SPC Some common questions

ASQ Automotive Division Webinar Series

SPC Some common questionsMay 27 8PM EDT

Presenter: John Katona


SPC Some common questionsMay 27 8PM EDT

Agenda:

5 Min Introduction5 Min Introduction70 Min presentation 10 Min Q&A

ASQ Automotive Division

ASQ Automotive Division is part of the American Society for

Quality (ASQ), the world’s leading authority on quality issues

since 1946.

ASQ Automotive Division has more than 3400 members

globally. Members include professionals from almost every globally. Members include professionals from almost every

discipline in the vehicle manufacturing and supplier business in

the automotive, heavy-truck, off-highway, agricultural, industrial

and construction equipment industries.


VISION

• To be the worldwide leader on quality issues related to the automotive industry.

MISSION

• To provide member value by identifying, communicating, and • To provide member value by identifying, communicating, and promoting quality knowledge, professional development and networking opportunities.


OBJECTIVES:

• Be a global provider of automotive quality knowledge and learning opportunities for advancing individual and organizational performance excellence.

• Engage, grow and retain members through new and improved communities and cutting-edge technologies.

• Develop and sustain a strong Council Leadership to support our members.

• Develop and sustain a strong Council Leadership to support our members.

Statistical Process Control

Some Common Questions

John Katona

Page 6

John Katona

Secretary


Question 1My process has several sources of

variation:Can I put them all on (1) chart?

Can I put all (12) nests on the same chart

Page 7

Can I put all (12) nests on the same chart or do I need (12) charts?

My mold makes (32) parts in every shot. Can I just grab any (4) parts from a shot

and maintain only a single chart?

Spindle A

Spindle B

Dotplot of Spindle A, Spindle B, Spindle C, Spindle D, Spindle E

Page 8

31.831.230.630.029.428.828.2

Spindle C

Spindle D

Spindle E

Data

Each symbol represents up to 2 observations.

Spindle A

Spindle B


Can I put all (5) of these spindles on the same chart?

Page 9

31.831.230.630.029.428.828.2

Spindle C

Spindle D

Spindle E

Data


Spindle A

Spindle B


Can I put all (5) of these spindles on the same chart?

Yes – they all have about the same average & spread.

Page 10

31.831.230.630.029.428.828.2

Spindle C

Spindle D

Spindle E

Data


9181716151413121111

31.0

30.5

30.0

29.5

29.0

Sample

Sample M

ean

__X=30.007

UC L=30.914

LC L=29.101

1

Xbar-R Chart of Spindle A, ..., Spindle E

Each Subgroup contains all (5) Spindles –

Data is from (5) identical distributions, all normal with mean 30 and standard deviation 0.7

Page 11

Sample

9181716151413121111

4

3

2

1

0

Sample

Sample Range

_R=1.571

UC L=3.323

LC L=0

1

9181716151413121111

31.0

30.5

30.0

29.5

29.0

Sample

Sample M

ean

__X=30.038

UC L=30.916

LC L=29.161

1 2 3 4 5

Xbar-R Chart of Spindle Data by Stages

Each Subgroup contains only (1) Spindle –

Data is from (5) identical distributions, all normal with mean 30 and standard deviation 0.7

Page 12

Sample

9181716151413121111

4

3

2

1

0

Sample

Sample Range

_R=1.521

UC L=3.216

LC L=0

1 2 3 4 5

Fixture 1

Fixture 2

Dotplot of Fixture 1, Fixture 2, Fixture 3, Fixture 4, Fixture 5

Can I put all (5) of these fixtures on the same chart?

Page 13

4236302418126

Fixture 3

Fixture 4

Fixture 5

Data


Fixture 1

Fixture 2


Can I put all (5) of these fixtures on the same chart?

NO – they have averages that are very different.

