Introduction: Data Science for Materials and

Manufacturing

Hongyi Xu
Mechanical Engineering & UTC Institute for
Advanced Systems Engineering
University of Connecticut

1
 Welcome!

• Course format: online course + live discussions

• Instructor: Hongyi Xu, Ph.D., Assistant Professor, Mechanical
Engineering & UTC Institute for Advanced Systems Engineering
• Email: [email protected]

2
 What’s exciting about this course

• Learn data science methods for

knowledge discovery in
manufacturing and materials-related
problems.
• Apply data mining and machine
learning techniques in prediction and
design optimization.

3
 Course Learning Objectives

1. Identify opportunities of applying data mining (DM) and machine learning

(ML) to solve material/manufacturing-relevant problems
2. Generate data of material/manufacturing systems
3. Select proper DM and ML methods for material/manufacturing-relevant
problems
4. Implement DM and ML algorithms in software
5. Perform data analysis for prediction
6. Conduct design optimization based on the knowledge learned from data
7. Apply the DM and ML methods to solve a real world engineering problem

4
 Modules of this Course
1. Introduction: Integrated Computational Materials Engineering (ICME)
2. Data generation by simulation process integration
3. Material image data analysis and learning
4. Data collection and labeling by Design of Experiments (DOE)
5. Surrogate modeling
6. Literature study and presentation
7. Comprehensive study with matminer
8. Optimization and model parameter calibration
9. Multi-fidelity modeling
10. Final project: progress update + final presentation
5
 Modules v.s. Learning Objectives
Course Learning Objectives (CLOs) Assessments
1. Identify opportunities of applying data mining (DM) and machine Module 1, Module 6, Module 10
learning (ML) to solve material/manufacturing-relevant problems
2. Generate data of material/manufacturing systems Module 2, Module 3, Module 4,
Module 10
3. Select proper DM and ML methods for material/manufacturing- Module 4, Module 5, Module 7,
relevant problems Module 8, Module 9, Module 10
4. Implement DM and ML algorithms in software Module 3, Module 4, Module 5,
Module 7, Module 8, Module 9,
Module 10
5. Perform data analysis to predict the responses Module 5, Module 9, Module 10
6. Conduct design optimization based on the knowledge learned Module 8, Module 10
from data
7. Apply the DM and ML methods to solve a real world engineering Module 10
problem
6
 Course Materials

• Lecture videos and slides

• Journal papers

• Sample codes

• Webpages

• Online videos

7
 Software used in this Course
Sample codes will be provided in the following formats:

https://software.uconn.edu/software/?licenseS
elect=University+Licensed&affiliationSelect=Stu Freely available online
dent

8
 Some online Tutorials of MATLAB

https://www.mccormick.northwestern.edu/documents/stude
nts/undergraduate/introduction-to-matlab.pdf

https://web.eecs.umich.edu/~aey/eecs451/matlab.pdf

9
 Some online tutorials of Python and ML packages

Python official tutorial: https://docs.python.org/3/tutorial/

https://cs231n.github.io/python-numpy-tutorial/
numpy tutorial https://s3.amazonaws.com/assets.datacamp.com/blog_assets/Num
py_Python_Cheat_Sheet.pdf

scikit-learn official tutorial https://scikit-learn.org/stable/tutorial/index.html

https://elitedatascience.com/keras-tutorial-deep-
Keras tutorial learning-in-python

10
 Assignments

• Homework: model implementation and report

• Literature study and presentation
o Assigned journal papers + literature search
• Final project
o Schedule a meeting with me to discuss your proposed
project topic

11
 Literature Study and Final Project
Literature study:
Choose one of the following three topics:
Topic 1: Integrated Computational Materials Engineering (ICME)
Topic 2: Design for Additive Manufacturing (DfAM)
Topic 3: Materials informatrics
Module 2: the reading materials will be provided
Module 6: Upload a pre-recorded presentation

Final project:
• Propose a topic for your final project
• You can work individually or in a team of 2 people
• Mid-term progress update
• Final presentation and report

12
 First assignment: install the software packages.

In the next video, we will introduce the concept of ICME.

13
Integrated Computational Materials Engineering (ICME)

Part 1: Computational Design of Structures and

Materials

14
Engineering Design: Structure Design Optimization

15
Example of Structure Design: Design for Crashworthiness

Source: https://www.computerhistory.org/makesoftware/exhibit/car-crash-simulation/ 16
 Structure design → Material Design

Usually, we choose materials for the structure design

Can we also design materials?

