You are on page 1of 325

Lean-Sigma Black Belt

Development program
Week 4 booklet for trainees
Devenir Green Belt

August 2015 1
CODE: GROUND RULES

Switch off your mobile phone !

Start on time !

Straight to the point !

Let people finish their thoughts !

No side conversations !

Everyone participates !

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
2
Designed by Safran Consulting
CHECK LIST BEFORE TAKE OFF

 Timing
 Start ?
 End ?
 Constraints for the last day

 Lunch
 Time & duration
 Logistics (where, how)?

 Breaks
 How many?
 How long?

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
3
Designed by Safran Consulting
TIPS

 Mark your books with your name

 Keep your name-tag

 Any concern… ask the facilitator!


This document and the information it contains belong to
Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
4
Designed by Safran Consulting
ROLES

Who is the
Time Keeper ?

Who maintains the


parking lot ?

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
5
Designed by Safran Consulting
SUMMARY

1. Coaching

2. Statistics for measure phase

3. Design of experiments and multiple linear regressions

4. DFSS and manufacturability

5. Road map projects

6. Case study

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
6
Designed by Safran Consulting
NOTES AND ADDITIONAL EXPLANATIONS

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
7
Designed by Safran Consulting
AGENDA WEEK 3

Day 1 Day 2 Day 3 Day 4

Introduction &
feedback on
Green Belt content
MEASURE & ANALYZE MEASURE & ANALYZE
Lean
Hypothesis Hypothesis
Transformations
testing testing
Lean Transformations

Lunch Lunch Lunch Lunch

EXERCISES
Practice with Minitab on
MEASURE & ANALYZE a simulation
Flow chart Hypothesis
Lean Transformations
Practical application testing SAMPLING
Power

Conclusion

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
8
Designed by Safran Consulting
AGENDA WEEK 4

Day 1 Day 2 Day 3 Day 4

Introduction &
ANALYSE
reminders week 1
Design of Experiments
MEASURE
(DoE)
Long term capability CASE STUDY
Role of BB manage non-normal DMAIC
distributions Exercise
Coach
Green Belts Multiple Regression
Rituals and Posture

Lunch Lunch Lunch Lunch

DFSS approach &


Coach manufacturability
Green Belts Introduction CASE STUDY
Rituals and Posture ANALYSE DMAIC
Design of Experiments Exercise
(DoE)
Coach LEADING A TEAM
Green Belts Roles in a successful
Planning your project team
Conclusion

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
9
Designed by Safran Consulting
ROLE OF A BLACK BELT

This document and the information it contains belong to Safran.


They must not be copied or communicated to any other person
without prior written authorisation from Safran.
BLACK BELT ROLE (AS A DEDICATED IMPROVEMENT PERSONNEL)

 Leadership
 Must demonstrate Leadership model behaviors
 Leadership of complex and high impact projects
 Challenges the sponsors by giving him support / coaching in his sponsor role
 Facilitates link between Sponsor & GB (remove blockers)
 According to company/location: may assume a MBB role
 Accompanies managers & sponsors to built their progress plan, implement S+ tools
 Supports the MBB in some of his/her missions
 Assumes MBB like role in his area of responsibility

 Coaching
 Mentors and coaches Green Belts
 Ensures referent role vs. tools
 Networking and knowledge creation/ sharing
 Participates actively to S+ networks (RCE, events, animation of local networks…)
 Supports Safran group improvement & excellence methodologies deployment: working
group, training and coaching implementation
 Always demonstrates & fosters the search and application of state of art solutions (better
way of working, benchmarking, challenging …)

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
11
Designed by Safran Consulting
BLACK BELT ROLE (WHEN BACK TO OPERATIONS)

 Leadership
 Strongly promotes the S+ approach in his area of responsibility and within
SAFRAN
 Becomes a model sponsor (when manager)

 Coaching
 Coach 1 GB per year (depending on Safran+ needs/ requirements)

 Networking and knowledge creation/ sharing


 Continues to follow and participate to the S+ community activities

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
12
Designed by Safran Consulting
COACHING A GREEN BELT

This document and the information it contains belong to Safran.


They must not be copied or communicated to any other person
without prior written authorisation from Safran.
WHAT IS GB COACHING

 Coaching is the support a BB provides to a GB (project manager)

 Coaching aims to
 develop the project manager’s autonomy (including from the Coach)
 find and apply appropriate solutions
 support the project leader in the use of their own resources

 Over a given period and within a fixed framework, the Coach


 Supports the project manager in attaining their set goals
 The Green Belt is responsible for the project’s results
 The Coach ensures the usage of the methods
 Practices active listening, reformulation and open questioning
 Examines the situation and assists in the research of options

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
14
Designed by Safran Consulting
WHY COACH?

 You must guide Green Belts in


 Choosing and using the tools
 Applying the method
 Being pro-active

 For Safran, you are also


 One of the pillars of the deployment of Lean Sigma
 Accountable for correct usage of the methodology to projects

 So, you must


 Be able to answer questions on methodology
 Notify the Green Belt and then the sponsor if the Lean Sigma agenda or
methodology is not properly followed
 Ensure that the tools are correctly used (the right tool, the right use)

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
15
Designed by Safran Consulting
WHY COACH? (CONTINUED)

 Offer another perspective


 Take a step back
 Check that all improvement possibilities are considered and analyzed
 Give the Green Belt a more strategic overview

 Identify possible connections between projects


 from your awareness of other projects underway

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
16
Designed by Safran Consulting
THE 5 AXES OF THE SAFRAN LEADERSHIP MODEL ARE BASED ON THE
SUCCESS OF OUR PROJECTS

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
17
Designed by Safran Consulting
THE 5 AXES OF THE SAFRAN LEADERSHIP MODEL ARE BASED ON THE
SUCCESS OF OUR PROJECTS

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
18
Designed by Safran Consulting
SOME EXAMPLES OF THE BENEFITS OF COACHING

 Offer a fresh eye to the project

 Help the project manager build their vision for the project

 Expand the range of possibilities

 Learn from errors

 Motivate

 Develop

 Strengthen newly acquired skills

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
19
Designed by Safran Consulting
THE 3 STAGES IN THE COACHING PROCESS

Closing the
Coaching at the
Coaching during the project project and
start of the project
the Coaching

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
20
Designed by Safran Consulting
1. STARTING WELL
This document and the information it contains belong to Safran.
They must not be copied or communicated to any other person
without prior written authorisation from Safran.
ROLES AND RESPONSIBILITIES

 Responsibilities of the GB coachee  Responsibilities of the BB Coach

 Head of project  Does not take over the GB’s role in


the project

 Use the tools from training  In charge of quality and the progress
 Obtain certification of the Coaching
 Acquire autonomy
 Be available for the project and  Committed to assisting the GB during
Coaching the entire project to help in:
 acquiring autonomy
 preparing for certification
 choosing the appropriate tools

 At the end of the Coaching, the GB  At the end of the Coaching, the BB
actively participates in the formal conducts a formal Coaching report
Coaching report with the BB with the GB

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
22
Designed by Safran Consulting
20’

You meet your coachee for


the first time …

1. Do the role play 2 by 2


Person A acts as the Coach and
Person B as the coachee (10’)

2. Debriefing (10’)
Note taking

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
24
Designed by Safran Consulting
WHO IS THE BB COACH?

 The Coach is a Black Belt or Black Belt trainee


 Trained (Black Belt and Coaching)
 Knows the Coaching good practices
 Has been in charge of at least 1 project as the project manager

 He deals with the GB formally

 At the start of the project, he structures the Coaching with the GB  “Coach
contract”
 Frequency of reviews
 Method of communication and of checking what the GB has learned
 Stages in Coaching
 If preferred, the BB can be accompanied by an MBB: whenever “guidance” is
needed.
 At the end, an opinion on the methodology section of the project must be given

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
25
Designed by Safran Consulting
Note taking

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
26
Designed by Safran Consulting
30’

Back to role playing


You meet your coachee for
the first time …
Reverse the roles for A and B

B = Black Belt role


• Explains how the Coaching will be carried
out

A = Green Belt role


• Listens to the BB in explaining the Coaching
program
• Asks questions for better
understanding/clarification if needed
Note taking

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
28
Designed by Safran Consulting
30’

What are your 3 essential


questions on the project to
ask the coachee?

In groups of 4, identify the TOP 3 questions to


ask at the start of project

Debriefing
Note taking

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
30
Designed by Safran Consulting
ROAD MAP 1
PROJECT FRAMEWORK

 In addition to framing the Coaching, it is important to frame the project

 The BB assists the GB in framing the project


With the GB, he assesses
 project feasibility (6 pack)
 the success factors
 sponsor involvement
 GB availability
 project team availability

 He validates passport compliance


 if it complies, he signs
 if not, he informs the MBB and, where necessary, supports the GB in
discussions with other signers

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
31
Designed by Safran Consulting
Note taking

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
32
Designed by Safran Consulting
This document and the information it contains belong to Safran.
They must not be copied or communicated to any other person
2. DURING THE PROJECT:
without prior written authorisation from Safran.
HOW TO COACH

 Important point
 Some project managers do not want to be instructed or watched.
 Others appreciate this kind of help
 You will have to reconcile the degree of autonomy on methodology,
depending on each individual’s nature and personality.

 The 3 types of Coaching sessions


 Formal exchanges
 Informal exchanges
 Sideline assistance

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
34
Designed by Safran Consulting
FORMAL EXCHANGES TO MOVE THE PROJECT FORWARD

 Arrangements
 Scheduled meetings of 1-2 hours with the GB

 At least once every two weeks


 Even with project managers who know the
method

 Analyze the work done during the period


 The content form and difficulties encountered

 Help the coachee plan their actions for the


next period
 The project manager must have a clear vision of
the activities to do for the next period

 Where specific difficulties exist, you agree


on
 actions to resolve them
 the people in charge of those actions
 when they are to be implemented

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
35
Designed by Safran Consulting
FORMAL EXCHANGES
GOOD PRACTICES

 Ideal conditions either face-to-face or over the phone


 2-person meeting
 You are available for as much time as necessary
 Relaxed and without disturbances (interruptions, phone calls…)
 With network access (to work files, Minitab…)
 Use a standard format for the follow-up

 Points to Watch
 Critical eye on the project’s agenda and the regular holding of project
meetings.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
36
Designed by Safran Consulting
THE COACH’S TOOLS/TEMPLATES

Done

Planned but missing Mentee Jean CEINTURE


program Project Reduce employee travel time
Summary of meeting Reduce time passed by 2h - Time remaining: N / A
Review of the actions foreseen at the previous session
Day

Week the milestone of the MEASURE Phase was happening. The project team
Month approached the validity of the measurements. The session allowed to work on
Person 1 techniques of animation of brainstorming
Person 2
Person 3 Next steps
Person 4 A1. Analyze Root Causes on ...
Person 5 A2. Create Transport Time Charts vs Transport Type
Person 6 …
Person 7 Next meeting: August 4th
Person 8
Person 9
Person 10 Current Stage: Analysis Process in current stage : 10%

Schedule for Coaching Model of Coaching minutes

Agenda
Review of the actions of the last
meeting
Follow-up
model for
Hard points or difficulties note taking
encountered

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
37
Designed by Safran Consulting
FOLLOW-UP FILES
SOURCE: TOOL KIT

Summary to be sent by the coachee

Mentee Jean CEINTURE


Project Reduce employee travel time
Summary of meeting Reduce time passed by 2h - Time remaining: N / A
Review of the actions foreseen at the previous session

the milestone of the MEASURE Phase was happening. The project team
approached the validity of the measurements. The session allowed to work
on techniques of animation of brainstorming

Next steps
A1. Analyze Root Causes on ...
A2. Create Transport Time Charts vs Transport Type

Next meeting: August 4th

Current Stage: Analysis Process in current stage : 10%

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
38
Designed by Safran Consulting
EXAMPLE OF NOTE TAKING
SOURCE: TOOL KIT
Agenda Discussion/Decision/Actions
Review of last meeting's discussion Ok
Difficulties encountered
Coach notes and key questions
Respect of project planning Ok
Activities for next period Sustainability Allows
work beforetothekeep
end oftrack of discussions to
the project
follow
- Formalizing what the done
has been sequence during
(standardization, Termstheof project
Reference) and is the basis for the final Coaching
- Measure of results ==> to transform into financial gains
report.
- Reunion purchase to open the debate on waste generations (this
subject will not be treated in itself but will remain to treat post-
project)
-Cloture meeting (with all the team for the good and bad things of
the project) + closure review with xxx (organize a visit to atalier)
-See for the celebration

Ambience and efficiency of the team Last team meeting


Availability of bb / gb on his project It begins to be hard ...
Method and tools
Alert (if necessary)

Duration of meeting 1 hr Remaining balance: N/A

Next meeting Date: July - 22 Hours: 11 Hours

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
39
Designed by Safran Consulting
INFORMAL EXCHANGES

 Location
 Meeting in the corridor, at the
coffee machine or on the job

 Objectives
 Establish open and
spontaneous communication
between you and the project
manager (or team members).
 Detect their state of mind and
possibly adapt the Coaching as
a consequence

 Note: Does not replace formal


exchanges

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
40
Designed by Safran Consulting
SIDELINE ASSISTANCE OF THE COACHEE

 On the project manager’s request, you


can be brought to assist by their side:
 Initiate a tool that the project
manager does not feel comfortable
with: VOC, IPO, PRIO, FMECA, …
 Note: always remember you are only
acting as support

 With the project manager, prepare in


advance your style of intervention
 At what moment do you take the
reins? For how long?
 Do not hesitate in giving back control to the
project manager once the team understands the
tool and is working independently.
 Does the project manager prefer
remarks during a team meeting or
during a debriefing session?
 The project manager must retain (or
strengthen) his position as project
manager.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
41
Designed by Safran Consulting
Note taking

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
42
Designed by Safran Consulting
30’

Your coachee informs you of a


delay with the project…

1. Role play
A = Black Belt role
B = Green Belt role

2. Debriefing
Note taking

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
44
Designed by Safran Consulting
THE 3 X’S OF A COACH
PROJECT MONITORING

3. Feedback

2. Questioning
and help in
discovering

1. Active
listening

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
45
Designed by Safran Consulting
COACHING SKILLS
ACTIVE LISTENING

 Listening is a fundamental skill for managers and coaches. Effective


communication depends on the quality of listening and needs constant
practice.

 Conditions for listening


 Remain silent and do not interrupt the speaker
 Listen to their point of view without judging
 Be curious
 Reduce your own internal dialog (turn off the mind chatter)
 Be in sync with the speaker

 Reformulate = message received


 Reformulate, i.e., repeat to the speaker what you understood
 Helps the coachee explore the problem themselves

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
46
Designed by Safran Consulting
COACHING SKILLS
RELEVANT QUESTIONING

 Relevant questioning is one way to lead a meeting

 Asking to better understand


 Closed questions lead the other person to take a position (yes or no) or to engage in discussion
 Open questions allow complete freedom to the other person
 to express a point of view
 to explore the subject

 Examples
 Discover what the situation is by asking
 What happened?
 What are the important points?
 What do you think?
 Examine the coachee’s expectations by asking
 What is your main goal?
 What problems did you come across?
 What are their consequences?
 Find out what the person thinks and feels by asking
 What do you think?
 What is motivating you?
 What do you lack ?