Page 14

4236302418126

Fixture 3

Fixture 4

Fixture 5

Data


9181716151413121111

48

36

24

12

0

Sample

Sample M

ean

__X=26.87

UC L=49.16

LC L=4.59

Xbar-R Chart of Fixture 1, ..., Fixture 5

Each Subgroup contains all (5) Fixtures –

When it looks too good to be true, it is too good to be true

Page 15

Sample

9181716151413121111

80

60

40

20

0

Sample

Sample Range

_R=38.63

UC L=81.69

LC L=0

9181716151413121111

48

36

24

12

0

Sample

Sample M

ean

__X=26.87

UC L=49.16

LC L=4.59

Xbar-R Chart of Fixture 1, ..., Fixture 5

Each Subgroup contains all (5) Fixtures –

When it looks too good to be true, it is too good to be true

Page 16

Sample

9181716151413121111

80

60

40

20

0

Sample

Sample Range

_R=38.63

UC L=81.69

LC L=04236302418126

Fixture 1

Fixture 2

Fixture 3

Fixture 4

Fixture 5

Data



9181716151413121111

40

30

20

10

0

Sample

Sample M

ean

__X=26.87UC L=27.82LC L=25.93

11111111111111111111

11111111111111111111

11111111111111111111

11111111111111111111

11111111111111111111

Xbar-R Chart of Fixture Data

Each Subgroup contains only (1) Fixture –

The averages for each fixture are very different

Page 17

Sample

9181716151413121111

3

2

1

0

Sample

Sample Range

_R=1.633

UC L=3.454

LC L=0

9181716151413121111

40

30

20

10

0

Sample

Sample M

ean

__X=3.38UC L=4.29LC L=2.47

1 2 3 4 5

Xbar-R Chart of Fixture Data by Stages

Each Subgroup contains only (1) Fixture –

The averages for each fixture are very different

Page 18

Sample

9181716151413121111

4

3

2

1

0

Sample

Sample Range

_R=1.580

UC L=3.341

LC L=0

1 2 3 4 5

Question 1My process has several sources of variation:

Can I put them all on (1) chart?

Can I put all (12) nests on the same chart or do I need (12) charts?

Do all (12) nests have the same average & spread?

My mold makes (32) parts in every shot. Can I just grab any (4)

Page 19

My mold makes (32) parts in every shot. Can I just grab any (4) parts from a shot and maintain only a single chart?

Do all (32) cavities have the same average & spread?

Answer:If the averages and spreads are the same, then yes.

Otherwise NO.

I have an engineering specification.Why do I need statistical control

limits?

Question 2

Page 20

limits?

Can’t I just put the spec or 70% of the spec on the control chart?

Critical Distinctions• Specifications apply to the parts.

• Specifications tell if a part meets customer requirements

• Specifications do not apply to the

process that makes the parts.

• Control limits apply to the


• Control limits do not tell if the

parts meet customer

requirements.

• Control Limits tell when the

Page 21


Specifications do not tell if the

process has changed.

• A process where all parts are

within specifications may or may

not be “In Control” (Predictable)

• Control Limits tell when the

process has changed.

• A process that is “In Control” is

predictable.

• A process that is “In Control” may

or may not be making parts within

specifications

LSL USL

LSL 26

Target *

USL 34

Sample Mean 30.0075

Sample N 500

StDev (Within) 0.708499

StDev (O v erall) 0.688751

Process Data

C p 1.88

C PL 1.89

C PU 1.88

C pk 1.88

O v erall C apability

Potential (Within) C apability

Within

Overall

Process Capability of Spindle Data

Is this process “In control” or “predictable”?

Page 22

33.632.431.230.028.827.626.4

Pp 1.94

PPL 1.94

PPU 1.93

Ppk 1.93

C pm *

PPM < LSL 0.00

PPM > USL 0.00

PPM Total 0.00

O bserv ed Performance

PPM < LSL 0.01

PPM > USL 0.01

PPM Total 0.02

Exp. Within Performance

PPM < LSL 0.00

PPM > USL 0.00

PPM Total 0.01

Exp. O v erall Performance

LSL USL

LSL 26

Target *

USL 34

Sample Mean 30.0075

Sample N 500

StDev (Within) 0.708499

StDev (O v erall) 0.688751

Process Data

C p 1.88

C PL 1.89

C PU 1.88

C pk 1.88

Pp 1.94

O v erall C apability

Potential (Within) C apability

Within

Overall

Process Capability of Spindle Data

Is this process “In control” or “predictable”?

The Specifications and distribution shape don’t reveal

anything about process stability or predictability

from one time period to the next.

Is this process

Changing from (1)

Page 23

33.632.431.230.028.827.626.4

Pp 1.94

PPL 1.94

PPU 1.93

Ppk 1.93

C pm *

PPM < LSL 0.00

PPM > USL 0.00

PPM Total 0.00

O bserv ed Performance

PPM < LSL 0.01

PPM > USL 0.01

PPM Total 0.02

Exp. Within Performance

PPM < LSL 0.00

PPM > USL 0.00

PPM Total 0.01

Exp. O v erall Performance

Changing from (1)

time period to the next?

