Skip to main content

La Metallurgia Italiana, n.9 settembre 2026

Page 1

La

Metallurgia Italiana

International Journal of the Italian Association for Metallurgy

n.09 Settembre 2026 Organo ufficiale dell’Associazione Italiana di Metallurgia. Rivista fondata nel 1909


La Metallurgia Italiana International Journal of the Italian Association for Metallurgy Organo ufficiale dell’Associazione Italiana di Metallurgia. House organ of AIM Italian Association for Metallurgy. Rivista fondata nel 1909

Direttore responsabile/Chief editor: Mario Cusolito Direttore vicario/Deputy director: Gianangelo Camona Comitato scientifico/Editorial panel: Marco Actis Grande, Ettore Anelli, Silvia Barella, Enrico Baroni, Paola Bassani, Shahab Bazri, Christian Bernhard, Massimiliano Bestetti, Wolfgang Bleck, Franco Bonollo, Irene Calliari, Riccardo Carli, Mariano Enrique Castrodeza, Emanuela Cerri, Vlatislav Deev, Andrea Di Schino, Donato Firrao, Piero Frittella, Berndt Kleimt, Carlo Mapelli, Susanne Michelic, Roberto Montanari, Marco Ormellese, Mariapia Pedeferri, Massimo Pellizzari, Annalisa Pola, Ulrich Prahl, Barbara Previtali, Dario Ripamonti Segreteria di redazione/Editorial secretary: Flynn Russo Comitato di redazione/Editorial committee: Federica Bassani, Gianangelo Camona, Mario Cusolito, Carlo Mapelli, Federico Mazzolari, Flynn Russo Direzione e redazione/Editorial and executive office: AIM - Via F. Turati 8 - 20121 Milano tel. 02 76 02 11 32 - fax 02 76 02 05 51 met@aimnet.it - www.aimnet.it Reg. Trib. Milano n. 499 del 18/9/1948. Sped. in abb. Post. - D.L.353/2003 (conv. L. 27/02/2004 n. 46) art. 1, comma 1, DCB UD Immagine in copertina: Shutterstock

Gestione editoriale e pubblicità Publisher and marketing office: siderweb spa sb Via Don Milani, 5 - 25020 Flero (BS) tel. 030 25 400 06 commerciale@siderweb.com - www.siderweb.com La riproduzione degli articoli e delle illustrazioni è permessa solo citando la fonte e previa autorizzazione della Direzione della rivista. Reproduction in whole or in part of articles and images is permitted only upon receipt of required permission and provided that the source is cited.

siderweb spa sb è iscritta al Roc con il num. 26116


La

Metallurgia Italiana

International Journal of the Italian Association for Metallurgy

n.09 Settembre 2026 Organo ufficiale dell’Associazione Italiana di Metallurgia. Rivista fondata nel 1909

Editoriale / Editorial

a cura di Prof.ssa Annalisa Pola, Università degli Studi di Brescia....................................................................... pag.05

Memorie scientifiche / Scientific papers Digitalizzazione e AI / Digitalization and AI

Cellular Automaton – Finite Volume model for the simulation of grain evolution during solidification P. Airoldi, R. Pirovano, A. Agne, V. Maguin, N.T. Niane ...................................................................................... pag.09

Manutenzione predittiva dei cuscinetti nei rulli immersi in linee di zincatura a caldo mediante realtà estesa e algoritmi di Intelligenza Artificiale A. Curci, G. Annicchiarico, M. Saccone, M. Pavan, A. Cecchinelli, F. Menchetti, V. Fantozzi, M. Della Santa,

n.09 Settembre 2026

I. Karakostas, N. Dimitriou, E. Sykianaki, K. Valakou, G. Margetis ....................................................................... pag.16

Anno 117 - ISSN 0026-0843

Attualità Industriale / Industry News Gemello digitale della colata continua: costruzione e calibrazione in ProCAST per la previsione di difetti e l’ottimizzazione di processo

R. Ottini, G. Mori, A. Mario, M. Bognolo, S. Gobbo, M. Marzaro, M. Bianchi, S. Buoro, F. Ricchini, C. Viscardi,

indice

L. Valente .................................................................................................................................................................................. pag.28

Sensor and digital twin solutions developed in DiGreeS project for improvement of scrap-based EAF steelmaking

B. Kleimt, B. Palm, G. Weides, K. Srivastava, F. van den Berg, F. Schrama, M. Heinrich, B. Wolter, S. Groenheide,

C. Noel, K. Winkler, F. Egger ................................................................................................................................................. pag.40

Laminazione / Rolling Towards workshop application of ring rolling simulations with embedded machine control K. van Putten, S. Stergianou, A. Gohr, A. Neumann, D. Michl, V. Horáček, M. Vindyš ....................... pag.52

New insights into the online LUS grain size measurements

D. Hoppe, T. Haschke, A. Sprock, C. Hassel, J. Hafer, L. Bäcke, J.-E. Thorberg, C. Jonsson, M. Malmström, F. Kneisel, M. Bärwald ...................................................................................................................... pag.58

Atti e notizie / AIM news

41° Convegno Nazionale AIM: ottant'anni al servizio della metallurgia in Italia ......... pag.71 Eventi AIM / AIM events ........................................................................................................................................ pag.74 Normativa / Standards ........................................................................................................................................... pag.78


International Forum

Tinplated Steels and Metals Packaging & Recycling >> more info

BERGAMO, 3-4 DECEMBER 2026 | CENTRO CONGRESSI GIOVANNI XXIII IFTSR 2026 will bring together industry leaders, researchers, recyclers, regulators, and brand owners to discuss the main technological, environmental, commercial, and regulatory challenges facing tinplate and metal packaging. The Forum will address the impact of anti-dumping measures, CBAM, energy costs, and raw material volatility on global supply chains and competitiveness. Key discussions will focus on advanced lacquer technologies, BPA-NI solutions, food contact compliance, corrosion mechanisms, and shelf-life performance. Special attention will be devoted to sustainability and circularity, including low-carbon production, eco-design, eco-labeling, recycled content valorization, and strategies for reducing the environmental footprint of metal packaging. The growing role of Artificial Intelligence and digital technologies in quality control, defect detection, predictive maintenance, and smart manufacturing will also be explored. Dedicated sessions will examine pre-consumer and post-consumer recycling, de-coating and sorting technologies, as well as the evolving regulatory framework related to bisphenols, PFAS, food contact materials, and the Packaging and Packaging Waste Regulation (PPWR). IFTSR 2026 aims to promote collaboration across the value chain and contribute to a shared roadmap for innovation, sustainability, regulatory compliance, and market resilience in the metal packaging sector.

DEADLINES

Submission of abstracts: .....................................................9 October 2026 Information on acceptance: .......................................... 23 October 2026 Opening online registration: ........................................... 23 October 2026 Submission of pdf presentations: ...........................16 November 2026 Early bird registration: .................................................... 16 November 2026 Submission of full papers (optional):..................After the Workshop

organised by

ORGANISING SECRETARIAT

Associazione Italiana di Metallurgia Via Filippo Turati 8 Milano - Italy t. +39 0276397770 . t. +39 0276021132 conference@aimnet.it www.aimnet.it

with the support of

>> more info


“

“… L’elevato numero di giovani tra i relatori e le relatrici [...] un segnale che non solo dimostra l’interesse delle nuove generazioni per la metallurgia, ma evidenzia anche come questo ambito possa continuare a crescere e affrontare i cambiamenti in atto con nuove energie e competenze.”

editoriale - editorial

“… The large number of young speakers […] this not only demonstrates the interest of younger generations in metallurgy, but also highlights the potential of the field to continue growing and addressing ongoing changes with new energy and expertise.”

Prof.ssa Annalisa Pola Università degli Studi di Brescia

METALLURGIA E DIGITALE

METALLURGY AND DIGITAL TECHNOLOGIES

Dal 9 all’11 settembre 2026 si è tenuto, presso l’Univer-

From 9 to 11 September 2026, the University of Brescia

sità degli Studi di Brescia, il 41° Convegno Nazionale di

hosted the 41st AIM National Conference. This year’s

AIM, un’edizione che ha assunto un significato speciale

edition had a special significance as, in addition to being

poiché, oltre a rappresentare uno dei principali appun-

one of the major events in the field, it coincided with

tamenti del settore, ha coinciso con l’80° anniversario di

AIM’s 80th anniversary. To celebrate this double anni-

AIM. Per celebrare questa doppia ricorrenza, si è scelto

versary, the technical and scientific programme was

di affiancare al programma tecnico-scientifico iniziative

complemented by cultural initiatives and social events,

culturali e momenti conviviali, organizzati con attenzio-

organised with a focus on social aspects, so as to pro-

ne anche alla dimensione sociale, così da offrire ulterio-

vide further opportunities to foster personal and pro-

ri occasioni per coltivare relazioni personali e profes-

fessional relationships and to reflect more deeply on

sionali e approfondire la riflessione sul contributo che

the contribution that the metallurgical community can

la comunità metallurgica può offrire alla collettività.

make to society.

Il Convegno ha registrato oltre 350 partecipanti e più di

The Conference attracted more than 350 participants

200 memorie proposte e si è caratterizzato per l’equili-

and over 200 submitted papers and was characterised

brio tra mondo accademico e industriale, con una platea

by a balanced representation of academia and industry,

composta da circa il 45% di persone provenienti dalle

with approximately 45% of attendees coming from uni-

università, circa il 46% dalle aziende e la quota restan-

versities, 46% from industry, and the remainder from

te da centri di ricerca e associazioni. Un bilanciamento

research centres and associations. An almost perfect

quasi perfetto, che riflette lo spirito di collaborazione

balance, reflecting the spirit of collaboration and inte-

e integrazione con cui il settore sta lavorando ed evol-

gration that is driving the development and evolution of

vendo.

the sector.

La Metallurgia Italiana - September 2026

pagina 5


editoriale - editorial Un altro dato interessante emerso dall’analisi del pro-

Another interesting aspect emerging from an analysis

gramma del Convegno è stato l’elevato numero di gio-

of the Conference programme was the large number

vani tra i relatori e le relatrici nelle diverse sessioni, con

of young speakers across the various sessions, with

quasi il 40% delle memorie illustrato da dottorandi,

almost 40% of the papers presented by PhD students,

assegnisti, studenti, ricercatori e giovani professioni-

research fellows, students, researchers, and young pro-

sti. Un segnale che non solo dimostra l’interesse delle

fessionals. This not only demonstrates the interest of

nuove generazioni per la metallurgia, nelle sue diverse

younger generations in metallurgy, in its many forms,

declinazioni, ma evidenzia anche come questo ambito

but also highlights the potential of the field to contin-

possa continuare a crescere e affrontare i cambiamenti

ue growing and addressing ongoing changes with new

in atto con nuove energie e competenze. E proprio per

energy and expertise. To recognise and encourage their

dare valore al loro entusiasmo, quest’anno si è voluto

enthusiasm, this year an award was introduced for the

introdurre un premio dedicato alle migliori presenta-

best presentations given by younger colleagues.

zioni dei colleghi più giovani.

The programme as a whole reflected the scale of the

Ma l’intero programma ha rispecchiato l’ampiezza delle

transformations currently taking place. The topics

trasformazioni in corso. I temi affrontati sono stati infatti

covered ranged from decarbonisation and the circular

numerosi e hanno spaziato dalla decarbonizzazione e

economy to traditional and advanced materials, from

dall’economia circolare ai materiali tradizionali e avan-

their characterisation and treatment to safety and digi-

zati, dalle loro caratterizzazione e trattamento, fino alla

talisation.

sicurezza e alla digitalizzazione.

Within this broad framework were the sessions devot-

In questo quadro si sono collocate le sessioni dedica-

ed to Artificial Intelligence, Digitalisation and Industry

te a “Intelligenza artificiale, Digitalizzazione e Industria

4.0, from which the three papers included in this issue

4.0”, dalle quali provengono le tre memorie raccolte in

were selected. Although addressing different scales

questo numero. Sebbene affrontino scale e problemi

and challenges, they share the common perspective of

differenti, condividono la prospettiva di trasformare dati

transforming data and models into useful tools to bet-

e modelli in strumenti utili per meglio comprendere i

ter understand phenomena, predict critical issues, and

fenomeni, prevedere le criticità e supportare la ricerca.

support research.

Il primo lavoro porta la digitalizzazione alla scala micro-

The first paper brings digitalisation to the microstruc-

strutturale attraverso un modello Cellular Automaton

tural scale through a Cellular Automaton Finite Volume

Finite Volume in grado di simulare nucleazione e cre-

model capable of simulating nucleation and dendritic

scita dendritica durante la solidificazione. Il contributo

growth during solidification. The contribution shows

rivela come la simulazione rappresenti sempre di più

how simulation is increasingly becoming a valuable

uno strumento di previsione utile ai ricercatori per ana-

predictive tool for researchers to investigate the influ-

lizzare l’influenza dei parametri di processo, con possi-

ence of process parameters, with potential applications

bili applicazioni sia in fonderia sia nella saldatura.

in both casting and welding.

Il secondo contributo riguarda la manutenzione predit-

The second contribution focuses on the predictive

tiva dei cuscinetti dei rulli immersi in una linea di zin-

maintenance of bearings supporting submerged rolls

catura a caldo, dove la misura diretta è particolarmente

in a hot-dip galvanising line, where direct measure-

complessa. La memoria illustra come segnali vibrazio-

ments are particularly challenging. The paper describes

nali, parametri di processo e algoritmi di apprendimen-

how vibration signals, process parameters, and ma-

to automatico siano stati combinati per stimare usura e

chine-learning algorithms were combined to estimate

vita utile residua, e mostra come la possibilità di antici-

wear and remaining useful life. It also shows how the

La Metallurgia Italiana - Settembre 2026

pagina 6


editoriale - editorial pare condizioni critiche riduca il rischio dell’insorgere

ability to predict critical conditions can reduce the risk

di situazioni potenzialmente pericolose e ad alto impat-

of potentially hazardous and high-impact situations,

to, sia per la produzione sia per la sicurezza.

both in terms of production and safety.

Il terzo lavoro descrive un gemello digitale di una mac-

The third paper describes a digital twin of a continuous

china di colata continua, costruito integrando geome-

casting machine, developed by integrating plant geom-

tria dell’impianto, parametri reali di processo e feno-

etry, actual process parameters, and thermo-metallur-

meni termo-metallurgici. Calibrato attraverso misure

gical phenomena. Calibrated using experimental mea-

sperimentali, il modello permette di analizzare porosità,

surements, the model enables the analysis of porosity,

segregazione e distribuzione dell’idrogeno, di valutare

segregation, and hydrogen distribution, as well as the

soluzioni operative alternative, mostrando così un utile

evaluation of alternative operating solutions. It there-

strumento predittivo per migliorare qualità e resa, ridu-

fore represents a useful predictive tool for improving

cendo prove in impianto, scarti e costi.

quality and yield, while reducing plant trials, scrap, and

Nel loro insieme, questi lavori presentati al Convegno

costs.

mostrano come dal processo alla manutenzione, fino

Taken together, these papers presented at the Con-

alla microstruttura, il digitale costituisca un ponte tra ri-

ference show how, from processes and maintenance

cerca e produzione e uno strumento per l’innovazione.

to microstructure, digital technologies can provide a bridge between research and production and serve as a

Buona lettura di queste prime memorie!

powerful tool for innovation. Enjoy reading this first selection of papers!

La Metallurgia Italiana - September 2026

pagina 7


ICRF 2026 13-15 October | Bardolino . Italy

www.aimnet.it/icrf2026

Advancing the Future of Steel Manufacturing Through Innovation and Sustainability

ABOUT ICRF

The global steel industry is at a pivotal moment, navigating the dual challenges of technological advancement and environmental responsibility. ICRF 2026 will serve as a critical forum for industry leaders, researchers, and innovators to explore the latest breakthroughs in improvement of product quality, process optimization, digital transformation, and AI-driven manufacturing. By bridging cutting-edge research with industrial applications, ICRF 2026 aims to shape a more sustainable, competitive, and innovative future for ingot casting, forging and ingot rolling.

CONFERENCE CHAIRPERSONS Jacopo Longhi Vienna - GIVA Group, Italy Alessandra Saleri - Forge Fedriga, Italy

TOPICS

The Conference offers a leading platform to present papers on one or more of the following general topics: • Ingot casting and remelting • Heat treatment • Rolling, forging and design of hot deformation processes • Production scheduling • Measurement, mechanical testing, non-destructive testing • Welding • Oil & gas industry and power generation applications • Aerospace applications • Numerical simulation, artificial intelligence and digital innovation • Environment, emissions and new production technologies

REGISTRATION OF THE ATTENDEES

To register, please go to the “registration of attendees” section at www.aimnet.it/icrf2026 CONFERENCE REGISTRATION FEES INCLUDE • Admittance to technical sessions & exhibition • Welcome aperitif on October 13 • Coffee break and Lunches

• Conference electronic proceedings • Conference dinner on October 14

Ingot Casting Rolling Forging 20

EXHIBITION & SPONSORSHIP

As an integral element of the event, the Conference will feature an exhibition, that will enable excellent exposure for products, technologies, innovative solutions or services. For any further information contact Siderweb: commerciale@siderweb.com

ORGANISING SECRETARIAT

5th International Conferen t. +39 0276021132 • +39 0276397770 icrf2026@aimnet.it • www.aimnet.it/icrf2026

SPONSORS Platinum Sponsor

Organised by

Golden Sponsor

With the support of


Scientific papers - Digitalization and AI

DOI 10.36146/2026_09_09

La

Italiana

Metallurgia

Cellular Automaton–Finite Volume model for the simulation of grain evolution during solidification P. Airoldi, R. Pirovano, A. Agne, V. Maguin, N.T. Niane

A model based on Cellular Automaton - Finite Volume (CAFV) for simulating grain nucleation and anisotropic growth during the solidification of alloys has been implemented in commercial software. This model operates on a sub-grid scale, capturing heterogeneous nucleation and competitive dendritic growth while incorporating orientation-depen-

International Journal of the Italian Association for Metallurgy

dent growth kinetics, thereby providing a physically consistent representation of grain selection mechanisms in directional solidification. Validation against cases reported in the literature confirms the model’s accuracy, offering a robust tool for optimizing process parameters in single-crystal castings and other advanced castings, with potential extensions to welding and additive manufacturing processes.

KEYWORDS: GRAIN EVOLUTION; DIRECTIONAL SOLIDIFICATION; SIMULATION. INTRODUCTION

The numerical simulation of dendritic grain structures

formed during solidification becomes important to re-

duce the risk of microstructural defects that negatively impact mechanical properties of the casting.

New methodologies including more detailed physics have been developed in last decades to support this analysis better than the traditional predictive criteria based CAFV is one of these methodologies: it is based on the

026

3D Cellular Automaton (CA) model solved on a micro-

scopic computational grid to reproduce nucleation and growth of grains at large scales (cm). This CA grid is su-

perimposed to a coarser finite volume (FV) mesh where the solution of the heat flow equation is computed; the

n. 6 giugno 2020

Organo ufficiale

dell’Associazione Italiana

di Metallurgia.

Rivista fondata nel 1909

solely on the thermal field and generally limited.

Paolo Airoldi, Raul Pirovano Flow Science Mediterranea, Como, Italy

Aboubakry Agne, Vincent Maguin, Ngadia Taha Niane Safran Tech, Châteaufort, France

coupling of these models constitutes the CAFV model. This work presents its implementation in the commercial

nce

software FLOW-3D® CAST. Modelling and results are de-

scribed in the following sections. MODELLING The CA modelling through simple local laws was first proposed in [1]. The model was coupled to the thermal solution obtained from a finite element (FE) mesh, leading to the first versions of the Cellular Automaton-Finite Element (CAFE) model [1,2]. Similarly, the CAFV (Cellular

La Metallurgia Italiana - September 2026

pagina 9


Memorie scientifiche - Digitalizzazione e AI Automaton – Finite Volumes) model couples the CA mod-

ber of cells in CA grid may require significant memory.

el with the solution of the heat transfer equation based on

Minimizing data storage is, therefore, a priority.

an FV mesh, and includes the representation of grain en-

A state index represents the state of a cell: -1 means the

velope through an octahedral form, the heterogeneous

cell is passive, 0 means the cell is liquid; otherwise, it co-

nucleation based on nucleation sites and the growth ki-

incides with the grain index; the cells having the same

netics along preferential directions of the dendrites.

grain index belong to the same grain.

At the initial time of the simulation, the uniform CA grid

Then, the nucleation sites are identified. The heteroge-

is generated: it requires a structured block of cubic cells

neous nucleation is a phenomenon of instantaneous for-

and the unique cell size l CA has to be of the order of the

mation of a solid grain occurring in the liquid bulk or at

secondary dendrite arm spacing; this way, this grid may be

the surface of the mold, when a critical undercooling is

much finer than FV mesh. The CA cells are grouped into

reached. Nucleation is based on the stochastic instanta-

CA blocks (a block contains the centre of its associated

neous nucleation model described in [2], which is based

cells), and these blocks are grouped into CA windows, that

on the definition of nucleation sites. They can be defined

are user-defined parallelepipeds identifying the zones of

in different ways: as points, surfaces or volumes. In the

interest where the CA grid overlaps the FV mesh: the nu-

first case, the properties of the nuclei are assigned deter-

cleation and growth can occur only inside them.

ministically. In the other cases, the number of nucleation

The use of blocks and windows allows to limit the compu-

sites is calculated as n maxV, where n max is the maximum grain

tational domain and to deactivate dynamically the parts of

density and V is a given volume (or area, for surfaces),

the domain which are still fully liquid or have completely

then they are randomly distributed among the CA cells.

solidified. This leads to save computational time [2] and is

The grain density depends on the undercooling ∆T, which

essential because without such constraints a large num-

is defined as:

where TL is the liquidus temperature and T is the tempera-

tribution described by the following probability density

where ∆Ta is the mean undercooling and ∆Tσ the standard

scribed Gaussian distribution. If a cell contains more than

ture of the nucleation site at the cell-centre.

function [1,5]:

This dependency is modelled through a Gaussian dis-

deviation. For a given alloy, these parameters are deter-

one nucleation site, only the smaller critical nucleation

mined experimentally.

undercooling is used.

