (Pp, Ppk, Cpm)
2
  
 x1  x 
 s
n

 
p
i n 1
Beneficio

Internal (Cp, Cpk)

𝑅
𝜎 =
𝑑

po
𝜎 =

em
𝑐

Ti
n A2 D3 D4 d2 c4
2 1,880 0 3,267 1,128 0.7979
3 1,023 0 2,574 1,693 0.8862
𝑀𝑅
𝜎 = 4 0.729 0 2,282 2,059 0.9213
1.128
5 0.577 0 2,114 2,326 0.9400
6 0.483 0 2.004 2,534 0.9515
7 0.419 0.076 1,924 2,704 0.9594
8 0.373 0.136 1,864 2,847 0.9650
9 0.337 0.184 1,816 2,970 0.9693
10 0.308 0.223 1,777 3,078 0.9727

© Copyright by Ricardo Espinosa. 17

17

Internal variation

Tim e : 8 12 4 8 12 4 8 12 4 8 12 4 8 12 4 8 12 4 8 12 4 8 12 4 8
Day: 1 1 1 2 2 2 3 3 3 4 4 4 5 5 5 6 6 6 7 7 7 8 8 8 9
S ubgro up 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

1 11.811 11.809 11.820 11.820 11.814 11.821 11.810 11.817 11.816 11.819 11.840 11.813 11.799 11.811 11.806 11.816 11.815 11.813 11.811 11.819 11.807 11.820 11.818 11.806 11.808

2
Readings

11.800 11.821 11.806 11.810 11.813 11.809 11.821 11.819 11.813 11.821 11.830 11.811 11.817 11.817 11.812 11.807 11.817 11.821 11.816 11.821 11.818 11.810 11.819 11.812 11.817

3 11.799 11.811 11.822 11.816 11.808 11.820 11.812 11.800 11.820 11.812 11.850 11.820 11.811 11.812 11.814 11.801 11.807 11.807 11.809 11.815 11.809 11.813 11.808 11.818 11.806

4 11.819 11.798 11.812 11.814 11.816 11.819 11.812 11.818 11.812 11.798 11.830 11.805 11.809 11.818 11.822 11.810 11.817 11.810 11.813 11.798 11.816 11.811 11.815 11.815 11.819

5 11.821 11.820 11.815 11.816 11.813 11.812 11.820 11.815 11.812 11.815 11.800 11.793 11.813 11.804 11.814 11.821 11.815 11.809 11.809 11.819 11.806 11.812 11.817 11.815 11.812

A v e ra g e 11.8 10 11.8 12 11.8 15 11.8 15 11.8 13 11.8 16 11.8 15 11.8 14 11.8 15 11.8 13 11.8 3 0 11.8 0 8 11.8 10 11.8 12 11.8 14 11.8 11 11.8 14 11.8 12 11.8 12 11.8 14 11.8 11 11.8 13 11.8 15 11.8 13 11.8 12

R a ng e 0 .0 2 2 0 .0 2 3 0 .0 16 0 .0 10 0 .0 0 8 0 .0 12 0 .0 11 0 .0 19 0 .0 0 8 0 .0 2 3 0 .0 50 0 .0 2 7 0 .0 18 0 .0 14 0 .0 16 0 .0 2 0 0 .0 10 0 .0 14 0 .0 0 7 0 .0 2 3 0 .0 12 0 .0 10 0 .0 11 0 .0 12 0 .0 13

n A2 D3 D4 d2 c4
𝑅
2 1,880 0 3,267 1,128 0.7979 𝜎 =
3 1,023 0 2,574 1,693 0.8862
𝑑
4 0.729 0 2,282 2,059 0.9213
Beneficio

5 0.577 0 2,114 2,326 0.9400 𝑅 0.01504

𝜎 = =
6 0.483 0 2.004 2,534 0.9515 𝑑 2.326
7 0.419 0.076 1,924 2,704 0.9594
8 0.373 0.136 1,864 2,847 0.9650
po
em

9 0.337 0.184 1,816 2,970 0.9693

Ti

10 0.308 0.223 1,777 3,078 0.9727

Internal sigma = 0.006466

© Copyright by Ricardo Espinosa. 18

18

9
 Total variation

Tim e : 8 12 4 8 12 4 8 12 4 8 12 4 8 12 4 8 12 4 8 12 4 8 12 4 8
Day: 1 1 1 2 2 2 3 3 3 4 4 4 5 5 5 6 6 6 7 7 7 8 8 8 9
S ubgro up 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

1 11.811 11.809 11.820 11.820 11.814 11.821 11.810 11.817 11.816 11.819 11.840 11.813 11.799 11.811 11.806 11.816 11.815 11.813 11.811 11.819 11.807 11.820 11.818 11.806 11.808

