everything you wanted to know about six-sigma but were afraid to ask! dave stewardson - isru ronald...
TRANSCRIPT
Everything you wanted to know about
Six-Sigma but were afraid to ask!
Dave Stewardson - ISRU
Ronald Does – The Netherlands
Soren Bisgaard - USA
Bo Bergman – Sweden
Ron Kennet – Israel
Oystein Evandt – Norway
Xavier Tort-Martorell - Spain
Pro-EnbisAll joint authors - presenters- are members of:
Pro-Enbis and ENBIS.
This presentation is supported by Pro-Enbis a Thematic Network funded under the ‘Growth’ programme of the European Commission’s 5th Framework research programme - contract number G6RT-CT-2001-05059
ENBIS
European Network for Business and Industrial Statistics
www.enbis.org
Overview
Brief resume of Six Sigma- Key concepts- Training- Execution
The Scientific Method Project selection “Quotes” Barriers – Overcoming these Critique of ISO 9000 Change programs Real reasons why six-sigma works Simple case study
Hoovering 30 m2 on 6-level means only 1 cm2 missed. 1/3.4 million part of the day equals 0.29 second 1/3.4 million part of the equator of the earth equals about 140
meter.
is the symbol for the standard deviation.
“6” is equivalent with 3.4 defects per million opportunities.
6 : new world
A new way of doing business?
Statistical background
Target =
Some Key measure
Statistical background
Target =
‘Control’ limits
LSL USL
Statistical background
Required Tolerance
Target =
LSL USL
Statistical background
Tolerance
Target =
Six-Sigma
LSL USL
ppm1350
ppm1350
Statistical background
Tolerance
Target =
LSL USL
ppm0.001
ppm1350
ppm1350
ppm0.001
Statistical background
Tolerance
Target =
Statistical background
But Six-Sigma allows for un-forseen ‘problems’ and longer term issues
when calculating
failure
error or
re-work rates
Assumes a process ‘shift’
LSL
0 ppm ppm3.4
USL
ppm3.4ppm
66803
Statistical background
Tolerance
Performance Standards
23456
30853766807
62102333.4
PPM
69.1%93.3%99.38%99.977%99.9997%
Yield
Processperformance
Processperformance
Defects permillion
Defects permillion
Long term yield
Long term yield
Current standardCurrent standard
World ClassWorld Class
Number of processesNumber of processes 3σ3σ 4σ4σ 5σ5σ 6σ6σ
110
100500100020002955
110
100500100020002955
93.3250.090.10000
93.3250.090.10000
99.37993.9653.644.440.200
99.37993.9653.644.440.200
99.976799.7797.7089.0279.2462.7550.27
99.976799.7797.7089.0279.2462.7550.27
99.9996699.996699.96699.8399.6699.3299.0
99.9996699.996699.96699.8399.6699.3299.0
First Time Yield in multiple stage process
Performance standards
Benefits of 6approach w.r.t. financials
-level Defect rate (ppm)
Costs of poor quality Status of the company
6 3.4 < 10% of turnover World class 5 233 10-15% of turnover 4 6210 15-20% of turnover Current standard 3 66807 20-30% of turnover 2 308537 30-40% of turnover Bankruptcy
Financial Aspects
Simple– Eliminate defects– Eliminate the opportunity to have defects
Complex– Vision– Metric (Standard measuring method)– Benchmark– Philosophy– Method– Tool for:
Customer satisfaction ‘Breakthrough’ improvements Continuous improvement
Employee involvement – Agressive goals
What is Six Sigma as a Concept?
A scientific and practical method to achieve improvements in a company
Scientific:• Structured approach.• Assuming quantitative data.
Practical:• Emphasis on financial result.• Start with the voice of the customer.
“Show me the data”
”Show me the money”
Six Sigma
Six Sigma Methods Production
DesignService
Purchase
HRM
Administration
QualityDepart.
Management
M & S
IT
Where can Six Sigma be applied?