A critical nucleation undercooling is attributed to each

The method of Box-Muller [4] is used for this operation:

cell containing a nucleation site, according to the pre-

where ∆Tnucl=TL-Tnucl is the critical undercooling, Tnucl is the

A new grain forms when the undercooling ∆T of a liquid

critical nucleation temperature, and u1 and u2 are two ran-

cell containing nucleation sites exceeds a prescribed crit-

dom numbers in (0,1).

ical nucleation undercooling:

La Metallurgia Italiana - Settembre 2026

pagina 10


Scientific papers - Digitalization and AI At this instant, a crystallographic orientation is given by a

the CA grid at nucleation sites and at the locations of the

set of 3 Euler angles.

growing grains. This allows to update the grain growth ve-

After the initialization of the nucleation sites, the resolu-

locity and increase their size, computing the growth and

tion in time and space starts. Typically, the time step used

capture dynamics; at the end of the iteration the nucle-

for the FV mesh is too big for CA grid: the CA advance-

ation of new sites is checked.

ment requires an inner loop over sub-time-steps, where

Figure 1 shows the structure of the CA algorithm, repro-

at each iteration the FV temperature is interpolated to

ducing the grain evolution, during one FV time step.

Fig.1 - Scheme of the CA algorithm. t and ∆t are micro-time-step and time-step, respectively / Schema dell’algoritmo

CA. t e ∆t sono rispettivamente il micro-time-step e il time-step.

The grain evolution modelling is now described in more

called micro-time-steps, used for the integration of the

detail. The growth of the dendritic network during a time-

growth kinetics velocity v and computed in the following

step cannot exceed the cell spacing. Consequently, the

way [2]:

time-step is subdivided into several sub-time-steps,

where α is a security coefficient to avoid having a growth

its own octahedron. Its growth is reproduced through a ki-

instant, and vmax=v(∆Tmax) is the velocity at maximum un-

kinetics law, which represents the velocity of the dendrite

of the envelope that exceeds the cell without being able to capture it and to manage multiple captures at the same

dercooling of all the growth cells in the domain. The value

netics law, guaranteeing the preferential growth directions 〈100〉 and the irrelevance of the grid on the orientation. The tips, is a function of the undercooling ∆T of the cell, and is

of α is in [0,1]. A grain is supposed to have an envelope of

usually fitted (in order to speed up the calculations) with a

octahedral shape, which is modified locally in order to re-

polynomial approximation, for instance:

where a is a coefficient and n is the exponent.

i.e. the diagonals of the octahedron, is obtained by inte-

Within a micro-time-step δt , the increment L of the radii,

grating the kinetics velocity of the dendritic tips [2]:

The evolution of the octahedron involves the neighbour-

octahedron envelope encloses the centre of a neighbour-

ing cells and, through the dynamic of capture the growth

ing liquid cell, this cell is captured by the grain. The de-

of the grain, proceeds inside the CA domain. When the

centred octahedron growth algorithm described in [3] is

flect the growth of the grain itself. Every involved cell has

La Metallurgia Italiana - September 2026

pagina 11


Memorie scientifiche - Digitalizzazione e AI used to avoid a strong anisotropy effect due to the cubic

VERIFICATION RESULTS

cell lattice.

The CAFV model has been verified against some analyt-

When a liquid cell starts to solidify, it becomes mushy. A

ical/experimental/numerical cases described in [1], [3],

grain is then associated with the cell, and the mushy zone

[4]. A couple of them are reported briefly here.

fraction is defined as in [5]. The modelling is completed taking into consideration the

The first test case presents a validation of the growth of

type of coupling CA-FV: in weak coupling, CA scheme is

2D multigrain, measuring the competition between the

used just as “output”, without affecting the temperature

grains [4]. It involves a 2D domain (mm2): 0 ≤ x ≤ 4.14 and

field and solid fraction in the FV mesh; in full coupling,

0 ≤ y ≤ 8.31. 3 nuclei are placed at the boundary y=0, the

instead, a two-way transfer of the computed variables is

properties are described in table 1.

executed between the two computational domains. Tab. 2 - Properties of grains for test 1 / Proprietà dei grani nel test 1. Alloy

succinonitrile-1.3%wt acetone

Nucleation undercooling Growth kinetics

m/s

Orientation for x ≤0.71 mm

4°

Orientation for 0.71 ≤ x ≤3.43 mm

30°

Orientation for 3.43 ≤ x ≤ 4.14 mm

11°

The thermal conditions are imposed thermal gradient of

figure 2); its theoretical value is 19°. The predicted results

rate of -0.1634 K/s. The aim is to measure the angle θd be-

in figure 2 (on the right), compared with the reference ones

1900 K/m along the vertical direction and imposed cooling

tween the central and right grains (see image on the left in

Fig.2 - Reference (left) and numerical (right) results / Risultato di riferimento (sinistra) e

numerico (destra).

La Metallurgia Italiana - Settembre 2026

using lCA=25μm (cell density of 1600 mm-2) are presented

(on the left).

Fig.3 - θd in function of the cell density /

θd in funzione della densità di celle.

pagina 12


Scientific papers - Digitalization and AI Figure 3 shows the convergence analysis, where different

In a 3D cylindrical domain with radius of 35 mm and height

used to study the evolution of θd. The results confirm that

nucleation occur, under an imposed profile temperature

cell sizes lCA (reported in function of the cell density) are

the numerical solution is very close to the experimental

one; the predicted angle θd matches the theoretical value even using a low cells density. The second test case pres-

ents the growth of 3D multigrain and the measurement of

of 173 mm the surface nucleation at x=0 and the volumetric at z=0 and an imposed heat flux at z=173 mm. The alloy is

AlSi7. Properties and parameters are reported in tables in [4].

the temperature [4].

Fig.4 - Grains evolution (reference on the left, numerical on the right) / Evoluzione dei grani

(riferimento sulla sinistra, numerico sulla destra).

Figure 4 shows the grain evolution in 3 instants, compar-

perimental (grey) and numerical (colored) temperature

ing the results (right) against the reference ones (left),

evolution measured at 6 equidistant probes along the ver-

while figure 5 reports the good agreement between ex-

tical axis.

Fig. 5 - Comparison between experimental (grey) and numerical (coloured) temperature profiles at 6 probes /

Confronto tra profilo di temperature sperimentale (grigio) e numerico (colorato) in 6 punti di campionamento.

La Metallurgia Italiana - September 2026

pagina 13


Memorie scientifiche - Digitalizzazione e AI Finally, a qualitative industrial case is mentioned. A ten-

is AM1, while its geometry, created ad hoc to study para-

sile test manufactured was conducted at Safran Aircraft

sitic grains that prevent the production of a single-crystal

Engines laboratory, following a Bridgman process to

piece, is shown in figure 6.

achieve a controlled solidification. The alloy of the casting

Fig. 6 - Geometry of industrial case / Geometria del caso industriale. Solidification starts at the base: many grains form and

single grain prevails. The solidification continues in the

grow, undergoing strong competition. In the selector

plate, where parasitic grains form can be identified by

only a few grains remain, reaching the cylinder where a

observing the surface of the plate.

Fig. 7 - Details of the base, the selector and the plate with parasitic grains (compared to experimental case) / Dettagli

della base, del selettore e del piatto con i grani parassiti (paragonato ad un caso sperimentale).

The comparison of the grain distribution between the ex-

industrial scientists to study the process parameters in

perimental result and the simulation shows a reasonable

their casting applications. Furthermore, this model can

agreement, as in figure 7; surely, exact correspondence is

be extended for other kinds of applications like weld-

not possible, due to the random nature of the solidifica-

ing. This version, implemented in commercial software,

tion. The shape, size and number of the parasitic grains are

demonstrates very good performance, satisfying the the-

influenced by the alloy-dependent and hard-to-measure

oretical features of the CAFV model through verification

properties as well: the CAFV model is useful to predict

tests from literature. Finally, a realistic case in industrial

their impact and calibrate them.

conditions gave good agreement as well.

CONCLUSION The CAFV model is a performing simulation tool that enables detailed predictions of microstructure, useful for

La Metallurgia Italiana - Settembre 2026

pagina 14


Scientific papers - Digitalization and AI REFERENCES [1] [2] [3] [4] [5]

Ch.-A. Gandin, M. Rappaz, A coupled finite element-cellular automaton model for the prediction of dendritic grain structures in solidification processes, Acta Metall. Mater. 42 2233, 1994. Ch.-A. Gandin, J.-L. Desbiolles, M. Rappaz, Ph. Thevoz, A three-dimensional cellular automation-finite element model for the prediction of solidification grain structures. Metallurgical and Materials Transactions A, 30(12):3153–3165, 1999. Ch.-A. Gandin, M. Rappaz, A 3d cellular automaton algorithm for the prediction of dendritic grain growth. Acta Materialia, 45(5):2187– 2195, 1997. T. Carozzani, Développement d’un modèle 3D Automate Cellulaire-Éléments Finis (CAFE) parallèle pour la prédiction de structures de grains lors de la solidification d’alliages métalliques. PhD thesis, 2012. H. Ben Hamouda, Modelling and Simulation of solidification structure in nickel-based superalloys: application AM1. Thesis, Ecole Nationale Supérieure des Mines de Paris, September 2012.

Il modello Cellular Automaton per la simulazione diretta dei grani durante la solidificazione Un modello basato su Cellular Automaton - Volumi Finiti (CAFV) per la simulazione della nucleazione dei grani e della crescita anisotropica durante la solidificazione delle leghe è stato implementato in un software commerciale. Tale modello opera su una scala sub-griglia, catturando la nucleazione eterogenea e la crescita dendritica competitiva e incorporando una cinetica di crescita dipendente dall’orientamento, fornendo così una rappresentazione fisicamente coerente dei meccanismi di selezione dei grani nella solidificazione direzionale. La validazione rispetto a casi di letteratura conferma la bontà del modello, offrendo uno strumento robusto per l’ottimizzazione dei parametri di processo nei getti monocristallini e in altri getti avanzati, con possibili estensioni ai processi di saldatura e di produzione additiva.

PAROLE CHIAVE: EVOLUZIONE DEI GRANI; SOLIDIFICAZIONE DIREZIONALE; SIMULAZIONE.

TORNA ALL'INDICE >

La Metallurgia Italiana - September 2026

pagina 15


Memorie scientifiche - Digitalizzazione e AI

DOI 10.36146/2026_09_16

Manutenzione predittiva dei cuscinetti nei rulli immersi in linee di zincatura a caldo mediante realtà estesa e algoritmi di Intelligenza Artificiale A. Curci, G. Annicchiarico, M. Saccone, M. Pavan, A. Cecchinelli, F. Menchetti, V. Fantozzi, M. Della Santa, I. Karakostas, N. Dimitriou, E. Sykianaki, K. Valakou, G. Margetis

Nelle linee di zincatura a caldo, i rulli immersi guidano il nastro attraverso il bagno di zinco e operano a temperature prossime a 470 °C e velocità fino a 180 m/min. Queste condizioni accelerano l’usura dei cuscinetti di strisciamento e possono causare fermi non programmati. Il caso studio, sviluppato presso Acciaierie d’Italia nell’ambito del progetto INDUX-R, integra accelerometri triassiali, parametri di processo e modelli di apprendimento automatico per stimare lo stato di usura e la vita residua dei cuscinetti. I risultati sono resi disponibili tramite strumenti di realtà estesa: VR per la formazione degli operatori e AR per il monitoraggio e il supporto alla manutenzione in campo.

PAROLE CHIAVE: MANUTENZIONE PREDITTIVA; INTELLIGENZA ARTIFICIALE; REALTÀ ESTESA; ZINCATURA A CALDO. INTRODUZIONE La manutenzione predittiva basata sull’analisi delle vibrazioni rappresenta uno dei punti fondamentali per prevenire rotture critiche di sistemi rotanti. La raccolta e l’elaborazione dei dati avviene tramite accelerometri, i cui segnali vengono elaborati in modo evoluto tramite analisi nel dominio del tempo e della frequenza, FFT, envelope analysis e tecniche tempo-frequenza come la trasformata wavelet [1]. Tali metodologie consentono di correlare le specifiche caratteristiche del segnale con fenomeni di usura e con le cause di rottura. Non sempre tali tecniche sono però applicabili come nel caso in esame dove è negata la possibilità di installare gli accelerometri nei punti più vicini agli organi rotanti e dove i cuscinetti usati sono di tipo non convenzionale. L’utilizzo degli algoritmi di machine learning ha rivoluzionato l’approccio alla manutenzione predittiva basata su vibrazioni. In particolare, le tecniche di riduzione della dimensionalità, come la Principal Component Analysis (PCA) [2], sono ampiamente

Antonio Curci, Giorgio Annicchiarico, Marco Saccone, Matteo Pavan, Andrea Cecchinelli Acciaierie d’Italia, Taranto (Italia)

Fernando Menchetti, Valerio Fantozzi, Martina Della Santa RINA Consulting - CSM S.p.A., Castel Romano (Italia)

Iason Karakostas, Nikolaos Dimitriou Centre for Research & Technology, Hellas (CERTH), Thermi-Salonicco (Grecia)

Eirini Sykianaki, Katerina Valakou, George Margetis

Institute of Computer Science, Foundation for Research and Technology-Hellas (FORTH), Creta (Grecia)

utilizzate per estrarre le caratteristiche più rilevanti dai segnali vibrazionali e migliorare le prestazioni dei modelli predittivi. A ciò si aggiunge l’utilizzo delle Convolutional Neural Networks (CNN) che rappresentano una delle tecniche più promettenti per l’analisi dei segnali vibrazionali. Tali modelli sono in grado di apprendere automati-

La Metallurgia Italiana - Settembre 2026

pagina 16


Scientific papers - Digitalization and AI camente le caratteristiche rilevanti direttamente dai dati

nel sito di Genova. Il processo di zincatura è uno dei più

grezzi o da rappresentazioni tempo-frequenza come gli

diffusi per la protezione anticorrosiva dei materiali metal-

spettrogrammi. Studi recenti mostrano che modelli CNN

lici ferrosi, in particolare dell’acciaio, e avviene tramite il

applicati alla diagnosi di guasti nei rotori possono rag-

riscaldamento del nastro in un forno ad atmosfera ridu-

giungere accuratezze superiori al 99% in condizioni con-

cente per prevenire l’ossidazione superficiale. Questo

trollate, mantenendo prestazioni elevate anche in presen-

consente di portare il nastro alla temperatura richiesta

za di variazioni operative [3]. L’addestramento delle reti

dalle specifiche metallurgiche e, dopo il raffreddamento

neurali, tuttavia, è vincolato a una quantità di dati che non

controllato, di farlo entrare nella vasca a una temperatura

è sempre disponibile (come nel caso dei cambi dei cusci-

di pochi gradi superiore a quella del bagno di zinco (450–

netti della zincatura). In questo caso vengono in aiuto altre

470 °C). Nel ba-gno avviene il rivestimento vero e proprio

tecnologie come le reti DESN (Deep Echo State Network)

del materiale; l’immersione del nastro e la correzione di

appartenenti alla famiglia del reservoir computing che uti-

eventuali sbandamenti sono gestite mediante i rulli im-

lizzano una rete neurale ricorrente con pesi fissi nel “re-

mersi. Il sistema è illustrato in figura 1, dove sono visibili

servoir”, mentre solo i pesi di output vengono addestrati,

il rullo principale e i rulli ausiliari impiegati, ossia il rullo

rendendo il modello computazionalmente efficiente. Le

correttore e il rullo stabilizzatore. I cuscinetti installati su

varianti profonde (DESN) estendono questo approccio

questi rulli costituiscono l’oggetto del caso studio svilup-

anche in presenza di un numero contenuto di dataset di

pato nel progetto INDUX-R [5] finanziato dal programma

addestramento migliorando la capacità di modellazione

Horizon (Horizon-CL4-2023-Human-01-22) nell’ambito

dei segnali complessi, risultando particolarmente efficaci

dell’accordo di sovvenzione n. 101135556.

nella prognostica dei guasti nei sistemi rotanti [4].

La presente pubblicazione riflette esclusivamente le opi-

Il caso studio di Acciaierie d’Italia (ADI) si inserisce a pie-

nioni degli autori. La Commissione europea non è respon-

no nell’ottica della manutenzione predittiva. Il tema esa-

sabile dell’uso che potrà essere fatto delle informazioni in

minato è relativo alla sostituzione dei cuscinetti dei rulli

essa contenute.

immersi della linea di zincatura a caldo ZIN/5, presente

Fig.1 - Schema e foto relativi ai rulli immersi della zincatura a caldo / Scheme and photos of sink

rolls for the hot dip galvanizing line.

Per ciascun rullo, sono montati dei cuscinetti (di striscia-

alla velocità della linea che può raggiungere i 180 m/min.

mento con inserti ceramici soggetti a usura) che operano

Attualmente, gli interventi di sostituzione dei cuscinetti

in condizioni estremamente critiche, a causa della tem-

sono periodici e la previsione di una rottura accidentale

peratura elevata (>450°C), della presenza di zinco fuso

risulta difficile perché affidata alla sensibilità dell’opera-

altamente corrosivo e delle vibrazioni elevata associata

tore di manutenzione; ciò può comportare una fermata

La Metallurgia Italiana - September 2026

pagina 17


Memorie scientifiche - Digitalizzazione e AI

accidentale della linea con aumento dei costi e rottama-

implementare il sistema di monitoraggio in continuo del-

zione del nastro in lavorazione.

le vibrazioni, è stato effettuato uno studio preliminare per definire la posizione dei sensori che dovevano essere in-

Analisi preliminare

stallati in corrispondenza della linea, poiché non era pos-

Per risolvere i limiti attuali di previsione della rottura, è

sibile installare gli accelerometri direttamente a contatto

stato sviluppato dal RINA-CSM con il supporto di ADI, un

con i cuscinetti. Sono state effettuate delle campagne di

modello di previsione della vita utile dei cuscinetti basa-

misurazione (a differenti valori di velocità della linea) av-

to sul monitoraggio in continuo delle vibrazioni dei rul-

valendosi di accelerometri portatili triassiali, presenti in

li non sui bracci immersi ma sulla struttura, tramite due

differenti zone dei rulli immersi (figura 2).

accelerometri triassiali posizionati a bordo macchina. Per

Fig.2 - Posizioni di misurazione delle vibrazioni sui due lati dell’assieme rullo immerso e rullo stabilizzatore /

Vibration measurement positions on both sides of the sink roll and stabilization roll assembly.

Fig.3 - Confronto tra la motion amplification e gli accelerometri nella zona del rullo immerso (punti 2 e 3) a 100 m/ min / Comparison between motion amplification and accelerometers in the area of the sink roll (points 2 and 3)

at 100 m/min.

La Metallurgia Italiana - Settembre 2026

pagina 18


Scientific papers - Digitalization and AI È stata effettuata un’analisi in frequenza, filtrando i segna-

rometrici (figura 3) ha evidenziato differenze significative

li con valori inferiori a 25 Hz ed elaborando i grafici degli

nella risposta dinamica delle diverse zone analizzate, con-

spettri, come è mostrato nella figura 3. Considerando la

sentendo di individuare le aree più sensibili alla trasmis-

variazione della velocità del nastro e analizzando i com-

sione delle vibrazioni. L’analisi ha inoltre confermato che

portamenti della struttura del rullo immerso (roll frame)

le vibrazioni del rullo correttore non presentano picchi

nelle posizioni ai lati della linea identificate con LO (lato

sostanzialmente diversi da quelli osservati sugli altri rulli,

operatore) e LM (lato motore) emerge come sia per gli assi

orientando la scelta definitiva verso l’installazione di due

delle componenti radiali sia per l’asse longitudinale, si ha

accelerometri nelle posizioni ritenute più rappresentative

un aumento delle ampiezze a partire dalla quinta armonica

e compatibili con i vincoli di temperatura e accessibilità.

(rispetto alla frequenza di rotazione del rullo); per la componente assiale, ciò è legato probabilmente più a un disal-

Utilizzo del sistema di acquisizione e sviluppo del mo-

lineamento angolare dovuto all’usura, mentre per quelle

dello per la previsione della vita residua

radiali all’usura diretta dei cuscinetti. Oltre alla velocità, è

Il sistema di acquisizione installato in campo è stato pro-

stata effettuata un’analisi di sensibilità sul tiro del nastro

gettato per acquisire segnali da due accelerometri trias-

in corrispondenza dei rulli, dove è stato osservato come

siali e trasferirli a un server di elaborazione dedicato, in-

un suo aumento comporta una variazione delle frequenze

tegrandoli con informazioni di processo. La scelta della

delle componenti rilevate a un tiro minore. Per completa-

posizione e dell’ancoraggio dei sensori è derivata dallo

re lo studio ci si è avvalsi inoltre di una tecnica basata sulla

studio precedentemente descritto.

motion amplification [6] per valutare il comportamento

Alcune componenti del sistema di acquisizione sono ri-

generale della struttura e decidere con maggiore preci-

portate in figura 4.

sione le posizioni definitive dell’installazione dei sensori,

Le vibrazioni acquisite dai sensori installati sul gruppo rul-

rispettando i vincoli di rilevazione accettabile delle vibra-

lo sono integrate con i principali parametri di processo,

zioni e una temperatura di esercizio inferiore ai 60°C. La

quali tensione e velocità del nastro, temperatura del ba-

motion amplification consente di rendere visibili piccoli

gno e geometria del prodotto. Questa integrazione con-

movimenti o variazioni temporali impercettibili in una se-

sente di contestualizzare il segnale rispetto alle condizio-

quenza video, applicando una decomposizione spaziale e

ni operative, distinguendo le variazioni associate allo stato

un filtraggio temporale dei fotogrammi, seguiti dall’ampli-

del cuscinetto da quelle indotte dalla linea, e costituisce la

ficazione del segnale nella banda di frequenza di interes-

base per costruire il dataset storico utilizzato nella stima

se. l confronto tra i video amplificati e gli spettri accele-

della vita utile residua.

Fig.4 - Foto del sistema di acquisizione delle vibrazioni (sensore posizionato sul lato LO e server per acquisizione dei dati) / Photos of the vibration acquisition system (sensor positioned on the LO side and data acquisition server).

Cuscinetti con inserti ceramici e boccole (fissate alle parti

superfici di contatto non è praticabile: sensori a contatto

terminali dei rulli) operano nello zinco fuso a 450-470 °C

con il bagno sarebbero esposti a shock termici, corrosio-

in presenza di scorie, gradienti termici e accessibilità fisica

ne e adesione dello zinco, con rischio di interferenza con

limitata. Di conseguenza, la strumentazione diretta delle

il moto del nastro e la qualità del rivestimento. La misura è

La Metallurgia Italiana - September 2026

pagina 19


Memorie scientifiche - Digitalizzazione e AI pertanto indiretta, rilevata da strutture accessibili e mec-

mancato utilizzo della vita disponibile o una loro sostitu-

canicamente collegate al gruppo rullo. L’aumento del

zione tardiva. Si è definito un indice di usura che consente

gioco tra boccole e cuscinetti modifica la risposta dina-

di pianificare l’intervento sulla base del degrado osserva-

mica dell’assieme rullo. Gli urti e le componenti armoni-

to, mantenendo un margine di sicurezza rispetto alla fine

che risultanti si propagano attraverso i perni e le strutture

vita teorica. Per fare ciò, è stata considerata una pipeline

di supporto fino ai punti di misura, senza che il processo

che elabora congiuntamente i segnali vibrazionali e i para-

venga perturbato [7].

metri di processo per ottenere un unico indice operativo dello stato del cuscinetto.

