The document outlines an agenda for a workshop on digital twin technology from a manufacturing quality perspective. The workshop will discuss using CAE simulation and artificial intelligence to create digital twins for closed-loop in-process quality improvement. Speakers will discuss applications in automotive and aerospace manufacturing and how digital twin technology can shorten lead times and ramp-up times by improving quality.

1. Presented by: Darek Ceglarek
WMG, University of Warwick, Coventry, UK
Authors: Pasquale Franciosa, Sumit Sinha, Manoj Babu, Emile Glorieux, Darek Ceglarek
4th February 2020 | Professor Lord Bhattacharya Building (NAIC), WMG | University of Warwick, Coventry, CV4 7AL
DIGITAL TWIN TECHNOLOGY
- A MANUFACTURING QUALITY PERSPECTIVE -
Get Connected: When CAE Simulation Meets Artificial Intelligence
Deep Learning Enhanced Digital Twin for Closed-loop In-Process Quality Improvement

2. REGISTRATION AND REFRESHMENTS
• Exhibition, demonstrations and engineering games
Welcome and opening remarks
• Dr David Bott, Principal Fellow, WMG
Digital twin technology – a manufacturing perspective
• Professor Darek Ceglarek, WMG
Academic and industry research collaboration – the Mathworks suite
• Bulat Khusainov, Technical Specialist, Mathworks
BREAK AND REFRESHMENTS
• Exhibition, demonstrations and engineering games
From concept to production implementation – the Remote Laser
Welding (RLW) process for aluminium door assembly
• Marcus Henry, Structures Research Manager, Jaguar Land Rover
• Sujit Chatterjee, Lead Engineer, AME, Jaguar Land Rover
LUNCH AND NETWORKING
• Exhibition, demonstrations and engineering games
09:30
10:00
10:15
10:45
11:15
11:30
12:00
Mathworks CAE simulation suite – case in point from
automotive and aerospace
• Bulat Khusainov, Technical Specialist, Mathworks
‘Variation Response Method’ CAE simulation suite – What is it?
• Dr Pasquale Franciosa, Associate Professor, WMG
BREAK AND REFRESHMENTS
• Exhibition, demonstrations and engineering games
‘Variation Response Method’ CAE simulation suite – Generating
training data for deep learning
• Dr Pasquale Franciosa, Associate Professor, WMG
Deep learning in manufacturing – predicting and preventing
manufacturing defects
• Sumit Sinha, PhD student, WMG
Wrap up and close
• Professor Darek Ceglarek, WMG
• Dr David Bott, Principal Fellow, WMG
END
13:00
13:45
14:15
14:30
15:30
16:15
16:30
A G E N D A
Get Connected: When CAE Simulation Meets Artificial Intelligence
4th February 2020 | Professor Lord Bhattacharya Building (NAIC), WMG | University of Warwick, Coventry, CV4 7AL
TECHNICAL SESSIONS

3. WMG @ University of Warwick, United Kingdom
WMG – an independent interdisciplinary academic department equivalent to Industrial & Systems Engg.
o An academic department of the University of Warwick
o Employing over 600 staff
o Working across research and education centres on the
Warwick campus
o Delivering education programmes in seven countries, and
collaborating globally on research and development
o An annual programme of £200m (industrial & in-kind
support)
o Strong relationships with over 1,000 global companies,
and supporting 1,800 SMEs
o Part of the HVM Catapult network HVM Catapult - Supports RTD &
technology maturation from concept
(TRL 3-4) to pre-production (TRL 6-7)

4. WMG @ University of Warwick, United Kingdom
New Development - ‘National Automotive Innovative Campus’ (NAIC)
• £150m investment - the biggest industrial & private sector investment in any UK university
(2017)
• A 33,000m2 collaborative research environment
• Unique infrastructure & national focus for R&D addressing UK Automotive Council agenda
• Creating 1,000 & attracting 3,000 further R&D jobs
• Bringing global R&D of major Tier 1 suppliers to co-locate
Supporting RTD and technology maturation from concept (TRL 3-4) to pre-
production (TRL 6-7)

5. DIGITAL LIFECYCLE
MANAGEMENT (DLM)
DLM research & technology development provides a transformative framework integrating Industry 4.0 enablers such as big data, cloud computing, robotics and AI:
- To create precise digital twins with capabilities to integrate data from interconnected systems with multi-disciplinary simulation of products & processes.
- To achieve resilient performance in production systems such as predict behaviour of process machinery to either optimise existing equipment for increased productivity & quality; or
introduce new equipment into an assembly line with minimum disruption.
DEEP LEARNING ENHANCED DIGITAL TWIN FOR
CLOSED-LOOP IN-PROCESS QUALITY CONTROL & IMPROVEMENT

