LSSGB

Technische Universität München
Methods and Tools for Process Improvement
Copyright
Dr. Reiner Hutwelker
reiner.hutwelker@TUM.de
© Dr. Reiner Hutwelker Lean Six Sigma 1

Six Sigma Technische Universität München
Six Sigma is a concept with several meanings –– itit links

links Practice
practice with Theory
methods and
and statistics
Statistics …
1. Improvement Program: increase customer satisfaction and reduce costs

A
• Problems of corporate products & services offer potentials for the improvement of processes Board
• These improvement potentials are identified by customer, by the management and by employees A1 A2 A3
value stream
• Potentials are defined as Six Sigma projects and then supervised by Sponsors from management
results
• Projects are implemented by Green-/ Black-Belts, according to the method and tools of the DMAIC cycle
2. Methodical Approach: chronologically linked tools for the implementation of the project
• The DMAIC-Cycle is a problem-solving approach in five phases for the implementation of projects Control Define
• The approach is based on an open box of chronologically linked rational-logical and statistical tools
Improve Measure
• The tools serve to define problems, analyze causes, develop solutions and control their success
Analyse
• Additional tools serve to prepare the business case and to manage the project
3. Statistical Basis: process capability, process control, problem-cause relations, improvements

- tolerance range -
• The sigma level is a performance measure of the process capability, i.e. the excellence of the outputs
Number of Outputs
• At a level of 6 sigma (6σ), only 3.4 errors are expected in 1 million outputs (Motorola’s short-term-sigma) Influences
in the
Process
• Control charts indicate the variation of the process performance, i.e. variation of the output over time
Measure of Output
• Statistical tests identify the main problems, the relationships to their causes and degree of improvement
… to analyse and realize potentials and to verify improvements

Six Sigma
Improvement Program

Six Sigma > 1. Improvement Program: factors of success and failure Technische Universität München
The quality of a Six Sigma program is determined by the management, …
Success factors for Six Sigma Programs:

• Involvement of the management via policy deployment and target agreement
• Specific career options for successful Green and Black Belts
Number of projects implemented per Green Belt:

• In successful programs: 1.6 projects per year (Mean)
• In terminated programs: 1.2 projects in the lifespan of the Six Sigma program (Mean)
Unsuccessful Six Sigma Programs:

• Termination after 5.7 years (Mean)
• Reasons for termination:
- new management appointed, e.g. after reengineering of the company or mergers
- no more suitable improvement topics identified, conviction that nothing more can be improved
Leyendecker, B., Schindewolf, S., Hutwelker, R., Weigel, H.: Erfolgsfaktoren für die Etablierung von Six Sigma (QZ 7/ 2011)
… career options for Belts and the continuous identification of suitable project topics
Six Sigma > 1. Improvement Program: Sources and realization of improvement potentials Technische Universität München
The sustainability of a Six Sigma corporate program is ensured by the continuous ...
Sources of improvement potentials Realization of improvement potentials

Management Customer Steering Board
 Targets for important key figures  Ratings  Management
 Deficiencies in key figures  Complaints  Process owner (as potential sponsors)
 Process efficiency and effectiveness  Purchases and cancellations  Master Black Belt
Analyses Employees Project Team

 Suggestions scheme  Clarifications on inputs  Sponsor
 Value stream analysis  Negative influences on the process  Green/Black Belt
 5S/ 5A  Deficiencies in outputs  Subject Matter Experts
implementation realized potential

of projects (customer satisfaction,
cost reduction)
pool of improvement topics evaluation &

selection of
topics
… identification of improvement topics and their implementation in projects

Six Sigma > 1. Improvement Program: Development, organization and maintenance Technische Universität München
An improvement program for operational excellence should be developed and ...

9. Objectives
Management Dashboard
11. end to end chain of related QAC performance indicators 1. Identify important outputs/ customer touch points
2. Analyse data on quality, availability, and consumption
3.
Influences Problems 5. Influences Problems (QAC), leading to the process costs
3. Pre-define obvious deviations of the outputs as problems
1. 2.
Output 4. Identify the associated process, its key inputs, and the
Output
Input
Input
team members
Process 6. Process
5. Pre-define obvious influences that trigger the problems
KPI’s KPI’s
4. 6. Structure the upstream process up to the first relevant
(external) input
functional team functional team
12. 7. Create a cross-functional team consisting of Lean Six
Sigma Green Belt candidates
7. cross-functional team 10. Projects 8. Structure further important processes from the last
output to the first relevant input
KPI’s 9. Determine objectives for important outputs (touch points)
10. Identify, evaluate and select improvement topics,
8. assign teams, implement projects
KPI’s 11. Derive QAC performance indicators, and target values
for intermediate outputs along the process
12. Integrate KPI’s to a management dashboard
13. Collect data continuously and adapt initial KPI’s,
8. 14. Continuously improve and/ or define further projects to
achieve the objectives/ operational excellence
KPI’s
15. Review project results, acknowledge achievements
… organized along cores processes, evaluated by KPI’s - with improvements triggered by deviations from requirements
Six Sigma Method

Six Sigma > Method >> Structure Observations Technische Universität München
Observe by asking questions – Classify to generalize and Describe to concretize
same family bakery products
generalize
different same type savory-pastry sweet-pastry bread
Classification
What is a Cookie ?
Object
What has a Cookie ?
Description
attributes ingredients form taste/ delivery/...

concretize
shape diameter height weight colour

(cm) (cm) (g) (RGB)
deviant conform values cylinder 10 0.7 10 239, 226, 178
Each of these assignments means: modelling and measuring the reality

Six Sigma > Method >> Problems Technische Universität München
Observations in the bakery assigned to the: Problems  Output – Attribute – Deviation
problem type descriptors primarily affected Problems (Y) of Outputs Description of Problems (Y)
- false, defective, insufficient, missing or unreliable
Object Attribute Deviation  Outputs are material/ immaterial objects, e.g.
- cookie / video / decision
Quality - not or only partially suitable for the targeted purpose
(internal)
Customer
Cookie taste bad Q  Objects are characterised by attributes, e.g.
- risky for safety and security
- taste / audience-retention / alternative
- not available in the required quantity (too much/ too few)

Cookie diameter > 10 cm Q  Attributes deviate from required state or value
- bad / discontinued / false alarm
Availability - not available at the required time (too early/ too late)
(internal)
Customer Cookie delivery 1 week early A
- not available at the required location
Classification of Problems (Y)
- time consuming in preparation or waiting times in between Cookie ingredients > 50g waste C  Problems are output attribute deviations in the categories:
- Quality (in a given or required attribute)
Consumption - wasteful in consumption of input or resources (active) Management
- losses of input or resources during creation (passive)

Cookie energy waste C - Availability (right quantity, time, and location)
- Consumption (losses or waste of inputs or resources)
Problems can be classified into the Categories of: Quality, Availability and Consumption
Six Sigma > Method >> Causes Technische Universität München
Observations in the bakery assigned to the: Causes (x), …
Causes (x) in the Process Classification of Causes (x)

 Negative influences from:
Order specification on taste ambiguous I - Inputs (xi),
- Activities (xa)
Chocolate nibbled from ingredients A - Methods (xm)
- Resources (xr)
trigger and/ or amplify problems
Date on calendar wrong R
Timer setup wrong R Method
Oven extremely preheated R Input Activity Output
Resource
Ingredients individually determined M
Cookies shaped variably M
… and classified into the Categories of: Input, Method, Resources and Activity
Six Sigma > Method >> Effects Technische Universität München
Observations in the bakery assigned to the: Effects (Z), …
Problems (Y) of Outputs Classification of Problems (Y)

Object Attribute Deviation  Problems are deviations in:
Quality - Quality (in a given or required attribute)
Problem (Y)
Cookie taste bad Q - Availability (right quantity, time, and location)
- Consumption (losses or waste of inputs or resources)
Cookie diameter > 10 cm Q  of (intermediate) outputs
Cookie delivery 1 week early A

Effects
(Z)
Cookie ingredients > 50g waste C
Costs
Satisfaction
Cookie energy waste C
Effects (Z) on Customer & Company Classification of Effects (Z)
Consumption Availability  Problems lead to effects in the categories:
Problem (Y) Problem (Y)
Customer dissatisfied S - Costs, resulting e.g. by inspection, rework, scrap, lost revenue
- Satisfaction, expressed e.g. by ratings, claims, migration
 and determine the financial potential of a project
Customer claims compensation C
We solve problems to reduce costs Company quality-costs increased C
NOT
Reduce costs to solve problems Company revenue losses C
… and classified into the Categories of: Costs and Satisfaction
Six Sigma > Method >> Model Observations Technische Universität München
Observations in the bakery assigned to the categories: Solutions (S), Causes (x), Problems (Y) and Effects (Z) and …
Causes (x) in the Process Problems (Y) of Outputs Classification of Problems (Y)
Object Attribute Deviation  Problems are deviations in:
Order specification on taste ambiguous I - Quality (in a given or required attribute)
Cookie taste bad Q - Availability (right quantity, time, and location)
Chocolate nibbled from ingredients A - Consumption (losses or waste of inputs or resources)
Cookie diameter > 10 cm Q  of (intermediate) outputs
Date on calendar wrong R Classification of Causes (x)
Cookie delivery 1 week early A  Negative influences in the upstream process from:
Timer setup wrong R - Inputs (xi),
Cookie ingredients > 50g waste C - Methods (xm),
- Resources (xr) and/ or
Oven extremely preheated R - Activities (xa)
Cookie energy waste C  trigger and/ or amplify problems
Ingredients individually determined M Effects (Z) on Customer & Company Classification of Effects (Z)
Cookies shaped variably M Customer dissatisfied S
 Problems lead to effects in the categories:
- Costs (e.g. by inspection, rework, scrap, lost revenue)
- Satisfaction (e.g. by ratings, claims, migration)
Solutions (S) to control Causes  and determine the financial potential of a project
Customer compensation claims C
Determine correct oven temperature a Classification of Solutions (S)
Use cookie blanks c Company quality costs increased C  Solutions aim to:
- adjust,
- circumvent or
Forbid nibbling e Company revenue loss C - eliminate
 causes (negative influences) of the problem
… within these categories further classified

Effects (Z), Problems (Y), Causes (x) and Solutions (S) hierarchically structured in a fault-tree
Effects (Z) Company revenue losses
Customer compensation claims (C)
Customer dissatisfied (S) Company quality costs increased (C)
Cookie taste bad Cookie delivery 1 w. early Cookie diameter > 10 cm Cookie ingredients Cookie energy waste
time
Problems (Y) (Q) (A) (Q) > 50 g waste (C) (C)
Triggering Date on calendar wrong Cookies shaped variably Oven extremely preheated
(R) (M) (R)
Causes (x)
.
Order specification on Ingredients individually Chocolate nibbled from
. taste ambiguous (I) determined (M) ingredients (A)
Timer setup wrong (R)
.
.
Root Causes (x)
Solutions (S) Determine correct oven

Use cookie blanks (c) Forbid nibbling (e)
temperature (a)
This modelled problem twin will be specified and statistically analysed to focus solutions on the important root causes
The Process-Problem-Solving Model …
Input Check Process Output Check

Methods - guide the Activity
Input ok? e.g. Algorithms/ Instructions/ Rules/ Expertise/ Habits Output ok?
yes: pass §! Operating yes: pass
Rules, instructions
no: reject Standards Recipe no: reject
Influences of Methods (xm)-
Influences of Activities (xa)

+
target range target range
Performance Performance
Objects
Activities - transform or transfer the Input
+ e.g. digital/ manual/ mechanical +
Influences of
Inputs (xi) Object Δ = Problem (Y)
- -
I OI OI OI O
+ -
Consumption
Consumption
Quality
Quality
Availability
Influences of Resources (xr)
Availability

Resources - execute the Activity
e.g. Employees/ Machines/ Tools/ Energy
Problem (Y) as deviation in:

Causes from: Input (xi), Methods (xm), Resources (xr), and Activities (xa)
Quality, Availability or Consumption
Effect (Z) on:

Solutions (S) to: adjust, circumvent or eliminate, the root-causes
Costs & Satisfaction
… links a Problem and its Effects, its Causes and its Solution to the underlying Process, its Inputs and Outputs
Three important Resolution Levels of a Process …
Project-Topic
1. Project-Topic: Determine the project field
Determine the critical output, the main activity and the first required inputs
Example: ingredients  baking  cookies
SIPOC VoC/ VoB

2. SIPOC/ Voice of Customer/ Voice of Business: Identify intermediate outputs
Divide the project field into 8-12 core process steps:
Example: process order  order; determine ingredients  shopping-list; …
Process-Analysis
3. Process-Mapping/ -Analysis: Identify the trigger/ amplifier of the problems
Divide the core process steps very detailed into input-activity-output units
Example: Order Content ambiguous; Chocolate nibbled from ingredients, …
… serve different Purposes in a Six Sigma Project

Six Sigma > 3. Statistical Basis >> Process Capability, Process Control, Analysis of Influences and Improvements Technische Universität München
Output requirements are specified by a tolerance range for each important attribute (LSL-USL)
xbar/μ
LSL LSL USL USL xbar = 0
s = 1
LSL =-6
USL = 6
Sigma-Level= 6
There are 6 standard

Frequency / Probability
6s / σ deviations distance
= from the mean to each
0,0000002% specification limit
defects
Process Capability: 6 sigma

(z-values from standard normal distribution)
xbar/μ
68% xbar = 0
s = 2
LSL =-6
USL = 6
95%
Sigma-Level= 3
There are 3 standard

3s / σ deviations distance
3 5 7 9 11 13 15 17 19 g = from the mean to each
-2σ -1σ μ+1σ +2σ 0,27%
defects
specification limit
-2s -1s Mean +1s +2s

(xbar) Process Capability: 3 sigma
(z-values from standard normal distribution)
The more outputs are within the Lower (LSL) and Upper Specification Limits (USL), the more capable the process is
Six Sigma > 3. Statistical Basis >> Control scattering and position of key performance indicators Technische Universität München
Six Sigma's goal is to align performance indicators of products and services …

scattering
not capable capable
Monday Tuesday
not capable
centering
Wednesday Thursday
capable
… to the center and to minimize their variability, according to customer requirements

Six Sigma > 3. Statistical Basis >> Comparison of defect rate and sigma level Technische Universität München
Percentages and sigma levels can be transformed into each other

… and they should be compared with data
Yield (%) Sigma Level Sigma Level measured in:
examples from everyday life 99% yield would mean: in reality happend: short-term
long-term
(short-term) (long-term)
short-term interval long-term interval
11 crashes of aircraft daily arriving and 37.6 million flights with 4 fatal accidents
Air traffic safety 99,9999894% (6,69) 5,19 adjustment:
departing Munich Airport (2015, worldwide) no adjustment
short-term interval short-term-data
necessary
Data were - 1,5s
139 victims/ 349 million passengers
U-train victims of violence 1 victim/ 100 passengers 99,9999602% (6,44) 4,94 measured in: adjustment:
(Germany) no adjustment
long-term interval long-term-data
necessary
Reference: Errors in typical paperback 6 sigma performance corresponds to: + 1,5s
1) 2)
3 errors per page; 900 errors per book 99,9996600% 6 4,5
(300 words/page; 300 pages/book) 1 error per 3,27 books
10,000 industrial accidents/ 1 million 14,76 industrial accidents/ 1 million
Work safety 99,9985240% (5,68) 4,18
working hours working hours (Germany)
10,000 counterfeit notes/ 1 million bank 50 counterfeit notes/ 1 million bank notes
Counterfeit money 99,9950000% (5,39) 3,89
notes (Europe)
Error in hospital treatment resulting in 190,000 errors resulting death/ 19 million 19,000 errors resulting death/ 19 million
99,9000000% (4,59) 3,09
death hospital treatments hospital treatments (Germany)
62000 revoked licenses / 6.2 million 158 revoked licenses/ 6.2 million visitors
Driving licences revoked for alcohol 99,9974516% 4,05 (2,55)
visitors on Oktoberfest 2017
Executed death penalty on innocent
1 innocent/ 100 executed 4 innocent/ 100 executed (USA) 96,0000000% (3,25) 1,75
people
1) Sigma Level is based on z-values of the standard normal distribution +/- 1,5s as adjustment for time-dependent performance drifts, according to Mikel J. Harry (Motorola)**
2) Sigma Level is based on z-values of the standard normal distribution according to Carl Friedrich Gauss
The appropriate target for a desired performance level varies depending on the industry, the process,
the requirements of management and the specification of customers.
Naturally, an automated process in production can achieve a higher yield than the sales process in a car
dealership. 99% yield can therefore be excellent in one process and alarming in another.
Each yield should thus be evaluated within meaningful and appropriate specification limits.
** Harry, M. J. (1988): The Nature of Six Sigma Quality; Motorola University Press
Examples of performance levels in everyday life, given in yield rates and sigma levels
Six Sigma projects follow a scientific approach …
Relationship assumptions
+ hypotheses
Difference
Reality + tests
Observation
x Y
of status &
deviations
allow to
- describe,
- explain and
C D - predict the reality
Project
Model I M Statistics
x Y Y=Af(x)
Input Process Output
Y Scatterplot Y vs. x
hypotheses This scientific approach

Cause Problem
is realized in the phases
Solution Effect
x DEFINE, MEASURE and ANALYSE
Process-Problem-Model Scales, Measurement
Assumptions about:
Relations & Differences
Hypotheses: Y= f(x)/ Y1=Y2 of the DMAIC cycle
Statistical Tests
Rational Analysis
… to solve the equation: Y= f(x) – the problem is a function of its negative influences
Six Sigma > Method >> DMAIC Technische Universität München
The DMAIC is a chronological guideline for problem solving …
Effects (Z) and Sustainability (Y, PM) Problems (Y), their Effects (Z) and Project Mgmt (PM)
 Process Performance (of critical Y)  Project-Topic & -Definition (Problems Y, Effects Z)
 Test of Improvements (Yt1 vs. Yt2)  SIPOC (project field, from first influence x to last problem Y)
 Financial/ other Benefits (Z) Control Define  Voice-to-Criticals (identify critical problems Y)
 Process-Management-Plan (PM)  Project-Charter (organize project, PM)
 Stakeholder Communication (reduce resistance, PM)
Solutions (S) Improve Measure Influences (x) and their Relations to Problems (Y)
 Solutions (eliminate root causes x’)  Input-Analysis (identify influences x)
 FMEA (reduce risks of solutions)  Process-Mapping/ Analysis (identify influences x)
 Action-Plan (adapt and specify solutions) Analyse  C&E Matrix (assumptions about x-Y relationships)
 Implement the Measures  Data-Collection-Plan (units, scale levels, sampling plan for x, Y)
 Hypotheses (formalized assumptions for risky x-Y-pairs)
Root Causes (x’) of Influences (x) of Problems (Y)
 Data Evaluation (data inspection and charts for x, Y)
 Process Performance (control over time & capability of Y)
 Test of Hypotheses (test relationships between influences x and problems Y)
 Root-Cause-Analysis (identify root causes x’ of influences x on problems Y)
... including a sequence of rational and statistical tools for Six Sigma projects

DMAIC > Sequence Technische Universität München
DEFINE: From the identification of problems (Y) to the Project-Charter
 Identify and define a new Six Sigma project (Project-Topic & Project-Definition)
Risk of Topic: 72%
- Identify a weakness in a process and/ or a deviation in a product/ service,
evaluate the relevance for the company and check the suitability for Six Sigma (Project-Topic)
Suitability
- Specify the problems with the product/ service and the underlying process, for Six Sigma:
89%
evaluate the effects of the problems on the customer and on the business (Project-Definition)
 Delimitate and structure the application area of the project (SIPOC) Supplier-Input-Process-Output-Customer

- Delimitate the area of the project by determining the first Input and the last output of the process
- Identify all important intermediate outputs of the process within the determined area
- Identify the underlying core process steps and their important inputs Input Process Output
- Identify the related internal/external supplier of the inputs and the internal/external customer of the outputs
 Listen to the customer and manager (VoC/VoB), filter their complaints and requirements (CCR/ VoC/VoB > CCR/CBR > CtQ
degree of satisfaction
delighter
CBR), prioritize the problems (KANO) and identify the Critical to Quality (CtQ) (Voice to Critical) My needs
and more is better
- Interview customer and manager about their needs and complaints on the (intermediate) outputs (VoC / VoB) complaints
- Translate complaints into requirements of customer and manager for the attributes of these outputs (CCR / CBR) must be
- Assign required attributes to the categories: Quality, Availability and Consumption (type of problem) fulfillment of Requirement
- Evaluate the importance of the requirements (KANO-Model) and determine the degree of their fulfillment (CtQ)
 Define the scope, define the targets, build the team and agree these conditions (Project-Charter) Project-Charter
- Summarize the relevance of the project for the customer and the company (business case) Business Problem
- Summarize information about the process, its important outputs and their problems Relevance
- Agree on the targets for the important requirements (CtQ´s) and on the scope of the project Scope/ Experts
Belt-Team
- Assemble the team with Sponsor, process owner, Black/ Green Belts, subject matter experts and controlling Targets
Management
Summary DEFINE: Important Problems (Y) are defined and a contract on the targets, scope and team of the project is agreed.
Outlook MEASURE: Identify the Influences (x) on the Problems (Y), develop related Hypothesis and a Data Collection Plan
Identification and Definition of a Six Sigma Project

DMAIC > Project-Topic Technische Universität München
sigmaGuide: content
Phase Tool Purpose
Summary Summary of the purpose of the tools in sigmaGuide
Define Part 1 Identify a topic for a Six Sigma Project

Project-Topic Identify Problems of the daily work as a potential for Improvement
Part 2 Define a Six Sigma Project
Process & Output Describe the Process and its Output
Problem Describe the Problem
Effect Describe the Effect of the Problem
Solution Outline Solution-Ideas - if already available
Project-Definition Summary: Process, Output, Problem and Effect
Define Part 3 Implement a Six Sigma Project

SIPOC Structure the Process in its important steps, with related Supplier, Inputs and Outputs and Customer
Voice to Critical Identify Voice of Business/ Customer (VoC/ VoB), Critical Business/ Customer Requirements (CCR/ CBR), Problems and CtQ´s
Voice to Critical (Summary) Summary: VoC, VoB, CCR, CBR, Problems and CtQ´s
Chart: Y CtQs Bar-Chart Problems and their Severity
Chart: Y CtQs Kano-Chart Fulfillment of requirements resulting satisfaction according to Kano-characteristics
Project-Charter Complete and sign the Project-Charter
Stakeholder-Communication Identify the Stakeholder and develop a Communication-Plan
Measure Input-Analysis Describe Inputs (xI) of the Process, Requirements on the Inputs and Deficiencies
Chart: Influence of xI Display: Strength of negative Influences of the Inputs (xI) on the Outputs (Y)
Process-Mapping and -Analysis Describe Activities of the Process (xP), related Inputs (xI) and Outputs (Y) and negative Influences
Chart: Influence of xP Display: Strength of negative Influences of the Activities (xP) on the Outputs (Y)
C&E Matrix xY Evaluate relationships between negative Influences of the Inputs (xI) and the Activities of the Process (xP) on the Outputs (Y)
Chart: C&E Heatmap Display: Risks of the Influences from Inputs (xI) and Activities (xP) on the Outputs (Y)
Operationalise Measurands of Inputs (xI), Activities (xP) and Outputs (Y);
Data-Collection-Plan
Recommendation for appropriate: Charts, Parameter, Process-Capability-Indices, Control-Charts, One-Sample-Tests
Hypothesis Overview of all automatically generated Hypothesis, prioritized by their Risk; Recommendation for appropriate statistical Tests
Analyse Process-Capability Calculate Process-Capability (Yield, DPMO, Pp/ Ppk, Sigma-Level, …)
Improve Solution-Ideas Develop Solution-Ideas to eliminate the Root-Causes
Chart: Solution-Selection Overview to Efforts, Benefits and Effect of Solutions
Action-Plan Specify Measures to implement the Solution-Ideas
FMEA Analyse the Risks of Measures (FMEA:= Failure Mode and Effects Analysis)
Control Process-Capability Calculate Process-Capability (Yield, DPMO, Pp/ Ppk, Sigma-Level, …)
Process-Management-Plan Define measures to sustainably maintain the process-improvements
Summary and benefits Summarize the results of the phases and demonstrate the financial and other benefits of the project
Please select the marked tool

The identification of a Project-Topic is the first step to a Six Sigma project
1/ 7 Please describe the weakness, which you noticed. 1/ 7 Please describe the weakness, which you noticed.
The bakery Cookie du Chef has several Problems. Most important: sometimes the cookies taste bad. Sometimes the cookies are delivered too early.
Additionally there is some waste of ingredients and energy.
2/ 7 Please summarize your description in the following two categories.