Without the

control chart you

don’t know.

9181716151413121111

31.0

30.5

30.0

29.5

29.0

Sample

Sample M

ean

__X=30.007

UC L=30.914

LC L=29.101

1


Page 24

Sample

9181716151413121111

4

3

2

1

0

Sample

Sample Range

_R=1.571

UC L=3.323

LC L=0

1 Is this process “predictable”?

9181716151413121111

31.0

30.5

30.0

29.5

29.0

Sample

Sample M

ean

__X=30.007

UC L=30.914

LC L=29.101

1


Is this process “predictable”?

Page 25

Sample

9181716151413121111

4

3

2

1

0

Sample

Sample Range

_R=1.571

UC L=3.323

LC L=0

1

Is this process “predictable”?

It looks pretty predictable.

9181716151413121111

31.0

30.5

30.0

29.5

29.0

Sample

Sample M

ean

__X=30.007

UC L=30.914

LC L=29.101

1


Are these parts “in Specification”?

Page 26

Sample

9181716151413121111

4

3

2

1

0

Sample

Sample Range

_R=1.571

UC L=3.323

LC L=0

1


9181716151413121111

31.0

30.5

30.0

29.5

29.0

Sample

Sample M

ean

__X=30.007

UC L=30.914

LC L=29.101

1



Page 27

Sample

9181716151413121111

4

3

2

1

0

Sample

Sample Range

_R=1.571

UC L=3.323

LC L=0

1


The Control Chart does not answer this question!

Question2 I have an engineering specification.

Why do I need statistical control limits?

1. Engineering spec. is for classifying parts as conforming or non

Page 28

parts as conforming or non conforming to Customer

Requirement, it does not signal process change.

2. Control limits signal process change. They do not classify parts as meeting

Customer Requirements.

I’m measuring “flatness” or “leak” or “roundness”.

Why do I have a Lower Control Limit? Shouldn’t it just be 0?

Page 29

I’m measuring “flatness” or “leak” or “roundness”.

Why do I have a Lower Control Limit? Shouldn’t it just be 0?

Page 30

This point is below the Lower Control Limit. This is unusual

compared to where the process ordinarily makes product.

Control Limits alert us to process changes and unusual events

I’m measuring “Weld Strength”.

Why do I have an Upper Control Limit?

Page 31

I’m measuring “Weld Strength”.

Why do I have an Upper Control Limit?

These point are above the Upper Control Limit. This is unusual

compared to where the process ordinarily makes product.


Page 32


Why are my control limits so narrow?Why would we control the process tighter than the specification??

Subgroup Size n=1

Page 33


Subgroup Size n=1

Page 34

Control limits tell us where the process ordinarily makes product.

Control limits are based on data from the process, not on the specification.

Control Limits alert us to process changes and unusual events.

A process that is very “Capable” will have control limits narrower than the specification.


Subgroup Size n=5


Page 35


Control limits are based on data from the process, not on the specification.

Control Limits alert us to process changes and unusual events.

A process that is very “Capable” will have control limits narrower than the specification.

Increasing the Subgroup Size will further “tighten” the control limits.

Why are my control limits so wide? We are allowing the process to vary way beyond the specifications!

Subgroup Size n=1

Page 36

Why are my control limits so wide?We are allowing the process to vary way beyond the specifications!

Subgroup Size n=1


Page 37

Control limits tell us where the process ordinarily makes product.Control limits are based on data from the process, not on the specification.

Control Limits alert us to process changes and unusual events.A process that is NOT “Capable” may have control limits wider than the specification.

This depends on Subgroup Size. With n=1, here the Control Limits are wider than the specification.

Why are my control limits so wide?We are allowing the process to vary way beyond the specifications!

Subgroup Size n=5


Page 38

Control limits tell us where the process ordinarily makes product.Control limits are based on data from the process, not on the specification.

Control Limits alert us to process changes and unusual events.A process that is NOT “Capable” may have control limits wider than the specification.

This depends on Subgroup Size. With n=5, the Control Limits are tighter than using n=1,but still wider than the Specification. Notice, that the process is still not “Capable”

Question 3

Cp, Cpk, Pp, Ppk???