6. DIGITAL LIFECYCLE
MANAGEMENT (DLM)
DLM research & technology development provides a transformative framework integrating Industry 4.0 enablers such as big data, cloud computing, robotics and AI:
- To create precise digital twins with capabilities to integrate data from interconnected systems with multi-disciplinary simulation of products & processes.
- To achieve resilient performance in production systems such as predict behaviour of process machinery to either optimise existing equipment for increased productivity & quality; or
introduce new equipment into an assembly line with minimum disruption.
DEEP LEARNING ENHANCED DIGITAL TWIN FOR
CLOSED-LOOP IN-PROCESS QUALITY CONTROL & IMPROVEMENT
GOAL
Towards Near-Zero Defects
Challenges:
Cell design with capabilities to simulate
‘production parts’
Quality Improvement with capabilities for
quality defects Root Cause Analysis (RCA)
Quality Defects
Data
Manufacturing
Closed-loop in-process (CLIP) quality improvement
CAE Simulations
Variation Response Method (VRM)
Artificial Intelligence
Deep learning for Manufacturing
Takes as input a set of
control parameters
Gives as output a
system response
Constraints such as
specification limits
System
Control parameters
have to be estimated
Given the output is
known
System
Constraints such as
specification limits
Design Simulation tools
(1) To support system optimisation
(2) To improve process capability
(3) Variation reduction
Quality Improvement tools
(1) To support system optimisation
(2) Intelligent root cause analysis of defects
(3) Preventive control actions

7. DIGITAL LIFECYCLE
MANAGEMENT (DLM)
For more information: Prof. Darek Ceglarek; Prof. Pasquale Franciosa, {d.j.Ceglarek, p.Franciosa}@warwick.ac.uk
DLM research & technology development provides a transformative framework integrating Industry 4.0 enablers such as big data, cloud computing, robotics and AI:
- To create precise digital twins with capabilities to integrate data from interconnected systems with multi-disciplinary simulation of products & processes.
- To achieve resilient performance in production systems such as predict behaviour of process machinery to either optimise existing equipment for increased productivity & quality; or
introduce new equipment into an assembly line with minimum disruption.
Business value
• Shorten lead time by.
• Increasing Right-First-Time
• Rapid deployment of new process
technologies.
• Reduction of installation and commissioning
time & cost.
• Shorten ramp-up time by.
• Shortening Mean-Time-to-Detection (MTTD)
of quality defects
• Shortening Mean-Time-to-Resolution
(MTTR) ) of quality defects
• Improved quality (6-sigma)
• Applications: automotive,
aerospace, consumer goods &
others.
What we do? Framework
Digital Twin for rapid scaling-up & deployment of new technology
AI tools
DEEP LEARNING ENHANCED DIGITAL TWIN FOR
CLOSED-LOOP IN-PROCESS QUALITY CONTROL & IMPROVEMENT
Quality in
closed-loop
Monitoring
in closed-loop
Smart Connected FactoryProduct & Process
Design
Products in Field
Goal:
Towards Zero-defects
Objective 2: Real-time optimization of
quality & operations performance
Objective 1: ‘Right First Time’
Towards Resilient System
Simulations
in closed-loop
Enterprise Analytics
In-use dataQuality Data
2
3
1
Device analytics
In-line monitoring
Digital
Lifecycle
Management
Multi-disciplinary
Defect Simulator
System
Configurator
RCA / CAPA

8. DIGITAL LIFECYCLE
MANAGEMENT (DLM)
For more information: Prof. Darek Ceglarek; Prof. Pasquale Franciosa, {d.j.Ceglarek, p.Franciosa}@warwick.ac.uk
DLM research & technology development provides a transformative framework integrating Industry 4.0 enablers such as big data, cloud computing, robotics and AI:
- To create precise digital twins with capabilities to integrate data from interconnected systems with multi-disciplinary simulation of products & processes.
- To achieve resilient performance in production systems such as predict behaviour of process machinery to either optimise existing equipment for increased productivity & quality; or
introduce new equipment into an assembly line with minimum disruption.
Business value
• Shorten lead time by.
• Increasing Right-First-Time
• Rapid deployment of new process
technologies.
• Reduction of installation and commissioning
time & cost.
• Shorten ramp-up time by.
• Shortening Mean-Time-to-Detection (MTTD)
of quality defects
• Shortening Mean-Time-to-Resolution
(MTTR) ) of quality defects
• Improved quality (6-sigma)
• Applications: automotive,
aerospace, consumer goods &
others.
• Configuration, optimization & control of assembly process with
deformable parts from design-to-production .
• Digital twin with embedded CAE simulations
• Multi-disciplinary optimization of system performance & configuration
Simulation
in closed-loop1
Monitoring
in closed-loop2
Quality
in closed-loop3
What we do? Framework
Digital Twin for rapid scaling-up & deployment of new technology
AI tools
DEEP LEARNING ENHANCED DIGITAL TWIN FOR
CLOSED-LOOP IN-PROCESS QUALITY CONTROL & IMPROVEMENT
Quality in
closed-loop
Monitoring
in closed-loop
Smart Connected FactoryProduct & Process
Design
Products in Field
Goal:
Towards Zero-defects
Objective 2: Real-time optimization of
quality & operations performance
Objective 1: ‘Right First Time’
Towards Resilient System
Simulations
in closed-loop
Enterprise Analytics
In-use dataQuality Data
2
3
1
Device analytics
In-line monitoring
Digital
Lifecycle
Management
Multi-disciplinary
Defect Simulator
System
Configurator
RCA / CAPA

