Qualitymetin/Or6302/Folios/om...X-bar, R-bar, p charts Process variability vs. Process...

utdallas.edu/~metin

1

Quality Management

Chapter 9-10

These slides are based in part on slides that come with Cachon & Terwiesch

book Matching Supply with Demand http://cachon-terwiesch.net/3e/. If you

want to use these in your course, you may have to adopt the book as a textbook

or obtain permission from the authors Cachon & Terwiesch.

http://cachon-terwiesch.net/3e/

utdallas.edu/~metin2

Learning Goals

Statistical Process Control

X-bar, R-bar, p charts

Process variability vs. Process specifications

Yields/Reworks and their impact on costs

Just-in-time philosophy


Steer Support for the Scooter


Steer Support Specifications

Go-no-go

gauge


0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

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

R

79.9

79.91

79.92

79.93

79.94

79.95

79.96

79.97

79.98

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

X-b

ar

Control Charts


Statistical Process Control (SPC)

SPC: Statistical evaluation

of the output of a process during production/service

The Control Process

– Define

– Measure

– Compare to a standard

– Evaluate

– Take corrective action

– Evaluate corrective action

» Apply this to Global Warming or lack of it, or to Temperature Volatility.


Not just the mean is important, but also the variance

Need to look at the distribution function

The Concept of Consistency:

Who is the Better Target Shooter?



Capability

AnalysisConformance

Analysis

Investigate for

Assignable Cause

Eliminate

Assignable Cause

Capability analysis

• What is the currently "inherent" capability of my process when it is "in control"?

Conformance analysis

• SPC charts identify when control has likely been lost and assignable cause

variation has occurred

Investigate for assignable cause

• Find “Root Cause(s)” of Potential Loss of Statistical Control

Eliminate assignable cause

• Need Corrective Action To Move Forward



Shewhart’s classification of variability:

– Common (random) cause

– Assignable cause

Variations and Control

– Random variation: Natural variations in the output of

process, created by countless minor factors

» temperature, humidity variations, traffic delays.

– Assignable variation: A variation whose source can be

identified. This source is generally a major factor

» tool failure, absenteeism


Common Cause Variation (low level)

Common Cause Variation (high level)

Assignable Cause Variation

Two Types of Causes for Variation


Mean and Variance

Given a population of numbers, how to compute the

mean and the variance?

deviation Standard

)(

Variance

Mean

},...,,{Population

1

2

2

1

21

N

x

N

x

xxx

N

i

i

N

i

i

N


Sample for Efficiency and Stability

From a large population of goods or services (random if possible) a sample is drawn.

– Example sample: Midterm grades of OPRE6302 students whose last name starts with letter R {60, 64, 72, 86}, with letter S {54, 60}

» Sample size= n

» Sample average or sample mean=

» Sample range= R

» Standard deviation of sample means=

x

population theofdeviation Standard: where

n

x


Sampling Distribution

Sampling distribution

Variability of the average scores of

people with last name R and S

Process distribution

Variability of the scores

for the entire class

Mean

Sampling distribution is the distribution of sample means.

Grouping reduces the variability.


Normal Distribution

Mean

95.44%

99.74%

x

at x. cdf normal )1,_,,(normdist:functions lstatistica Excel

at x. pdf normal )0,_,,(normdist:functions lstatistica Excel

devstmeanx

devstmeanx

normdist(x,.,.,0)

Probab

normdist(x,.,.,1)


Cumulative Normal Density

)_,,(norminv :prob""at cdf offunction Inverse

)1,_,,(normdist:at x (cdf)function Cumulative

:functions lstatistica Excel

devstmeanprob

devstmeanx

0

1

x

normdist(x,mean,st_dev,1)

prob

norminv(prob,mean,st_dev)


Normal Probabilities: Example

If temperature inside a firing oven has a normal

distribution with mean 200 oC and standard deviation of

40 oC, what is the probability that

– The temperature is lower than 220 oC

=normdist(220,200,40,1)

– The temperature is between 190 oC and 220oC

=normdist(220,200,40,1)-normdist(190,200,40,1)


Control Limits for Consistency not Correctness

Sampling

distribution

Process

distribution

Mean

LCL

Lower

control

limit

UCL

Upper

control

limit

Process is in control if sample mean is between control limits.