2
Readings

11.800 11.821 11.806 11.810 11.813 11.809 11.821 11.819 11.813 11.821 11.830 11.811 11.817 11.817 11.812 11.807 11.817 11.821 11.816 11.821 11.818 11.810 11.819 11.812 11.817

3 11.799 11.811 11.822 11.816 11.808 11.820 11.812 11.800 11.820 11.812 11.850 11.820 11.811 11.812 11.814 11.801 11.807 11.807 11.809 11.815 11.809 11.813 11.808 11.818 11.806

4 11.819 11.798 11.812 11.814 11.816 11.819 11.812 11.818 11.812 11.798 11.830 11.805 11.809 11.818 11.822 11.810 11.817 11.810 11.813 11.798 11.816 11.811 11.815 11.815 11.819

5 11.821 11.820 11.815 11.816 11.813 11.812 11.820 11.815 11.812 11.815 11.800 11.793 11.813 11.804 11.814 11.821 11.815 11.809 11.809 11.819 11.806 11.812 11.817 11.815 11.812

A v e ra g e 11.8 10 11.8 12 11.8 15 11.8 15 11.8 13 11.8 16 11.8 15 11.8 14 11.8 15 11.8 13 11.8 3 0 11.8 0 8 11.8 10 11.8 12 11.8 14 11.8 11 11.8 14 11.8 12 11.8 12 11.8 14 11.8 11 11.8 13 11.8 15 11.8 13 11.8 12

R a ng e 0 .0 2 2 0 .0 2 3 0 .0 16 0 .0 10 0 .0 0 8 0 .0 12 0 .0 11 0 .0 19 0 .0 0 8 0 .0 2 3 0 .0 50 0 .0 2 7 0 .0 18 0 .0 14 0 .0 16 0 .0 2 0 0 .0 10 0 .0 14 0 .0 0 7 0 .0 2 3 0 .0 12 0 .0 10 0 .0 11 0 .0 12 0 .0 13

 2
 
 x1  x 
 s
n

 
p
i n 1 Beneficio

Total sigma = 0.006150

Calculation with Excel:

po
em
Ti
DEVEST.M
STDEV.S

© Copyright by Ricardo Espinosa. 19

19

Calculation Of Capability And Performance Indices

© Copyright by Ricardo Espinosa. 20

20

10
 Specification Limits
Tolerance =
USL-LSL
Nominal

The Specification Limits

apply to individual
measurements not
averages

LSL USL
When you have both upper (USL) and
Nominal is
lower specification (LSL) limits, this is the
better
most commonly used characteristic type.

© Copyright by Ricardo Espinosa. 21

21

Specification Limits

The Specification Limits

apply to individual
measurements not
averages

LSL
When there is only one specification limit,
Maximize
the LSL.

© Copyright by Ricardo Espinosa. 22

22

11
 Specification Limits

The Specification Limits

apply to individual
measurements not
averages

LSL USL
When there is only one specification limit,
Minimize
the USL

© Copyright by Ricardo Espinosa. 23

23

Process capability
• Cpk -The Capability index for a stable process. The estimation of Sigma is based on the internal
variation (R-bar / d2 or S-bar / c4

𝑅 𝑀𝑅
𝜎 = 𝜎 =
𝑑 1.128

𝑠
𝜎 =
𝑐

• Ppk- The performance index. The estimation of sigma is based on the total variation.

2
  
 x1  x 
 s
n

 
p
i n 1

Ref: AIAG PPAP Reference Manual Fourth Edition © Copyright by Ricardo Espinosa. 24

24

12
 Process capability
• Process Capability is determined by variation that comes from common causes
• To be acceptable, a process must be in a state of statistical control and the capability (common cause
variation) must be less than tolerance.

LSL USL

Measurements
Individual

6

© Copyright by Ricardo Espinosa. 25

25

Process Capability (Internal Variation)

LSL USL
The Cp index reflects the
Cp = 1 potential of the process if the
average were perfectly centered
between the specification limits