GE “Service company”Examples
Approving a credit card application Installing a turbine Lending money Servicing an aircraft engine Answering a service call for an appliance Underwriting an insurance policy Developing software for a new CAT product Overhauling a locomotive
DMAIC
Define Select a project
Measure Prepare for assimilating information
Analyze Characterise the current situation
Improve Optimise the process
Control Assure the improvements
Six-Sigma - A “Roadmap” for improvement
• Belongs to the middle management• Is well-educated• Project is related to his daily activities• May prioritise his work• Well motivated• Willing to change• Has good social skills
Black BeltBlack Belt
Improvement potential: € 50 000
Execution
Training (1 week)Training (1 week)
Work on project(3 weeks)
Work on project(3 weeks)
ReviewReview
Define
Measure
Analyze
Improve
Control
Throughput time projectThroughput time project
4 months (full time)4 months (full time)
Classic Training strategy
In Spain5 x three day sessions
Includes weekend
More ‘Homework’
Heaver individual support
Fewer advanced methods
It is their view that some training is not assimilated by delegates and that some items do not fit the need of some delegates
Training
In Poland
5 x 5 day sessions
But with 5 weeks between training sessions not 3 weeks
Extra Support via on-line materials
Individual support stepped up
Training
In Sweden
4 x 5 day day sessions
3 week gaps as in America
Less emphasis on top-down
Perceived to be more need for buy-in by staff than in America
Training
Black BeltBlack Belt
TrainingTraining
ApplicationApplication
ReviewReview
MBBMBB
MBB,Champion
MBB,Champion
MBB,Champion
MBB,Champion
Project execution
Traditional Six Sigma
-Project leader is obliged to make an effort.
-Set of tools .
-Focus on technical knowledge.
-Project leader is left to his own devices.
-Results are fuzzy.
-Safe targets.
-Projects conducted “on the side”.
-Black Belt is obliged to achieve financial results.
-Well-structured method.
-Focus on experimentation.
-Black Belt is coached by champion.
-Results are quantified.
-Stretched targets.
-Projects are top priority.
Conducting projects
Black Belt is given the required resources
-Training in statistical methods.
-Time to conduct his project!
-Software to facilitate data analysis.
-Permissions to make required changes!!
-Coaching by a champion – or external support.
Project support
In other words the Black Belt is
-Empowered.
-In the sense that it was always meant!
-As the theroists have been saying for years!
Project support
7. Screen potential causes.8. Discover variable relationships.9. Establish operating tolerances.
10. Validate measurement system.11. Determine process capability.12. Implement process controls.
DMAIC procedure
4. Establish product capability.5. Define performance objectives.6. Identify variation sources.
1. Select CTQ characteristic.2. Define performance standards.3. Validate measurement system.
Measure
Analyze
Improve
Control
Define
Roadmap to improvement
Statistics
Methods for the collection, presentation and analysis of data.
Based on mathematics and mathematical modelling.
Major role is played by uncertainty / variation.
Statistical approach to quality improvement:1. Explain predict control.2. All ideas are empirically tested before they are accepted.
1. Y = f(X1, X2, … , Xn).
2. “Show me the data”.
1. Y = f(X1, X2, … , Xn).
2. “Show me the data”.
Basic approach
Data, measurements, observations
Hypotheses (potential leverage variables)
Cre
ativ
eth
inki
ng
Critical
thinking
TestingExploratory
study
Learning by scientific method:
Scientific method
Scientific method (after Box)
INDUCTION INDUCTION
DEDUCTION DEDUCTION
DataFacts
TheoryHypothesisConjectureIdeaModel
Check
Plan
DoAct
Plan
DoCheck/Study
Act
Deming Cycle
The Scientific Process
Key elements:– Formulation of the problem– Collection of data– Experimentation– Generation of ideas from patterns in data–
hypothesis generation– Making predictions from hypothesis– Comparing predictions with real data– Making inferences from the data
Exploratory study:At first we search -- like a detective -- in the data for traces of potential leverage variables. We must not be critical. It is more important to find all leverage variables.
Testing:Then we determine -- like a judge -- which of the potential leverage variables are indeed important. We do this by conducting an experiment.
How to discover potential leverage variables:
Exploit available knowledge:• FMEA• Cause and effect diagram• Technical literature
Collection and analysis of data:• Control chart• Boxplot• Scatter diagram
The search for root causes
Practical solution Statistical solution
Statistical problemPractical problem
Y = f(X1, X2, …, Xn)
Approach to improve
Problem fixing vs. explanation
Define
Select:- the project - the process- the Black Belt- the potential savings- time schedule- team
Project selection
Is management’s responsibility.