...? ...? products/ services, that we create have a poor quality
please select an answer please select an answer
3/ 7 Please name the products or services you mean (singular form).

Cookie
4/ 7 How often does the quality defect occur? never x always
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
please mark the applicable box with x
5/ 7 Who is directly affected by the negative impact of the quality defect?

...? The quality defect directly affects the internal/ external customer
please select an answer
6/ 7 How severe is the negative impact on the internal/ negligible x disastrous
1 2 3 4 5 6 7 8 9 10
external customer? 1 2 3 4 5 6 7 8 9 10
please mark the applicable box with x
7/ 7 Which proportion of the solution can your own no contribution x complete solution
1 2 3 4 5 6 7 8 9 10
department probably contribute with its own 1 2 3 4 5 6 7 8 9 10
resources? please mark the applicable box with x
The video: Cookie du Chef shows, that there are improvement potentials in this bakery
Every employee is able to describe observed weaknesses in the work environment

The Project-Topic summarizes the given information in statements and evaluations

o k
Please check the summary of our dialogue.
Summary:
The bakery Cookie du Chef has several Problems. Most important: sometimes the cookies taste bad. Sometimes the cookies are delivered too early.
Cookies - i.e. Products/ services, that we create - have a poor quality. The quality defect occurs very often and has a strong impact on the internal/ external
customer. The problem can be solved with a very big contribution by the own department.
Relevance of the topic: 35%

Suitability for method: Six Sigma
Solvable by own department up to: 80%
If this summary of answers does not seem to make sense to you, please correct your answers.
If this summary seems to make sense to you, then please go to section: Process & Output
or send this File to:
Please check the summary of our controlled dialogue which shows your answers in context.
If the content of this or subsequent summaries is incorrect or pointless, then please change your given answers.
Based on this summary, the management can decide whether further investigations are necessary
Criteria for Lean Six Sigma Project-Topics

Focus: Problem of the Output
Type of problem: Deviation in Quality, Availability and Consumption
Effect of the problem on: (internal) Customer and Business
Measurability:
Statistical tests: Sample size > 30 xY value pairs collectable within a month
Project benefit: 20 - 60k€ for a Green Belt and > 100k€ for a Black Belt project
Implementation time: Not longer than half a year (typically)
Persons to be involved: - Customer for interviews
- Experts for workshops (Process-Mapping; Root-Cause-Analysis & Solutions)
- Employees to implement the developed measures
For the Lean Six Sigma certification you must solve a Quality, Availability and Consumption Problem
DMAIC > Project-Definition Technische Universität München
Project-Definition: Summary of all collected information about the topic, process and output, problems, …
100% 100%
90% 90%
Project-Definition 80% 80%
70% 70%
60% 60%
Wichtigkeit/ Verfügbarkeit/
50% 50%
The bakery Cookie du Chef has several Problems. Most important: sometimes the cookies taste bad. Sometimes the cookies are delivered too early. Importance Availability
40% 40%
30% 30%
20% 20%
10% 10%
0% 0%
COOKIE - i.e. products/ services, that we create - have a poor quality. The quality defect occurs very often and has a strong impact on the internal/ external 0% 10% 20% 30% 40% 50% 60% 70% 80% 90%100% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90%100%
customer. The problem can be solved with a very big contribution by the own department.
Dringlichkeit/ Urgency Qualität/ Quality
Relevance of the topic: 35% Summary: Voice of Customer

The satisfaction of the external customers with the:
Suitability for method: Six Sigma
- Quality of COOKIE is: 20%.
Solvable by own department up to: 80% - Availability of COOKIE is: 40%.
Section 1: Process and Output Section 4: Solution
Summary: Solution Idea to 1. Problem
more sugar in the mix

The Product COOKIE is a tangible final Output for external Customers and is in the Creation Process BAKE COOKIES within a year 13 - 52 times
generated. Important Input of the Process to generate the Product COOKIE is: BUTTER, SUGAR, FLOUR, CHOCOLATE.
Solution Idea to 2. Problem
Section 2: Problem
Summary:
1. Problem: COOKIE TASTE BAD. COOKIE fulfills the requirement on Quality (is error-free) in 30%. Solution Idea to 3. Problem
2. Problem: COOKIE DELIVERY TOO EARLY. COOKIE fulfills the requirement on Availability (just in time) in 70%.
additional Information
3. Problem: COOKIE INGREDIENTS WASTED. COOKIE fulfills the requirement on efficient utilisation of means (no waste of Input, Resources) in 50%.
Your additional comments, advices, feedback … are very appreciated.
Section 3: Effect
Summary: Voice of Business

The satisfaction of the process-owners with the Consumption in the Creation Process of the COOKIE is: 40%.
The total costs of the specified 3 problems are estimated by 100€ / year. Personal Data
They are primarily the result of quality costs due to scrap and additional expenditure. First Name Reiner Surname Hutwelker
The solution of the problems is rated as: Unit Cookies Location Munich
- major URGENT (80%-Level) Telephone 123456789 eMail reiner.hutwelker@sigmaLogic.de
- major IMPORTANT (80%-Level)
effects, available solution ideas and some details about the author – ready for evaluation by the Six Sigma Board
SIPOC, Voice-to-Criticals, Project-Charter, Stakeholder Communication

DMAIC > SIPOC Technische Universität München
Specify the project limits by the output with the last problem and the first influential input …
SIPOC
Process-
Practice Tips:
Supplier Input (xI) Process (xMR) Output (Y) Customer
Step  Indicate the subsequent
1 Customer (Child) Request (cookie-type & delivery-date) check Disposability Order (cookie-type; delivery-date) Logistic & Procurement Service
development stages of the objects
2 Order (cookie-type; delivery-date)
determine necessary & missing
Ingredients
Shopping-List Logistic & Procurement Service in Input and Output by adding the
respective latest characteristic
3 Supermarket Ingredients assemble Ingredients Ingredients (complete) Logistic & Procurement Service
attribute of the object in brackets,
4 Start-Signal prepare Workspace Workspace (clean) Production Support e.g.: Cookies (baked).
5 Ingredients (complete) weigh Ingredients Ingredients (weighed) Production Support  Typical number of (core) Process-
6 Ingredients (weighed) knead Ingredients Dough (ball) Chef du Cookie Steps in a SIPOC: 8 – 12
7 Dough (ball) roll out Dough Dough (sheet) Chef du Cookie  To shift an entry: never use cut &
paste – instead use: copy & paste
8 Dough (sheet) cut out Cookies Cookies (raw) Chef du Cookie
and delete unnecessary cells
9 Cookies (raw) bake Cookies Cookies (baked) Chef du Cookie
 If an Output in one step serves as
10 Tin Factory Tin pack Cookies in Tin Cookies (boxed) Production Support an Input in a subsequent step, then
11 Cookies (boxed) deliver Cookies Cookies (delivered) Logistic & Procurement Service copy the Output and paste it as new
Input
12
 If more than one object enters the
SIPOC process as Input, then only enter
the new external object as Input
The SIPOC is a table of the project-field. Its purpose is to:
- define the limits of the project, from the first Input to the last Output. The last Output should be the Output with the chronological last Problem from the Problem-Statement of the
 Describe the Activities of the
Project-Definition. Select the first Input such that the area between the first Input and the last Output covers all triggers of the last Problem. Process with a verb, followed by a
- structure the project-field from the first Input (top left) to the last Output (bottom right) line by line with the important Input-Process-Output steps. noun, e.g.: bake Cookies
… and structure the process-field row by row by the important process-steps

(Supplier - Input - Process - Output - Customer)
DMAIC > SIPOC Technische Universität München
Instructions
1. Start with the Output
Outputs are the results of Activities in the Process-Steps. They can be material Objects (e.g. Cookie) or immaterial (e.g. Order, Decision).
Sometimes for the immaterial Outputs it is not immediately clear, how the Output can be identified and discriminated from the Activity of the Process-Step. The following
guidelines might be helpful.
In the area of Process-Improvement its necessary to work with specific examples to get a realistic idea of the situation.
So at first develop an idea of the Output from a concrete example or even better: take a photo or a screenshot. If you only have a general idea of the Output in mind (e.g.
Management, Information) you might also not be able to specify its Requirements and Problems.
If you visualize at least one concrete example of the Output, you will be able to describe the Output with just one single noun (e.g. Order) or a composite noun (e.g. Shopping-
List). This is important, because now you are able to:
- identify a given class of Outputs (e.g. the class of Shopping-Lists, the class of Orders, the class of Cookies),
- specify the variation of the Outputs within their class (e.g. Cookie salty vs. Cookie sweet) and
- count Outputs (one, two, … Cookies).
This is a prerequisite, to later measure the attributes of the Outputs (see below).
In the different development steps of the Process sometimes the same noun is used for the different intermediate Outputs. In this case it is helpful to specify the status in
brackets to discriminate its development steps (e.g. Cookie (raw), Cookie (baked), Cookie (boxed), Cookie (delivered)).
Please specify all intermediate Outputs, between the first Input and the final Output of the Project, order them chronologically and enter them into the Output-Column of your
SIPOC.
2. Specify the Process-Steps

Start with the last Output (bottom right) and identify the upstream Process-Step that leads to this Output.
Please take care, to describe the activity of the Process-Step by using a precise verb (e.g. weigh Ingredients). This will assure that you do can differentiate it from another
Output or an Input.
It is not necessary here to describe the details of the Process. This will be done later in the Process-Mapping-Analysis. Purpose of the SIPOC is to structure the field of the
Project and to identify the most important Inputs, Activities and Outputs.
3. Specify the Input

Each Input of a Process-Step is an Output of an upstream Process-Step. For this please describe the Inputs also with just one single noun (e.g. Tin).
If an intermediate Output becomes the Input of a downstream Process-Step within the SIPOC, you should again write it down to show the flow of the Process.
If several Inputs come into the Process, then try to combine them in just one noun (e.g. Ingredients).
In the Input-Analysis (see below) the Inputs, which result as Outputs from within the SIPOC, will be shown in [brackets] to indicate, that they do not need to be evaluated.
4. Specify Supplier and Customer

Identify the Customer of those Outputs, which are delivered outside the SIPOC - at least the Customer of the Output in the bottom right corner.
Identify the Supplier of those Inputs, which are supplied from outside into the SIPOC - at least the Supplier of the Output in the upper left corner.
(Extract from sigmaGuide)

DMAIC > Voice to Criticals >> Structure and purpose Technische Universität München
Voice to Critical leads from the Voice of the Customer (VoC)/ Voice of Business (VoB) via the …
Voice/ statements of customer and manager (VoC/ VoB) Managers often complain about a lack
of process transparency and expect
about their wishes and complaints on an output more insight from the project.
This is not a problem - but is one of
Derivation of output attributes the standard tasks of a Six Sigma
underlying the wishes and complaints project due to the MEASURE- and
ANALYSE- and CONTROL-phase.
Definition of Critical Requirements of Customer (CCR) It makes sense to ask for more precise
and Business, i.e. manager (CBR) ideas about exactly what is to be made
transparent and take it into account in
the Process-Management-Plan.
Definition of the critical deviation of the output attribute as problem
Managers sometimes express
requests for specific solutions
Determination of the problem type (Quality; Availability; Consumption) already in the DEFINE phase.
Please note that determining a
Determination of the Kano-category of the problem solution in this phase contradicts the
(must-be; more-/ less-is-better; delighter) idea of DMAIC and a precise analysis
of the problems. Ask them to postpone
the solution ideas until the IMPROVE
Estimation of the frequency with which the requirements are met phase.
An exception are quick wins with
Calculation of the severity of the problem and their prioritization guaranteed success, which can be
leads to the problems that are critical to quality (CtQ) y
implemented immediately.
… Critical Requirements of Customers and Business (CCR/ CBR) to the critical problems (CtQ)
DMAIC > Voice to Criticals >> Procedure Technische Universität München
Procedure of the Voice-to-Criticals
Procedure to collect the necessary information, e.g. in interviews: Sources to collect the Voice of
Customer (VoC):
1. Select a relevant source and collect answers at least to the following questions: 1. Internal/ external customer
2. Which of these SIPOC outputs is causing you trouble? 2. Internal quality control,
3. Complaints (sales/ customer
3. What do you require from this output? care),
4. What complaints do you have about these outputs? 4. Warranty cases,
5. Dealers,
5. Which quality of output does your request and complaint relate to?
6. Customer satisfaction surveys,
6. How should this property be pronounced in the best case?
7. Social media and
7. How is this characteristic pronounced in the worst cases? 8. Own experiences and
experiences of friends/ colleagues
8. What is their satisfaction if your expectations are fully met?
9. What is their satisfaction if your expectations are not met?

Sources to collect the Voice of
10. What percentage of the outputs received meets your requirements? Business (VoB):
• Personal interviews with:
- Process owner
- Sponsor (if not the same)
Use as many information sources that are available to you – interviews are preferred
DMAIC > Voice to Criticals >> Kano-Model Technische Universität München
Identify the critical Requirements of the Customer and the Business (CtQ‘s), …
The KANO-Model serves to:

Customer is ...
- Classify the requirements of the Customer (CCR) and Business
excited delighter (CBR) and
- fitness
assure ecxitement
- Identify the critical requirements (CTQ´s= Critical to Quality), i.e.
delighter enhancing severity of the output problems.
excitement requirement - harmless
to teeth There are three categories, specified by their characteristics:
- eco-friendly 1. must be: These are the basic requirements of the customer/
more is better business which have to be fulfilled to prevent dissatisfaction.
indifferent/ performance requirement more is better
moderately - fresh 2. more/ less is better: These are the performance require-
satisfied - inexpensive ments of the customer/ business which should be fulfilled either
prevent dissatisfaction
must be - low carb as high/ as low as possible or straight to a certain value within
basic requirement a specific range.
(= Critical To Quality) must be
3. delighter: These are the excitement requirements, important
- mellow
for marketing but unimportant for Six Sigma, because they are
- crunchy
not necessarily expect, but lead to excitement if given.
- sweet
angry - non-toxic Focus is on the must be´s. These attributes determine the CTQ´s
of your product/ service, because they cannot be compensated by
0% Requirement fulfilled to … 100% other attributes of your product/ service.
… from the Severity of the must-be Problems of the Outputs

DMAIC > Voice to Criticals >> Tool Technische Universität München
Please answer the questions and check the summary
Y_01
Whose Requirement do you want to specify? Customer Source of Requirement: Customer
Please select an answer.
Which Output does the the Customer want to evaluate? Cookies (baked) Output: Cookies (baked)

What does the Customer require from COOKIES (BAKED)?
Cookies taste like those from Grandma Voice of Customer (VoC): Demand
COOKIES (BAKED) should …
Please quote the statement of the Customer.
Which deviation of COOKIES (BAKED) from the Requirement is
Cookies taste of nothing Voice of Customer (VoC): Complaint
problematic for the Customer ?
Please quote the statement of the Customer.
Which attribute of COOKIES (BAKED) is addressed in this
taste Attribute of COOKIES (BAKED): TASTE
Demand and Complaint?
Please enter just one Noun.
How should TASTE be according to the Requirement of the Critical Customer Requirement (CCR): COOKIES
mellow-crunchy-sweet (BAKED) TASTE MELLOW-CRUNCHY-SWEET
Customer?
Please enter just one adjective.
Problem: COOKIES (BAKED) TASTE CRUMBLY-
Which deviation of TASTE is problematic? crumbly-bland BLAND
Please enter just one adjective.

To which category does the Requirement TASTE MELLOW-
Quality (Faultlessness/ Fulfilment of Purpose) Requirement-Category of TASTE: Quality
CRUNCHY-SWEET vs. CRUMBLY-BLAND belong?
Which relevance does the Requirement COOKIES (BAKED)
TASTE MELLOW-CRUNCHY-SWEET have according to the Must-Be Kano-Category of TASTE: Must-Be
Kano-Model?
To what extend is the Requirement on TASTE of COOKIES Requirement for TASTE of COOKIES (BAKED) is
20% fulfilled to: 20%
(BAKED) actually fulfilled?
Please enter a value between: 0% - 100%.
Y_01: Output: Cookies (baked) | Voice of Customer (VoC): Demand: Cookies taste like those from Grandma; Complaint: Cookies taste of nothing | Critical Customer
Requirement (CCR): COOKIES (BAKED) TASTE MELLOW-CRUNCHY-SWEET | Problem: COOKIES (BAKED) TASTE CRUMBLY-BLAND | Requirement-Category of
TASTE: Quality | Kano-Category of TASTE: Must-Be | Requirement for TASTE of COOKIES (BAKED) is fulfilled to: 20% | The Severity of being unsatisfied with
TASTE of COOKIES (BAKED) is: 92% (Rank 1/ 5).
If the statements in the summary are pointless or cumbersome, please adapt the answers
DMAIC > Voice to Criticals >> Summary and details Technische Universität München
Summary and details of VoC/ VoB, Critical Requirements, CTQ´s for all Y´s
Summary: Voice of Customer (VoC), Voice of Business (VoB), Critical Requirements (CCR/ CBR), Problems, Severity, KANO and CtQ-Rank
Critical Business Requirement (CBR) or Critical to Quality
Y Voice of … Problem Kano-Category Severity
Critical Customer Requirement (CCR) (CtQ) Rank
Y_01 Cookies (baked) Cookies taste of nothing Customer CCR: Cookies (baked) taste mellow-crunchy-sweet Cookies (baked) taste crumbly-bland Must-Be 92% 1
Cookies (delivered) delivery > 1 hour too early/

Y_02 Cookies (delivered) delivered too early Customer CCR: Cookies (delivered) delivery in time More/Less-Is-Better 60% 2
late
Cookies (boxed) energy-consumption > 10%

Y_03 Cookies (boxed) waste of energy Management CBR: Cookies (boxed) energy-consumption minimal More/Less-Is-Better 30% 3
waste
Y_05 Cookies (baked) Cookies are too big Management CBR: Cookies (baked) diameter > 9cm and < 10 cm Cookies (baked) diameter > 10 cm More/Less-Is-Better 30% 3
Cookies (boxed) ingredient-consumption > 10%