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
47
Designed by Safran Consulting
Note taking

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
48
Designed by Safran Consulting
30’

Back to role playing


Your coachee informs you of a delay
with the project…
1. Role play
A = Black Belt role
B = Green Belt role

2. Debriefing
COACHING SKILLS
THE PRACTICE OF FEEDBACK

Definition Goal

Feedback is an exchange of
Feedback assists in improving
suggestions between two
performance through relevant
people based on
and practical suggestions.
observations

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
50
Designed by Safran Consulting
COACHING SKILLS
HOW TO PROVIDE EFFECTIVE FEEDBACK?

To provide effective feedback, you must


be convinced of its benefits
Why do people not
give feedback?  Feedback is a gift and a right; it:
- Helps frame expectations and priorities
• Fear of confrontation - Erases the fear of the unknown
- Helps to fill skills gaps
• Time constraints - Shows the fields in which to take
corrective measures
• Bad experiences
 An accomplished organization
in the past
requires accurate and regular feedback

• Fear of hurting or damaging a  Feedback has advantages for the company:


relationship employees know they are on the right track

 On the other hand, the price paid for mediocre or


• Fear of not being liked non-existent feedback is high
 Arbitrarily-defined performance criteria
• Fear of not understanding  Confusion
or of being misinterpreted  Loss of the investment in time
 Anxiety
 Decline in work quality

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
51
Designed by Safran Consulting
ONE POSSIBLE METHOD FOR
GIVING FEEDBACK

“I noticed something going on between you


Start with the facts and so-and-so”, “that you had some
words…”

“I felt uneasy because…”


Give your perspective on how those facts
have affected you

Count slowly to three while looking at the


Allow a moment of silence for this to sink person
in; listen to the other, make sure the other
has accepted the facts and the impact on
you before continuing

“Here’s what you could have done to avoid


Offer a suggestion on what could have such a reaction…”
been done differently

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
52
Designed by Safran Consulting
60’

The GB is currently working on


solutions based on the initial
measures. The MEASURE
phase has not yet been
approved. What do you say?

1. Role play
A = Black Belt role
B = Green Belt role

2. Debriefing

3. Reverse the roles


Note taking

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
54
Designed by Safran Consulting
DURING THE PROJECT:
SUMMARY

 The BB verifies the GB’s progress with the project


 keeps himself up to date with the project through the GB
 helps to identify and categorize the GB’s difficulties in the project
 helps to identify and categorize the way the difficulties are solved or not
 helps to identify the comfort zones and areas of progress of the GB
 checks the use of tools (hard and soft) from the GB’s training (choice, use, mastery)

 The BB regularly identifies the GB’s needs


 support in the assimilation of training modules
 support in their implementation within the project framework
 according to the needs identified by the BB:
 he himself assists or
 he turns to an MBB

 The BB ensures the proper working of the Coach program


 He ensures that the GB writes up and returns the session’s CR

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
55
Designed by Safran Consulting
Note taking

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
56
Designed by Safran Consulting
This document and the information it contains belong to Safran.
They must not be copied or communicated to any other person
3. PROJECT TOUCHDOWN
without prior written authorisation from Safran.
40’

Your coachee asks to start the


certification process before the
Control phase has finished
A = Black Belt role
• Notes that the GB is not doing any Control phase
actions
• Tells him firmly that it cannot be certified and argues
the point
• Following GB’s explanations, continues to reiterate that
a GB must complete the C phase of the project
• Then, listens to the GB
• At the end, talks about how he felt being in this
conflict situation

B = Green Belt role


• Attempts to convince the Black Belt that the project is
completed
CONFLICT RESOLUTION

Can’t compromise Compromise at 2nd


at 1st level level

Possible
I want to get solution
to my new

Area for innovation


position
Why do you
want to
leave? Possible
solution

What do you Possible


want to get solution
from the
I want GB to project?
finish the
project Possible
solution

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
59
Designed by Safran Consulting
PROJECT CLOSURE IS EXPECTED FROM THE BEGINNING

 The BB prepares the GB for the project’s end by:


 Identifying the remaining tasks and those responsible (GB or other)
 Preparing and passing responsibility over to the project recipients
 allocating objectives for the future
 anticipating possible difficulties
 ensuring their autonomy

 The BB checks the “Control” stage with the GB:


 He ensures that the GB has performed the knowledge transfer
 He makes sure that the GB has thought of the support measures for the
change
 He ensures that the GB has scheduled the updating of function/document
procedures.

9
This document and the information it contains belong to
Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
60
Designed by Safran Consulting
COACHING CLOSURE

 The BB remains in his role and turns to an MBB if needed


 Remains Coach as long as the project is not completed (or abandoned)
 He does not “catch” nor “rescue” an incomplete project or one in danger of failing
 He teaches how to close a project and finish the Coaching

 The BB closes the Coaching


 At the end of the Coaching, the BB conducts a formal Coach report with the GB
whom he requests to be an active participant in the report

 They review the project and Coaching stages


 He emphasizes the strong points of the GB, the improvements made and the steps
still to be taken.

 The BB provides feedback to the GB regarding the Coaching


 The GB provides feedback to the BB regarding the Coaching

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
61
Designed by Safran Consulting
This document and the information it contains belong to Safran.
They must not be copied or communicated to any other person
CONCLUSION
without prior written authorisation from Safran.
COACHING ALSO BENEFITS THE COACH

 Accelerated expansion of knowledge


 On an organizational level, by the numerous projects Coached
 On a technical level, by the many tools used in the projects of the
coachees

 Development of behavioral tools useful for project managing and


management itself
 Active listening
 Reformulation
 Relevant questioning
 Feedback
 Learning a new autonomy process
 Framing the Coaching
 Support Project framework
 Support Project monitoring
 Assist in Project closure
 Coaching closure

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
63
Designed by Safran Consulting
Process Capability
SHORT AND LONG TERM VARIABILITY

64 /
Ce document et les informations qu’il contient sont la propriété de Snecma. Ils ne doivent pas être copiés ni communiqués à un tiers sans l’autorisation préalable et écrite de Snecma.
GAUSSIAN CURVE

68%

95%

99.73 %

m - 3 Mean -  Mean +  m+3


m-2 Mean m+2

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 65
Designed by Safran Consulting
2 TYPES OF VARIABILITY

Specifications
Target
1
2

Time

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 66
Designed by Safran Consulting
CAPABILITY
INDEXES 1
2

Time

 Usual indexes for capability (based on the tolerance)

Tolerance
Cp  1
Short term variability

Tolerance
Pp 
Long term variabili ty 2

Min (distance between mean and specificat ions)


Ppk 
Half long term variabili ty 2
Rq: variability calculation depends on the distribution.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 67
Designed by Safran Consulting
CAPABILITY INDEXES (CONT)

 Non Conformity Rate (NCR)


 For Discrete data we can use bad parts ratio

 For Continuous data, we may use the distribution

NCR

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 68
Designed by Safran Consulting
CAPABILITY EVOLUTION

Usage of long term variability instead


Cp of short term

Pp Decreases with the offset

Ppk

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 69
Designed by Safran Consulting
RELATIONSHIP BETWEEN
LONG TERM NCR & PP/PPK

Ppk 0.7 1 1.1 1.3


Pp

0.7 35 729
0.9 18 348
1 17 912 2 700
1.1 17 868 1 509 967
1.3 17 864 1 351 487 96
1.5 17 865 1 350 483 48
2 17 864 1 350 483 48
2.6 17 864 1 350 483 48
Unit: Parts per million (ppm) Hypothesis: gaussian curve

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 70
Designed by Safran Consulting
PROCESS CAPABILITY
CUSTOMER SPECIFICATIONS

 The specifications of our customer are usually predetermined values


given by the customer.
 They used to express the limits of acceptance (lower and upper specification
limits).

 It is wise to check whether these limits are really what the customer needs
or wants.
 Too often these specification do not correctly represent the customers needs
and wants.

 Most of the times, customers have Upper and Lower specification limits
 Sometimes only an upper or a lower specification is given.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 71
Designed by Safran Consulting
10’

Before continuing, lets validate our


knowledge...

Process steps
• Answer the questions with true or false
• Debrief
QUIZ CAPABILITY: TRUE OR FALSE

Phrase Choice
The capability indexes are valid only if the statistical distribution is normal True - False
Long term capability is always greater than the short term capability True - False
Cp and Cpk are short-term capability indicators True - False
If the distribution is normal, we can calculate the percentage of failure if True - False
you know Pp and Ppk
Cp is always greater than or equal to Cpk True - False
For a normal distribution, there is a probability of 99.73% to find a value True - False
between average +/-3 sigma
One can calculate Cp, Cpk, Pp, Ppk from a same sample True - False

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
73
Designed by Safran Consulting
PROCESS PARAMETERS
IS OUR PROCESS STABLE ?

 Suppose we would shoot with the catapult. We have rubber band 1


coming form supplier A and rubber band 2 coming form supplier B.

 Both rubber bands are initially capable of shooting a golf ball 200 cm far
(with specific settings of the catapult).
 Rubber band 1 is of good quality, its elastic characteristics remain stable
over time.
 Rubber band 2 on the contrary, is of poor quality (not stable). Its strength
weakens very fast.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 74
Designed by Safran Consulting
PROCESS PARAMETERS
IS OUR PROCESS STABLE ?

Stable process Nonstable process

230 230
UCL=226.8 UCL=226.8
220 220

210 210

Distance (cm)
Distance (cm)

200 Mean=199.8 200 Mean=199.8

190 190

180 180
LCL=172.7
LCL=172.7 170 1
170 1 1 1111 1 111111111111 1
1 1 11 1
160 1 1 1 11 1 1 111 1 1
160 1
1 1 1 1 1 11111
150 1
150
0 100 200 300
0 100 200 300
Observation Number Observation Number

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 75
Designed by Safran Consulting
PROCESS PARAMETERS
IS OUR PROCESS STABLE ?

 To know whether a process is stable or not, we need to calculate the short


term and the long term standard deviation.

 The long term standard deviation S is calculated with the formula:

 X 
n 2

i X
s i 1
n 1

 The short term standard deviation  is calculated out of the moving range
of the data:
ˆ  mR d
2

Note: for short term variation, we use the range between 2


consecutives data points

for long term variation, we use the gap of each value compared
to the overall average.
This document and the information it contains belong to
Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
76
Designed by Safran Consulting
PROCESS CAPABILITY
MOVING RANGE

 The moving range is the difference (in absolute value) between two
consecutive points.

 Example: lets look at only 5 consecutive shots from the catapult

181 199 209 199 200


mR
18 10 10 1

mR  (18  10  10  1) / 4  9.75

 We can now calculate the average moving range

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 77
Designed by Safran Consulting
PROCESS CAPABILITY

 Process parameters:
 Suppose we shot 300 shots with our catapult. The average is 200 cm. The process is
stable (the rubber band does not deteriorate in time): see previous graph.
 Let’s calculate the long term and the short term standard deviation for this dataset.

Long term standard deviation Short term standard deviation

 X 

n 2

i X
s i 1
 8.7 ˆ  mR d  9.71.128  8.6
n 1 2

 For stable processes the long term standard deviation is equal to the short term standard
deviation.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 78
Designed by Safran Consulting
PROCESS CAPABILITY

 Process parameters:
 Suppose we shot 300 shots with our catapult. The average is 200 cm. The process is not
stable (the rubber band deteriorates very fast in time): see previous graph.
 Let’s calculate the long term and the short term standard deviation for this dataset.

Long term standard deviation Short term standard deviation

 X 

n 2

i X ˆ  mR d  9.71.128  8.6
s  12.8
i 1 2

n 1
 For non-stable processes the long term standard deviation
IS NOT equal to the short term standard deviation.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 79
Designed by Safran Consulting
PROCESS CAPABILITY

Stable process Non Stable process


230 230
UCL=226.8 UCL=226.8
220 220


210 210

Distance (cm)
Distance (cm)

200 Mean=199.8 200 Mean=199.8

190 190

180 180
LCL=172.7 LCL=172.7
170 170 1 1 1 1111 1 1111111111111
1 11 11 11 1 11 1
160 160 1 11 1 1 1 11
1 11 1 1 111 1
150 11
150
0 100 200 300 0 100 200 300
Shot Number Shot Number

Moving Range Chart for Stable Moving Range Chart for Nonstabl

1 1
UCL=31.64 UCL=31.58
30 30

 Moving Range
Moving Range

20 20

10 R=9.665
10 R=9.685

0 LCL=0
0 LCL=0

0 100 200 300


0 100 200 300
Observation Number
Observation Number

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
80
Designed by Safran Consulting
CAPABILITY INDEXES: CP & CPK

 Based on the short term variability

INDEX LABEL FORMULA COMMENT

USL  LSL  The process average is


Potential

Short Term (ST) not taken into account.


Cp 
Capability Index 6

CpL
Lower ST X  LSL

 USL is not taken into
account
Capability Index
3

CpU Upper ST USL  X  LSL is not taken into


account
Capability Index 3

 
Current

CpK Short Term This is the complete


min C pL ,C pU
Capablity Index “picture”.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 81
Designed by Safran Consulting
PERFORMANCE INDEXES: PP & PPK

 Based on long term variability

INDEX LABEL FORMULA COMMENT


The process average is
USL  LSL 
Potential

Pp Performance index not taken into account.


6S

PpL Lower performance index


X  LSL  USL is not taken into
account
3S

PpS
Upper performance index
USL  X  LSL is not taken into
account
3S

  This is the complete


Current

PpK min PpL , PpS “picture”.


Performance index

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 82
Designed by Safran Consulting
PAST PERFORMANCE & PREDICTION

 When we are calculating Cpk, we take an image of the process.


 “What could this process do if it remained stable ?”

 When we compute Ppk, we ask the question:


 “What has this process done ?”

 Using the observed historical data, combined with the assumption that data
are normally distributed, we try to make an estimate of the future
performance or capability of the process.

 Even if a process has never delivered products out of specification, it is


possible to determine that the same process will create some defects in the
future.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 83
Designed by Safran Consulting
20’
Let's take the case of the catapult. The
objective is to shoot balls of squash in
a target between 170 and 230 mm of
the catapult. We have 3 process to
compare

Process steps
• Be aware of the information available
for each process and calculate (without
minitab) Pp and Ppk for each process
• Conclude
• Debrief
CALCULATION OF THE INDICES

 The 3 following processes are:


 Stable (standard deviations short and long-term are identical)
 Normally Distributed

 Measured data:
 Process 1: average = 200 cm & Sigma Long Term = 9 cm
 Process 2: average = 200 cm & Sigma Long Term = 15 cm
 Process 3: average = 215 cm & Sigma Long Term = 9 cm

 Calculations:
Process Pp Ppk
1
------------ = ------------ =

2
------------ = ------------ =

3
------------ = ------------ =

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
85
Designed by Safran Consulting
20’
Now use samples to quantify the
stability of a process

Process steps
• Open 'Stable Non Stable.mtw' in
Minitab
• Analyze the graphics and capability for
the STABLE and Non-STABLE variables
• Upper and lower limits given by
customer are 230 and 170 respectively

• What are your conclusions?