These limits have nothing to do with product specifications!


Setting Control Limits:

Hypothesis Testing Framework

Null hypothesis: Process is in control

Alternative hypothesis: Process is out of control

Alpha=P(Type I error)=P(reject the null when it is true)=

P(conclude out of control when in control)

Beta=P(Type II error)=P(accept the null when it is false)

P(conclude in control when out of control)

If LCL decreases and UCL increases, we accept the null more easily. What happens to – Alpha?

– Beta?

Not possible to target alpha and beta simultaneously,

– Control charts target only a desired level of Alpha.


Type I Error=Alpha

Mean

LCL UCL

/2 /2

Probability

of Type I error

st_dev)mean,,2/1norminv(UCL

st_dev)mean,,2norminv(LCL

-α

α/

The textbook uses Type I error=1-99.74%=0.0026=0.26%.

Sampling distribution


Time

Process

Parameter

Upper Control Limit (UCL)

Lower Control Limit (LCL)

Center Line

• Track process parameter over time

- mean

- percentage defects

• Distinguish between

- common cause variation

(within control limits)

- assignable cause variation

(outside control limits)

• Measure process performance:

how much common cause variation

is in the process while the process

is “in control”?

Statistical Process Control: Control Charts


Control Chart

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

UCL

LCL

Sample number

Mean

Out of

control

Normal variation

due to chance

Abnormal variation

due to assignable sources

Abnormal variation

due to assignable sources


Observations from Sample Distribution

Sample number

UCL

LCL

1 2 3 4


Number of

Observations

in Sample

Sample size (n)

Factor for X-

bar Chart

(A2)

Factor for

Lower

control Limit

in R chart

(D3)

Factor for

Upper

control limit

in R chart

(D4)

Factor to

estimate

Standard

deviation, (d2)

2 1.88 0 3.27 1.128

3 1.02 0 2.57 1.693

4 0.73 0 2.28 2.059

5 0.58 0 2.11 2.326

6 0.48 0 2.00 2.534

7 0.42 0.08 1.92 2.704

8 0.37 0.14 1.86 2.847

9 0.34 0.18 1.82 2.970

10 0.31 0.22 1.78 3.078

Parameters for computing UCL and LCL

the Table method


Period x1 x2 x3 x4 x5 Mean Range

1 1.7 1.7 3.7 3.6 2.8 2.7 2 2 2.7 2.3 1.8 3 2.1 2.38 1.2 3 2.1 2.7 4.5 3.5 2.9 3.14 2.4 4 1.2 3.1 7.5 6.1 3 4.18 6.3 5 4.4 2 3.3 4.5 1.4 3.12 3.1 6 2.8 3.6 4.5 5.2 2.1 3.64 3.1 7 3.9 2.8 3.5 3.5 3.1 3.36 1.1 8 16.5 3.6 2.1 4.2 3.3 5.94 14.4 9 2.6 2.1 3 3.5 2.1 2.66 1.4

10 1.9 4.3 1.8 2.9 2.1 2.6 2.5 11 3.9 3 1.7 2.1 5.1 3.16 3.4 12 3.5 8.4 4.3 1.8 5.4 4.68 6.6 13 29.9 1.9 7 6.5 2.8 9.62 28 14 1.9 2.7 9 3.7 7.9 5.04 7.1 15 1.5 2.4 5.1 2.5 10.9 4.48 9.4 16 3.6 4.3 2.1 5.2 1.3 3.3 3.9 17 3.5 1.7 5.1 1.8 3.2 3.06 3.4 18 2.8 5.8 3.1 8 4.3 4.8 5.2 19 2.1 3.2 2.2 2 1 2.1 2.2 20 3.7 1.7 3.8 1.2 3.6 2.8 2.6 21 2.1 2 17.1 3 3.3 5.5 15.1 22 3 2.6 1.4 1.7 1.8 2.1 1.6 23 12.8 2.4 2.4 3 3.3 4.78 10.4 24 2.3 1.6 1.8 5 1.5 2.44 3.5 25 3.8 1.1 2.5 4.5 3.6 3.1 3.4 26 2.3 1.8 1.7 11.2 4.9 4.38 9.5 27 2 6.7 1.8 6.3 1.6 3.68 5.1

Average

3.81

5.85

n

xxxX n

...21

},...,min{

},...,max{

21

21

n

n

xxx

xxxR

• Collect samples over time

• Compute the mean:

• Compute the range:

as a proxy for the variance

• Average across all periods

- average of means,

- average of ranges,

• Normally distributed

The X-bar Chart: Application to Call Center

X

R


• Define control limits

0.58 comes from the Table.