What’s all this alphabet soup

about?

Page 39

Why are there (4) of these

indices??

Cp Pp

Process CAPABILITY (adjusted for targeting) Cpk can improve to Cp

if I can adjust my

Process PERFORMANCE (adjusted for targeting)

Ppk can improve to Pp if I

Page 40

Cpk Ppk

if I can adjust my process average so it is

in the middle of the specifications.

can adjust my process average so it is in the

middle of the specifications.

Cp Pp

Process PERFORMANCE Pp can improve to Cp if I can

stabilize my process on the Control Chart. (Even if I don’t re-target to the middle of the

specifications.)

Process PERFORMANCE

Page 41

Cpk Ppk

Process PERFORMANCE (adjusted for targeting)

Ppk can improve to Cpk if I can stabilize my process

on the Control Chart.

(Even if I don’t re-target to the middle of the specifications.)

Cp Ppk

Process PERFORMANCE (adjusted for “targeting)

Ppk can improve to Cp if I can stabilize my process on the Control Chart and

Page 42

Cp Ppkon the Control Chart and also re-target to the middle

of the specifications.

Content Application

Variation within

subgroups only

1. Short Term

Capability

2. Diagnostic use

___

Rσd2 = /d2

Page 43

σn−1= i=1

n

Σ (xi-X)2

n-1

Variation both

within & between

subgroups

2. Diagnostic use

1. Predicted

Performance

Cp=Total Tolerance

6σd2

The minimum Upper Specification - X

Capability Indices – Include Within Group Variation Only

“Cp shows how good Ppkcould be if the process were targeted

within the specifications and stable on the control chart”

Page 44

Cpk=The minimum

of either

3σd2

or

- Lower Specification

3σd2

Upper Specification - X

X

Cpk will be worse than Cp if the process is not centered within

the specifications.“Cpk shows how good Ppk could be if

the process was just stable on the control chart”

Performance Indices – Include Both Within Group &Between Group Variation

Pp=Total Tolerance

6σn−1

Pp will be worse than Cp if the process is unstable on the control chart

“Pp shows how good Ppk could be if the process was just targeted

within the specifications.”

Page 45

Ppk=The minimum

of either

3

or

- Lower Specification

3

Upper Specification - X

X

σn−1

σn−1

464136312621161161

0.834

0.831

0.828

Sample Mean

__X=0.829989

UCL=0.833553

LCL=0.826424

0.8400.8370.8340.8310.8280.8250.822

LSL USL

LSL 0.82

USL 0.84

Specifications

Process Capability Sixpack of M.830 s.0025

Xbar Chart

R Chart

Capability Histogram

Normal Prob PlotA D: 0.159, P: 0.950

Cp, Cpk, Pp, and Ppk are all virtually equal.How can that be?

Page 46

464136312621161161

0.016

0.008

0.000

Sample Range

_R=0.00618

UCL=0.01307

LCL=0

5045403530

0.835

0.830

0.825

Sample

Values

0.840.830.82

Within

O v erall

Specs

StDev 0.00265691

C p 1.25

C pk 1.25

Within

StDev 0.00262862

Pp 1.27

Ppk 1.27

C pm *

O v erall

1

Last 25 Subgroups

A D: 0.159, P: 0.950

Capability Plot

464136312621161161

0.834

0.831

0.828

Sample Mean

__X=0.829989

UCL=0.833553

LCL=0.826424

0.8400.8370.8340.8310.8280.8250.822

LSL USL

LSL 0.82

USL 0.84

Specifications


Xbar Chart

R Chart


Normal Prob PlotA D: 0.159, P: 0.950

Process average is targeted very close to the center of the specifications & points on both X-bar and Range charts indicate decent process stability.

Note that Cp, Cpk, Pp, and Ppk are all virtually equal.

Page 47

464136312621161161

0.016

0.008

0.000

Sample Range

_R=0.00618

UCL=0.01307

LCL=0

5045403530

0.835

0.830

0.825

Sample

Values

0.840.830.82

Within

O v erall

Specs

StDev 0.00265691

C p 1.25

C pk 1.25

Within

StDev 0.00262862

Pp 1.27

Ppk 1.27

C pm *

O v erall

1

Last 25 Subgroups

A D: 0.159, P: 0.950

Capability Plot

464136312621161161

0.8300

0.8275

0.8250

Sample Mean

__X=0.826120

UCL=0.829249

LCL=0.822991

0.8400.8370.8340.8310.8280.8250.822

LSL USL

LSL 0.82

USL 0.84

Specifications


Xbar Chart

R Chart


Normal Prob Plot

Process average is off-target from the center of the specifications & points on both X-bar and Range charts indicate decent process stability.

