AI4EAFRoute: Artificial Intelligence Supporting Innovative EAF Route Methods for Process and Plant Management to Enhance Performance and Component Effectiveness

Rese­arch Fund of Coal & Steel, 1 July 2026 — 30 June 2030

Project description

This pro­ject aims to enhan­ce the per­for­mance, relia­bi­li­ty, and sus­taina­bi­li­ty of the Elec­tric Arc Fur­nace (EAF) rou­te by inte­gra­ting advan­ced moni­to­ring tech­no­lo­gies with arti­fi­ci­al intel­li­gence (AI) tools. Through four tar­ge­ted use cases, the pro­ject will tack­le key pro­cess chal­lenges, inclu­ding scrap mel­ting manage­ment, opti­miza­ti­on of elec­tri­cal para­me­ters, refrac­to­ry lining ero­si­on, water leaka­ge in panels, event reco­gni­ti­on and pre­dic­tion, and over­all per­for­mance manage­ment. For each use case, dedi­ca­ted AI models will be deve­lo­ped using real indus­tri­al data and lever­aging mul­ti­mo­dal inputs such as acou­stic, volu­metric, visu­al, and ther­mal signals. The approach fol­lo­wed will include the moni­to­ring and data ana­ly­sis of equip­ment sta­tus in par­al­lel with the typi­cal approach used for pro­cess moni­to­ring and manage­ment, to stu­dy the influence of the adopted ope­ra­ting prac­ti­ces on equip­ment con­di­ti­on and dura­bi­li­ty, and to eva­lua­te the impact of com­po­nent inef­fec­ti­ve­ness on pro­cess per­for­mance. Both eva­lua­tions will be trans­la­ted into sim­pli­fied pre­dic­tion rules, to be used also as part of the gui­de­lines for a Decis­i­on Sup­port Sys­tem. The­se models will be inte­gra­ted into an advan­ced AI-based Decis­i­on Sup­port Sys­tem (DSS), desi­gned to pro­vi­de fur­nace ope­ra­tors with real-time recom­men­da­ti­ons. This will enable more pre­cise pro­cess con­trol, redu­ced unplan­ned main­ten­an­ce, exten­ded equip­ment life­time, and signi­fi­cant cost and ener­gy savings. The pro­ject advan­ces the sta­te of the art on seve­ral fronts. In Com­pu­ter Visi­on, it will intro­du­ce novel tech­ni­ques for robust image ana­ly­sis under the harsh con­di­ti­ons of EAF envi­ron­ments. In time-series ana­ly­sis, it will app­ly advan­ced AI methods to acou­stic signal pat­terns, enab­ling the ear­ly detec­tion of inef­fi­ci­en­ci­es and fail­ures. By com­bi­ning the­se inno­va­tions, the pro­ject will deli­ver a holi­stic and pre­dic­ti­ve approach to EAF manage­ment that is not curr­ent­ly available on the mar­ket. Ulti­m­ate­ly, the pro­ject will streng­then both the decar­bo­niza­ti­on and the com­pe­ti­ti­ve­ness of the Euro­pean steel indus­try by opti­mi­zing EAF pro­ces­ses and manage­ment. In doing so, it direct­ly sup­ports the objec­ti­ves of the RFCS pro­gram and the Euro­pean Green Deal, through impro­ved resour­ce effi­ci­en­cy, hig­her metal­lic yield, and redu­ced CO₂ emissions.

Project goals

The con­sor­ti­um has iden­ti­fied four key use cases to be inves­ti­ga­ted, each con­tri­bu­ting to the over­all goal of opti­mi­zing the manage­ment of the Elec­tric Arc Fur­nace (EAF) mel­ting pro­cess. For each use case, a dedi­ca­ted Arti­fi­ci­al Intel­li­gence (AI) model will be deve­lo­ped to address spe­ci­fic pro­cess chal­lenges and enable more effec­ti­ve decision-making:

  • EAF scrap mel­ting manage­ment: This use case focu­ses on iden­ti­fy­ing rela­ti­onships bet­ween the type of scrap and the mel­ting per­for­mance, using volu­metric data ana­ly­sis for scrap bas­ket obtai­ned through radar mea­su­re­ments or 3D laser scan­ner and acou­stic sen­sor for mel­ting process.
  • EAF elec­tri­cal manage­ment: This use case aims to deve­lop an AI-based model that lever­a­ges the novel Arc Qua­li­ty Index (AQI) for opti­miza­ti­on of the EAF ope­ra­ti­on. By inte­gra­ting high-speed mea­su­re­ments of elec­tri­cal varia­bles, phy­sics-based soft-sens­ing, and AI models trai­ned on expert know­ledge, the sys­tem will esti­ma­te arcs sta­bi­li­ty and their covera­ge by slag, and recom­mend opti­mal trans­for­mer tap set­tings, elec­tro­de posi­ti­on, and C and O2 injec­tion during vary­ing mel­ting stages.
  • EAF com­pon­ents and events manage­ment: The fol­lo­wing sec­tion broad­ly encom­pas­ses three major goals: local panel manage­ment, events detec­tion for alert manage­ment, and over­all lining and elec­tro­des management.
  • EAF glo­bal per­for­mance ana­ly­sis: this use case aims to demons­tra­te how pro­cess data (elec­tri­cal con­sump­ti­on, metal­lic yield, heat dura­ti­on, steel and slag com­po­si­ti­on, and tem­pe­ra­tures) tog­e­ther with the digi­ta­liza­ti­on of equip­ment sta­tus, dama­ges, and main­ten­an­ce ope­ra­ti­ons can be lever­a­ged to train AI algo­rith­ms for pro­duc­tion optimization.

 

The out­puts from the­se four AI-dri­ven use cases will be inte­gra­ted and fur­ther ana­ly­sed by a hig­her-level AI model. This final model will ser­ve as the foun­da­ti­on for a Decis­i­on Sup­port Sys­tem (DSS), desi­gned to assist fur­nace ope­ra­tors in real time, impro­ving pro­cess con­trol, effi­ci­en­cy, and equip­ment lifetime.

Project participants

Contact

Amit Sharma, M.Sc.

 

+49 241 80–26071

Dr.-Ing. Thomas Echterhof

 

+49 241 80–25958

Funding

Fun­ded by the Euro­pean Uni­on. Views and opi­ni­ons expres­sed are howe­ver tho­se of the author(s) only and do not neces­s­a­ri­ly reflect tho­se of the Euro­pean Uni­on or Euro­pean Rese­arch Exe­cu­ti­ve Agen­cy. Neither the Euro­pean Uni­on nor the gran­ting aut­ho­ri­ty can be held respon­si­ble for them.