•Constants are taken from a table

• Identify assignable causes:

- point over UCL

- point below LCL

- many (6) points on one side of center

• In this case:

- problems in period 13

- new operator (CSR) was assigned

0

2

4

6

8

10

12

1 3 5 7 9 11 13 15 17 19 21 23 25 27

UCL=X +A2 ×R =3.81+0.58*5.85=7.19

LCL=X -A2 ×R =3.81-0.58*5.85=0.41

CSR 1 CSR 2 CSR 3 CSR 4 CSR 5

mean 2.95 3.23 7.63 3.08 4.26 st-dev 0.96 2.36 7.33 1.87 4.41

Control Charts: The X-bar Chart

The Table method


Range Control Chart

ranges sample of average theof multipleA

ranges sample of average theof multipleA

3

4

RDLCL

RDUCL

Multipliers D4 and D3 depend on n and are available in the Table.

EX: In the last five years, the range of GMAT scores of incoming PhD class is

88, 64, 102, 70, 74. If each class has 6 students, what are UCL and LCL for

GMAT ranges?

079.6*0 159.279.6*2

.0D ,2D 6,nFor .6.795/)74701026488(

34

34

RDLCLRDUCL

R

Are the GMAT ranges in control?


0

2

4

6

8

10

12

1 3 5 7 9 11 13 15 17 19 21 23 25 27

0

5

10

15

20

25

30

1 3 5 7 9 11 13 15 17 19 21 23 25 27

X-B

ar

R

Control Charts: X-bar Chart and R-bar Chart

For the Call Center


X-bar and Range Charts: Which?

UCL

LCL

UCL

LCL

R-chart

x-Chart Detects shift

Does not

detect shift

(process mean is

shifting upward)Sampling

Distribution


X-bar and Range Charts: Which?

UCL

LC

L

LC

L

R-chart Reveals increase

x-Chart

UCL

Does not

reveal increase

(process variability is increasing)Sampling

Distribution


• Compute the standard deviation of the sample averages• stdev(2.7, 2.38, 3.14, 4.18, 3.12, 3.64, 3.36, 5.94, 2.66, 2.6, 3.16, 4.68, 9.62,

5.04, 4.48, 3.3, 3.06, 4.8, 2.1, 2.8, 5.5, 2.1, 4.78, 2.44, 3.1, 4.38, 3.68)=1.5687

• Use type I error of 1-0.9974

Control Charts: The X-bar Chart

The Direct method

8.531.5687)9987,3.81,norminv(0.

st_dev)mean,/2,-norminv(1UCL

-0.911.5687)0013,3.81,norminv(0.

st_dev)mean,/2,norminv(LCL

0.0026

Recall LCL=0.41 and UCL=7.19 in the Table method.


Tolerances/Specifications

– Requirements of the design or customers

Process variability

– Natural variability in a process

– Variance of the sample means coming from the process

Process Capability

Let us Tie Tolerances and Variability

Process capability

– Process variability relative to specifications

Capability = Process specifications / Process variability


Process Capability:

Specification limits are not control chart limits

Lower

SpecificationUpper

Specification

Process variability matches

specifications

Lower

Specification

Upper

Specification

Process variability well within

specificationsLower

Specification

Upper

Specification

Process variability exceeds

specifications

Sampling

Distribution

is used


Process Capability Ratio

When the process is centered, process capability ratio

A capable process has large Cp.

Example: The standard deviation, of sample averages of the midterm 1

scores obtained by students whose last names start with R, has been 7.

The SOM requires the scores not to differ by more than 50% in an exam.

That is the highest score can be at most 50 points above the lowest score.

Suppose that the scores are centered, what is the process capability ratio?

Answer: 50/42

X

pC6

levelion specificatLower - levelion specificatUpper


Process mean

Lower

specification

Upper

specification

+/- 3 Sigma

+/- 6 Sigma

3 Sigma and 6 Sigma Quality

Can you reduce variability so much that 12 Sigmas fit within specification limits?