Y_04 Cookies (boxed) waste of ingredients Management CBR: Cookies (boxed) ingredient-consumption minimal More/Less-Is-Better 20% 5
waste
Y_06
Details: Voice of Customer (VoC), Voice of Business (VoB), Critical Requirements (CCR/ CBR), Problems, Severity, KANO and CtQ-Rank
Y_01: Output: Cookies (baked) | Voice of Customer (VoC): Demand: Cookies taste like those from Grandma; Complaint: Cookies taste of nothing | Critical Customer Requirement (CCR): COOKIES (BAKED) TASTE MELLOW-CRUNCHY-SWEET | Problem: COOKIES (BAKED) TASTE CRUMBLY-BLAND | Requirement-Category of TASTE: Quality | Kano-Category of TASTE: Must-Be |
Requirement for TASTE of COOKIES (BAKED) is fulfilled to: 20% | The Severity of being unsatisfied with TASTE of COOKIES (BAKED) is: 92% (Rank 1/ 5).
Y_02: Output: Cookies (delivered) | Voice of Customer (VoC): Demand: Cookie delivery on requested time; Complaint: delivered too early | Critical Customer Requirement (CCR): COOKIES (DELIVERED) DELIVERY IN TIME | Problem: COOKIES (DELIVERED) DELIVERY > 1 HOUR TOO EARLY/ LATE | Requirement-Category of DELIVERY: Availability | Kano-Category of DELIVERY: More/Less-
Is-Better | Requirement for DELIVERY of COOKIES (DELIVERED) is fulfilled to: 40% | The Severity of being unsatisfied with DELIVERY of COOKIES (DELIVERED) is: 60% (Rank 2/ 5).
Y_03: Output: Cookies (boxed) | Voice of Business (VoB): Demand: low energy consumption; Complaint: waste of energy | Critical Business Requirement (CBR): COOKIES (BOXED) ENERGY-CONSUMPTION MINIMAL | Problem: COOKIES (BOXED) ENERGY-CONSUMPTION > 10% WASTE | Requirement-Category of ENERGY-CONSUMPTION: Consumption | Kano-Category of ENERGY-
CONSUMPTION: More/Less-Is-Better | Requirement for ENERGY-CONSUMPTION of COOKIES (BOXED) is fulfilled to: 70% | The Severity of being unsatisfied with ENERGY-CONSUMPTION of COOKIES (BOXED) is: 30% (Rank 3/ 5).
Y_04: Output: Cookies (boxed) | Voice of Business (VoB): Demand: low consumption of ingredients; Complaint: waste of ingredients | Critical Business Requirement (CBR): COOKIES (BOXED) INGREDIENT-CONSUMPTION MINIMAL | Problem: COOKIES (BOXED) INGREDIENT-CONSUMPTION > 10% WASTE | Requirement-Category of INGREDIENT-CONSUMPTION: Consumption | Kano-
Category of INGREDIENT-CONSUMPTION: More/Less-Is-Better | Requirement for INGREDIENT-CONSUMPTION of COOKIES (BOXED) is fulfilled to: 80% | The Severity of being unsatisfied with INGREDIENT-CONSUMPTION of COOKIES (BOXED) is: 20% (Rank 5/ 5).
Y_05: Output: Cookies (baked) | Voice of Business (VoB): Demand: Cookies fit exactly into the tin; Complaint: Cookies are too big | Critical Business Requirement (CBR): COOKIES (BAKED) DIAMETER > 9CM AND < 10 CM | Problem: COOKIES (BAKED) DIAMETER > 10 CM | Requirement-Category of DIAMETER: Quality | Kano-Category of DIAMETER: More/Less-Is-Better | Requirement for
DIAMETER of COOKIES (BAKED) is fulfilled to: 70% | The Severity of being unsatisfied with DIAMETER of COOKIES (BAKED) is: 30% (Rank 3/ 5).
Summary & Details: Voice of Customer (VoC), Voice of Business (VoB), Critical Requirements (CCR/ CBR), Problems, Severity, KANO and CtQ-Rank
The summary of the "Voice to Criticals" lists the important results of this tool in a table.
The details of the "Voice to Criticals" lists all given information for each Y of this tool also in a table.
Both tables can be included in the Project-Story-Book.
Please sort the table: Summary with: Ctrl Shift S
This information about Y is the anchor of your project

DMAIC > Voice to Criticals >> Charts Technische Universität München
Bar-Chart with the Severity of the Problems (Y) …
The Kano-Chart: Y CtQ aligns the Problems to the

modified Kano characteristics of Must-Be, More/
The bar-chart: Y CtQ shows Less is Better and Delighter.
Original characteristics
the Problems and the Severity of their impacts
of the Kano model
on the Customer and on the Business The basis for the calculation is the:
- fulfillment of requirements and the
- inverted Severity of the problems (= Satisfaction)
… and adapted Kano-Chart with the fulfillment of requirements and calculated satisfaction
DMAIC > Project-Charter Technische Universität München
The contract for your project with business-base, VoC/ VoB and solution-ideas
Project-Name
Project-Charter Improve the performance of Cookie production The smart-criteria are helpful in formulating the targets:
Business-Case Process & Output
Product/ Service: Cookie

The Product COOKIE is a tangible final Output for external Customers and is in the Creation Process BAKE
COOKIES within a year 13 - 52 times generated. Important Input of the Process to generate the Product
s: specific -
COOKIE is: BUTTER, SUGAR, FLOUR, CHOCOLATE.
Process: bake cookies it must be clear what exactly the target state is
Voice of Customer (VoC) Problems
The satisfaction of the external customers with the: Y_01 | Cookies (baked) taste crumbly-bland
m: measurable -
- Quality of COOKIE is: 20%. Y_02 | Cookies (delivered) delivery > 1 hour too early/ late
a measure and a target value or value corridor should indicate
- Availability of COOKIE is: 40%. Y_03 | Cookies (boxed) energy-consumption > 10% waste
Voice of Business (VoB) Solution-Ideas

the success criterion
The satisfaction of the process-owners with the Consumption in the Creation Process of the COOKIE is:
more sugar in the mix
40%.
The total costs of the specified 3 problems are estimated by 100€ / year. a: attractive -
They are primarily the result of quality costs due to scrap and additional expenditure.
The solution of the problems is rated as:
an incentive can additionally make the target attractive for the
- major URGENT (80%-Level) / - major IMPORTANT (80%-Level) team and motivate for rapid implementation
Comment Comment
In Scope Out of Scope Management

r: realistic -
in: Vanilla Cookies
Chocolate Cookies
out: Chips Sponsor Mr. B. Aking Supplier Supermarket
targets that are too high lead to fear or frustration, targets
in: out: Bread A_ccountable Chef du Cookie Customer Charlotte
in: out: Controlling Mrs M. Oney Customer Lili that are too low lead to boredom
in: out: ...? ...?
Targets Timeline Experts
Y_01 Improve taste of Cookies (grade of 1,5 in customer rating)
Y_02 Deliver ordered Cookies on time (+/- 1h)
30 October 20xx
30 October 20xx
Black-Belt Mr B. B. Elt
Green-Belt Y. Ou
Master-Black-Belt Reiner
...?
t: terminated -
Y_03 Reduce consumption of energy (- 20%) 30 October 20xx Expert Mrs D. Ough ...? a deadline for the target is important to speed up and control
Y_04 Reduce consumption of ingredients (- 20%) 30 October 20xx Expert Mr E. Quipment ...?
Timeline Define Measure Analyse Improve Control* Control End
the project implementation
Target-Date: 03.08.20xx 31.08.20xx 28.09.20xx 26.10.20xx 23.11.20xx 01.06.20xy 28.06.20xy
Completion-Date:
Evaluation: days remaining : 58 days remaining : 86 days remaining : 114 days remaining : 142 days remaining : 170 days remaining : 360 days remaining : 387
Please select the problems and add scope, team members targets and deadlines
DMAIC > Project-Charter Technische Universität München
Instructions
Project-Charter
The Project-Charter is the contract for the project that is concluded between all parties involved.
The most important existing information is bundled into one field each:
- Business-Case,
- VoC/ VoB,
- Process & Output,
- Solution ideas.
Please supplement:
Project name
This should succinctly sum up the purpose of the project.
Problems
In the list boxes of the problems field, select a maximum of three problems to be processed in the project, preferably the first three from the Voice-to-Critical (Summary).
Scope
To avoid later misunderstandings and to keep the scope of the project appropriate,
it should be clear and acceptable to all parties involved what is withIN the scope and what is OUTside the scope of the project.
Management & Experts

Here the sponsors, process-owner, belts, experts, controlling, important suppliers and customers can be named as team members.
Targets
In a sense, targets are problems that are upside down.
Therefore, the Yn is already displayed here according to the ranking of the problems Yn and a corresponding target is to be formulated for each problem.
The smart criteria are helpful in formulating the targets:

s: specific - it must be clear what exactly the target state is.
m: measurable - a metric and a target value/corridor of values indicate what the success criterion is.
a: attractive - an incentive can also make the destination attractive for the team.
r: realistic - targets that are too high lead to fear or frustration, targets that are too low lead to boredom.
t: terminated - a deadline for the target is important to accelerate and control project implementation.
Timeline
The deadlines of the DMAIC phases and the dates of the actual completion of each phase are entered here.
An indicator then shows the remaining time and the timeout for each phase.

DMAIC > Stakeholder-Communication >> Support and Resistance Technische Universität München
Assume that the resistance to and support for your project is normally distributed
Number of employees/ Probability
resistance neutral support
open enemy underground fighter spectator companion enthusiast

Types of communication to win
Where does the resistance come from? What can you do against resistance?
Not knowing Communicate
and increase support for your
don’t see the need for improvement, no sense recognizable, communicate basics, facts, opportunities and risks on the topic, project:
distorted perception, bad experience with change give continuous insight into decisions and progress • Newsletter,
Not able Qualify
lack of technical and methodical skills, share easy to understand examples, • eMail,
no say in the matter, fear of making mistakes familiarize with the theoretical basis and tools
• Personal talk,
Not willing Motivate
own benefit unclear, preference of routine, arrogance, identify fears, recognize participation, • Team discussion,
fear of additional workload and loss of influence balance incentives and negative consequences
Not permitted Integrate • DMAIC phase-steering and the
single antagonistic forces, group pressure, interview, open workshops for participation, • Final presentation
secret rules of the game, self-imposed barriers clarify roles and tasks, delegate responsibility
Take any expression of resistance seriously - it contains the key to gaining support
DMAIC > Stakeholder-Communication Technische Universität München
Continuous communication about and integration into the project supports its implementation
Power to support the …
Degree of your Influence on this
Rank (Power x Interest x

Interested in …
Stakeholder-Analysis and Communication-Plan How do you want to win the support of this Person?
Influence)
Person
Who in the company is positively/
Type of Type of Type of
Select one of your targets negatively affected by the Pseudonym … target-achievement Frequency Frequency
communication communication communication
achievement of this target? (Name)
Y_01 | Improve taste of Cookies (grade of 1,5 in customer rating) Mrs Sponsor Spo 10 10 newsletter monthly eMail weekly personal talk 5 1
Y_05 | Reduce consumption of ingredients and energy (- 20%) Chef du Cookie CdC 1 9 newsletter monthly eMail weekly personal talk 7 2
Y_03 | Adapt cookies to tin diameter (99% yield) Mr D.E. Fender DEF 10 1 newsletter monthly eMail weekly personal talk 1 4
Y_01 | Improve taste of Cookies (grade of 1,5 in customer rating) Mr A. Pathetic AP 2 2 newsletter monthly eMail weekly personal talk 3 3
...? ...? ...? ...? ...? ...? ...? ...? ...?
Stakeholder-Analysis and Communication-Plan
The Stakeholder-Analyses supports to identify important Stakeholder

in the environment of your project.
The Chart Stakeholder- The Communication-Plan helps to win the support of these persons.
Analysis shows the Interest in
and Power to support the 1. Select one of your targets
Unfortunately, Excel
achievement of targets for 2. Identify one of the Persons, which is positively or negatively affected by the achievement of this target.
sometimes does not
display the names of defined Stakeholders. 3. For anonymization, give this person a pseudonym.
the bubbles, so please 4. Rate the:
supplement them The bubble sizes represent - degree of interest of this Person in your targets
manually. - degree of power of this Person to support your targets
the degree to positively
5. Determine the media of communication with which you want to win this person to support your goals
influence Stakeholders for the
6. Set the dates for the respective communication
targets of the project.
7. Rate the:
- degree of your influence to win this Person to support your targets
The rank order shows the arithmetic product: Power x Interest x Influence
Copy the Chart: Stakeholder in your Project-Story-Book and add the pseudonyms to the bubbles
The communication-plan can help to win the support of important stakeholders

Input-Analysis, Process-Mapping/ -Analysis, C&E-Matrix, Data-Collection-Plan, Hypotheses

Measure: From the identification of influences on the problems to related hypotheses
 Identify the negative influences of the input: Input-Analysis Input-Analysis
- Focus on the necessary inputs (xi) for the output (Y) from the SIPOC Input
- Specify the requirements for the inputs (xi) Requirement

- Classify the inputs (xi) to one of the categories of: Quality, Availability and Consumption
- Specify the deviations of the inputs (xi) from the requirements and estimate their probability of occurrence negative Influences on Problems
 Map the process and identify its negative influences: Process-Mapping & -Analysis Process-Mapping and Analysis
- Divide the process-steps into activities from the first input (xi) to the last output (Y), specified in the SIPOC Process-Steps > Activities
- Specify the inputs (xi) and outputs (Y) of the activities as well as their associated methods (xm) and resources (xr)
Input, Output, Methods & Resources
- Specify the negative influences of methods (xm) and resources (xr) in the activities on the problems of the outputs (Y)
- Estimate the probability of occurrence of the negative influences (xm; xr) negative Influences on Problems
 Identify the relationships between the influences (xi; xm; xr) on the problems (Y): C&E Matrix C&E Matrix
Problems (Y)
- Estimate the impact strength of the negative influences of the inputs (xi) on the problems (Y) Y1 Y2 Y3 Yn
3
- Estimate the impact strength of the negative influences of the activities (xm; xr) on the problems (Y) Input
(Xi)
Xi1
Xi2 1
- Identify the strongest influences xi, xm and xr on the problems (Y) Xp1
Xin
2
4
Process-Steps
- Check the overall determination of the problems (Y) by the influences xi, xm and xr (Xp) Xp2
7
5
Xpn
 Plan the measurement/ collection of data (Data Collection Plan) Data Collection Plan
- Operationalize the influences xi, xm and xr and problems (Y) as measurable variables
Operationalisation Graphical Display
Y1
Problems
Y2
- Determine the scale-level of the variables, specification limits and targets

(Y)
Yn
Xi1
Input
- Plan a Measurement-System-Analysis (MSA) if necessary and determine sample size (if the statistical test is determined) (Xi) Xi2
Xin
- Plan graphical representation of the variables, determine parameters, the process capability indices and control charts
Process- Xp1
Steps Xp2
(Xp) Xpn
 Relate problems (Y) and influences (xi, xm, xr): Statistical Hypothesis Ya= Yb; Y= f(x) Hypothesis
- Statistical hypothesis are automatically formulated on the basis of the given information, with: There is a/ no Difference
Risk: 14%
- type of the hypothesis (difference/ relationship) and in: the degree of: (Y) …
between: Levels of (x)
- the relevance of each hypothesis for the problem (risk)
- a suggestion for an appropriate statistical test to check the hypothesis Test: ANOVA
Summary MEASURE: Data collected and their analysis determined

Outlook ANALYSE: Display variables, test the hypotheses and find the root causes of influences (x) on problems (Y)
DMAIC > Input-Analysis Technische Universität München
Please answer the questions and get an overview …
xI_01 The Input Analysis examines all

Which Input is necessary for the Process BAKE COOKIES? Request (cookie-type & delivery-date) Input Inputs which come from outside into
Please select an answer. the SIPOC.
What do you require from REQUEST (COOKIE-TYPE &
unambiguous, complete and understandable Requirement
DELIVERY-DATE)? Outputs of one Process-Step in the
Please enter your answer. SIPOC going as Input into another
To which category does the Requirement UNAMBIGUOUS,
Quality (Faultlessness/ Fulfilment of Purpose) Requirement-Category
Process-Step are represented in
COMPLETE AND UNDERSTANDABLE belong?
[brackets].
Which deviation of REQUEST (COOKIE-TYPE & DELIVERY-
DATE) from the Requirement is problematic for the ambiguous, incomplete or not understandable negative Influence You do not need to evaluate them
Process?
Please enter your answer.
again, because you did it already in
How often does the negative Influence REQUEST (COOKIE- the Voice to Criticals.
TYPE & DELIVERY-DATE) AMBIGUOUS, INCOMPLETE OR 5% Probability of Occurrence
NOT UNDERSTANDABLE occur?
Please enter a value between: 0% - 100%. This information will be used in the
C&E Matrix as negative Influences
on the Output.
The Chart Influence of xi

displays the Inputs (xi) and
the probability of their
deviation from Requirements.
… of the negative Influences from the Inputs (xi) on the Problems of the Output (Y).
DMAIC > Input-Analysis Technische Universität München
Instructions
Input-Analysis
In this Chapter we will:

- focus on the Inputs from the SIPOC,
- specify the Requirements for the Inputs,
- collect deviations from the Requirements and
- evaluate the frequency, how often the Inputs deviate from the Requirements.
In the SIPOC the Process BAKE COOKIES was already structured in its basic Process-Steps.
The Inputs of these Input-Process-Output-Sequences may come from two different sources:
- Inputs from Suppliers, which enter the Process from outside and
- Inputs which result in the course of the Process, as an Output of a preceding Process-Step.
For this Input-Analysis only the external Inputs are relevant. Internal Inputs are seen as Outputs and analysed in the Voice-to-Criticals.
Please note that:

- internal Inputs which result as an Output from a preceding Process-Step , are [ excluded ] in the List-Boxes.
- external Inputs are presented without brackets in the List-Boxes. They should be analysed here.
Please answer the questions to the overall 16 possible Inputs.

DMAIC > Process-Mapping-Analysis Technische Universität München
Specify the activities in the process-steps, assign inputs, outputs, methods, resources and …
Departments Activities  Process-Mapping-Analysis of: baking cookies
Customer transmit request evaluate cookies
Logistic & determine cookie- purchase missing

deliver cookies
Procurement type & delivery-date ingredients
determine
Production
necessary &
Support missing ingredients
Chef du Cookie preheat the oven …
order(cookie-type;
Inputs request (cookie)
delivery date)
Shopping-List start-signal …
Methods calendar-entry personal habit personal habit cooking-standards …
Resources oven …
Order(cookie-type; Ingredients
Outputs delivery date)
Shopping-List
(complete)
…
influences on …
wrong cookie-type wrong ingredients

Quality determined
…
determined
negative
necessary
Availability ingredients not …
available
clarification with
Con- oven preheated too
the customer is …
sumption early/ hot
laborious
Frequency 0% 10% 5% 5% 5% …
… identify the triggering influences in the methods (xm) and resources (xr) of the problems (Y)
Standard for the workshop: key questions and procedure
Key questions for the workshop Procedure for the workshop
Involved departments 1. Keep calm – if you get excited: speak slower, breathe evenly, deeply
1. Who is involved in the process?
2. Invite participants and inform the sponsor
Activities
3. Ask for a guided tour of the process, get the workshop materials, prepare
2.a What is typically done first? the room
2.b What happens next? (variant) 4. Take care of your participants
Input & Output 5. Take the guided tour of the process
3. Which input is necessary for the activity?
6. Repeat the following microprocess with for each of the key questions (left)
4. Which output results from the activity? for each activity, until you have received a valid answer:
a) ask one of the standard key questions,
Methods & Resources
b) check the answer,
5. Which resources execute or support the execution the activity? c) summarize relevant aspects of the answer and
6. Which methods guide the execution? d) let the participants check your understanding of the answer
Influences & Frequency 7. Note the relevant information of the answers on different stickers each:
a) Activities: describe them as precisely as possible with: verb & noun
7. Which influences from the methods and resources on or by the
b) Input, Output, Methods & Resources: write their names or
activities negatively affected the quality, the availability of the output
“personal habits”, if e.g. methods are missing
or the consumption and waste of input and resources in the past?
c) negative influences: Identify the last direct influence, i.e. the
8. How often (%) did the most relevant and most important of these last trigger in the causal chain that causes or contributes to a defined
influences affect the output in the past? problem, check: If influence occurs, then problem arises/increases?
Stay to these question, repeat them, do not vary them, 8. In between thank your experts for their participation and their answers
because you might get a different answer than intended
9. Take a picture of yourself and the team in front of your workshop flip charts
Please apply this standard and make your own experiences before changing it
Transfer the information of the Process-Mapping-Analysis after the workshop to sigmaGuide

Process-Mapping-Analysis of the Process: bake cookies
Please specify the Process-Steps in detailed Activities the format: Verb + Noun (e.g.: weigh Ingredients)
Who … … does what?
1. Activity 2. Activity 3. Activity 4. Activity 5. Activity
transmit request for

1. Process-Step Customer (Child) cookies
determine cookie-type purchase missing

2. Process-Step Logistic & Procurement Service & delivery-date ingredients
determine necessary &

3. Process-Step Production Support missing ingredients
preheat the oven
4. Process-Step Chef du Cookie
5. Process-Step ...?
Input:
Which Inputs are necessary to start the
Activity?
...? Request (cookie)
Order (cookie-type;
delivery-date)
Ingredients Start-Signal The Chart Influence of xP graphically display the activities and the
Which Instructions/ Rules direct how to
probability of their deviation from requirements.
Methods: calendar-entry personal habit personal habit cooking-standards
perform the Activity?
Which Equipment/ Machines/ Tools operate

Resources: oven
or support the Activity?
Order (cookie-type;
Output: Which Output results from the Activity? ...?
delivery-date)
Shopping-List Ingredients (complete) ...?
… the Quality (Faultlessness/ Fulfilment of wrong cookie-type wrong ingredients