CAPABILITY CALCULATION

 Use command:
 STAT>QUALITY TOOLS>CAPABILITY ANALYSIS >NORMAL

Options allow you to view


different informations

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 87
Designed by Safran Consulting
CAPABILITY INDEXES WITH MINITAB

Units for defects


may be in % or ppm
Within  short term
Overall  Long term

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 88
Designed by Safran Consulting
Note taking

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
89
Designed by Safran Consulting
CAPABILITY INDEXES FOR « STABLE »

Short term = Long term

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 90
Designed by Safran Consulting
CAPABILITY INDEXES FOR « UNSTABLE »

Short term << Long term

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 91
Designed by Safran Consulting
CAPABILITY ANALYSIS REMINDERS

 Analyzing data
 Pp = Ppk  Process is centered long term
 Cp = Cpk  Process is centered short term
 Cpk = Ppk  Process is stable
 Cpk > 1  Capable short term
 Ppk > 1  Capable long term

 Minitab
 Cp and Cpk are done with moving average
 Pp and Ppk are done the classical way

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 92
Designed by Safran Consulting
SHIFTING THE MEAN

 If we know the rubber band deteriorates over time, we can improve the
capability and performance of the process by changing the settings of the
catapult (+15 cm). We are centring our process.

 In this way the new average will be 200 cm (185 cm +15 cm). However, the
long term standard deviation will always be bigger for a non stable process
then the short term standard deviation. So Ppk, will be lower then Cpk.

 On the next page we see how this new (non stable) process performs in
relation to the customer specifications.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 93
Designed by Safran Consulting
SHIFTING THE MEAN

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 94
Designed by Safran Consulting
30’
The knowledge of the short and long term
capability indicators is a source of information
for the ANALYZE Phase
Process steps
Establish groups of 2-3 people
• Define groups rather ADMINISTRATIVE and
other PRODUCTION
• Conduct a brainstorming session on the
following topic:
"What are the factors that have an impact on
the short-term variation? “

• Renew for Long Term variability


• Compare and conclude
SHORT TERM Influential factors LONG TERM Influential factors

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
96
Designed by Safran Consulting
CONCLUSION FOR THE CAPABILITY

 Capability indicators require normal distributions to


be interpretable

 There are 2 families of capability indicators


 The short-term capability describes the
potential of the process if it remains stable over
time
 The long-term capability describes the overall
current performance (viewed by customers)

 The analysis of the indicators of capability


allows to:
 Check if the process is stable or not
 Check the potential of the process
 Narrow down the list of causes to investigate in
the ANALYZE phase (short and long term
variations are generated by different causes)

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
97
Designed by Safran Consulting
Note taking

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
98
Designed by Safran Consulting
Non-normal distributions
99 /
Ce document et les informations qu’il contient sont la propriété de Snecma. Ils ne doivent pas être copiés ni communiqués à un tiers sans l’autorisation préalable et écrite de Snecma.
NON-NORMAL DISTRIBUTIONS

 Importance of the normal distribution


 Many statistical tools can be applied with relevance only if the distribution is
normal.
 Control charts / Capability / Calculation of sigma level
 Hypothesis testing: t-test, paired t-test, ANOVA, regression ...
 Design of Experiments

 What to do?
Causes of non-normality Possible Solution
The data indicates two or more groups of Understand / find the cause of these groups and
variables then analyze by subgroup
The nature of the characteristic (fatigue, Use the most suitable type of distribution (Weibull,
flow time, roughness, ...) lognormal, ...)
Clean the database and solve the cause of special
Outliers
causes
Other Cases Use a transformation (cf slides to come)

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
100
Designed by Safran Consulting
NON-NORMAL DISTRIBUTIONS
Histogramme de Distance (cm)

40

30

Effectif
20

10

0
310 320 330 340 350 360 370
Distance (cm)

Solution:
50 Stratifying by 70

60

40
type of elastic
50

Result: the two 40

Effectif
30
Effectif

30
20 distributions are
20

10
normal 10

0
0 310 320 330 340 350 360 370
310 320 330 340 350 360 370 Elastique Neuf
Elastique Usé

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
101
Designed by Safran Consulting
EXAMPLES OF NORMAL AND NON-NORMAL DISTRIBUTIONS

Comments
• The non-normality can result
in asymmetric data, the
impossibility of having data
less than a (truncated)
threshold, the character
either bimodal and uniform
distribution

• This type of distribution can


often be corrected via data
transformation

Source: assistant Minitab


This document and the information it contains belong to
Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
102
Designed by Safran Consulting
TRANSFORMING DATA

 How?
 Instead of using the raw data (Y) in our analyses, we will use processed
data.

YTrans  Y 

The form
changes

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
103
Designed by Safran Consulting
TRANSFORM THE DATA IN Y

 The  of the above formula may have different values between - 5 and +
5. Current values are:

 YTrans  Y 

-2 1 2
Y
-1 1 Note:
Y
When there is a negative lambda, specification limits need
-0,5 1 to be inverted when doing the transformation
Y
0 Log10 (Y )
0,5 Y
1 Y
2 Y2

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
104
Designed by Safran Consulting
20’

Carry out a case study for the


transformation of data

Process steps
• Open the file "reclamations.mtw"
• Conduct a descriptive study and transform
the data to make them normal
• What is the process capability knowing
that the upper specification limit is 20
days?
• Debrief
TRANSFORM THE DATA - USE MINITAB

 STAT > CONTROL CHART > BOX-COX TRANSFORMATION

Each claim is independent.

The data have therefore


been taken one by one (and
not in sub groups))

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
106
Designed by Safran Consulting
TRANSFORM THE DATA - USE MINITAB

 Specify the column in which you want to store for example transformed
data: "transfer (days)"

STAT > CONTROL CHART > BOX-COX


TRANSFORMATION > OPTIONS

You can let Minitab use optimal lambda or


enter a particular value of lambda in our
example,  = 0,3

The processed data will be automatically


stored in the column “new days".

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
107
Designed by Safran Consulting
NOTES PAGE

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
108
Designed by Safran Consulting
Solution
PROCESSING OF THE VARIABLE 'DAY'

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
109
Designed by Safran Consulting
Solution

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
110
Designed by Safran Consulting
ADDITIONAL EXERCISE

Temps de Cycle 2.mtw

 Is data normally distributed?

 Otherwise, transform the distribution.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
111
Designed by Safran Consulting
CONCLUSION OF NON-NORMAL DISTRIBUTIONS

 In the event of non-normal distribution,


the first step is to analyze data
 Outliers ?
 Bimodal distribution ?
 Characteristic type according to another
known distribution?

 Data transformation is to change the


values Y into Y
 The specifications LSL and USL then
become LSL and USL
 The analyzes, tests and regression will be
performed with the transformed data.
 If the transformation is not enough, we
must look for other causes of non-
normality

 In the case of capability tests, Minitab


directly transforms data
This document and the information it contains belong to
Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
112
Designed by Safran Consulting
Optimize your test strategy
DESIGN OF EXPERIMENTS
113 /
Ce document et les informations qu’il contient sont la propriété de Snecma. Ils ne doivent pas être copiés ni communiqués à un tiers sans l’autorisation préalable et écrite de Snecma.
ACTIVITIES OF ANALYZE PHASE - ANALYZE

Process Analysis
 The 7 wastes
Value added / No value
Capabilities Analysis

Graphic Analysis
Basic graphics
 Box plot / whiskers chart
 Correlation diagrams

Statistic Analysis
 Hypothesis testing
 Analysis of Variance (ANOVA)
 Design of Experiments
 Linear Regression
 other regressions Root Causes
 5 Why’s
 Fishbone / Ishikawa Diagram

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
114
Designed by Safran Consulting
WHY THE DESIGN OF EXPERIMENTS?

 We have already learned to process data  Design of Experiments is a method to


with the following conventional tools select and analyze a series of
 Run chart, control chart tests/experiments. They are intended to
 Correlation, regression, ANOVA  Select the "best trials" to launch
 Master the test conditions to get
the more information from tests
 Measure the influence of
parameters on one or more
responses
 Limitations of conventional tools
 Error sensitivity or completeness of the  Limitations of design of experiments
data  Requires to preselect the tests
 Sensitivity to other unmeasured  The cost of conducting certain
factors tests
 A difficult analysis if the original  Requires discipline in conducting
database is poorly constructed the experiments

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
115
Designed by Safran Consulting
LEARNING OBJECTIVES

Are Are NOT

Being able to use


DOE at a good level

Allowing a dialogue
Becoming an expert in stats
with an expert or a statistician
Limiting Lean-Sigma
Understanding the results
only to DOE
and being able to explain them
Doing tests/trials for
Being able to run this
the sake of doing tests/trials
type of project

Better understanding of DOE


module in Minitab

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 116
Designed by Safran Consulting
CLASSICAL WAY OF THINKING

Base Composition
1

Change one factor


2

If there is improvement,
keep the change and test
3 another factor

If the result doesn’t


improve, start again and try
4 another factor

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 117
Designed by Safran Consulting
LET’S USE THE CATAPULT EXAMPLE TO EXPLAIN NOTATIONS &
DEFINITIONS

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 118
Designed by Safran Consulting
THE CLASSIC (CONSERVATIVE FACTOR) APPROACH

Description Ball Hinge Arm Stop Shooting Distance Comments


Angle
Baseline
1. Initial - - - - 130 cm

The ball change improves the


2. Ball Test + - - - 190 cm shooting distance. We keep the
ball '+' to future trials

The hinge change does not give


a better result. It will remain '-' for
3. Hinge Test + + - - 160 cm the upcoming trials

The change of level to stop the


4. Arm Test + - + - 210 cm arm of the Catapult gives a better
result.

The angle '+' gives a better result.


According to these tests, the
5. Angle Test + - + + 220 cm largest shooting distance will be
obtained with the combination 5

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
119
Designed by Safran Consulting
WHAT ARE THE DISADVANTAGES ?

 Reliability?
 Usually we get a result that is a combination of unconscious important. (Is
that the best combination ? )
 We do not take into account the variability of the extent our findings

 Understanding the phenomenon?


 No idea / forecasts of untested combinations
 Little or no visibility into interactions ( coupled effects ) between factors .
 We do not know the influence of other factors not listed .

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
120
Designed by Safran Consulting
Design of Experiments
1. WHAT IS IT?
DESIGN OF EXPERIMENTS
WHAT'S THE POINT?

I X1 O
Y2
N U
P X2 Process P
U Y1 U
T X3 T
S S

Design of Experiments

X1 X1 X1
Q
< X2 > Y  X P  X  X R
X3 X2 Y1 1 1 2 3
X2 X2

Determine the important Identify Correlation Establish Quantify the relationships


factors relationships interactions between inputs and outputs

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
122
Designed by Safran Consulting
DESIGN OF EXPERIMENTS: CONTENT
• Identify settings
• Choose the approach
• Plan
THE
• Experiment
PROCESS • Analyze
• Conclude and verify
• Implement and consolidate

• Full factorial design

• ½ fractional factorial Plan


THE
• Factorial: other fractions TOOLS

• filtering

• Response Surface

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
123
Designed by Safran Consulting
Design of experiments
2. DESIGN OF EXPERIMENTS
FULL FACTORIAL DESIGNS

« To determine the behavior of a subject to disturbances, a passive observer is not


satisfactory. It is necessary to induce these disturbances

George E.P. Box

124 /
Ce document et les informations qu’il contient sont la propriété de Snecma. Ils ne doivent pas être copiés ni communiqués à un tiers sans l’autorisation préalable et écrite de Snecma.
FULL FACTORIAL APPROACH
WHAT’S FOR?

 Simultaneously change several factors.


 Often with two levels per factor (low - high)

 Determine the important factors and isolate the variations of common


causes using
 Trials randomization
 Repetitive trials

 Easy & simplified analysis of the results.


 Facilitated by Minitab

 Reduce the effects of uncontrolled factors during experimentation (also


called hidden factors)
 Random trials

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 125
Designed by Safran Consulting
CATAPULT EXAMPLE

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
126
Designed by Safran Consulting
DEFINITIONS

 Factors (Xs): Input Variable that we have chosen to study


 Hinge of the elastic,
 Starting Angle,
 Type of ball.

 Level of the Factor: Values or levels that we wish to test for one of the
factors
 Hinge: Position 1 or 2,
 Angle: 0° or 30°,
 Ball: squash or Basketball.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 127
Designed by Safran Consulting
DEFINITIONS (CONT.)

 Combination: Definition of each level for each factor. Ex:


 Hinge Position = 1 & Angle = 30° & Type of Ball = Basketball

 TRIAL: Experience by testing a combination of factors. Ex:


 1 shoot with Hinge Position = 1 & Angle = 30° & Type of Ball = Basketball

 Most of the time, we test several trials, at different moments, for the same
combination of factor levels.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 128
Designed by Safran Consulting
10’
We want to know if the temperature of
the cutting oil, cutting speed and the
operator have an influence on
roughness... For testing, we will test 2
temperatures: 22 and 28 degrees, for
speed: 30-40 m/min and the operator:
Jean & Paul

Process steps
• Fill in the fields on the next page by naming
settings, levels and combinations
COMPLETE THE FOLLOWING FIELDS

Factor(s) Levels of each factor

Response(s)

2 examples of combinations

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
130
Designed by Safran Consulting
FACTOR APPROACH: NOTATION

 Convention to design 2 factor levels:


 LOW level (or standard) = -
 HIGH level = +

Example

Id Hinge Ball Angle Id Hinge Ball Angle

1 Pos 1 Squash 0° 1 - - -

2 Pos 2 P-P 30° 2 + + +

Explicit Notation (not coded) Conventional Notation (coded)

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 131
Designed by Safran Consulting
NOTES AND ADDITIONAL EXPLANATIONS

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
132
Designed by Safran Consulting
10’

Use the catapult to practice...