9. DIGITAL LIFECYCLE
MANAGEMENT (DLM)
For more information: Prof. Darek Ceglarek; Prof. Pasquale Franciosa, {d.j.Ceglarek, p.Franciosa}@warwick.ac.uk
• In-line & In-process quality monitoring to accelerate RCA –incl. off-line
& in-line programming of robotic 3D scanner;
• In-use data from intelligent assets & 3rd party devices.
• Statistical Quality Control for Point-Clouds & big data.
DLM research & technology development provides a transformative framework integrating Industry 4.0 enablers such as big data, cloud computing, robotics and AI:
- To create precise digital twins with capabilities to integrate data from interconnected systems with multi-disciplinary simulation of products & processes.
- To achieve resilient performance in production systems such as predict behaviour of process machinery to either optimise existing equipment for increased productivity & quality; or
introduce new equipment into an assembly line with minimum disruption.
Business value
• Shorten lead time by.
• Increasing Right-First-Time
• Rapid deployment of new process
technologies.
• Reduction of installation and commissioning
time & cost.
• Shorten ramp-up time by.
• Shortening Mean-Time-to-Detection (MTTD)
of quality defects
• Shortening Mean-Time-to-Resolution
(MTTR) ) of quality defects
• Improved quality (6-sigma)
• Applications: automotive,
aerospace, consumer goods &
others.
• Configuration, optimization & control of assembly process with
deformable parts from design-to-production .
• Digital twin with embedded CAE simulations
• Multi-disciplinary optimization of system performance & configuration
Simulation
in closed-loop1
Monitoring
in closed-loop2
Quality
in closed-loop3
What we do? Framework
Digital Twin for rapid scaling-up & deployment of new technology
AI tools
DEEP LEARNING ENHANCED DIGITAL TWIN FOR
CLOSED-LOOP IN-PROCESS QUALITY CONTROL & IMPROVEMENT
Quality in
closed-loop
Monitoring
in closed-loop
Smart Connected FactoryProduct & Process
Design
Products in Field
Goal:
Towards Zero-defects
Objective 2: Real-time optimization of
quality & operations performance
Objective 1: ‘Right First Time’
Towards Resilient System
Simulations
in closed-loop
Enterprise Analytics
In-use dataQuality Data
2
3
1
Device analytics
In-line monitoring
Digital
Lifecycle
Management
Multi-disciplinary
Defect Simulator
System
Configurator
RCA / CAPA

10. DIGITAL LIFECYCLE
MANAGEMENT (DLM)
For more information: Prof. Darek Ceglarek; Prof. Pasquale Franciosa, {d.j.Ceglarek, p.Franciosa}@warwick.ac.uk
• In-line & In-process quality monitoring to accelerate RCA –incl. off-line
& in-line programming of robotic 3D scanner;
• In-use data from intelligent assets & 3rd party devices.
• Statistical Quality Control for Point-Clouds & big data.
DLM research & technology development provides a transformative framework integrating Industry 4.0 enablers such as big data, cloud computing, robotics and AI:
- To create precise digital twins with capabilities to integrate data from interconnected systems with multi-disciplinary simulation of products & processes.
- To achieve resilient performance in production systems such as predict behaviour of process machinery to either optimise existing equipment for increased productivity & quality; or
introduce new equipment into an assembly line with minimum disruption.
Business value
• Shorten lead time by.
• Increasing Right-First-Time
• Rapid deployment of new process
technologies.
• Reduction of installation and commissioning
time & cost.
• Shorten ramp-up time by.
• Shortening Mean-Time-to-Detection (MTTD)
of quality defects
• Shortening Mean-Time-to-Resolution
(MTTR) ) of quality defects
• Improved quality (6-sigma)
• Applications: automotive,
aerospace, consumer goods &
others.
• Data-driven Root Cause Analysis (RCA) of quality (6-sigma & No-Fault-
Found) defects using in-line and in-process data.
• RCA-driven Corrective & Preventive Actions (CAPA).
• CAPA-driven continuous improvement.
• Multi-physics simulator of manufacturing defects.
• Configuration, optimization & control of assembly process with
deformable parts from design-to-production .
• Digital twin with embedded CAE simulations
• Multi-disciplinary optimization of system performance & configuration
Simulation
in closed-loop1
Monitoring
in closed-loop2
Quality
in closed-loop3
What we do? Framework
Digital Twin for rapid scaling-up & deployment of new technology
AI tools
DEEP LEARNING ENHANCED DIGITAL TWIN FOR
CLOSED-LOOP IN-PROCESS QUALITY CONTROL & IMPROVEMENT
Quality in
closed-loop
Monitoring
in closed-loop
Smart Connected FactoryProduct & Process
Design
Products in Field
Goal:
Towards Zero-defects
Objective 2: Real-time optimization of
quality & operations performance
Objective 1: ‘Right First Time’
Towards Resilient System
Simulations
in closed-loop
Enterprise Analytics
In-use dataQuality Data
2
3
1
Device analytics
In-line monitoring
Digital
Lifecycle
Management
Multi-disciplinary
Defect Simulator
System
Configurator
RCA / CAPA

