Webinar: Machine Learning Steel Properties Simulation
10 December 2020
EIT RawMaterials consortiums EndureIT and CorTools have the common goal to develop machine learning based software for the purpose of steel characteristics simulation, focusing on embrittlement and corrosion respectively. Extended simulation capabilities would support a larger degree of steel optimization and have incalculable effects on the steel industry and the society.
On 10 December 2020, EndureIT and CorTools project partner Ferritico will host a public webinar together with EIT RawMaterials and the project consortiums. The webinar will focus on machine learning steel properties simulation and present the solutions being developed in the projects.
Industrial steel development, manufacturing and implementation are time-consuming, inefficient and manual processes, but digitalization has the potential to change this. These processes are characterized by significant trial-and-error and physical testing as conventional physical modelling-based simulation software cannot simulate with satisfying accuracy for all steels and processes.
Research shows that a data-driven machine learning approach can outperform physical modelling, in terms of simulation accuracy and effectiveness, for many steel characteristics. Hence, machine learning-based steel simulation is becoming an important complement to physical-based modelling.
One of the key success factors for developing machine learning software for steel simulation is data availability. Therefore, collaboration and setting up consortiums to pool data where each partner e.g. has data for a certain steel composition range and property is of great importance.
Ferritico, a Spin-off from steel research made at KTH Royal Institute of Technology, is developing simulation software for the steel and manufacturing industry, combining metallurgical know-how and physical modelling with development of large databases and machine learning models.
EIT RawMaterials CorTools project is developing a corrosion resistance prediction model. Consortium partners are VTT, Outotec, Boliden, Outokumpu, Tecnalia, ZAG and DMS.
EIT RawMaterials EndureIT project is developing an embrittlement prediction model. Consortium partners are KTH, CEA and Outokumpu.