Process steps
Based on the defined system previously can you:

• Establish all combinations:


• 3 factors: 1. Hinge / 2. Angle / 3. Ball with
two levels each
• In coded notation
• Translate these combinations into explicit
notation (uncoded):
• Hinge levels: Pos 1 - Pos 2
• Angle levels: 0 ° - 30 °
• Ball levels: Squash - Table tennis
EXPLICIT NOTATION (NOT CODED) CONVENTIONAL NOTATION CODED

Hinge Angle Ball Hinge Angle Ball

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
134
Designed by Safran Consulting
FULL FACTORIAL DESIGN 2 LEVELS ^ 3 FACTORS = 2 X 2 X 2 = 8 TRIALS

EXPLICIT NOTATION (NOT CODED) CONVENTIONAL NOTATION CODED


Hinge Angle Ball Hinge Angle Ball

Position 1 30 Squash - - -
Position 2 30 Squash + - -
Position 1 60 Squash - + -
Position 2 60 Squash + + -
Position 1 30 Basketball - - +
Position 2 30 Basketball + - +
Position 1 60 Basketball - + +
Position 2 60 Basketball + + +
Solution
This document and the information it contains belong to
Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
135
Designed by Safran Consulting
2K CONVERSION TABLE ACCORDING TO STANDARD ORDER
Note:

X1 Changes with every run


(- + - +)

X2 Double(redo) the previous run


Change every second run
(- - + + - - + +)

X3 Double(redo) the previous run


Change every 4th run
(- - - - + + + +)

Etc…

This is called the standard order.


During the experiment, the tests are
not done in this order. However, it is a
convenient way to establish all the
runs for a certain DOE.

# runs = 2k =2*2*…*2
 hence ‘factorial’

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
136
Designed by Safran Consulting
3 FACTORS SYSTEM: 3-D REPRESENTATION

 The easy way to represent a 3 factors system is the CUBE.

 Each of its corners represent a set of factors level


 1 corner = 1 trial
Trial Hinge Angle Ball

7 8 1 Position 1 30 Squash
2 Position 2 30 Squash

5 6 3 Position 1 60 Squash
4 Position 2 60 Squash
Angle

3 4
5 Position 1 30 Basketball
6 Position 2 30 Basketball
1 2
Hinge 7 Position 1 60 Basketball
8 Position 2 60 Basketball

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 137
Designed by Safran Consulting
ONE FACTOR AT A TIME OR DOE?
EXAMPLE WITH 3 FACTORS - SOLUTION

One factor at a time Design of Experiment


Angle

Angle
Hinge Hinge

= Coverage of the trials

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
138
Designed by Safran Consulting
3
DESIGN OF EXPERIMENT STEPS

139 /
Ce document et les informations qu’il contient sont la propriété de Snecma. Ils ne doivent pas être copiés ni communiqués à un tiers sans l’autorisation préalable et écrite de Snecma.
DESIGN OF EXPERIMENTS: THE APPROACH

Process Steps Deliverables


Think 1 • Ensure basics
Structure plan o Which response should be measured?
experience and confirm o Which factors should be analyzed?
the inputs / outputs o What approach are we choosing?
o Which factor levels?

Plan 2 • Define the test conditions


Master the experimental o Make the trials randomly
protocol o Plan the tests
Experiment 3 • Testing and Data Collections
Conduct the tests o Do the trials
o Collect the data

Analyze 4 • Translation of the statistical results in conclusions for the process


to extract useful o Analyze the Data
information from test o Make conclusions
o Check the conclusions

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
140
Designed by Safran Consulting
1
ENSURE BASICS

 What is the response to measure?


 The response to measure is the variable you want to improve in your project
Lean-Sigma

 Quiz: True or False?

Phrase Choice
Preferably, a response has continuous data True or False
There is only one response per experiment True or False
The response is defined with precision True or False
A response is an uncontrolled factor True or False

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
141
Designed by Safran Consulting
1
WHAT ARE THE FACTORS?

 In a DMAIC or DMADV project, design of experiments benefits from


filtering during the analyze phase
 Use the influential X or a first selection of prioritized X

X can be any
 Continuous variable
 Discrete Variable

Note: if your experience level is outside a Lean Sigma project, use the tools
of the DEFINE and MEASURE steps to identify the factors that you want to
study in the experimental design
• Use brainstorming to list all possible variables.
• Prioritize X test with the matrix causes / effects

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
142
Designed by Safran Consulting
WHICH LEVELS/SETTINGS TO CHOOSE FOR THE FACTORS?

 General Rule
 Use key values to clearly identify the effect of a factor

 The chosen levels should be based on


 Experience
 Physical properties
 Actual technical knowledge

Example:
A chemical factory wishes to study what the effect of the temperature of a batch
and the treatment time is on the size of the granules that they produce. To
identify the significant effects the project group used key values. Rather than
comparing the granules with sugar, they compared with one granule the same
size as the reactor.

The study group finally found the equation, but had to stop the production line
during 4 days to clean the reactor.
This document and the information it contains belong to
Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 143
Designed by Safran Consulting
NOTES AND ADDITIONAL EXPLANATIONS

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
144
Designed by Safran Consulting
EXCERCISE: RESPONSE, FACTORS AND LEVELS/SETTINGS

Ball

4
Stop-Position of the 3 Hinge of elastic
arm
2
4 3 1
5 2
Starting Angle
6 1

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 145
Designed by Safran Consulting
SYNTHESIS STEP "REFLECT"

 List of factors, levels and response

Arm Distance between catapult


# Hinge Point Ball Angle Position and point of impact

Type Factor Factor Factor Factor Response

1 Position 2 Squash 150 Position 2 -

2 Position 3 Basketball 170 Position 3 -

 Calculate the number of possible combinations:

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
146
Designed by Safran Consulting
4
PLAN THE EXPERIMENTS

147 /
Ce document et les informations qu’il contient sont la propriété de Snecma. Ils ne doivent pas être copiés ni communiqués à un tiers sans l’autorisation préalable et écrite de Snecma.
DESIGN OF EXPERIMENTS: THE APPROACH

Process Steps Deliverables


Think 1 • Ensure basics
Structure plan o Which response should be measured?
experience and confirm o Which factors should be analyzed?
the inputs / outputs o What approach are we choosing?
o Which factor levels?

Plan 2 • Define the test conditions


Master the experimental o Make the trials randomly
protocol o Plan the tests
Experiment 3 • Testing and Data Collections
Conduct the tests o Do the trials
o Collect the data

Analyze 4 • Translation of the statistical results in conclusions for the process


to extract useful o Analyze the Data
information from test o Make conclusions
o Check the conclusions

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
148
Designed by Safran Consulting
PLAN: REPLICATES & RANDOMIZATION

 To plan the experiments, we take into account 2 key elements


 Several trials will have the same combination levels = REPLICATES
 The order of the trials will be launched RANDOMLY

 What is the interest for each of these elements?

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 149
Designed by Safran Consulting
WHY DO WE REPLICATE THE TRIALS?

 This means reproducing the trials 2 or more times with the same
levels/settings of factors to:
 Identify the measuring error: the variation identified between trials done with
the same levels/settings
 Identify the importance of the tested factors: the visible effect, is this due to
the factor modification or the variability of the common causes?

 Measure the impact of the condition of modifications not only on the average
response but also on the dispersion

Note: To replicate a trial does not mean to measure the same response twice,
BUT rather to run a trail for a second time in the same combination
level/setting and measure the new response.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 150
Designed by Safran Consulting
10’

Example of the Catapult:


REPLICATION TRUE or FALSE?

Case Choice
2 same tested combinations at the beginning and at the end of
True - False
the tests
2 successive shots without adjusting the settings True - False
2 successive shots with adjustment / settings True - False
2 shots with different combinations True - False
NOTES AND ADDITIONAL EXPLANATIONS

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
152
Designed by Safran Consulting
HIDDEN FACTOR

A hidden factor has an important effect on the response but is not known
as an important factor yet

 It is hidden because
 Still unknown,
 Its influence is judged negligible,
 No data is available

 Randomization & replication


will help to
 Identify hidden factors effects
 Be more robust to hidden factors effects

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
153
Designed by Safran Consulting
WHY DO RANDOMIZED TRIALS?

 Example  Ifwe consider that


 On a system of wire-  The quality of the wire is constant,
drawing(pulling) on steel cables  The drawing speed is constant,
the response is the diameter of  The drawn product is constant,
the wire once drawn out over
time.  Then
 The increase in diameter from the
Diameter of the wire

wear of the die


at the output

 The wear depends on the quality


of drawing soap used to lubricate

Time
This document and the information it contains belong to
Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
154
Designed by Safran Consulting
WHY DO WE RANDOMIZE THE TRIALS?

 If the company wants to test 2 different types of soaps, type A and B, 2


approaches are possible
Diameter of the wire coming out

Diameter of the wire coming out


Type A
1 Type B 1,24
2
1,22

1,20

1,18 1,18

1,16 1,16
Type A
1,14 1,14 Type B

Type A is used during the first week Type A is used alternatively with
(10 workstations), type B is used in the type B.
following week (10 workstations)

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 155
Designed by Safran Consulting
RANDOM IS RANDOM

 RANDOM means:

- different than the standard order.

- not in a chosen or adapted order.

 REALLY means:

- randomly, drawn out of a hat.

- practically, use the random function of Minitab.

Smoothes the impact of the Helps with validation of


hidden factors on all the factors. statistical conclusions from
experiments

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 156
Designed by Safran Consulting
10’

Use the catapult to practice...

Process steps
Follow the instructions below to create the
experimental design of 4 factors at 2 levels

Arm Distance between catapult


# Hinge Point Ball Angle Position and point of impact

Type Factor Factor Factor Factor response

1 Position 2 Squash 150 Position 2 -

2 Position 3 Basketball 170 Position 3 -


NEW DESIGN OF EXPERIMENT
CREATE A FULL FACTORIAL DESIGN

 STAT > DOE > FACTORIAL > CREATE FACTORIAL DESIGN

Default

Design with Click here


4 factors

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 158
Designed by Safran Consulting
NEW DESIGN OF EXPERIMENT
SELECT THE NUMBER OF REPLICATES

Full Factorial
Design

Number of
replicates

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 159
Designed by Safran Consulting
NEW DESIGN OF EXPERIMENT
FILL-UP THE FACTORS DIALOG BOX

Select the type of data: text (discrete) or


numeric (continuous)

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
160
Designed by Safran Consulting
NEW DESIGN OF EXPERIMENT
RANDOMIZE

 By default, the design is randomized.

 For the sake of this exercise, we are going to use the Options menu ... to
ensure we have the same « random ization »

2010 is a figure to initiate the randomization. 2 or more randomization


starting with the same seed will provide the same result

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 161
Designed by Safran Consulting
NEW DESIGN OF EXPERIMENT
VISUALIZE THE DESIGN

DoE summary

32 trials because
(4 factors ** 2 levels) *
4 replicates = 4 * 2 = 32

Standard order (not Randomized order for


relevant) the trials

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 162
Designed by Safran Consulting
DESIGN OF EXPERIMENTS: THE APPROACH

Process Steps Deliverables


Think 1 • Ensure basics
Structure plan o Which response should be measured?
experience and confirm o Which factors should be analyzed?
the inputs / outputs o What approach are we choosing?
o Which factor levels?

Plan 2 • Define the test conditions


Master the experimental o Make the trials randomly
protocol o Plan the tests
Experiment 3 • Testing and Data Collections
Conduct the tests o Do the trials
o Collect the data

Analyze 4 • Translation of the statistical results in conclusions for the process


to extract useful o Analyze the Data
information from test o Make conclusions
o Check the conclusions

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
163
Designed by Safran Consulting
DRIVING TESTS

A lack of rigor in carrying out the tests and data collection would result in
errors:
 Some will be detected during the analysis part,
 Unfortunately, others will not be detected and will lead to erroneous
conclusions.

 Typical errors during this phase


 Lack of explanation on the importance of the process and the
expected rigor
 No presence of black Belt in the field to answer questions
 No respect for the random order
 Non-compliance with desired settings in the plan
 Lack of comments in the test forms

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
164
Designed by Safran Consulting
COLLECT DATA

 It’s often useful to collect and filter the data rapidly. This allows to detect eventual
errors of:
 Data entries,
 Setting,
 Data manipulation,…

 We can redo the questionable test under acceptable conditions.

 Good Practice
 Start with the 2 most extreme trials to test the measurement, the
feasibility and the maximum amplitude observed on the response
 Quickly create first graphics
 Follow the steps to anticipate replicated measurement errors

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
165
Designed by Safran Consulting
NOTES AND ADDITIONAL EXPLANATIONS

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
166
Designed by Safran Consulting
100’

Use the catapult to practice...

Process steps
• Conduct 32 tests of your rigorous experimental
design
• Collect information in Minitab

Arm Distance between catapult


# Hinge Point Ball Angle Position and point of impact

Type Factor Factor Factor Factor Response

1 Position 1 Squash 150 Position 2 -

2 Position 2 Basketball 170 Position 3 -


NOTES AND ADDITIONAL EXPLANATIONS

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
168
Designed by Safran Consulting
DESIGN OF EXPERIMENTS
5. ANALYSIS WITH THE HELP OF
MINITAB

169 /
Ce document et les informations qu’il contient sont la propriété de Snecma. Ils ne doivent pas être copiés ni communiqués à un tiers sans l’autorisation préalable et écrite de Snecma.
DESIGN OF EXPERIMENTS: THE APPROACH

Process Steps Deliverables


Think 1 • Ensure basics
Structure plan o Which response should be measured?
experience and confirm o Which factors should be analyzed?
the inputs / outputs o What approach are we choosing?
o Which factor levels?

Plan 2 • Define the test conditions


Master the experimental o Make the trials randomly
protocol o Plan the tests
Experiment 3 • Testing and Data Collections
Conduct the tests o Do the trials
o Collect the data

Analyze 4 • Translation of the statistical results in conclusions for the process


to extract useful o Analyze the Data
information from test o Make conclusions
o Check the conclusions

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
170
Designed by Safran Consulting
ANALYZE RESULTS

 Before embarking on the analysis of your results, let's see a few


complementary concepts. Find the corresponding definition:

Terms Definitions

Residuals   Quantifying the response according to


the levels of the factors (forecast)

Main effect of a factor   Difference between the measured value


and the forecast calculated with the
predictive equation

Interaction   Synergy between factors

Predictive Equation   Expected change in response, resulting


from the change in the level of a factor

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
171
Designed by Safran Consulting
RESIDUALS IN THE CASE OF A FULL FACTORIAL DESIGN

 Residual = value of the response - value average replicas (for the same
combination of factor levels)

 Example: take 2 made replicas whose results are:


 Rep1: 230 cm
 Rep2: 242 cm

 Calculate the average of the 2 replicas

Average=

 Calculate the residuals


for rep1 rep2 Res1 = Rep 1 – Average =

Res2 = Rep 2 – Average=


This document and the information it contains belong to
Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
172
Designed by Safran Consulting
MAIN EFFECTS
INTERPRETATION

 Definition
 It is the impact of a factor on the result Y.
 The main effects will allow us to prioritize the factors, or even of neglecting
some

 What does this effect of 20 min mean?

cm

260
Going from position 2 to
Distance

position 3 correspond to an
increase of 20 min

240

Position 2 Position 3
Position
This document and the information it contains belong to
Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 173
Designed by Safran Consulting
MAIN EFFECT: VALUE OF EFFECT - INTERPRETATION .

 What is the ratio?