11. DIGITAL LIFECYCLE
MANAGEMENT (DLM)
For more information: Prof. Darek Ceglarek; Prof. Pasquale Franciosa, {d.j.Ceglarek, p.Franciosa}@warwick.ac.uk
• In-line & In-process quality monitoring to accelerate RCA –incl. off-line
& in-line programming of robotic 3D scanner;
• In-use data from intelligent assets & 3rd party devices.
• Statistical Quality Control for Point-Clouds & big data.
DLM research & technology development provides a transformative framework integrating Industry 4.0 enablers such as big data, cloud computing, robotics and AI:
- To create precise digital twins with capabilities to integrate data from interconnected systems with multi-disciplinary simulation of products & processes.
- To achieve resilient performance in production systems such as predict behaviour of process machinery to either optimise existing equipment for increased productivity & quality; or
introduce new equipment into an assembly line with minimum disruption.
Business value
• Shorten lead time by.
• Increasing Right-First-Time
• Rapid deployment of new process
technologies.
• Reduction of installation and commissioning
time & cost.
• Shorten ramp-up time by.
• Shortening Mean-Time-to-Detection (MTTD)
of quality defects
• Shortening Mean-Time-to-Resolution
(MTTR) ) of quality defects
• Improved quality (6-sigma)
• Applications: automotive,
aerospace, consumer goods &
others.
• Data-driven Root Cause Analysis (RCA) of quality (6-sigma & No-Fault-
Found) defects using in-line and in-process data.
• RCA-driven Corrective & Preventive Actions (CAPA).
• CAPA-driven continuous improvement.
• Multi-physics simulator of manufacturing defects.
• Configuration, optimization & control of assembly process with
deformable parts from design-to-production .
• Digital twin with embedded CAE simulations
• Multi-disciplinary optimization of system performance & configuration
Simulation
in closed-loop1
Monitoring
in closed-loop2
Quality
in closed-loop3
What we do? Examples
Rapid scaling-up & deployment of new technology
CAE Simulators embedded
into digital twin & VR
ConvolutionalNeuralNets
Reinforcementlearning
AI tools
Fuzzy models
Spatio-temporalmodels
Genetics Algorithms
Polynomial Chaos
Rule-based systems
Digital twin
& VR (TRL 3-4)
Lab demo
(TRL 6)
Pre-production
pilot trials
(TRL7)
DEEP LEARNING ENHANCED DIGITAL TWIN FOR
CLOSED-LOOP IN-PROCESS QUALITY CONTROL & IMPROVEMENT
Scale-up simulator &
In-process/inline monitoring
RCA and CAPA
simulator and controller

12. DIGITAL LIFECYCLE
MANAGEMENT (DLM)
For more information: Prof. Darek Ceglarek; Prof. Pasquale Franciosa, {d.j.Ceglarek, p.Franciosa}@warwick.ac.uk
• In-line & In-process quality monitoring to accelerate RCA –incl. off-line
& in-line programming of robotic 3D scanner;
• In-use data from intelligent assets & 3rd party devices.
• Statistical Quality Control for Point-Clouds & big data.
DLM research & technology development provides a transformative framework integrating Industry 4.0 enablers such as big data, cloud computing, robotics and AI:
- To create precise digital twins with capabilities to integrate data from interconnected systems with multi-disciplinary simulation of products & processes.
- To achieve resilient performance in production systems such as predict behaviour of process machinery to either optimise existing equipment for increased productivity & quality; or
introduce new equipment into an assembly line with minimum disruption.
Business value
• Shorten lead time by.
• Increasing Right-First-Time
• Rapid deployment of new process
technologies.
• Reduction of installation and commissioning
time & cost.
• Shorten ramp-up time by.
• Shortening Mean-Time-to-Detection (MTTD)
of quality defects
• Shortening Mean-Time-to-Resolution
(MTTR) ) of quality defects
• Improved quality (6-sigma)
• Applications: automotive,
aerospace, consumer goods &
others.
• Data-driven Root Cause Analysis (RCA) of quality (6-sigma & No-Fault-
Found) defects using in-line and in-process data.
• RCA-driven Corrective & Preventive Actions (CAPA).
• CAPA-driven continuous improvement.
• Multi-physics simulator of manufacturing defects.
• Configuration, optimization & control of assembly process with
deformable parts from design-to-production .
• Digital twin with embedded CAE simulations
• Multi-disciplinary optimization of system performance & configuration
Simulation
in closed-loop1
Monitoring
in closed-loop2
Quality
in closed-loop3
What we do? Examples
Laser WeldingRobotic
AI for rapid scaling-up & deployment of new technology
Fixture
CAE Simulators embedded into digital twin & VR
AI tools
Genetics Algorithms
Polynomial Chaos
Rule-based systems
Digital twin
& VR (TRL 3-4)
DEEP LEARNING ENHANCED DIGITAL TWIN FOR
CLOSED-LOOP IN-PROCESS QUALITY CONTROL & IMPROVEMENT
Lab demo
(TRL 6)
Pre-production
pilot trials
(TRL7)
Scale-up simulator &
In-process/inline monitoring
RCA and CAPA
simulator and controller