• Estimate standard deviation:

• Or use the direct method with the excel function stdev()

• Look at standard deviation relative to specification limits

= R/d2

3

Upper

Specification (USL)

Lower

Specification (LSL)

X-3A X-2A X-1A X X+1A X+2 X+3A

X-6B X X+6B

Process A

(with st. dev A)

Process B

(with st. dev B)

Cp P{defect}

1 0.33 0.317=31.7%

2 0.67 0.0455

3 1.00 0.0027=0.27%

4 1.33 0.0001

5 1.67 0.0000006

6 2.00 2x10-9

The Statistical Meaning of Six Sigma

L

e

s

s

V

a

r

i

a

b

i

l

i

t

y


Another Chart: Use of p-Charts

p=proportion defective, assumed to be known

When observations can be placed into two categories.

– Good or bad

– Pass or fail

– Operate or don’t operate

– Go or no-go gauge


• Estimate average defect

1 300 18 0.060

2 300 15 0.050

3 300 18 0.060

4 300 6 0.020

5 300 20 0.067

6 300 16 0.053

7 300 16 0.053

8 300 19 0.063

9 300 20 0.067

10 300 16 0.053

11 300 10 0.033

12 300 14 0.047

13 300 21 0.070

14 300 13 0.043

15 300 13 0.043

16 300 13 0.043

17 300 17 0.057

18 300 17 0.057

19 300 21 0.070

20 300 18 0.060

21 300 16 0.053

22 300 14 0.047

23 300 33 0.110

24 300 46 0.153

25 300 10 0.033

26 300 12 0.040

27 300 13 0.043

28 300 18 0.060

29 300 19 0.063

30 300 14 0.047

p =0.052

SizeSample

pp )1( ̂ = =0.013

pUCL= + 3 ̂

pLCL= - 3 ̂ =0.091

=0.014

Period n defects p

Attribute Based Control Charts: The p-chart

Think of printing defective pages

• Define control limits

• Estimate Standard Deviation


0.000

0.020

0.040

0.060

0.080

0.100

0.120

0.140

0.160

0.180

13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

Attribute Based Control Charts: The p-chart

utdallas.edu/~metin 39

Inspection

» Before a costly operation, PhD comp. exam before candidacy

» Before an irreversible process, firing pottery

» Before a covering process, painting, assembly

» After some use: now 30,000 flights for cracks in the fuselage;

was 60,000 flights before April 1, 11 Southwest incidence.

Centralized vs. On-Site, my friend checks quality at cruise lines

Inputs Transformation Outputs

Acceptance

samplingProcess

control

Acceptance

sampling

Where/When » Raw materials

» Finished products


Process Bottleneck

Based on labor and

material cost

MarketEnd of

Process

Defect

occurred Defect

detected

Cost of

defect

$ $ $Based on sales

price (incl. Margin)

Recall, reputation,

warranty costs

Discovery of Defects and the Costs

CPSC, Segway LLC Announce Voluntary Recall to

Upgrade Software on Segway™ Human Transporters

The following product safety recall was conducted by the firm in

cooperation with the CPSC.

Name of Product: Segway Human Transporter (HT)

Units: Approximately 6,000

Recall AlertU.S. Consumer Product Safety Commission

Office of Information and Public Affairs

Washington, DC 20207, September 26, 2003

Defect

detected

Defect

detected


Examples of Inspection Points

Type of

business

Inspection

points

Characteristics

Fast Food Cashier

Counter area

Eating area

Building

Kitchen

Accuracy

Appearance, productivity

Cleanliness

Appearance

Health regulations

Hotel/motel Parking lot

Accounting

Building

Main desk

Safe, well lighted

Accuracy, timeliness

Appearance, safety

Waiting times

Supermarket Cashiers

Deliveries

Accuracy, courtesy

Quality, quantity


The Concept of Yields

90% 80% 90% 100% 90%

Line Yield: 0.9 x 0.8 x 0.9 x 1 x 0.9

Assuming that yields are independent.

What if failure rate is higher at the second stage for the items that pass the first stage?