Projects may be selected according to:
1. A complete list of requirements of customers.
2. A complete list of costs of poor quality.
3. A complete list of existing problems or targets.
Project selection
1. Requirements,2. Costs,3. Problems.
1.Collect data 2.Arrange the information
3.Give priority-Financial benefits-Expected throughput time of the project-Severity of the problem
321
Project prioritization
Before a simple case study a few quotes - some important
issues - then some why’s?
“the most important initiative GE has ever undertaken”.
Jack WelchChief Executive OfficerGeneral Electric
• In 1995 mandated each GE employee to work towards achieving 6 sigma• The average process at GE was 3 sigma in 1995• In 1997 the average reached 3.5 sigma • GE’s goal is to reach 6 sigma by 2001• Investments in 6 sigma training and projects reached 45MUS$ in 1998, profits increased by 1.2BUS$
• In 1995 mandated each GE employee to work towards achieving 6 sigma• The average process at GE was 3 sigma in 1995• In 1997 the average reached 3.5 sigma • GE’s goal is to reach 6 sigma by 2001• Investments in 6 sigma training and projects reached 45MUS$ in 1998, profits increased by 1.2BUS$
General ElectricGeneral Electric
“At Motorola we use statistical methods daily throughout all of our disciplines to synthesize an abundance of data to derive concrete actions….How has the use of statistical methods within Motorola Six Sigma initiative, across disciplines, contributed to our growth? Over the past decade we have reduced in-process defects by over 300 fold, which has resulted in a cumulative manufacturing cost savings of over 11 billion dollars”*.
Robert W. GalvinChairman of the Executive CommitteeMotorola, Inc.
MOTOROLMOTOROLAA
*From the forward to MODERN INDUSTRIAL STATISTICS by Kenett and Zacks, Duxbury, 1998*From the forward to MODERN INDUSTRIAL STATISTICS by Kenett and Zacks, Duxbury, 1998*From the forward to MODERN INDUSTRIAL STATISTICS by Kenett and Zacks, Duxbury, 1998*From the forward to MODERN INDUSTRIAL STATISTICS by Kenett and Zacks, Duxbury, 1998
“Six Sigma is making war on defects”
Bill Smith, Motorola
“If an employee is not enthusiastic about Six Sigma, GE is simply not the right company for that person”
Jack Welch, General Electric
“If all we have is spirit, we will lose to the US”
President Idei, Sony
Some more Quotes
Even more Quotes
“Six-Sigma is remarkable – it has made managers start to adopt those simple and efficient methods that they have all needed desperately ever since they were developed back in the 1920s”
Translated from Oystein Evandt (Norway)
“Six-sigma’s focus on the bottom line provides the missing ingredient in Deming’s philosophy”
KnowledgeKnowledgeManagementManagement
The Six Sigma InitiativeThe Six Sigma Initiativeintegrates these effortsintegrates these efforts
Black Belt training programs may includeBlack Belt training programs may includeBlack Belt training programs may includeBlack Belt training programs may include
• 6 sigma principles• Quality Improvement• Quality by Design• Quality Control• Teamwork• Effective presentations• QFD/VOC • Statistical thinking• Process mapping• Barriers to breakthroughs• JMP, MINITAB…..
• Gage R&R• SPC• SPC Strategy• Risk Management• FMEA• Statistical Inference• Design Of Experiments• DOE Strategy• Bootstrapping• Robust Designs• System Thinking
Barrier #1: Engineers and managers are not interested in mathematical statistics
Barrier #2: Statisticians have problems communicating with managers and engineers
Barrier #3: Non-statisticians experience “statistical anxiety” which has to be minimized before learning can take place
Barrier # 4: Statistical methods need to be matched to management style and organizational culture
Barrier #1: Engineers and managers are not interested in mathematical statistics
Barrier #2: Statisticians have problems communicating with managers and engineers
Barrier #3: Non-statisticians experience “statistical anxiety” which has to be minimized before learning can take place
Barrier # 4: Statistical methods need to be matched to management style and organizational culture
Barriers to implementation
Technical Technical SkillsSkills
Soft SkillsSoft Skills
StatisticiansStatisticiansMaster Master
Black BeltsBlack BeltsBlack BeltsBlack Belts
Quality Improvement Quality Improvement FacilitatorsFacilitators
BBBBBBBBMBBMBB
Leadership Group
Processes, internal and external customers
Team 1 Team 2 Team 3
BBBBBBBB BBBBBBBB BBBBBBBB
MBBMBB
The The 6 Sigma6 Sigma Project Structure Project Structure
KPAISRUIBISENBISCAMT
Comparing three recent developments in
“Quality Management”
ISO 9000 (-2000) EFQM Model Quality Improvement and Six
Sigma Programs
ISO 9000
Proponents claim that ISO 9000 is a general system for Quality Management
The de facto applications seem to be – an excessive emphasis on Quality Assurance, and – standardization of already existing systems with
little attention to Quality Improvement It would have been better if improvement
efforts had preceded standardization
Critique of ISO 9000
Bureaucratic, large scale Focus on satisfying auditors, not customers Certification is the goal; the job is done when certified Little emphasis on improvement The return on investment is not transparent Main driver is:
– We need ISO 9000 to become a certified supplier, – Not “we need to be the best and most cost effective supplier to
win our customer’s business” Corrupting influence on the quality profession
EFQM Model
A tool for assessment: Can measure where we are and how well we are doing
Assessment is a small piece of the bigger scheme of Quality Management:– Planning – Control – Improvement
EFQM provides a tool for assessment, but no tools, training, concepts and managerial approaches for improvement and planning
The “Success” of Change Programs?