Elaborazione dei dati e stima dello stato di usura

Il flusso di calcolo è organizzato in cinque livelli funzio-

L’obiettivo è stimare in modo progressivo la vita utile del

nali: acquisizione, sincronizzazione e controllo della qua-

cuscinetto consumato e fornire un supporto quantitativo

lità, correzione delle condizioni operative, costruzione

alla pianificazione della manutenzione, evitando sosti-

dell’indice di usura, stima della vita residua e supporto alla

tuzioni anticipate o tardive che possono determinare un

manutenzione. L’architettura è riportata in figura 5.

Fig.5 - Pipeline per la stima dell’usura e il supporto alla manutenzione: dai dati acquisiti all’indice unificato e alla

stima della vita residua / Pipeline for wear estimation and maintenance support: from acquired data to the unified

index and remaining-life.

Preparazione e qualità dei dati

delle variabili di processo è gestita distinguendo tra dati

I dati dei sensori lato motore (LM) e lato operatore (LO)

osservati, interpolati (nel caso di brevi interruzioni) e sti-

sono sincronizzati con le variabili di processo mediante

mati. Infatti, per i periodi più estesi si utilizzano condizioni

associazione temporale, mentre i periodi di fermo sono

operative di riferimento ricavate dalle campagne storiche.

riconosciuti attraverso l’intensità vibrazionale ed esclusi

Questa gestione e il passaggio dalle sorgenti alle caratte-

dall’accumulo delle ore operative e del danno.

ristiche validate è illustrato in figura 6.

Per evitare anomalie sul calcolo dell’usura, la continuità

Fig.6 - Acquisizione, sincronizzazione e trattamento della qualità dei dati prima dell’estrazione delle caratteristiche di degrado / Data acquisition, synchronization and quality handling before degradation feature extraction.

È noto che le vibrazioni dipendono dallo stato del cusci-

ni di tiro e velocità possono modificare il segnale anche in

possibile osservare dall’analisi preliminare; infatti, variazio-

Per separare questi effetti, una regressione robusta stima

netto e dalle condizioni istantanee della linea, come è stato

La Metallurgia Italiana - Settembre 2026

assenza di degrado.

pagina 20


Scientific papers - Digitalization and AI la risposta vibrazionale associata alle condizioni operative, cosicché la componente attribuibile al processo venga

sottratta dal segnale misurato. Questo approccio evita la divisione diretta delle vibrazioni per il carico, che potrebbe amplificare il rumore durante i transitori o in condizioni operative ridotte.

Le caratteristiche corrette vengono confrontate con il

comportamento iniziale del cuscinetto mediante statisti-

che robuste. L’analisi considera indicatori legati all’intensità vibrazionale, all’impulsività e alla forma della distribuzione del segnale. Poiché il componente monitorato è un

cuscinetto radente, non vengono utilizzate le frequenze caratteristiche proprie dei cuscinetti volventi [8].

Per calcolare un indice di usura unificato RUL (Remaining

Useful Life), si considera in prima istanza la combinazione

delle variazioni delle vibrazioni misurate sui tre assi. Ciò

è il carico normale equivalente ricavato dalle

variabili di processo,

bile la stima anche in caso di temporanea assenza di uno

dei segnali. Eventuali disaccordi tra le misure restano disponibili come diagnostica separata. È noto che l’usura di un cuscinetto radente dipende dal carico applicato e dallo

scorrimento accumulato; una formulazione che interpreta ciò è ispirata alla legge di Archard, nella quale il volume

di usura è proporzionale al prodotto tra carico normale e distanza di scorrimento, a parità di materiale e regime tri-

bologico [9]. Poiché il carico normale e la velocità di scorri-

mento nel contatto immerso non possono essere misurati direttamente, vengono sostituiti da grandezze equivalenti

derivate dal tiro, dalla geometria del rullo e dalla velocità della linea. L’esposizione non rappresenta pertanto una

Indice di usura unificato

dove

riduce la sensibilità al rumore locale e mantiene disponi-

è una velocità di scorrimento

stima del volume usurato, ma un indicatore empirico delle

sollecitazioni accumulate. L’algoritmo la esprime, in forma compatta, come

è il tempo di funzionamento. L’esposizione viene combinata con le evidenze vibrazionali su entrambi i sensori. La

equivalente, proporzionale alla velocità della linea, e

struttura dell’indice può essere sintetizzata come

con coefficienti non negativi calibrati sulle campagne ma-

dell’esercizio. Una soglia prudenziale, posta prima della

nutentive disponibili.

fine vita teorica, identifica la regione nella quale program-

L’indice è limitato a un intervallo convenzionale compre-

mare la sostituzione [10].

so tra la condizione di cuscinetto nuovo e la fine vita teori-

La figura 7 riassume la combinazione delle informazioni

ca. È inoltre vincolato a essere monotono, poiché la quo-

di processo e delle evidenze provenienti dai due sensori.

ta di vita consumata non può diminuire con il proseguire

Fig.7 - Fusione delle evidenze LM e LO con l’esposizione cumulativa per ottenere un unico indice di usura

monotono, con diagnostica separata dei sensori / Fusion of LM and LO evidence with cumulative exposure to obtain

a single monotonic wear index, with separate sensor diagnostics.

La Metallurgia Italiana - September 2026

pagina 21


Memorie scientifiche - Digitalizzazione e AI L’addestramento del modello si basa sulle campagne

Stima della vita residua e pianificazione della manu-

concluse, per le quali è noto il momento della sostituzio-

tenzione

ne; invece, quelle ancora in corso non partecipano all’ad-

Il valore corrente dell’indice e la sua velocità di crescita

destramento e sono impiegate soltanto per l’inferenza.

consentono di stimare la vita operativa residua, mentre

I riferimenti del modello vengono congelati al termine

l’intervallo d’incertezza distingue una traiettoria ben sup-

dell’addestramento e l’utilizzo di nuovi campioni non

portata dai dati da una valutazione basata in parte su in-

modifica quindi retroattivamente i valori di usura già cal-

formazioni mancanti o incoerenti. Si specifica che l’indice

colati. Questa proprietà consente di utilizzare l’indice du-

non costituisce un comando automatico di sostituzione,

rante il monitoraggio online senza dipendere dalla durata

ma fornisce un riferimento all’operatore di manutenzione

futura della campagna. Infine, l’incertezza associata alla

per programmare la sostituzione dei cuscinetti (Human in

stima tiene conto della completezza dei dati di processo,

the loop). Nel calcolo viene introdotto un margine di in-

della quota di esposizione stimata, della disponibilità dei

certezza conosciuto dall’operatore così da mantenere un

sensori e della loro coerenza.

approccio conservativo rispetto alla stima della vita utile dei componenti e lasciare parte della decisione alla sensibilità dell’operatore. La logica di supporto alla decisione è rappresentata in figura 8.

Fig.8 - Utilizzo dell’indice di usura e della vita residua per pianificare la sostituzione, riducendo sia il sottoutilizzo

dei cuscinetti sia il rischio di fermo non programmato / Use of the wear index and remaining life to plan replacement

while reducing both premature bearing disposal and unplanned-stop risk.

Dal confronto tra le campagne disponibili, si nota come la

le diverse velocità di progressione dell’usura e colloca le

pendenza delle curve al variare dell’esposizione cumula-

campagne concluse in prossimità della regione prevista

tiva risulti differenti a causa dei differenti valori di velocità

per la sostituzione, mentre la campagna in corso rimane

e del tiro applicato (figura 9).

in una fase intermedia (figura 10).

La rappresentazione in ore di marcia rende inoltre visibili

Fig.9 - Traiettorie dell’indice di usura unificato rispetto all’esposizione cumulativa / Unified wear-index trajectories La Metallurgia Italiana - Settembre 2026

versus cumulative operating exposure.

pagina 22


Scientific papers - Digitalization and AI

Fig.10 - Evoluzione dell’indice unificato durante le campagne in funzione delle ore di marcia / Evolution of the

unified index during the campaigns as a function of operating hours.

Integrazione con la extended reality (VR/AR)

l’orientamento dello sguardo dell’operatore, proiettando

Per visualizzare i risultati del framework di previsione del-

metriche critiche sullo stato dei componenti, modelli 3D

la RUL basato sull’IA, vengono integrate tecnologie XR

accurati e dati predittivi direttamente nelle loro esatte po-

(realtà estesa) a supporto sia della formazione degli ope-

sizioni spaziali.

ratori sia delle attività di manutenzione sulla linea di zin-

La RUL stimata, calcolata dal modello di machine learning

catura a caldo. L’approccio proposto introduce una sorta

a partire dai dati dei sensori di vibrazione, viene visualizza-

di “visione a raggi X” virtuale della vasca di zincatura, con-

ta in tempo reale nel campo visivo dell’operatore tramite

sentendo agli operatori di visualizzare lo stato dei cusci-

un’interfaccia che adotta soluzioni di tipo adattivo (Adap-

netti immersi, altrimenti non accessibili né visibili durante

tive UI), regolando dinamicamente il livello di dettaglio in

il normale funzionamento dell’impianto. Per realizzare

base alla criticità di ciascun componente (figura 11).

ciò in un contesto industriale, il sistema impiega l’ana-

Quando un cuscinetto si avvicina alla fine della sua vita

lisi della prospettiva in prima persona (egocentric view)

utile, il sistema evidenzia il componente, fornisce infor-

mediante moduli di Computer Vision e l’uso di AprilTag

mazioni più dettagliate ed emette avvisi in tempo reale

(marker di riferimento a matrice ad alto contrasto) per la

per segnalare agli operatori eventuali condizioni critiche.

localizzazione. Scansionando tali marker, il dispositivo

Inoltre, gli operatori possono consultare una dashboard

AR localizza l’utente nello spazio fisico dell’impianto; da

di supporto alle decisioni che offre una panoramica com-

qui, il sistema traccia in tempo reale i movimenti precisi e

pleta dello stato della linea di zincatura (figura 12).

Fig.11 - Stima e visualizzazione in tempo reale della RUL / Real-time RUL estimation and visualization.

La Metallurgia Italiana - September 2026

pagina 23


Memorie scientifiche - Digitalizzazione e AI

Fig.12 - Dashboard di supporto alle decisioni con i dati sull’impianto (sinistra) e sui sensori di vibrazione (destra) /

Decision-support dashboard with information about the plant (left) and the vibration sensors (right).

Per ciascun cuscinetto, l’interfaccia visualizza la RUL sti-

L’applicazione di realtà virtuale (VR) è incentrata sulla for-

mata, lo stato di salute, la data dell’ultima sostituzione, le

mazione degli operatori per la sostituzione dei cuscinet-

specifiche del componente e indicatori di manutenzione

ti del rullo immerso (sink-roll) all’interno di un gemello

codificati a colori che segnalano i componenti che richie-

digitale della linea di zincatura. L’ambiente di formazione

dono attenzione. Gli operatori possono altresì monitorare

riproduce la procedura di manutenzione attraverso una

i parametri chiave del processo – tra cui larghezza del na-

guida interattiva passo-passo, fornisce assistenza in caso

stro, velocità della linea, temperatura del bagno, spessore

di errori dell’utente e registra le statistiche sulle prestazio-

e tensione del nastro – insieme ai relativi trend storici. La

ni per ogni fase dell’addestramento. L’ambiente virtuale

dashboard consente inoltre il monitoraggio in tempo reale

è realizzato a partire da modelli 3D ricostruiti dei compo-

dei sensori di vibrazione (incluso lo stato della connessio-

nenti critici dell’impianto, consentendo agli operatori di

ne e le metriche di vibrazione), permettendo agli operato-

esercitarsi in sicurezza sulle procedure di manutenzione

ri di verificare la qualità dei dati acquisiti che alimentano il

senza interferire con la produzione né esporre il persona-

modello di previsione della RUL. L’applicazione di realtà

le a condizioni operative pericolose, come illustrato nella

aumentata (AR) è collegata alla piattaforma digitale di ma-

figura 13.

nutenzione tramite un middleware che aggiorna continuamente i risultati delle previsioni e i dati di processo.

Fig.13 - Manutenzione dei cuscinetti attraverso realtà virtuale / VR application, maintenance of bearings.

La Metallurgia Italiana - Settembre 2026

pagina 24


Scientific papers - Digitalization and AI CONCLUSIONI

Dal punto di vista industriale, l’utilizzo di questa tecno-

Il lavoro svolto conferma la validità dell’approccio propo-

logia può generare benefici rilevanti. La disponibilità di

sto per la manutenzione predittiva dei cuscinetti dei rulli

una stima quantitativa dello stato di usura consente di ot-

immersi nelle linee di zincatura a caldo. Il modello svilup-

timizzare lo sfruttamento dei rulli e dei relativi cuscinetti,

pato ha mostrato un comportamento positivo, riuscendo

riducendo sia le sostituzioni premature sia il rischio di in-

a combinare segnali vibrazionali e parametri di processo

terventi tardivi. Ne deriva una migliore pianificazione delle

in un indice di usura coerente, monotono e utilizzabile

attività manutentive, una maggiore continuità produttiva

per la stima della vita utile residua. Tale risultato è parti-

e una conduzione più sicura dell’impianto. La possibilità

colarmente significativo perché dimostra l’applicabilità di

di anticipare condizioni critiche permette infatti di evitare

tecnologie di monitoraggio avanzato e algoritmi di Intel-

situazioni potenzialmente pericolose e ad alto impatto, sia

ligenza Artificiale anche in condizioni operative molto di-

per la produzione sia per la sicurezza dei lavoratori.

verse da quelle tipicamente considerate nei casi standard

L’integrazione con tecnologie di realtà estesa rappresen-

di manutenzione predittiva, caratterizzate da elevata tem-

ta inoltre un elemento distintivo dell’approccio propo-

peratura, accessibilità limitata dei componenti e impossi-

sto. La visualizzazione in AR delle informazioni preditti-

bilità di misurare direttamente lo stato dei cuscinetti.

ve e l’utilizzo della VR per la formazione introducono un

I risultati ottenuti costituiscono una base solida per le

nuovo modo di coinvolgere il lavoratore nell’impiego di

successive attività di validazione. Sono infatti previsti test

tecnologie digitali avanzate, rendendo più comprensibi-

in campo con il diretto coinvolgimento dei lavoratori,

le e operativo il contributo dell’Intelligenza Artificiale. In

finalizzati a verificare ulteriormente il comportamento

questo senso, il sistema non si limita a fornire un supporto

degli stimatori in condizioni operative reali e a raccogliere

alla manutenzione, ma contribuisce a costruire maggiore

feedback sull’utilizzabilità delle informazioni generate dal

fiducia nell’uso dell’IA in ambiente industriale, favorendo-

sistema. Questa fase sarà fondamentale per consolidare

ne uno sfruttamento più sicuro, efficace e orientato alle

l’affidabilità del modello, valutare la robustezza della stima

reali esigenze operative.

della vita residua e favorire l’integrazione dello strumento nei processi manutentivi ordinari dell’impianto.

BIBLIOGRAFIA [1] [2] [3]

[4] [5] [6] [7] [8] [9] [10]

J. Kulkarni. “Vibration-Driven Predictive Maintenance of Rotating Equipment Using Machine Learning”, International Journal for Research in Applied Science and Engineering Technology. 13. 2025. https://doi.org/10.22214/ijraset.2025.71321 ] M. Ghazali, M. Hazwan, W. Rahiman, “Vibration Analysis for Machine Monitoring and Diagnosis: A Systematic Review”, Shock and Vibration, 2021. https://doi.org/10.1155/2021/9469318 ] I. Keshta, M. E. Majeed, T. Raad Al-Shaikhli, A.-H. Adel, K. A. Jabbar and M. Soni. “AI-Based Predictive Maintenance for Machine Vibration and Condition Monitoring in Industrial Applications”, 4th OPJU International Technology Conference (OTCON) on Smart Computing for Innovation and Advancement in Industry 5.0. 1-7, 2025. DOI: 10.1109/OTCON65728.2025.11070934 ] X. Li, F. Bi, L. Zhang, X. Yang, G. Zhang, “An Engine Fault Detection Method Based on the Deep Echo State Network and Improved Multi-Verse Optimizer”, Energies 2022, 15, 1205. https://doi.org/10.3390/en15031205 ] INDUX-R – Transforming European Industrial Ecosystems through eXtended Reality enhanced with human-centric AI and secure, 5G-enabled IoT https://indux-r.eu/ ] H.-Y. Wu, M. Rubinstein, E. Shih, J. Guttag, F. Durand, W. T. Freeman, “Eulerian Video Magnification for Revealing Subtle Changes in the World”, ACM Transactions on Graphics, 31(4), 1–8, 2012. https://doi.org/10.1145/2185520.2185561 ] J. Dai, L. Tian, H. Chang, “An Intelligent Diagnostic Method for Wear Depth of Sliding Bearings Based on MGCNN”, Machines, 12(4), 266, 2024. https://doi.org/10.3390/machines12040266 ] M. M. Khonsari, E. R. Booser, Applied Tribology: Bearing Design and Lubrication, 3rd ed., John Wiley & Sons, 2017, Chapters 8 and 17. https://doi.org/10.1002/9781118700280 ] I. M. Hutchings, P. Shipway, Tribology: Friction and Wear of Engineering Materials, 2nd ed., Butterworth-Heinemann, 2017, Section 5.3, Simple Theory of Sliding Wear: The Archard Wear Equation. ISBN 978-0-08-100951-2. ] Y.-S. Zhao, P. Li, Y. Kang, Y.-B. Zhao. “A Health Indicator Enabling Both First Predicting Time Detection and Remaining Useful Life Prediction: Application to Rotating Machinery”, Measurement, 235, 114994, 2024. https://doi.org/10.1016/j.measurement.2024.114994

La Metallurgia Italiana - September 2026

pagina 25


Memorie scientifiche - Digitalizzazione e AI

Predictive maintenance of bearings in sink rolls of Hot-Dip Galvanizing Lines using extended reality and Artificial Intelligence lgorithms In hot-dip galvanizing lines, sink rolls guide the strip through the zinc bath and operate at temperatures close to 470 °C and speed up to 180 m/min. These conditions accelerate sliding-bearing wear and may cause unplanned downtime. The case study, developed at Acciaierie d'Italia within the INDUX-R project, combines triaxial accelerometers, process parameters and machine-learning models to estimate bearing wear and remaining useful life. Results are delivered through extended-reality tools: VR supports operator training, while AR provides field monitoring and maintenance assistance.

KEYWORDS: PREDICTIVE MAINTENANCE; ARTIFICIAL INTELLIGENCE; EXTENDED REALITY; HOT DIP GALVANIZING.

TORNA ALL'INDICE >

La Metallurgia Italiana - Settembre 2026

pagina 26


edizione 2026

L’ing. Carlo Longaretti è nato il 17 novembre 1924 e la sua vita professionale rappresenta un’esperienza emblematica di interazione tra mondo della ricerca e produzione industriale. E’ stato ricercatore universitario, rappresentante italiano presso la Comunità Europea del Carbone e dell’Acciaio per la gestione dei progetti di ricerca, fu il primo ad introdurre in Italia la produzione e l’utilizzo delle ghise sferoidali e nel 1952 con il cugino Amanzio è stato fondatore della FGS, la fonderia di acciai e ghise speciali sita in Treviglio, tuttora attiva, che ha amministrato sino alla morte avvenuta il 29 marzo 2019. Carlo Longaretti è stata una persona determinata e geniale, in grado di integrare nel proprio lavoro le competenze elettrotecniche, strutturali e metallurgiche attraverso le quali ha sviluppato impianti fusori, tecnologie di solidificazione e nuove leghe per la resistenza alle alte temperature. Grazie alla sua generosità e a quella della famiglia, per ricordare una delle più insigni figure che si sono stagliate nel panorama dell’industria metallurgica italiana del XX e XXI secolo, l’AIM istituisce il Premio Carlo Longaretti. Il Premio, nella sua terza edizione, verrà conferito in occasione dell’International Workshop Carlo Longaretti 2026 a Bergamo il 3 dicembre 2026 ed include un contributo di €50.000 per sostenere il vincitore nello sviluppo della propria innovazione. L’oggetto dell’innovazione proposta per il concorso può essere un nuovo materiale di natura metallica, oppure un nuovo materiale anche non metallico che abbia un impatto positivo nella produzione o nell’applicazione delle leghe metalliche oppure una nuova tecnologia metallurgica o a servizio del settore metallurgico.

Come concorrere Il concorso per il premio è rivolto ai Soci AIM (a titolo personale o aziendale) e aperto a diplomati che abbiano frequentato o stiano frequentando un Istituto Tecnico Superiore, studenti di corsi di laurea, laureati, laureati magistrali, dottori di ricerca che dovranno presentare in lingua italiana o inglese: • curriculum vitae; • una descrizione dettagliata dal punto di vista tecnico e scientifico dell’innovazione proposta (massimo 15 pagine), che includa: - elementi che validino l’innovazione e che escludano che si tratti di un’idea di carattere teorico senza prospettive concrete di sviluppo (es. dati sperimentali anche su scala ridotta, immagini, filmati ecc.). - elementi che comprovino l’originalità dell’innovazione proposta; - elementi che indichino quali esigenze intenda soddisfare la propria innovazione e quale è l’impatto atteso; • lista delle proprie pubblicazioni, in particolare di quelle relative all’innovazione proposta. Il termine di presentazione delle domande, da trasmettersi per email (info@aimnet.it) alla Segreteria AIM, è fissato al 15 ottobre 2026.

�946 2026

Via F. Turati 8 . Milano t. +39 0276397770 / +39 0276021132 info@aimnet.it . www.aimnet.it

PREMIO

Bando Premio Carlo Longaretti


Attualità industriale - Industry news

DOI 10.36146/2026_06_28

Gemello digitale della colata continua: costruzione e calibrazione in ProCAST per la previsione di difetti e l’ottimizzazione di processo

R. Ottini, G. Mori, A. Mario, M. Bognolo, S. Gobbo, M. Marzaro, M. Bianchi, S. Buoro, F. Ricchini, C. Viscardi, L. Valente La colata continua è un processo chiave nella produzione siderurgica, poiché influisce direttamente sulla qualità del prodotto. Il presente lavoro descrive lo sviluppo di un gemello digitale di una macchina di colata continua dello stabilimento di Padova di Acciaierie Venete. Il modello è stato realizzato da Ecotre Valente con software ProCAST seguendo una metodologia consolidata. Sono stati raccolti e integrati i reali parametri di processo. È stata costruita una modellazione 3D completa dell’impianto dalla lingottiera al taglio billette. Il sistema include diverse sezioni: raffreddamenti, stirrer elettromagnetici e rulli. Il modello ottenuto è stato calibrato confrontando dati simulati con le misure reali e la calibrazione ha mostrato una buona corrispondenza tra il modello e realtà. Nel lavoro sono state analizzate criticità come porosità, segregazioni e distribuzione dell’idrogeno. Il gemello digitale ottenuto consentirà quindi di ottimizzare processo e qualità riducendo scarti e costi.