13. DIGITAL LIFECYCLE
MANAGEMENT (DLM)
For more information: Prof. Darek Ceglarek; Prof. Pasquale Franciosa, {d.j.Ceglarek, p.Franciosa}@warwick.ac.uk
• In-line & In-process quality monitoring to accelerate RCA –incl. off-line
& in-line programming of robotic 3D scanner;
• In-use data from intelligent assets & 3rd party devices.
• Statistical Quality Control for Point-Clouds & big data.
DLM research & technology development provides a transformative framework integrating Industry 4.0 enablers such as big data, cloud computing, robotics and AI:
- To create precise digital twins with capabilities to integrate data from interconnected systems with multi-disciplinary simulation of products & processes.
- To achieve resilient performance in production systems such as predict behaviour of process machinery to either optimise existing equipment for increased productivity & quality; or
introduce new equipment into an assembly line with minimum disruption.
Business value
• Shorten lead time by.
• Increasing Right-First-Time
• Rapid deployment of new process
technologies.
• Reduction of installation and commissioning
time & cost.
• Shorten ramp-up time by.
• Shortening Mean-Time-to-Detection (MTTD)
of quality defects
• Shortening Mean-Time-to-Resolution
(MTTR) ) of quality defects
• Improved quality (6-sigma)
• Applications: automotive,
aerospace, consumer goods &
others.
• Data-driven Root Cause Analysis (RCA) of quality (6-sigma & No-Fault-
Found) defects using in-line and in-process data.
• RCA-driven Corrective & Preventive Actions (CAPA).
• CAPA-driven continuous improvement.
• Multi-physics simulator of manufacturing defects.
• Configuration, optimization & control of assembly process with
deformable parts from design-to-production .
• Digital twin with embedded CAE simulations
• Multi-disciplinary optimization of system performance & configuration
Simulation
in closed-loop1
Monitoring
in closed-loop2
Quality
in closed-loop3
What we do? Examples
Laser WeldingRobotic
AI for rapid scaling-up & deployment of new technology
Fixture
CAE Simulators embedded into digital twin & VR
In-lineIn-processScale-up Sim
Scale-up simulator & In-process/inline monitoring
AI tools
Fuzzy models
Spatio-temporalmodels
Genetics Algorithms
Polynomial Chaos
Rule-based systems
Digital twin
& VR (TRL 3-4)
Lab demonstrator (TRL6)
System tested on ~80 parts
DEEP LEARNING ENHANCED DIGITAL TWIN FOR
CLOSED-LOOP IN-PROCESS QUALITY CONTROL & IMPROVEMENT
Pre-production
pilot trials
(TRL7)
RCA and CAPA
simulator and controller

