Research

Research Library for Metal Transformation Intelligence

Paper summaries, technical explainers, process intelligence notes, and AI method explainers. Each one is grounded in a real industrial metal-transformation problem and links to the open models, datasets, and benchmarks it informs.

8 results

Sort

Technical explainer12 min

From feedstock to metal outcome: a shared framing for transformation intelligence

A cross-metal framing that connects feedstock, chemistry, process conditions, and equipment behaviour to final metal outcomes. It is the conceptual backbone shared across the platform's process maps.

General metal transformation

Read

Process intelligence note7 min

Oxygen, morphology, and metal powder qualification

How oxygen pickup, sphericity, and flowability interact to determine whether a powder batch qualifies, and how classification models can screen batches before costly downstream use.

Metal powdersTitanium & specialty alloys

Read

AI method explainer8 min

Impurity breakthrough in lithium purification: a risk-modelling primer

A primer on framing battery-grade impurity control as a calibrated risk-classification problem, including evaluation with ROC-AUC and Brier score and why calibration matters for qualification decisions.

Lithium

Read

Process intelligence note9 min

Predicting recovery and selectivity in rare earth solvent extraction

A structured look at how stage pH, phase ratio, and feed assay drive recovery and selectivity in rare earth separation, and which features carry the most predictive signal for recovery and purity tasks.

Rare earths

Read

Technical explainer7 min

Inclusion risk and melt chemistry: what predicts aluminium cast quality

How charge composition, melt treatment, and scrap blend interact to drive inclusion risk and cast quality, and how chemistry-prediction models can flag off-target melts before casting.

AluminiumScrap & recycled metals

Read

Process intelligence note10 min

Slag losses and impurity behaviour in copper smelting

How matte grade, slag chemistry, and furnace conditions govern copper losses to slag, and how loss-prediction models connect furnace behaviour to recovery and cathode quality.

Copper

Read

AI method explainer11 min

Tabular models for metal transformation: baselines that hold up

A method explainer on why gradient-boosted trees and calibrated tabular models remain strong baselines for metal-transformation tasks, and how to report results so they are comparable on benchmarks.

General metal transformation

Read

Technical explainer6 min

Impurity accumulation and circular metal quality

Why impurities accumulate irreversibly in recycled loops and how blend planning becomes a quality decision, with a look at contamination-risk prediction for secondary metals.

Scrap & recycled metalsAluminium

Read

Contribute research

Share an explainer or discuss a method

Research here is open and non-confidential. Contribute a summary, propose a method explainer, or discuss evaluation with other builders.

Join the Community Explore Benchmarks