11 1 Name Work/educational experience Background/classes taken in math, quality, continuous improvement, statistics, SPC, designed experiments

111

Name Work/educational experience Background/classes taken in

math, quality, continuous improvement, statistics, SPC, designed experiments

Expectations

Introduction

222

Chapter 1

333

What is quality?

Quality = Performance

Expectation

Fitness for use

Conformance to specifications

Producing the best results

Total customer satisfaction

Exceeding customer

expectations

Excellent products or services

444

Kano Model

The Kano model relates three factors to their degree of implementation or level of implementation, as shown in the diagram. 1) Basic ("must be") factors2) Performance ("more is better") factors3) Delighter ("excitement") factors.

The degree of customer satisfaction ranges from disgust, through neutrality, to delight.

555

Quality

Cost

Delivery

Responsiveness

Safety

… what about other factors?

666

1. Each team will be handed 20 cards.

2. Each team will have three operators, each of whom will drop one card at a time onto a target area.

3. The method of drop will be to hold the card at arms length over the target area or not. Only those cards that fall completely within the target area may move on.

4. The goal is to deliver 20 completed products or units to the customer.

5. Metrics-1. # of good units per station (A)2. # of cards used per station (B)3. Total time of exercise (C)4. Total # of defects (D)

Exercise: Why control a process?

Station 1 Station 2 Station 3 CustomerStart

777

Exercise cont…

A = # of good units B = # of cards

A1= B1=

A2= B2=

A3= B3=

FPY: Y1 = A1/B1 =

Y2 = A2/B2 =

Y3 = A3/B3 =

RTY = Y1 * Y2 * Y3 =

Total Cost = ($10 * D) + ($2 *[B1+B2+B3])=

Average cost per unit = Total cost / 20 =

Average cycle time = C / 20 =

888

RTY chart

99-100%90-99%0-90%

# of Steps +/- 3 Sigma +/- 4 Sigma +/- 5 Sigma +/- 6 Sigma1 93.32% 99.379% 99.9767% 99.99966%7 61.63% 95.733% 99.837% 99.998%

10 50.09% 93.961% 99.767% 99.997%20 25.09% 88.286% 99.535% 99.993%40 6.29% 77.944% 99.072% 99.986%60 1.58% 68.814% 98.612% 99.980%80 0.40% 60.753% 98.153% 99.973%

100 0.10% 53.637% 97.697% 99.966%150 0.00% 39.282% 96.565% 99.949%200 0.00% 28.769% 95.446% 99.932%300 0.00% 15.431% 93.248% 99.898%400 0.00% 8.277% 91.100% 99.864%500 0.00% 4.439% 89.002% 99.830%600 0.00% 2.381% 86.952% 99.796%700 0.00% 1.277% 84.949% 99.762%800 0.00% 0.685% 82.992% 99.728%900 0.00% 0.367% 81.081% 99.694%

1,000 0.00% 0.197% 79.213% 99.661%1,200 0.00% 0.057% 75.606% 99.593%3,000 0.00% 0.000% 49.704% 98.985%

17,000 0.00% 0.000% 1.904% 94.384%38,000 0.00% 0.000% 0.014% 87.880%70,000 0.00% 0.000% 0.000% 78.820%

150,000 0.00% 0.000% 0.000% 60.050%

(Distribution shifted +/- 1.5 Sigma)Overall Yield vs. Sigma

a.k.a. LeanSigma

999

Interesting Quote

“Quality control by statistical methods is now so extensively applied in all lines of industry, and in all sections of the United States, that everyone who is interested in

manufacturing should also have a definite interest in the methods.”

-Control Charts, E.S. Smith - 1947

101010

First, it is nothing new. Developed in the 1920’s.

Description: Involves the use of statistical signals to identify sources of variation, to maintain or improve performance to a higher quality level

What is SPC?

Process Control Statistical

111111

Quality is a must Detection mentality is out the

door Must build quality in Quality is a part of all job

functions

Why SPC?

121212

Training Quality

Planning Design

Review Quality

system audits

Continuous improvement

Technical data review

Process validation

Marketing research

Customer surveys

Field trials Supplier

quality planning

SPC Process

Control

The Cost of QualityPreventative Costs:

131313

The use of statistical signals to maintain or improve the process

The Prevention Model

Process Shipment

Figure 1.3

Output

Inspect with SPCAnalyze

Continueor Improve

141414

Let’s build a prevention model for taking a college course.