 This is the difference between the results of the factor related to the level
and the average
 That is half of the effect
 In this case the coefficient equals + 10 cm.

cm

260 Position 3 ' adds ' 10 cm to the


Distance

10 cm average.

10 cm Position 2 ' subtracts' 10 cm-


240 from the average .

Position 2 Position 3
Accrochage

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
174
Designed by Safran Consulting
RESIDUALS

 By hypothesis, the residuals are considered as the reflection of variations


of common causes, at experiment level.

 We suppose the following concerning residuals:


 Residuals should be independent, normally distributed,
with a mean equal to 0
 Residual plots should show no pattern relative to any factor
 Independent over time (= stable)
 Are not linked to the studied factor levels/settings
 Residuals will have the same variance over predicted values

 Note: in full factorial designs, to determine the residuals, it’s necessary to


do several replicates

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 175
Designed by Safran Consulting
INTERACTION - WHAT DOES THIS MEAN?

 We mention INTERACTION between factors when:


 The Impact of a factor on the response is different according to the level/settings of another factor

 Example
 Number of private car accidents, equipped with ABS systems or not, as a function of the presence of
frost/ice
Distance to stop a car

Frost/Ice

Factor B: Temperature

No frost/ice

yes no
Factor A: ABS Systems

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 176
Designed by Safran Consulting
INTERACTION: INTERPRETATION

Strong
Interaction
Case where
Response
the factor Average
No Interaction
has a
interaction
principal
Factor B
positive
effect
Factor A The steeper the difference in angle,
the stronger the interaction

Case where the


Response

Réponse
Factor B
factor does Réponse
Factor B
NOT have a Factor B

prinicpal postive
effect Factor A Factor A Factor A

No Average Strong
interaction Interaction Interaction

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 177
Designed by Safran Consulting
INTERACTION

Interaction Plot for Lead Time


Data Means

200 Language
English
190 German

180

170
In this example there’s no
160
Mean

interaction, because the


150
lines are parallel
140

130

120

110
Olympe Naxos
Products

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 178
Designed by Safran Consulting
PREDICTIVE EQUATION

 This analysis is based on a model also called a predictive equation. We


expect to find something like
 Y = general average + coefficient A [Factor A] + coefficient B [Factor B]
 For example, Y = 160 + 30 [Language] + 10 [Product]

 Figures in [ ] are the coded values ​( -1 or +1 depending on level)

 Exercise: What would be the distance for the model and combination below?

Arm
# Average Hinge Point Ball Angle Position Distance

Coefficients 240 10 -20 10 0 -

Basketbal
Combo. 1 Position 3 170 Position 3 ___________
l

Combo. 2 Position 3 Squash 170 Position 3 ___________


This document and the information it contains belong to
Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 179
Designed by Safran Consulting
PREDICTIVE EQUATION WITH INTERACTION TAKEN INTO ACCOUNT

 for example, with the interaction AB, the model becomes

Y (Distance) = overall average


+ coefficient A . [Factor A] + coefficient B . [Factor B]
+ coefficient C . [Factor C] + coefficient . [Factor D]
+ coefficient AB . [Factor A] . [Factor B]

 Values between [ ] are encoded values (-1 or +1, depending on the level)

Note: The values of the coefficients for A, B, C, D, AB, AC, AD,... are
dependent on test results and are automatically calculated by Minitab

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
180
Designed by Safran Consulting
ANALYZE DATA

 Represent the data


 Use the descriptive graphics
 Study residuals

 Represent the data


 Use the descriptive graphics
 Study residuals

 Represent the data


 Use the descriptive graphics
 Study residuals

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
181
Designed by Safran Consulting
CHRONOLOGICAL REPRESENTATION

 Use data from


DoE Devis 1.mtw

 Menu
 GRAPH > TIME SERIES PLOT > SIMPLE

The graph does not show


any particular
abnormalities (gaps from
average etc.)

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 182
Designed by Safran Consulting
RESIDUALS
 It is necessary to verify the residuals values (point diagram) before the data analysis is done, to
ensure that the hypothesis is kept/conserved . Otherwise we’ll face an erroneous risk of
interpretation.

Evolution in terms of
15 15
? the time function, a

Résiduels
Résiduels
Time
10 10
hidden factor linked to
5 5

Impact 0 0
time influence the result
-5 -5

- 10 - 10

Time Time
The variation evolve in
terms of the average value.
15 15
Try the log, square root or
Residuals

Residuals
10 10

Impact of 5 5
reversed transformation
result values 0 0 function in order to
-5 -5 understand
- 10 - 10

162 171 173 186 197 204 217 305 162 171 173 186 197 204 217 305
Response average Response average

15 15
Not a normal distribution
Detect why or try to use
N Score
N Score

10 10

Normal 5 5
a transfomation on the
Distribution 0 0
response.
-5 -5

- 10 - 10

-15 -15
-15 -10 -5 0 5 10 -15 -10 -5 0 5 10 15
15
Résidus Résidus

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 183
Designed by Safran Consulting
ANALYSIS OF RESIDUALS

 Use data from


DoE Devis 1.mtw

 Menu
 STAT > DOE > FACTORIAL > ANALYZE FACTORIAL DESIGN > GRAPHS
> FOUR IN ONE

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
184
Designed by Safran Consulting
FOUR GRAPHS IN ONE

The adjusted values are the combined averages of the


different factor levels/settings

These graphs allow us to


analyze the residuals No
abnormalties detected.

The distribution of the residuals


represent the Gaussian
distribution. In this example the
number is small.

The quantity of data


is too small to obtain the
Gaussian curve

The graphs of the residuals are OK

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 185
Designed by Safran Consulting
NOTES AND ADDITIONAL EXPLANATIONS

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
186
Designed by Safran Consulting
ANALYZE DATA

 Represent the data


 Use the descriptive graphics
 Study residuals

 Represent the data


 Use the descriptive graphics
 Study residuals

 Represent the data


 Use the descriptive graphics
 Study residuals

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
187
Designed by Safran Consulting
MINITAB DOE GRAPHS
CUBE PLOT

 Command:
 STAT > DOE > FACTORIAL > FACTORIAL PLOT

Select the 3 types of


graphics For each type of chart, specify the
configuration:Configuration SETUP

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
188
Designed by Safran Consulting
DOE GRAPHS - MAIN EFFECTS

The diagram shows the


effect once a factor is
changed ( low to high
level
The more the angle is
steep, the more significant
the effect.
If the segment is almost
horizontal, there’s almost
no effect during a change
in the factor.
To know whether the effect
is significant, consult the
p-value.
Overall average

The Language effect is more significant


than the Product effect

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 189
Designed by Safran Consulting
FINAL PREDICTIVE EQUATION

If segments are parallel,


there's no interaction.
If the segments intersect,
there is a very strong
interaction.

The angles are different


Therefore, search non-
parallel segments.

Interaction is taken into


account

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 190
Designed by Safran Consulting
ANALYZE DATA

 Represent the data


 Use the descriptive graphics
 Study residuals

 Represent the data


 Use the descriptive graphics
 Study residuals

 Represent the data


 Use the descriptive graphics
 Study residuals

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
191
Designed by Safran Consulting
ANALYSIS OF A FACTORIAL DESIGN

STAT > DOE > FACTORIAL > ANALYSE FACTORIAL DESIGN > TERMS

Switch the words in


the right column

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
192
Designed by Safran Consulting
ANALYSIS OF A FACTORIAL
DESIGN

Select:
Graphs
GRAPHS Select Half Normal

Select Pareto

Default Value

Select 4 Graphs in 1

Then: OK

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 193
Designed by Safran Consulting
PARETO DIAGRAM

This diagram show


the relative
importance of the
Main effects and
interactions

We will analyze/study the


data later (T value)

The AB interaction (Language * The effects on the right of the red line
Products) is not significant are statistically significant
(it’s = 0)

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 194
Designed by Safran Consulting
HALF-NORMAL PROBABILITY PLOT

The Factors not generating enough


effect are the common(current) causes
of variation, and will be close to the
left (black points).
The Main effects are the special
causes of variation and will be far from
the right (red squares).

With the half normal probability plot all


the effects, even the negative ones,
are represented on the right. It’s the
absolute value that will interfere, as
with Pareto.

Effects of A (language) & B (products)


are significant

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 195
Designed by Safran Consulting
FOUR GRAPHS IN ONE

 These graphs allow us to analyze the residuals.


No abnormalities are detected.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 196
Designed by Safran Consulting
RESULTS IN THE SESSION SHEET

Factorial Fit: Lead Time versus Language, Products

Estimated Effects and Coefficients for Lead Time (coded units)


Constant = overall
average.
Term Effect Coef SE Coef T P
Constant 160.000 0.7906 202.39 0.000
Language 60.000 30.000 0.7906 37.95 0.000
Products 20.000 10.000 0.7906 12.65 0.000
Language*Products 0.000 0.000 0.7906 0.00 1.000

Effects calculation
P-value for Ho:
Coefficients for the predictive « Effect is 0 »
equation.

Values found on the half-normal


probability plot

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
197
Designed by Safran Consulting
P-VALUE?

 Significance of the p-value:


 When p is low, Ha is preferred.
(when p is low, we reject the null hypothesis H0)
 What is the nul-hypothesis in this case?

 H0: Changes from lower to higher levels don’t change the effect.
or

 H0: This factor does not explain the variation in the process.

 A significant factor shows an effect when they change level.

 A factor is significant when we reject the H0.

 The significant factors have a p-value < 0,05.


This document and the information it contains belong to
Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 198
Designed by Safran Consulting
P-VALUES

 Remarks:
 To calculate the p-value, Minitab should estimate the error

 If there’s no repetition, Minitab can not estimate the error and therefore won’t
give a p-value.

 You may eliminate terms from the model using


 STAT > DOE > ANALYSE FACTORIAL DESIGN > TERMS,

 Minitab is going to use removed terms to estimate the error and then the p-
value

 Which terms should you eliminate? Use the Pareto diagram or the Normal
Probability plot to make an important decision.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 199
Designed by Safran Consulting
NOTES AND ADDITIONAL EXPLANATIONS

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
200
Designed by Safran Consulting
60’

Use the catapult to practice...

Process steps
Perform the analysis:
• Run chart of the responses,
• Residuals diagrams,

Then we will analyze:


• Diagram of the main effects
• Pareto chart
• interactions
• p values ​

What are your conclusions ?


NOTES AND ADDITIONAL EXPLANATIONS

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
202
Designed by Safran Consulting
THE PREDICTION EQUATION

 From the coefficients of the factors generated by Minitab, we can write an


equation:
 That quantifies the relationship between the output variable and the factors,
 Which allows us to predict the value of the output variable depending on the
levels of factors in the model ?.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
203
Designed by Safran Consulting
VERIFY THE CONCLUSIONS

 It is highly desirable to carry out confirmation tests.


 Make a small number of tests at recommended levels to see if they produce
the expected results.

 Implement recommended changes and carry out a monitoring of the


process
 Use control charts to check if the desired results are achieved (see steps
following DMAIC improve, control)

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
204
Designed by Safran Consulting
30’

Use the catapult to practice...

Process steps
• From all of your data, the facilitator will give
you a targeted distance

• Use the prediction equation to calculate the


correct setting to achieve this distance

• Do a confirmatory test
NOTES AND ADDITIONAL EXPLANATIONS

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
206
Designed by Safran Consulting
DESIGN OF EXPERIMENTS
6 FRACTIONAL DESIGNS

207 /
Ce document et les informations qu’il contient sont la propriété de Snecma. Ils ne doivent pas être copiés ni communiqués à un tiers sans l’autorisation préalable et écrite de Snecma.
KEY POINTS

 In full factorial designs, when the number of factors increases the number
of trials increases exponentially .

 Exercise: calculate the number of tests of a full factorial design for the
following cases
Number of factors Number of trials for a Full Factorial Design
5 ______________
6 ______________
7 ______________
8 ______________

 Is there a way to reduce the number of tests while having always valid
conclusions?

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
208
Designed by Safran Consulting
FRACTIONAL FACTORIALS: HOW TO REDUCE THE NUMBER OF TRIALS

Lets suppose that we can only do 4 trials?

Which trials will we choose?

7 8

5 6 ?
Angle

3 4

1 2
Hinge

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 209
Designed by Safran Consulting
FRACTIONAL FACTORIALS
HOW TO REDUCE THE NUMBER OF TRIALS

 We may imagine that it would be interesting to

 Ensure that each factor is tested the same number of times at the same level
(what we call a balanced design)

 The design should cover the biggest possible experimental range

 If one of the factors is considered as negligible, the result will be a full


factorial test on the other factors

 Fractional factorial designs allows it !

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 210
Designed by Safran Consulting
FRACTIONAL FACTORIAL DESIGNS

Id A B C AB AC BC ABC
1 - - - + + + -
2 + - - - - + +
3 - + - - + - +
4 + + - + - - -
5 - - + + - - +
6 + - + - + - -
7 - + + - - + -
8 + + + + + + +
 1: global average
 3: Main effects A, B, C
8 elements of information can be
 3: 2 factor interactions: AB, AC, BC
extracted from this plan
 4: 3 factor interactions: ABC

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
211
Designed by Safran Consulting
FRACTIONAL FACTORIAL DESIGNS

 We suppose that the interaction effect of the 3 factors is not significant.

 We choose the 4 combinations that corresponds to A + for ABC (or - for


ABC).

Id A B C AB AC BC ABC
1 - - - + + + -
2 + - - - - + +
3 - + - - + - +
4 + + - + - - -
5 - - + + - - +
6 + - + - + - -
7 - + + - - + -
8 + + + + + + +

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
212
Designed by Safran Consulting
PLAN FRACTIONAL FACTORIAL ½ = 4 TRIALS

Id A B C AB AC BC ABC
2 + - - - - + +
3 - + - - + - +
5 - - + + - - +
8 + + + + + + +

Note: this 4-run design has the following properties: the plan is balanced
(when A is +, there are as many of + and - for B and C), neglecting C, plan
with A and B is complete, etc...

Attention: 4 pieces of information can be extracted from this plan. We see


particular columns A and BC are identical, just like B and AC then C and
AB. These effects will be combined in the analysis.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
213
Designed by Safran Consulting
HOW TO CREATE A FRACTIONAL FACTORIAL PLAN WITH MINITAB

 STAT > DOE > FACTORIAL > CREATE FACTORIAL DESIGN…

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
214
Designed by Safran Consulting
½ FRACTIONAL DESIGNS

Confusions appear in the session window

The experiments appear in the Worksheet

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 215
Designed by Safran Consulting
CONFUSION

 The effect is the result of the actions of father and son working together, mixed
together.

 In this example we can suppose that the action of the father is stronger than the
son. In real examples we don’t know.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 216
Designed by Safran Consulting
CONFUSION (CONT)

 “We thought we studied the effect of the factor C, in reality though, the
effect that we observed is the result of factor C + interaction of the 2
factors AB".