PAROLE CHIAVE: COLATA CONTINUA; GEMELLO DIGITALE; PROCAST; SOLIDIFICAZIONE; SEGREGAZIONE, POROSITÀ; IDROGENO; STIRRER ELETTROMAGNETICI. INTRODUZIONE La crescente digitalizzazione dei processi siderurgici consente oggi di affiancare alla pratica industriale tradizionale strumenti numerici avanzati in grado di descrivere e prevedere il comportamento termo-metallurgico dei prodotti durante tutte le fasi produttive, tra cui la colata continua. In tale contesto, il concetto di gemello digitale assume un ruolo strategico: non si tratta soltanto di una rappresentazione geometrica dell’impianto, ma di un modello numerico calibrato sui dati reali di processo, capace di riprodurre le condizioni operative e di supportare decisioni tecniche orientate al miglioramento della qualità. L’attività descritta nel presente lavoro si inserisce in un

Riccardo Ottini, Giacomo Mori, Andrea Mario, Marco Bognolo, Sandro Gobbo, Matteo Marzaro, Marco Bianchi, Stelvio Buoro Acciaierie Venete S.p.A.

Flavio Ricchini, Cristian Viscardi, Lorenzo Valente Ecotre Valente S.r.l.

più ampio progetto di digitalizzazione di una delle macchine di colata continua di Acciaierie Venete S.p.A., condotto con il sostanziale contributo di Ecotre Valente utilizzando il software ProCAST. Il caso studio riguarda il tondo colato con diametro 200 mm in acciaio 42CrMoS4, i cui parametri di colata sono riportati in tabella 1. L’obiettivo principale è costruire un modello digitale dell’impianto reale, verificarne la coerenza termica trami-

La Metallurgia Italiana - Settembre 2026

pagina 28


Industry news - Attualità industriale te confronto con dati sperimentali e utilizzarlo per preve-

forme, porosità da ritiro, segregazione e concentrazione

dere criticità metallurgiche quali solidificazione non uni-

di idrogeno.

Tab.1 - Principali parametri di processo / Main process parameters. Parametro

Valore

Profilo colato

Tondo 200 mm

Grado acciaio

42CrMoS4

Surriscaldo tipico

+40°C

Velocità di colata

1.2 m/min

ACCIAIO OGGETTO DELLO STUDIO

dell’acciaio sono state calcolate in ProCAST a partire dalla

Il profilo selezionato per lo studio è un tondo di diametro

composizione chimica mediante il modulo Computherm.

200 mm in acciaio 42CrMoS4. Le proprietà termo-fisiche

Tab.2 - Composizione chimica dell’acciaio utilizzato nel modello / Chemical composition of the steel used

for the model.

Elemento

wt%

C

0.400

Si

0.200

Mn

0.790

P

0.010

S

0.030

Cr

1.030

Mo

0.150

Cu

0.180

MODELLAZIONE DELL’IMPIANTO

La digitalizzazione geometrica dell’impianto è il primo

La modellazione tridimensionale ha interessato una linea

passaggio fondamentale, poiché permette di associa-

della macchina di colata continua per sezioni piccole dal-

re alle diverse zone della linea le corrette condizioni di

la lingottiera fino alla zona dell’ossitaglio. Sono state in-

scambio termico, raffreddamento e forze in gioco (forze

cluse le principali sezioni impiantistiche: lingottiera (cri-

elettromagnetiche degli stirrer, forza gravitazionale, forze

stallizzatore + raffreddamento primario), M-EMS (mould

dei rulli di raddrizzatura, etc.). In questo modo il modello

electromagnetic stirrer), Loop 1 del raffreddamento se-

non rappresenta una condizione teorica semplificata, ma

condario, Loop 2 del raffreddamento secondario, F-EMS

una replica numerica dell’impianto industriale reale.

(final electromagnetic stirrer), rulli di raddrizzatura, cappe intermedie e cappe finali (figura 1).

La Metallurgia Italiana - September 2026

pagina 29


Attualità industriale - Industry news

Fig.1 - Layout digitale della linea di colata continua / Digital layout of the continuos casting strand. CRISTALLIZZATORE E RAFFREDDAMENTO PRIMARIO

troflusso rispetto alla direzione di colata. I parametri di

La lingottiera è costituita da un classico cristallizzatore

esercizio utilizzati sono: portata ≈ 3000 l/min, ΔT tra tem-

curvo in rame collocato in un circuito di raffreddamento

peratura in uscita e in ingresso ≈ 4°C e pressione pari ≈ 8

primario ed è stata modellata considerando il passaggio

bar.

dell’acqua tra cristallizzatore e convogliatore in con-

Fig.2 - Velocità dell’acqua del raffreddamento primario (sinistra) e temperatura in ingresso e in uscita (destra) / Water

speed in primary cooling (left) and inlet and outlet water temperature (right).

Per la descrizione fluidodinamica del circuito è stato uti-

RAFFREDDAMENTO SECONDARIO

lizzato il modulo ProCAST CFD Cooling Channels, fina-

Il raffreddamento secondario, costituito da sola acqua, è

lizzato alla valutazione della velocità dell’acqua e dello

suddiviso in due loop. Il loop 1 è composto da 24 ugelli

scambio termico nel sistema primario. La velocità dell’ac-

con una portata complessiva di 40 l/min, ingresso dell’ac-

qua tra cristallizzatore e convogliatore calcolata dal mo-

qua a temperatura ambiente e pressione di ≈ 2 bar. Il loop

dello è di circa 10 m/min (in linea con quanto atteso).

2 consta di 48 ugelli, ha una portata di 30 l/min, acqua in

Lo scostamento tra la temperatura in uscita misurata e la

ingresso a temperatura ambiente e pressione di circa 1

temperatura in uscita calcolata dal modello si aggira intor-

bar.

no a 0.1°C

La Metallurgia Italiana - Settembre 2026

pagina 30


Industry news - Attualità industriale

Fig.3 - Dettaglio del raffreddamento secondario. Loop 1 (sinistra), loop 2 (centro) e profilo termico (destra) /

Secondary colling details. Loop 1 (left), loop 2 (centre) and thermal profile (right).

La figura 3 riporta il dettaglio del raffreddamento seconda-

attrezzata la linea di colata: M-EMS (mould electromagne-

rio ed il profilo termico ottenuto dalla simulazione.

tic stirrer) in zona lingottiera e F-EMS (final electromagnetic stirrer) a valle del secondario. Sia M-EMS sia F-EMS

STIRRER ELETTROMAGNETICI

sono costituiti da 6 bobine e sfasatura trifase di 120°.

Il modello include i due stirrer elettromagnetici con cui è

Fig.4 - Dettaglio del M-EMS (sinistra) e del F-EMS (destra) / EMS details, M-EMS on the left and F-EMS on the right. La simulazione degli stirrer è stata realizzata mediante il

do e sulla distribuzione degli elementi durante la solidifi-

modulo ProCAST EMX, utilizzato per descrivere l’effetto

cazione, utilizzando i parametri di processo riportati nella

dello stirring elettromagnetico sul moto del metallo liqui-

tabella.

Tab.3 - Parametri indicativi di funzionamento degli stirrer / Indicative operating parameters of the stirrers. Stirrer

Posizione

Corrente [A]

Frequenza [Hz]

M-EMS

Lingottiera

100

2

F-EMS

Dopo secondario

250

15

La Metallurgia Italiana - September 2026

pagina 31


Attualità industriale - Industry news CALIBRAZIONE TERMICA DEL MODELLO

visibile, e tali misure sono state confrontate con quanto re-

sviluppo del gemello digitale, poiché consente di verificare

le temperature delle superfici superiori delle carpenterie

La calibrazione termica rappresenta una fase chiave nello la corrispondenza tra il comportamento simulato e quello

dell’impianto reale. Sulla base della termica vengono poi

infatti modellati tutti gli altri aspetti (segregazione, porosità, etc.).

Nel caso in esame, il primo controllo ha riguardato il delta di temperatura del circuito primario: lo scostamento misu-

rato è riportato nel paragrafo precedente. La calibrazione è stata anche poi verificata confrontando le temperature superficiali della billetta in diversi punti della linea. Per fare

questo sono state eseguite varie rilevazioni con pirometro,

tipicamente nelle zone più accessibili della macchina di colata continua dove la superficie del blumo risultasse ben

stituito dal modello. Inoltre, sono state confrontate anche delle cappe intermedie e finali.

I risultati finali sono ripotati in tabella 4. Nel complesso, la

correlazione tra simulazione e misura risulta soddisfacente per l’utilizzo del modello a fini predittivi. Gli scostamenti

osservati sono contenuti nelle zone ad alta temperatura, mentre diventano relativamente più rilevanti nelle zone

finali a bassa temperatura (soprattutto sulla carpenteria superiori delle cappe finali), dove possono incidere maggiormente aspetti legati all’irraggiamento, alla convezione

ambientale, alla precisione della misura pirometrica e alla modellazione più o meno fedele delle cappe isolanti.

Tab.4 - Confronto tra temperature reali misurate e simulate / Measured and simulated billet surface

temperature.

Posizione

Scostamento %

Blumo – dopo raffreddamento secondario (loop 2)

0.9

Blumo – prima di F-EMS

1.0

Blumo – tra cappe intermedie e finali

2.0

Blumo – post cappe finali

0.8

Carpenteria superiore cappe intermedie

1.6

Carpenteria superiore cappe finali

-19.8

Fig.5 - Carpenteria superiore delle cappe intermedie e finali, zona di esecuzione delle misure di temperatura / Sections where temperature measurements are taken for upper metalwork of the intermediate and

terminal hoods.

La Metallurgia Italiana - Settembre 2026

pagina 32


Industry news - Attualità industriale RISULTATI DELLA SIMULAZIONE – EVOLUZIONE TER-

stensione della vena liquida e la condizione metallurgica

MICA E FRAZIONE SOLIDA

in corrispondenza delle zone interessate dagli stirrer elet-

Una volta calibrato, il modello è stato utilizzato per pri-

tromagnetici, in particolare lo stirrer finale in quanto nella

ma cosa per analizzare la distribuzione della temperatura

zona interessata dallo stirrer di lingottiera tutto il materia-

superficiale (figura 6) e l’evoluzione della frazione solida

le è ancora in fase liquida (a eccezione del sottile strato di

lungo la billetta. Queste due simulazioni consentono di

pelle in formazione).

osservare l’avanzamento del fronte di solidificazione, l’e-

Fig.6 - Profilo di temperatura superficiale del blumo lato intradosso / Intradox surface temperature profile

of the bloom.

In particolare, l’analisi della frazione solida in corrispon-

zione è rilevante perché permette di valutare l’efficacia

denza del F-EMS evidenzia la presenza di una vena fluida

dell’agitazione elettromagnetica in una zona in cui il ma-

con frazione solida prossima al 70%, equivalente a una

teriale mantiene ancora una significativa componente li-

frazione liquida di poco superiore al 30%. Tale informa-

quida (figura 7).

Fig.7 - Frazione solida all’interno del blumo in corrispondenza del F-EMS (sinistra) e velocità del campo di

forza generato dallo stirrer finale nella stessa posizione (destra) / Solid fraction within the bloom at the F-EMS

(left) and speed of the force field generated by the final stirrer in the same position (right).

RISULTATI DELLA SIMULAZIONE – POROSITÀ DA

dei prodotti di colata continua, nei quali l’ultima zona a

RITIRO

solidificare tende a coincidere con l’area centrale, sog-

La simulazione ha evidenziato la presenza di porosità da

getta a deficit di alimentazione e a concentrazione delle

ritiro localizzate lungo l’asse della billetta. Questa tipolo-

discontinuità volumetriche.

gia di difetto è coerente con la dinamica di solidificazione

La Metallurgia Italiana - September 2026

pagina 33


Attualità industriale - Industry news

Fig.8 - Porosità da ritiro / Shrinkage porosity. La possibilità di conoscere presenza e probabile entità

lo indichi che le porosità non siano continue (figura 8) è

delle porosità assiali consente di collegare il difetto alle

conforme alla reale struttura di solidificazione come os-

condizioni locali di solidificazione e di valutare possibili

servata attraverso macrografie con persolfato d’ammonio

interventi su raffreddamento secondario, velocità di co-

(figura 9) ed è frutto dei fenomeni di solidificazione.

lata o configurazione dello stirring. Il fatto che il model-

Fig.9 - Macrografia longitudinale di un acciaio 42CrMoS4 grezzo di colata prodotto dalla macchina di colata

continua oggetto dello studio / Longitudinal macroetching of an as-cast 42CrMoS4 bloom produced by the

continuous casting machine under investigation.

RISULTATI DELLA SIMULAZIONE – CONCENTRAZIO-

li di raddrizzatura). I risultati mostrano, come atteso, una

NE DI IDROGENO

concentrazione maggiore al centro del prodotto ma an-

Il modello ha permesso di analizzare la distribuzione della

che una presenza in superficie, attribuita al raffreddamen-

concentrazione di idrogeno nel blumo, tale analisi è stata

to repentino della “pelle” che limita la diffusione dell’i-

effettuata in una zona completamente solida (dopo i rul-

drogeno. Tuttavia, il fatto che il blumo sia ancora caldo al

La Metallurgia Italiana - Settembre 2026

pagina 34


Industry news - Attualità industriale termine della colata continua promuove un certo grado

al centro del blumo, non è però dimostrato quale possa

di diffusione dalla superficie verso l’atmosfera esterna di

essere la soglia di attenzione per l’innesco di tali cricche,

quell’idrogeno intrappolato negli strati corticali, se pre-

considerando anche la presenza delle segregazioni e del-

sente. Mentre l’aspetto più deleterio per la formazione

le porosità da ritiro che possono fungere da ulteriori trap-

di cricche da idrogeno è il concentramento di idrogeno

pole per l’idrogeno che va accumulandosi al centro.

Fig.10 - Distribuzione dell’idrogeno nel blumo nel caso standard / Hydrogen distribution in the bloom in

standard condition.

Nel modello sono state simulate due condizioni. Nella

za in paniera è stato aumentato del 45% fino a un livello

prima condizione è stato impostato un valore di idroge-

ritenuto non conforme e che, sulla base dei dati storici e

no nel liquido di partenza (in paniera) tipico per gli acciai

sull’esperienza di Acciaierie Venete, richiede contromisu-

colati su questa macchina di colata, valore che non genera

re per il rischio cricche (tipicamente ricottura dei grezzi di

allarmi e che non fa scattare contromisure a valle della co-

colata per favorire la diffusione dell’idrogeno). In questo

lata continua (figura 10). In tale situazione si raggiungono

secondo contesto il valore raggiunto al centro del blumo

concentrazioni a cuore di circa 1.3 ppm.

si attesta intorno a 1.8 ppm.

Per la seconda condizione (figura 11) il valore di parten-

Fig.11 - Distribuzione dell’idrogeno nel blumo nel caso standard / Hydrogen distribution in the bloom in

standard condition.

La Metallurgia Italiana - September 2026

pagina 35


Attualità industriale - Industry news RISULTATI DELLA SIMULAZIONE – SEGREGAZIONE

nio (figura 12) e di altri elementi chimici (Cr, Mn, Mo, Ni e

DEGLI ELEMENTI

Si).

La simulazione ha considerato la segregazione del carbo-

Fig.12 - Profilo di distribuzione del carbonio dopo il F-EMS / Carbon profile after the F-EMS. Per il carbonio è stata ottenuta una forma del profilo qua-

torno a metà raggio. L’analisi sul campione reale (figura

litativamente simile a quella sperimentale (figura 13): un

13, destra) è stata effettuata con metodo IR per combu-

picco centrale di segregazione positiva, seguito da un

stione (analizzatore LECO CS744, ASTM E1019-18).

calo e da un ulteriore picco positivo di minore entità in-

Fig.13 - Confronto tra profilo di carbonio calcolato con ProCAST (sinistra) e profilo misurato (destra) /

Comparison between the carbon profile calculated using ProCAST (left) and the measured one (right).

La Metallurgia Italiana - Settembre 2026

pagina 36


Industry news - Attualità industriale Il confronto con i dati sperimentali deve essere interpre-

le potenzialità di ProCAST, riguarda la valutazione dell’ef-

tato in modo qualitativo data la differenza di contenuto

fetto degli stirrer elettromagnetici. Il confronto tra con-

nominale; tuttavia, il comportamento riscontrato è coe-

figurazioni con entrambi gli stirrer spenti ed entrambi gli

rente (differenza percentuale tra nominale e massimo per

stirrer accesi (figura 14) mostra che, in assenza di stirring,

calcolato ≈10%; per misurato ≈14%). La presenza del pic-

la segregazione positiva al centro risulta grandemente ac-

co a metà raggio è dovuta all’azione degli stirrer elettro-

centuata, mentre l’utilizzo degli stirrer modifica il campo

magnetici, un’azione troppo pronunciata può infatti dare

di moto del metallo liquido e conseguentemente la distri-

luogo alla cosiddetta banda bianca, localizzata solitamen-

buzione del carbonio.

te a metà raggio. RISULTATI DELLA SIMULAZIONE – EFFETTO DEGLI STIRRER SULLA SEGREGAZIONE Una applicazione del gemello digitale, che ben dimostra

Fig.14 - Profilo di distribuzione del carbonio a stirrer spenti (sinistra) e accesi (destra) / Carbon profile without

EMS (left) and with EMS (right).

Questa evidenza conferma il potenziale del modello

DISCUSSIONE, CONCLUSIONI E SVILUPPI FUTURI

come strumento di analisi causale: variando virtualmente

I risultati ottenuti evidenziano come la costruzione di un

il set-up di M-EMS e F-EMS è possibile studiare l’effetto

gemello digitale calibrato possa rappresentare un ausilio

dei parametri elettromagnetici sulla solidificazione e sulla

determinante per la comprensione del processo di colata

qualità interna del prodotto, senza dover effettuare prove

continua. La buona corrispondenza tra dati reali e simu-

onerose direttamente sull’impianto industriale.

lati nella calibrazione termica costituisce il presupposto per utilizzare il modello non solo in modalità descrittiva,

La Metallurgia Italiana - September 2026

pagina 37


Attualità industriale - Industry news ma anche in modalità predittiva. Dal punto di vista me-

e processo, utile per valutare scenari alternativi di raffred-

tallurgico, l’asse del blumo si conferma una zona critica

damento, velocità di colata e set-up degli stirrer, con l’o-

per la formazione di porosità da ritiro, concentrazione di

biettivo di ridurre gli scarti, aumentare la resa e migliorare

idrogeno e segregazione positiva del carbonio. La simu-

l’efficienza complessiva della macchina di colata continua.

lazione permette di visualizzare tali fenomeni in modo integrato, collegandoli all’evoluzione della frazione solida,

Considerando i risultati ottenuti i prossimi sviluppi, par-

al profilo termico e all’azione degli stirrer elettromagne-

zialmente già in corso, riguarderanno:

tici. L’analisi della segregazione evidenzia inoltre il valore del modello come strumento di confronto tra condizioni operative alternative. Anche quando il confronto con i dati sperimentali è limitato a una valutazione qualitativa, come per la segregazione di carbonio (figura 13), la coerenza della forma del profilo consente di ottenere indicazioni

-

modifica di alcuni aspetti operativi, fra cui la posizione

-

ottimizzazione dei parametri finalizzata all’ulteriore

-

e i parametri di gestione dello stirrer finale; miglioramento della qualità interna del prodotto; verifica della possibilità di estensione della gamma dimensionale colabile dalla macchina di colata.

utili sull’evoluzione del fenomeno. RINGRAZIAMENTI Un ulteriore aspetto rilevante riguarda la possibilità di

Acciaierie Venete S.p.A. desidera ringraziare il personale

condurre analisi di sensitività, come mostrato per la con-

di Ecotre Valente per la preziosa collaborazione e inoltre

centrazione iniziale di idrogeno in paniera. Questo ap-

ringraziare Auril & Gond S.r.l. per l’assistenza nella mo-

proccio permette di stimare in anticipo l’impatto di varia-

dellazione degli stirrer elettromagnetici.

zioni metallurgiche o operative, riducendo tempi, costi e rischi associati alla sperimentazione diretta in impianto. Riassumendo, il lavoro ha permesso di sviluppare un gemello digitale della macchina di colata continua per sezioni piccole di Acciaierie Venete S.p.A. per il tondo 200 mm in acciaio 42CrMoS4, utilizzando ProCAST, la metodologia e le competenze di Ecotre Valente. Il modello include le principali sezioni dell’impianto, dalla lingottiera alla zona di taglio, e integra raffreddamento primario, raffreddamenti secondari, stirrer elettromagnetici, rulli e cappe isolanti. Le simulazioni hanno consentito di prevedere criticità metallurgiche quali porosità da ritiro assiale, concentrazione di idrogeno al centro, segregazione del carbonio e influenza degli stirrer sulla distribuzione degli elementi. Il gemello digitale si configura quindi come una piattaforma efficace per l’ottimizzazione di prodotto

BIBLIOGRAFIA [1] [2] [3] [4]

© Keysight Technologies, ProCAST 2026.0 User’s Guide, Revision A, November 2025. O. Ludwig, M. Aloe, P. Thevoz, State of the art in modelling continuous casting, Lausanne, 2009. O. Koeser, M. Aloe, L. Valente, «Simulazione di Colata Continua», Pressocolate & Tecniche Fusorie, pp. 64-66, Dicembre 2006. G. Couturier, J.-L. Desbiolles, M. Rappaz, «Effect of volatile elements on porosity formation in solidifying alloys», Modelling and Simulation in Materials Science and Engineering, vol. 14, n. 2, pp. 253-271, 2006.

La Metallurgia Italiana - Settembre 2026

pagina 38


Industry news - Attualità industriale

Digital twin of continuous casting: ProCAST model development and calibration for defect prediction and process optimization Continuous casting is a key process in steel production, as it directly affects product quality. This paper describes the development of a digital twin of a continuous casting machine at Acciaierie Venete’s Padua plant. The model was created using ProCAST software and the Ecotre Valente methodology. Actual industrial process parameters were collected and integrated into the model. A complete 3D model of the plant was constructed, from the mold to the cutting section. The system includes several sections: cooling systems, electromagnetic stirrers, and rollers. The model was calibrated by comparing simulated data with actual measurements and the calibration showed good agreement between the model and reality. Critical issues such as porosity, segregation, and hydrogen distribution were analyzed in this work. The digital twin will therefore enable the optimization of the process and quality, reducing scrap and costs.