Yield of Resource = rate Flow

resource the atcorrectly processed units of rate Flow

Yield of Process = rate Flow

resources theof allat correctly processed units of rate Flow


Rework / Elimination of Flow Units

Step 1 Test 1 Step 2 Test 2 Step 3 Test 3

Rework



Rework:

Defects can be corrected

Same or other resource

Leads to variability

Examples:

- Readmission to Intensive Care Unit

Loss of Flow units:

Defects can NOT be corrected

Leads to variability

To get X, we have to start X/y units

Examples:

- Interviewing

- Semiconductor fab

ReworkRework Rework


Why Having a Process is so Important:

Two Examples of Rare-Event Failures

Case 1: No rework loops

• Airport security

• Safety elements (e.g. seat-belts)

Case 2: Process has built-in rework loops

• Double-checking

1 problem every

10,000 units

99% correct

Indepen-

dence from

the first

Indepen-

dence from

the first two

“Bad” outcome happens

with probability (1-0.99)3,

That is (0.01) 3.

Good

Bad

99% 99%

99%

1% 1%

1%

You can add rework loops to improve quality.

If checks are dependent, this is less effective.

“Bad” outcome only happens

Every 100*10,000 units

That is probability of (0.01)3

1 problem

every 1 unit


Rare events are not so rare:Jetliner Crash due to Engine Icing

Engine flameout due to crystalline icing: Engine stops for 30-90 secs and hopefully starts again.

Suppose 150 1-engine flameouts over 1990-2005 and 15 2-engine flameouts over 2002-2005. What are the annualized 1- and 2-engine flameouts?

10=150/15 and 5=15/3

Let N be the total number of widebody jetliners flying through a storm per year. Assume that engines ice independently to compute N.

Set Prob(2 engine icing)=Prob(1 engine icing)2

(5/N)=(10/N)2 which gives N=20 ???

There are 1200 widebody jetliners worldwide. It is safe to assume that each flies once a day. Suppose that there are 2 storms on their path every day, which gives us about M=730 (=2*365) widebodyjetliner and storm encounters every year. How can we explain M=730 > N=20?

Engines do not ice independently. With M=730, Prob(1 engine icing)=10/730=1.37% and

Prob(2 engine icing)=5/730=0.68%.

Because of dependence,

Prob(2 engine icing) >> Prob(1 engine icing) 2 .

Unjustifiable independence leads to underestimation of the failure probabilities in operations, finance, engineering, flood control, nuclear power plants, etc.

May 11, 2011, the Emirates A-380

from Dubai was struck by lightning

as it approached Heathrow airport


Rare events are not so rare:

An Earthquake and a Power Loss

Unjustifiable independence leads to underestimation of the failure probabilities in nuclear power plants.

An earthquake in Japan or a tornado in Texas can cause

both a power outage and a structural damage in a nuclear reactor.Alabama’s Browns Ferry reactor had a faulty cooling valve replaced right before April 27 2011 tornados

that cut off the power to the reactor which was idled for a while (>2 weeks) afterwards.

Squaw Creek Reservoir

for cooling water

Pacific Ocean

for cooling water


Just-in-Time Philosophy

Pull the operations rather than pushing them

– Inventory reduction

– JIT Utopia

» 0-setup time

» 0-non value added operations

» 0-defects

Discover and reduce process variability


Push vs Pull System

What instigates the movement of the work in the system?

In Push systems, work release is based on downstream demand

forecasts

– Keeps inventory to meet actual demand

– Acts proactively

» e.g. Making generic job application resumes today (e.g.: exempli gratia)