The larger the index

Cp, better!
Cp > 1
𝑈𝑆𝐿 − 𝐿𝑆𝐿 Voice of the Customer
𝐶 =
6𝜎 Voice of the Process

Cp <1

© Copyright by Ricardo Espinosa. 26

26

13
 Process Capability
Process Capability (Cp)

Calculation of Cp

𝑈𝑆𝐿 − 𝐿𝑆𝐿
𝐶 =
6𝜎
Mean

Cpk calculation

X−L𝑆𝐿 USL−X
Cpk = Min( , )
3𝜎 3𝜎

© Copyright by Ricardo Espinosa. 27

27

Process Capability Indices

Variation within subgroups (Internal)
𝑠 𝑀𝑅
𝑅 𝜎 = 𝜎 =
𝜎 = 𝑐 1.128
𝑑

𝑈𝑆𝐿 − 𝐿𝑆𝐿 n A2 D3 D4 d2 c4
C =
6𝜎 2 1,880 0 3,267 1,128 0.7979

3 1,023 0 2,574 1,693 0.8862

4 0.729 0 2,282 2,059 0.9213

USL−X
CPU = 5 0.577 0 2,114 2,326 0.9400
3𝜎 6 0.483 0 2.004 2,534 0.9515

7 0.419 0.076 1,924 2,704 0.9594

X−L𝑆𝐿 8 0.373 0.136 1,864 2,847 0.9650

CPL = 9 0.337 0.184 1,816 2,970 0.9693
3𝜎
10 0.308 0.223 1,777 3,078 0.9727

Cpk = Min(CPU, CPL)

© Copyright by Ricardo Espinosa. 28

28

14
 Process Performance Indices
2
 
 x1  x  Total process variation (Total)
n 
 p s 
i n 1

𝑈𝑆𝐿 − 𝐿𝑆𝐿 Taguchi index

P =
6𝜎
𝑈𝑆𝐿 − 𝐿𝑆𝐿
Cpm =
USL−X 6 𝜎 + (𝑇 − 𝑥 )
PPU =
3𝜎

T = Nominal
X−L𝑆𝐿
PPL =
3𝜎

Ppk = Min(PPU, PPL)

© Copyright by Ricardo Espinosa. 29

29

Xbar-R Process Capability Study with Minitab

© Copyright by Ricardo Espinosa. 30

30

15
 Xbar-R Process Capability

Stat> Quality Tools> Capability Sixpack> Normal

© Copyright by Ricardo Espinosa. 31

31

Xbar-R Process Capability

© Copyright by Ricardo Espinosa. 32

32

16
 Process Capability Indices

𝑅 0.418
𝜎 = = = 0.1799 Variation within subgroups (internal)
𝑑 2.326

. .
𝐶 = = = 1.85
∗ .

USL−X 23.5 − 22.5308

CPU = = = 1.80
3𝜎 3 ∗ 0.1799

X−𝐿𝑆𝐿 22.5308 − 21.5

CPL = = = 1.91
3𝜎 3 ∗ 0.1799

Cpk = Min(CPU, CPL) = 1.8

© Copyright by Ricardo Espinosa. 33

33

Process Performance Indices

2
 
 x1  x 
 p s   
n
=0.1887 Total variation (Overall) 𝑈𝑆𝐿 − 𝐿𝑆𝐿
i n1 Cpm =
6 𝜎 + (𝑇 − 𝑥)
. .
P = = = 1.77 . .
∗ . Cpm = =1.75
, ( . . )

USL−X 23.5 − 22.5308

PPU = = = 1.71
3𝜎 3 ∗ 0.1887

X−𝐿𝑆𝐿 22.5308 − 21.5

PPL = = = 1.82
3𝜎 3 ∗ 0.1887

Ppk = Min(PPU, PPL) = 1.71

© Copyright by Ricardo Espinosa. 34

34

17
 I-MR Process Capability Study
with Minitab

© Copyright by Ricardo Espinosa. 35

35

Process Capability I-MR

Stat> Quality Tools> Capability Sixpack> Normal

© Copyright by Ricardo Espinosa. 36

36

18
 Process Capability I-MR

© Copyright by Ricardo Espinosa. 37

37

Process Capability Indices

𝑅 1.272
𝜎 = = = 1.127 Variation within subgroups (internal)
𝑑 1.128

𝐶 = = = 1.48
∗ .

.
CPU = = =1.43
∗ .

−𝐿 𝑆 𝐿 .
CPL = = =1.53
∗ .

Cpk = Min(CPU, CPL) = 1.43

© Copyright by Ricardo Espinosa. 38

38

19
 Process Performance Indices
2
 
 x1  x 
 p s   
n
=1.280 Total variation (Overall) 𝑈𝑆𝐿 − 𝐿𝑆𝐿
i n1 Cpm =
6 𝜎 + (𝑇 − 𝑥)
P = = = 1.30 . .
∗ . Cpm = =1.75
, ( . . )

USL−X 40 − 35.172
PPU = = = 1.26
3𝜎 3 ∗ 1.280

X−𝐿𝑆𝐿 35.172 − 30
PPL = = = 1.35
3𝜎 3 ∗ 1.280

Ppk = Min(PPU, PPL) = 1.26

© Copyright by Ricardo Espinosa. 39

39

• Questions?

• Thanks a lot

© Copyright by Ricardo Espinosa. 40

40

20