An analytical predictor machine learning corrector scheme for modeling lateral flow in hot strip rolling

Aug 4, 2026

Rolling is a metal forming process that converts heated slabs into strips with targeted dimensions and mechanical properties, where hot rolling induces plastic deformation and longitudinal elongation. Due to plastic incompressibility, transverse expansion occurs as spread or lateral flow, whose accurate prediction is essential for sustainability and product quality. Conventional finite element models are accurate but slow, while analytical models are fast but insufficiently accurate for optimal control. A hybrid analytical predictor–machine learning corrector framework is developed to rapidly and accurately predict spread by coupling analytical initial predictions with data-driven corrections trained on high-fidelity FE results, while preserving computational efficiency.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages10

Reading Time25 min

AccessFree to download

Introduction

Motivation for hybrid physics–machine learning approaches

Predictor–corrector concept for spread modeling

Materials Research Proceedings 54 (2025) 2002-2011 [https://doi.org/10.21741/9781644903599-215](https://doi.org/10.21741/9781644903599-215) An analytical predictor machine learning corrector scheme for modeling lateral flow in hot strip rolling Amirali HASHEMZADEH 1,a* , Frederic E. BOCK2, b , Camile HOL3,c , Koen SCHUTTE3,d , Antonella COMETA 1,e , Celal SOYARSLAN 1,4,f, Benjamin KLUSEMANN2,5,g, Ton VAN DEN BOOGAARD 1, h 1Chair of Nonlinear Solid Mechanics, University of Twente, The Netherlands 2 Institute of Materials and Process Design, Helmholtz-Zentrum Hereon, Geesthacht, Germany 3Tata Steel, Research and Development, IJmuiden, The Netherlands 4 Fraunhofer Innovation Platform, University of Twente, The Netherlands 5 Institute for Production Technology and Systems, Leuphana University Lüneburg, Lüneburg, Germany [a](aa.hashemzadeh@utwente.nl)[a.hashemzadeh@utwente.nl](aa.hashemzadeh@utwente.nl), bfrederic. bock@hereon.de, [c](ccamile.hol@tatasteeleurope.com)[camile.hol@tatasteeleurope.com](ccamile.hol@tatasteeleurope.com), [d](dKoen.Schutte@tatasteeleurope.com)[Koen.Schutte@tatasteeleurope.com](dKoen.Schutte@tatasteeleurope.com) , [e](ea.cometa@utwente.nl)[a.cometa@utwente.nl](ea.cometa@utwente.nl), [f](fc.soyarslan@utwente.nl)[c.soyarslan@utwente.nl](fc.soyarslan@utwente.nl), [g](gbenjamin.klusemann@leuphana.de)[benjamin.klusemann@leuphana.de](gbenjamin.klusemann@leuphana.de), [h](ha.h.vandenboogaard@utwente.nl)[a.h.vandenboogaard@utwente.nl](ha.h.vandenboogaard@utwente.nl) Keywords: Hot rolling, Finite Element Model, Predictor-corrector Modeling, Machine Learning Abstract. Rolling is a metal forming process where slabs are passed through rollers to produce strips with specific dimensions and mechanical properties. This process is performed in hot or cold formats. In hot rolling, the workpiece is initially heated above its recrystallization temperature. During the hot rolling process, plastic deformation occurs as the material’s thickness decreases and elongation takes place along the longitudinal axis of the workpiece. Due to the incompressibility of plastic deformation, the material also expands in the transverse direction, a phenomenon known as spread or lateral flow. Modeling spread is crucial for sustainability considerations and meeting customer expectations regarding the quality of the final product. Current prediction methodologies, such as the accurate but slow Finite Element (FE) method orthe fast but inaccurate analytical metal forming analysis, are impractical for optimal control. To tackle these challenges, hybrid frameworks have emerged as a promising alternative. The present work aims to develop a fast and accurate model for predicting spread in hot rolling. Specifically, machine learning improves analytical models by leveraging data from a high-fidelity FE model. Initially, we review analytical models for spread, which address key aspects of the problem’s physics. To generate the ground truth (GT) space, an automated FE model for hot strip rolling is created. Moreover, the model’s sensitivity to both process and material parameters is investigated. In the Analytical Predictor Machine Learning Corrector scheme, the analytical models generate initial predictions of GT. In the correction step, a data-driven machine learning model is used torefine these predictions by compensating for deviations from high-fidelity FE simulations. The proposed hybrid framework improves the accuracy of the existing analytical models while preserving their computational efficiency. Introduction The use of machine learning in materials mechanics and processing can be a key enabler to accelerate the development of novel materials designs, techniques, and systems. Along the socalled process-structure-property-performance chain, a vast number of advances has been achieved, as summarized in [1] . While data-driven techniques play a vital role in advancing Content from this work may be used under the terms of the

FAQs

Why is modeling lateral flow (spread) important in hot strip rolling?

Spread prediction is crucial to meet sustainability goals and customer requirements by ensuring the quality of the final strip product.

What limitation exists in current prediction methods for hot rolling?

Finite element methods are accurate but computationally slow, while fast analytical models tend to be inaccurate for effective optimal control.

How does the proposed analytical predictor machine learning corrector scheme work?

Analytical models provide initial predictions of ground truth spread, and a machine learning correction refines them by compensating for deviations from high-fidelity FE simulations.

Aug 4, 2026

Upload by Theodora