Session 9

Quality Management

OMM 5704 Session 9 Handout

1 of 34

CYCLE TIME
THROUGHPUT

CAPACITY

THE OPS QUADRANGLE

INVENTORY

Some TPS terms

Roma-ji andon goesu heijunka jidoka kaizen kanban S English Andon Lamp 5S (standing for Seiri, Seiton, Seiso, Seiketsu, Shitsuke)

Leveled Production Quality in Station (Jidoka avoids mindless automation) Continuous Improvement (persistence in achieving Just in Time, Jidoka and Heijunka) Kanban, Kanban Card, Kanban System Fail-safe Devices, literally avoiding the effects of distraction
Chayet

pokayoke

OMM

OMM 5704 Session 9 Handout

2 of 34

OMM

OMM

1940

2-3 hr setups

Taiichi Ohno: TPS manual 1950

OMM 5704 Session 9 Handout

manual system compete w/o long prod. runs

Sort (Seiri)

Shine (Seiso)

Autonomation + JIT

Sustain (Shitsuke)

Streamline (Seiton)

1960

15 min setups

MRP
Materials Requirement Planning Orlicky, Wight & Plossl

Standardize (Seiketsu)

3 min setups MRP Crusade

150 Implementations

1970

5S

Eliminate unnecessary tools / parts

8,000 Implementations

A bit of History

1980

Quality
NBT $1.2 B ~ 1/3 of software sales!

JIT
Manufacturing Resources Planning MRP + Modules

MRP II

TQM

ERP 1990 Lean SCM

Supply Chain Management SAP Enterprise Resources Planning MRP II + modules, client/server

Clean the work area, keep it tidy and organized

Womack et al.

ISO 9000

Agile 2000

Six Sigma

Arrange work for orderly ow, remove waste / non value add

Ensure adherence to rules to prevent backsliding, visual workplace

Ensure uniform processes for interchangeability, regular maintenance

Hopp & Spearman 2004

Lean Six Sigma SCM ?

5

Chayet

Chayet

3 of 34

Benets of Heijunka (balanced production, batch size of one)

1. Lowers ! to suppliers 2. Balances the line 3. Shortens response time to customers

OMM

Chayet

Benets of Heijunka (balanced production, batch size of one) 1. Lowers ! to suppliers

Without Heijunka 10,000 Wagons Months 1,3, 5,...

10,000 Sedans Months 2,4,6,...

OMM

Chayet

OMM 5704 Session 9 Handout

4 of 34

Benets of Heijunka (balanced production, batch size of one) 1. Lowers ! to suppliers

With Heijunka

OMM

Chayet

Benets of Heijunka (balanced production, batch size of one) 2. Balances the line
Without Heijunka 2 min 1 min 1 min
1 2 min

w =

1 min

2 min

2 min

s =

1 2 min

OMM

Chayet

10

OMM 5704 Session 9 Handout

5 of 34

Benets of Heijunka (balanced production, batch size of one) 2. Balances the line
With Heijunka 2 min 1 min 1 min
= 2 1 = 3 min 1.5 min

1 min

2 min

2 min

OMM

Chayet

11

Benets of Heijunka (balanced production, batch size of one)

With Heijunka: Plant Robust with respect to Product Mix changes but not with respect to volume changes

OMM

Chayet

12

OMM 5704 Session 9 Handout

6 of 34

Heijunka Requires

Short setup/changeover times Coordination with suppliers

OMM

Chayet

13

Single Minute Exchange of Dies (SMED)

Process improvement to reduce setup/changeover times

OMM

Chayet

14

OMM 5704 Session 9 Handout

7 of 34

Cultural/Geographic Differences...

OMM

Chayet

15

A River Analogy: Lowering WIP Inventory

TOO MUCH SPACE INVENTORY

MISSED DUE DATES MACHINE BREAKDOWNS LARGE LOT SIZES

THE ROCKS ARE QUALITY, PRODUCTIVITY, AND RESPONSIVENESS PROBLEMS

LONG LEAD TIMES SCRAP AND REWORK

Chayet

POOR VENDOR QUALITY

UNRELIABLE DELIVERIES

OMM

16

OMM 5704 Session 9 Handout

8 of 34

Jidoka - Quality at the Source

Andon

OMM

Chayet

17

Jidoka - Quality at the Source

Andon

OMM

Chayet

18

OMM 5704 Session 9 Handout

9 of 34

Jidoka - Quality at the Source

Detect defect when it occurs

OMM

Chayet

19

Poke Yoke: Fool Proof

Before Kaizen

After Kaizen Poke Yoke

Work (in reverse) Machine Chuck

Work (correct)

Spindle

OMM

Chayet

20

OMM 5704 Session 9 Handout

10 of 34

Identifying Defects at the Source Lowers Costs

Defects Found at: Own Process Next Process End of Line Final Inspection End Users Hand

Cost to the Company:

Impact to the Very Company: Minor

Minor Delay

Rework Reschedule of work

Signicant Rework Delay in Delivery Additional Inspection

Warranty Costs Administrative Costs Reputation

OMM

Loss of Market Share

Chayet

21

Responsibility Reversal

Traditionally...
Operators Responsible for: Throughput

Management Responsible for:

Quality

OMM

Chayet

22

OMM 5704 Session 9 Handout

11 of 34

Responsibility Reversal

In TPS...
Operators Responsible for: Quality

Management Responsible for:

Throughput

OMM

Chayet

23

TPS / JIT / Lean

A Complete Integrated System: Dont Cherry-Pick portions

OMM

Chayet

24

OMM 5704 Session 9 Handout

12 of 34

TPS/JIT Summary
1. All processes driven to be in control and capable (customer expectations) 2. Problems are natural: Opportunities to learn 3. Every activity must add value (whole system) 4. Heijunka (connect value chain from customers to suppliers) 5. Human infrastructure: 6. It is a System:
1. No cherry pick, optimize whole system 2. Pragmatism (no absolutes, e.g., zero inventory) 3. Robust (with respect to uncertainty) 4. A system deteriorates without constant attention!
Fact based vs. blame - based.

OMM

Chayet

25

Quality Management
Saw it on the tube Bought it on the phone Now youre home alone Its a piece of crap. I tried to plug it in I tried to turn it on When I got it home It was a piece of crap. Neil Young

OMM

Chayet

26

OMM 5704 Session 9 Handout

13 of 34

Denitions of Quality

Transcendent: Innate excellence Product-Based: Function of product attributes User-Based: Based on customer preferences / values Process-Based: Conformance to specications Value-Based: Jointly determined by performance and price

OMM

Chayet

27

Variability and Quality

Variability: Machine setups, failures, uctuations in product mix, operator behavior, etc.

Quality

Variability

OMM

Chayet

28

OMM 5704 Session 9 Handout

14 of 34

TQM versus SPC

TQM: Qualitative management focus SPC: Quantitative engineering focus

OMM

Chayet

29

Quality as a Competitive Dimension

1970s Cost 1980s Quality 1990s Speed 2000s ?

OMM

Chayet

30

OMM 5704 Session 9 Handout

15 of 34

Cost of Quality
Cost increases with quality
(Jurans Optimal conformance level)

or Quality is free ?
(Crosby)

OMM

Chayet

31

Quality Management Programs

Demings 14 points Jurans 7 step breakthrough sequence Crosbys 14 point program

OMM

Chayet

32

OMM 5704 Session 9 Handout

16 of 34

Learning Mechanisms
Effective Operations Management requires the installation and monitoring of mechanisms for learning about solving problems.

OMM

Chayet

33

Tools for Managing Quality

Jidoka (e.g., andon cord pulls, visual control, cleanliness) 5 Whys Teamwork / empowerment / coaching Worker rotation Standard procedures (e.g., ISO 9000) Learning Mechanisms Statistical process control (SPC)

OMM

Chayet

34

OMM 5704 Session 9 Handout

17 of 34

An Experiment

OMM

Chayet

35

Ideally

Aim
OMM Chayet

36

OMM 5704 Session 9 Handout

18 of 34

Actually

Aim
OMM Chayet

37

Experiment
1/2 of time 1/2 of time

OMM

Aim

Aim

Chayet

38

OMM 5704 Session 9 Handout

19 of 34

Experiment

xed aim point

two aim points

OMM

Chayet

39

Experiment
1/3 of time 1/3 of time xed aim point 1/3 of time three aim points

Aim
OMM

Aim

Aim
Chayet

40

OMM 5704 Session 9 Handout

20 of 34

Summary

xed aim point

two aim points

three aim points

OMM

Chayet

41

Common Pitfall
Attempting to manage stable process (e.g., experiment) increased variance!

need to distinguish between common and special variation

OMM

Chayet

42

OMM 5704 Session 9 Handout

21 of 34

Types of Variability

Common or natural variability

multitude of relatively small and purely random variations

Special or assignable-cause variability

variation that could indicate problems, changes in system

OMM

Chayet

43

Statistical Control
If process exhibits only common or natural variability is stable and in statistical control A process subject to special or assignable-cause variability is said to be unstable or out of control

OMM

Chayet

44

OMM 5704 Session 9 Handout

22 of 34

Normal Distribution

mean 68.27 % 95.45 % 99.73 %

OMM

Chayet

45

Control Charts
UCL

Center Line

LCL

OMM

Chayet

46

OMM 5704 Session 9 Handout

23 of 34

Control Charts
UCL = Upper Control Limit = mean + z LCL = Lower Control Limit = mean - z Traditionally and in practice z = 3 so Probability of being within CLs is 99.73%