17
Microstructural Material Design: Microstructure – Property

Damping Thermal
Microstructure plasticity

Computational “Microstructure-
mediated” design
(McDowell, D. L., Olson, G., B., 2008)

 Advanced  Material  System design

simulation informatics methods

18
 Example: Nano-particle reinforced Polymer
Image processing Statistical
characterization

1 pixel = 11.9 nm

Microscopic image 3D microstructure

of heterogeneous reconstruction for
microstructures property simulation

Microstructure Property

Xu, H., Dikin, D.A., Burkhart, C. and Chen, W., 2014. Descriptor-based methodology for statistical characterization and 3D
reconstruction of microstructural materials. Computational Materials Science, 85, pp.206-216.
Xu, H., Li, Y., Brinson, C. and Chen, W., 2014. A descriptor-based design methodology for developing heterogeneous
microstructural materials system. Journal of Mechanical Design, 136(5). 19
 Example: Multi-layer Uni-directional Fiber Composites
Yu, Composites Par B: Engineering, 2015
Uni-Directional Tapes/sheets: Fibers run
parallel to roll length and are held together
by matrix resin

• Number of layers
• Fiber orientation
• Layer thickness
• …

20
Example: Woven and Stitched Carbon Fiber Composites
Woven Fabrics
Plain weave Multi-Axial Non
Crimp Fabrics

Twill weave

Tri-ax braid

Layers of fiber with

different orientation
are stitched together.
21
Integrated Computational Materials Engineering (ICME)

Part 2: ICME workflow

22
Integrated Computational Materials Engineering (ICME)

Product
Design

Materials Manufacturing
System Processes

23
Process-Microstructure-Property: Chopped Fiber Composites

Chips cut from a

long UD fiber tow Chip Compression molding
(short fiber tows)

Initial charge
(chips + resin)

Chip orientation = Fiber orientation

24
 Process-Microstructure-Property: Chopped Fiber Composites
• Compression force profile • Fiber chip size
• Temperature • Fiber volume fraction
Compression molding Fiber orientation Microstructure Property simulation (e.g.
simulation: material flow distribution of reconstruction fracture)
this element

Input predict properties to this element

(for upper scale analysis)

• Li, Y., Chen, Z., Xu, H., Dahl, J., Zeng, D., Mirdamadi, M. and Su, X., 2017. Modeling and simulation of compression molding process for
sheet molding compound (SMC) of chopped carbon fiber composites. SAE International Journal of Materials and Manufacturing, 10(2),
pp.130-137.
• Chen, Z., Tang, H., Shao, Y., Sun, Q., Zhou, G., Li, Y., Xu, H., Zeng, D. and Su, X., 2019. Failure of chopped carbon fiber Sheet Molding
Compound (SMC) composites under uniaxial tensile loading: Computational prediction and experimental analysis. Composites Part A:25
Applied Science and Manufacturing, 118, pp.117-130.
Process-Microstructure-Property: Preforming of Woven Composites
Punch

Die cavity

Source:
https://www.sciencedirect.com/topics/materials-
Zhang, W., Gao, J. and Cao, J., 2020. Blank geometry design for carbon fiber
science/textile-composite
reinforced plastic (CFRP) preforming using finite element analysis (FEA). Procedia
26
Manufacturing, 48, pp.197-203.
A Full ICME Framework: Carbon Fiber Composites for Vehicle Lightweighting

27
Data Flow in the ICME Framework

28
 Data Management and Model Validation

Constituent Level: Material Characterization

Thermosetting Resin Carbon Fiber
Filament
• Chemical Structure • Thermal Properties
• Rheological Properties • Mechanical Properties Carbon Fiber
Fabric

Coupon-Level (Molded Plagues): Model Correlation

ICME Database

ICME Models
Viscoplastic Behavior Voids Interfacial Energy
• Multi-axial
Fiber Orientation Fatigue
• High-Strain Rate
• Elevated Temperature Fiber Distribution Fracture