“Performance improvement efforts … have as much impact on
operational and financial results as a ceremonial rain dance has on the weather”
Schaffer and Thomson,Harvard Business Review (1992)
Change Management:Two Alternative Approaches
Activity Based Programs
Result Oriented Programs
ChangeManagement
Reference: Schaffer and Thomson, HBR, Jan-Feb. 1992
Activity Centered Programs Activity Centered Programs: The pursuit of
activities that sound good, but contribute little to the bottom line
Assumption: If we carry out enough of the “right” activities, performance improvements will follow– This many people have been trained– This many companies have been certified
Bias Towards Orthodoxy: Weak or no empirical evidence to assess the relationship between efforts and results
No Checking with Empirical Evidence, No Learning Process
ISO 9000
Data
Hypothesis
Deduction Induction
An Alternative: Result-Driven Improvement Programs
Result-Driven Programs: Focus on achieving specific, measurable, operational improvements within a few months
Examples of specific measurable goals:– Increase yield
– Reduce delivery time
– Increase inventory turns
– Improved customer satisfaction
– Reduce product development time
Result Oriented Programs: Project based Experimental Guided by empirical evidence Measurable results Easier to assess cause and effect Cascading strategy
Why Transformation Efforts Fail! John Kotter, Professor, Harvard Business
School Leading scholar on Change Management Lists 8 common errors in managing
change, two of which are: 1. Not establishing a sense of urgency2. Not systematically planning for and
creating short term wins
Six Sigma Demystified*
Six Sigma is TQM in disguise, but this time the focus is:– Alignment of customers, strategy,
process and people– Significant measurable business results– Large scale deployment of advanced
quality and statistical tools– Data based, quantitative
*Adapted from Zinkgraf (1999), Sigma Breakthrough Technologies Inc., Austin, TX.
Keys to Success*
Set clear expectations for results Measure the progress (metrics) Manage for results
*Adapted from Zinkgraf (1999), Sigma Breakthrough Technologies Inc., Austin, TX.
Six Sigma
The precise definition of Six Sigma is not important; the content of the program is
A disciplined quantitative approach for improvement of defined metrics
Can be applied to all business processes, manufacturing, finance and services
Focus of Six Sigma*
Accelerating fast breakthrough performance Significant financial results in 4-8 months Ensuring Six Sigma is an extension of the
Corporate culture, not the program of the month
Results first, then culture change!
*Adapted from Zinkgraf (1999), Sigma Breakthrough Technologies Inc., Austin, TX.
Six Sigma: Reasons for Success
The Success at Motorola, GE and AlliedSignal has been attributed to:
– Strong leadership (Jack Welch, Larry Bossidy and Bob Galvin personally involved)
– Initial focus on operations – Aggressive project selection (potential savings in
cost of poor quality > $50,000/year)– Training the right people
The right way!
