information sciences to fuel the data age of materials science
DESCRIPTION
A presentation given at Novelis R&D in Kennesaw,Ga on Wednesday August 28 2013. The presentation was organized by Babak Raeisinia. The presentation provides a scope of what emerging information science, data science, and microstructure informatics techniques can used to drive the Materials Genome Initiative.TRANSCRIPT
Information Sciences to Fuel the Data Age of Materials Science
Tony Fast Materials Data Analyst
MINED Materials Informatics for Engineering Design
Georgia Institute of Technology
MATERIALS SCIENCE IS BIG DATA
Ad Hoc Standards, Silos, Integration, Software, Equipment, Data Formats, Ideas
The Materials Innovation Network is a collaborative environment built to fuel the Materials Genome Initiative by managing users and their digital data for microstructure driven materials development and improvement. SOCIAL NETWORK Manages users, projects, and expert communities engaged in materials science related efforts. A melting pot for materials scientists, big data, and integration. CODE REPOSITORY A platform with an embedded versioning system to develop codes and deployable tools for the MATIN community at large. This platform will enable good coding practices and rapid delivery of academic utilities to market quickly. DATABASE The database is the unifying feature of MATIN. This graph database is specifically designed to store nearly all types of materials datasets, maintain data provenance, semantically query metadata, learn design patterns ( or workflows ), support the big data generation, and establish a federated database with access control for academia, industry, and national labs.
Database
Information
Models Analytics
Codes Users
Social Network Versioning Control Database API
Collaborative Workspace
TEAM 1
Information / Code Upload
Query/Download
Analytics
Model Development Dat
a P
rove
nanc
e New Data
Object
Workflow
Data Object
Metadata
Collaborative Workspace
TEAM 2
Information / Code Upload
Query/Download
Analytics
Model Development
Metadata
MATIN will enable • Collaboration • Standarization • Electronic
Recording • Data Management
and Federated • High Value Testing • Knowledge
Transfer
STRUCTURE INFORMATICS WORKFLOW
PHYSICS BASED MODELS SIMULATION EXPERIMENT
MICROSTRUCTURE (MATERIAL) SIGNAL MODULES
ADVANCED & OBJECTIVE STATISTICAL MODULES
DATA MINING MODULES
VALUE ASSESSMENT
INTE
LLIG
EN
T D
ES
IGN
OF
EX
PE
RIM
EN
TS Microstructure Informatics is a data-
driven system to mine structure-property/processing connections from experimental and simulation materials science information. The system is agnostic to material system and length scale, objectively quantifiable, and rapidly iterates in less cycles for both materials improvement and discovery.
Microstructure signal modules are (semi) automated tools to identify local and effective microstructure features.
Aluminum in Epoxy Titanium
EMMPM - BlueQuartz
Bamboo
Martensitic Steel SiC/SiC Al-Cu Solidification
ADVANCED & OBJECTIVE STATISTICAL MODULES
THE MICROSTRUCTURE IS A SAMPLE IN AN IMMENSE STATISTICAL POPULATION.
α-β Titanium
SPATIAL STATISTICS
fthh ' =
mshms+t
h '
Dts∑t t
t
Statistical correlations between random points in space/time which reveal systematic patterns in the microstructure. Contains the original µS within a translation & inversion.
CURRENT APPLICATIONS metals, polymers, fuel cells, cmc, md, & a bunch of other things
TYPES OF SIGNALS sparse, experimental, simulation, heterogeneous, surface, bulk
DATA MINING MODULES
Microstructure Material Processing Property
Mining modules are machine Learning solutions to extract rich bi-directional structure-property/processing linkages from materials & microstructure datasets. Mining modules create structure taxonomies, homogenization and localization relationships, ground truth comparison between simulation and experiment, materials discovery, and materials improvement.
Objective Microstructure Classification of α-β Titanium Images àStatisticsààMine with Principal Component Analysis
Mechanical Deformation of Polymer Chains
Molecular Dynamics of Aluminum Atoms
MPL
GDL
X-CTàFinite Element ModelingàStatisticsàà Regression to connect the statistics with diffusivity values from FEM
Bottom-up Homogenization Relationships
FEM"ε=5e-4"
Meta-modeling with Materials Knowledge Systems Top-down localization relationships
ps = athms+t
h
h∑
t∑
The MKS design filters that capture the effect of the local arrangement of the microstructure on the response. The filters are learned from physics based models and can only be as accurate as the model never better.
OTHER APPLICATIONS"Spinodal Decomposition, Grain Coarsening, Thermomechanical, Polycrystalline
Top-Down Localization Relationships for High Contrast Composites
The MKS is a scalable, parallel meta-model that learns from physics based models to enable rapid simulation at a cost in accuracy.
N2 vs. Nlog(N) complexity
It learns top-down localization relationships to extra extreme value events and enables multiscale integration.
Structure-Processing MKS Processing History
Structure-Property Homogenization
Structure-Property Localization
Illustrating Integration
Data enables bidirectional S-P/P, multiscale integration, and higher throughput
OTHER DOMAINS THAT WILL FUEL BIG DATA IN MATERIALS SCIENCE information gain theory, digital signal processing, regressions, statistics,
high performance computing, cloud computing, databases, mobile devices, a connected community