Component-Level: Model Validation

Viscoplastic Fiber Orientation/ Porosity Fracture Fatigue
Behavior Distribution

Assembly-Level ( Surrogate Prototype Parts): Model Validation

Stiffness/NVH Strength/Deflection Crash Durability

29
 A Sample List of Deliverables of an ICME Project

• Verified Material constitutive models

• Virtual material testers: microstructure RVE
• Implemented Material Models in FEA software
o Processing simulation
o Property prediction
• Implemented structure performance models in FEA software
• Integrated ICME design optimization workflow in optimization software
• Recommended ICME designs
• Experimental characterization data (validation)

30
 Demo: Design of Hat Section
• Objective:
• Minimize the weight of the hat section assembly, while keeping the attributes no
worse than the baseline design (1.5mm steel hat section)
• Design variables:

31
 Demo: Design of Hat Section

Requirement direction Threshold

r1_disp ≤ 1.13e-4 (mm)
r3_peeq ≤ 0.1%
r3_tsaiwu ≤ 1.0
r3_disp ≤ 1.0 (mm)
r4_peak ≤ 31.4 (kN)
r4_disp ≤ 13.51 (mm )

32
 Summary

Knowledge Discovery: Data

Mining & Machine Learning

Multi-fidelity Metamodeling
& Optimization

Pareto
frontier
Uncertainty
Propagation
Inputs
Outputs

Statistical characterization & Uncertainty

Stochastic Reconstruction Quantification 9
x 10 Copula vs. Data
1.65

1.6

1.55

1.5

Stress
1.45

1.4

r
1.35

1.3

0 0.1 0.2 0.3 0.4 0.5

Strain

33
Application of ICME in other areas

34
 3D printing of Alloys by Powder Bed Fusion

Source:
https://www.sciencedirect.com/topics/engineering/powder-bed-fusion
https://gfycat.com/faroffbreakablehatchetfish-engineering-titanium-dmls-qcml-eos
35
Multiphysics nature of metal powder bed fusion process

King, Wayne, Andrew T. Anderson, Robert M. Ferencz, Neil E. Hodge, Chandrika Kamath, and Saad A. Khairallah. "Overview of modelling and
simulation of metal powder bed fusion process at Lawrence Livermore National Laboratory." Materials Science and Technology 31, no. 8 (2015):
957-968. 36
 Material Data by Experiments
Pore size VED=28.66

Yang et al. MSEA, 2018: irregular shape, lack of fusion, 400um

In layer

Excessive VED + poor heat dissipation

Yang et al. MSEA, 2018:

keyholing, >30um, round
Contour scan region, edge of core,
island boundary
Marchese et al., Metals 2018
(probably) lack of fusion, 200um

VED=37.88

Yang et al. MSEA, 2018: Round trapped gas, ~5um

Weingarten et al, 2015: gas is mostly H2, can be removed by drying

Volumetric Energy Density = P/(vht)

P: Laser power, v: scan speed, t: layer thickness
h: hatching distance
37
Area Energy Density = P/(vh)
 Material Data by Experiments

• As built SS 316L

Wang et al., Nature Mater, 2018

38
 Material Data by Simulations

Numerical microstructure model for material property simulation

Li, Y., Xu, H., Lai, W.J., Li, Z., Su, X., A Multiscale Material Modeling Approach to Predict the Mechanical Properties of Powder Bed Fusion
(PBF) Metal, ASTM International’s STP: Selected Technical Papers, Fourth ASTM Symposium on Structural Integrity of Additive
Manufactured Materials & Parts, 2020 39
 Inputs for SLM Process Simulation
• Meshing
• Material properties
• Machine information: powder recoating sequence, laser scan path, material deposition of the printing head,
arc weld power, etc.

Zhang, Qi, Jiawen Xie, Zhenyuan Gao, Tyler London, David Griffiths, and Victor Oancea. "A metallurgical phase transformation framework applied to
SLM additive manufacturing processes." Materials & Design 166 (2019): 107618. 40
 Material Data by Simulations

Spatial distribution of metallurgical phases in the printed structure

Zhang, Qi, Jiawen Xie, Tyler London, David Griffiths, Imran Bhamji, and Victor Oancea. "Estimates of the mechanical
properties of laser powder bed fusion Ti-6Al-4V parts using finite element models." Materials & Design 169 (2019): 107678. 41
Predict the Structural Performances considering Local Properties