Plan for “quick wins”– Find good initial projects - fast wins
Establish resource structure– Make sure you know where it is
Publicise success– Often and continually - blow that trumpet
Embed the skills– Everyone owns successes
RoastRoast
CoolCool
GrindGrind
PackPack
Coffeebeans
Sealed coffee
Moisture content
Moisture content
Savings:-Savings on rework and scrap-Water costs less than coffee
Potential savings:500 000 Euros
Case study: project selection
Measure
1. Select the CTQ characteristic
2. Define performance standards
3. Validate measurement system
Case study: Measure
Measure
Moisture contents of roasted coffee
1. CTQ
- Unit: one batch- Defect: Moisture% > 12.6%
2. Standards
Case study: Measure
Gauge R&R studyGauge R&R study
3. Measurement reliability
Measurement system too unreliable!
Case study: Measure
So fix it!!
Analyze
4. Establish product capability
5. Define performance objectives
6. Identify influence factors
Case study: Analyze
USLUSL
USLUSL
Improvement opportunities
CT
Q
CT
Q
CT
Q
CT
Q
Diagnosis of problem
-Brainstorming-Exploratory data analysis
6. Identify factorsMaterialMachineMan
Method Measure-ment
MotherNature
Amount ofadded water
Roastingmachines
Batchsize
Reliabilityof Quadra Beam
Weatherconditions
Moisture%
Discovery of causes
Control chart for moisture%
Discovery of causes
- Roasting machines (Nuisance variable)
- Weather conditions (Nuisance variable)
- Stagnations in the transport system (Disturbance)
- Batch size (Nuisance variable)
- Amount of added water (Control variable)
Potential influence factors
A case study
Improve
7. Screen potential causes
8. Discover variable relationships
9. Establish operating tolerances
Case study: Improve
- Relation between humidity and moisture% not established
- Effect of stagnations confirmed
- Machine differences confirmed
7. Screen potential causes
Design of Experiments (DoE)
8. Discover variable relationships
Case study: Improve
Experiments are run based on: IntuitionKnowledgeExperiencePowerEmotions
Possible settings for X1
Po
ssible se
ttings fo
r X2
X: Settings with which an experiment is run.
X
X
XX
X
X
X
Actually:• we’re just trying • unsystematical• no design/plan
How do we often conduct experiments?How do we often conduct experiments?
Experimentation
A systematical experiment: Organized / disciplineOne factor at a timeOther factors kept constant
Procedure:
XX XX OX X X X X
X: First vary X1; X2 is kept constant
O: Optimal value for X1.
X: Vary X2; X1 is kept constant.
: Optimal value (???)
X
X
X
X
X
X
X
Possible settings for X1
Po
ssible se
ttings fo
r X2
Experimentation
One factor (X)
low high
X1 21
Two factors (X’s)
low
high
high
X2
X1
22
high
Three factors (X’s)
low highX1
X3
X2
23
Design of Experiments (DoE)
Experiment:
Y: moisture%
X1: Water (liters)X2: Batch size (kg)
A case study: Experiment
Feedback adjustments for influence of weather conditions
A case study
9. Establish operating tolerances
A case study: feedback adjustments
Moisture% without adjustments
A case study: feedback adjustments
Moisture% with adjustments
Control
10. Validate measurement system (X’s)
11. Determine process capability
12. Implement process controls
Case study: Control
long-term < 0.280
Objective
long-term = 0.532
Before
long-term < 0.100
Result
Results
Benefits of this project
long-term < 0.100
Ppk = 1.5This enables us to increase the mean to 12.1%
Per 0.1% coffee: 100 000 Euros saving
Benefits of this project:
1 100 000 Euros per year
Benefits
Approved by controller
- SPC control loop- Mistake proofing- Control plan- Audit schedule
12. Implement process controls
Case study: control
- Documentation of the results and data.
- Results are reported to involved persons.
- The follow-up is determined
Project closure
- Step-by-step approach.
- Constant testing and double checking.
- No problem fixing, but: explanation control.
- Interaction of technical knowledge and experimentation methodology.
- Good research enables intelligent decision making.
- Knowing the financial impact made it easy to find priority for this project.
Six Sigma approach to this project
Re-cap I!
Structured approach – roadmap Systematic project-based improvement Plan for “quick wins”
– Find good initial projects - fast wins Publicise success
– Often and continually - blow that trumpet Use modern tools and methods Empirical evidence based improvement
Re-cap II!
DMAIC is a basic ‘training’ structure Establish your resource structure
– Make sure you know where external help is
Key ingredient is the support for projects
- It’s the project that ‘wins’ not the training itself
Fit the training programme around the company needs
– not the company around the training
Embed the skills– Everyone owns the successes