 We suppose that:
 Main effect > interactions of the 2 factors

 This can be false in certain cases.

Note: To confuse C with AB, can result in other confusions:


A = BC,
B = AC.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 217
Designed by Safran Consulting
NOTES AND ADDITIONAL EXPLANATIONS

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
218
Designed by Safran Consulting
10’
½ FRACTIONAL DESIGN
EXAMPLE WITH 4 FACTORS
Plan – 2: LEVELS
1/2 fractionnaire 4 facteurs -IN 8 TRIALS
2 niveaux - 8 essais
A B C D AB AC AD BC BD CD ABC ABD ACD BCD
- - - - + + + + + + - - - -
+ + - - + - - - - + - - + +
+ - + - - + - - + - - + - +
- + + - - - + + - - - + + -
+ - - + - - + + - - + - - +
- + - + - + - - + - + - + -
- - + + + - - - - + + + - -
+ + + + + + + + + + + + + +

Process steps
• List the combined effects?
• What are the assumptions you ask for analysis (when combined effects) ?

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 219
Designed by Safran Consulting
NOTES AND ADDITIONAL EXPLANATIONS

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
220
Designed by Safran Consulting
CONFUSION

 We studied the effect of the D factor although in reality the effect that we
observed is the result of "Factor D” + interaction of the 3 factors ABC".

 We suppose that:
 Main effects > Interactions of 2 factors > Interactions of 3 factors

 This can be false in some cases.

 Note: To confuse D with ABC lead to other confusions:


 B = ACD, C = ABD, A = BCD
 AB = DC, AC = BD, BC = AD

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 221
Designed by Safran Consulting
INFORMATION ON THE FRACTION FACTORIAL PLANS

 Advantages of the fractional factorial designs:


 The number of trials is half of the full factorial design

 Disadvantages of the fractional factorial experiment:


 The interactions on a high level can be confused with the interactions on a
lower level (level 2)

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 222
Designed by Safran Consulting
HOW TO CREATE A FRACTIONAL FACTORIAL DESIGN IN MINITAB

 Example with 6 factors


STAT > DOE > FACTORIAL > CREATE FACTORIAL DESIGN…

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 223
Designed by Safran Consulting
HOW TO CREATE A FRACTIONAL FACTORIAL DESIGN IN MINITAB

Design resolutions describe how much the effects in a fractional


factorial design are aliased with other effects.

This is a VI resolution.

Minitab shows the structures of alias (confusions):

1+5

2+4

3+3

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 224
Designed by Safran Consulting
AVAILABLE DESIGNS OF EXPERIMENT

 STAT > DOE > FACTORIAL > CREATE FACTORIAL DESIGN… > DISPLAY
AVAILABLE DESIGNS

Red: will not create a model


Yellow: Acceptable model
Green: Good model

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
225
Designed by Safran Consulting
RESOLUTION: DESCRIPTION

What Indicates the quality of the Fractional Factorial Design.


High resolution  ability to distinguish a large number of effects

Why Choose the plan knowingly (associated risk)

When During the generation of the plan

Who The person who generates the plan (Black Belt or specialist)

Comment With Minitab, it is calculated from the combined effect chain shorter. For
example, if A = BC, then resolution = # factors = 3

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
226
Designed by Safran Consulting
15’

Check our understanding of the


concept of confusion

Process steps
• Calculate the resolution of plans on the
following page
• Answer to the true or false quiz
EXERCISES

Combined effects Resolution


A + BD ______________
AB + CD ______________
A + BCD ______________
AB + CDE ______________

Phrase Choice
The higher the resolution, the better the design (plan) True - False

A resolution III is sufficient to know the interaction effects True - False

It is possible to have a resolution V True - False

If the couple A + BC has a significant effect, we say that BC is the


True - False
influential factor

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
228
Designed by Safran Consulting
60’

What is the difference between the results of a


Full Factorial Design and a Fractional Factorial
Design?
Process steps
• Resume the Catapult exercise
• Build ½ Fractional Factorial Design for the 4
factors studied in the Full Factorial Design. Do
not add replicates.
• Retrieve the results of the Full Factorial Design
and put the values of distance in the ½ Fractional
Factorial Design (for selected combinations)
• Analyze the plan and compare the conclusions
obtained between the 2 approaches (full and ½
fractional)
• Debrief
NOTES

Parameters Coefficient of the complete plan Coefficient of ½ fractional


plan

Hinge (A) ______________ ______________

Ball (B) ______________ ______________

Angle (C) ______________ ______________

Arm (D) ______________ ______________

Adj. R-square ______________ ______________

S ______________ ______________

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
230
Designed by Safran Consulting
DESIGN OF EXPERIMENTS
7. CONCLUSION
CONCLUSION OF THE DESIGN OF EXPERIMENTS CHAPTER

 Adjust the number of runs required,


understand and optimize processes
 Improve the process
 To predict a response
 A Fractional Factorial Design is an
interesting alternative to reduce the
number of tests when seeking only to
order the influences of factors

 Talk about experience


 The choice of factor levels and the
identification of responses require some
experience and skills in the relevant field.

 Key success factors


 Stable processes
 Defined and stable measurement systems
 Replicated runs
 Runs led in a random order
This document and the information it contains belong to
Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
232
Designed by Safran Consulting
Statistical Modeling
MULTIPLE LINEAR REGRESSION
233 /
Ce document et les informations qu’il contient sont la propriété de Snecma. Ils ne doivent pas être copiés ni communiqués à un tiers sans l’autorisation préalable et écrite de Snecma.
MULTIPLE LINEAR REGRESSION

What Statistical tool for the prediction of a response from several factors

Why Understanding the sources of variability in a process


Optimize the factors of a process
Remove the analysis of non-significant variables

When During analyze or improve (Improve) when there are historical data collected
in the form of a table

Who The person who generates the project or specialist

Comment With Minitab, following the step by step process

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
234
Designed by Safran Consulting
MULTIPLE LINEAR REGRESSION Multiple regression - Catapulte.mtw

 Show data (matrix chart)

 Make a regression
 Calculate VIF values.
 If VIF is greater than 5, remove factors in the model until the VIF are lower than 5.

 Build Model
 Watch P values for the factors.
 Remove insignificant factors.

 Analyze residuals
 Analyzing the residual plots.
 Study unusual cases.

 Do we have a model?
 Watch the P value for the regression.
 Looking adjusted R² and S

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
235
Designed by Safran Consulting
MULTIPLE LINEAR REGRESSION

 First, we focus on the continuous input variables X and we will achieve


a matrix chart to see their correlation with output variable Y.
Diameter

55

50
Weight
The correlation graphs are
45 organized in a matrix.
25.0

22.5

20.0
Room_Temperature One can visualize the relationships
and variable profiles of continuous
60
inputs with each other and with the
45
output variable.
Drawback_Angle

30
4000

It is a superficial approach to
3000 eradicate the sources of error
Shooting_Distance

2000

32.0 34.5 37.0 45 50 55 20.0 22.5 25.030 45 60

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 236
Designed by Safran Consulting
MULTIPLE LINEAR REGRESSION

 Use the Minitab command


 Graph > Matrix plot

 First we will focus on the continuous input


variables X’s and make a matrix plot to see how
they relate to the output variable Y.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
237
Designed by Safran Consulting
MULTIPLE LINEAR REGRESSION

 Correlation:
 With the matrix plot we have been able to visualise the relationships between
different input and output variables.

 In order to quantify this relationship with “facts and figures”, we can express
this relationship with the correlation coefficient.

 When we are dealing with multiple variables we will use the correlation
matrix:
 >Stat>Basic Statistics>Correlation>

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 238
Designed by Safran Consulting
MULTIPLE LINEAR REGRESSION
Select the input and
output relevant variables
(which appear to be
linked)

To have the P values for the


coefficients of correlation.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 239
Designed by Safran Consulting
MULTIPLE LINEAR REGRESSION

In this correlation matrix we can see the


Pearson correlation coefficient from all pairs
of variables.
Correlations: Diameter, Weight, Room Tempera, Drawback Ang, Shooting Dis

Diameter Weight Room Tempera Drawback Ang


Weight 0.902
0.000

Room Tempera 0.059 0.037 The correlation coefficient between weight of


0.574 0.724
the ball and the room temperature is 0.037
Drawback Ang -0.194 -0.192 0.186 (non significant because of high P value).
0.061 0.064 0.072

Shooting Dis -0.397 -0.369 0.163 0.895


0.000 0.000 0.116 0.000
Strong positive and significant correlation
between shooting distance and drawback
angle.

R value
P value

P value indicates whether there is enough evidence to


say there is a significant correlation, where as R
indicates the intensity of the correlation.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 240
Designed by Safran Consulting
MULTIPLE LINEAR REGRESSION: THE MODEL

 With single linear regression (1 input variable) the model looked like:

Y  aX  b
 When we are dealing with multiple input or process variables, our model
expands to:

Y  a1X1  a2 X 2  .......  ak X k  b

 Y = output variable, X1, X2, X3 = input variables


 Slope for Xk. If all other X’s are constant, the value of the output Y increases
with this amount for each increase of Xk with one unit.
 B = Intercept: predicted value of Y when all X=0 (little practical value).

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 241
Designed by Safran Consulting
MULTIPLE LINEAR REGRESSION Multiple regression - Catapulte.mtw

 Show data (matrix chart)..

 Make a regression
 Calculate VIF values.
 If VIF is greater than 5, remove factors in the model until the VIF are lower than 5.

 Build Model
 Watch P values for the factors.
 Remove insignificant factors.

 Analyze residuals
 Analyzing the residual plots.
 Study unusual cases.

 Do we have a model?
 Watch the P value for the regression.
 Looking adjusted R² and S

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
242
Designed by Safran Consulting
MULTIPLE LINEAR REGRESSION

 Limits multiple regressions


 Instead of an experimental design, historical databases are not optimized for
statistical analysis.
 It is common and normal for several input variables that are highly correlated

 Consequences
 Brings instability and errors in the calculation of the regression coefficients
 It is unclear which factor is really important

 How to deal with this?


 We will deal with this problem by dropping highly correlated input variables
from the model.
 We will find these variables by using the Variance Inflation Factors (VIF’s)
when performing the multiple regression…

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
243
Designed by Safran Consulting
MULTIPLE LINEAR REGRESSION: CO-LINEARITY

 Output = Water consumption ; input variables are Age and Height of the
children. As we can see in the matrix plot, both age and height are highly
correlated. Because of this, it is not possible to determine which input variable
is responsible for the water consumption per day. Is it Age or is it Height (or
both)?

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 244
Designed by Safran Consulting
MULTIPLE LINEAR REGRESSION: CO-LINEARITY

 Variance Inflation Factor (VIF) :


 The VIF is a number that expresses the amount of multi co-linearity for one input
variable with the rest of the input variables that are used in the model.

 Suppose there are k input variables: X1, X2, …Xj and Xk. To find the VIF for input
variable Xj, we will regress Xj on the remaining k-1 input variables. Out of this
regression we can calculate Rsquared for this input variable and put it into the formula
(see above).
1
VIF j 

1  R 2j 
 If there is a high correlation from Xj with the remaining input variables, R² will get close
to 1 and the VIF will be higher. If there is no correlation at all, R² wil be 0 and the VIF
will be 1.

 Rule of use
 If the VIF is higher than 5, we have a problem of multi co-linearity and input variables
should be dropped from the regression model in order to avoid multi co-linearity.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 245
Designed by Safran Consulting
MULTIPLE LINEAR REGRESSION

 In our test case of water consumption of young children we assume there


is a strong correlation between age and height. Calculate the VIF for
height:

Regression Analysis: Height versus Age


1
VIFj  The regression equation is

1  R j  Height = 28.3 + 1.35 Age

 squared
 Predictor
Constant
Coef
28.3446
SE Coef
0.5957
T
47.58
P
0.000
Age 1.35169 0.01755 77.02 0.000
1
 S = 2.98373 R-Sq = 95.5% R-Sq(adj) = 95.4%

1  .955 Analysis of Variance

Source DF SS MS F P
 22 Regression
Residual Error
1
282
52806
2511
52806
9
5931.53 0.000

Total 283 55317

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 246
Designed by Safran Consulting
MULTIPLE LINEAR REGRESSION

 Doing the multiple regression


 Find the data in dataset;
 Stat > Regression > Regression: Select the output variable

Select the input variables

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 247
Designed by Safran Consulting
MULTIPLE LINEAR REGRESSION

 Look for input variables that are highly correlated to other input parameters. Once you
have found them, eliminate them from the model and perform the regression analysis
again.

 Hint: you can turn on the VIF’s by using the command:


 Stat > Regression > Regression > Options.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 248
Designed by Safran Consulting
MULTIPLE LINEAR REGRESSION
INTERPRETING RESULTS (PART 1)

Regression Analysis: Shooting Distance versus Diameter, Weight, ...

The regression equation is


Shooting Distance = 2852 - 64.9 Diameter + 3.9 Weight + 4.32 Room Temperature
+ 44.8 Drawback Angle

• Conclusions are not valid as long as the


VIF’s are bigger than 5.
Predictor Coef SE Coef T P VIF
Constant 2852.5 428.8 6.65 0.000 • We will drop “Diameter” as input variable
in our regression model and redo the
Diameter -64.94 24.13 -2.69 0.009 5.4 regression.
Weight 3.88 16.42 0.24 0.814 5.4
Room Temperature 4.318 9.185 0.47 0.639 1.0
Drawback Angle 44.755 2.235 20.02 0.000 1.1

S = 174.510 R-Sq = 85.3% R-Sq(adj) = 84.7%

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 249
Designed by Safran Consulting
MULTIPLE LINEAR REGRESSION Multiple regression - Catapulte.mtw

 Show data (matrix chart)..

 Make a regression
 Calculate VIF values.
 If VIF is greater than 5, remove factors in the model until the VIF are lower than 5.

 Build Model
 Watch P values for the factors.
 Remove insignificant factors.

 Analyze residuals
 Analyzing the residual plots.
 Study unusual cases.

 Do we have a model?
 Watch the P value for the regression.
 Looking adjusted R² and S

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
250
Designed by Safran Consulting
P-VALUE ANALYSIS

 As for analyses in the design of experiments, analysis of P-values allows


to remove the insignificant factors
Coefficients

Term Coef SE Coef T-Value P-Value VIF


Constant 2551 428 5.96 0.000
Weight -35.78 7.48 -4.79 0.000 1.04
Temperature 2.59 9.47 0.27 0.785 1.04
Angle 45.13 2.31 19.56 0.000 1.08

Comments

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
251
Designed by Safran Consulting
MULTIPLE LINEAR REGRESSION Multiple regression - Catapulte.mtw

 Show data (matrix chart)..

 Make a regression
 Calculate VIF values.
 If VIF is greater than 5, remove factors in the model until the VIF are lower than 5.

 Build Model
 Watch P values for the factors.
 Remove insignificant factors.

 Analyze residuals
 Analyzing the residual plots.
 Study unusual cases.