KEYWORDS: CONTINUOUS CASTING; DIGITAL TWIN; PROCAST; SOLIDIFICATION; SEGREGATION; POROSITY; HYDROGEN; ELECTROMAGNETIC STIRRERS.

TORNA ALL'INDICE >

La Metallurgia Italiana - September 2026

pagina 39


Attualità industriale - Industry news

DOI 10.36146/2026_09_40

Sensor and digital twin solutions developed in DiGreeS project for improvement of scrap-based EAF steelmaking B. Kleimt, B. Palm, G. Weides, K. Srivastava, F. van den Berg, F. Schrama, M. Heinrich, B. Wolter, S. Groenheide, C. Noel, K. Winkler, F. Egger On the path toward low-carbon steelmaking and increased circularity, the Electric Arc Furnace (EAF) is pivotal for European steelmakers. Within the EU-funded Horizon project “Demonstration of Digital twins for a Green Steel value chain (DiGreeS)” an integrated digitalisation approach is developed across the steel value chain, to leverage process data, Artificial intelligence (AI) techniques and human experience for seamless industrial integration. This paper presents sensor-based and digital twin solutions being developed in DiGreeS for two key challenges in scrap-based EAF steelmaking: (i) automated verification of scrap feedstock quality and (ii) real-time assessment and control of foamy slag conditions. The approach combines novel sensing technologies, including multi-modal scrap characterisation and EAF process monitoring, with machine learning models embedded in a digital twin framework. First project results demonstrate the potential of the proposed methods to improve scrap quality assessment and enhance process efficiency in EAF operation.

KEYWORDS: EAF PROCESS; DIGITALISATION; SENSORS; SCRAP; SLAG FOAMING CONTROL. INTRODUCTION: EAF STEELMAKING USE CASES IN DIGReeS The transition towards low-carbon steel production, as outlined in the EU Green Deal and Circular Economy Action Plan [1], requires a substantial increase in low-quality scrap utilisation in Electric Arc Furnace (EAF) steelmaking. However, decreasing quality of scrap input and the complexity of the EAF process introduce significant challenges for both feedstock quality control and energy-efficient furnace operation. In particular, unnoticed variations in scrap compositions and in process conditions directly affect steel quality, energy efficiency, and productivity. In current industrial practice, scrap purchasing and classification remain largely based on established supplier relationships and manual inspection. As a result, the bulk chemical composition of heterogeneous scrap types

Bernd Kleimt, Birgit Palm

VDEh-Betriebsforschungsinstitut GmbH, Düsseldorf, Germany

Gerd Weides

Saarstahl AG, Völklingen, Germany

Kinshuk Srivastava

Stahl-Holding-Saar GmbH & co kgaa., Dillingen, Germany

Frenk van den Berg, Frank Schrama Tata Steel, IJmuiden, Netherlands

Matthias Heinrich, Bernd Wolter

Fraunhofer Institute for Nondestructive Testing, Saarbrücken, Germany

Stefan Groenheide, Clement Noel Spectral Industries, Delft, Netherland

Klemens Winkler, Florian Egger K1-MET GmbH, Linz, Austria

such as Heavy Melting Scrap (HMS) remains highly uncertain at the time of charging. To mitigate this, techniques such as Laser-Induced Breakdown Spectroscopy

La Metallurgia Italiana - Settembre 2026

pagina 40


Industry news - Attualità industriale (LIBS) [2] enable fast, contactless elemental analysis and

The paper is structured accordingly: the first part focuses

are increasingly explored in research projects and indus-

on sensor-based scrap characterisation and data fusion,

trial environments. However, their application to large,

while the second part addresses the digital twin-based

bulk scrap flows is still limited, particularly with respect

monitoring and control of EAF process conditions, with

to representative, truck-scale composition assessment

particular emphasis on foamy slag behaviour.

under realistic plant conditions. CHARACTERISATION OF HEAVY MELTING SCRAP In parallel, EAF process monitoring relies on established

(HMS)

measurement systems such as off-gas analysis and arc

The characterisation of heavy melting scrap with large

signals, which provide indirect information on process

scrap pieces of high inhomogeneity is today based on

conditions. While these signals are routinely used, their

manual and thus highly empirical visual inspection. In

integration into real-time predictive and control-oriented

some plants also random samples are analysed with hand-

frameworks remains limited, and key process phenom-

held XRF sensors. However, such measurements are time

ena—such as foamy slag formation—are still strongly

consuming when making them representative for the full

influenced by operator experience. Additional sensing

scrap load and are therefore typically only taken occa-

approaches, including structure-borne and airborne vi-

sionally. Consequently, the bulk chemical composition of

bration measurements, offer further insight into process

HMS remains uncertain.

dynamics but are not yet systematically combined with existing data sources for real-time control.

To address this limitation, DiGreeS develops an automated, AI-driven dual-sensor fusion system for in-situ ele-

To address these challenges, the Horizon Europe proj-

mental analysis of a full truck load which will be set up at

ect “Demonstration of Digital Twins for a Green Steel

the premises of Tata Steel in The Netherlands. The system

value chain (DiGreeS)” develops integrated sensor and

is designed to estimate the bulk chemical composition of

ultimately digital twin solutions for improved monitoring

HMS, with an initial focus on the critical tramp element

and control along the steel production chain. Within this

copper, which significantly affects downstream steelmak-

framework, this paper presents two complementary de-

ing processes and final product quality. To achieve an in-

velopments targeting scrap-based EAF steelmaking.

line characterisation, the system uses a combination of

First, a multi-sensor system for automated characterisa-

images generated by two RGB cameras for geometrical

tion of HMS is introduced, combining RGB-camera-based

information with the surface analysis from a Laser Induced

imaging for geometric analysis with LIBS-based surface

Breakdown Spectroscopy (LIBS) system. The sensor set-

composition measurements. Through machine learn-

up will be installed on a gantry system for scanning a truck

ing-based data fusion, the approach enables the estima-

upon delivery, enabling to estimate the average copper

tion of bulk chemical composition at truck scale, over-

concentration as a bulk value for the whole scrap load.

coming the limitations of conventional sampling-based

This approach aims to minimise unexpected impurity lev-

methods.

els in scrap feedstock and thus to improve the crude steel quality. The overall concept is sketched in figure 1 togeth-

Second, a sensor-integrated digital twin approach for EAF

er with a sketch showing the envisaged dimensions and

operation is presented, combining off-gas analysis, EAF

the distance between sensors and scrap load. The single

power signals, and vibration measurements within a hy-

components will be described hereafter.

brid modelling framework. This approach allows the online prediction of foamy slag conditions based on physics-based and data-driven modelling and provides a basis for a digital twin enabling improved control of carbon and oxygen injection.

La Metallurgia Italiana - September 2026

pagina 41


Attualità industriale - Industry news

Fig.1 - Concept and dimensions of sensor system for characterisation of heavy melting scrap. Dual RGB camera system

the image analysis of the stereo camera system allows to

The dual RGB camera system is used to capture geometric

calculate the filling degree of the truck and the scrap pile

information of the scrap load, including the size and shape

height, see figure 2. This image data provides the basis for

of the scrap pieces as well as of the height of scrap pile.

segmentation of individual scrap objects and spatial refer-

Several images are acquired and stitched to generate a full

encing of compositional measurements performed by the

picture of the scrap load surface regarding distribution of

LIBS system.

size and shape of the different scrap pieces. Furthermore,

Fig.2 - Stereo camera image for scrap pile height detection: Half full (right) and full (left) scrap box. LIBS system

focused laser beam creates a small region around the op-

The LIBS sensor will be used for remote, contactless sur-

timal focus, in which the power density still surpasses the

face analysis of the scrap pieces including the separation

required one to create a plasma, called the plasma depth.

of coatings from base material [3, 4]. LIBS is a laser-based

For this system, it is spread on 10 mm. Furthermore, the

spectroscopy method that allows for non-contact testing

HMS is very heterogeneous in shape and size, which

of the components [3]. Each individual LIBS measure-

means the laser focus must be moved at high speed due to

ment reveals the elemental composition of the HMS un-

topographic variations. To focus the laser light and collect

der investigation. The LIBS data will be timestamped and

the resulting plasma, the optical configuration is therefore

localized on the sample surface based on external posi-

of vital importance. The autofocus system ensures that the

tion measurements of the system. The laser installed in

material is always in focus for optimal measurement con-

the LIBS system is a Q-switched laser which operates at

ditions. The long-range LIBS sensor is a telescope-based

1645 nm and a 1000 Hz pulse frequency. To create a plas-

optical configuration to optimize light collection from the

ma on the sample surface the energy density must exceed

laser-induced plasma and to relay the focus rapidly over,

roughly 1 GW/cm [5]. To reach this power threshold, the

relatively, large distances (see figure 3).

2

laser light needs to be focussed by means of optics. The

La Metallurgia Italiana - Settembre 2026

pagina 42


Industry news - Attualità industriale

Fig.3 - Optical configuration of LIBS system. This optical setup plays a central role in enhancing sig-

low-concentration elements.

nal quality, especially under challenging industrial condi-

The LIBS system is equipped with an analysis computer to

tions. The reflective telescope configuration utilizes two

directly analyse the captured spectra and convert these to

mirrors: 1) a large mirror (M1) that focuses the laser light

quantitative elemental concentrations based on an anal-

on a sample and collects the broadband emission from

ysis model. The analysis model is created after a calibra-

the plasma plume generated on the sample surface; 2) a

tion campaign with certified reference material (CRM) that

secondary mirror (M2), positioned along the optical axis

cover the spread of expected elemental concentration in

that creates a divergent bundle of the laser light towards

the HMS (see figure 4), illustrating a strong correlation

M1 and reflects and focuses the collected plasma light, di-

between the ground truth data and the estimation based

recting it towards an off-axis optical path where the spec-

on the LIBS signal intensity. The calibration campaign ad-

trometers reside. M2 is mounted on a moving stage that

dressed the following key needs: model training, correc-

enables it to reposition itself rapidly causing a change in

tion of matrix effects and validation of quantification ac-

focal distance from the sensor to the sample. A distance

curacy. Besides the calibration of the LIBS system using

sensor in the LIBS sensor provides the required position

CRMs, trials in laboratory have been performed to deter-

data for the movable stage. This design allows the system

mine the impact of moisture and of surface coating.

to achieve high collection efficiency, critical for detecting

Fig.4 - Calibration curve obtained on Cu using reference materials, at different laser shot distances and in different moisture conditions.

La Metallurgia Italiana - September 2026

pagina 43


Attualità industriale - Industry news To ensure operator safety, a laser safety zone must be

(NOHD) of 8 m is relatively low for this optical power

defined. The focal distance of the system is 3 to 8 m

[5], which is achieved through a combination of the laser

(physically limited by optics and autofocus stage stroke).

wavelength used and the optical design.

However, due to the nature of steel, which can be highly reflective, the eye safety should be based on the distance

After calibration and pre-trials, the LIBS sensor has been

from the excepted plasma generation site to account for

mounted together with the stereo camera system on a

a possible specular reflection. Thus, the LIBS system has

gantry for simultaneous shape detection and surface anal-

been independently classified as safe 8 m away from the

ysis of the scrap pieces. Figure 5 shows the system in a

possible specular reflection, due to an energy density in

laboratory set-up which has been used for first analysis

the reflected laser beam too low to cause eye damage

trials.

after that distance. The Nominal Ocular Hazard Distance

Fig.5 - LIBS analyser mounted on a gantry: Top view (left) and bottom view with stereo camera (right).

Edge processing of multi-modal sensor information

an image of the whole truck load. Then a segmentation

As LIBS measurements are inherently limited to the sur-

algorithm identifies individual scrap pieces from the RGB

face of the material, the resulting data require interpreta-

images. Subsequently, LIBS measurements are spatially

tion in a spatial and statistical context to enable a repre-

assigned to the corresponding objects by merging the

sentative bulk composition estimation. This is particularly

segmented mask with the LIBS measurement points. This

relevant for HMS which is rather inhomogeneous in terms

procedure with the different steps is schematically shown

of size and shape of the different scrap pieces. Thus, the

in figure 6.

data, which provide the surface analysis of single scrap pieces, and the image information for shape and size analysis of the different scrap pieces is processed by a machine learning (ML) algorithm to classify the different HMS pieces. First, the different images of the RGB cameras taken during scanning of the truck are stitched to provide

La Metallurgia Italiana - Settembre 2026

pagina 44


Industry news - Attualità industriale

Fig.6 - Scheme for alignment of sensor information for assessment of surface scrap composition. The dual camera system will finally provide information

7). The primary output of the system is the estimated mean

on the height of the scrap load in the truck. Based on the

elemental concentration of the scrap batch, with a first ap-

above-described mapping with piece-wise extrapolation

plication for the Cu content. Therefore, the combination

of the surface analysis, geometrical and compositional in-

of sensor information and their ML-based analysis repre-

formation are combined to derive an estimate of the bulk

sent the digital twin for the bulk analysis of the scrap load.

chemical composition of the entire scrap load (see figure

Fig.7 - Principle of extrapolation to assign the LIBS analysis to the bulk material.

La Metallurgia Italiana - September 2026

pagina 45


Attualità industriale - Industry news Given the heterogeneous nature of HMS, the represen-

ASSESMENT AND CONTROL OF FOAMY SLAG QUALI-

tativity of surface measurements is a key aspect. The ap-

TY IN ELECTRIC ARC FURNACES

proach leverages the relatively low fill volume of HMS

To provide a high electrical power input in the EAF it is

scrap deliveries, which results in a high accessible sur-

necessary to avoid any radiation losses of the electric arcs.

face-to-volume ratio, allowing a substantial portion of the

In high power furnaces these arcs can reach open length

material to be characterised during scanning.

up to 500 mm. It is a well-established technique to cover the arcs by the metallurgical necessary slag. To improve

Validation approach

the shielding, the slag can be foamed, which increas-

To evaluate the performance of the system, a campaign

es its height significantly and guarantees an effective arc

with a demonstrator set-up is planned in which several

shielding. Metallurgical slags in the EAF process mainly

truck loads are analysed using the developed sensor sys-

consist of CaO and MgO provided by slag formers and the

tem. The estimated bulk compositions will be compared

oxidation reaction products SiO2, Al2O3 and FeO, which

against reference values obtained from handheld XRF

are the result of the reaction of Fe and other elements in

measurements on selected scrap samples and from sub-

the liquid pool with the hypersonic injected oxygen. The

sequent meltdown analyses. This combined validation

foamy slag is generated by the reduction of FeO in the slag

approach enables assessment of both local measurement

by injecting coal, which leads to the formation of CO-bub-

accuracy and the reliability of bulk composition estima-

bles in the slag (figure 8).

tion under industrial conditions.

Fig.8 -Reactions to generate foamy slag. Adequate control of the foaming process and the stabili-

the analysis of the arc harmonics or structure-borne vibra-

ty of the foamy slag is essential for an effective electrical

tion sensors [6-9] have been applied to assess the foamy

energy input in the EAF and therefore an important task

slag quality. In DiGreeS, a hybrid dynamic process model

for EAF process control. As foaming is controlled by the

driven by continuous multi-modal online sensing will be

reactions between Fe, O2 and C, it is necessary to control

applied to foamy slag control. The sensors and process

the injection of coal and oxygen in a proper way to ensure

models for assessment of the foamy slag quality during

a good foamy slag quality. Today, mainly fixed operating

the refining phase of the EAF process will be installed at

patterns are used, and the quality of the foaming process

the AC EAF of Saarstahl Ascoval. The overall concept is

quite often depends on the individual operator skills rath-

sketched in figure 9, and the single components will be

er than real-time measured information of the process

described hereafter.

status. So far, only isolated sensors like acoustic sensors,

La Metallurgia Italiana - Settembre 2026

pagina 46


Industry news - Attualità industriale

Fig.9 -Concept for foamy slag control.

Multi-modal sensor system for assessment of the foamy

adapters and protective housings to collect the vibrations

The multi-modal sensor system for assessment of the

trodes. Each of the two interchangeable EAF vessels will

slag quality

foamy slag quality consists of four components (see fig-

ure 10): (i) Rogowski coils, which measure the arc currents across a very wide frequency band, are already installed at

each phase of the high current system. The signals will be analysed by Fourier transformation or Wavelet analysis to

detect the arc current harmonics and short-term fluctuations of the current. This measurement data provides in-

sights into the stability of the electric arc and the formation of gas bubbles, thus providing information about the melting process and the formation (height and stability) of the foamy slag.

Two types of acoustic emission (AE) sensors will also be in-

stalled. (ii) An air-borne (AB) AE sensor, realised by a dust-

proof microphone will be installed several meters away from the EAF. The temperature at this location will be max.

50 °C. This AB-AE sensor will allow to detect the character-

of the furnace in the immediate vicinity of the three elecbe equipped with 3 permanently attached sensors. These SB-AE sensors are temperature stable up to 100 °C, with

a frequency range up to 5-10 kHz and possible amplitudes

of up to 10 g. The SB-AE sensors will detect the acoustic signature of the electric arc, which is damped by the slag in the vessel and will provide information about the foamy slag height and the melting behaviour. In detail, the SB-AE

sensor signals are related to the EAF process stability in different stages of the process and to the foamy slag height

in the refining phase of the furnace operation. This information will be correlated with the input from the evaluated

information from the high-current system to improve the quality and liability of the already used measurement tech-

nique. (iv) Finally, an already installed laser-based optical off-gas measurement system [10] allows the in-situ online

analysis of CO and CO2 with an accuracy of +/- 2% of the

istic noise evolution during the melting and refining phase,

absolute value with a time delay of less than one second.

into the environment, superimposed on the ambient noise.

the EAF, as shown in figure 7, allowing a real-time analysis

by the ear of an experienced operator, although the cutoff

tion of the EAF duct with no direct contact to the hot and

tended compared to human hearing. (iii) In addition, three

measurement process. The system allows to monitor the

part of the water-cooled panels with permanently fixed

and CO2 formation from the steel bath, which induces the

i.e. acoustic noise originating from the EAF and emitted

This sensor subsystem is positioned in the off-gas duct of

This mimics the acoustic quality analysis of the foamy slag

of the off-gas stream. It covers a large part of the cross-sec-

frequency (> 20 kHz) and sensitivity are significantly ex-

dust-loaded off-gas, therefore there is no distortion of the

structure-borne AE sensors will be mounted on the upper

evolution of the decarburisation process by the level of CO

La Metallurgia Italiana - September 2026

pagina 47


Attualità industriale - Industry news foaming of the slag. The provided measurement informa-

to provide information on the energy losses via the off-gas.

control of supersonic oxygen and coal injection processes.

improve step by step the quality of the calculated results.

tion is also crucial for real-time process monitoring and In addition, off-gas temperature and velocity are measured

This will deliver valuable input for the process model to

Fig.10 - Components of the multimodal sensor system: Rogowski coils, microphones, vibration sensors and off-gas analysis.

Dynamic EAF process model

The model was validated with process data of around 300

An existing dynamic EAF process model [11], which is

heats and already provides accurate results for the evolu-

based on a dynamic energy and mass balance calculation

tion of the melt temperature as well as for carbon and ox-

in combination with thermodynamic calculations for met-

ygen content. It will be extended with a hybrid model part

allurgical reactions like decarburisation and dephosphori-

for determination of the properties of the foamy slag with

sation, has been adapted to the operation practice and

focus on the foamy slag height, which will in turn affect

process data which are available at the EAF of Saarstahl As-

the efficiency of the electrical energy input and the energy

coval. The model uses cyclically acquired and event-driv-

losses. The structure of the extended EAF process model

en process data to monitor on-line the evolution of the

with input and output data is shown in figure 11.

most important process state values like melt temperature as well as steel and slag amount and composition.

Fig.11 - Structure of the extended dynamic EAF process model.

Figure 12 shows the simulation results for steel tempera-

tent for an example heat of the Saarstahl Ascoval furnace.

ture, meltdown degree as well as carbon and oxygen con-

La Metallurgia Italiana - Settembre 2026

pagina 48


Industry news - Attualità industriale

Fig.12 - Simulation results for melt temperature and meltdown degree (right) and carbon and oxygen content (left) for a Saarstahl Ascoval example heat.

For the determination of the properties of the foamy slag a hybrid process model will be developed. The already existing, physics-based EAF model includes a dynamic slag

balance calculation, which provides the evolution of slag amount and composition. The model part for the decar-

burisation process allows to calculate the intensity of CO bubbles generation. This enables a first rough estimation

of the foamy slag properties. In addition, the data of the

multi-modal sensor system will feed an AI model for dynamic foamy slag quality assessment. The estimation of

the foamy slag quality from the physics-based model will be en-hanced by a correction term provided from the AIbased model for a comprehensive hybrid model for the

foamy slag height which allows an enhanced on-line process monitoring via a digital twin. The principle of this hybrid model is shown in figure 13.

Fig.13 - Hybrid model for prediction of the foamy slag height. Built on this real-time information of the foamy slag perfor-

be collected and analysed. AI-methods will be used to de-

oxygen injection as well as electrical power parameters will

the different EAF process steps like melting and refining.

mance, a dynamic control and optimisation of carbon and be developed to increase the efficiency of the electrical energy input.

Validation approach

After implementation of the multimodal sensor system at the EAF of Saarstahl Ascoval, sensor and process data will

La Metallurgia Italiana - September 2026

termine characteristic patterns of the measured signals for

These patterns will be correlated with already used parameters from the EAF control system.

Additionally, for training of the hybrid model for assess-

ment of the foamy slag performance, the furnace operators

will provide for a larger amount of heats their judgement on the foamy slag height by a classification system. The pre-

pagina 49


Attualità industriale - Industry news trained hybrid model will then be integrated in the overall dynamic EAF process model, which will be implemented

for on-line process monitoring and further fine-tuning at the Saarstahl Ascoval plant.

provide input for control of power input and the injection parameters for coal and oxygen.

CONCLUSIONS AND EXPECTED INDUSTRIAL IMPACT

CURRENT PROJECT STATE AND NEXT STEPS

To date, the sensors for scrap characterisation have been

selected, adapted and calibrated, and a first version of the analysis software has been developed. The whole system

has already been tested in a laboratory set-up and will be transferred to a demonstrator set-up for scanning complete trucks later in the project.

Regarding slag foaming control, the existing dynamic process model has been adapted to the Saarstahl Ascoval EAF.