Purpose) of the Output?
The Process-Mapping-Analysis lets you detail all the process-steps of
determined determined
Which Influences
of the:
- Methods and
… the Availability (right Quantity just in
Time) of the Output?
necessary ingredients
not available
the SIPOC into specific activities, their inputs, outputs, the executing
- Resources
negatively affect: resources and the guiding methods
… the Consumption and Waste of Input and/ clarification with the oven preheated too
or Resources? customer is laborious early/ hot
Anzahl der Abteilungen The related negative influences on the output and their probability continue
How often are the Activities affected by these negative
Influences?
0% 10% 5% 5% 5%
the preparation of hypothesis in the C&E Matrix.
The related chart displays all identified influences and their frequency in the past
DMAIC > C&E Matrix Technische Universität München
Specify the impact strength of each negative influence in the …
Severity 92% 60% 30% 20% 30%
Output (Y)
Kano-Category Must-Be More/Less-Is-Better More/Less-Is-Better More/Less-Is-Better More/Less-Is-Better
Results for: Impact of Influences (xI & xP) on the Outputs (Y)
C&E Matrix Y_01 | Problem: COOKIES (BAKED)
TASTE CRUMBLY-BLAND
Y_02 | Problem: COOKIES
(DELIVERED) DELIVERY > 1 HOUR Problems
Y_03 | Problem: COOKIES (BOXED)
ENERGY-CONSUMPTION > 10%
Y_04 | Problem: COOKIES (BOXED)
INGREDIENT-CONSUMPTION > 10%
Y_05 | Problem: COOKIES (BAKED)
DIAMETER > 10 CM
Problems (= TOO EARLY/ LATE WASTE WASTE
Effects)
(Yi) Product Sum of the Impact
of each Influence (xI & xP)
Percentual Impact of each
Influence (xI & xP) on all
Ranking of the Impact of
each Influence (xI & xP) on
on all Outputs (Y) Outputs (Y) all Outputs (Y)
Influences from Input (xI) (= Causes) Probability Rank
xI_01: Input: Request (cookie-type & delivery-date) | Requirement: unambiguous, complete and understandable | Requirement-
Category: Quality (Faultlessness/ Fulfilment of Purpose) | negative Influence: ambiguous, incomplete or not understandable
5% 2 40% 0,02 1% 10
xI_02: Input: Start-Signal | Requirement: given at the right time | Requirement-Category: Availability (right Quantity just in Time)
| negative Influence: given too late or too early
10% 1 80% 0,05 3% 8
xI_03: Input: Tin | Requirement: diameter target: 110, LSL: 107 USL: 113 mm | Requirement-Category: Quality (Faultlessness/
Fulfilment of Purpose) | negative Influence: diameter < LSL or > USL
1% 3 5% 100% 0,01 0% 14
xI_04: Input: Ingredients | Requirement: expiry date met | Requirement-Category: Quality (Faultlessness/ Fulfilment of
Purpose) | negative Influence: expiry date exceeded
1% 3 80% 0,01 0% 13
Influences from Process-Step (xMR) (= Causes) Probability Rank
xMR_01: Activity: transmit request for cookies | Input: ./. | Methods: ./. | Resources: ./. | Output: ./. | Influence on Quality: ./. |
Influence on Availability: ./. | Influence on Consumption: ./.
0% 90% 0,11 7% 4
xMR_02: Activity: determine cookie-type & delivery-date | Input: Request (cookie) | Methods: calendar-entry | Resources: ./. |
Output: Order (cookie-type; delivery-date) | Influence on Quality: wrong cookie-type determined | Influence on Availability: ./. |
Influence on Consumption: clarification with the customer is laborious Influences
10% 6 40%
Y= f(x) 0,02 1% 10
xMR_03: Activity: determine necessary & missing ingredients | Input: Order (cookie-type; delivery-date) | Methods: personal
habit | Resources: ./. | Output: Shopping-List | Influence on Quality: wrong ingredients determined | Influence on Availability: ./.
| Influence on Consumption: ./.
(xi, xm, xr)
5% 7 90% Basis for Hypotheses 0,08 5% 5
xMR_04: Activity: purchase missing ingredients | Input: Ingredients | Methods: personal habit | Resources: ./. | Output:
Ingredients (complete) | Influence on Quality: ./. | Influence on Availability: necessary ingredients not available | Influence on 5% 7 90% 0,03 2% 9
Consumption: ./.
xMR_05: Activity: preheat the oven | Input: Start-Signal | Methods: cooking-standards | Resources: oven | Output: ./. |
Influence on Quality: ./. | Influence on Availability: ./. | Influence on Consumption: oven preheated too early/ hot
5% 7 100% 0,02 1% 12
Product Sum of the Determination of each Output (Y) by the Influences (xI & xP) 4,8823 0,1830 0,0150 0,1401 0,0050 2 100% 5
Results for: Determination of Outputs (Y) by
Percentual Determination of each Output (Y) by the Influences (xI & xP) 93% 4% 0% 3% 0% 100%
Influences (x)
Ranking of the Determination of each Output (Y) by the Influences (xI & xP) 1 2 4 3 5
Guiding question to fill the cells is:

The C&E Matrix:
- Lets you estimate the impact strength of each influence on the problems.
How strong (0-100%) is the impact of the negative influence (x_n) on the problem (Y_n)?
- Every entry in one of the cells at the intersection of input/ activity of process-steps and output
with:
thus is an assumption about a causal relationships between an influence (X) and a problem (Y).
- 0% meaning, that the problem (Y_m) is independent from the influence (x_n)
- This assumptions about causal relationships are the basis for the formulation of hypothesis.
- 100% meaning, that the influence (x_n) always triggers or increases the problem (Y_m)
… inputs (xi) and activities (xm; xr) on the problems of the outputs (Y).
DMAIC > C&E Matrix Technische Universität München
Instructions
C&E Matrix
The C&E Matrix serves to indicate causal relationships between the negative Influences of the Inputs (Xi) and Activities (Xp) (= Causes)
on the Problems of the Output (Y) (= Effects).
1. Within the C&E Matrix every Influence (x) in the rows is contrasted with every Problem (Y) in the columns.
In the xY-intersection cell the strength of each impact can be estimated within a range from 0%-100%.
2. The sum of the impacts of an Influence (x), relativized to its probability of occurrence and the severity of the influenced Problems (Y)
indicates the overall impact strength of each Influence (column U).
3. The rank of each Influence shows its importance in relation to the other Influences and
thus its importance for the subsequent Hypothesis (see below) (column W).
4. The sum of the impacts of all Influences (x) on a Problem (Y), relativized to their probability of occurrence,
indicates the determination of the Problem by its Influences (row 42).
5. The rank of each Problem shows which Problem is relatively strong (high rank) and which Problem is only weakly "explained" by the specified Influences.
This ranking can thus show, where in the Inputs or the Process-Steps additional Influences should be looked for (row 44).
Please evaluate the strength of the presumed impact of the Influences (xI and xP) on the Outputs (Y).
- To do so enter a value between 1 .. 100% in the intersection of the focused cells.
- It does not matter, whether you assume a positive relationship (the greater x, the greater Y) or a negative relationship (the smaller x, the greater Y).
In this tool its a question of the absolute strength of the relationship.
- If there is no relationship at all between an x and an Y, the leave the cell empty.
Result per row: Overall strength of every Influence (x), added up over all Outputs (Y)
Result per column: Determination of every Output (Y) by all Influences (x)

DMAIC > C&E Matrix >> C&E Heatmap Technische Universität München
The Chart C&E Heatmap gives an overview about number and dispersion …
Chart: C&E Heatmap Severity 92% 60%

The Chart C&E Heatmap shows the risk of each
Y_01 | Problem: COOKIES (BAKED) Y_02 | Problem: COOKIES
Problems (=
influence x problem relationship.
Output (Y)
The cells indicate the strength of each relationship between influences (xI and xP) and the related TASTE CRUMBLY-BLAND (DELIVERED) DELIVERY > 1 HOUR
Effects)
Outputs (Y) as Risks (Probability x Severity). TOO EARLY/ LATE
The Risks are the basis for prioritizing of the corresponding Hypothesis between x and Y.
(Nothing needs to be entered here)
The percentual risk-value combines the:
Influences from Input (xI) (= Causes) Probability D E
- probability of occurrence of an influence,
xI_01: Input: Request (cookie-type & delivery-date) | Requirement: unambiguous, complete and understandable | Requirement-
Category: Quality (Faultlessness/ Fulfilment of Purpose) | negative Influence: ambiguous, incomplete or not understandable
5% 10,63% - strength of its impact on the problem and
- severity of the problem.
xI_02: Input: Start-Signal | Requirement: given at the right time | Requirement-Category: Availability (right Quantity just in Time)
10% 8,55%
| negative Influence: given too late or too early
xI_03: Input: Tin | Requirement: diameter target: 110, LSL: 107 USL: 113 mm | Requirement-Category: Quality (Faultlessness/
1%
The colours represent the relative risk level for the influence-
Fulfilment of Purpose) | negative Influence: diameter < LSL or > USL
problem pairs:
xI_04: Input: Ingredients | Requirement: expiry date met | Requirement-Category: Quality (Faultlessness/ Fulfilment of
Purpose) | negative Influence: expiry date exceeded
1% 11,76% high risk (> 10%)
medium risk (1% < risk < 10%)
Influences from Process-Step (xMR) (= Causes) Probability
low risk (< 1%)
xMR_01: Activity: transmit request for cookies | Input: ./. | Methods: ./. | Resources: ./. | Output: ./. | Influence on Quality: ./. |
Influence on Availability: ./. | Influence on Consumption: ./.
0% for triggering the problem.
xMR_02: Activity: determine cookie-type & delivery-date | Input: Request (cookie) | Methods: calendar-entry | Resources: ./. |
Output: Order (cookie-type; delivery-date) | Influence on Quality: wrong cookie-type determined | Influence on Availability: 13,72%
delay in case of unavailability of the customer | Influence on Consumption: clarification with the customer is laborious
10%
Thus the risks of the xY-relations give an overview about the
xMR_03: Activity: determine necessary & missing ingredients | Input: Order (cookie-type; delivery-date) | Methods: personal
habit | Resources: ./. | Output: Shopping-List | Influence on Quality: wrong ingredients determined | Influence on Availability: ./. 5% 23,92%
number and dispersion of the hypothesis (see hypothesis).
| Influence on Consumption: ./.
xMR_04: Activity: purchase missing ingredients | Input: Ingredients | Methods: personal habit | Resources: ./. | Output:
Ingredients (complete) | Influence on Quality: ./. | Influence on Availability: necessary ingredients not available | Influence on 5% 7,46%
Consumption: ./.
xMR_05: Activity: preheat the oven | Input: Start-Signal | Methods: cooking-standards | Resources: oven | Output: ./. |
5%
Influence on Quality: ./. | Influence on Availability: ./. | Influence on Consumption: oven preheated too early/ hot
… of risks resulting from the causal relationships between influences and problems
DMAIC > Data-Collection-Plan >> Scales of Measurement Technische Universität München
Measuring means assigning a value of an underlying scale …
Cardinal-Scale: Punctuality:
Relationship  Attribute of object: quantitative Values: - ∞ – + ∞ (continuous)
 Values of scale: discrete or continuous -5 -4 -3 -2 -1 0 1 2 3 4 5
Difference
 Value order: value= sum of intervals
Reality
 Intervals: unit defined, like values
Number of daily eMails:
Values: 0 – ∞ (discrete)
0 1 2 3 4 5 6 7 8 9 10
Ordinal-Scale:
Model Project Statistics  Attribute of object: quantitative Satisfaction with Product:
Y= f(x)
Scatterplot Y vs. x
 Values of scale: discrete Values:  – 
Cause Problem
 Value order: ranking  <  <  <  < 
Operationalization  Intervals: undefined Personal state of mind:

Solution Effect
Values: bad/ indifferent/ good
Process-Problem-Model Scales & Measurement
Attributes: Cardinal Scales:
Nominal-Scale:

amount of information
x: Baking Time x: 10 vs. 25 min

Y: Cookie Diameter Y: 100 vs. 75 mm  Attribute of object: qualitative


Values of scale: attributive categories
Value order: undefined Means of transport:
 
 Intervals: undefined Values:
Operationalization is the mapping of an on foot/ bicycle/ car/ bus/ plane  
object’s attribute to a scale,
so that the measured data represent the
Gender:
attribute validly, reliably and precisely. Values: male / female  
… to the state of an object's attribute
DMAIC > Data-Collection-Plan >> Scales of Measurement Technische Universität München
Attributes of objects mapped on scales with different attributes and parameters
weight of cookies:
Values: continuous length: cm
The Scale of a measurand determines the:
- gram (g)
race duration: - Parameter of central tendency
Value order: value= sum of intervals cycle time: hours processing time of
hours : minutes : …
Intervals: unit defined, continuous costs: $/ €/ £/ ¥/ … cookies: - Parameter of scattering
Average Deviation,
Cardinal Scale
Mean
Standard Deviation,
- minutes (min) - Appropriate statistical test
(Xbar)
Variance
Values: discrete number of children in Number of phone number of defect Values of a higher scale-level can always be
Value order: value= sum of intervals a family: calls: cookies:
transformed to a lower scale level, e.g. if you
Intervals: unit defined, discrete 0…n 0…n 0…n
transform the real age of people (Cardinal
quantitative
Scale) into age categories (Nominal Scale).
satisfaction
FMEA Risk-Priority-
assessment: Risk of a Problem:
****
Number (RPN):
0% - 100%
This reduction of scale-level reduces the
1 - 1000 already available information, because the
Values: discrete Range, smileys
Value order: ranking Ordinal Scale Median Percentiles, parameters of a Nominal-Scale (e.g. Mode)
Intervals: undefined Interquartile Range
grades: complexity of carry less information then the parameters of a
rating of taste: Cardinal-Scale (e.g. Mean).
ABCDEF/ projects:
****
123456 project X > project Y
The same is true for the related statistical
election: types of different: types of errors: tests, where e.g. the Chi2-Test has less power
means of - products (Y) - burnt than the ANOVA.
number of different transportation/ - errors (Y) - salty
- locations (x)
Values: attributive categories values in sample/ parties/ persons/ …
- shifts (x)
-… Thus always plan your measurement on the
qualitative Value order: undefined Nominal Scale Mode number of different highest possible level and avoid transfor-
physical condition:
Intervals: undefined possible values in
population
healthy/ sick avaliability of a
Recipe available and
mations to a lower scale as far as possible.
method (x) for certain
used? (yes/ no)
gender: activity (yes/ no) Data on a lower scale-level cannot be
male/ female transformed to a higher scale level
Type of attribute of
Parameters of central Parameters of Example of everyday Example from work
the object of Attributes of the Scale Name of Scale Example for cookies
tendency scattering life environment
measurement
The higher the scale level of a measurement, the more information is carried by the data
DMAIC > Data-Collection-Plan Technische Universität München
Operationalize the measurement of the variables x and Y, …
The Data Collection Plan lists all Outputs (Y)

and their Influences (xi and Xp) with the
ranking of their relevance.
The task is to operationalize the related

variables and prepare them for the
measurement.
This means to transfer the subject of

measurement into measurands and specify
the variables, i.e. the:
- subject of measurement
- measurand
- units and its
Which Data about the
- scale level
How should the Data be
collected?
Is a Measurement-System-
Analysis (MSA) necessary?
circumstances should
additionally be collected?
How large should the Sample
Size be?
Where should the Data be
collected? (Location/
For which Time Interval
should the Data be
Which Variable-Name will
you assign to the
In which File will the Data be
stored?
Who is responsible for the
collection of the data?
- sample size
(Blocking-/ Condition-
Variables)
Source) collected? (Start/ End) Measurand?
- type of Data Collection and the
- conditions of measurement
- responsible persons
collect new data no

cookie-type (Vanilla vs.
Chocolate)
50 shop 1 month Y_01_Taste Cookie_du_Chef_01.xls Green Belt As a result some recommendations are given
Parameter of Central Dispersion
to their graphical display and first analysis of
Graphical Representation Process-Capability Control-Charts Test of one Sample vs. Limit/ Target
Tendency Parameter
the variables.
Pp/ Ppk; Cp/ Cpk / Z-Value (Sigma-
X-bar/ R-Chart (if n_total > 80 and

Error-Bar-Chart/ Interval-Diagram
X-bar/ S-Chart (if n_total > 80 and
Chi-Square goodness-of-fit-Test
Time-Series-Chart/ Run-Chart/
Error Rate/ DPU/ ppm/ z-Value
1P-Test (Percentage vs. Limit/
t-Test (Mean vs. Limit/ Target)

(Percentage vs. Limit/ Target)
Standard Deviation/ Variance
DPO/ DPMO/ z-Value (Sigma-
Variance-Test (Variance vs.

Bar-Chart/ Pareto-Chart
The colours represent the suitability of charts,

if n_subsamples <= 8)
p-Chart (if ok vs. ko is
p-Chart (if ok vs. ko is

I/ MR-Chart (if N < 80)
if n_subsamples > 8)
discriminated)
discriminated)
parameters, indices and tests:

(Sigma-Level)
Limit/ Target)
Line-Chart
Histogram
Box-Plot
Quartile
Median
Target)
Modus
Level)
Level)
suitable
Mean
interpretation under reserve

200,00% 200,00% 100,00% 100,00% 100,00% 200,00% 100,00% 200,00% 200,00% 200,00% 200,00% 200,00% 200,00% 200,00%
… for single case raw data on the highest available scale-level

DMAIC > Data Collection Plan Technische Universität München
Instructions
Data Collection Plan
In the Data Collection Plan the Outputs (Y), Influences of the Inputs (Xi) and the Processes (Xp) are operationalised.
In this Chapter you can determine the:

- subject for the measurement (What should be measured?)
- measurand and the unit of measurement (e.g. Time (hrs.))
- target and specification limits
- scale level of the data (Nominal-, Ordinal- and Cardinal-Scale)
- Type of data collection (collect new or existing data)
- decision whether a Measurement-System-Analysis should be executed or not
- conditions and circumstances, which should be additionally collected? (Blocking-/ Condition-Variables)
- sample size
(can be calculated with the procedure: Power & Sample Size only after the determination of the statistical test (see Hypothesis))
- location and source of data collection
- time interval in which the data should be collected (eventually with a start- and end-date of measurement)
- name of the variables in the data collection table
- name of the data collection table
- responsible persons for the data collection
On the basis of the scale level recommendations are given for:

- Graphical Representation
- Parameter of Central Tendency
- Dispersion Parameter
- Process-Capability
- Control-Charts
- Test of one Sample vs. Limit/ Target
With these information all requirements are given to form statistical Hypothesis.
In the next chapter you will find a list of automatically formulated Hypothesis with recommended Statistical Tests for their examination.

DMAIC > Hypotheses >> Translation of assumptions into hypotheses Technische Universität München
Observations of the reality are transformed into logical assumptions …
Relationship-Hypothesis: Y
…
Relationship  Assumption: If x, then Y
 Attributes of
Difference
Reality x: 2 … n levels (categories on nominal scale)
Y: 2 … n levels (categories on nominal scale)
 Hypothesis: x
 9 1
There is a relationship between gender (x) and the type of preferred snack (Y)
 Test: e.g. Chi2-Test  2 8
Relationship-Hypothesis:
Model Project Statistics  Assumption: The x, the Y
Y= f(x) Scatterplot Y vs. x  Attributes of
x: discrete or continuous data (ordinal or cardinal scale)
Cause Problem
Y: discrete or continuous data (ordinal or cardinal scale)
Hypotheses for x and Y
Solution Effect
 Hypothesis:
Hypothesis: Hypothesis: There is a relationship between the baking time (x) and cookie diameter (Y)
Relationship between H0: rxY = 0  Test: e.g. Correlation, Regression
baking time (x) and HA: rxY ≠ 0
cookie diameter (Y)
Difference-Hypothesis:
 Assumption: Difference in Y, dependent from type/ level of (x)
 Attributes of
x: 2 … n levels (categories on nominal scale)
A hypothesis is the translation of an Y: discrete or continuous data (ordinal or cardinal scale)
assumption about objects in reality into an  Hypothesis:
empirically testable format with the There is a difference in the weight of cookies (Y) between cookie types (x)
measurands x and Y.  Test: e.g. t-Test, ANOVA
… and then translated into hypotheses about relationships or differences

DMAIC > Hypothesis >> Translation of relationship/ difference hypotheses into statistically testable hypotheses Technische Universität München
The logical content of each hypothesis on relationships or differences …
... and translated into the format of statistically testable hypotheses

In reality, we observe relationships and
differences that lead us to assumptions of
Null-Hypothesis (H0) Alternative-Hypothesis (HA)
causal connections between x and Y.
From the assumption about reality…
These assumptions about reality are
Relationship-Hypothesis
represented in the model as hypothesis about:

There is no relationship between x and Y There is a relationship between x and Y
- Relationships between x and Y or
Example - Differences in Y between levels of x.

based on the scale level of x and Y.
There is a/no relationship between baking time (x) and cookie size (Y) These relationship and difference hypotheses
can be statistically tested when translated into
... to the statistically testable formulation the format of statistical hypotheses. In the:
- Null-Hypothesis (H0) no relationship or
Assumption H0: rxy = 0 HA: rxy ≠ 0 difference is formulated
about reality From the assumption about reality… - Alternative-Hypothesis (HA) a
modeled as ... relationship or difference is formulated.
There is no difference There is a difference
Difference-Hypothesis
- in the degree of: Y - in the degree of: Y Therefore, the Data-Collection-Plan already
- between the levels of: x (xi, xj, ...) - between the levels of: x (xi, xj, ...) defines a) the type of hypothesis as
relationship or difference and b) determines
Example
the appropriate statistical tests.
There is a/ no difference
The statistical tests now checks the Null-
- in: the taste of cookies (Y) Hypothesis and indicates whether the H0 is
- between: types of cookies (xi) (e.g. Vanilla (x1) vs. Chocolate (x2) vs. ...) confirmed or should be rejected in favour of
... to the statistically testable formulation HA, based on the calculated p-value of the test
and the alpha-level, that we have determined
H0: Yxi = Yxj HA: Yxi ≠ Yxj in advance.
… is formally divided into the hypotheses H0 vs. HA for their statistical analysis
DMAIC > Hypothesis >> Scale levels, appropriate hypotheses and statistical tests Technische Universität München
Selection of the appropriate statistical test for the examination of hypotheses …
Y The scale level of x and Y

Data in 2 Levels Data in > 2 Levels Data Rank Ordered Data discrete or continuous
determines the content of the
(Nominal-Scale) (Nominal-Scale) (Ordinal-Scale) (Cardinal-Scale) hypothesis and the selection of
Relationship Hypothesis Relationship Hypothesis Difference Hypothesis Difference Hypothesis
suitable statistical tests.
Data in 2 Levels
(Nominal-Scale)
Chi-Square-Test Chi-Square-Test Wilcoxon-Mann-Whitney-Test t-Test
Based on the determined scale
Relationship Hypothesis Relationship Hypothesis Difference Hypothesis Difference Hypothesis level in the Data-Collection-Plan,
Data in > 2 Levels sigmaGuide automatically
(Nominal-Scale)
Chi-Square-Test Chi-Square-Test Kruskal-Wallis-Test ANOVA generates hypotheses for xY
x Relationship Hypothesis Relationship Hypothesis Relationship Hypothesis Relationship Hypothesis
pairs and suggests suitable
Data Rank Ordered statistical tests.
(Ordinal-Scale) Rank Correlation (Spearman) /
Binary-Logistic-Regression Nominal-Logistic-Regression Rank Correlation (Spearman)
Ordinal-Logistic-Regression
Relationship Hypothesis Relationship Hypothesis Relationship Hypothesis Relationship Hypothesis