14. DIGITAL LIFECYCLE
MANAGEMENT (DLM)
For more information: Prof. Darek Ceglarek; Prof. Pasquale Franciosa, {d.j.Ceglarek, p.Franciosa}@warwick.ac.uk
• In-line & In-process quality monitoring to accelerate RCA –incl. off-line
& in-line programming of robotic 3D scanner;
• In-use data from intelligent assets & 3rd party devices.
• Statistical Quality Control for Point-Clouds & big data.
DLM research & technology development provides a transformative framework integrating Industry 4.0 enablers such as big data, cloud computing, robotics and AI:
- To create precise digital twins with capabilities to integrate data from interconnected systems with multi-disciplinary simulation of products & processes.
- To achieve resilient performance in production systems such as predict behaviour of process machinery to either optimise existing equipment for increased productivity & quality; or
introduce new equipment into an assembly line with minimum disruption.
Business value
• Shorten lead time by.
• Increasing Right-First-Time
• Rapid deployment of new process
technologies.
• Reduction of installation and commissioning
time & cost.
• Shorten ramp-up time by.
• Shortening Mean-Time-to-Detection (MTTD)
of quality defects
• Shortening Mean-Time-to-Resolution
(MTTR) ) of quality defects
• Improved quality (6-sigma)
• Applications: automotive,
aerospace, consumer goods &
others.
• Data-driven Root Cause Analysis (RCA) of quality (6-sigma & No-Fault-
Found) defects using in-line and in-process data.
• RCA-driven Corrective & Preventive Actions (CAPA).
• CAPA-driven continuous improvement.
• Multi-physics simulator of manufacturing defects.
• Configuration, optimization & control of assembly process with
deformable parts from design-to-production .
• Digital twin with embedded CAE simulations
• Multi-disciplinary optimization of system performance & configuration
Simulation
in closed-loop1
Monitoring
in closed-loop2
Quality
in closed-loop3
ARTIFICIAL INTELLIGENCE (AI) APPLICATIONS IN MANUFACTURING:
CLOSED-LOOP IN-PROCESS QUALITY CONTROL & IMPROVEMENT
What we do? Examples
Laser WeldingRobotic
AI for rapid scaling-up & deployment of new technology
Fixture
CAE Simulators embedded into digital twin & VR
In-lineIn-processScale-up Sim
Scale-up simulator & In-process/inline monitoring
CAPARCAFault ID
RCA and CAPA simulator and controller
ConvolutionalNeuralNets
Reinforcementlearning
2nd shift1st shiftLaunchPre-Launch
6-sigmaQuality
New production Launch
AI tools
Fuzzy models
Spatio-temporalmodels
Genetics Algorithms
Polynomial Chaos
Rule-based systems
2018 Innovista Award from
JLR for the Most
Innovative Piloted
Technology
Digital twin
& VR (TRL 3-4)
Lab demonstrator (TRL6)
System tested on ~80 parts
Pre-production pilot trials (TRL7)
System tested on ~3000 parts

15. Laser Welding
process parameters
Off-line
programming of
robotic joining and
handling
Fixture/jigs
Simulations
in closed-loop1
Determining optimum fixture design for a batch of non-ideal parts
Video explaining design optimization posted on the Youtube – RLW Navigator
Closed-loop In-Process Quality Improvement
Simulations in Closed-loop

16. Laser Welding
process parameters
Off-line
programming of
robotic joining and
handling
Fixture/jigs
Simulations
in closed-loop1
Gap= ~0.0 mm Gap= ~0.4 mm Gap= ~0.8 mm
AA5182; 1.5 mm to 2.2 mm; 4 m/min; 3.8kW (avg power)
Automatic process parameters selection for part-to-part gap bridging
Determine optimum laser parameter selection
What does it do?
(1) Definition of optimum process
parameters(i.e., power, speed), based
on defined output criteria (weld
quality, cycle time, etc.)
(2) Automatic identification of feasible
process
windows
Video explaining design optimization posted on the Youtube – RLW Navigator
Closed-loop In-Process Quality Improvement
Simulations in Closed-loop

17. Laser Welding
process parameters
Off-line
programming of
robotic joining and
handling
Fixture/jigs
Simulations
in closed-loop1
SCALABILITY: Capability for Scaling process parameters for varying welding speed
Gap= ~0.0 mm Gap= ~0.4 mm Gap= ~0.8 mm
AA5182; 1.5 mm to 2.2 mm; 4 m/min; 3.8kW (avg power)
ROBUSTNESS: Automatic process parameters selection for part-to-part
gap bridging
Closed-loop In-Process Quality Improvement
Simulations in Closed-loop

18. Industry challenges:
• Multi-disciplinary ‘simulation analytics’ to
configure, optimize & control assembly
process from design-to-production.
• Platform to integrate both product & process
models with measurement data.
• Applications: automotive, aerospace.
What we do:
• Configuration, optimization & control of assembly process
with deformable parts from design-to-production.
• In-line monitoring to accelerate root cause analysis (RCA).
• In-use data from intelligent assets & 3rd party devices.
• Data-driven Root Cause Analysis (RCA).
• RCA-driven Corrective & Preventive Actions (CAPA).
• CAPA-driven continuous improvement.
• Real-time optimization of machines, processes & plants.
• Machine learning for predictive capabilities.
Business Value:
• Improved interaction between
product and process design.
• Reduced engineering cost by
30%.
• Improved dimensional quality
20%.
• Reduction of installation,
commissioning and time to
launch.
• Rapid deployment of current or
new process (Remote Laser
Welding, RLW).
Current Industrial best-practice
‘Right-First-Time’ Design of RLW: Application to SUV door assembly
process - eliminated more than 100 engineering changes which
affected a total of 22 tooling elements or stiches.
Improvement obtained
using the developed “multi-
disciplinarydefect
simulator”
Laser Welding
process parameters
Off-line
programming of
robotic joining and
handling
Fixture/jigs
Simulations
in closed-loop1
Video explaining design optimization posted on the Youtube – RLW Navigator
Closed-loop In-Process Quality Improvement
Simulations in Closed-loop

