Vit Meeting Room

CONSULTANCY

 

MACHINE LEARNING 

STRATEGY

Ferritico conducts pre-studies to identify steel development, manufacturing and implementation process machine learning opportunities. We provide digitalization recommendations, i.e. how out-of-the-box and customized machine learning models can be combined to enhance steel product quality metrics, increase manufacturing output or to track the cause of product defects.  

consulting4.png

DATA 
MANAGEMENT

Customer empirical data digitalization and database development

Open literature data collection and local data set integration

 

Synthetic data generation using i.a physical modelling and first principles calculations

 

Data quality control design and technical implementation

 

Data pipeline solution development, i.e. automize process of converting raw process data

 

Data storage, e.g. develop and host databases and search capabilities for customer datasets.

dataCollectionWhite.png
 
 

NOVEL STEEL 
DEVELOPMENT

Novel steel recipe development based on requirement specification. Ferritico team stand-alone or as an integrated part of customer R&D to educate in ICME methodology and Ferritico machine learning simulation work-flow.

Steel optimization, i.e. fine tuning recipe to boost properties, reduce cost or decrease environmental impact

 

Alloy element replacement projects, i.e. refine recipe in order to reduce cost or dependency on critical alloying elements while maintaining mechanical properties.

 

Production ramp-up consultancy projects, i.e. simulation supported process to maintain properties when going from pilot to production scale

ml.jpg

FEDERATED

LEARNING

The Ferritico federated learning solution helps our customers to leverage local steel data sets without having to share and distribute data.  Ferritico central prediction models are reinforced through local data consumption at the customer site  and then deployed as web services. The federated learning solution enables customer collaboration without data sharing and provides the collaborators a boosted model at license discount in relation to model contribution.   

ml.png
 

FINITE ELEMENT METHOD 

MATERIALS DATA

Finite Element Method (FEM) simulation quality correlates with input materials data quality. Improved FEM simulations implies less trial-and-error during design and enables product performance and sustainability optimization. Ferritico provides materials data API:s where steel materials data can be consumed as a service for the purpose of boosting FEM simulations

femBlue2.png