Data discrete or
continuous Product-Moment-Correlation
Rank Correlation (Spearman) /
(Cardinal-Scale) Binary-Logistic-Regression Nominal-Logistic-Regression (Pearson) / General
Ordinal-Logistic-Regression
Regression
… depending on the determined scale level of the involved variables x and Y

DMAIC > Hypothesis Technische Universität München
Automatically generated (univariate) hypothesis for Y and xi, xm and xr …
1. Risk Y_01: Output: Cookies (baked) [ Ranking Position of: rating-scale (grades 1 .. 6) ]
There is a/ no Difference in the degree of: Y_01: Output: Cookies (baked) [ Ranking Position of: rating-scale (grades 1 .. 6) ] between the Levels of: xI_01: Input: Request (cookie) [ Levels of:
10,63% Cookie-Type (Vanilla, Chocolate) ] .
Difference Hypothesis Wilcoxon-Mann-Whitney-Test/ t-Test
There is a/ no Relationship between: xMR_07: Activity: weigh ingredients [ Degree of: Chocolate weight (grams/ g) ] and: Y_01: Output: Cookies (baked) [ Ranking Position of: rating-scale
46,24% (grades 1 .. 6) ] according to the Principle: The larger the value of x, the larger (resp. smaller) is the value of Y.
Relationship Hypothesis Rank Correlation (Spearman)/ Ordinal-Logistic-Regression/ General Regression
There is a/ no Difference in the degree of: Y_01: Output: Cookies (baked) [ Ranking Position of: rating-scale (grades 1 .. 6) ] between the Levels of: xMR_08: Activity: knead ingredients [ Levels
69,26% of: Chef ate chocolate from weighed portion (yes, no) ] .
Hypothesis
In this Chapter we will:

- summarize the most important information from the DEFINE- and MEASURE-Phase and formulate Hypothesis.
These Hypothesis are based on:

- the xY-pairings of the C&E Matrix,
- the information from the Data Collection Plan and
- the prioritization by the related Risks from the C&E Heatmap.
To sort and actualize the Hypothesis in a consecutive order please press: Ctrl + s
… their risk to cause a problem and recommended statistical tests

Data Evaluation, Process Performance, Test of Hypotheses, Root Cause Analysis

ANALYSE: From the visualization of problems (Y) to the identification of their root causes (x’)
 Evaluate collected data Histogram Pareto-Diagram

Relationship
- Check the plausibility of the collected data and eliminate invalid data
Summary Report for Yt1 _Cookie-Weight
Anderson-Darling Normality Test
Pareto Chart of Y_Type_of_Cookie-Problem
A-Squared 0,22 50
P-Value 0,832
1 00
Mean 20,951
StDev 2,333
Variance 5,444 40
Y_Problem -Frequen cy
Skewness -0,1 631 07
Kurtosis -0,1 48408
80
N 60
- Display the data graphically, e.g. with a histogram, pareto-diagram (see the recommendations in the Data-Collection-Plan)
Minimum 1 5,048 30
Percen t
1 st Quartile 1 9,398 60
Median 20,899
3rd Quartile 22,591
Maximum 26,1 72
Difference
95% Confidence Interval for Mean
20 40
20,348 21 ,553
95% Confidence Interval for Median
15 18 21 24 20,260 21 ,71 6 10 20
95% Confidence Interval for StDev
1 ,978 2,846
0 0
s r ht
Reality
Y_Type_of_Cookie-Problem te ce y er
as an orm ie
nt olo eig
nc th
Q
_T st Q
_F ed Q
_C
_W iste O
Y_ ub Y_ ngr Y_ Q ons
95% Confidence Intervals _S _I Y_ _C
Y _Q Y _Q
Mean Y_Q
Y_Problem-Frequency 16 8 5 5 4 4 3 2
Median
Percent 34,0 17,0 10,6 10,6 8,5 8,5 6,4 4,3
20,4 20,7 21 ,0 21 ,3 21,6 21 ,9
Cum % 34,0 51 ,1 61 ,7 72,3 80,9 89,4 95,7 100,0
 Calculate the process capability and monitor process performance over time Process Capability Control Chart
Model Project Statistics
- Calculate the process capability, e.g. DPMO, Pp/ Ppk, Sigma Level (see the recommendations in the Data-Collection-Plan)
Process Capability Report for Yt1 _Cookie-Weight I-MR Chart of Y_Consumption_per_Day
Summary Report
Is the process mean stable? Comments
Evaluate the % of out-of-control points.
LSL Target USL The process mean may not be stable. 5 (15,6%) data points are out of control
0% > 5% on the I chart. Keep in mind that you may see 0,7% out-of-control points by
Process Data Overall chance, even when the process is stable.
LSL 18 Within
Y= f(x)
Target 20 Yes No
USL 22 Overall Capability 15,6%
Sample Mean 20,9507 Z.Bench 0,1 8
Sample N 60 Z.LSL 1 ,26
StDev(Overall) 2,3331 5 Z.USL 0,45
Scatterplot Y vs. x
- Display the process performance over time, e.g. I/ MR Chart (see the recommendations in the Data-Collection-Plan)
StDev(Within) 2,04775 Ppk 0,1 5
Cpm 0,26

Potential (Within) Capability Individual and Moving Range Charts
Investigate any out-of-control points.
Z.Bench 0,31
25
Z.LSL 1 ,44 UCL=23,92
Z.USL 0,51
Individual Value
Cpk 0,1 7
10 _
X=5,89
-5
LCL=-12,13
UCL=22,14
20
Moving Range
1 5,0 1 6,5 1 8,0 1 9,5 21 ,0 22,5 24,0 25,5
Performance
10
__
MR=6,78
Cause Problem
Observed Expected Overall Expected Within 0 LCL=0
Hypothesis for x and Y

PPM < LSL 1 00000,00 1 02992,26 74799,56 1 4 7 10 13 16 19 22 25 28 31
PPM > USL 266666,67 326450,62 3041 79,60
PPM Total 366666,67 429442,89 378979,1 5 N: 32 Mean: 5,8947 StDev(within): 6,0087 StDev(overall): 4,0946
Control limits are estimated using the StDev(within).
Solution Effect
Hypothesis: Hypothesis:
Relationship between H0: rxY = 0
baking time (x) and HA: rxY ≠ 0
cookie diameter (Y)
 Test hypotheses Difference Relationship
- Select the hypotheses with the highest risks of x-Y pairs (see the risks in the hypotheses) 27,5
Boxplot of Y_Cookie_Weight
22
20
Scatterplot of Y_Brightness_of_Cookie vs x_Baking_Time
- Test the selected hypotheses (see the recommended statistical tests in the hypotheses)
25,0
Y_Brightness_of_Cookie
18
Y_Cookie_Weight
22,5
16
By testing the hypotheses we can

20,0
14
- Prioritize the problems (Y) by difference hypotheses

1 7,5 12
10
1 5,0
t_1 t_2 10 12 14 16 18 20
time_of_measurement x_Baking_Time
- Identify causes of the triggering influences (x > x' >...) with relationship hypotheses, if data are available confirm or reject our modelled
assumptions about causal
relationships in reality.
 Identify root causes Root-Cause-Analysis However, the statistics do not
- If the root causes could already be identified by statistical tests based on data, then go to IMPROVE provide any proof of causality,
- Identify causal chains from the influences (x) of problems to their root causes (x') using root cause analysis but merely support the usefulness
of our assumptions in the model.
Summary ANALYSE: root causes (x’) of problems (Y) identified

Outlook IMPROVE: develop solutions to eliminate, adjust or circumvent the root causes (x’)
DMAIC > Visual inspection and graphical evaluation of collected data Technische Universität München
Input of collected data, correction of invalid and useless data and …

Excel Minitab
1. 1. B 1.
A C
4. Summary Report for Y_Cookie-Taste
Y_07_cookieDiameter x_15_cookieType x_16_Date Pareto Chart of x_15_cookieType
10,5 Vanilla 28.08.2018 2. Anderson-Darling Normality Test
A-Squared 8,74
P-Value <0,005
10,2 Vanilla 28.08.2018 20 100
28.08.2018
1. Mean
StDev
3,2200
1 ,4976
9,9 Vanilla
2. Variance
Skewness
2,2428
0,4251 0
8,7 Chocolate 29.08.2018 80 Kurtosis -1 ,00065
N 200
13,8 Chocalote 29.08.2018 15 Minimum 1 ,0000
1 st Quartile 2,0000
11,1 Chocalote 29.08.2018 3. Median
3rd Quartile
3,0000
4,0000
11,0 Chocolate 29.08.2018 60
Percent
2. Maximum 6,0000
Count
95% Confidence Interval for Mean
10,1 Vanilla 30.08.2018 10 3,01 1 2 3,4288
95% Confidence Interval for Median
9,8 Vanilla 30.08.2018 40 1 2 3 4 5 6 2,0000 3,0000
5. 10,O Vanilla 30.08.2018 4. 95% Confidence Interval for StDev
30.08.2018
5. 1 ,3638 1 ,6607
10,3 Vanilla 5
9,9 Vanilla 30.08.2018 20
10,1 Vanilla 30.08.2018
2. 1. 95% Confidence Intervals
13,9 Chocolate 31.08.2018 0 0 5.
8,8 Chocolate 31.08.2018
x_15_cookieType Chocolate Vanilla Chocalote 2. Mean
Count 10 9 1
9,1 Chocolate 31.08.2018 Percent 50,0 3. 45,0 5,0 Median
8,7 Chocolate 31.08.2018 Cum % 50,0 95,0 100,0 2,00 2,25 2,50 2,75 3,00 3,25 3,50
8,6 Chocolate 31.08.2018

8,9 Chocolate 31.08.2018
1. Distribution of the: Taste ratings in a Histogram with a
14,1 Chocolate 31.08.2018
3. corresponding expected normal distribution curve
Chocolate 31.08.2018
(1= very good; 6= very bad)
1. Name problems and influences according to the Data-Collection-Plan, 2. Anderson-Darling Normality Test shows a significant
1. Count, percentage proportions and cumulative percentage
with a meaningful name, with a Y in front of the problems and an x in result; (p< 0,005), meaning: data are not normally
proportions of cookie types in the sample
front of the influences, supplemented by index numbers. distributed. Additionally important statistics are shown:
2. Wrong spelling - Chocalote - and therefore input errors.
2. Delete additional rows you may have inserted to structure the table. Mean, Standard Deviation, Median, …
Correct wrong entries in the raw data before you proceed.
3. Make sure that related data columns have the same length. If 3. Data are in the expected range, from 1 to 6; data outside
3. The frequency of chocolate and vanilla cookies in the
necessary, delete unneeded entries at the end of a column. this range would be invalid
sample is approximately the same
4. Format the data in the Excel table if, e.g., Minitab uses the T to 4. Box-Plot of the Distribution (Histogram seen “from above”)
display an attributive variable, although they are numeric data. 5. Confidence Intervals for the Mean and Median. It means,
5. Correct obvious input errors, like the wrong O instead of the correct that with a probability of 95% the Mean/ Median in the
zero, which are indexed in Minitab with an asterisk as missing data. population will be within this range.
… plausibility check of the data in the pareto-chart and histogram

Process control refers to the actual performance in relation to "historically" derived control limits
Prozess Control: Process Capability:

 Focuses on systematic anomalies of  Focus on the location (e.g. mean) and
individual outputs over time dispersion (e.g. standard deviation)
of outputs over time
 Relates the sequence of outputs to
control limits (e.g. xbar +/- 3s) and  Relates the location and dispersion of the
signals in the data outputs to specification limits and/or a target
value (:= target)
 Control limits are calculated from the
USL LSL variation of the outputs and thus reflect  Specification limits are based on requirements
the outputs of the past from "outside", e.g. from customer
UCL output LCL
 Depending on the extent of variation  Depending on the requirement and the value
and the range of values, the following range, the following are relevant:
are relevant: - Upper Specification Limit (USL)
- Upper Control Limit (UCL) - Lower Specification Limit (LSL)
- Lower Control Limit (LCL). - Target value
 Systematic influences are identified via  The level of process capability indicates:
the control limits and further tests on - how large the distance of the outputs to the
signals specification limit(s) is and
(patterns and trends) - how well the target is met.
Process capability refers to the actual performance in relation to externally specified limits
DMAIC > Process Performance > Process Control Technische Universität München
Distribution and time series of the taste ratings for Chocolate and Vanilla Cookies
Graphical Summary of Y_Cookie-T by x_Cookie-T Before/After I-MR Chart of Y_Cookie-Taste by x_Cookie-Type

Summary Report Summary Report
Distribution of Data Data in Time Order 1. Was the process standard deviation reduced? Comments
1. Compare center, shape, and variability. 2. Look for patterns and trends. Investigate any outliers (marked in red). 0 0,05 0,1 > 0,5 After a process change, you may want to test whether the standard deviation or
Chocolate Chocolate mean changed:
Yes No
6
• The standard deviation was not reduced significantly (p > 0,05).
P = 1,000
• The mean is significantly higher (p < 0,05). Make sure the direction of the shift
is an improvement.
Consider whether the change in the mean has practical implications.
4
2. Did the process mean change?
0 0,05 0,1 > 0,5
Yes No
P < 0,001
2
Y_Cookie-T
Vanilla
6
Vanilla 3. Vanilla Chocolate
UCL=5,565
5,0
Individual Value
_
X=3,98
4 2,5 LCL=2,395
0,0
2 4. 4
Moving Range
0 1 2 3 4 5 6 7 0 25 50 75 100
Y_Cookie-T 2 UCL=1,947
__
MR=0,596
Normality Test 0 LCL=0
Group N Mean 95% CI StDev 95% CI Min Median Max P Decision
1 21 41 61 81 101 121 141 161 181
Chocolate 100 3,98 (3,6536; 4,3064) 1,6452 (1,4445; 1,9112) 1 4 6 <0,005 Fail
Vanilla 100 2,46 (2,3019; 2,6181) 0,79671 (0,6995; 0,9255) 1 2 4 <0,005 Fail
Stage N Mean StDev(Within) StDev(Overall)
Vanilla 100 2,46 0,70743 0,79671 Control limits use
Chocolate 100 3,98 0,52833 1,6452 StDev(Within)
1. Histograms for Chocolate and Vanilla Cookies with the expected I-MR (Individual/ Moving Range) Control Chart for before-after
overlaid normal distribution curve (N=100), indicating comparisons, applied to compare the Taste of Van.- and Choc.-Cookies
a) deviation from the normal distribution (p< 0,005) 1. The standard deviations of both cookie types do not differ
b) different means (3,98 vs. 2,46), significantly (p= 1)
c) different standard deviations (1,645 vs. 0,796), 2. The means of both cookie types differ significantly (p< 0,001)
d) within the expected data range from: 1= very good to 6= very bad 3. The Individual Values chart shows many outliers (red), i.e. ratings
2. Time series plots for Chocolate and Vanilla Cookies: the course of that are outside the control limits, overall indicating a shift in the
data shows, that the rating of Vanilla Cookies is stationary over time, ratings from good to bad and thus a decrease of the performance
varying around the value of 2, while the rating of the Chocolate- 4. The Moving Range chart shows some noticeable changes from
Cookies decreases in the range from 2 (good) to 6 (very bad) cookie to cookie (red)
The time series plot and the I-MR chart show, that the Taste of Chocolate Cookies (Y) decreases over time
DMAIC > Process Performance > Process Capability Technische Universität München
Calculate the capability of your process for …

Indices of Process Capability
Calculation of Process Capability based on Units, Defects and Opportunities for Defects Data is nominal scaled:
Definitions Symbol Enter Your Data
Units U 1.000.000 If you know the number of units and defects then you can calculate: Yield, DPU, ppm and the Sigma
Defects D 3,40 Level
Opportunities for a defect O 1 If you know the number of units, defects and the opportunities for defect, then you can calculate:
Number of operation steps m DPO, DPMO and the Sigma Level
defective Units Symbol Calculation Result
Defects per Unit DPU D/ U 0,0000034
Defect Parts per Million PPM D/ U x 10^6 3,40
Defects per Unit
Total Opportunity TOP UxO 1.000.000
Defects per Unit Opportunity DPO DPU/ O 0,0000034
Defects per million Opportunity DPMO DPO x 10^6 3,4000000
without consideration of the Opportunities
Yield (%) (1- DPU)* 100 99,9996600
Defect (%) 100 - Yield 0,0003400
for prediction of long-term Sigma-Level from short-term measurement Sigma-Level (long-term) z-Value 4,50
for prediction of short-term Sigma-Level from long-term measurement Sigma-Level (short-term) z-Value + 1,5 6,00
with consideration of the Opportunities
Yield (%) (1- DPO)* 100 99,9996600
Defect (%) 100 - Yield 0,0003400
for prediction of long-term Sigma-Level from short-term measurement Sigma-Level (long-term) z-Value 4,50
for prediction of short-term Sigma-Level from long-term measurement Sigma-Level (short-term) z-Value + 1,5 6,00
Conversion of Yield% into corresponding z-Values (Sigma-Level) and vice versa Data is given as percentual Yield or as Sigma Level:
Enter Your Data Sigma-Level (long-term) Yield (%)
Conversion: Sigma - Yield 4,50 99,99966023269%
(Data from long-term study) 0,00 50,00000000000%
Sigma-Level (short-term) Yield (%)
Conversion: Process-Sigma - Yield 6,00 99,99966023269%
(Data from short-term study) 1,50 50,00000000000%
Calculation: Pp/ Ppk and Sigma-Level Data is cardinal scaled and normal distributed:
Sigma-Level (solely based on Pp/ Ppk and Sigma Level as an approximation of Z.bench (Interpretation of ordinal scaled, normal
Parameter Enter Your Data Pp
dispersion - unusual) distributed data under reservation)
Lower Specification Limit (LSL) 1,00 2,47 0,82
Upper Specification Limit (USL) 4,00 Sigma-Level (based on position & Mikel Harry (1988) of Motorola found in long-term studies that processes vary more over time and change their position because
dispersion - usual)
Ppk there is a greater chance of extreme values occurring. Typically, long-term studies have a shift of 1.5s compared to short-term
Mean (xbar) 3,98
studies. To predict long-term performance from a short-term study, the short-term performance is subtracted 1.5s. Conversely,
Standard Deviation 1,65 0,03 0,01 this results in a z value of 4.5 Sigma (a long-term measurement) Motorola's 6 Sigma (for a short-term measurement).
** Harry, M.J. (1988): The Nature of Six Sigma Quality; Motorola University Press
… nominal scaled values, percent values and cardinal scaled values

DMAIC > Process Performance > Process Capability Technische Universität München
How good is the taste of the critical Chocolate-Cookies (Y)
Capability Analysis for Y_Choc-Cookie-Taste

Summary Report
The process capability analysis shows:
2. How capable is the process? 1. Customer Requirements 1. The customer requirements are defined as specification limits on the
0 6
Z potential = 0,03
Upper spec
Target
4
1,5 rating scale of taste (see Data Collection Plan)
Low High Lower spec *
Z actual = 0,01 - Upper specification limit (USL) := 4
Process Characterization
(Taste should not be rated worse than 4:= adequate)
4. Does the process mean differ from 1 ,5? Mean 3,98 - Target:= 1,5 (:= very good/ good)
0 0,05 0,1 > 0,5 Standard deviation (overall) 1,6452
Yes No 3. Actual (overall) capability

Pp * 2. The Z-Values indicate the actual and the potential Process Capability:
P < 0,001
Ppk 0,00
Z.Bench 0,01 - Z actual= actual Sigma Level= 0,01
% Out of spec 49,52
PPM (DPMO) 495150 - Z potential= potential Sigma Level= 0,03
Actual (Overall) Capability
Are the data below the limit and close to the target? (The rating of taste of Chocolate Cookies is far away from 6 Sigma)
5. Target USL
Comments
• The process mean differs significantly from the target (p < 0,05). 3. Different Parameter of Process Capability:
• The defect rate is 49,52%, which estimates the percentage of parts
from the process that are outside the spec limits. - Pp (process performance) : not calculated due to the missing LSL
Actual (overall) capability is what the customer experiences.
- Ppk (k= katayori= centre) = 0,00
Potential (within) capability is what could be achieved if process shifts
and drifts were eliminated. - Z.Bench = 0,01
- % Out of specification = 49,52%
- PPM / DPMO = 495150
These parameter focus on different aspects of process capability
0 1 2 3 4 5 6 7 4. The (one-sample-t-) Test indicates a significant difference between
the mean of the rating of taste (= 3,98) and the target (= 1,5). We can
Because the rating data are not normal distributed, the statements conclude, that our next cookies would also not achieve the target, if
about the process capability should be interpreted with caution. we do not change the process.
But we are sure, that we have to improve the taste at least of the 5. The histogram shows a bimodal distribution and the Anderson-Darling
Chocolate Cookies. Normality Test (not shown here) indicates a significant result,
meaning that the data for the Chocolate Cookies are also not
The bimodal distribution with the two peaks indicates, that there normally distributed. This means: interpretation of the results under
were already conditions for a better process capability (see the reserve
shift between the first and last cookies in the time series plot).
Focus is now on the difference of Taste between Chocolate- vs. Vanilla-Cookies (Y)
DMAIC > Test of Hypothesis Technische Universität München
1. Hypothesis: There is a difference in Taste (Y) between Cookie-Types (x)
There is a/ no Difference in the degree of: Y_01: Output: Cookies (baked) [ Ranking Position of: rating-scale (grades 1 .. 6) ] between the Levels of: xI_01: Input: Request (cookie) [ Levels of:
10,63% Cookie-Type (Vanilla, Chocolate) ] .
2-Sample t Test for Y_Cookie-Taste by x_Cookie-Type

Summary Report
The tests of the hypothesis with the t-Tests shows:
1. Do the means differ?

