The Decay of Six Sigma and the

Rise of Quality 4.0 for

Manufacturing Innovation
October 13, 2022

Carlos Escobar, PhD

Learning Objectives

• In this session you will:

⎻ Become familiar with the technologies
and challenges of smart manufacturing

⎻ Understand the limitations of six sigms

to drive innovation in manufacturing

⎻ Learn a Quality 4.0 initiative and an

evolved problem-solving strategy that
guides its implementation
 Professional experience
Bio • Consultor Quality 4.0 Institute, present.
• Postdoc, Tec de Monterrey, 2021-present.
• Senior researcher, General Motors –
Manufacturing Systems Research Lab, Global
R&D, 2015-2021.
• Research assistant, New Mexico State
University, 2014-2015.
Education
• Master in Management, Harvard Extension
School, 2020-2023
• PhD in Engineering Sciences (DCI), Tec de
Monterrey, 2019.
• Master in Industrial Engineering, New Mexico
State University, 2016.
• Master in Quality Systems, Tecnológico de
Monterrey, 2005.
• Ingeniería Industrial, Tecnológico de Ciudad
Juarez, 2001.
Benefits of Projects
• 32 scientific publications in top journals.
• Presented across United States in top
conferences
• Book, Machine Learning in Manufacturing:
Quality 4.0 and the Zero Defects Vision,
Elsevier Publishing, 2023

• SHPE STAR of Today, SHPE Technical

Achievement Recognition, 2021
• Research interest lies within the 98%
percentile compared with the 2015 cohort
Relevant remarks

• Industrialization of AI is today a megatrend that

dominates the business landscape
⎻ 92% leaders are increasing their investments in AI

• Lack of powerful people is one of the biggest challenges

⎻ 79% of young professionals receives training outside of formal
schooling to acquire the necessary skills for Industry 4.0

• Leaders do not have a clear vision about how to create

value of out of AI
⎻ 80–87% of the big data projects never generate a sustainable solution
Agenda
• Smart manufacturing

• Six Sigma

• Artificial intelligence innovation

• Binary classification of quality

• Quality 4.0

• Learning Quality Control

⎻ Applications

⎻ Evolved Problem-Solving Strategy

• Case study

• Six Sigma vs Quality 4.0

• Take-aways
Smart Manufacturing

7
Smart Manufacturing

Definition
Emerging form of production integrating manufacturing assets of
today and tomorrow with sensors, computing platforms,
communication technology, control, simulation, data intensive
modelling and predictive engineering

Sources: Andrew Kusiak (2018) Smart manufacturing, International Journal of Production Research, 56:1-2, 508-517, DOI: 10.1080/00207543.2017.1351644
https://internetofthingsagenda.techtarget.com/definition/smart-manufacturing-SM
Smart Manufacturing

Technologies
⎻ Industrial internet of things
⎻ Robotics

⎻ Artificial intelligence
⎻ 3D-printing

⎻ Cloud storage and computing

Smart Manufacturing

Characteristics
• Generates more data than any other sector of the
economy
• Processes exhibit:
⎻ Increasing complexity
⎻ Transient sources of variation Militate against traditional
quality control methods
⎻ Hyper-dimensional feature spaces
⎻ Non-Gaussian pseudo-chaotic behavior
Six Sigma

11
Six Sigma

Evolution of manufacturing quality control movement

• Scientific method was adapted into a problem-solving
strategy
Six Sigma

• For the last 30 years, six Sigma has been widely deployed
across industries to innovate
• DMAIC (Define, Measure, Analyze, Improve, Control), a 5-
step problem solving strategy
• Six Sigma mastery levels are defined by a color convention
Six Sigma

Organization have merged traditional quality philosophies to create

high conformance production environments (e.g., 4𝜎 - 6,210
DPMOs).
 Six Sigma

Decay of six Sigma

• Six Sigma interest has declined

Traditional quality philosophies

have plateaued, and they
have stagnated with little innovation
to offer to the manufacturing
industry.
Source: Avigdor Zonnenshain and Ron S Kenett. Quality 4.0—the challenging future of quality engineering. Quality
Engineering, 32(4):614–626, 2020.

Source: The Decay of Six Sigma and the Rise of Quality 4.0 in Manufacturing
Innovation. Escobar et al. Submitted to the Journal of Quality Engineering, 2022.
 AI innovation
Quality control is the most cited priority
for the application of this technology in manufacturing

16
AI innovation

Research
⎻ AIand computing revolutions will accelerate discoveries up
to 10 times faster and with 10 times less cost, IBM
⎻ AI-powered systems to replace manual quality inspection
processes
▪ Enabling over-burdened and fatigued inspectors to do more in less
time
⎻ Defect-free processes
Sources: https://research.ibm.com/blog/what-is-accelerated-discovery
https://darwinai.com/how-ai-powered-inspection-can-bring-manufacturing-back-to-america/
Escobar, Carlos, Jorge Arinez, and Ruben Morales-Menendez. Process-monitoring-for-quality-a step forward in the zero defects
vision. No. 2020-01-1302. SAE technical paper, 2020.
AI innovation

Vision systems
⎻ Surpassed human capabilities in 2015

Source: https://www.pfmjournal.org/journal/Figure.php?xn=pfm-2018-00030.xml&id=f1-pfm-2018-00030&number=32&p_name=0752_32
Binary Classification of Quality

19
Binary classification of quality

Geometric problem
• Use product/process quality characteristics (features)
to project manufacturing items into a hyper-dimensional
space where the classes are separated
• MLA, e.g., Support Vector Machine (SVM), finds the
optimal separating hyperplane
Binary classification of quality