19. Quality
in closed-loop3
What is it:
• No-Fault-Found (NFF), a term for when faults
are known to exist yet evade efforts to
identify them - a pervasive problem in high
tech
• Developed data analytics for NFF root cause
analysis called Fault Region Localization (FRL)
How is it unique:
• The developed NFF analytics for NFF root cause isolation
helped in elimination of a top-5 warranty problems in
one mobile phone OEM.
• The NFF analytics utilizes heterogeneous data: (i)
warranty & (ii) manufacturing to identify both candidate
KCs explaining the NFF and also determine their fault
and normal conditions windows .
Business Value:
• Airlines voted NFF "the most
important issue" by the annual
Avionics Maintenance Conference
in 2004. It continues to be among
the top-most items AMC discusses
each year, but there are no
systematic root cause analysis
approaches.
• Mobile phone: 1 in 7 phones are
returned by customers – 60% have
NFF (WDSGlobal Report, 2006).
Application:
No-fault found (NFF) root cause analysis at Motorola IDEN mobile phone
manufacturing :
NFFBattery
NFFAudio
Display
Poweroff
TxDefect
Warranty Faults
Importance
Assembly process
Top-5 warranty faults
IDEN mobile phobne
Results of the FRL Analytics
for NFF Battery failure
• Number of Variable: 170
• No. of Normal Sample: 1500 (as ‘*’)
• Faulty Samples: 20 (as ‘*’)
KC2:SpecsRegion
Key Characteristics (KC1): Specs Region
No-Fault-Fault
Region (NFFR)
Boundary
Region (BR)
Normal Region =
Specs Region – BR – NFFR
Closed-loop In-Process Quality Improvement
Quality in Closed-loop

20. OUR RTD JOURNEY
RTD JOURNEY, IMPACT AND ACKNOWLEDGMENTS
TRL 4-6
TRL 5-7-8
TRL 2-4
Demo+Pilot
RLW for Al
CAE Solutions:
In-process closed-loop
process quality control
TRL 2-4
CAE Solutions:
assembly simulations
(non-ideal compliant parts
Solutions+Demo
RLW for Steel
IR cell @Solihull (tests on ~3000 partsAR cell @WMG (tests on ~80 parts)
Up-scaling
In-line QC (GD&T) In-process QC (weld)
VR with embedded CAE simulation

21. Engineering Game
Illustration
New Applications of
3D Optical Scanner
-In-line Monitoring
BIW
RoofFrame w/ Cowl
Cowl LH Frame w/ cross MBRs Cowl LH
RR Cross MBRBIW Frame
RoofBowsLH Side Fr Comp RH Side FrameUB 3 Re-spot
SF Outer&Inner A-Pillar
SF Inner Assem
RR W/H Outer Fr Door Hinge SF Inner PanelThrough Side Fr. Outer
SF Outer Assem
T-Lamp Mount. SF Outer Sub
SF Outer Panel B-Pillar Inner RR Door Locker
Cowl Top UB Line 2
UB Line 2-0 Cowl Side
UB 1 Dash Fl Plenum
UB Line 1-1 Body Side Sills Dash Floor Plenum
RR Wheel House UB Line 1-0
RR Floor Ladder Front Rail Sub
Frt Floor Panel Dash Panel
Engine Compt 3 Frt W/H LH&RH
Engine Compt 2 Frt Low Frt Xmbr
Engine Compt 1 Frt Low RR Xmbr
Front Sill Hydroform Pnl
Multi-station Assembly Assembly stations
Laser Welding
process
parameters
Off-line
programming
of robots
Fixture/jigs
Fixture design for a batch of non-ideal parts
In-line 3D Optical Scanner Station
Metrology
Modelling
Robot
Simulation
1
2
1
,2,1,
,22221
,11211
1
2
1






































mmmnmnnn
m
m
nn KCC
KCC
KCC
ccc
ccc
ccc
KPC
KPC
KPC






MatrixSOVA
Fixture
Modelling
Point-clouds data
Key Product Characteristics (KPCs)
Root Cause Analysis
Isolate Defective KCC(s)
Key Control Characteristics (KCCs)
KCC 2KCC 1
KCC 3
KCC 4
KCC 5