Zhang, Qi, Jiawen Xie, Tyler London, David Griffiths, Imran Bhamji, and Victor Oancea. "Estimates of the mechanical
properties of laser powder bed fusion Ti-6Al-4V parts using finite element models." Materials & Design 169 (2019): 107678. 42
 Multiscale modelling strategy for metal powder bed
fusion additive manufacturing process

King, Wayne, Andrew T. Anderson, Robert M. Ferencz, Neil E. Hodge, Chandrika Kamath, and Saad A. Khairallah. "Overview of modelling and simulation
43
of metal powder bed fusion process at Lawrence Livermore National Laboratory." Materials Science and Technology 31, no. 8 (2015): 957-968.
 ICME of Alloys printed by Powder Bed Fusion

King, Wayne, Andrew T. Anderson, Robert M. Ferencz, Neil E. Hodge, Chandrika Kamath, and Saad A. Khairallah. "Overview of modelling and 44 simulation
of metal powder bed fusion process at Lawrence Livermore National Laboratory." Materials Science and Technology 31, no. 8 (2015): 957-968.
Summary

Wang, Chengcheng, X. P. Tan, Shu Beng Tor,

and C. S. Lim. "Machine learning in additive
manufacturing: State-of-the-art and
perspectives." Additive Manufacturing (2020):
101538.

45
 Li-ion Battery Materials

Zhu, J., Li, W., Wierzbicki, T., Xia, Y. and Harding, J., 2019. Deformation and failure of lithium-ion batteries treated as a discrete layered
46
structure. International Journal of Plasticity, 121, pp.293-311.
 Manufacturing Processes & Microstructures

Electrodes:
Calendaring

Polymer separator:
Stretching & annealing

Meyer, Chris, Henrike Bockholt, Wolfgang Haselrieder, and Arno Kwade. "Characterization of the calendering process for compaction of
electrodes for lithium-ion batteries." Journal of Materials Processing Technology 249 (2017): 172-178.
Duquesnoy, M., Lombardo, T., Chouchane, M., Primo, E. and Franco, A.A., 2020. Accelerating Battery Manufacturing Optimization by
Combining Experiments, In Silico Electrodes Generation and Machine Learning
47
Yu, Wei-Ching, and Shawn E. Hux. "Method of making a trilayer battery separator." U.S. Patent 5,952,120, issued September 14, 1999.
Computational Microstructure Modeling for Property Prediction

LiNi1/3Mn1/3Co1/3O2 cathode

Electrochemical modeling

Polypropylene separator

Mechanical modeling

Lu, Xuekun, Antonio Bertei, Donal P. Finegan, Chun Tan, Sohrab R. Daemi, Julia S. Weaving, Kieran B. O’Regan et al. "3D microstructure design
of lithium-ion battery electrodes assisted by X-ray nano-computed tomography and modelling." Nature Communications 11, no. 1 (2020): 1-13.
Pan, Zhexin, Juner Zhu, Hongyi Xu, Tobias Sedlatschek, Xiaowei Zhang, Wei Li, Tao Gao, Yong Xia, and Tomasz Wierzbicki. "Microstructural
deformation patterns of a highly orthotropic polypropylene separator of lithium-ion batteries: Mechanism, model, and theory." Extreme Mechanics
48
Letters (2020): 100705.
 Parametric Microstructure Design
Particle size Particle shapes

Spatial arrangement

He, S., Habte, B.T. and Jiang, F., 2016. LBM prediction of effective electric
and species transport properties of lithium-ion battery graphite anode. Solid
State Ionics, 296, pp.146-153.
Kespe, M. and Nirschl, H., 2015. Numerical simulation of lithium‐ion battery
performance considering electrode microstructure. International Journal of
Energy Research, 39(15), pp.2062-2074.
Mai, W., Yang, M. and Soghrati, S., 2019. A particle-resolved 3D finite
element model to study the effect of cathode microstructure on the
behavior of lithium ion batteries. Electrochimica Acta, 294, pp.192-209.
49
Machine Learning-driven Process-Microstructure-Property Analysis
and Design

Deep
neural
network

Duquesnoy, Marc, Teo Lombardo, Mehdi Chouchane, Emiliano Primo, and Alejandro A. Franco. "Accelerating Battery
Manufacturing Optimization by Combining Experiments, In Silico Electrodes Generation and Machine Learning." (2020). 50
 Summary

Microstructure

Process Property

51