OMM

Chayet

47

A Process in Control
UCL

Center Line

LCL

OMM

Chayet

48

OMM 5704 Session 9 Handout

24 of 34

Periodicity
UCL

Center Line

LCL

OMM

Chayet

49

Trends
UCL

Center Line

LCL

OMM

Chayet

50

OMM 5704 Session 9 Handout

25 of 34

Hugging of a Control Line

UCL

Center Line

LCL

OMM

Chayet

51

Capable Processes
USL UCL

Center Line

LCL LSL
USL = Upper Specication Limit, LSL= Lower Specication Limit
OMM Chayet

52

OMM 5704 Session 9 Handout

26 of 34

Non Capable Processes

UCL

USL Center Line LSL

LCL
USL = Upper Specication Limit, LSL= Lower Specication Limit
OMM Chayet

53

Process Control and Capability

All processes are driven to be in control and capable

Capable In Control Not Capable Out of Control

OMM

Chayet

54

OMM 5704 Session 9 Handout

27 of 34

55

Six Sigma

OMM

Chayet

56

OMM 5704 Session 9 Handout

28 of 34

Motorolas Six Sigma Program

In 1989 Motorola recognized problems with quality Motorola Quality Goal Motorola won Baldridge Quality Award in 1998
Improve product & service quality by factor of 10 by 1989 Improve by factor of 100 by 1991 Achieve Six Sigma capability by 1992

OMM

Chayet

57

Cell Phone
1200 processes in product At 3design limits:

Prob of defect in one process = 1" P ( "3 # Z # 3 ) = .0027

.0027 $ 1,000,000 = 2,700 ppm (parts per million)

Prob of zero defect in nal product = (1-0027)1200
= .039

1" .039 = 96.1% of nal products have at least one defect!

OMM

Chayet

58

OMM 5704 Session 9 Handout

29 of 34

Idea
Suppose we want 99.6% of nal product defect free. Find the ppm defect rate (per component) that achieves this: Solve (1-p)
1200 = .996

for p yields p = 3.34 ppm

Want design limits such that P (Outside of z Get z ! 4.65.

) = 3.34 ppm

OMM

Chayet

59

Plan for a 1.5 sigma shift

!1.5 +1..5

mean

OMM

Chayet

60

OMM 5704 Session 9 Handout

30 of 34

Plan for a 1.5 sigma shift

!1.5 +1..5

mean

OMM

Chayet

61

Plan for a 1.5 sigma shift

Find z so that the ppm defect (per component) is no more than p = 3.34 ppm (i.e., 99.6% of nal product defect free) EVEN IF there is a shift in the mean of 1.5

Motorola instituted such policy (accounts for cases when mean

of process not aligned with design value)

This puts even greater strain on design limits Solving for 1" P [ "(z " 1.5)
OMM

# Z # (z " 1.5) ) = 3.4 ppm

Get z " 1.5 ! 4.65 or z ! 6.15

Chayet

62

OMM 5704 Session 9 Handout

31 of 34

Four Sigma versus Six Sigma

Four Sigma Lost articles of mail (per hour) 15 minutes of unsafe drinking water Incorrect surgeries per week 2 short or long landings at major airports 20,000 per day 5,000 per day Six Sigma 7 every 8.75 years 1.7 every 10 years

OMM

Chayet

63

Six Sigma Program

A statistically-based, structured methodology for identifying and eliminating causes of errors in a process. DMAIC: Dene, Measure, Analyze, Improve, and Control

OMM

Chayet

64

OMM 5704 Session 9 Handout

32 of 34

The Role of a Black Belt

Mentor Teach Coach Transfer Identify Inuence

OMM

Chayet

65

OMM 5704 Session 9 Handout

33 of 34

Washington University Olin Business School OMM 5704 Operations Management Fall B 2013, Chayet

Session 9: Lessons
1. Toyota Production System / Just in Time: We discussed some of the principles of the Toyota Production System (TPS). The philosophy is to reduce waste and variability. In response to variability the rm will hold excess inventory, capacity, or backlogs, all of which are detrimental to rm prots. We discussed Heijunka (batch sizes of one), Just in Time, and Poke-Yoke (fool proof systems), and mentioned how these ideas have inuenced American manufacturing and service industries since the 1980s. 2. Jidoka: Quality at the source. Addressing quality problems is far less expensive to the rm if these are detected when and where they occur rather than after theyve reached the consumer. Eective operations management requires that managers install and monitor mechanisms for learning about and solving problems as quickly as possible to avoid separating them from their causes. 3. Common and Special Variability: Common variations are statistically predictable while special variations arise from assignable causes. A process is in statistical control if all of its variability is of the common type. Using a simple experiment we illustrated how a stable process can become unstable as a result of over-control. 4. Process Control Charts: Process control charts are a practical tool for monitoring the variability of a process, and can be applied broadly to almost any performance statistic imaginable. Recall that just because a process is within control limits, it does not mean it is in control. If out-of-control patterns persist, then management should take action to investigate the assignable causes. A process is capable if it meets the customers requirements with regularity. Although all capable processes are in control. a process may be in control but not be capable. 5. Six Sigma: A process is operating at six sigma if the variations within plus/minus six sigma constitute being free of defects. This high degree of process capability is required for every subcomponent of a nal product in order to ensure that the probability of a defect free nal product is high. A process can be made more capable by reducing variability, or altering the customer requirements. Six Sigma programs formalize and structure problem solving around these process-control ideas for implementation throughout the organization.

OMM 5704 Session 9 Handout

34 of 34