In Pull systems, work release is based on actual demand or the

actual status of the downstream customers

– May cause long delivery lead times

– Acts reactively

» e.g. Making a specific resume for a company after talking to the recruiter


Push/Pull View of Supply Chains

Procurement,Manufacturing andReplenishment cycles

Customer Order

Cycle

Customer

Order Arrives

Push-Pull boundary

PUSH PROCESSES PULL PROCESSES


Direction of production flow

upstream downstream

Authorize

production

of next unit

Pull Process with

Kanban Cards


Bro

wser

err

or

Number of

defects

Cause of Defect

Absolute

Number Percentage Cumulative

Browser error 43 0.39 0.39

Order number out of sequence 29 0.26 0.65

Product shipped, but credit card not billed 16 0.15 0.80

Order entry mistake 11 0.10 0.90

Product shipped to billing address 8 0.07 0.97

Wrong model shipped 3 0.03 1.00

Total 110O

rder

nu

mb

er

ou

t

off

seq

uen

ce

Pro

du

ct

sh

ipp

ed

, b

ut

cre

dit

card

no

t b

ille

d

Ord

er

en

try

mis

take

Pro

du

ct

sh

ipp

ed

to

billin

g a

dd

ress

Wro

ng

mo

del

sh

ipp

ed

100

50

Cumulative

percents of

defects

100

75

50

25

Pareto Principle or 20-80 rule

Errors in the shipping process


•It is not enough to look at “Good” vs “Bad” Outcomes

•Only looking at good vs bad wastes opportunities for learning; especially as failures become

rare (closer to six sigma) you need to learn from the “near misses”

Reduce Variability in the Process

Taguchi: Even Small Deviations are Quality Losses

Lower

Specification

Limit

Target

value

Quality

Loss

High

Low

Performance

Metric Target

value

Quality

Loss

Performance

Metric, x

Upper

Specification

Limit

Traditional view of Quality loss Taguchi’s view of Quality loss

Performance

Metric


• Double-checking

• Fool-proofing, Poka yoke (see Toyota)

• Computer plugs

• Set the watch 5 mins ahead

• Process recipe

• Recipes and checklists help standardize

Accommodate Residual (Common) Variability

Through Robust Design

• Team : Are all the nurse and surgeons present

• Learn names of the team

• Surgical tools clean

• Sponges new

• ….

• ….

Book to Read: The Checklist Manifesto:

How to Get Things Right

By Atul Gawande, Metropolitan Books


Materials

MachinesSpecifications /

information

People

Vise position

set incorrectly

Machine tool

coordinates set incorrectly

Vice position shifted

during production Part clamping

surfaces corrupted

Part incorrectly

positioned in clamp

Clamping force too

high or too low

Cutting

tool worn

Dimensions incorrectly

specified in drawing

Dimension incorrectly coded

In machine tool program

Material

too soft

Extrusion stock

undersized

Extrusion die

undersized

Extrusion

rate

too high

Extrusion temperature

too high

Error in

measuring height

Steer support

height deviates

from specification

Ishikawa (Fishbone) Diagram


Summary


X-bar, R-bar, p charts

Process variability vs. Process specifications

Yields/Reworks and their impact on costs

Just-in-time philosophy


Jesica Santillan died after a bungled heart-lung transplant in 2003. In an operation

Feb. 7, Jesica was mistakenly given organs of the wrong blood type.

Her blood type was 0 Rh+.

Organs come from A Rh- blood type.

Her body rejected the organs, and a matching transplant about two weeks later

came too late to save her. She died Feb. 22 at Duke University Medical Center.

Line of Causes leading to the mismatch

• On-call surgeon on Feb 7 in charge of pediatric heart transplants,

James Jaggers, did not take home the list of blood typesLater stated, "Unfortunately, in this case, human errors were made during the

process. I hope that we, and others, can learn from this tragic mistake."

• Coordinator initially misspelled Jesica’s name

• Once UNOS (United Network for Organ Sharing) identified Jesica,

no further check on blood type

• Little confidence in information system / data quality

• Pediatric nurse did not double check

• Harvest-surgeon did not know blood type

Process Failure in Healthcare:

The Case of Jesica Santillan


- As a result of this tragic event, it is clear to us at Duke that we need to have

more robust processes internally and a better understanding of the

responsibilities of all partners involved in the organ procurement process.

William Fulkerson, M.D., CEO of Duke University Hospital.

- We didn’t have enough checks.

Ralph Snyderman, Duke University Hospital

Jesica is not the first death in organ transplantation because of blood type mismatch.