2. Individual Samples 1. A significant difference in the rated Taste between Chocolate vs.
0 0,05 0,1 > 0,5 Statistics Chocolate Vanilla
Sample size 100 100 Vanilla Cookies (p < 0,001)
Yes No Mean 3,98 2,46
P < 0,001 95% CI (3,654; 4,306) (2,3019; 2,6181)
Standard deviation 1,6452 0,79671
The mean of Chocolate is significantly different from the mean of
Vanilla (p < 0,05).
2. Statistics for the Taste of the two types of cookies:
Difference Between Samples - Vanilla: n= 100, mean/ xbar= 2,46, s= 0,796
3. Statistics *Difference
- Chocolate: n= 100, mean/ xbar= 3,98, s= 1,645
Difference 1,52
4. 95% CI for the Difference
Is the entire interval above or below zero?
95% CI (1,1587; 1,8813)
*Difference = Chocolate - Vanilla

3. Difference between the means in the sample (1,52) and the
Comments
corresponding confidence interval (CI= 1,1587; 1,8813) for the
• Test: You can conclude that the means differ at the 0,05 level of
0,0 0,5 1,0 1 ,5 2,0 significance.
• CI: Quantifies the uncertainty associated with estimating the difference in
population **
means from sample data. You can be 95% confident that the true difference
is between 1,1587 and 1,8813.
• Distribution of Data: Compare the location and means of samples. Look
5. Distribution of Data
Compare the data and means of the samples. for unusual data before interpreting the results of the test. 4. The Interval Diagram plots the difference between the means and the
Chocolate
related confidence interval for the difference
Vanilla
5. The two histograms show the distribution of the rating data, with their
means and the related confidence intervals for the means
1 2 3 4 5 6
** Data are not normally distributed; interpretation of results under reserve
The t-Test shows, that there is a significant difference of 1,52 grades on the rating scale of Taste.
Our customer likes the Vanilla Cookies more than the Chocolate Cookies. The Chocolate Cookies have a problem!
Focus is now on the negative influences (x) on the Taste of Chocolate Cookies (Y)
2. Hypothesis: There is a relationship between the amount of chocolate (x)

and the rating of Taste (Y)
There is a/ no Relationship between: xMR_07: Activity: weigh ingredients [ Degree of: Chocolate weight (grams/ g) ] and: Y_01: Output: Cookies (baked) [ Ranking Position of: rating-scale
46,24% (grades 1 .. 6) ] according to the Principle: The larger the value of x, the larger (resp. smaller) is the value of Y.
Relationship Hypothesis Rank Correlation (Spearman)/ Ordinal-Logistic-Regression/ General Regression
Regression for Y_Choc-Cookie-Taste vs x_Chocolate-Weight_pre

Y: Y_Choc-Cookie-Taste
Summary Report The tests of the hypothesis with the Regression Analysis shows:
X: x_Chocolate-Weight_pre
Fitted Line Plot for Linear Model 1. A significant relationship between the weight of chocolate in the
1. Is there a relationship between Y and X? 2. Y = 6,433 - 0,5653 X
0 0,05 0,1 > 0,5

Cookies and the rating of their Taste (p < 0,001)
6
Yes No
2. The Fitted Line Plot (scatter plot with a regression line) shows a linear
Y_Choc-Cookie-Taste
P < 0,001
The relationship between Y_Choc-Cookie-Taste and
x_Chocolate-Weight_pre is statistically significant (p < 0,05).
4 relationship between the weight of chocolate in the cookies (x) and the
rating of their Taste according to the regression equation:
2
Choc-Cookie-Taste= 6,433 - 0,5653 x Weight of chocolate in cookie
3. % of variation explained by the model (The more chocolate the better the rating of taste)
0% 1 00%
0,0 2,5 5,0 7,5 1 0,0
Low High x_Chocolate-Weight_pre 3. The strength of the relationship between weight of chocolate and
R-sq = 96,51%
96,51% of the variation in Y_Choc-Cookie-Taste can be explained by the
Comments Taste is expressed by R-square/ R2 (Determination-Coefficient).
regression model. The fitted equation for the linear model that describes the
relationship between Y and X is:
In this example 96,51% of the variation of the rating of the Taste can be
Y = 6,433 - 0,5653 X
If the model fits the data well, this equation can be used to predict
explained by the variation in chocolate weight, i.e.: the weight of
4. Correlation between Y and X
Y_Choc-Cookie-Taste for a value of x_Chocolate-Weight_pre, or find
the settings for x_Chocolate-Weight_pre that correspond to a desired chocolate is a strong determinant of the rating of Taste.
value or range of values for Y_Choc-Cookie-Taste.
-1 0 1
Perfect Negative No correlation Perfect Positive A statistically significant relationship does not imply that X causes Y.
4. The negative correlation of r= -0,98 (r2= 96%, see 3.) confirms the
-0,98 negative relationship in the regression equation and the Fitted Line Plot:
The negative correlation (r = -0,98) indicates that when
x_Chocolate-Weight_pre increases, Y_Choc-Cookie-Taste tends to the higher the value of x, the lower (better) the value of Y
decrease.
** Data are not normally distributed; interpretation of results under reserve
The regression analysis shows, that there is a significant and very strong relationship between the amount of chocolate in a Chocolate
Cookie and the rating of its Taste: the more the better – within the investigated range of the scales.
Our customer likes the chocolate in the chocolate cookies. Why does the amount chocolate in the cookies vary?
Focus is now on the influences (x) on the variation of the amount of chocolate in Chocolate Cookies (Y)
3. Hypothesis: There is a difference in the Taste of a Chocolate Cookie (Y)

between the conditions the Chef eats the chocolate from ingredient vs. not (x)
There is a/ no Difference in the degree of: Y_01: Output: Cookies (baked) [ Ranking Position of: rating-scale (grades 1 .. 6) ] between the Levels of: xMR_08: Activity: knead ingredients [ Levels
69,26% of: Chef ate chocolate from weighed portion (yes, no) ] .
2-Sample t Test for Y_Choc-Cookie-Taste by x_Test_of_Chocolate

Summary Report The tests of the hypothesis with the t-Tests shows:
1. Do the means differ? Individual Samples 1. A significant difference in the Taste of the Chocolate-Cookies
0 0,05 0,1 > 0,5 2. Statistics
Sample size
no
20
yes
80
between the conditions: Chef ate chocolate from ingredients:
Yes
P < 0,001
No Mean
95% CI
1,5
(1,260; 1,740)
4,6
(4,3384; 4,8616)
yes vs. no (p < 0,001)
The mean of no is significantly different from the mean of yes (p <
0,05). 2. Statistics for the two conditions:
Difference Between Samples
- ate Choc: no: n= 20, mean/ xbar= 1,5, s= 0,51299
95% CI for the Difference

Difference -3,1 - ate Choc: yes n= 80, mean/ xbar= 4,6, s= 1,1757
4. Is the entire interval above or below zero?
95% CI (-3,4479; -2,7521)
*Difference = no - yes
3. Difference between the means in the sample (- 3,1) and the
Comments
corresponding confidence interval (CI= -3,45; -2,75) for the
-3 -2 -1 0 significance.
population
is between -3,4479 and -2,7521.
5.
Distribution of Data • Distribution of Data: Compare the location and means of samples. Look
for unusual data before interpreting the results of the test.
4. The Interval Diagram plots the difference between the means as well
Compare the data and means of the samples.
no as the confidence interval for the difference**
5. The two histograms show the distribution of the rating data, with their
yes means and the related confidence intervals of the means.
**Data are not normally distributed; interpretation of results under reserve
1 2 3 4 5 6
The t-Test shows, that there is a significant difference of 3,1 grades on the rating scale of Taste.
If the Chef eats the already weighed chocolate, then the rating of taste of the resulting Chocolate-Cookies decreases!
Focus is now on the root causes (x´) of eating chocolate from the ingredients (x)
4. Hypothesis: There is a difference in: the amount of chocolate in (x) in the Chocolate Cookies
between the conditions: Chef eats the chocolate from the ingredients vs. not (x)
4. Graphical
There is a difference in: (Y) between: Levels of (x) Scale Level of Y Scale Level of x
Representation:
Statistical Test:
The amount of chocolate in (x) in the Chef eats chocolate from the ingredients
cardinal nominal Box-Plot t-Test
Chocolate Cookies vs. not (x)
2-Sample t Test for x_Chocolate-Weight_pre by x_Test_of_Chocolate

Summary Report The tests of the hypothesis with the t-Tests shows:
1. Do the means differ? Individual Samples 1. A significant difference in: the amount of chocolate in the Chocolate
0 0,05 0,1 > 0,5 2. Statistics
Sample size
no
20
yes
80
Cookies (x) between the conditions: Chef eats chocolate from the
Yes No Mean
95% CI
8,7644
(8,521; 9,008)
3,2337
(2,7895; 3,6780)
ingredients: yes vs. no (x) (p < 0,001)
P < 0,001
The mean of no is significantly different from the mean of yes (p <
0,05). 2. Statistics for the two test conditions:
Difference Between Samples
- ate Choc: no: n= 20, mean/ xbar= 8,76; s= 0,51
95% CI for the Difference

Difference 5,5307 - ate Choc: yes n= 80, mean/ xbar= 3,23; s= 1,99
4. Is the entire interval above or below zero?
95% CI (5,0313; 6,0301)
*Difference = no - yes
3. Difference between the means in the sample (5,53) and the
Comments
corresponding confidence interval (CI= 5,03; 6,03) for the population
0,0 1,5 3,0 4,5 6,0 significance.
means from sample data. You can be 95% confident that the true difference 4. The Interval Diagram plots the difference between the means as well
is between 5,0313 and 6,0301.
5. Distribution of Data • Distribution of Data: Compare the location and means of samples. Look
for unusual data before interpreting the results of the test.
as the confidence interval for the difference**
no
5. The two histograms show the distributions of the rating data, with their
means and the related confidence intervals of the means.
yes
**Data are not normally distributed; interpretation of results under reserve
0 2 4 6 8 10
The t-Test shows, that there is a significant difference of 5,5 grams chocolate in Chocolates Cookies.
If the Chef eats chocolate from the ingredients, then the weight of chocolate in Chocolate Cookies decreases!
The result confirms: Eating chocolate from the ingredients (x) reduces the taste of Cookies (Y)
DMAIC > Root Cause Analysis Technische Universität München
Analysis strategy: Prioritize the problems and focus on important problems …
Process-Mapping-Analysis of the Process: bake cookies Basic prioritization criterion:

Please specify the Process-Steps in detailed Activities the format: Verb + Noun (e.g.: weigh Ingredients) • Severity of Problems (Voice-to-Critical)
Who … … does what?
1. Activity 2. Activity 3. Activity 4. Activity 5. Activity 6. Activity 7. Activity Prioritize between categories:
Process-Steps Chef determine order weigh ingredients knead ingredients roll out the dough shape cookies bake cookies box cookies
• Quality problems have priority
Ingredients Cookie diameter Cookie weight • Quality problems can lead to availability
weight wrong wrong wrong problems for necessary corrections and
… the Quality order misinterpreted scale wrong calibrated variable manual sizing feedback loops
Which Influences of the: • Corrections and feedback loops consume at
- Methods and
- Resources
negatively affect:
… the Availability
customer not available
for clarification of order +waiting for missing
ingredients + laborious manual
kneading + laborious manual
shaping + oven occupied and not
available + tins not available
least human resources
• Availability problems often result from the
sum of many small delays in several
… the Consumption
clarification with the
customer +clarifications about
missing ingredients + laborious manual
kneading + laborious manual
shaping
activities. Quality problems are often
triggered by exactly one activity
Prioritize within categories:
Pareto Chart of Y_Cookie-Defects Pareto Chart of Y_Cookie-Defects • The Pareto diagram gives an initial
500
Cookie weight wrong indication of the priority of the problems
100 100
Y_Number_Of_Defects
100
based on their frequencies or costs
Y_Costs_of_Defects
80 80 400
80
300 • If there are dependencies between the
Percent
Percent
60 60
60
40 40 200 problems, the most frequent or expensive
40
100
does not need to have the highest priority
20 20 20
Cookie diameter wrong
0 0 0 0
Methods for prioritization:
Y_Cookie-Defects Y_Cookie-Defects t
d ng ng ng en nt ed ba
d ng ke
n
rn ng ng de
d
• Depended problems: Relationships and
ba ro ro ro ok
r
ed ro bu ro ro
e br bu te w ro w w ee
st tw rw tw ie ce s t b ie t r c
ta h e h ie k ex ta gh ie ok gh et
e ex priorities can be uncovered by plausibility
ie ei
g et ei
g ok o ie ei ok ei te
ok am Co te ok Co Co m
da
w
di
w Co da w w ia
Co
o kie
ie ie
nts ir y Co
ok
ie
ie
nts
ie
d
pi
ry considerations, a corresponding hierarchy
Co ok ed ex
p
Co ed ok ex
Co gr ts gr Co ts tree, confirmed by correlation/ regression.
n n
In
ed
ien I
ed
ie Ingredients weight
gr gr Start with the root problem.
Y_Number_Of_Defects 36 23 16 10
In
8 5 2 Y_Costs_of_Defects 288 46 40 25
In
20 16 1
wrong
Percent 36,0 23,0 16,0 10,0 8,0 5,0 2,0 Percent 66,1 10,6 9,2 5,7 4,6 3,7 0,2 • Independent problems: Pareto chart, Chi2-
Cum % 36,0 59,0 75,0 85,0 93,0 98,0 100,0 Cum % 66,1 76,6 85,8 91,5 96,1 99,8 100,0
Prioritization according to test, t-Test or ANOVA help to identify the
Prioritization according to problem frequency Prioritization according to problem costs problem dependencies largest deviation as the starting point.
… between and within problem categories (Quality, Availability, Consumption)

DMAIC > Root Cause Analysis Technische Universität München
Analysis strategy: Combine quantitative statistical results and logical arguments …

Root Cause Analysis by Hierarchy Tree Analysis strategy
• Create a tree for each problem
Problem Cookie Taste bad • Identify the trigger for this problem and
(Y1) break it down to its root causes
• Combine statistical results from the
Problem-Focus
(Test: Correlation/ Regression) Difference Hypothesis causal chain hypotheses tests with experts'
Relationship Hypothesis
(Prioritization by: Chocolate-Cookie Vanilla-Cookie
Pareto Diagram, 2-Sample- 2. 1. Y1 =/ ≠ Y2 … top-down assumptions, if no data are available
Taste bad (Test: Chi2, t-Test/ ANOVA) Taste bad
Proportion, t-Test, ANOVA (Y1a) Δ= 1,52 grades (Y1b) 5 x WHY …? BECAUSE …! • Ask “Why?” questions to a given cause
Y= f(x)
or separate hierarchy tree)

and note the “Because!” answers as
R2= 96,5% The logical links between 1. Why? n. If …,
causes determine the scope
assumptions of underlying causes
1. Cause-Level Because! then …!
(Causation by trigger)
Amount of chocolate of necessary solutions: • Typically, 5 “Why -Because“ levels are
too low (x1) - or: select both
recommended to identify the root causes
- and: select (at least) one
• Test the assumptions statistically if
2. Why?
corresponding data are available
2. Cause-Level Chef did NOT nibble on 3. Δ= 3,1 grades in Taste Chef nibbles on Expiry date of Because!
between conditions Scale incorrectly • For quality problems sometimes more
(intermediate causation) Chocolate of Chocolate of ingredients exceeded
Δ= 5,5 grades in Amount calibrated (x1.3)
ingredients (x1.1) 4. ingredients (x1.2) (x1.4) than 5 cause-levels are needed
between conditions
Reality check Reality check
3. Why? 3. If …,
• Availability and consumption problems
Chef ignores the
or Chef has grown and and
Chef has too few Because! then …! sometimes need less than 3 levels
3. Cause-Level now reaches the
(intermediate causation)
kitchen standards opportunities to nibble • Stop the analysis (of a branch) if:
ingredients on the
(x1.2.1)
table (x1.2.2)
on chocolate (x1.2.3) - Reality checks are necessary
- Root causes are identified
4. Why? 2. If …,
Because! then …!
- The next cause will be related to a
4. Cause-Level Information in recipe Body growth is part of Manager has restricted
person which allows finger pointing
exceeds the Chef's the development of the consumption of
(intermediate causation)
capabilities (x1.2.1.1) young Chefs (x1.2.2.1) chocolate (x1.2.3.1) - unchangeable facts of reality are
identified
Unchangeable
fact of reality Develop solutions to n. Why? 1. If …, - Identified causes explain 80% of the
Manager
n. Cause-Level Too much chocolate is - Eliminate, Because! then …! problem frequency
overestimated the - Adjust or
causal chain
(Causation by root cause) unhealthy for young
cognitive capabilities - Circumvent (root) causes bottom-up • Question assumed causes that contain:
Chefs (x1.2.3.1.1)
of the Chef (x1.2.1.1.1) 5 x IF … THEN …! no, not, none or the prefix un- (Solutions?)
Finger pointing
… to identify root causes by (statistically confirmed) assumptions about causal relationships

Development and selection of Solutions, Measures and risk prevention, Implementation

IMPROVE: From the development of solution ideas for the (root) causes (x)
to the implementation of suitable measures
 Develop solution ideas Solution-Ideas
- Focus on the selected root causes of the root cause analysis

- Find solution ideas to eliminate, adjust or circumvent these root causes Cause Solution Problem
 Evaluate the solution ideas and select suitable and beneficial solutions Solution-Selection
- Discuss pros and cons of the solution ideas and derive practicable solutions (Creativity-Techniques)
- Evaluate solution ideas according to their effort/ benefit ratio (Solution-Selection-Matrix)
 Specify solutions as measures and minimize their risks Action-List and FMEA
- Specify measures for the selected solutions and decide: Who? Does what? Until when? (Action-List) FMEA
(Failure Mode and Effects Analysis)
Risk-Analysis Improvement new Risk-Analysis
Probability of
Probability of
the Problem
the Problem
Detection of
Detection of
Severity of
R
Severity of
R
the Effect
the Effect
Cause
Cause
- Identify and minimize risks that may arise from the measures (FMEA)
actual controls to detect the Failures/
potential Failures/ Problems
Problems
potential Effects of the Failures/ Problems potential Causes of the Failure/ Problem P Countermeasures P
N N
Measure-No.
Measure (What has to be done?)
10= catastrophic
10= catastrophic
Risk-Priority-
Risk-Priority-
1= always -
1= always -
10= always
10= always
1= never -
1= never -
10= never
10= never
Number
Number
1= no -
1= no -
Rating:
Rating:
Rating:
Rating:
Rating:
Rating:
Which Failures/ Problems can result from the By which existing Controls could the Failure/ How could the trigger of the Failure/ Problem, i.e.
Measures? Problem be detected, before it occurs? What is the Effect of the Failure/ Problem? Which Influence triggers the Failure/ Problem? their Root-Causes be eliminated
...? ...? ...? ...? ...? ...?
Let the Chef weigh 10 g of chocolate for her immediate brushing of teeth after check of
1. can cause caries periodic dental prophylaxis 5 tooth loss 7 sugar/ bacterial plaque 10 350 7 2 5 70
special Chocolate Check chocolate
...? ...? ...? ...? ...? ...?
...? ...? ...? ...? ...? ...?
...? ...? ...? ...? ...? ...?
...? ...? ...? ...? ...? ...?
...? ...? ...? ...? ...? ...?
...? ...? ...? ...? ...? ...?
...? ...? ...? ...? ...? ...?
...? ...? ...? ...? ...? ...?
 Decide on measures and implement them Decide and Implement
- Present the measures to your Sponsor and let him/ her decide on their implementation
- Implement measures

Summary IMPROVE: Suitable measures specified and then implemented
Outlook Control: Verification of the improvement, development of a process management plan, completion of the project
DMAIC > Solution-Ideas Technische Universität München
Develop solution ideas to eliminate, adjust or circumvent the root causes,
Creativity techniques Morphological Analysis Design of Experiments (DoE)

Attributes Characteristics
6 participants note 3 ideas each on their worksheet. Then they pass it to the
6-3-5 Method Amount of chocolate (x2)
respective neighbours. They add their own 3 connotations. After moving 5 dough type yeast dough shortcrust pastry puff pastry
(Brainwriting)
times the worksheets arrive at the source again. Result: 108 ideas in 30 min.
basic flavour sweet salty spicy sour 80g (x2.1) 100g (x2.2) 120g (x2.3)
Separation of motives guiding cognitive processes into six consecutive
Six Thinking Hats
phases, marked by different colours (hats for the particpants) diameter 5 cm 10 cm 15 cm 20 cm
chocolate (x1)
full milk Y1 ratings
New design variants for given objects, such as products and services, based
Type of
Morphological Analysis (x1.1) Y2 costs
on new combinations of the characteristics of their attributes shape round oval rectangular quadradratic
Problem solving method to overcome trade-offs or dilemma between dark
TRIZ topping chocolate dragées nuts tomatoes (x1.2)
contradictory objectives or elements
Overview of selected creativity techniques Example of a morphological analysis Example of a Design of Experiments
Search for Solutions

Root-Causes (Branches of Hierarchy Tree) Solution Ideas Objection Solution
Chef has too few opportunities to nibble on chocolate
Keep the Chef away from the bakery Chef will protest ./.
(x1.2.3)
Checking the quantity of chocolate just before mixing it
Manager has restricted the consumption of chocolate Considerably higher effort, further waiting times
Y_01 | into the dough and, if necessary, compensating for the ./.
(x1.2.3.1) expected at the scale
Problem: missing chocolate
Cookies Unclear how much more chocolate has to be weighed; Chef may weigh herself an extra portion of chocolate;
Too much chocolate is unhealthy for young Chefs Weigh more chocolate in advance to compensate for
Taste bad the amount will vary if the Chef has different amounts she is also informed about the importance of the
(x1.2.3.1.1) the Chef's nibbling
of nibbles in each case ingredients for the taste of the cookies
Chef documents the baking process for standard recipes
Information in recipe exceeds the Chef's capabilities Translate standard recipes for children; detailed Acceptance of standards are low if they specified
with her own photos and drawings; Manager coaches
(x1.2.1.1) description of activities illustrated with images externally and then prescribed
the creation of this standard baking book for the Chef
Search for Solutions by combining brainwriting and Six Thinking Hats
… identify objections and reformulate ideas to practicable solutions