 Do we have a model?
 Watch the P value for the regression.
 Looking adjusted R² and S

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
252
Designed by Safran Consulting
VALIDATION OF RESIDUALS

 The results of a multiple regression analysis are valid only if residuals:

 Are normally distributed with a mean value of 0.

 Have a random profile over time (not time-dependent).

 Have a value that is not a function of the output response Y.

 Have a value that does not depend on the X values.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
253
Designed by Safran Consulting
MANIPULATION MINITAB

 Graphic selection

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
254
Designed by Safran Consulting
1 3

2 4

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
255
Designed by Safran Consulting
NOTES AND ADDITIONAL EXPLANATIONS ON RESIDUALS

1 3

2 4

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
256
Designed by Safran Consulting
MULTIPLE LINEAR REGRESSION Multiple regression - Catapulte.mtw

 Show data (matrix chart)..

 Make a regression
 Calculate VIF values.
 If VIF is greater than 5, remove factors in the model until the VIF are lower than 5.

 Build Model
 Watch P values for the factors.
 Remove insignificant factors.

 Analyze residuals
 Analyzing the residual plots.
 Study unusual cases.

 Do we have a model?
 Watch the P value for the regression.
 Looking adjusted R² and S

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
257
Designed by Safran Consulting
ANALYSIS MODEL

Analysis of Variance Comments on the value of P


Source DF Adj SS Adj MS F-Value P-Value Model
Regression 2 15548797 7774399 241.18 0.000
Weight 1 743497 743497 23.07 0.000
Angle 1 13038592 13038592 404.49 0.000
Error 91 2933342 32235
Total 93 18482139

Model Summary

S R-sq R-sq(adj) R-sq(pred)


179.540 84.13% 83.78% 83.05%

Comments on S, R and R squared adjusted square

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
258
Designed by Safran Consulting
MULTIPLE LINEAR REGRESSION

 Output of the multiple regression: R-squared


 As already explained before: this is the amount (%) of variation in the output that is
explained by the variation in the input variables.

SSregression SSerror
Rsquared   1
SStotal SStotal

 Every predictor that is added to the model will increase the value of R-squared even if
it is of no real value to the model and thus gives a better prediction of the output.

 However this does not mean that this model is of any practical value. If the ratio gets
closer to 1, the R-squared value might become artificially high (close to 100 %). The
model perfectly fits these data, but as soon as new data have to be predicted, the
model will fail.

 In order to compensate for the number (p) of predictors in the model compared to the
sample size (n) of the regression dataset, the value R-squared adjusted is introduced.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 259
Designed by Safran Consulting
MULTIPLE LINEAR REGRESSION

 Output of the multiple regression: R-squared adjusted


 Where n = number of sampling points in the dataset and p = number of
predictors (input variables) in the regression model.

 SS error 
 
 n  p  
adjusted
Rsquared  1
 SStotal 
 
 n  1 
 When do we have enough input variables to reasonably explain the output
variation ? Typical values in industrial processes should obtain R squared
adjusted values above 70 %

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 260
Designed by Safran Consulting
CONCLUSION - MULTIPLE LINEAR REGRESSION

 It is possible to create patterns with


more than one factor
 Multiple linear regression is a
generalization analysis tools used in
experimental designs

 Historical data handling requires more


precautions
 Checking multi collinearity factors
 Detailed study of residuals
 Study of R squared values

 This technique can be generalized to


discrete factors and responses as well
as discrete nonlinear studies

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
261
Designed by Safran Consulting
This document and the information it contains belong to
Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
262
Designed by Safran Consulting
Develop new products, services or
processes using Lean Sigma
263 /
Ce document et les informations qu’il contient sont la propriété de Snecma. Ils ne doivent pas être copiés ni communiqués à un tiers sans l’autorisation préalable et écrite de Snecma.
AGENDA

 Introduction

 Back on Six Sigma Performance

 Design to reach Six Sigma Level

 Methodological key tools

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 264
Designed by Safran Consulting
LEARNING OBJECTIVES

 To understand the implications of your design decisions in a world where


variation is everywhere

 To review the basic Six Sigma concepts, and how these concepts are
applied in the design environment

 To introduce Design for Six Sigma, including:


 The DFSS process
 The tools associated with DFSS
 The relationship with DMAIC

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 265
Designed by Safran Consulting
SUMMARY OF DFSS OBJECTIVES

 Six Sigma strives to prevent defects through reducing variation.


 We can only achieve 6s quality by reducing the variation inherent in our
designs

 Design for Six Sigma seeks to balance customer requirements against


manufacturing or service capabilities.

 DFSS tools help us to:


 Predict the quality of our designs
 Design quality into our new products/services
 Prioritize and improve the quality of our existing products and services.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 266
Designed by Safran Consulting
AGENDA

 Introduction

 Back on Six Sigma Performance

 Design to reach Six Sigma Level

 Methodological key tools

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 267
Designed by Safran Consulting
TWO DIFFERENT PERSPECTIVES ON QUALITY
Historical Perspective Modern Perspective

Uniformity around a target On target with minimum variation


The specifications are met. The specifications are met.
The Pilot lands on the runway every time. The Pilot lands as closely as she can to
the center of the runway every time.

Which pilot would you rather fly with?


What are the implications of this change in philosophy?

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
268
Designed by Safran Consulting
WHY AIMING FOR THE TARGET IS A GOOD IDEA

 Rolled Throughput Yield (RTY) is the probability that...


 A service will be made up of a number of operations,
 A service will pass through an entire process, or
 A service will meet all of its performance requirements

 ..without rework and without incurring any defects


Process Steps Units In Units Out Defectives Yield
Process A 134 125 9 93%
Process B 125 116 9 93%
Process C 116 108 8 93%
Process D 108 100 8 93%

Process A Process B Process C Process D


93% 93% 93% 93%
= 74.8% RTY

RTY= Y1 x Y2 . . . Yn (n = number of process steps)

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 269
Designed by Safran Consulting
WHY SIX SIGMA IS THE GOAL

 For complex products, processes and systems, 6s is necessary to


produce and perform defect-free more than 90% of the time.

Impact of Complexity on Rolled Throughput Yield


100%
90% 6
Rolled Throughput Yield

80%
70%
5
60%
50%
40%
30% 4
20%
10%
0%
0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000 10,000
Number of Parts/Process Steps

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 270
Designed by Safran Consulting
AGENDA

 Introduction

 Back on Six Sigma Performance

 Design to reach Six Sigma Level

 Methodological key tools

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 271
Designed by Safran Consulting
THE “5 SIGMA WALL”

6
Redesign
Sigma Level

“5 Wall” Benefit


5

4 Process Improvements
Plus Product Redesign to Match
Improved Process Capability
Process Improvements
3 Only
Time

Break through the “5 wall” by redesign for


manufacturability

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
272
Designed by Safran Consulting
VISIBLE / GENERATED COSTS
70%

50%

25%

20%
15%

10% 5% 5%

Design Material Work Overhead


Source: EADS
This document and the information it contains belong to
Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
273
Designed by Safran Consulting
WHERE SHOULD WE FOCUS?

1000:1
PRODUCT
100:1
DEFINITION
PRODUCT
10:1
DESIGN
PROCESS
DESIGN
1:1
PRODUCTION
PRODUCT
IMPROVEMENT

LOW VISIBILITY HIGH VISIBILITY


HIGH RETURNS LOW RETURNS

Considering variation up front in the design


maximizes our investment opportunity

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
274
Designed by Safran Consulting
15’
2 products have been proposed by the
Development teams.

These two products meet the same


specifications and similar functional
performances .
A
What are the advantages / disadvantages of
each proposal?

Process steps
• In groups, analyze both proposals and
complete table of the advantages of each
solution (10 ')
B • Share results (5 ')
Advantages of solution A

Advantages of solution B

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
276
Designed by Safran Consulting
 
MANUFACTURABLE DESIGN - SOLUTION

• More difficult to manufacture • Fewer components, easier to produce,


• More components buy, order, exchange, disassemble
• Longest assembly • Simplified and more robust assembly
• Weight biggest • Lighter
• Difficult to keep the number of different parts • Fewer key features in the process

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
277
Designed by Safran Consulting
ADDITIONAL NOTES AND EXPLANATIONS

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
278
Designed by Safran Consulting
DESIGNED FOR SIX SIGMA
- Easy to fabricate parts
- One assembly direction “top
down”
- No adjustments required - Standard parts (one screw
- No hidden features type)

- Test direction access - Parts are self-guiding


from top
- Avoid tangle with use of
- Sub-assemblies reduce fixtures
handling of small hard to
grip parts - Symmetry in two axis

- Holes large enough -Die cast with minimal amount


(straightness issues of holes (debris chip)
if too deep) -Standard cutters
-Guide features
- Common datum’s for all
fixtures
- One common plane for - bottom rails for conveyor
assembly
- Tabs for robotic lift
This document and the information it contains belong to
Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
279
Designed by Safran Consulting
AGENDA

 Introduction

 Back on Six Sigma Performance

 Design to reach Six Sigma Level

 Methodological key tools

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 280
Designed by Safran Consulting
PLAN AND MANAGE THE PROJECT
PHASES TYPICAL CONTENTS
DEFINE • Project Charter, Business Case, Budget, Scope
• Team List,
• Project Plan,
MEASURE • Customer Research Plan, Customer Needs, Competitive Performance
• QFD Matrix, CTQs, Design Scorecards

ANALYZE • Functional Analysis,


• Process Maps (High-Level),
• Benchmarking Plan, Benchmarking Report,
• Conceptual Design,
• Design Requirements, Design Scorecards
DESIGN • Detailed Process Maps,
• Facilities Layout,
• Information System Design, Job/Task Design,
• Materials Lists, CTP Specifications,
• FMEA, Capability Analyses, Design Scorecards
VERIFY • Pilot Plan, Pilot Results,
• Implementation And Transition Plans,
• Control Plans, Procedures, Design Scorecards
MISC Meeting Minutes, Design Control Process, Communication Plan, Tollgate
Reviews
This document and the information it contains belong to
Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
281
Designed by Safran Consulting
RELATIONSHIP WITH DMAIC
SHOULD WE USE DMAIC OR DMADV?

DMAIC/DMADV Transition Points:

Define Measure Analyze Improve Control

Yes Yes No
Is
Does a Incremental New or redesigned
Process Improvement product/
Exist? Enough? service?

No No Yes

Define Measure Analyze Design Verify

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
282
Designed by Safran Consulting
Example in this
formation
IF APPLICATIONS

• Meeting the Expectations of Customers


Product design to achieve • Be specific about conditions of use
and maintain a level of • Being robust to variations in production
performance

• Meeting the Expectations of Customers


Design for • Be specific about conditions of use
Manufacturability • Being robust to variations in production
(Conception Fabricable) • Take into account the feedback to ensure a better cost /
quality / manufacturing lead times

• Meeting the Expectations of Customers


Create a new • Be specific about conditions of use
administrative service or • Innovate by creating more value with the proposed service
deliverable

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
283
Designed by Safran Consulting
NOTES

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
284
Designed by Safran Consulting
DFSS IN PRODUCT DESIGN
DFSS DESIGN - A SIMPLIFIED EXAMPLE = PEN

Process steps Deliverables


Define 1 • Project Charter
Define the project • Developing a pen for use in manufacturing plants for a multi-media use
Identify and prioritize
project performance
criteria 2 • Project Performance Criteria
• Development costs < 230 k €
• Production series: April 3, 2016
• Quality first 6 months of production <100 ppm

3 Multidisciplinary team

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
286
Designed by Safran Consulting
DFSS DESIGN - A SIMPLIFIED EXAMPLE = PEN

Process steps Deliverables


Measure 4 • Product Functional Analysis
Rate / formalizing the • Remove ink on media
important features for • Anchoring product on clothing or pocket
Quality ( CTQs ) • Easy to use
• Waterproof during the non - use

5 • Product Performance Criteria


• 20,000 lines of writing
• 165g max
• Unit cost < 0.97 cts
• Writing on media available in factory

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
287
Designed by Safran Consulting
DFSS DESIGN - A SIMPLIFIED EXAMPLE = PEN

Process Steps Deliverables


Analyze 5 • Creation of concepts / design at the system level
Propose and select a • Benchmark competition
concept of product / • 3 pen design concepts
service or macro-
processes 6 Comparative analysis using the Pugh matrix

Voice of the Customer Voice of the Business

Manufacturability

Re-use of current
sub-components
Multi support
Lines of text

Unit cost
Weight
Scores
Weight 4 2 1 3 3 4 Total + Total S Total -
Best competitor S S S S S S 0 17 0

Pen 1 ++ ++ + ++ + + 26 0 0
Pen 2 ++ + ++ + + + 22 0 0
Pen 3 ++ + -- + - S 13 4 5

7 Decision on the concept

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
288
Designed by Safran Consulting
DFSS DESIGN - A SIMPLIFIED EXAMPLE = PEN

Process Steps Deliverables


Design 8 • Detailed design of the selected pen
Detailing and optimize • Design and creation of a behavioral model (X vs Y)
the product / service or • FMEA product
process • Tolerance and probabilistic sensitivity analysis functions tolerances

9 • Manufacturability Analysis
• Comparative tolerance vs forecast capability

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
289
Designed by Safran Consulting
DFSS DESIGN - A SIMPLIFIED EXAMPLE = PEN

Process Steps Deliverables


Verify 10• Validation of results vs. forecasts
Check that the solution • Validation of the measurement system characteristics
meets the needs of the • Measurement of produced parts and calculating the actual
Customer robustly and capability
durable • Product optimization if necessary

11• Update control plans

12• Lessons learned and knowledge capitalization

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
290
Designed by Safran Consulting
DFSS IMPLEMENTATION FOR A
CHANGE / NEW PRODUCTION LINE
 DEFINE
 Define the project & develop the project charter
 Achieving the project schedule
DFSS COURSE OF A PROJECT
 Assess project risks FOR A NEW PROCESS
 Establish the communication plan

 MEASURE
 Analyze expectations (surface, stocks, flow time, shops, line flexibility)
 Define KPIs end of project and establish the scorecard

 ANALYZE alternatives
 Matrix Product / Process VSM of the existing process (in the case of a line reconfiguration)
 5S of the current process (in the case of a line reconfiguration)
 Establish new process naked macro vision
 Brainstorming, spaghetti diagrams of processes and risk analysis solutions
 Pugh matrix for comparison and selection of the solution) Strategy for transfer ( order of macro operations )

 DESIGN ( detailed view of the process )


 Workshops by sub-processes: Define precisely the needs and implementation taking into account the tools,
tables, water inlets, air and electricity and the internal flow of sub-processes
 Plan of detailed implementation plan and conduct change

 VERIFY ( Implementation and verification of results )


 Lead the transition project or creation of the line
 Check the results and adjust / optimize if necessary
 Capitalize on lessons learned, define the progress plan the coming months and close the project

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
292
Designed by Safran Consulting
CONCLUSION
THE APPROACH: DESIGN FOR SIX SIGMA

 Quality control by controlling the


variability
 To reach the 6 level, action is needed
from the design

 Design the product / process respecting


 Expectations of Customers
 Industrial or organizational constraints

 DFSS tools help


 Provide the quality of our product or
process since day 1
 Ensure that our assumptions are
confirmed when implementing

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
294
Designed by Safran Consulting
Developing a successful team
BELBIN®
295 /
Ce document et les informations qu’il contient sont la propriété de Snecma. Ils ne doivent pas être copiés ni communiqués à un tiers sans l’autorisation préalable et écrite de Snecma.
LEARNING OBJECTIVES

 Understand the roles that exist in


teams

 Understand your own preferred


role in a team

 Understand how to use team roles


as a black belt

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 296
Designed by Safran Consulting
HISTORY

 Developed by Dr Meredith Belbin in the 1970’s at Henley Management


College. Belbin and his team tried to develop a ‘model’ that demonstrated
what makes an effective team. Their discoveries were developed into the
questionnaire you have just filled out

 They have now been used all over the world

 Belbin defines a team role as;

“A tendency to behave, contribute and interrelate with others in a


particular way”

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 297
Designed by Safran Consulting
INTRODUCTION
 Successful teams contain members with a mixture of key attributes; they are not
collections of similar people. It is the variety of people which makes the effective
group.