Process Failure in Healthcare:

The Case of Jesica Santillan


Step 1: Define and map processes

- Jaggers had probably forgotten the list with blood groups 20 times before

- Persons involved in the process did not double-check,

everybody checked sometimes

- Learning is triggered following deaths / process deviations are ignored

Step 2: Reduce variability- quality of data (initially misspelled the name)

Step 3: Robust Design- color coding between patient card / box holding the organ

- information system with no manual work-around

- let the technology help

RFID tagged patients: Tag includes blood type and other info

Electronic medicine box: Alarming for the obsolete medicine

The Three Steps in the Case of Jesica


1. Management Responsibility

2. Quality System

3. Contract review

4. Design control

5. Document control

6. Purchasing / Supplier evaluation

7. Handling of customer supplied material

8. Products must be traceable

9. Process control

10. Inspection and testing

11. Inspection, Measuring, Test Equipment

12. Records of inspections and tests

13. Control of nonconforming products

14. Corrective action

15. Handling, storage, packaging, delivery

16. Quality records

17. Internal quality audits

18. Training

19. Servicing

20. Statistical techniques

Examples: “The design process shall be planned”,

“production processes shall be defined and planned”

How do you get to a Six Sigma Process?

Do Things Consistently (ISO 9000)


Zero Inventories

Zero Defects

Flexibility / Zero set-ups

Zero breakdowns

Zero handling / non

value added

Just-in-time Production

• Kanban

• Classical Push

• “Real” Just-in-time

Mixed Production

Set-up reduction

Autonomation

Competence and Training

Continuous Improvement

Quality at the source

Organization MethodsPrinciples

The System of Lean Production

Pardon the French, caricatures are from Citroen.


• Avoid unnecessary inventory

• To be seen more as an ideal

• To types of (bad) inventory:

a. resulting from defects / rework

b. absence of a smooth process flow

• Remember the other costs of inventory (capital, flow time)

Invento

ry in p

rocess

Buffer argument:

“Increase inventory”

Toyota argument:

“Decrease inventory”

Principles of Lean Production:

Zero Inventory and Zero Defects


71

2345

68

ITAT=7*1 minute

3

1

2

4

ITAT=2*1 minute

Good unit

Defective unit

ITAT: Information Turnaround Time


• Flexible machines with short set-ups

• Allows production in small lots

• Real time with demand

• Large variety

• Maximize uptime

• Without inventory, any breakdown

will put production to an end

• preventive maintenance

Avoid Non-value-added activities,

specifically rework and set-ups

Principles of Lean Production:

Zero Set-ups, Zero NVA and Zero Breakdowns


Push: make to forecast Pull: Synchronized production

Pull: Kanban

• Visual way to implement a pull system

• Amount of WIP is determined by

number of cards

• Kanban = Sign board

• Work needs to be authorized by demand

• Classical MRP way

• Based on forecasts

• Push, not pull

• Still applicable for

low cost parts

• Part produced for specific

order (at supplier)

• shipped right to assembly

• real-time synchronization

• for large parts (seat)

• inspected at source

Methods of Lean Production: Just-in-time


InventoryInventoryInventory

CycleInventory

Production with large batches

End of

Month

Beginning of

Month

Cycle


End of

Month

Beginning of

Month

Cycle


End of

Month

Beginning of

Month

Cycle


End of

Month

Beginning of

Month

Inventory

End of

Month

Beginning of

Month

Produce Sedan

Produce Station wagon

End of

Month

Beginning of

Month

Produce Sedan


End of

Month

Beginning of

Month

Produce Sedan


End of

Month

Beginning of

Month

Production with small batches

Methods of Lean Production:

Mixed Production and Set-up reduction


• Create local decision making rather

than pure focus on execution

• Use machines / tools, but avoid the

lights-off factory

• Automation with a human touch

• Cross training of workers

• Develop problem solving skills

Organization of Lean Production:

Autonomation and Training


• Solve the problems where they occur

- this is where the knowledge is

- this is the cheapest place

• Traditional: inspect and rework

at the end of the process

• Once problem is detected, send

alarm and potentially stop

the production

Own Process Next Process End of Line FinalInspection

End User

$ $ $ $ $

• Rework

• Reschedule

• very minor • minor

delay

• Significant

Rework

• Delayed

delivery

• Overhead

• Warranty

cost

• recalls

• reputation

• overhead

Defect found

Defect fixed

Organization of Lean Production: Continuous Improvement and Quality-at-the-source

Date post:	27-Sep-2020
Category:	Documents
Upload:	others
View:	4 times
Download:	1 times

Qualitymetin/Or6302/Folios/om...X-bar, R-bar, p charts Process variability vs. Process...

Documents