DMAIC > Solutions Technische Universität München
Select the problem, transfer the root causes, the related solutions and evaluate the solutions
Cause determines
• Select the problem
the Problem to:
Problem-Costs
Determination
Costs of the
Reduction of
Rank (Effort/
Rank
Kano-
Category
Problem/ • Estimate the quality costs of the problem
Benefit)
Year:
Sum of
Benefit
• Transfer the root causes of the branches of the
Effort
Problem Root-Causes Solutions
x1.2.3 Chef has too few opportunities to nibble on chocolate
Chef may weigh herself an extra portion of chocolate; she is also informed about the importance of
hierarchy tree
100,00 €
x1.2.3.1 Manager has restricted the consumption of chocolate
x1.2.3.1.1 Too much chocolate is unhealthy for young Chefs
75% the ingredients for the taste of the cookies
8 4 1 75 €
• Estimate the percentual determination of the
x1.2.1.1 Information in recipe exceeds the Chef's capabilities 5%
Chef documents the baking process for standard recipes with his own photos and drawings;
Manager coaches the creation of this standard baking book for the Chef
4 10 2 5 € problem by root causes.
Y_01 | Problem: COOKIES ...? ...? ./. € • Transfer the identified solutions
1 Must-Be (BAKED) TASTE CRUMBLY-
BLAND
80% ...?
...?
...?
...?
./.
./.
€
€
• Evaluate the relative benefits and efforts of the
...? ...? ./. € solutions
...? ...? ./. €
...? ...? ./. € The ranking and the reduced costs give you hints
...?
...?
...?
...?
./.
./.
€
€
for the selection of appropriate solutions.
Solutions
The Chart Solution-Selection shows

all Solutions in an: After the identification of the Root-Causes the Solutions can now be developed.
Effort x Benefit diagram. Depending on the Type of Cause the Solution can serve to:
- eliminate,
- adjust or
The size of bubbles correspond to - circumvent a Root-Cause.
the effect of the solutions,
At first focus on the Problems with the highest Priority.
Unfortunately, Excel i.e. the reduced costs. 1. Calculate the Quality-Costs, which are caused by the Problem or at least estimate them.
sometimes does not
2. Enter the identified Root-Causes.
display the names of
the bubbles, so please Recommended are all solutions of 3. Evaluate how strong the Problem is determined by each Root-Cause.
4. The Sum of Determination indicates the portion, how strong all Root-Causes together determine the Problem.
supplement them the lower half triangle, where the 5. Develop a Solution-Ideas which could manage the Root-Cause.
manually. benefit is greater than the effort. 6. Evaluate the Benefit of the Solution, i.e. the contribution of the Solution to eliminate, adjust or circumvent the Root-Cause.
7. Evaluate the Effort for the Solution, i.e. the necessary consumption (time/ costs) to implement the Solution.
8. The Effect shows the product: Quality-Cost x Determination of the Problem, i.e.:
Nevertheless, there may be reasons
If the Solution is suitable, then the Quality-Costs of the Problem will be reduced by this amount.
to choose solutions of the upper half 9. The Rank of a Solution results from the relation: Effort/ Benefit, i.e.:
triangle as well. The higher the Effort of a Solution is in relation to the Benefit of the Solution, the lower is its rank.
The Chart: Solution Selection shows all Solutions in a Effort x Benefit Diagram.
The size of bubbles correspond to the Effect of the Solutions, i.e.: reduced Quality-Costs.
The ranking and the chart: Solution-Selection prioritize the solutions for their selection
DMAIC > Action-Plan Technische Universität München
Specify measures to implement the solutions
Action-Plan
Rank (Effort/
Measure-No.
of Problem-
Reduction
Benefit)
Costs
Costs of Decision on Implementation-

Solutions Measure (What has to be done?) Result (What will be achieved?) Risk-Reduction-Measure (from FMEA) Implementation Cost center Deadline Responsibility implementation Status in %
1. Purchase of additional necessary chocolate

Chef may weigh herself an extra portion of chocolate; she is also informed about the importance of 2. Let the chef weigh an additional 10g of chocolate for a nibble Immediate brushing of teeth after each test of
1 75 €
the ingredients for the taste of the cookies
1.
3. Provide this extra portion on a separate plate
Integrity of ingredients ensured
chocolate
10,00 € SiSi123456789 15.11.20xx Chef yes 60%
4. The chef is informed about the importance of the ingredients for the taste at every baking
1. Purchase of necessary resources for documentation (camera, notebook, pens, etc.)

2. Introduction to operating the camera
Chef documents the baking process for standard recipes with his own photos and drawings; 3. Determining the activities in the process that are to be documented Notebook with step-by-step documented standard Pilot the new standard based on one concrete
2 5 €
Manager coaches the creation of this standard baking book for the Chef
2.
4. Printout of photos in the office recipes recipe
60,00 € SiSi123456789 31.12.20xx Manager of Chef yes 20%
5. Creation of the standard (photos, drawings, texts)
6. Training of the new standard
./. € ...? ...?
Influences that can make implementation more difficult: Action-Plan

• Individual measures are too complex  Break them down into specific tasks and
put them in their chronological order
• Expected result of implementation is unclear  Determine the expected result In the action-plan the solutions are specified.
• Diffuse responsibility for implementation  Determine responsibilities and deadlines
• Resources for implementation are necessary  Inform your Sponsor and let him/her decide 1. Specify in the measures exactly what needs to be done to implement the solutions.
• Resistance to implementation  see Stakeholder-Communication
• Implemented measures are not accepted  Have the Sponsor present the measures to all 2. Specify the result, which will be achieved by the implementation.
affected employees before implementation
• Return to previous habits  Let the Sponsor check compliance with
3. Perform an FMEA (see FMEA).
new standards after their implementation
Risks of the measures are reduced by:
• Carrying out an FMEA 4. Estimate the costs of the task implementation and define the cost center for the payment.
• Exemplary implementation for a product/service on a line, at a location (piloting)
5. Define a deadline for the implementation.
Decision for the implementation:
The decision for the implementation of the measures is made by the Sponsor (process owners) 6. Define a responsible person for every measure.
within the IMPROVE Steering presentation.
Prepare this presentation so well that decisions can be made at the end of the presentation if
possible. It is useful to inform participants in advance about the measures to be taken. 7. Monitor the progress of implementation.
Make sure that a risk analysis (FMEA) for the measures is performed before implementation
DMAIC > FMEA Technische Universität München
Specify the potential failures/ problems of the specified measures, their causes and effects
FMEA
Risk-Analysis Improvement new Risk-Analysis
(Failure Mode and Effects Analysis)
Probability of
Probability of
the Problem
the Problem
Detection of
Detection of
Severity of
Severity of
R R
the Effect
the Effect
Cause
Cause
actual controls to detect the Failures/
potential Failures/ Problems
Problems
potential Effects of the Failures/ Problems potential Causes of the Failure/ Problem P Countermeasures (integrated in Action-Plan) P
N N
Measure-No.
Measure (What has to be done?)
10= disastrous
10= disastrous
1= each time -
1= each time -
Risk-Priority-
Risk-Priority-
1= minimal -
1= minimal -
10= always
10= always
1= never -
1= never -
10= never
10= never
Number
Number
Rating:
Rating:
Rating:
Rating:
Rating:
Rating:
Which Failures/ Problems can result from the By which existing Controls can the Failure/ How could the trigger of the Failure/ Problem, i.e.
Measures? Problem be detected, before it occurs? Which Effect results from the Failure/ Problem? Which Influence triggers the Failure/ Problem? their Root-Causes be eliminated?
1. Purchase of additional necessary chocolate

2. Let the chef weigh an additional 10g of chocolate for a nibble
Immediate brushing of teeth after each test of
1. 3. Provide this extra portion on a separate plate can cause caries periodic dental prophylaxis 5 tooth loss 8 sugar/ bacterial plaque 10 400 8 2 5 80
chocolate
4. The chef is informed about the importance of the ingredients for the
taste at every baking
1. Purchase of necessary resources for documentation (camera, notebook,

pens, etc.)
2. Introduction to operating the camera
New standard has gaps or is still ambiguous and/ Pilot the new standard based on one concrete
2. 3. Determining the activities in the process that are to be documented Chef still prefers to work after spontaneous ideas Close monitoring of the Chef during baking 3 cookies do not meet customer requirements 7 7 147 7 6 1 42
or not motivating enough for our Chef recipe
4. Printout of photos in the office
5. Creation of the standard (photos, drawings, texts)
6. Training of the new standard
increase
1 detectability 10
FMEA (Failure Mode and Effects Analysis) is a problem-solving procedure to evaluate and reduce decrease existing detection
risks of products/ services, their components and processes. 1 probability 10
controls
The risk is indicated by the Risk-Priority-Number (RPN) which varies between 1 ...1000.
CAUSE PROBLEM
Risks with an RPN > 100 should be reduced by:
- Increasing the detectability of the problem before its occurrence and

- Decreasing the probability of occurrence of the cause of the problem counter-measure EFFECT
The FMEA here has been adapted her to evaluate the risks of the developed measures. 1 severity 10
Risk Priority Number (RPN) = detectability x severity x probability
Calculate the related risk and reduce high risks (RPN> 100) by countermeasures
DMAIC > FMEA Technische Universität München
Instructions
FMEA (Failure Mode and Effects Analysis)
The FMEA, like the DMAIC itself, includes a complete problem-solving cycle. By multiplying the subjective assessments, the Risk-Priority-Number (=RPZ) results:
FMEA is typically used to reduce the risk of components (e.g. the brake on the car) (product
RPN= Detectabilty x Probability x Severity
FMEA) or process steps (process FMEA).
In Six Sigma, it serves to identify and reduce the possible risks of the formulated measures. The RPN can vary between 1 (lowest risk) and 1000 (highest risk).
As a rule of thumb, the risk of a formulated Measure with an EPC > 100 should be reduced
by countermeasures.
The FMEA here also consists of: Possible countermeasures include:
- a problem that may result from a Measure, - increase the Detectability of the Problem
- an effect of the problem, e.g. on the customer or the business, - reduce of the probability of occurrence of the Cause
As a rule, it is assumed that the Severity of the Effect cannot be reduced if the Problem
- a cause that triggers the problem and
occurs, even if, for example, an airbag could mitigate the Effect of the brake failure.
- countermeasures.
After formulating countermeasures to increase the Detectability of the Problem and/or

reduce the Probability of occurrence of the Cause, exactly the Detectability and Probability
In addition to the identification of problems, causes, effects is evaluated:
are re-evaluated.
If this assessment leads to the RPN becoming < 100, then the risk is generally considered
- the Detectability of the Problem before it occurs
controlled and the countermeasures can be implemented.
- the Probability of the Cause
In sigmaGuide, the countermeasures are linked to the action list and are thus part of the
- the Severity of the Effect
Measures already formulated.
Each component is rated on a 10-level rating scale (ordinal-scale), subjectively but typically
by experts. This means:
- Detectability= 1: high detectability; Detectability= 10: low detectability
- Probability= 1: low probability; Probability= 10: high probability
- Severity= 1: low severity; Severity= 10: high severity

Data-Evaluation, Process-Performance, Improvements & Benefits, Process-Management, Project Completion

Control: From the evaluation of the improvement to sustainable success
 Graphical Evaluation and Process-Performance (see the recommendations in the Data Collection Plan) Process Capability Control Chart
- Inspect the data with descriptive graphs and charts, e.g. Pareto chart, Boxplot, Time series plot
Process Capability Report for Yt1 _Cookie-Weight I-MR Chart of Y_Consumption_per_Day
Summary Report
Is the process mean stable? Comments
LSL Target USL The process mean may not be stable. 5 (15,6%) data points are out of control
Process Data Overall chance, even when the process is stable.
LSL 18 Within
Target 20 Yes No
USL 22 Overall Capability 15,6%
Sample Mean 20,9507 Z.Bench 0,1 8
Sample N 60 Z.LSL 1 ,26
StDev(Overall) 2,3331 5 Z.USL 0,45
StDev(Within) 2,04775 Ppk 0,1 5
- Monitor process performance over time, e.g. I/ MR Chart

Cpm 0,26
Potential (Within) Capability Individual and Moving Range Charts
Z.Bench 0,31
25
Z.LSL 1 ,44 UCL=23,92
Z.USL 0,51
Individual Value
Cpk 0,1 7
10 _
X=5,89
-5
LCL=-12,13
UCL=22,14
20
- Calculate process capability, e.g. Pp/ Ppk, Sigma Level
Moving Range
10
1 5,0 1 6,5 1 8,0 1 9,5 21 ,0 22,5 24,0 25,5 __
MR=6,78
Performance
Observed Expected Overall Expected Within 0 LCL=0
PPM < LSL 1 00000,00 1 02992,26 74799,56 1 4 7 10 13 16 19 22 25 28 31
PPM > USL 266666,67 326450,62 3041 79,60
PPM Total 366666,67 429442,89 378979,1 5 N: 32 Mean: 5,8947 StDev(within): 6,0087 StDev(overall): 4,0946
Control limits are estimated using the StDev(within).
 Verify the improvement statistically Test the Difference
- Test the effect of the measures using difference hypotheses for the important output variables
Boxplot of Y_Cookie_Weight
27,5
25,0
- Calculate the financial and estimate the non-financial benefits on the basis of significant differences
Y_Cookie_Weight
22,5
- Compare your results with your targets from the Project-Charter

20,0
1 7,5
1 5,0
t_1 t_2
time_of_measurement
 Ensure sustainability of the improved process performance Process-Management & Standards
- Create a Process Management Plan

- Establish a Continuous Improvement Process (CIP)
 Finish the project Story-Book & Belt Certification
- Summarize your lessons learned, indicate detected potentials for improvement

- Finish and present your Six Sigma Project-Story-Book and handover the project
Summary Control: Verify the improvement and ensure its sustainability

Outlook: next Project?
DMAIC > Evaluation of new collected data Technische Universität München
The overall number of defect Cookies was reduced by 38%

Boxplot of Y_Cookie Taste before; Y_Cookie Taste after_1st week Graphical analysis descriptive statistics:
1a. Pareto Chart of Y_Cookie-Defects_before 2. Summary Report
Y_Number_Of_Defects_before
100 100
Distribution of Data by Group
Compare the center and the variability across samples. Identify any outliers.
1. Pareto charts
80 80
• Frequencies of Cookie defects in comparison
Percent
60 60 6
a) before vs. b) after improvement (N= 100)
40 40
4
• Overall number of errors decreased by 38%:
20 20
- 83% reduction in: Taste of Cookies
0 0
Y_Cookie-Defects_before d ng ng ng en rn
t
ed
- 80% reduction in: Weight of ingredients
ba ro ro ro ok ed
2
e br bu
st tw rw tw ie ce
ta e x • Thus it came to a shift of the ranking of the
Data
h e h i k e
ie ei
g et ei
g ok o
te
ok w am w Co Co da
Co
o k ie
ie
di
ie
n ts
piry
0
error frequencies.
Co ok ed ex
Co gr ts
In ien • Primary problems now:
ed
gr - Cookie weight wrong
In -2
Y_Number_Of_Defects_before 36 23 16 10 8 5 2
Percent 36,0 23,0 16,0 10,0 8,0 5,0 2,0 - Cookie diameter wrong.
Cum % 36,0 59,0 75,0 85,0 93,0 98,0 100,0
-4
Y_Cookie_Taste_before Y_Cookie_Taste_after_1st
Before / After comparison Cookie defect frequencies 2. Boxplot
a. Y_Cookie Taste Y_Cookie Taste b. • Rating of Cookie taste a) before (N= 100) vs.
1b. Pareto Chart of Y_Cookie-Defects_after Statistics before after (1st week)
b) 1st week after improvement (N= 32)
N 100 32
70
Y_Number_Of_Defects_after
Mean 2,46 3,9219

60 100 StDev
Minimum
1,2012
1
2,4695
-3
• The mean taste decreased from:
50 80 Maximum 5 6
a) xbar= 2,46 (before) to
Percent
40
60 b) xbar= 3,92 (after) (1= very good; 6= very bad)
30
20
40 Before/ After (1st week) comparison of Cookie Taste rating • The dispersion of taste also increased in the
10 20
1st week, from:
0 0 • The overall number of defects in Cookies was reduced by 38% (100  62). a) s= 1,20 (before) to
Y_Cookie-Defects_after t
ng ng en ba
d
rn ed ng
ro ro ok bu ed ro
h tw e rw e
br
ta st
e
k ie x ce t w • The focused defects in taste were reduced by 83% b) s= 2,47 (after)
g et ok
i o e gh
ei ki
e
Co te ei
w am Co oo da w
ok ie
ie
di C iry n ts • Additionally the defects in the weight of the ingredients were reduced by 80%. • The range of data changed from :
p ie
Co ok ex ed
Co ts gr
1st
• In the week the average taste rating even decreased. One reason is that the a) 1= very good to 5= bad to
ien In
ed b) 1= very good to 6= very bad.
gr process was not stable in the 1st week after implementation ( control chart).
In
Y_Number_Of_Defects_after 23 16 8 6 5 2 2
Percent 37,1 25,8 12,9 9,7 8,1 3,2 3,2
• Potential for improvement is now mainly in: Deviations in cookie weight. • There is at least one invalid value (-3)
Cum % 37,1 62,9 75,8 85,5 93,5 96,8 100,0
In the 1st week however, the taste ratings even decreased from xbar= 2,46 to xbar= 3,92
DMAIC > Process Performance >> Control Chart Technische Universität München
Patterns in the data of the 1st week after implementation indicate, that the process …
I-MR Chart of Y_Choc-Cookie-Taste first Week Look for these patterns:
Summary Report
Chart Test Out-of-Control Points

Is the process mean stable? Global Trend Cyclical
Comments
The process mean may not be stable. 6 (18,8%) data points are out of control
chance, even when the process is stable.
Yes No 1a. I Test 1: Outside control limits 1

Test 2: Shift in mean 10-14
18,8%
Shifts Drifts 1b. MR Test 1: Outside control limits 2

Individual and Moving Range Charts
1a.10 UCL=10,27 Test for signals from the Minitab Assistant I-MR Control Chart
Individual Value
5 _
X=3,92
Oscillation Mixture
0
LCL=-2,43 Test for signals from the Minitab Statistics I-MR Control Chart
1b.10
UCL=7,80
1a. Individual Value (I) Chart of Y_Choc-Cookie-Taste_post (Tests from I-MR-Chart from statistics menu)
Moving Range
5
Excessive Out TEST 1. One point more than 3,00 standard deviations from center line.
__
MR=2,39 of Control Test Failed at points: 1
0 LCL=0
1 4 7 10 13 16 19 22 25 28 31
TEST 2. 7 points in a row on same side of center line.
Test Failed at points: 8; 9; 10; 11; 12; 13; 14
N: 32 Mean: 3,9219 StDev(within): 2,1162 StDev(overall): 2,4695
Control limits are estimated using the StDev(within). TEST 3. 5 points in a row all increasing or all decreasing.
Test Failed at points: 15; 16; 17; 18; 19; 20
TEST 4. 12 points in a row alternating up and down.
Data of controlled processes are typically normally distributed. Test Failed at points: 31; 32
The variation of the data is based exclusively on random common causes.
1b. Moving Range (MR) Chart of Y_Choc-Cookie-Taste_post
Specific influences can systematically affect the process. Tests of the control charts can identify TEST 1. One point more points more than 3,00 standard deviations from center line.
some signals in the data and assign them to certain patterns. Test Failed at points: 2
TEST 2. 18 and 12 points in a row on same side of center line.
Each systematic event will be analysed for its root causes to develop appropriate solutions! Test Failed at points: 9; 10; 11; 12; 13; 14; 15; 16; 17; 18; 19; 20; 27; 28; 29; 30; 31; 32
… was not yet under control and needs to be analysed for specific causes.
DMAIC > Process Performance >> Process Capability Technische Universität München
The before vs. after comparison of the process capability shows a significant improvement
Before/After Capability Comparison for Y_Choc-Cookie_pre vs Y_Choc-Cookie_post

Summary Report
The Process Capability Analysis shows:
Customer Requirements
Reduction in % Out of Spec
1.
100% % Out of spec was reduced by 100% from 2,66%
to 0,00%.
Lower Spec
*
Target
1,5
Upper Spec
4
1. The customer requirements are defined as specification limits on the
rating scale of taste:
3. Was the process standard deviation reduced? 2. Process Characterization
- Upper specification limit (USL): 4 (see Data-Collection-Plan)
0 0,05 0,1 > 0,5 Statistics Before After Change
Yes No
Mean
StDev(overall)
2,46
0,79671
1,65
0,47937
-0,81
-0,31734
Taste should be rated better than 4:= adequate
P < 0,001
Actual (overall) capability
- Target: 1,5 (see Data-Collection-Plan)
4. Did the process mean change?
Pp
Ppk
*
0,64
*
1,63
*
0,99 The mean of taste rating should be ≤1,5:= very good/ good)
0 0,05 0,1 > 0,5 Z.Bench 1,93 4,90 2,97
% Out of spec 2,66 0,00 -2,66
PPM (DPMO) 26621 0 -26621
Yes
P < 0,001
No
2. Improvement of statistical parameters:
xbar: 2,46  1,65 - Difference: 0,81
5. Actual (Overall) Capability
Are the data below the limit and close to the target?
Comments s: 0,80  0,48 - Difference: 0,32
Before: Y_Choc-Cookie_pre
Target USL
After: Y_Choc-Cookie_post
Before
• The process standard deviation was reduced significantly (p < 0,05). Improvement of capability indices:
• The process mean changed significantly. It is now closer to the target (p <
0,05). Ppk: 0,64  1,63 - Difference: 0,99
Actual (overall) capability is what the customer experiences.
Z.Bench: 1,93  4,90 - Difference: 2,97
After
Potential (within) capability is what could be achieved if process shifts and
drifts were eliminated. % Out of spec: 2,66  0,00 - Difference: 2,66 (100%)
3. The standard deviation is significantly reduced (p< 0,001)

(2 Sample Standard Deviation Test/ F-Test)
4. The difference in the means: is significant (p< 0,001)

• Significant and practically relevant increase of the taste (2 Samples t-Test)
judgement, from 2.46 to 1.65.
• The variability of taste was also significantly reduced. 5. Histograms for the Before vs. After comparison:
→ We offer customers a better and more consistent taste experience. The Anderson-Darling normality test shows a significant result.**
• This improvement is reflected in the process capability, increasing

approximately 3 Sigma Levels, from 1.93  4.9 Sigma.
• The performance of the process is now closer to the goal (1.5).
(see below: 1-Sample t-test for difference to target)
** interpretation of results under reserve
Focus is now on the statistical significance and practical relevance of the improvement
DMAIC > Improvements and Benefits >> Test of Hypothesis Technische Universität München
Hypothesis: There is a/no difference in: Taste (Y) between: the states before vs. after improvement (x)
2-Sample t Test for the Mean of Y_Choc-Cookie-Taste_pre (1 ) and Choc-Cookie-Taste_post (2) The tests of the hypothesis with the t-Test shows:
Summary Report
3. Do the means differ?