The novel sensors have been selected and are currently installed at the furnace. After commissioning and the devel-

opment of sensor data preprocessing algorithms, the training phase of the hybrid process model will start. A Human

Machine Interface will be developed to support the operators by providing information on the process evolution in-

cluding the status of the foamy slag to keep the process in optimal conditions. By comparison of the saved patterns, the control parameters and the human interaction with the

control system, the system shall achieve valuable infor-

mation about the quality of the running process and shall

The already performed preparation work for integration

of novel sensors, process models and AI/ML algorithms

in DiGreeS demonstrates a highly promising pathway to sustainable EAF steelmaking. By transferring these components into a unified digital twin architecture, the project

is advancing to TRL7 demonstrators in operational envi-

ronments. Ultimately, these AI-driven solutions for digi-

talization of the EAF steelmaking process are expected to facilitate a higher percentage of HMS use without affecting quality, and a significantly improved foamy slag perfor-

mance to decrease energy consumption and thus the CO₂ footprint of EAF steelmaking. ACKNOWLEDGMENT

The DiGreeS project is funded by the European Union in the Horizon EUROPE program within CSP under grant

agreement no. 101178079. Views and opinions expressed are however those of the author(s) only and do not nec-

essarily reflect those of the European Union. Neither the European Union nor the granting authority can be held responsible for them.

REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11]

European Commission. (2019). The European Green Deal (COM(2019) 640 final) EUR-Lex - 52019DC0640 - EN - EUR-Lex Harmon, R., Russo, R., “Laser-Induced Breakdown Spectroscopy”, Treatise on Geochemistry, Vol. 15, 2014, p. 245-272 Rodolfa, K., Cremers, D., “Capabilities of surface composition analysis using a long laser-induced breakdown spectroscopy spark”, Applied Spectroscopy, 58(4), 2004, p. 367–375 Diaz-Romero D., Van den Eynde S., et al., “Real-time classification of aluminum metal scrap with laser-induced breakdown spectroscopy using deep and other machine learning approaches”. Spectrochimica Acta Part B: Atomic Spectroscopy, Vol. 196, Oct. 2022 Cremers D.A., Radziemski L.J., Handbook of Laser-Induced Breakdown Spectroscopy, 2nd Ed., Wiley, 2013, 426 pp, Chapter 2 and Appendix A Nikolaev A.A., Tulupov P.G., Ivekeev V.S.; “Comparative Analysis of Modern Electric Control Systems of Electric Arc Furnaces”, 2020 International Conference on Electrical Power Engineering (UralCon), 2020, 464 - 468 Torres-Renteria A., Damián-Cuallo M, Mayo-Maldonado J., Micheloud-Vernackt O., “Analysis of electric arc furnaces efficiency via frequency spectrum-based arc coverage detection”, Ironmaking and Steelmaking, Vol. 44, No. 4 2027, p. 255 Matschullat, T., Rieger, D., Krüger, K., Döbbeler, A., “Foaming slag and scrap melting in electric arc furnace – a new and very precise detection method with automatic carbon control”, Arch. of Metallurgy and Materials 53 (2008) 2, 399-403 Sedivy, Ch., Krump, R., “Tools for foaming slag operation at EAF steelmaking”, Archives of Metallurgy and Materials 53 (2008) 2, 405409. Graf A., "Decarbonization by Process Optimization - Off-Gas Results Using Deep View Infra-Red Technology", 7th European Steelmaking and Application Days (ESTAD), Verona, Oct. 2025 Kleimt B., Krieger W., Mier D., Arteaga A., Unamuno I., “Model-Based Decision Support System for Electric Arc Furnace (EAF) Online Monitoring and Control”. Metals 2023, Vol. 13, Issue 8, 1332 TORNA ALL'INDICE >

La Metallurgia Italiana - Settembre 2026

pagina 50


Scientific papers - Rolling

DOI 10.36146/2026_09_51

Towards workshop application of ring rolling simulations with embedded machine control K. van Putten, S. Stergianou, A. Gohr, A. Neumann, D. Michl V. Horáček, M. Vindyš

Radial axial (RAW) ring rolling is a well-established incremental forging process which produces seamless rings and shells, both with rectangular as well as with profiled cross sections. Due to the complex interactions between the mutually influencing radial and axial roll gaps, the ring position, and the numerous simultaneous tool movements, finite element (FE) simulation has become an established method for process modelling and analysis. Although FE ring rolling simulation coupled with closed-loop control is sometimes reported as state of the art, the process control was only approximated before. Due to this, very large rings and large profiled rings have proven almost impossible to simulate. A direct coupling between FE-simulation and the real RAW machine control overcomes these difficulties by utilizing the sophisticated control algorithms enabling real-world virtual production of such very large rings. To simulate ring rolling processes as close to reality as possible the SMS group’s real machine control (CARWIN®) is embedded in the FE-simulation. Feasibility has been proven and the model is validated by industrial ring rolling processes. The industrial relevance of these coupled ring rolling simulations has been demonstrated at Bohemia Rings. The machine control including interface to the FE-simulation ‒ Rolltech RPS ‒ operates independently and can be implemented in various commercial FE-software. A roll-out to SMS group ring rolling machine operators, with emphasis on accessibility for SME, is available.

KEYWORDS: RING ROLLING; PROCESS MODELLING; MACHINE CONTROL; CLOSED LOOP; CONTROL; FINITE ELEMENT SIMULATION; VIRTUAL PRODUCTION. INTRODUCTION Radial axial (RAW) ring rolling is a well-established incremental forging process which produces seamless rings and shells with outstanding productivity, both with rectangular as well as with profiled cross sections, with high demands in terms of mechanical properties and microstructure. Rolled rings and shells are applied in many industrial fields, e.g. wind power, nuclear industry, general construction of vessels, flanges, valves and bearings. Due to the complex interactions between the mutually

Koos van Putten, Stefan Stergianou, Alexander Gohr, Alexander Neumann, Dennis Michl SMS group, Germany

Václav Horáček, Marek Vindyš Bohemia Rings

influencing radial and axial roll gaps, the ring position, and the many simultaneous tool movements, finite element (FE) simulation has been established as a method for process modeling and analysis. However, many parameters of the ring rolling process are highly dependent on changes during the process. Consequently, realistic

La Metallurgia Italiana - September 2026

pagina 51


Memorie scientifiche - Laminazione modeling requires adaptive adjustments of tool move-

the start of the rolling process is managed by the control

ments during the rolling process. This is not possible with

system. In the model setup, only realistic boundary con-

conventional pre-defined tabular specifications of tool

ditions for all degrees of freedom present in real ring roll-

movements in the simulation and highlights the need for

ing processes are assumed; additional stabilization mea-

integrating control within the simulation model. Although

sures are not required.

the FE simulation of ring rolling combined with closedloop control is sometimes considered state-of-the-art

In the second step, the control module, including the

[1-3], process control has mostly been approximated.

communication interface and machine parameters, known

As a result, simulating the rolling of very large rings and

as “Rolltech RPS,” is deployed. The geometry and pro-

large profiled rings has proven to be nearly impossible. A

cess specifications for the rolling process (preform and

direct coupling between FE simulation and the real RAW

rolling geometry, rolling curve specifications, ring growth

machine control overcomes these difficulties by utilizing

speeds, etc.) are entered identically to the input at the

sophisticated control algorithms that already enable the

control panel in the control cabin of the real ring rolling

production of such challenging rings in an industrial set-

machine. Based on this, the module accounts for process

ting. To address this, the SMS group integrated its actual

limits as well as the constraints of the radial-axial ring roll-

machine control system (CARWIN®) into the FE simula-

ing machine, making it suitable for rings of all dimensions,

tion through a bi-directional interface, forming a closed-

including very large ones, with both rectangular and pro-

loop control system.

filed cross-sections. Instead of determining optimized, potentially fictitious load paths for individual tools [5], the

MODEL SETUP

comprehensive simulation of all machine functionalities

The model setup is essentially carried out in two steps.

allows for the nearly complete determination of ideal process parameters for real rolling operations, without prior

First, the FE model is set up using commercial simulation

test rolling. This is achieved independently of the plant or

software tailored for metal forming and specifically ring

software supplier, ensuring that the user’s specific exper-

rolling processes. The geometries of the tools and the

tise remains protected at all times.

ring blank can be imported from CAD software or Rolltech Profiles [4]. Optionally, the result of a previous pro-

Another advantage of the module is its independence

cess stage, such as upsetting and piercing, is transferred

from the simulation software used, allowing it to be com-

as preform into the ring rolling simulation. An approxi-

bined with solutions from various manufacturers. The

mate initial positioning of the tools should be performed

simulations presented here were conducted using Simu-

during model creation, while precise tool positioning at

fact Forming [6, 7].

Fig.1 -Technological concept of embedding the online ring rolling machine control.

La Metallurgia Italiana - Settembre 2026

pagina 52


Scientific papers - Rolling During the simulation—specifically the solution pha-

the full range of capabilities available on real machines is

se—data exchange occurs before and after each time in-

accessible in the simulation.

crement between the FE model and the control module

The coupling does not exclude the integration of custom

(figure 1). Measurement data captured in the FE model,

subroutines. Furthermore, microstructure simulations, as

such as geometry, rolling forces, and roll drive torques,

described in [8], can continue to be applied with the mo-

are transmitted to the control system. The control system

dule.

then determines the current state of the process and subsequently calculates new target specifications for tool

VALIDATION BY MEANS OF ROLLING A RING WITH A

movements, aligned with the capabilities of the machine.

RECTANGULAR CROSS-SECTION

The ring rolling process can be monitored both in the

The model was validated in two ring rolling processes

simulation software and in a “Quick Viewer”, where the

with identical settings and specifications. First, a rolling

ongoing recording of process data during the simulation

process for a 184 kg ring was simulated, expanding from

can be tracked. The visualization of this data, such as the

an outer diameter of 464 mm to 1418 mm. Subsequently,

current ring growth speed, forces, and drive torques, al-

two rings, as similar as possible, were rolled on the real

lows for an assessment of the actual state within the pro-

counterpart of the simulated machine, an SMS radial-axial

cess and the decision on whether manual intervention

ring rolling mill of type RAW 100/100-1500-500, using the

might be necessary. Such manual intervention, typically

same setting parameters and specifications. The simula-

possible through handwheels on the real machine, is also

tion results were then compared with those from the real

possible in Rolltech RPS via digital handles, ensuring that

rolling operations.

Fig.2 - Comparison between measured and calculated rolling force, ring geometry, and rolling curve.

La Metallurgia Italiana - September 2026

pagina 53


Memorie scientifiche - Laminazione A comparison of the logger records indicates small vol-

aligns well with the real progressions (figure 2). The com-

ume differences due to production (forming of the pre-

parison of the rolling curves shows that the volume loss

form) compared to the simulation, in which the ideal ring

due to scale loss is absent in the simulation but does not

volume matching the trials was used. Nevertheless, the

lead to significant deviations. The calculated rolling forc-

77.3 seconds rolling time of the model falls exactly be-

es also agree well with the measured forces (figure 2).

tween the real measured times of 75.6 and 77.6 seconds. Furthermore, the geometry progression, measured by

The comparison confirms that the simulation with the in-

means of the outer diameter (OD), the inner diameter (ID),

tegrated control system effectively models and predicts

the height (h), and the wall thickness (s), in the model

real-world ring rolling process.

ROLLING OF A RING WITH PROFILED CROSS SECTION

Fig.3 - Comparison distribution in the cross-section of the profiled ring within the radial roll gap before and after the ring rolling process.

Fig.4 - Effective plastic strain distribution from the FE simulation with coupled machine control of the rolling process for a 4.3-ton complex profiled ring featuring a symmetrical outer profile and an asymmetrical inner profile on a RAW 500/400. An example of the capabilities is demonstrated by the de-

ter, i.e. in the same heat, by transposing the main roll and

sign of the second stage of a complex profiled ring with a

the mandrel via special functions of the SMS system, or

symmetrical outer profile and an asymmetrical inner pro-

after reheating and replacing the tools with profiled ones,

file, see figure 3. The 4.3-ton ring is initially rolled rect-

the ring is rolled to have a profiled cross-section and the

angularly to approximately 2.1 meters in outer diameter

target dimensions of approximately 3.1 meters in outer

and 330 mm in height on a RAW 500/400. Immediately af-

diameter while maintaining the same height. This second

La Metallurgia Italiana - Settembre 2026

pagina 54


Scientific papers - Rolling rolling stage is completed in approximately 380 seconds.

type RAW 1250/800 9000-1600, maintaining a constant

Manual displacement of the ring using a handwheel, com-

height of approximately 1.5 meters, expanding in diam-

bined with the selection of an appropriate intermediate

eter from 1.8 meters to 6.2 meters. The model was en-

geometry determined through simulation, the complex

hanced with ring carrier rolls, which are installed on the

profile is fully filled, and the occurrence of retractions on

corresponding machine to prevent sagging of rings with

the front faces of the profiled ring is minimized. In the FE

large diameters. The initial mesh of the shell consisted of

model, the ring’s mesh began with 45,000 elements and

47,900 elements, expanding to 166,700 elements during

grew to 67,000 elements during the simulation. Figure 4

the simulation. The rolling process was completed in ap-

illustrates the situation at the end of the process, where

proximately 380 seconds. The unstable ring geometry,

the ring meets all requirements for roundness and pro-

with a wall thickness of less than 100 mm at the end of

file filling. This confirms that the ring can be successfully

the process, requires extremely precise coordination of

manufactured using the described process sequence and

tool movements, particularly the guidance of the ring by

management.

the centering rolls, to prevent collapse. Figure 5 shows the final result: the ring successfully achieved the target

ROLLING OF A GIANT NON-PROFILED RING

dimensions, exhibiting uniform deformation around the

To further demonstrate the system capabilities, a 22-ton

circumference, and both ovality and conicity are within a

ring was rolled on an SMS radial-axial ring rolling mill of

range of a few millimeters.

Fig.5 - Effective plastic strain distribution from the FE simulation with coupled control of the rolling process for a 22-ton shell with a final diameter of 6.2 meters and a height of 1.5 meters on a RAW 1250/800 9000-1600.

INDUSTRIAL APPLICATION AT BOHEMIA RINGS

the rolling mill operator sees and sets at the control pan-

The industrial relevance of these coupled ring rolling sim-

el during operation, it enables direct know-how transfer

ulations has been demonstrated at Bohemia Rings. In 2025,

from simulation to the workshop and significantly im-

Bohemia Rings acquired Simufact Forming and Rolltech

proves process understanding.

RPS to (on the one hand) simulate and optimize current production of non-profiled rings made from various alloys

As one example among many simulated processes, the

and (on the other hand) develop new products within its

rolling of a 351 kg thin-walled aluminum alloy (AW7075)

portfolio. Bohemia Rings operates an SMS radialaxial ring

ring is presented here. The preform and ring dimensions

rolling mill, type RAW 250/200 4200–660; its specifications

are given in table 1. The initial preform mesh comprised

and parameters are fully implemented in the machine con-

27,900 elements and increased to 30,700 during the simu-

trol. As the Rolltech RPS interface closely reflects what

lation. The real rolling process was completed in approx-

La Metallurgia Italiana - September 2026

pagina 55


Memorie scientifiche - Laminazione imately 203 s, while the simulation predicted 224 s. Force

At Bohemia Rings, ring rolling simulations with embedded

development in the radial and axial roll gaps, ring height,

machine control have progressed from basic ring rolling

and outer diameter over time have all been successfully

analysis toward a high-performance tool for developing

validated. The deviation between the final dimensions of

new products with workshop-oriented application.

the real and simulated ring is below 1.5% (table 1). Tab.1 - Preform and ring dimensions of thin-walled aluminum alloy ring, rolled and simulated by Bohemia Rings. dimension

preform

ring

Mm

CARWIN target mm

real process mm

simulation mm

deviation %

outer diameter

790

2209

2210.3

2208.9

0.063

inner diameter

230

2023

2021.5

2022.4

-0.045

wall thickness

319

93

94.4

93.2

+1.27

height

280

196

197.5

196.3

+1.2

Fig.6 - Effective plastic strain distribution from the FE simulation by Bohemia Rings with Simufact Forming and coupled control by Rolltech RPS of the rolling process for an aluminum alloy thin-walled ring with a final diameter of 2.2 meters and a height of 196 millimeters on a RAW 250/200 4200-660. CONCLUSIONS To achieve realistic simulation of ring rolling processes, especially for large rings and large profiled rings, direct coupling between FE simulation and the actual RAW machine control is indispensable. The SMS group has embedded its real machine control system (CARWIN®) fully into the FE simulation through a bi-directional interface as a closed-loop control system. This coupled model cre-

• • • •

through industrial trials. Consideration of the process limits as well as the constraints of the radial-axial ring rolling machine. Interface and operation analogous to the control panel in the control cabin. Applicable for ring rolling simulation of non-profiled as well as profiled ring cross-sections. Includes all functionalities of the simulated machine,

ates a digital twin of the ring rolling process with the fol-

such as manual ring displacement and automatic ring

lowing advantages:

positioning control.

•

Realistic simulation of the ring rolling process in very close agreement with reality, successfully validated

La Metallurgia Italiana - Settembre 2026

•

Ability to consider intervention by handwheels during simulation directly.

pagina 56


Scientific papers - Rolling

• •

Enables direct transfer of conclusions from simula-

the simulation software used and can be combined with

tion results towards the real physical process.

various commercial FE software tailored for metal form-

Industrial relevance has been demonstrated at Bo-

ing and specifically ring rolling processes. The coupling

hemia Rings, where ring rolling simulations with em-

between “Rolltech RPS” and the FE software is straight-

bedded machine control have progressed from basic

forward and emphasizes accessibility for small and medi-

ring rolling analysis toward a high-performance tool

um-sized enterprises (SMEs).

for developing new products with workshop-oriented application.

As of June 2025, a fully integrated coupling between “Rolltech RPS” and Simufact Forming (versions 2025.2

OUTLOOK

and onward) has been established. Additionally, efforts

The “Rolltech RPS” control module, designed for integra-

are underway to ensure compatibility with other software

tion into finite element simulations, is available for SMS

providers, such as Forge NxT.

radial-axial ring rolling mill operators. It is independent of

REFERENCES [1] [2] [3] [4] [5] [6] [7] [8]

Jenkouk, V.; Hirt, G.; Franzke, M.; 3D-FE simulation of ring rolling with integrated closed-loop tool motion control, Proceedings ICRF 1 (2012) Schwich, G.; Seitz, J.F.; Jenkouk, V.; Hirt, G.; Application of finite element analysis considering the complex tool kinematics of axialprofiling and dishing in ring rolling, Proceedings - ICRF 2 (2014) Schwich, G.; Jenkouk, V.; Hirt, G.; Realistic modelling of the tool kinematics of radial-axial ring rolling machines in finite element simulation, Proceedings – ESAFORM 19 (2016) Rolltech Profiles, SMS group, www.sms-group.com/plants/ring-and-wheel-rolling-machines Sartori, A.; Ricci, M.; Brun, M.; Konnerth, U.; Enhancing ring rolling simulation accuracy with the machine digital twin, Proceedings – IFM (2024), ISBN 978-88-98990-35-1, digital Simufact Forming 2024, Hexagon Manufacturing Intelligence, https://hexagon.com/products/simufact-forming An Innovate Approach to Automated Simulation of Full 3D Ring Rolling Process and Other Incremental Forming Processes, Proceedings – Metal Forming (2010), steel research international, volume 81 (2010), number 9, p.p. 202 – 205 Back, A.; van Putten, K.; Krämer, A.; Kumar, R.; FE-Simulation von Prozessketten mit Berücksichtigung der Mikrostrukturentwicklung, massivUMFORMUNG, September 2018, p.p. 36 – 41

TORNA ALL'INDICE >

La Metallurgia Italiana - September 2026

pagina 57


Memorie scientifiche - Laminazione

DOI 10.36146/2026_09_58

New insights into the online LUS grain size measurements D. Hoppe, T. Haschke, A. Sprock, C. Hassel, J. Hafer, L. Bäcke, J.-E. Thorberg, C. Jonsson, M. Malmström, F. Kneisel, M. Bärwald

In a collaborative initiative, an innovative laser-ultrasonic (LUS) device was installed at SSAB’s Hot Strip Mill (HSM) in Borlänge. This measuring device, along with the software for signal processing and evaluation, was developed, constructed, and implemented through a partnership involving SSAB, Swerim, EMG, and SMS group. Positioned downstream of the last rolling stand in the finishing mill, the austenite grain size is measured immediately after rolling. The LUS measurements were conducted on various steel grades with different dimensions and process parameters. The calculated process data for the tested strips were sourced from the SMS process models (PSC® pass schedule model and CSC cooling section model) and underwent comprehensive analysis alongside the measured process data and austenite grain size. This analysis encompasses the entire process, beginning with the discharge of slabs from the reheating furnaces, through the roughing mill, coil box, finishing mill, and cooling section, and concluding with the downcoiler. Extensive recalculations using the models PSC® and CSC were performed to quantify the correlation between the measured austenite grain size and the process conditions of the rolled strips. The detailed recalculations demonstrate a strong correlation with the measured grain sizes. Based on the results, for example, variations in the transfer bar thickness were carried out, leading to a more equal grain size along the strip length, which has also a noticeable effect on the subsequent cold rolling process. In the near future, the LUS measurements could be fully integrated into the Level 2 automation system, thus enabling the use of the austenite grain size within the PSC® pass schedule model, the CSC cooling section model, and the MPM microstructure property model. Consequently, these models could utilize the austenite grain size as a setpoint for direct process control in the hot rolling mill, thereby improving control of product properties and minimizing downgrading.

KEYWORDS: LASER ULTRASONIC; HOT ROLLING MILL; MICROSTRUCTURE; PROCESS MODEL; AUSTENITE GRAIN SIZE; MEASURING DEVICE; MICROALLOYED STEEL. INTRODUCTION The manufacture of microalloyed high-strength steels is a complex process requiring advanced technologies and the corresponding know-how to attain the desired mechanical properties. These steels often contain small amounts of alloying elements such as niobium, vanadium, titanium, molybdenum, and boron, which have a significant influence on the microstructure of the steel. With this composition, the steels demonstrate an excellent balance of strength, toughness, and weldability. These properties make them particularly suitable for demanding applications in various industries, such as the construction industry for load-bearing structures, the

Dietmar Hoppe, Thomas Haschke, August Sprock, Christoph Hassel, Joachim Hafer SMS group GmbH, Hilchenbach, Düsseldorf

Linda Bäcke, Jan-Erik Thorberg, Christer Jonsson SSAB EMEA AB, Borlänge, Sweden

Mikael Malmström

Swerim AB, Stockholm, Sweden

Franziska Kneisel, Matthias Bärwald EMG Automation Wenden, Germany

automotive industry for lighter and safer vehicle bodies, and the pipeline industry for transporting oil and gas. A

La Metallurgia Italiana - Settembre 2026

pagina 58


Scientific papers - Rolling remarkable feature of these steels is their high strength,

and other microstructural variables during hot rolling, as

with values ranging from 275 to 750 MPa combined with

it functions at high temperatures and with moving ob-

good formability [1, 2].

jects. It has been demonstrated that laser ultrasonic can analyze the grain structure in real time, including the di-

In addition to the chemical composition used, the micro-

rect measurement of dynamic recrystallization during de-

structure of a material mainly determines the mechani-

formation, which is crucial for process control [4, 18, 19].

cal and physical properties, including strength, ductility, hardness, resistance to corrosion, and wear. Of particu-

The data obtained with laser ultrasonics can be directly

lar importance is the austenite grain size, found at high

processed in process models, which are used in hot strip

temperatures during hot forming. Smaller austenite grain

production for plant control during rolling and cooling.

sizes impede the spread of dislocations, which increases

The setup model (PSC®, Pass Schedule Calculation) cal-

strength, makes crack propagation more difficult, and si-

culates the load distribution during forming and defines

multaneously improves ductility. This effect is deliberate-

the pass schedule to achieve the desired strip thickness.

ly exploited, particularly with microalloyed steels.

