Research
Theory, experiments, models, state estimation, uncertainty, AI augmentation and autonomous decision-making across all five thrusts.
SCIENCE → METHODS → EVIDENCELed by Franck Adjogble · Adjunct Professor · CMU Materials Science & Engineering
Adjogble
Research Group
Materials Intelligence, Digital Twins & Autonomous Manufacturing
Advancing materials-centric digital twins through physics, artificial intelligence, and human expertise for intelligent, energy-aware, and increasingly autonomous manufacturing.
5research thrusts
MP-DTcore construct
R→E→Iresearch · education · translation
Explore the material-process intelligence loop
Academic homeCMU Department of Materials Science and EngineeringARG is a faculty-led research group in the Department of Materials Science and Engineering.
Energy collaborationWilton E. Scott Institute for Energy InnovationCollaboration on energy, sustainability and industrial decarbonization.
Manufacturing collaborationManufacturing Futures InstituteCollaboration on advanced manufacturing and digital systems.
Research thesis
From evolving material state
to trustworthy manufacturing intelligence.
Materials do not experience manufacturing as isolated unit operations. Their state evolves continuously through thermal, mechanical, chemical and temporal histories. The industrial system that produces them is equally interconnected.
ARG asks how knowledge of that evolving physical state can be transformed into trustworthy computational intelligence capable of improving — and ultimately autonomously controlling — the way materials are manufactured.
Selected research question
How can the evolving physical, metallurgical and microstructural state of a material be continuously represented through its processing history?
Core scientific construct
Material-Process
Digital Twin
The Material-Process Digital Twin is the scientific core of ARG: a continuously evolving computational representation that connects what processing does to a material, how its structure changes, what properties emerge, and how performance can be predicted and controlled.
01Processingthermal · mechanical · chemical history
→
02Structurephase · microstructure · defects · state
→
03Propertiesmechanical · thermal · functional response
→
04Performancequality · reliability · product behavior
Sensing → State Estimation → Physics → Data → Digital Twin → Prediction → Decision → Control
Sensing→State Estimation→Physics→Data→Digital Twin→Prediction→Decision→Control
↳ feedback from physical operation, validation and learning returns to the twin
ENERGY→CARBON→SUSTAINABILITY→INDUSTRIAL RESILIENCE
Cross-cutting framework · spans all five thrusts
AI-Augmented,
Human-Accountable Research
ARG openly studies and uses advanced artificial intelligence — including generative AI, large language models, multimodal foundation models, scientific machine learning and agentic systems — as both an enabling research methodology and an object of systematic scientific study.
AI may assist reasoning, modeling, coding, analysis, simulation, visualization, documentation, and scientific exploration, while scientific claims, validation, attribution, research integrity, data governance, and final research decisions remain the responsibility of human researchers.
01Physical Systemmaterial · process · equipment · plant
02Digital Twin Intelligencestate · model · prediction · control logic
03Artificial Intelligencelearning · reasoning · generation · agents
04Human Intelligencehypothesis · judgment · validation · accountability
Knowledge synthesis→Hypothesis→Model→Software & simulation→Data analysis→Digital twin→Experiment→Verification→Industrial validation→Communication
ValidityWhen does AI-generated scientific reasoning remain valid?
TraceabilityHow are sources, transformations, prompts, models and decisions made reproducible?
Physics interactionHow should AI cooperate with physics-based and hybrid models?
UncertaintyHow should confidence and model limitations affect recommendations or control?
GovernanceHow are sensitive data, attribution, integrity and human oversight preserved?
Systems-scale extension
Materials-to-Enterprise
Twin Continuum
The Material-Process Digital Twin provides the scientific core. This continuum shows how its intelligence scales from material state to process, equipment, plant, energy system and enterprise decision-making.
MATERIAL STATE × PHYSICS / METALLURGY
Constitutive & metallurgical state
Scientific representation of the physical state and its evolution.
One research system · three outward-facing dimensions
Research creates the science.
Education develops capability.
Industrial translation tests reality.
The five research thrusts are not isolated from teaching or industry. Each generates scientific questions, student work and industrial validation, with evidence flowing back into the research program.