Note that Cp & Pp are virtually equal as are Cpk & Ppk; however Cpk & Ppk are degraded from Cp & Pp as the process is off-target..

Page 48

464136312621161161

0.010

0.005

0.000

Sample Range

_R=0.00542

UCL=0.01147

LCL=0

5045403530

0.830

0.825

0.820

Sample

Values

0.8350.8300.8250.820

Within

O v erall

Specs

StDev 0.00233212

C p 1.43

C pk 0.87

Within

StDev 0.00232063

Pp 1.44

Ppk 0.88

C pm *

O verall

R Chart

Last 25 Subgroups

Normal Prob PlotA D: 0.207, P : 0.866

Capability Plot

252321191715131197531

6.8

6.4

6.0

Sample Mean

__X=6.4553UCL=6.5050LCL=6.4055

7.257.006.756.506.256.005.75

LSL USL

LSL 5.95

USL 6.95

Specifications

11111

11

1

1

1

1

1

11

1

11111

1

1

1

1

1

Process Capability Sixpack of Data Stacked

Xbar Chart

R Chart


Normal Prob Plot

Question: Is this process “capable.”

Page 49

252321191715131197531

0.2

0.1

0.0

Sample Range

_R=0.0862

UCL=0.1823

LCL=0

252015105

7.0

6.5

6.0

Sample

Values

8765

Within

O v erall

Specs

StDev 0.0370682

C p 4.5

C pk 4.45

Within

StDev 0.36429

Pp 0.46

Ppk 0.45

C pm *

O verall

R Chart

Last 25 Subgroups

Normal Prob PlotA D: 16.396, P : < 0.005

Capability Plot

252321191715131197531

6.8

6.4

6.0

Sample Mean

__X=6.4553UCL=6.5050LCL=6.4055

7.257.006.756.506.256.005.75

LSL USL

LSL 5.95

USL 6.95

Specifications

11111

11

1

1

1

1

1

11

1

11111

1

1

1

1

1


Xbar Chart

R Chart


Normal Prob Plot

Question: Is this process “capable.”Yes it is “capable”, but not “performing”

Page 50

252321191715131197531

0.2

0.1

0.0

Sample Range

_R=0.0862

UCL=0.1823

LCL=0

252015105

7.0

6.5

6.0

Sample

Values

8765

Within

O v erall

Specs

StDev 0.0370682

C p 4.5

C pk 4.45

Within

StDev 0.36429

Pp 0.46

Ppk 0.45

C pm *

O verall

R Chart

Last 25 Subgroups


Capability Plot

252321191715131197531

6.8

6.4

6.0

Sample Mean

__X=6.4553UCL=6.5050LCL=6.4055

7.257.006.756.506.256.005.75

LSL USL

LSL 5.95

USL 6.95

Specifications

11111

11

1

1

1

1

1

11

1

11111

1

1

1

1

1


Xbar Chart

R Chart


Normal Prob Plot

Process average is targeted at the center of the specifications so Cp=Cpk & Pp=Ppk. However the X-bar chart is very unstable, so Pp<Cp and Ppk<Cpk.

Page 51

252321191715131197531

0.2

0.1

0.0

Sample Range

_R=0.0862

UCL=0.1823

LCL=0

252015105

7.0

6.5

6.0

Sample

Values

8765

Within

O v erall

Specs

StDev 0.0370682

C p 4.5

C pk 4.45

Within

StDev 0.36429

Pp 0.46

Ppk 0.45

C pm *

O verall

R Chart

Last 25 Subgroups


Capability Plot


Thank you for attending the Webinar

For video from this webinar please check our website:Video will be delayed by a few weeks:

www.asq-auto.org

Twitter: @ASQautomotive

New Website

Twitter: @ASQautomotiveFacebook: ASQ Automotive division

Slides are posted at: http://www.slideshare.net/oldeckQuestions & Comments: [email protected]

SPC Some common questions

Documents