Key aspects here are the ductility of the material and the rolling temperature, which are important for the recrystal-

By adding niobium (Nb) and titanium (Ti), the austenite

lization behavior of the material during forming. The cool-

grain size can be effectively controlled and reduced al-

ing model (CSC, Cooling Section Control) determines the

ready during hot rolling. These elements form fine pre-

volume of water required to attain a specific temperature

cipitates that inhibit grain growth during heating and roll-

distribu-tion in the cooling section. Here, the exact de-

ing (grain-boundary pinning) and result in finer austenite

scription of the transformation process and the energy

grains. Dissolved niobium accumulates predominantly

balance play a significant role in solving the transient heat

at austenite grain boundaries and delays their movement

conduction equation, as the austenite grain size influenc-

(solute-drag effect), thus restricting the growth of the

es the transformation behavior [5].

austenite grains [3, 20, 21]. This work describes the evaluation of process data for the In hot rolling, steel is formed at high temperatures (800-

roughing and finishing mills collected during hot rolling

1200°C) and a desired final thickness is attained. This leads

at SSAB’s HSM in Borlänge and the comparison between

to a refinement of the austenite grain, which has a positive

simulated and measured austenite grain sizes. It demon-

effect on the mechanical properties. A precise determina-

strates how the austenite grain size develops along the

tion of the microstructural changes occurring in the mate-

strip length under various process parameters [6]. In ad-

rial and of the austenite grain structure during hot rolling

dition, the influence of the measured austenite grain size

is important not only for achieving the desired dimension

on the required water quanti-ties in the cooling section is

but also for the resulting material properties. Innovative,

examined.

non-contact measurement methods could enable precise process control and significantly increase product quality

METHODOLOGY AND PARAMETERS

and efficiency in the hot rolling process.

Figure 1 shows the HSM at SSAB in Borlänge viewed from left to right, starting with the reheating furnaces, the

This is where laser-ultrasonic measuring method comes

roughing mill, the coil box, the finishing mill, the cooling

into play, a non-contact technique that uses lasers to

section, and the downcoilers.

generate and detect ultrasonic waves. The method utilizes pulses from an excitation laser and a second laser to detect the waves, allowing precise analysis of the material properties without physical contact. This technology is particularly suitable for measuring the austenite grain size

La Metallurgia Italiana - September 2026

pagina 59


Memorie scientifiche - Laminazione

Fig.1 - HSM SSAB, Borlänge - Process data - Simulation of the microstructure evolution. The newly installed LUS measuring device, shown in fig-

rometer (see figure 1).

ure 2, is located in the HSM below the finishing mill py-

Fig.2 - LUS measuring device: (a) during the test phase at EMG’s workshop in Wenden, showing the power

supply (gray) and the laser unit (yellow); (b) installed at SSAB’s HSM in Borlänge, positioned on vertical rails with a working distance of 60 cm from the rolled strip.

Figure 2(a) shows the new LUS measuring device during

al. [4, 18]: (a) generation of broadband ultrasound by laser

the test phase at EMG’s workshop in Wenden. Shown are

ablation, (b) detection of the surface response for multi-

the power supply (gray) and the optical head with the la-

ple backwall echoes, (c) calculation of the frequency-de-

sers (yellow). The measuring device can be moved verti-

pendent attenuation, and (d) conversion to grain size via a

cally along the visible rails, as shown in figure 2(b). During

calibration curve. These steps are illustrated schematical-

rolling, it is moved into position beneath the rolled strip

ly in figure 3.

with a working distance of 60 cm.

The total ultrasonic attenuation, α, can be decomposed

Grain size determination by laser ultrasonics follows four sequential steps, as described in detail by Malmström et

La Metallurgia Italiana - Settembre 2026

into three contributions [4]:

pagina 60


Scientific papers - Rolling

where

represents material absorption,

diffraction-related attenuation, and

where

describes the

depends on the scattering regime (Rayleigh:

; stochastic: is the frequency,

where

is the

),

is the mean grain diameter,

is the temperature, and

accounts

frequency-dependent grain scattering. The scattering contribution follows:

for temperature-dependent elastic anisotropy. In the frequency range of approximately 1-40 MHz relevant here, setting

provides a practical linearization [4], reduc-

ing the total attenuation to:

is a constant offset comprising material absorption and other non-scattering contributions, and:

Since only the third-order attenuation coefficient,

, contains information of grain size, the mean grain diameter is

obtained as [4]:

where

is a temperature- and material-

dependent transfer function. In practice, this calibration function is determined experimentally by comparing LUS

attenuation measurements on reference samples with grain sizes obtained by optical microscopy or by EBSD with parent-grain reconstruction [4].

Fig.3 - Steps in LUS grain size determination: (a) generation of broadband ultrasound by laser ablation, (b)

detection of the surface response from multiple backwall echoes, (c) calculation of the frequency-dependent

attenuation,

, and (d) conversion to grain size via calibration curve. Adapted from Malmström et al. [4].

The L2 online setup models of the roughing and finishing

(see also figure 1 - Process Data). The same applies to the

mills each calculate the pass schedules and transmit them

L2 online setup model of the CSC cooling section, which

to the L1 control systems, with defined interfaces facili-

calculates the required water quantities in the individual

tating communication between the models themselves

zones of the cooling section based on the input variables

La Metallurgia Italiana - September 2026

pagina 61


Memorie scientifiche - Laminazione and also transmits these to the L1.

If the slab contains microalloying elements, such as nio-

The online process data from the PSC® RM and FM setup

bium, the precipitation behavior is simulated as a function

models were recalculated using the model PSC® Offline

of temperature, recrystallization kinetics, and deforma-

in order to generate a closed representation of the operat-

tion. The niobium content remaining in solution at the

ing mode. Based on the closed time-temperature-defor-

end of the finishing mill, for example in the penultimate

mation history, it is possible to simulate the microstruc-

stand, is included in the calculation of the critical degree

ture evolution from the furnace exit to the finishing mill

of deformation and other model variables.

pyrometer, where the measured grain size can then be compared with the simulated austenite grain after the last

Governing Equations for Microstructure Evolution

hot-rolling deformation.

The simulation of microstructure evolution within the

The simulation of microstructure evolution starts from a

PSC® pass schedule model follows well-established

standard austenite grain size at the slab extraction tem-

physically based formulations for recrystallization kinet-

perature upon exit from the reheating furnace. It is as-

ics, grain growth, and precipitation. The core equations

sumed that all microalloying constituents, such as niobi-

are briefly summarized below.

um, are fully dissolved. During transport to the roughing stand, potential grain growth and the temperature evolu-

Recrystallization Kinetics (JMAK)

tion are calculated. Furthermore, grain refinement (i.e.,

The fraction recrystallized,

recrystallization during forming in the roughing and fin-

described by the Johnson-Mehl-Avrami-Kolmogorov

ishing mills) and possible grain growth after full recrystal-

(JMAK) equation [11, 12]:

, as a function of time, , is

lization are taken into account [7-9].

where

is the time for 50% recrystallization and

is the

Avrami exponent, which depends on the nucleation and growth mechanism. The parameter

where

,

,

constants, tion, and

, and

pressed as a function of the initial grain size , the strain rate , and the temperature

, the strain

[13, 20]:

is commonly ex-

are empirically determined material

is the activation energy for recrystallizais the universal gas constant.

Grain Growth Following complete recrystallization, grain growth is governed by a modified grain-growth law that accounts for the retarding influence of second-phase particles present in the microstructure [14, 20]:

where

is the grain diameter at time ,

growth exponent, nential constant, and

is the grain-

is a material-dependent pre-expois the activation energy for grain

present—such as those formed by microalloying elements—grain boundary migration is impeded by a Zener drag pressure,

[14, 15]:

boundary migration. When finely dispersed particles are

La Metallurgia Italiana - Settembre 2026

pagina 62


Scientific papers - Rolling

where

is the volume fraction of particles,

boundary energy per unit area, and

is the grain

is the mean particle

radius. Grain growth effectively ceases when the driving pressure for growth falls below grain size,

, yielding a limiting

:

Precipitation Kinetics – Nucleation and Growth of

titanium, or vanadium plays a central role in controlling

Second-Phase Particles

austenite grain size and recrystallization behavior. The

In microalloyed steels, the precipitation of carbides, ni-

nucleation rate

trides, and carbonitrides of elements such as niobium,

classical nucleation theory (CNT) [16, 17]:

where

is the Boltz-

diffusion, and

is the activation energy for solute

for nucleation:

is the pre-exponential factor,

mann constant,

of precipitate particles is described by

is the critical Gibbs free-energy barrier

is the precipitate-matrix interfacial energy, and

which is a function of the degree of supersaturation

is the volumetric driving force for precipitation,

the relevant solute elements in the austenite matrix:

Here,

where

is the molar volume of the precipitate phase.

The supersaturation

is defined as the ratio of the actual

where

denotes the concentration of solute species

and

is the corresponding stoichiometric coefficient,

making this formulation applicable to a wide range of precipitate chemistries.

of

ionic activity product of the dissolved alloying elements to the temperature-dependent solubility product of the respective precipitate [17]:

cleation, growth, and coarsening—are frequently represented in an analogous JMAK framework. For the case of strain-induced precipitation, as described by Dutta, Palmiere, and Sellars [9], and further corroborated for Nb-V

The overall precipitation kinetics—encompassing nu-

multi-component systems [21]:

where

during rolling for 287 strips with the following parameters

is the precipitated fraction,

5% precipitation, and

is the time for

is the precipitation Avrami expo-

nent. LUS measurements were conducted in the hot strip mill

La Metallurgia Italiana - September 2026

(table 1), consistent with online measurement approaches reported for comparable industrial hot strip mill installations [19]:

pagina 63


Memorie scientifiche - Laminazione Tab.1 - Process parameters and chemical composition ranges of the 287 strips investigated. Furnace temperature:

1180 - 1290 °C

C content:

< 0.39 wt.-%

Final rolling temperature:

820 - 920 °C

Mn content:

< 2.3 wt.-%

Slab thickness:

217 - 223 mm

Nb content:

< 0.045 wt.-%

Transfer bar thickness:

26 - 31 mm

Ti content:

< 0.1 wt.-%

Final thickness:

2 - 12 mm

N content:

< 0.007 wt.-%

Strip width:

815 - 1600 mm

RESULTS

the strip; here, however, due to conservative safety mar-

Using the example of strip A, figure 4 shows the results of

gins, the LUS measurement starts after the strip head end

the recalculation of the operation, i.e., the rolling forces in

has passed the measuring point and ends before reach-

roughing (a) and finishing mill (d)-(i), temperatures (c) and

ing the strip tail end. Hence, some LUS data are missing at

(k), and speeds (b) and (j) and the comparison of the mea-

both ends of the strip. The available measured values are

sured and recalculated austenite grain size (l). In principle,

synchronized over the strip length.

the LUS device can measure from the first centimeter of

Fig.4 - Recalculation of pass schedule items for strip A: rolling forces in roughing (a) and finishing mill (d)-

(i), temperatures (c) and (k), and speeds (b) and (j), including a comparison of the measured and recalculated austenite grain size (l) (red - measurement results; blue - calculation results of model PSC® Offline).

In general, there is a very good match between the mea-

calculated strip speed (j) at the finishing mill pyrometer

sured rolling forces, rolling speeds, and temperatures.

is in good agreement with the measured values, which is

Only the recalculated transfer bar temperature (c) is ap-

reflected in the precise time mapping of the online pass

proximately 15-20°C below the measurement. The re-

schedules within the model PSC® Offline system. The

La Metallurgia Italiana - Settembre 2026

pagina 64


Scientific papers - Rolling finishing strip temperature (k) shows small deviations

lated austenite grain size for 12 strips (B - M), which again

between the calculation and measurement. Of particular

demonstrate a very good match. Strips B and C each show a

note is the good match between the measured and recal-

small gradient both in the measurement and in the recalcu-

culated austenite grain size (l), which stands at on the or-

lation, with the recalculated austenite grains approximately

der of 10 µm. This means that PSC Offline, based on the

1-2 µm larger than the measured grains. In the case of strip

and precipitate formation, thereby enabling accurate re-

this discrepancy has not yet been definitively determined.

®

recalculated pass schedules, is well suited to simulate the metallurgical processes of grain growth, recrystallization, production of the measured LUS austenite grain size.

D, the calculated austenite grain size is approximately 15 μm and the measured size around 12 μm. The reason for

Figure 5 shows comparisons of the measured and recalcu-

Fig.5 - Comparison of the measured and recalculated austenite grain size of 12 strips: (a) Strip B, (b) Strip C, (c) Strip D, (d) Strip E, (e) Strip F, (f) Strip G, (g) Strip H, (h) Strip I, (i) Strip J, (j) Strip K, (k) Strip L, (l) Strip M. (red measurement results; blue - calculation results of model PSC® Offline).

Figure 6 shows comparisons of the measured and recal-

furnace. It has not yet been definitively clarified whether

culated austenite grain size for 7 strips (N - T), which again

this represents an inhomogeneity in the austenite grain

demonstrate a very good match. A conspicuous feature of

size upon exiting the furnace or whether the phenome-

figure 6 is strip Q with its pronounced waveform. The pro-

non may be attributable to the differences in the dissolved

nounced waveform, with an amplitude of approximately

state of Nb.

3 μm, may be due to cold spots as the material exits the La Metallurgia Italiana - September 2026

pagina 65


Memorie scientifiche - Laminazione

Fig.6 - Comparison of the measured and recalculated austenite grain size of 7 strips: (a) Strip N, (b) Strip O, (c) Strip

P, (d) Strip Q, (e) Strip R, (f) Strip S, (g) Strip T. (red - measurement results; blue - calculation results of model PSC® Offline).

The waveform of strip Q can also be seen in the recalcula-

tion of the austenite grain size. Here, the sensitivity of the LUS measuring device is particularly evident, as is the capability of the model PSC® Offline to reproduce this behavior.

Based on the recalculations shown, further optimization

calculations were performed to standardize the austenite

grain size during rolling. Figure 7(a) shows the progression

of the measured and recalculated austenite grain size of strip U, which serves as reference. At a transfer bar thick-

ness of 26 mm, the measured values show a slight curva-

ture in the first half of the strip followed by an increase from 13 to approximately 17 µm. During optimization, the trans-

fer bar thickness has proven to be a potential influencing

parameter on the recrystallization behavior and, consequently, the austenite grain size; therefore, it was adjusted in small increments, i.e., increased, during the rolling program.

Fig.7 - Increase in the transfer bar thickness: (a) measured and recalculated austenite grain size of strip U (transfer bar thickness: 26 mm); (b) measured and recalculated austenite grain size of strip V (transfer bar thickness: 31 mm); (c) strip strength during subsequent cold rolling - comparison of strips U and V.

La Metallurgia Italiana - Settembre 2026

pagina 66


Scientific papers - Rolling Figure 7(b) shows the result of increasing the transfer bar

thickness from 26 to 31 mm for strip V, i.e., the progression of the measured and recalculated austenite grain sizes. In

both the calculation and the measurement, the austenite grain size remains at an almost constant level of approx-

imately 14 µm; in other words, its progression is clearly more uniform compared with strip U. From a metallurgical point of view, this somewhat smaller and more uniform

austenite grain size originates from the higher overall degree of deformation in the finishing mill resulting from the

greater transfer bar thickness. This evener curve is also apparent in downstream processes in the tandem cold mill, as reflected by the more consistent strength (figure 7(c)).

Strip U shows a distinct waveform with scatter in the cold yield point, i.e., a strength of approximately 750-780 MPa, which is not the case for strip V; here, the strength level remains almost constant at around 750 MPa.

The austenitization conditions, particularly the austeniti-

zation temperature, the holding time, and the deformation

above the transformation temperatures, have a consider-

able influence on the transformation temperatures and the

resulting microstructure of steel grades. This influence can

be attributed above all to the grain size of the austenite. A

higher austenitization temperature and/or longer holding

time generally lead to a coarser austenite grain size. If the carbides are not completely dissolved, a low austenitiza-

tion temperature or a holding time that is too short will lead to insufficient homogenization of the austenite. Particular attention should be paid when adding Nb and Ti, as the sta-

ble carbides and carbonitrides are difficult to dissolve and

at the same time constitute effective grain-growth inhibi-

tors for the austenite. This results in a finer austenite grain during forming [10]. The CSC cooling-section setup model

considers the austenitization conditions (PA) as an input variable for calculating the transformation temperatures.

Figure 8 shows the influence of grain size—the grain size

previously used in the model (upper section of the figure) and the currently measured grain size (lower section)—on the calculated water quantity for three strip thicknesses. In the model, a grain size of approximately 33 µm has been

used until now, while the current measured grain size is approximately 13 µm. This difference affects the calculated water quantity.

Fig.8 - Recalculated water quantities in the cooling section. Compared with the reference water quantity (100%, initial

more water is required. For other materials, however, it has

grain size: 33 µm), 13-15% more water is required when

been found that there are no, or negligible, changes in the

the recalculation is performed with the measured austenite

required water quantity when the measured austenite grain

grain size. From a metallurgical point of view, smaller aus-

size is included as an input variable in the CSC cooling-sec-

tenite grains lead to a larger grain-boundary surface area,

tion model.

which in turn forms preferential nucleation sites for ferrite and causes ferrite to form at an earlier stage during cooling.

CONCLUSION

The increased number of nucleation sites accelerates the

The newly developed LUS grain-size measuring device

phase transformation from austenite to ferrite; therefore,

was installed and successfully tested in the hot strip mill at

La Metallurgia Italiana - September 2026

pagina 67


Memorie scientifiche - Laminazione SSAB, Borlänge. For nearly 290 strips covering a wide field

To achieve improved uniformity of the austenite grain size,

of chemical composition and geometrical parameters, the

an increase in the intermediate strip thickness was derived

austenite grain size was measured directly during rolling un-

from optimization calculations and successfully verified

der normal process conditions. Recalculations of the pass

during rolling tests. Moreover, the more uniform austenite

schedules using the process model PSC® Offline showed

grain size had positive effects on subsequent process steps;

a very good match with both the measured rolling forces

for example, the strength of the tandem cold mill was more

and temperatures and the recalculated and measured aus-

consistent. Recalculations of the required water quantities

tenite grain sizes, i.e., with the simulation of grain growth,

using the CSC cooling-section model and the measured

recrystallization, and precipitation. For some strips, the

austenite grain size as input showed that some grades re-

measurements showed slight gradients that could also be

quired more water due to smaller austenite grains. Going

reproduced in the recalculation of the austenite grain size;

forward, the recalculations of the pass schedules and the

however, the causes of the observed deviations require

simulation of the microstructure evolution will be further

further investigation. Of particular note is the pronounced

optimized and validated by LUS measurements, building

waveform of the measured austenite grain size on some

on recent advances in the real-time measurement of dy-

strips, which is likely attributable to cold spots as the mate-

namic recrystallisation by laser ultrasonics. Further rolling

rial exits the furnace. This high sensitivity of the LUS mea-

trials and series of measurement campaigns with optimized

suring device is also reflected in the process model PSC®

rolling conditions are planned.

Offline through the recalculation of the austenite grain size. Nomenclature Abbreviations Abbr.

Definition

Abbr.

