MultiSensEAF: Multi-Sensor Systems for an optimized EAF Process Control

In the elec­tric steel­ma­king rou­te, the EAF pro­cess is the pha­se most cri­ti­cal in terms of ener­gy con­sump­ti­on, metal­lic los­ses, and cost. A main pro­blem is the lack of know­ledge regar­ding the cha­rac­te­ristics of the char­ged scrap and its mel­ting beha­viour insi­de the fur­nace as well as the con­di­ti­on of the slag and steel bath.

The Mul­ti­Sen­sE­AF pro­ject addres­ses the steel pro­cess and pro­cess-chain opti­miza­ti­on via instru­men­ta­ti­on, detec­tion of pro­per­ties of pro­ducts, model­ling, con­trol and auto­ma­ti­on, inclu­ding digi­ta­liza­ti­on, appli­ca­ti­on of big data, arti­fi­ci­al intelligence.

The over­all objec­ti­ve will be imple­men­ted by a three-step pro­cess. New­ly deve­lo­ped and addi­tio­nal off-the-shelf sen­sors (OES, acou­stic, load cells, came­ra) will be instal­led at indus­tri­al EAFs to crea­te inno­va­ti­ve mul­ti-sen­sor sys­tems moni­to­ring the cri­ti­cal aspects of the EAF pro­cess. Data coll­ec­ted by the­se mul­ti-sen­sor sys­tems will be com­pared with pro­cess KPIs and the acou­stic/OES-based inves­ti­ga­ti­on of the mel­ting evo­lu­ti­on, an opti­mi­zed scrap mix can be deter­mi­ned to redu­ce ener­gy los­ses due to an unde­si­ra­ble mel­ting pro­gress. The infor­ma­ti­on gai­ned will be exploi­ted by uti­li­zing a machi­ne lear­ning approach and will be incor­po­ra­ted into pro­cess con­trol and decis­i­on sup­port sys­tems, pre­ven­ting exces­si­ve oxi­da­ti­on or over­hea­ting of the steel. The deeper pro­cess know­ledge crea­ted by the mul­ti-sen­sor sys­tems and soft sen­sors will be uti­li­zed in KPI and model-based pro­cess manage­ment and optimization.

By rea­li­zing the pro­po­sed impro­ve­ments to the EAF pro­cess, the ener­gy and resour­ce con­sump­ti­on and by extend the cost of the steel pro­duc­tion can be decreased while a hig­her metal­lic yield and pro­duc­ti­vi­ty is achie­ved. The impro­ved effi­ci­en­cy is rela­ted to a signi­fi­cant reduc­tion of CO2 emis­si­ons and thus con­tri­bu­tes to glo­bal sus­taina­bi­li­ty and the Green Steel initia­ti­ve. The trans­fera­bi­li­ty of the results to other plants is ensu­red by appli­ca­ti­on of the mul­ti sen­sor sys­tem at two EAFs with dif­fe­rent elec­tri­cal sup­p­ly sys­tems (AC/DC), capa­ci­ties (80t/140t) and gene­ral characteristics.

New and addi­tio­nal sen­sors and pro­cess con­trol systems

To con­duct the Mul­ti­Sen­sE­AF pro­ject, seve­ral tech­no­lo­gies and methods are appli­ed. In the figu­re abo­ve an over­view of the new and addi­tio­nal sen­sors, the mul­ti-sen­sor sys­tem as well as the pro­cess con­trol sys­tems to be deve­lo­ped and tes­ted is presented.

The final sys­tems will com­pri­se of inno­va­ti­ve mul­ti-sen­sor sys­tems estab­lished by com­bi­ning exis­ting sen­sors with addi­tio­nal off-the-shelf and new pro­xi­mi­ty sen­sors. Based on the data deli­ver­ed by the hard­ware sen­sors, new soft sen­sors will be deve­lo­ped employ­ing part­ly also machi­ne lear­ning and AI methods to cha­rac­te­ri­se the scrap, moni­tor scrap melt­down, detect hot heel and slag con­di­ti­ons as well as moni­tor melt decar­bu­ri­sa­ti­on. The data pro­vi­ded by hard­ware and soft sen­sors will be employ­ed to deve­lop, imple­ment, and test inno­va­ti­ve scrap mix opti­mi­sa­ti­on and injec­tion opti­mi­sa­ti­on sys­tems in dyna­mic pro­cess control.

Over­view of the pro­ject part­ners involved