 He was invited to investigate why some teams did so much better than others on
the course’s major management game.

 Belbin began by applying a battery of tests and inventories to individual team


members but gradually he became aware of the effect of the mixture of people
within the team.

 Teams with an appropriate balance of attributes won; teams missing some key
roles tended to lose.

 This empirical research has been successfully transferred to organisational


settings and Belbin’s conclusions are being used to increase team effectiveness
in a number of large companies.

 Bibilography:

“Management Teams – Why They Succeed or Fail” published by Heineman.


This document and the information it contains belong to
Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 298
Designed by Safran Consulting
TEAM CONSTITUTION
NOBODY IS PERFECT, BUT A TEAM CAN BE.

 Despite expertise of team members, team can miss the target


 Discussion on details
 No decision making
 After deciding, team still criticize choices

 Team consistency appears more important than individual expertise.


Team achieves better results when
 Team members can organize and structure the team
 There is no dominant member
 Facilitator is effective

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
299
Designed by Safran Consulting
ROLES WITHIN A TEAM

 While no man or woman can possibly have all the talents, skills and
knowledge necessary to successfully complete a major project by
themselves, it has been shown that a well structured team can achieve a
very great deal.

 In addition to the functional roles connected with the task, there are some
very clear roles associated with the working processes within the team.
These can be classified into Outward Looking and Inward Looking.

Outward Looking Inward Looking


Chairman Coordinator
Plant Monitor Evaluator
Resource Investigator Team Worker
Shaper Completer

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 300
Designed by Safran Consulting
UNDERSTANDING TEAM ROLES

 You will be working in teams

 As the change agent you will have to:


 recognise missing roles and try to either assign someone to it
 or take on the role yourself

 You will have to build teams and the Belbin questionnaire is an excellent
tool for discussion and development about teams

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 301
Designed by Safran Consulting
QUESTIONNAIRE – EXAMPLE

 What I believe I can contribute to a team:


Sentences Pts

a) I think I can quickly see and take advantage of new opportunities

b) I can work well with a very wide range of people

c) Producing ideas is one of my natural assets

d) My ability rests in being able to draw people out whenever I detect they have something of
value to contribute to group objectives
e) My capacity to follow through has much to do with my personal effectiveness

f) I am ready to face temporary unpopularity if it leads to worthwhile results in the end


g) I am quick to sense what is likely to work in a situation with which I am familiar
h) I can offer a reasoned case for alternative courses of action without introducing bias
Total 10

 7 different questionnaires have to be answered to evaluate your role in a


team,10 points to spread over 8 attitudes
 A radar will summarize the behavior of the team as a whole

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 302
Designed by Safran Consulting
15’

What are our preferences? If we


were to be part of the same team,
how could we be more efficient?

Process steps
• Each person completes the questionnaire
• Calculate your score on the last page

• Form 2 teams and fill the sheet Excel with data


from the team
• Analyze the results of the team
• What are the strengths of your team?
• What are the weak points of your team?
• What actions should you take to improve this
team?
NOTES AND ADDITIONAL EXPLANATIONS

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
304
Designed by Safran Consulting
CHAIRMAN (CH)
Contribution Allowable Weaknesses
Mature, confident, a good chairperson. Clarifies Can often be seen as manipulative. Off
goals, promotes decision-making, delegates well. loads personal work.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
305
Designed by Safran Consulting
PLANT (PL)

Contribution Allowable Weaknesses


Creative, imaginative, unorthodox. Solves Ignores incidentals. Too pre-occupied to
difficult problems. communicate effectively.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
306
Designed by Safran Consulting
RESOURCE INVESTIGATOR (RI)

Contribution Allowable weaknesses


Extrovert, enthusiastic, communicative. Over-optimistic. Loses interest once initial
Explores opportunities. Develops contacts. enthusiasm has passed.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
307
Designed by Safran Consulting
SHAPER (SH)
Contribution Allowable Weaknesses
Challenging, dynamic, thrives on pressure. The Prone to provocation. Offends people’s feelings.
drive and courage to overcome obstacles.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
308
Designed by Safran Consulting
COORDINATOR (CO)

Contribution Allowable Weaknesses


Disciplined, reliable, conservative and efficient. Somewhat inflexible. Slow to respond to new
Turns ideas into practical actions. possibilities.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
309
Designed by Safran Consulting
MONITOR-EVALUATOR (ME)

Contribution Allowable Weaknesses


Sober, strategic and discerning. Sees all Lacks drive and ability to inspire others.
options. Judges accurately.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
310
Designed by Safran Consulting
TEAM WORKER (TW)

Contribution Allowable Weaknesses


Co-operative, mild, perceptive and diplomatic. Indecisive in crunch situations.
Listens, builds, averts friction.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
311
Designed by Safran Consulting
COMPLETER FINISHER (CF)

Contribution Allowable Weaknesses


Painstaking, conscientious, anxious. Searched Inclined to worry unduly. Reluctant to delegate.
out errors and omissions. Deliver on time.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
312
Designed by Safran Consulting
ROLES IN A SUCCESSFUL TEAM

 Everyone has a preferred team role

 As Black Belts you will need to take


different roles and be aware of how
to make your teams effective

 Using Belbin type analysis is an


excellent method to engage and
discuss team working in your teams

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
313
Designed by Safran Consulting
This document and the information it contains belong to
Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
314
Designed by Safran Consulting
DETAILS ON BELBIN TEAM ROLES

315 /
Ce document et les informations qu’il contient sont la propriété de Snecma. Ils ne doivent pas être copiés ni communiqués à un tiers sans l’autorisation préalable et écrite de Snecma.
CHAIRMAN (CH)

 The “Chairman” controls and organises the activities of his / her team, making best use of the
resources available.
 The “Chairman” is good at encouraging people to give their best, by helping them identify their
role and contribution and providing positive feedback on their performance. He / she is usually
adept at smoothing over disagreements using a blend of tact and firm control.
 He / she co-ordinates the use of resources from both within and outside the team and keeps
peoples’ efforts orientated towards the teams’ goals. He / she is good at identifying
weaknesses in the teams’ make up and if necessary expending his / her own role to cover
these weaknesses. He / she also delegates.
 When the going gets tough, the “Chairman” can exercise personal self discipline as acting as
a focal point for group effort.
 If the “Chairman” is also a manager, then he / she is in a position to employ his / her talents
overtly. In more junior positions, he / she should support team structure and co-ordination
without threatening the appointed leader.
 The “Chairman” has to avoid rigidity and obstinacy posing as grit and determination. He / she
also experiences difficulty in recognising superior ability in other team members. He / she
should also be careful not to abdicate from the leadership role in the face of powerful
competition.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
316
Designed by Safran Consulting
PLANT (PL)

 The “Plant” acts as a prime source of ideas and innovation for his / her team.

 He / she should concentrate his / her attention on basic strategies and major
issues, including formulating new ideas relevant to the team’s objectives.

 The “Plant” is also valuable in looking for possible breaks in approach to the
problems with which the group has been confronted for some time. However, he /
she should take care in timing his / her contributions; presenting his / her
proposals at appropriate moments to assist their positive reception.

 If the “Plant” is also a manager, he / she must exercise considerable self-


discipline in listening to his / her team comments on his / her proposals
(particularly those with monitoring / evaluating skills). He / she must also take
care that the stresses of management does not stifle his / her creative input.

 In less senior roles, the “Plant” should take care to use his / her creativity in the
interest of the team, rather than on personal issues. He / she must be prepared
for criticism and having ideas rejected but he / she should not become inhibited
about putting his / her ideas forward especially in a dominant or over critical
group.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
317
Designed by Safran Consulting
RESOURCE INVESTIGATOR (RI)

 The “Resource Investigator” explores outside resources and develops contacts of


use to his / her team.
 The “Resource Investigator” has an ability to get on with people quickly and easily
and extend the range of the teams contacts. He / she has an interest in new ideas and
methods which leads him / her to explore possibilities outside his / her immediate
working environment.
 He / she may also expand the role of the teams’ point of contact with outside bodies
by keeping up to date on all developments that may be relevant to the team.
 He / she may also maintain good relationships with the team and encourage fellow
team members to make best use of their talents – especially in times of pressure and
crisis.
 The “Resource Investigator” should avoid relaxing too much when the pressure of
work eases and allowing his / her sociability to lead him / her into unproductive use
of time.
 He / she may also get too involved with his / her own ideas at the expense of
exploring others and may reject ideas or information before submitting them into his /
her team for an opinion.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
318
Designed by Safran Consulting
SHAPER (SH)

 The person with the highest SH score should be allocated the role of the
“Shaper”. He / she gives shape to the groups efforts and ensures that it
keeps on track, keeps together and achieves its goals:
 gives shape to efforts and its application
 seeks patterns in discussion
 tries to unite ideas, objectives and practical considerations
 anxious to reach decisions and action stages
 results oriented, very competitive, intolerant of vagueness
 may be seen by outsiders as arrogant and abrasive
 can steamroller less confident members of the group

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
319
Designed by Safran Consulting
COORDINATOR (CO)

 The “Coordinator” translates the team’s plans into a practical working brief and
carries out that brief in a systematic fashion.

 The “Coordinator” is good at maintaining a steady systematic approach whatever


the pressures (or lack of them) that exists. He / she perseveres in the face of
difficulty and strives to meet targets. He / she ensures that the team’s tasks have
been structured and the objectives clearly outlined.

 The “Coordinator” is also good at providing practical support and back up to


other team members.

 If a manager, the “Coordinator’s” strengths lie in his / her concern for clarifying
objectives in practical terms and maintaining structure in a team.

 The “Coordinator” may lack flexibility; he / she must strike a balance between
perseverance and adaptability. He / she is a pragmatist and may find him / herself
unconstructively criticizing other team members’ ideas. He / she must also
ensure that his / her strong sense of personal identity does not lead him / her
into competing with other team members.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
320
Designed by Safran Consulting
MONITOR-EVALUATOR (ME)

 The “Monitor - Evaluator” analyses ideas and suggestions from within and outside the team and
evaluates their feasibility and practical value against his / her teams’ objectives.
 The “Monitor - Evaluator” has a high level of critical thinking ability which he / she must use
constructively in his / her teams interest. He / she can achieve a valuable blend of experimentation
and critical appraisal. He / she can build on others suggestions and help develop their ideas to a
relevant and practical fruition.
 The “Monitor - Evaluator” should be able to make firm but tactful cases against unsound approaches
and choose the appropriate moment for airing these concerns with the teams. The “Monitor -
Evaluator” should take care to develop a close working relationship with the teams’ “Plant” if there is
one.
 A successful “Monitor - Evaluator” combines high critical thinking with qualities of fair-mindedness,
practicality and receptivity to change. The role is often combined with another team role. If the
“Monitor - Evaluator” is also the team leader, he / she must take care that he / she does not dominate
the other members of the team and stifle their contribution.
 If the “Monitor - Evaluator” is a junior officer, he / she must develop ways of making his / her points
heard and not appearing as a threat. He / she must avoid becoming unduly sceptical and cynical.
 He / she must also ensure that his / her critical thinking is not used to his / her own advantage rather
than that of his / her team. He / she must also take care that he / she does not destructively debunk
other ideas. In such a case he / she might lower his / her team’s morale by being too critical or
damning. He / she must also ensure that his / her critical powers do not outweigh his / her receptivity.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
321
Designed by Safran Consulting
TEAM WORKER (TW)

 The “Team Worker” helps individual members achieve and maintain team
effectiveness.

 The “Team Worker” is good at observing the strengths and weaknesses of team
members and in supporting them in their strengths (e.g. building on
suggestions). He / she is good at underpinning their shortcomings by personal
assistance.

 The “Team Worker” is concerned to improve communications between team


members and to foster a sense of team spirit by setting an example in team
member behaviour.

 If a manager, the “Team Worker” should concentrate on developing others and


delegating. At a junior level, he / she should act as “a behind the scenes” helper.

 The “Team Worker” should avoid competing for status or dominance in the
group or siding with one member against another. He / she should also take care
not to be too ostentatious in the exercise of his / her team member function.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
322
Designed by Safran Consulting
COMPLETER FINISHER (CF)
 The “Completer” ensures that all the teams’ efforts are as near perfect as possible and that
nothing is overlooked.
 The “Completer” keeps an eye open for mistakes of all kinds and especially those that may fall
between the responsibilities of two team members.
 He / she is good at choosing areas of work in which finishing qualities are important and at
looking for mistakes in detail that may spoil the finished product. The “Completer” actively
searches for aspects of the work which need a more than usual degree of attention.
 The “Completer” constantly endeavours to raise the standard of all the teams’ activities by
vigilance and help as required and maintains a sense of urgency within the team.
 If the “Completer” is also a manager, he / she should be careful to keep his / her interference
with subordinates to a minimum and to pay careful attention to delegation.
 At a more junior level, he / she must employ tact and discretion and avoid earning a reputation
for “not seeing the wood for the trees”. He / she may well have a nervous drive that must be
controlled and directed if it is to have positive results.
 As a team member, the “Completer” has qualities of conscientiousness and perseverance
which will ensure that projects are completed to standard and to schedule. Usually, his / her
sense of duty prevents personal status from interfering with his / her work towards achieving
objectives. The “Completer” must be careful not to place undue emphasis on detail at the
expense of the overall plan and direction. There is a chance that he / she can affect team
morale by excessive worrying or destructive criticism.

This document and the information it contains belong to


Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran.
323
Designed by Safran Consulting
This document and the information it contains belong to
Safran. They must not be copied or communicated to any
other person without prior written authorisation from Safran. 324
Designed by Safran Consulting

You might also like