Program information

Program aim

To achieve the Dutch circularity goals, the use of raw materials must be drastically reduced by 2030. This is particularly relevant for steels due to the extremely high ecological footprint of their primary production process. Improved circularity can be achieved by increasing the ratio of recycled material (scrap) in the production process. However, by increasing the scrap-ratio, impurities are introduced that induce variations in the mechanical properties of the steel, causing problems in the manufacturing of steel products. DEPMAT aims to develop data-enhanced physical material models which are sufficiently predictive to guide adjustments of process settings, to achieve constant output quality with varying impurity content, enabling higher scrap ratios in steel production and processing.

The Dutch mission to realize a fully circular economy in 2050, and by 2030 a reduction of 50% of raw material use is problematic for the steel industry. These targets are particularly urgent for high quality steel strip, e.g., used for (battery-) cans and automotive applications, whose production relies on the blast furnace process where ‘virgin’ steel is produced via a carbon-based iron ore reduction process. Apart from being non-circular by definition (iron ore is obtained by mining) the blast furnace reduction process puts a high burden on the environment: oxygen atoms in the ore are captured by carbon such that apart from iron, huge quantities of CO2 are produced. Thus, both from a circularity and a carbon footprint perspective, it is evident that we need to ensure that reduced iron, which from their reduction process already carries a high environmental load, is not lost.

Problem statement


The principal cause limiting the proportion of scrap in high-quality steels, is that scrap can contain a large number of elements other than iron. Introduction of scrap causes composition variations in the steel which induce property variations. These variations are undesirable. Steel production and processing lines, both at steel producers and customers, are carefully tuned towards the steel properties, especially for high-quality steels. Variations of the properties lead to increased rejection rates, both at steel producers and their customers, and in the worst case can lead to a line stop in production. 

Steel properties are governed by both composition and thermo-mechanical history. In steel processing, deviations due to composition variations could therefore in principle be compensated with adjustments of the processing parameters. For example, a steel containing an element which has a hardening effect, could be deformed at a relatively high temperature, which has a softening effect. State of the art manufacturing lines indeed have the flexibility to adjust process settings within (limited) process windows. However, today this flexibility is deployed to optimize the robustness of the production process and/or to optimize the end-product properties for a carefully controlled, standardized material. The steel composition is kept constant as much as possible because models predicting properties from steel composition and processing parameters are not sufficiently quantitatively predictive to tune processing process parameters towards obtaining the required properties for steels with significant composition variations. Apart from the lack of accuracy,
existing physical material models can be extremely time-consuming which makes them unfit to govern processing conditions on a batch-to-batch or in-line basis.

Our vision is to close the life cycle for steel, ensuring that no reduced iron atom is wasted. This can be achieved by re-introducing scrap as a base-material in steel production, which is currently already done. However, only low quality steels can be produced using entirely recycled material, using for example the electric arc furnace. The current processing and manufacturing technologies for high quality steels cannot deal with the composition variations that accompany the increase in recycled material content.

The ultimate vision of success of this program is that material producers and users are deploying higher fractions of recycled material without compromising on product quality. This by adjusting processing and manufacturing parameters, potentially in-line, based on data-enhanced physical materials modelling. These data-enhanced physical material models will be faster and more accurate than existing purely physical models and can quickly adapt to varying material compositions.

Goals and objectives


The implementation of the program results will give the involved steel manufacturers the possibility of increasing their scrap content further. The models developed will provide valuable insight into the effects of the increased material variability, and thus the possibility to better control the processing conditions.

The increased scrap content automatically results in a reduction in raw material use. As such, the emissions accompanying the process will decrease accordingly.

If successfully implemented, several outcomes will not be necessarily connected to the introduction of higher scrap fractions and can be considered as valuable spin-off: (1) with the implemented models products can be produced within tighter tolerance margins; (2) the possibility to adjust process settings to steer product properties will allow producers and manufacturers to improve product properties by adjusting process settings; (3) Instead of specifying a steel in terms of average properties with certain error margins, steel properties can be specified from batch-to-batch, or even more locally by specifying variations within each steel coil. Together with the outcome of using steel with higher scrap content (low price of scrap as raw material), these three outcomes will support the competitiveness of the companies in which the data enhanced physical materials model-based strategy is implemented.