Separating
hyperplane
 Binary classification of quality
Patterns tend to exist in hyper-
dimensional spaces
• No single feature separates the classes
• Feature 1 and 2 cannot separate the
classes
• Same situation if features 1 and 3, and
2 and 3 are combined.
• Classes cannot be separated in one- or
two-dimensional spaces
Binary classification of quality

Confusion Matrix
• Table with two rows and two columns that summarizes
the prediction ability of a classifier Reduce efficiency
(reevaluation/rework – hidden factory)

Generate warranty events Main concern

• From a statistical perspective

o FPs refer to the alpha error (α)
o FNs refer to the beta error (β)
Quality 4.0

24
 Quality 4.0

Definition

Sources: Escobar, Carlos A., Megan E. McGovern, and Ruben Morales-Menendez. "Quality 4.0: a review of big data challenges in manufacturing." Journal of Intelligent Manufacturing (2021): 1-16.
Quality 4.0
Problem
Solving
Areas of knowledge PSS
Strategy

Source: Escobar, Carlos A., et al. "Quality 4.0—Green, Black and Master Black Belt Curricula." Procedia Manufacturing 53 (2021): 748-759.
Quality 4.0

2020 survey findings

⎻ 66% of the respondents believe that Quality 4.0 will
significantly affect their operations within five years.
⎻ 16% have started to implement a Quality 4.0 initiative.
⎻ 20% respondents have started to plan for implementation.
⎻ 63% respondents have not even reached the planning stage
yet.
⎻ zero respondents have done a full implementation
Sources: Quality 4.0 survey by BCG, ASQ, and DGQ, https://asq.org/quality-resources/quality-4-0#Study
 Quality 4.0

End–to–end implementation

Sources: https://www.emerald.com/insight/2040-4166.htm
Quality 4.0
Competencies

Source: Escobar, Carlos A., et al. "Quality 4.0—Green, Black and Master Black Belt Curricula." Procedia Manufacturing 53 (2021): 748-759.
Learning quality control (LQC)

30
Learning quality control
Conceptual framework
• Monitoring systems aimed at
real-time defect detection
⎻ Binary classification problem
• Empirical knowledge discovery
⎻ Process
redesign and
improvement
⎻ Augment human intelligence
(boost troubleshooting and root-
case-analyses)
Learning quality control

Quality 4.0 into the 4th industrial revolution context

4th Industrial Revolution

Smart
Manufacturing
Quality 4.0

LQC
 Learning quality control
Defective Rework/
Application Scrap
(TP) revaluation Good
(TN)

• Improve automation Defective

items (TP,FP)
• Reduces manual
inspections Manufactured Good
Process Good
items (good,defective) LQC
PMQ
items (TN,FN)
CustomersWorn-out
(TN)
• Enables smart Customers
systems and Highly Predictive
A few
DPMO Defect
unbalanced Main
decisions model
(classifier)
(FN)
concern

• Boosts traditional Observational data

(IBD)
Machine
learning
Warranty

quality control Historical data

methods Offline learning Online deployment

 Learning quality control
Evolved Problem-Solving Strategy

PSS

Source: Escobar, Carlos A., Megan E. McGovern, and Ruben Morales-Menendez. "Quality 4.0: a review of big data challenges in manufacturing." Journal of Intelligent Manufacturing (2021): 1-16.
 Learning quality control
Evolved Problem-Solving Strategy
⎻ Guides the implementation of LQC systems
⎻ Addresses the main challenges of the application of AI for QC
▪ Transient sources of variation
▪ Hyperdimensional spaces
▪ Highly unbalanced classes
⎻ Increases the chances of success
⎻ Helps to select the right projects
⎻ Systematic way of driving innovation
Source: Escobar, Carlos A., Megan E. McGovern, and Ruben Morales-Menendez. "Quality 4.0: a review of big data challenges in manufacturing." Journal of Intelligent Manufacturing (2021): 1-16.
Case study

36
Case study
⎻ 10 samples
⎻ Three features
⎻ Four defects
⎻ Separate in three dimensions
Case study

⎻ Univariate control charts

⎻ No out of specifications detected
Case study

⎻ Multivariate control charts

⎻ One FP each
Six Sigma vs Quality 4.0

40
Six Sigma vs Quality 4.0

• Artificial Neural Networks, it has shown better predictive

modeling and optimization performance than that of RSM.
• Curse of dimensionality, most traditional methods can be
highly affected by high-dimensional data.
• Computation time, most traditional statistical methods were
not developed considering this challenge.
Six Sigma vs Quality 4.0

• Vision systems, deep learning was is not part of the six Sigma
quality tools. Visual/manual inspections still prevail.
• Control charts, control charts have shown significant
limitations in detecting patterns in hyperdimensional spaces.
• Problem solving strategy, six Sigma DMAIC methodology
does not cover all the steps required to develop an ML-based
solution.
Six Sigma vs Quality 4.0

Differences
Take-aways

44
Take-aways

⎻ First
companies to successfully deploy Q4.0 will
competitively position themselves among the most
advanced and influential companies in the world
⎻ Founding techniques and paradigms of Six Sigma are not
up to the task of handling the challenges posed by smart
manufacturing
⎻ Although six Sigma innovation capacity has plateaued, it is
still necessary for engineers to be trained in this
methodology
Take-aways

⎻ Research community should be free from Six Sigma and

should stop trying to fit the ML cycle into the DMAIC
problem-solving strategy
⎻ Quality engineers trained in Q4.0 are in a good position to
take the lead again because of their problem-solving
abilities, business acumen, and strong quantitative
reasoning
 Questions?

Carlos Escobar– Phone: 248-805-5467 – Email: [email protected]