Definition

CNT

Classical Nucleation Theory

LUS

Laser Ultrasonic

CSC

Cooling Section Control

MPM

Microstructure Property Model

EBSD

Electron Backscatter Diffraction

PSC®

Pass Schedule Calculation

FM

Finishing Mill

RM

Roughing Mill

GLUS

Gleeble and Laser-Ultrasonics

L1

Level‑1 automation system

HSM

Hot Strip Mill

L2

Level‑2 automation system

JMAK

Johnson-Mehl-Avrami-Kolmogorov

Latin Symbols Symbol

Definition

Unit

Pre‑exponential factor (Arrhenius‑type kinetics, general)

model‑dependent

Pre‑exponential factor for nucleation rate Empirical material constant (recrystallization / growth model)

—

Empirical material constant (recrystallization / growth model)

—

Empirical material constant (grain growth / recrystallization kinetics)

model‑dependent

Concentration of solute species

wt.-%

Initial austenite grain diameter

µm

Mean austenite grain diameter

Limiting grain size due to Zener pinning

µm µm

Ultrasonic frequency

MHz

Volume fraction of precipitate particles

—

Pre‑exponential constant for grain growth

model‑dependent

Nucleation rate of precipitates Boltzmann constant

La Metallurgia Italiana - Settembre 2026

J

pagina 68


Scientific papers - Rolling Temperature-dependent elastic anisotropy factor

model‑dependent

Solubility product

—

Zener‑pinning constant / particle pinning coefficient

model‑dependent

Grain‑growth exponent

—

Grain‑boundary mobility Initial number density Nucleation rate (time derivative of

)

Avrami exponent for recrystallization

—

Avrami exponent for precipitation

—

Frequency‑dependence exponent (ultrasonic attenuation)

—

Mean precipitate particle radius

µm

Universal gas constant Supersaturation ratio

νi

J

—

Time

s

Time for 50 % recrystallization

s

Time for 5 % precipitation

s

Temperature

K / °C

Stoichiometric coefficient of solute species in the respective precipitate phase

—

Molar volume of precipitate phase Recrystallized fraction

—

Precipitated fraction

—

Greek Symbols Symbol

Definition

Unit

Model constant (ultrasonic scattering / kinetic relation)

—

Grain‑boundary or interfacial energy

J

Grain‑boundary energy per unit area

J

Precipitate-matrix interfacial energy

J

LUS calibration transfer function True strain

model‑dependent —

Strain rate Precipitate-matrix interfacial energy

Pa

Critical Gibbs free‑energy barrier for nucleation

J

Volumetric driving force for precipitation

J

Activation energy for solute diffusion

J

Activation energy for recrystallization

J

Activation energy for grain‑boundary migration

J

Activation energy for grain growth

J

Laser‑Ultrasonic Attenuation Terms Symbol

Definition

Total ultrasonic attenuation Material absorption attenuation

Unit dB dB

Diffraction‑related attenuation

dB

Grain‑scattering attenuation

dB

Third‑order grain‑scattering coefficient Zener drag pressure

La Metallurgia Italiana - September 2026

dB J

pagina 69


Memorie scientifiche - Laminazione REFERENCES [1] [2] [3] [4]

[5] [6]

[7] [8] [9] [10] [11] [12]

[13] [14] [15] [16] [17] [18]

[19]

[20] [21]

Davis, Joseph R. (2001). “High-Strength Low-Alloy Steels”. Alloying: Understanding the Basics. ASM International. p. 193. ISBN 9781615030637 Degarmo, E. Paul; Black, J T.; Kohser, Ronald A. (2003), Materials and Processes in Manufacturing (9th ed.), Wiley, ISBN 0-471-656534. Lenard, J.G.; Pietrzyk, M.; Cser, L. (1999). “Mathematical and Physical Simulation of the Properties of Hot Rolled Products”, ISBN 9780-08-042701-0 Malmström, M.; Jansson, A.; Hutchinson, B.; Lönnquist, J.; Gillgren, L.; Bäcke, L.; Sollander, H.; Bärwald, M.; Hochard, S.; Lundin, P. (2022). “Laser-Ultrasound-Based Grain Size Gauge for the Hot Strip Mill”, Appl. Sci. 2022, 12(19), 10048; https://doi.org/10.3390/ app121910048 Gorni, A. (2016). “Modelling the Microstructural Evolution During Hot Strip Rolling of Niobium Microalloyed Steels”, 4th International Conference on Thermomechanical Simulation - SimPro 2016, Ranchi (India), February 2016, 186-195. At: Ranchi, India Hoppe, D.; Haschke, T.; Sprock, A.; Hassel, Ch.; Hafer, J.; Bäcke, L.; Thorberg, J.-E.; Jonsson, Ch.; Malmström, M.; Kneisel, F.; Bärwald, M. (2024) “Insights from the online LUS grain size gauge and the potential for future process control”, 6th International Workshop on Laser-Ultrasound for Metals (LUS4Metals) at Swerim in Stockholm, Sweden May 22 to 23, 2024 Fernandez, A. I., Uranga, P., Lopez, B., Rodriguez, J. M. (2000), ISIJ International, Vol. 40 No. 9, pp. 893-901 Siciliano, F. (2000) “Mathematical modeling of the hot strip rolling of microalloyed Nb, multiply-alloyed Cr-Mo, and plain C-Mn steels”, Metallurgical and Materials Transactions A https://doi.org/10.1007/S11661-000-0287-8 VOLUME 31A, 511 Dutta, B.; Palmiere, E.J.; Sellars, C.M. (2001) “Modelling the kinetics of strain induced precipitation in Nb microalloyed steels” Volume 49, Issue 5, 14, Pages 785-794 Miettinen, Y. “Prediction of Austenite decomposition and heat release during cooling of low alloyed steels”, TKK V B107, 1995 Avrami, M. (1939). "Kinetics of Phase Change I". Journal of Chemical Physics, 7(12), 1103-1112. https://doi.org/10.1063/1.1750380 Johnson, W.A.; Mehl, R.F. (1939). "Reaction Kinetics in Processes of Nucleation and Growth". Transactions of the American Institute

of Mining and Metallurgical Engineers, 135, 416-458. Sellars, C.M.; Whiteman, J.A. (1979). "Recrystallization and grain growth in hot rolling". Metal Science, 13(3-4), 187-194. https://doi. org/10.1179/030634579790433484 Gladman, T. (1997). The Physical Metallurgy of Microalloyed Steels. The Institute of Materials, London. ISBN 978-1-86125-054-0. Zener, C. (cited in Smith, C.S.) (1948). "Grains, Phases, and Interfaces: An Interpretation of Microstructure". Transactions of the American Institute of Mining and Metallurgical Engineers, 175, 15-51. Christian, J.W. (2002). The Theory of Transformations in Metals and Alloys. Pergamon Press, Oxford. ISBN 978-0-08-044019-4. Bhadeshia, H.K.D.H.; Honeycomb, R. (2006). Steels: Microstructure and Properties. 3rd ed., Butterworth-Heinemann. ISBN 978-07506-8084-4. Malmström, M.; Bäcke, L.; Magnusson, H.; Lönnqvist, J.; Hutchinson, B. (2025). "Investigation of Dynamic Recrystallisation of Steel

During Hot Compression Using Gleeble and Laser-Ultrasonics (GLUS)", Metallurgical and Materials Transactions A, Vol. 56. https:// doi.org/10.1007/s11661-025-07834-0 Cousin, G.; Meilland, P.; Damoiselet, F.; Legrand, N.; Naumann, N.; Ayeb, A. (2023). "Online Grain Size Measurement by Laser Ultrasonics in a Hot Rolling Mill", Proceedings of the 13th European Conference on Non-Destructive Testing (ECNDT 2023), 3-7 July

2023, Lisbon, Portugal. https://doi.org/10.58286/28204 Mohrbacher, H. (2026). "Application of Microalloying for Controlling Recrystallization and Grain Growth During Downstream Steel Processing." Metallurgical and Materials Transactions A, Vol. 57, pp. 2548-2571. https://doi.org/10.1007/s11661-026-08172-5 Ma, Q.; Yin, S.; Shang, C.; Liu, Q.; Li, B.; Jia, S. (2025). “Recrystallization and Second-Phase Precipitation in Nb-V Microalloyed Steels: A Thermal Simulation Study." Materials, Vol. 18, No. 13, Article 3069. https://doi.org/10.3390/ma18133069

TORNA ALL'INDICE >

La Metallurgia Italiana - Settembre 2026

pagina 70


Atti e notizie - AIM news

41° Convegno Nazionale AIM: ottant’anni al servizio della metallurgia in Italia Dal 9 all’11 settembre 2026 si è tenuto presso il Dipartimento di Ingegneria Meccanica e Industriale dell’Università degli Studi di Brescia il 41° Convegno Nazionale AIM, ormai consolidato come evento capillare per il settore metallurgico in Italia. Questa edizione in particolare – tra le più partecipate degli ultimi anni con oltre 350 persone tra ricercatori, docenti, professionisti e rappresentanti del mondo industriale – ha avuto un sapore ancora più speciale coincidendo con l’ottantesimo anniversario dell’Associazione Italiana di Metallurgia. Nel corso delle tre giornate sono state presentate oltre 200 memorie scientifiche, dedicate a una decina di aree tematiche che hanno attraversato l’intera filiera della

metallurgia; ad aprire i lavori scientifici, dopo i saluti del Presidente AIM, del Rettore dell’Università di Brescia e delle autorità politiche e istituzionali, sono state due testimonianze che hanno intrecciato storia e futuro del settore: quella del dr. Stefano Karadjov, direttore del Museo di Santa Giulia, dedicata alla Vittoria Alata, e quella della dr.ssa Laura Tolettini di Feralpi Group, sullo sguardo di genere nella siderurgia del futuro. Queste le parole della prof.ssa Annalisa Pola, presidente del Convegno e curatrice dell’editoriale per questo numero de La Metallurgia Italiana: “Per noi accademici, far parte di AIM vuol dire avere diverse occasioni di incontro e di confronto, con la possibilità di coordinarci al meglio. […] Il Convegno Nazionale AIM è una di queste, in cui tutti i docenti, i giovani dottorandi e borsisti che fanno parte del settore metallurgico possono interfacciarsi non solo con altri ricercatori universitari, ma anche con esperti e studiosi provenienti dai centri di ricerca e dalle aziende. Questo Nazionale in particolare, il 41°, si svolge a Brescia, una città famosa come punto nevralgico per la metallurgia e la storia dei metalli”. Il Convegno ha proposto anche momenti di convivialità tra i partecipanti, tra cui l’aperitivo “MEETallurgy”, una visita guidata al Museo di Santa Giulia alla scoperta della Vittoria Alata e la cena sociale in Franciacorta. A contribuire inoltre alla buona riuscita dell’organizzazione, è stato molto importante il servizio catering della co-

La Metallurgia Italiana - September 2026

pagina 71


Atti e notizie - AIM news operativa sociale Alborea di Brescia, che impiega e offre formazione in ambito ristorazione a persone detenute e in condizione di fragilità; nelle parole del Presidente Angelo Maiolo: “Ogni servizio rappresenta per noi l’impegno a offrire qualità al cliente e, nello stesso tempo, a trasformare il lavoro in un’opportunità di crescita, autonomia e riscatto per le persone. Per questo essere riconosciuti per la qualità del nostro lavoro significa anche dare valore alla dimensione sociale che lo rende possibile”. Il 41° Convegno Nazionale ha offerto inoltre riconoscimento tangibile al mondo della ricerca scientifica: durante la cerimonia d’apertura sono stati conferiti il Premio Aldo Daccò 2026 a Gianluca Di Egidio dell’Università di Bologna e il Premio Felice De Carli 2026 ex aequo a Elena Messinese del Politecnico di Milano e a Mattia Franceschi del

durante la sessione conclusiva, i numerosi giovani che si

CENIM-CSIC. Visto inoltre l’alto numero di contributi da

sono distinti con la propria presentazione con un attesta-

parte di giovani (under 32) afferenti a Università e Centri di

to di merito, denominato “AIM Young Researcher Presen-

ricerca, il Comitato organizzativo ha deciso di premiare,

tation Award”.

La Metallurgia Italiana - Settembre 2026

pagina 72


THE INTERNATIONAL EVENT DEDICATED TO THE STEEL INDUSTRY 11-12-13 MAY 2027 | HALLS 8-12-16 FIERA MILANO RHO, MILAN ITALY

Scan here to secure your place at the 2027 edition

POWERED BY

www.madeinsteel.it

SPONSORED BY


Atti e notizie - AIM news

Eventi AIM / AIM events

www.aimnet.it

Giornata di Studio Le tecnologie laser per l’industria del futuro Cazzago San Martino (BS) c/o Castellini S.p.A. - 16 ottobre 2026 >> SCOPRI DI PIÙ

5th International Conference on INGOT CASTING, ROLLING & FORGING Bardolino, Verona - 13-15 October 2026 >> MORE INFO

Metallurgia per non metallurgisti (Webinar Zoom) – 20-21-27-28 ottobre, 3-4-10-11-17-18-24-25 novembre 2026 >> SCOPRI DI PIÙ

Casi pratici di failure analysis Milano, Centro Congressi Fast - 22 ottobre 2026 >> SCOPRI DI PIÙ

Giornata di Studio Materiali e tecnologie per i cicli combinati di nuova generazione Prestazioni, flessibilità e transizione ecologica Centro Congressi FAST, Milano - 18 novembre 2026 >> SCOPRI DI PIÙ

La Metallurgia Italiana - Settembre 2026

pagina 74


Atti e notizie - AIM news

Workshop High-Strength Steels. Properties, Applications and the Future Acciai alto resistenziali. Proprietà, applicazioni e futuro Vicenza c/o Università degli Studi di Padova - 19 November 2026

Per ulteriori informazioni rivolgersi alla Segreteria AIM, e-mail: info@aimnet.it, oppure visitare il sito internet www.aimnet.it

>> MORE INFO

Tinplated Steels and Metals Packaging & Recycling - IFTSR 2026 IFTSR - International Forum Bergamo - 3-4 December 2026 >> MORE INFO

ECHT 2027 & 32nd IFHTSE World Congress The Industry meeting point for the international heat treatment and materials science network Milano (Italia) - 14-16 April 2027 >> MORE INFO

GNC2027 - Giornate Nazionali sulla Corrosione e Protezione (Università di Messina - Messina, 14-16 giugno 2027 >> SCOPRI DI PIÙ

Acciai sotto la lente: fondamenti, trattamenti e caratterizzazione Bari c/o Politecnico di Bari - 17-18 giugno 2027 >> SCOPRI DI PIÙ

La Metallurgia Italiana - September 2026

pagina 75


Atti e notizie - AIM news

www.aimnet.it

ABRASION 2027 - Conference on Abrasion Wear Resistant Cast Iron And Forged Steel For Rolling and Pulverizing Mills Povo - Trento at University of Trento - 1-3 September 2027 >> MORE INFO

ECCC 2027 - 12th European Continuous Casting Conference Milano c/o NH Milano Congress Centre - 26-28 October 2027 >> MORE INFO

VISITA IL SITO AIM

La Metallurgia Italiana - Settembre 2026

SCARICA IL CALENDARIO EVENTI AIM

pagina 76


Normativa / Standards Norme UNSIDER pubblicate da UNI nei mesi di luglio e agosto 2026 Norme UNSIDER pubblicate da UNI nei mesi di luglio e agosto 2026 UNI EN ISO 14577-1:2026 Materiali metallici – Prova di penetrazione strumentata per la determinazione della durezza e altri parametri dei materiali – Parte 1: Metodo di prova UNI EN ISO 14577-2:2026 Materiali metallici – Prova di penetrazione strumentata per la determinazione della du-

strumentata per la determinazione della du-

ghness (ISO 12135:2021, including corrected

rezza e altri parametri dei materiali – Parte 1:

version 2022-08)

Metodo di prova EN 10365:2026 UNI EN 10365:2017

Hot rolled steel channels, I and H sections –

Profili a U di acciai laminati a caldo, travi I e H –

Dimensions and masses

Dimensioni e masse ISO 19901-2:2026 UNI EN ISO 14577-2:2015

Specific requirements for offshore structures

Materiali metallici – Prova di penetrazione

– Part 2: Seismic design

strumentata per la determinazione della durezza e altri parametri dei materiali – Parte 2:

ISO 19901-7:2026

Verifica e taratura delle macchine di prova

Specific requirements for offshore structures – Part 7: Stationkeeping systems for floating offshore structures and mobile offshore units

Norme UNSIDER pubblicate da CEN e ISO nei mesi di luglio e agosto 2026

Oil and gas industries including lower carbon

rezza e altri parametri dei materiali – Parte 2: Verifica e taratura delle macchine di prova UNI EN 10365:2026 Profili a U di acciai laminati a caldo, travi I e H – Dimensioni e masse UNI EN ISO 14577-3:2026 Materiali metallici – Prova di penetrazione strumentata per la determinazione della durezza e altri parametri dei materiali – Parte 3: Taratura dei blocchetti di riferimento

ISO 15589-1:2026 energy – Cathodic protection of pipeline tran-

EN ISO 19901-7:2026

sportation systems – Part 1: On-land pipelines

Specific requirements for offshore structures – Part 7: Stationkeeping systems for floating

EN ISO 19901-1:2026

offshore structures and mobile offshore units

Specific requirements for offshore structures

(ISO 19901-7:2026)

– Part 1: Metocean design and operating considerations (ISO 19901-1:2026)

EN ISO 19901-2:2026 Specific requirements for offshore structures

ISO 19905-4:2026

– Part 2: Seismic design (ISO 19901-2:2026)

Site-specific assessment of mobile offshore units – Part 4: Jack-up installation and removal

CEN/TR 18341:2026

at a site

District heating and cooling systems – Sup-

Norme UNSIDER ritirate con sostituzione da UNI nei mesi di luglio e agosto 2026

UNI EN ISO 14577-3:2015 Materiali metallici – Prova di penetrazione strumentata per la determinazione della durezza e altri parametri dei materiali – Parte 3: Taratura dei blocchetti di riferimento UNI EN ISO 14577-1:2015 Materiali metallici – Prova di penetrazione

La Metallurgia Italiana - September 2026

plementary information on usage of CEN/TC

ISO 19901-1:2026

107 documents

Specific requirements for offshore structures – Part 1: Metocean design and operating con-

EN ISO 14577-2:2026

siderations

Metallic materials – Instrumented indentation test for hardness and materials parameters –

ISO 17078-2:2026

Part 2: Verification and calibration of testing

Oil and gas industries including lower carbon

machines (ISO 14577-2:2026)

energy – Drilling, production and injection equipment – Part 2: Flow-control devices for

EN ISO 12135:2026

side-pocket mandrels

Metallic materials – Unified method of test for the determination of quasistatic fracture tou-

pagina 77


ISO 16630:2026

prEN 10025-3 rev

Metallic materials – Sheet and strip – Hole

Hot rolled products of structural steels – Part

expanding test

3: Technical delivery conditions for normalized/normalized rolled weldable fine grain structural steels

ISO 9441:2026 Steel – Determination of niobium content

–

4-(2-Pyridylazo)-resorcinol

(PAR)

spectrophotometric method ISO 6892-2:2026

ISO/DIS – agosto e settembre 2026 prEN – progetti di norma europei prEN 545

prEN 10025-2 rev

Ductile iron pipes, fittings, accessories and

Hot rolled products of structural steels – Part

their joints for water pipelines – Requirements

2: Technical delivery conditions for non-alloy

and test methods

structural steels

Metallic materials – Tensile testing – Part 2: Method of test at elevated temperature

Progetti UNSIDER in inchiesta prEN e

prEN 10048 prEN 10025-4 rev

Hot-rolled narrow steel strip – Tolerances on

Hot rolled products of structural steels – Part

dimensions and shape

4: Technical delivery conditions for thermo-

Progetti UNSIDER messi allo studio dal

CEN (Stage 10.99) – agosto e settembre

mechanical rolled weldable fine grain structu-

prEN 10111

ral steels

Continuously hot rolled low carbon steel

2026

sheet and strip for cold forming – Technical prEN 10025-5 rev

delivery conditions

prEN ISO 10416 rev

Hot rolled products of structural steels – Part

Oil and gas industries including lower carbon

5: Technical delivery conditions for structural

prEN ISO 16708

energy – Drilling fluids – Laboratory testing

steels with improved atmospheric corrosion

Oil and gas industries including lower carbon

resistance

energy – Pipeline transportation systems -

prEN 1560 rev

Reliability-based limit state methods (ISO/

Founding – Designation system for cast iron –

prEN ISO 21809-2 rev

Material symbols and material numbers

Oil and gas industries including lower carbon

DIS 16708:2026)

energy “External coatings for buried or sub-

prEN 489-1

prEN 1559-1 rev

merged pipelines used in pipeline transporta-

District heating pipes – Bonded single and

Founding – Technical conditions of delivery –

tion systems” – Part 2: Fusion-bonded epoxy

twin pipe systems for buried hot and coldwa-

Part 1: General

coatings” Single-layer and multilayer systems

ter networks – Part 1: Joint casing assemblies and thermal insulation for hot and cold-water

prEN 10025-6 rev

prEN ISO 6506-1 rev

networks with metal service pipes in accor-

Hot rolled products of structural steels – Part

Metallic materials “Brinell hardness test” –

dance with EN 13941-series.

6: Technical delivery conditions for flat pro-

Part 1: Test method

ducts of high yield strength structural steels in the quenched and tempered condition

prEN ISO 20482 prEN ISO 6506-2 rev

Metallic materials – Sheet and strip – Erichsen

Metallic materials “Brinell hardness test” –

cupping test (ISO/DIS 20482:2026)

prEN 10149 rev

Part 2: Verification and calibration of testing

Hot rolled flat products made of high yield

machines

strength steels for cold forming – Technical delivery conditions

La Metallurgia Italiana - Settembre 2026

EN ISO 683-3:2022/prA1 Heat-treatable steels, alloy steels and fre-

prEN ISO 6506-3 rev

e-cutting steels – Part 3: Case-hardening

Metallic materials “Brinell hardness test” –

steels – Amendment 1 (ISO 683-3:2022/DAM

Part 3: Calibration of reference blocks

1:2026)

pagina 78


Progetti UNSIDER al voto FprEN e ISO/

ISO/FDIS – progetti di norma internazio-

Technical delivery conditions

FprEN – progetti di norma europei

ISO/FDIS 25319

ISO/DIS – progetti di norma internazio-

FprEN 12680-4

(sponge Iron) and briquette – Iron (III) chlori-

Founding – Ultrasonic testing – Part 4: Phased

de titrimetric method

EN 10338:2025/prA1 Hot rolled and cold rolled non-coated products of multiphase steels for cold forming –

nali

FDIS – agosto e settembre 2026

nali

Determination of metallic Fe in reduced iron

array ultrasonic testing of steel castings ISO/DIS 25640.2

ISO/FDIS 12490

Respiratory equipment – Breathing appara-

FprEN 1564

Oil and gas industries including lower carbon

tus performance requirements for diving and

Founding – Ausferritic spheroidal graphite

energy – Mechanical integrity and sizing of

hyperbaric applications

cast irons

actuators and mounting kits for pipeline val-

ISO/DIS 25374

FprEN 12681-1

Oil and gas industries including lower carbon

Founding – Radiographic testing – Part 1: Film

ISO/FDIS 1099

energy – Steam injection for thermal recovery

techniques

Metallic materials – Fatigue testing – Axial for-

ves

– Determination of two-phase relative permeability in heavy oil reservoirs ISO/DIS 25373

ce-controlled method FprEN 12681-2 Founding – Radiographic testing – Part 2: Te-

ISO/FDIS 25408-2

chniques with digital detectors

Testing method for bead wire – Part 2: Adhe-

Oil and gas industries including lower carbon

sion test

energy – Multi-component for thermal reco-

EN 1092-1:2018/FprA1

very – Design of thermal fluid huff and puff

Flanges and their joints – Circular flanges for

workflow

pipes, valves, fittings and accessories, PN designated – Part 1: Steel flanges

ISO/DIS 20482 Metallic materials – Sheet and strip – Erichsen

FprEN 18213-1

cupping test

District heating and district cooling pipes – Bonded single and twin pipe systems for

ISO/DIS 16708

directly buried hot and cold water networks –

Oil and gas industries including lower carbon

Part 1: Qualification testing of fitter

energy – Pipeline transportation systems – Reliability-based limit state methods

FprEN 18213-2 District heating and district cooling pipes

ISO 683-3:2022/DAmd 1

– Bonded single and twin pipe systems for

Heat-treatable steels, alloy steels and fre-

directly buried hot and cold water networks –

e-cutting steels – Part 3: Case-hardening ste-

Part 2: Qualification testing of PE-Welder

els – Amendment 1

La Metallurgia Italiana - September 2026

pagina 79


12th european conference on continuous casting

e c

12

c c

save the date

Milan - Italy 26-28 October 2027

The 12th European Continuous Casting Conference (ECCC 2027) will be held in Milan, Italy, from 26 to 28 October 2027, bringing together the leading minds and innovators shaping the future of continuous casting. Organized by AIM, the Italian Association for Metallurgy, ECCC 2027 is recognized as one of the most important international forums for plant operators, technology providers, researchers, academics, and service companies active in the continuous casting sector. Join us in Milan and be part of the conversations, innovations, and partnerships that will drive the next generation of continuous casting excellence.

Organised by


Turn static files into dynamic content formats.

Create a flipbook
La Metallurgia Italiana, n.9 settembre 2026 by aimnet3 - Issuu