  • WP leader: M.G.D. Geers (TU/e-MoM)
  • Co-applicants: E.S. Perdahcioglu (UT-NSM), J.P.M. Hoefnagels (TU/e-MoM), A.H. van den Boogaard (UT-NSM), R.H.J. Peerlings (TU/e-MoM)
  • Requested research positions: 3 PhDs
  • Duration of project: 4 years


  • PhD 1a: Towards hybrid model formulations: data-driven solutions exploiting constitutive equations
    Supervisors: M.G.D. Geers (TU/e-MoM), R.H.J. Peerlings (TU/e-MoM), J.P.M. Hoefnagels (TU/e-MoM)
  • PhD 1b: Physically consistent data-driven constitutive models by machine learning
    Supervisors: E.S. Perdahcioglu (UT-NSM), A.H. van den Boogaard (UT-NSM)
  • PhD 1c: Bridging synthetic RVE data to micromechanical tests for efficient data-coupled models
    Supervisors: J.P.M. Hoefnagels (TU/e-MoM), M.G.D. Geers (TU/e-MoM), R.H.J. Peerlings (TU/e-MoM)
  • WP leader: J. Havinga (UT-NSM)
  • Co-applicants: A.H. van den Boogaard (UT-NSM), Dr. D. Zarouchas (TUD-ASM), Prof.dr. J. Schmidt-Hieber (UT-MOR), Dr. P.K. Mandal (UT-HS)
  • Requested research positions: 3 PhDs and 2 Postdocs
  • Duration of project: 4.5 years


  • PhD 2a: Inline hybrid modelling in cold rolling and forming
    Supervisor: J. Havinga (UT-NSM), A.H. van den Boogaard (UT-NSM)
  • PhD 2b: Inline probabilistic state estimation and model correction
    Supervisors: J. Havinga (UT-NSM), Dr. P.K. Mandal (UT-HS), A.H. van den Boogaard (UT-NSM)
  • PhD 2c: Hybrid modelling for inline estimation of oxide layer formation
    Supervisor: Dr. D. Zarouchas (TUD-ASM)
  • Postdoc 2d: Microstructure estimation with electromagnetic sensing
    Supervisors: J. Havinga (UT-NSM), L. Abelmann (UT-RAM)
  • Postdoc 2e: Mathematical foundations of hybrid modelling for complex manufacturing processes
    Supervisors: Prof.dr. J. Schmidt-Hieber (UT-MOR), J. Havinga (UT-NSM)
  • WP leader: Dr. S. Kumar (TUD-MSE)
  • Co-applicants: M.G.D. Geers (TU/e-MoM), M.H.F. Sluiter (TUD-MSE)
  • Requested research positions: 3 PhDs
  • Duration of project: 4 years


  • PhD 3a: Fast micromechanical models for (data-enhanced) materials engineering
    Supervisors: M.G.D. Geers (TU/e-MoM), Dr. S. Kumar (TUD-MSE)
  • PhD 3b: Fast models for microstructure evolution in steel
    Supervisors: K. Bos (TUD-MSE), Dr. S. Kumar (TUD-MSE)
  • PhD 3c: Physics-based modelling of segregation of tramp elements and its effects on processes
    Supervisors: M.H.F. Sluiter (TUD-MSE), Dr. S. Kumar (TUD-MSE)
  • WP leader: Dr. L.Y. Chen (TUD-DS)
  • Co-applicants: Dr. M.H.F. Sluiter (TUD-MSE), A.H. van den Boogaard (UT-NSM)
  • Requested research positions: 2 PhDs
  • Duration of project: 4.5 years


  • PhD 4a: AI Process Model
    Supervisor: L.Y. Chen (TUD-DS), K. Bos (TUD-MSE)
  • PhD 4b: Meta/Transfer Learning
    Supervisor: L.Y. Chen (TUD-DS), K. Bos (TUD-MSE)
  • WP leader: Prof. van den Boogaard (UT-NSM)
  • Co-applicants: Dr. D. Zarouchas (TUD-ASM), L.Y. Chen (TUD-DS), J. Havinga (UT-NSM)
  • Requested research positions: None; researchers of all other WPs will contribute to WP5.
  • Duration of project: 4.5 years

Work packages and research positions

This website uses cookies to ensure you get the best experience on our website. Learn more.