Exercise: Prevention Model

__________________

__________________

Output

Inspect with SPCAnalyzeImprove

______________

______________

______________

151515

Purchasing Appraisal Costs

Receiving/Incoming Inspection and Test

Measurement Equipment

Qualification of Supplier Product

Source Inspection and Control Programs

Manufacturing Appraisal Costs

Planned Inspections, Tests, Audits

Checking Labor Product or Service

Quality Audits Review of Test and

Inspection data Other Quality

Evaluations Laboratory Support

Inspection and Test Materials

Set-up Inspections and Tests

Depreciation Allowances

Measurement Equipment Expense

Maintenance and Calibration Labor

Outside Endorsements and Certifications

External Appraisal Costs

Field Performance Evaluations

Special Product Evaluations

Evaluations of Field Stock and Spare Parts

Process Control Measurement

The Cost of Poor QualityAppraisal Costs:

161616

An attempt to achieve quality by inspecting the quality into the product through 100% inspection

The Detection Model

Process Inspection Shipment

Repair/ Rework

Scrap or wasteFigure 1.2

What happens when your process improves?

171717

Exercise: Detection Model

______________

______________

______________

______________

______________

Let’s build a detection model for taking a college course.

181818

Appraisal exercise 

The defects in this story appear as the letter “S.”How many defects or S’s (capital or lower case) can

you detect in the story below?     

Bubba-Gump Shrimp Company

 It was a simple test to sample the number of shrimp secured by each of the fishing vessels after passing the second level of serious FDA inspection techniques. Susan was the first of several student inspectors to have the occasion to assay the new sampling system. Susan first separated seventy random sized shrimp from the sample received from the FDA. Those seventy were then weighed on a special scale. Susan then posted this weight on a job log. The sample of seventy shrimp is then returned to the same FDA sample. The weight of the seventy shrimp is then sent to the large sample scale were the first sample of shrimp has been assembled. The sample from the fishing vessel is then weighed and a series of automatic calculations determines the best number of shrimp in the sample from the comparison of Susan’s seventy shrimp sample.

191919

Reliability on visual inspection methods?

1. How many S's did you find in the story after reading it only once? ___________________________________________________  2. How many S's did you find in the story after reading it through the second time? ___________________________________________________ 3. What was your range from your first to second reading? ___________________________________________________ 4. How is this similar to “real life” inspection systems?______________________________________________________________________________________________________ 5. What is the most cost-effective way to not have non-conformances pass through the system?_____________________________________________________________________________________________________

See example 1.1, page 6, in the text book

202020

This is so interesting!.......can you read this note?

I cdnuolt blveiee taht I cluod aulaclty

uesdnatnrd waht I was rdgnieg.

THE PAOMNNEHAL PWEOR OF THE HMUAN MNID

Aoccdrnig to rscheearch at Cmabrigde Uinervtisy, it deosn't mttaer in waht oredr the ltteers in a word are, the olny iprmoatnt tihng is taht the frist and lsat ltteer be in the rghit pclae. The rset can be a taotl mses and you can sitll raed it wouthit a porbelm. Tihs is bcuseae the huamn mnid deos not raed ervey lteter by istlef, but the wrod as a wlohe. Amzanig huh?

Something weird I thought you might like....

212121

The Cost of Poor QualityFailure Costs (internal or external):

• Scrap• Rework• Customer Complaint Investigation• Returned Goods• Retrofit Costs• Recall Costs• Warranty Claims• Liability Costs• Penalties• Customer/User Goodwill • Other Failure Costs

222222

Minimize cost Attain a consistent process Allow everyone to contribute to

process improvement Help to make economical

decisions

Goals of SPC

232323

Use the seven basic tools◦ Flowchart◦ Pareto chart◦ Checksheet◦ Cause-and-effect diagram◦ Histogram◦ Control Chart◦ Scatterplot

How?

242424

1. Analyze where SPC should be done

2. Work on decreasing any obvious variability

3. Gage R&R4. Make sampling plan5. Create control chart – allow

only common cause variability6. Run the process7. Calculate process capability8. Improve if necessary or control

process9. Pre-control10. Continue to improve

How to apply SPC to a process

252525

DOE – Design of Experiments or referred as Designed Experiments

A systematic change to process variables to find the best combination to produce a quality product

What is DOE?

Process

Input Output

262626

Will help to gain knowledge in:

Improving process performance Reducing costs Understanding relationships

between variables Understanding how to optimize

processes

Why use DOE ?

272727

Let’s start with an example:

Data

18 16 30 29 28 21 17 41 8 1732 26 16 24 27 17 17 33 19 1831 27 23 38 33 14 13 26 11 2821 19 25 22 17 12 21 21 25 2623 20 22 19 21 14 45 15 24 34

Fuel Economy of 50 automobiles (in mpg)

282828

Histogram of MPG:Fuel Economy

0

2

4

6

8

10

12

14

16

0 to <6 6 to <12 12 to <18 18 to <24 24 to <30 30 to <36 36 to <42 42 to <48 48 to <54 54 to <=60

mgp

Nu

mb

er

of

Ca

rs

What causes variation in fuel economy?

DOE is about discovering and quantifying the magnitude of cause and effect relationships.

We need DOE because intuition can be misleading.

292929

What about other factors - and noise?

303030

To explain how we can model data experimentally, let’s take another look at the mileage data and see if there’s a factor that might explain some of the variation.