Education
Course modules, supervised student work, capstones, Digital Twin software and undergraduate, MS and PhD research opportunities.
LEARN → BUILD → VALIDATEIndustrial Translation
Industrial Digital Twin applications, manufacturing use cases, validation, software demonstrators, technology transfer and sponsored research.
PILOT → DEPLOY → LEARNResearch ↔ Education ↔ Industrial Translation ↔ Research
Flagship work
Research designed to
survive contact with reality.
Steel and metals manufacturing are ARG's principal initial scientific and industrial testbed, building on established expertise, course activities, data and industrial relationships. The science extends well beyond steel.
Metallurgical Corridor Digital Twin
How can material state be propagated through thermal and mechanical processing while remaining synchronized with the operating plant?
Physics-Informed Materials Intelligence
Hybrid scientific learning systems that use data and AI without discarding materials physics, uncertainty, provenance or expert knowledge.
View research questions →Credible, Knowledge-Driven Digital Twins
Connecting physical assets, material genealogy, models, uncertainty, provenance and operational decisions through traceable knowledge structures.
Explore credibility layer →Beyond the initial steel testbed: ARG methods extend to additive manufacturing, aluminum, battery materials, composites, critical materials, semiconductor-material processing and other advanced manufacturing systems.
Education studio · CMU MSE
Principles of Digital Twins in Materials Science & Advanced Manufacturing
Students treat the Digital Twin as an engineering and scientific system rather than a visualization. The course integrates scientific models, industrial data, software architecture, interactive representations, uncertainty, AI-assisted development and human decision-making.
Theory→Scientific Model→Industrial Data→Digital Representation→Decision Support
Digital Twin Studioresearch → classroom → prototype
VisualizationReact · Three.js · Scientific UI
EngineeringPython · C# · Optimization
Industrial dataOPC UA · MQTT · Time Series
ScienceMaterials · Physics · Process Models
Supervised workStudent research · assignments · capstones · thesis-linked development
SoftwareDigital Twin applications · simulations · educational tools · demonstrators
Research pathwayUndergraduate · MS · PhD research opportunities
Industrial Translation
From scientific theory to operating plant.
A twin becomes meaningful when assumptions, data, uncertainty and decisions remain credible under real operating conditions. Industrial translation provides the validation loop between research questions and manufacturing reality.
01Scientific modelRepresent the physics
→
02Digital twinConnect model + system
→
03ValidationQuantify credibility
→
04DecisionHuman + machine reasoning
→
05DeploymentLearn from operation
industrial Digital Twin applicationsmanufacturing use casesindustrial validationsoftware demonstratorsindustry collaborationtechnology transfersponsored research
E
Research Thrust 5 · In collaboration with the Scott Institute
The manufacturing twin
is also an energy twin.
Temperature, transformation, productivity, quality, energy consumption and carbon intensity are tightly coupled in materials processing.
ARG studies unified twins that expose these interactions to support energy-aware operation, industrial decarbonization, sustainability and resilience, in collaboration with the Wilton E. Scott Institute for Energy Innovation.
Explore energy-aware manufacturing research →Research evidence
Publications, demonstrations & insights.
Selected work connecting publications, student work, software demonstrators and industrial validation to ARG's research thrusts and the Material-Process Digital Twin.
Request a publication or demonstration →People
The group.
ARG brings together faculty, students and industrial collaborators working across materials science, manufacturing, digital twins, artificial intelligence and energy.
Group lead
Franck Adjogble
Adjunct Professor · Department of Materials Science and Engineering · Carnegie Mellon University
Leads ARG's research program in materials intelligence, materials-centric digital twins and autonomous manufacturing, and teaches Principles of Digital Twins in Materials Science and Advanced Manufacturing.
Students & researchers
Undergraduate, MS and PhD students interested in digital twins, materials processing, scientific machine learning or autonomous manufacturing are welcome to get in touch.
Research opportunities →Build with us
Research that crosses
disciplinary boundaries.
Contact
Adjogble Research Group
Department of Materials Science and Engineering
Carnegie Mellon University
5000 Forbes Avenue · Pittsburgh, PA 15213
Carnegie Mellon University
5000 Forbes Avenue · Pittsburgh, PA 15213