22. Engineering Game
Illustration
New Applications of
3D Optical Scanner
-In-line Monitoring
BIW
RoofFrame w/ Cowl
Cowl LH Frame w/ cross MBRs Cowl LH
RR Cross MBRBIW Frame
RoofBowsLH Side Fr Comp RH Side FrameUB 3 Re-spot
SF Outer&Inner A-Pillar
SF Inner Assem
RR W/H Outer Fr Door Hinge SF Inner PanelThrough Side Fr. Outer
SF Outer Assem
T-Lamp Mount. SF Outer Sub
SF Outer Panel B-Pillar Inner RR Door Locker
Cowl Top UB Line 2
UB Line 2-0 Cowl Side
UB 1 Dash Fl Plenum
UB Line 1-1 Body Side Sills Dash Floor Plenum
RR Wheel House UB Line 1-0
RR Floor Ladder Front Rail Sub
Frt Floor Panel Dash Panel
Engine Compt 3 Frt W/H LH&RH
Engine Compt 2 Frt Low Frt Xmbr
Engine Compt 1 Frt Low RR Xmbr
Front Sill Hydroform Pnl
Multi-station Assembly Assembly stations
Laser Welding
process
parameters
Off-line
programming
of robots
Fixture/jigs
Fixture design for a batch of non-ideal parts
In-line 3D Optical Scanner Station
Metrology
Modelling
Robot
Simulation
1
2
1
,2,1,
,22221
,11211
1
2
1






































mmmnmnnn
m
m
nn KCC
KCC
KCC
ccc
ccc
ccc
KPC
KPC
KPC






MatrixSOVA
Fixture
Modelling
Point-clouds data
Key Product Characteristics (KPCs)
Root Cause Analysis
Isolate Defective KCC(s)
Key Control Characteristics (KCCs)
KCC 2KCC 1
KCC 3
KCC 4
KCC 5
CAESimulations
Variation Response Method (VRM)
Takes as input a set of
control parameters
Gives as output a
system response
Constraints such as
specification limits
System
Root Cause Analysis
Artificial Intelligence
Deep learning for Manufacturing
Control parameters
have to be estimated
Given the output is
known
System
Constraints such as
specification limits

23. Engineering Game
Life demo/competition
CAESimulations
Variation Response Method (VRM)
Takes as input a set of
control parameters
Gives as output a
system response
Constraints such as
specification limits
System
Root Cause Analysis
Artificial Intelligence
Deep learning for Manufacturing
Control parameters
have to be estimated
Given the output is
known
System
Constraints such as
specification limits

24. Name/Position Comments
Company/Division
Email/Telephone
DOES NOT EXIST
in my company
EXIST (not used)
in my company
INCREMENTAL
CHANGE
STEP
CHANGE
SIGNIFICANT
CHANGE
A. Presented software tools capabilities will provide for my company:
EXIST (limited use)
in my company
EXIST (common use)
in my company
(1) My company would be Interested
in the software tools as:
As ‘stand alone’ software
As ‘plug-in’ software
As ‘Integrated’ with software in my company
(2) Potential applications of interest:
• Automotive(please specify)
• Aerospace (please specify))
• Other (please specify)
Please mark in the table to the left your
response to the following two items:
A
Feedback
Simulation Tools: (1) Variation Response method (VRM); & (2) Deep learning in Manufacturing (DL in Mfg)

25. Name/Position Comments
Company/Division
Email/Telephone
Short term
(<1year)
Medium term
(<3years)
Long terms
( >3 years)
B. TIMELINE: I am interested in using the presented software tools:B
As ‘stand alone’ tool As ‘plug-in’ tool As ‘integrated’ tool
Feedback
Simulation Tools: (1) Variation Response method (VRM); & (2) Deep learning in Manufacturing (DL in Mfg)

26. Name/Position Comments
Company/Division
Email/Telephone
Short term
(<1year)
Medium term
(<3years)
Long terms
( >3 years)
As ‘stand alone’ tool As ‘plug-in’ tool As ‘integrated’ tool
Feedback
Simulation Tools: (1) Variation Response method (VRM); & (2) Deep learning in Manufacturing (DL in Mfg)
C. SKILLS and TRAINING Needs:B

27. MAIN SOFTWARE CONTRIBUTORS AND DEVELOPERS:
Pasquale Franciosa (Warwick University), Abhishek Das (Warwick University), Manoj Babu (Warwick University), Emile Glorieux (Warwick University), Sumit Sinha
(Warwick University), Darek Ceglarek (Warwick University), Avishek Pal (Microsoft), Gerardo Beruvides (Hitachi), Vahid Shahi (University of Tehran), Ferdinando
Vitolo (University of Naples Federico II), Salvatore Gerbino (University of Campania), Achille Pesce (FCA), Chris Esposito (University of Naples Federico II)
27
Thanks for your attention!
Any questions and/or comments?
Dr. Pasquale Franciosa
Prof. Darek Ceglarek
WMG, University of Warwick, UK
WMG, University of Warwick, UK
Email: p.franciosa@warwick.ac.uk / d.j.ceglarek@warwick.ac.uk
M: +44(0) 7440022523 / 44(0) 7824540721
T: +44(0)24765 73422 / 44(0) 24765 72681
Acknowledgements