Draw a scatter diagram for the following data:

Let’s talk about regression

313131

Mileage data with vehicle weight:

The variable called ‘weight’ is known as a ‘factor’ and is plotted on the x-axis.

The variable called ‘mileage’ is known as the response, and is plotted on the y-axis. It’s sometimes called ‘Y’.

Weight (lbs) Mileage(mpg)3000 182800 212100 322900 172400 313300 142700 213500 122500 233200 14

Observation12345

10

6789

323232

Scatter Diagram Form:

Mile

age

Vehicle Weight (lbs)

333333

Yours might look something like:

Scatter Chart (Weight vs mpg)

05

101520253035

1900 2400 2900 3400 3900

Weight

mpg

343434

If you draw a best fit line and figure out an equation for that line, you would have a ‘model’ that represents the data.

Regression analysis

Scatter Chart (Weight vs mpg)

y = -0.0152x + 63.507

R2 = 0.9191

05

101520253035

1900 2400 2900 3400 3900

Weight

mp

g

353535

Looking at correlation from a Scatter diagram:

‘Correlation’ is a fancy word for how well the model predicts the response from the factors.

363636

Is there really an effect?

373737

There are basically two ways to understand a process you are working on.

One factor at a time (OFAT) DOE

Each have their advantages and disadvantages. We’ll talk about each.

Understanding a system

383838

To illustrate the need for experimental design, let’s consider how two known (based on years of experience) factors affect gas mileage, tire size (T) and fuel type (F).

Why DOE an OFAT example

Fuel Type Tire size

F1 T1

F2 T2

393939

Step 1:Select two levels of tire size and two kinds of fuels.

Step 2: Holding fuel type constant, test the car at both tire sizes.

One–at–a-time design

Fuel Type

Tire size Mpg

F1 T1 20

F1 T2 30

404040

Since we want to maximize mpg the more desirable response happened with T2.

Step 3: Holding tire size at T2, test the car at both fuel types.

One–at–a-time design

Fuel Type

Tire size Mpg

F1 T2 30

F2 T2 40

414141

At first glance the ideal setting looks like F2 and T2 at 40mpg.

However this experimental method did not test the interaction effect of tire size and fuel type.

One–at–a-time design

Fuel Type

Tire size Mpg

F1 T2 30

F2 T2 40

424242

Suppose that the untested combination F2T1 would produce the results below.

There is a different slope so there appears to be an interaction. A more appropriate design would be to test all four combinations.

One–at–a-time design

0

10

20

30

40

50

60

70

T1 T2

Tire Size

mpg

F2

F1

434343

What about the other factors - and noise?

444444

We need a way to investigate the relationship(s) between variables

We need to distinguish the effects of variables from each other (and maybe tell if they interact with each other)

We need to be able to quantify the effects...

...so we can predict, control, and optimize processes.

What we need to answer

454545

Objectives of an Experimental Design Obtain the

maximum amount of information using a minimum amount of resources

Determine which factors shift the average response, which shift the variability and which have no effect

Build empirical models relating the response of interest to the input factors

464646

So how do we do it?

PLANNING

DESIGN

ANALYSIS

CONFIRMATION

474747

DOE uses purposeful changes of the inputs (factors) in order to observe corresponding changes in the output (response).

Use an IPO – they are real important here.

DOE to the rescue!!

484848

Here’s what one looks like:

Run X1 X2 X3 X4 Y1 Y2 Y3 Y-bar SY

1 - - - -2 - - + +3 - + - +4 - + + -5 + - - +6 + - + -7 + + - -8 + + + +

494949

To ‘design’ an experiment, means to pick the points that you’ll use for a scatter diagram.

The basics

Run A B

1 - -

2 - +

3 + -

4 + +

In tabular form, it would look like:

High (+)

Low (-)

Fa

cto

r B

Se

ttin

gs

Factor A Settings High (+)Low (-)

(-,+)

(+,-)

(+,+)

(-,-)

505050

The difference in the average Y when A was ‘high’ from the average Y when A was ‘low’ is the ‘factor effect’

Res

pons

e -

Y

Factor ALow High

Average Y when A was set ‘high’

Average Y when A was set ‘low’

The differences are calculated for every factor in the experiment.

Okay - a little math must be done. But a computer helps to keep it simple.

Measuring an “Effect”

515151

When the effect of one factor changes due to the effect of another factor, the two factors are said to ‘interact’.

more than two factors can interact at the same time, but it is rare.

Res

pons

e - Y

Factor ALow High

B = High

B = Low

No interaction:

Resp

onse

- Y

Factor ALow High

B = High

B = Low

Slight

Re

spo

nse

- Y

Factor ALow High

B = High

B = Low

Strong

Looking for interactions

525252

Poor experimental discipline Measurement error Other errors Too much variation in the

response Aliases (confounded) effects Inadequate model Something changed

Reasons why a model might not confirm:

- And: - There may not be a true

cause-and-effect relationship.

535353

Is there really an effect?