1. Individual Samples 1. Statistical parameters for the conditions of taste:
0 0,05 0,1 > 0,5 Statistics Y_Choc-Coo_1 Y_Choc-Coo_2
Sample size 100 100 - pre (before): N= 100, Mean/ xbar= 2,46, s= 0,796
Yes No Mean 2,46 1,65
P < 0,001 95% CI (2,302; 2,618) (1,5549; 1,7451) - post (after): N= 100, Mean/ xbar= 1,65, s= 0,479
The mean of Y_Choc-Coo_1 is significantly different from the mean of
Y_Choc-Coo_2 (p < 0,05).
Difference Between Samples 2. Difference between the means in the sample (0,81) and the
corresponding confidence-interval (CI= 0,626; 0,993) for the
Difference 0,81
4. 95% CI for the Difference
Is the entire interval above or below zero?
95% CI (0,62639; 0,99361)
population **
*Difference = Y_Choc-Coo_1 - Y_Choc-Coo_2
Comments
3. Difference in the taste between chocolate cookies pre (before) vs.
0,00 0,25 0,50 0,75 1,00 significance.
post (after) improvement is significant (p < 0,001)
is between 0,62639 and 0,99361.
• Distribution of Data: Compare the location and means of samples. Look
for unusual data before interpreting the results of the test. 4. The Interval Diagram plots the difference between the means and its
Y_Choc-Coo_1
confidence interval as well as the difference= 0 for the H0
Y_Choc-Coo_2
5. The two Histograms show the distribution of the rating data, with their
means and the related confidence intervals of the means**
1 2 3 4
** Data are not normally distributed; interpretation under reserve
The t-Test shows, that there is a significant improvement of 0,81 grades in the rating of Taste.
Our customer will like the chocolate Cookies better now. But there are still potentials for improvement!
Focus is now on the degree how good the target for the chocolate Cookies was achieved
DMAIC > Improvements and Benefits >> Test of Hypothesis Technische Universität München
Hypothesis: There is a/no difference: in Taste (Y) between: the ratings and the target (Y’)
1 -Sample t Test for the Mean of Y_Choc-Cookie_post The tests of the hypothesis with the 1-Sample t-Test shows:
Summary Report
1. Statistics for the comparison of taste:

2. Is the mean greater than 1 ,5?
1. - N= 100,
Statistics
0 0,05 0,1 > 0,5
Sample size 100
- Mean/ xbar= 1,65, confidence interval (1,57; 1,73)
Yes No
Mean
90% CI
1,65
(1,5704; 1,7296) - s= 0,479
Standard deviation 0,47937
P = 0,001
The mean of Y_Choc-Cooki is significantly greater than the target (p <
Target 1,5 - target= 1,5
0,05).
2. Difference between the rated taste of chocolate Cookies vs. the

target is significant (p < 0,001)  Deviation from the target
Where are the data relative to the target?
1 ,5 Comments 3. Histogram with target (1,5) and the mean with its confidence interval
• Test: You can conclude that the mean is greater than 1,5 at the 0,05
level of significance.
• CI: Quantifies the uncertainty associated with estimating the mean
from sample data. You can be 90% confident that the true mean is There is a significant difference between rating of taste and the target for
between 1,5704 and 1,7296, and 95% confident that it is greater than
1,5704.
• Distribution of Data: Compare the location of the data to the target.
taste, i.e. the target is still not yet reached, indicated by the target value
Look for unusual data before interpreting the test results.
falling outside the confidence interval of the mean.**
** Data are not normally distributed; interpretation under reserve

1 2
The 1-Sample t-Test shows, that we nearly reached the target of Taste with our process improvements.
This means, that there is still potential for improvement.
This target should continue to be achieved in order to remain competitive.
After all before-after comparisons tested, we will now deduce the benefits of the improvements
DMAIC > Improvements and Benefits >> Financial and Other Benefits Technische Universität München
Summarise the main results - from the problems to the measures implemented
Summary and benefits
Problems Root Causes Implemented Measures Financial Benefits Other Benefits Derivation of financial benefits based on:
x1.2.3 Chef has too few opportunities to nibble on chocolate
x1.2.3.n Chef may weigh an additional portion of 10 g of - 80€ (estimated); • Significant differences confirmed by e.g.:
x1.2.3.1 Manager has restricted the consumption of - Standardized process in the bakery;
chocolate
chocolate so that she can nibble while baking
- expected: 256€ (confirmed) - 2-Sample t-Test
Y_01 | Cookies (baked) Cookies taste of nothing
x1.2.3.1.1 Too much chocolate is unhealthy for young Chefs
x1.2.3.n Immediate brushing of teeth after nibbling - Customer can expect an improved
x1.2.1.1 Information in recipe exceeds the Chef's
x1.2.1.1 Standard recipe with detailed and illustrated - worst case: 212€ (confirmed)
and consistent taste experience - ANOVA
process description - best case: 300€ (confirmed)
capabilities
- 2-Sample % Defective (% out of spec.)
Y_02 | Cookies (delivered) delivered too early
negative Influences of the: Input (xi), Methods (xm) and the Resources (xr) Problem (Y) in category - Chi Square % Defective (% out of spec.)
on the Output are the starting point to identify their underlying Root-Causes Quality, Availability or Consumption
• Estimate reduced costs and/ or increased
Solutions (S) to eliminate, adjust or circumvent the Root-Causes
Effect (Z) on profit based on the lower limit (worst case),
Y_03 | Cookies (boxed) waste of energy
Costs (VoB) & Satisfaction (VoC) the expected costs (typical case) and upper
limit (best case) of the confidence interval
• Prepare Pareto charts to compare the costs
(before) with the expected costs (after)
Pareto Chart of Y_Cookie-Defects_before Pareto Chart of Y_Cookie-Defects_after

500 200
Y_Costs_of_Defects_before
Y_Costs_of_Defects_after
400
100 100 Costs (before) Costs (after) Savings
150 80
80
300
Percent
Percent
60 60
100 Cookie taste bad 288,00 € 48,00 € 240,00 €
200
40 40
50
100 20 20 Ingredients weight
20,00 € 4,00 € 16,00 €
Y_Cookie-Defects_before
0
t
0 0 0 wrong
d g n ng ng d Y_Cookie-Defects_after d t
ba on ke rn de ba ng ke
n
rn ng ng de
d
r ro bu ro ro ee ro ro bu ro ro
s te t w b ie t w r w c ste w b e w w c ee
ta x t i er t
gh ie ok gh et
e e ta gh ie ok et gh ex expected 256,00 €
ie ei ok Co ei te ie ei ok ei te
ok w Co w ia
m
da ok w Co Co ia
m w da
C o ie t s d y o ie d ts
ok en ie pi
r C
ok ie en ry
di ok di pi
Co re ex Co ok ex
g Co ts Co gre ts
In ie
n In en worst case 212,48 €
ed di
gr g re
I n In
Y_Costs_of_Defects_before 288 46 40 25 20 16 1 Y_Costs_of_Defects_after 48 46 40 25 16 4 1
Percent 66,1 10,6 9,2 5,7 4,6 3,7 0,2 Percent 26,7 25,6 22,2 13,9 8,9 2,2 0,6
Cum % 66,1 76,6 85,8 91,5 96,1 99,8 100,0 Cum % 26,7 52,2 74,4 88,3 97,2 99,4 100,0 best case 299,52 €
Assign the financial/ other benefits to the measures and have them independently confirmed
DMAIC > Process-Management-Plan Technische Universität München
Install appropriate control charts to identify deviations in the “new” process, …
Process-Management-Plan Define measures to sustainably maintain the process-improvements

Outputs (Y)
Ranking of
In which time intervals will How many data points u-Chart (if ok vs. different Which control limits should Who is responsible for
Target and specification How large will the sample xbar-R Chart (if N > 100 and xbar-S Chart (if N > 100 and p-Chart (if ok vs. ko is Who is responsible for In which document is the
Outputs (Y) Measurand Unit Scale-Level the control chart be should the control chart I-MR Chart (if N <= 100) defect opportunities are be used? (LCL; Center-Line; maintaining the reaction
limits (USL; LSL) size be in each time interval? if subgroup size <= 8) if subgroup size > 8) discriminated) creating the control charts? reaction plan specified?
actualized? represent? discriminated) UCL) plan?
Output (Y) Data from Data-Collection-Plan
I-Chart:
UCL: 2.324
Center-Line: 1.438
Y_01 | Problem: COOKIES (BAKED) TASTE Target: 1.5 Data Rank Ordered (Ordinal- 25 data points; no LCL: 0.551
1
CRUMBLY-BLAND
rating-scale grades 1 .. 6
LSL: 4 Scale)
weekly 25 25 Mrs X Reaction-Plan.xlsx Mr Y
subgrouping MR-Chart:
UCL: 1.089
Center-Line: 0.333
LCL: 0
Target: 0 hrs Sample size > data points!

Y_02 | Problem: COOKIES (DELIVERED) Data discrete or continuous 50 data points; subgroup
2
DELIVERY > 1 HOUR TOO EARLY/ LATE
time hrs LSL: - 1 hrs
(Cardinal-Scale)
monthly 500 50 Select a smaller time
size: 10
USL: + 1 hrs interval?
Sample size > data points!

Y_03 | Problem: COOKIES (BOXED) ENERGY- Target: 0 kw/h Data discrete or continuous 60 data points; subgroup
3 amount of electricity kw/h quarterly 120 60 Select a smaller time
CONSUMPTION > 10% WASTE USL: + 1 kw/h (Cardinal-Scale) size: 2
interval?
Reproduction of a simple control chart with Excel

The Process-Management-Plan ensures the sustainability of the implemented measures by:
a. Monitoring of the future process performance with control charts for the critical outputs, with Y_Cookie-
UCL Center Line LCL
determined values for their control limits (UCL, LCL) and center line and a schedule for their Taste
-3 2,324 1,438 0,551
actualisation 6 2,324 1,438 0,551
6 2,324 1,438 0,551
b. Responses to performance drops with a reaction plan, which determines, how signals in the 6 2,324 1,438 0,551
control charts are to be treated by whom (additional document, not part of sigmaGuide) 6 2,324 1,438 0,551
6 2,324 1,438 0,551
• The results of the control charts should be discussed regularly, daily or weekly in the first weeks 6 2,324 1,438 0,551
2,324 1,438 0,551
after implementation, in order to identify weak points immediately and react to them 6
6 2,324 1,438 0,551
• Important part of the reaction plan are regular meetings of a Continuous Improvement Team, 6 2,324 1,438 0,551
5,5 2,324 1,438 0,551
which analyses the causes of the signals and develops solutions 5 2,324 1,438 0,551
4,5 2,324 1,438 0,551
• The process owner should know and accept the reaction plan and place the control charts visibly 4 2,324 1,438 0,551
in his office. A copy should be visible to all employees on a blackboard (Visual Management) 3,5 2,324 1,438 0,551
3 2,324 1,438 0,551
… and develop a reaction plan, to solve these “new” problems and ensure sustainability
DMAIC > Process-Management-Plan >> Control limits and center line for future monitoring Technische Universität München
Determination of appropriate control limits for control charts
I-MR Chart of Y_Choc-Cookie-Taste_Week_1-2 I-MR Chart of Y_Choc-Cookie-Taste_Week_2 I-MR Chart of Y_Choc-Cookie-Taste_Week_1-2

Summary Report Summary Report Summary Report
Is the process mean stable? Comments Is the process mean stable? Comments Is the process mean stable? Comments
Evaluate the % of out-of-control points. Evaluate the % of out-of-control points. Evaluate the % of out-of-control points.
The process mean may not be stable. 16 (33,3%) data points are out of control The process mean is stable. No data points are out of control on the I chart. The process mean may not be stable. 25 (52,1%) data points are out of control
0% > 5% on the I chart. Keep in mind that you may see 0,7% out-of-control points by 0% > 5% 0% > 5% on the I chart. Keep in mind that you may see 0,7% out-of-control points by
chance, even when the process is stable. chance, even when the process is stable.
Yes No Yes No Yes No

33,3% 0,0% 52,1%
Individual and Moving Range Charts

Investigate any out-of-control points. 2a. Individual and Moving Range Charts
3a. Individual and Moving Range Charts
3b.
8
2a. UCL=7,36
2,4 UCL=2,324
Individual Value
Individual Value
Individual Value
4 _ UCL=2,32
_
X=3,16 1,6 _
X=1,438
0
3b. X=1,44
LCL=0,55
0
LCL=-1,03
0,8
LCL=0,551 -4
10 1b. UCL=1,089 10
1,0
Moving Range
Moving Range
Moving Range
5 UCL=5,16 5
0,5
__
MR=0,333
__
MR=1,58
UCL=1,09
__
0 LCL=0 0,0 LCL=0 0 MR=0,33
LCL=0
1 6 11 16 21 26 31 36 41 46 1 3 5 7 9 11 13 15 1 6 11 16 21 26 31 36 41 46
N: 48 Mean: 3,1630 StDev(within): 1,3987 StDev(overall): 2,1758 N: 16 Mean: 1,4375 StDev(within): 0,29551 StDev(overall): 0,51235 N: 48 Mean: 3,0938 StDev(within): 1,5090 StDev(overall): 2,3467
Control limits are estimated using the StDev(within). Control limits are estimated using the StDev(within). Historical values for the control limits and center lines are used.
1. Determination of the preliminary control limits and center line for an improved process:
a) Select: Estimate from the data
b) Omit points outside the control limits, if detected
1a. 2. Determination the control limits and center line for an improved process with already 3a.
controlled periods:
a) Copy the data of the controlled period in a new variable
b) Select: Estimate from the data (see 1b)
1b. 3. Use of calculated control limits and center line for future periods
a) Copy the calculated data in the fields for: Known values
b) The calculated limits and the center line of the controlled period now serve as the
reference for future data
For the first try: Estimate from the data - For sustainable monitoring: Use known values from controlled periods
DMAIC > Review and Outlook Technische Universität München
Lessons learned and …

What I learned in the course of the project, concerning:
Lessons learned
1. Subject matter/ Product:
• Structure your lessons learned according to
the seven topics and summarize them by
2. Process: short statements
3. Methods/ Tools:
4. People/ Teams:
5. Management:
6. Finance:
7. Company:
Potentials/ topics for further improvements:

Further potentials
Pareto Chart of Y_Cookie-Defects_after
1. 70
• List the remaining potentials from your
Y_Number_Of_Defects_after
60 100
project, problems you have not worked on,
50 80
promising topics that were outside your
Percent
40
60
2.
30
20
40 scope or topics you have noticed additionally
10 20
0 0
Y_Cookie-Defects_after t
ng ng en ba
d
rn ed ng
ro ro ok bu ed ro
tw w br st
e
ie ce w
3. ei
gh
am
et
er
oki
e
k ie
ta
C ook
te
ex
e ig
ht
w
di Co o da w
ie Co ry nt
s
ok ie pi ie
Co ok ex ed
Co ts gr
n In
ie
ed
gr
In
4. Y_Number_Of_Defects_after
Percent
23
37,1
16
25,8 12,9
8
9,7
6 5
8,1 3,2
2 2
3,2
Cum % 37,1 62,9 75,8 85,5 93,5 96,8 100,0
… potentials for further improvements

DMAIC > Project Progress and Evaluation Technische Universität München
Progress of the project …

Phase Tool Business-Project Standard-Project Your check MBB grade
General Project-Story-Book All headers and footers of the slides are specified, results described, interpretations provided, conclusions drawn yes 5
Define Part 1 Identify a topic for a Six Sigma project Progress of your project
Project-Topic Identify Problems of the daily work as a potential for Improvement yes 7 • Please document your project
Part 2 Define a Six Sigma project progress in sigmaGuide. If you
Project-Definition Summary: Process, Output, Problem and Effect yes 8 were only able to edit a task
Define Part 3 Implement a Six Sigma project partially, then use the entry:
SIPOC Structure the Process in its important steps, with related Supplier, Inputs and Outputs and Customer yes 10
partially.
Interviews conducted with customers, the sponsor and Interviews conducted with representatives of the municipality/
Voice to Critical yes 10
managers garbage collector
Project-Charter Complete and sign the Project-Charter yes 10
Stakeholder-Communication Identify the Stakeholder and develop a Communication-Plan yes 10
Measure Input-Analysis Describe Inputs (xI) of the Process, Requirements on the Inputs and Deficiencies yes 10 Evaluation of your project
Process-Mapping and -Analysis
Workshop conducted & documented (photos: Process- If possible: Workshop conducted & documented (photos:
yes 10 • After project completion, a Master
Mapping & team)) Process-Mapping & team (municipality/ garbage removal))
C&E Matrix xY Evaluate relationships between negative Influences of the Inputs (xI) and the Activities of the Process (xP) on the Outputs (Y) yes 10
Black Belt will evaluate the quality
Data-Collection-Plan
Operationalise Measurands of Inputs (xI), Activities (xP) and Outputs (Y); of your results.
yes 10
Recommendation for appropriate: Charts, Parameter, Process-Capability-Indices, Control-Charts, One-Sample-Tests
• Here you can achieve 0-10 points
Hypothesis Overview of all automatically generated Hypothesis, prioritized by their Risk; Recommendation for appropriate statistical Tests yes 10
per task.
Analyse Data Evaluation Plausibility of data checked and data graphically displayed yes 10
• If you receive at least 5 points in all
Process Performance Process performance determined, with capability indices and control charts yes 10
tasks, your project will be certified
Hypotheses Tests Hypotheses statistically tested yes 10 (50%).
Workshop conducted & documented (photos: Root-Cause- If possible: Workshop conducted & documented (photos: Root-
Root-Cause-Analysis
Analysis & team)) Cause-Analysis & team (municipality/ garbage removal))
yes 10 • If you earn more than this 50%, you
Improve Solutions Develop Solutions to eliminate, adjust or circumvent the Root-Causes yes 10 will be credited for your Black Belt
Action-Plan Specify Measures to implement the Solution-Ideas yes 10 certification if you want to go for
FMEA Analyse the Risks of Measures (FMEA:= Failure Mode and Effects Analysis) yes 10
this next level.
Implementation Measures implemented (at least one for each problem) yes 10
• With the best result - 10 points in
each task - you will e.g. receive a
Control Data Evaluation Plausibility of data checked and data graphically displayed yes 10
credit of 50%.
Process Performance Process performance determined, with capability indices and control charts yes 10
• In this case your Black Belt project
Hypotheses Tests Hypotheses statistically tested yes 10
doesn't have to be as demanding
Process-Management-Plan Define measures to sustainably maintain the process-improvements yes 10 and complex as it would be without
Summary and Benefits Summarize the results of the phases and demonstrate the financial and other benefits of the project yes 10 credits.
General Lessons Learned Lessons learned summarized and topics identified for future projects yes 10
Evaluation: 96% excellently passed (46% Black Belt Bonus)
… and its evaluation for certification by your Master Black Belt

End of this Course

Six Sigma - Methods and Tools for Process Improvement

LSSGB

Uploaded by

Copyright:

Available Formats

LSSGB

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

LSSGB

Uploaded by

Copyright:

Available Formats

Technische Universität München

Methods and Tools for Process Improvement

© Dr. Reiner Hutwelker Lean Six Sigma 1

Six Sigma is a concept with several meanings –– itit links

1. Improvement Program: increase customer satisfaction and reduce costs

3. Statistical Basis: process capability, process control, problem-cause relations, improvements

… to analyse and realize potentials and to verify improvements

© Dr. Reiner Hutwelker Lean Six Sigma 3

The quality of a Six Sigma program is determined by the management, …

Success factors for Six Sigma Programs:

Number of projects implemented per Green Belt:

Unsuccessful Six Sigma Programs:

Sources of improvement potentials Realization of improvement potentials

Analyses Employees Project Team

implementation realized potential

pool of improvement topics evaluation &

… identification of improvement topics and their implementation in projects

An improvement program for operational excellence should be developed and ...

Six Sigma Method

© Dr. Reiner Hutwelker Lean Six Sigma 7

Observe by asking questions – Classify to generalize and Describe to concretize

same family bakery products

attributes ingredients form taste/ delivery/...

shape diameter height weight colour

deviant conform values cylinder 10 0.7 10 239, 226, 178

Each of these assignments means: modelling and measuring the reality

Observations in the bakery assigned to the: Problems  Output – Attribute – Deviation

- not available in the required quantity (too much/ too few)

- losses of input or resources during creation (passive)

Observations in the bakery assigned to the: Causes (x), …

Causes (x) in the Process Classification of Causes (x)

Observations in the bakery assigned to the: Effects (Z), …

Problems (Y) of Outputs Classification of Problems (Y)

Cookie delivery 1 week early A

… within these categories further classified

Effects (Z) Company revenue losses

Customer compensation claims (C)

Customer dissatisfied (S) Company quality costs increased (C)

Problems (Y) (Q) (A) (Q) > 50 g waste (C) (C)

Solutions (S) Determine correct oven

The Process-Problem-Solving Model …

Input Check Process Output Check

Influences of Activities (xa)

Problem (Y) as deviation in:

Effect (Z) on:

Three important Resolution Levels of a Process …

SIPOC VoC/ VoB

… serve different Purposes in a Six Sigma Project

There are 6 standard

Process Capability: 6 sigma

There are 3 standard

-2s -1s Mean +1s +2s

Six Sigma's goal is to align performance indicators of products and services …

… to the center and to minimize their variability, according to customer requirements

Percentages and sigma levels can be transformed into each other

Six Sigma projects follow a scientific approach …

hypotheses This scientific approach

The DMAIC is a chronological guideline for problem solving …

© Dr. Reiner Hutwelker Lean Six Sigma 21