PLENARY LECTURES

With great pleasure we would like to introduce fantastic presenters of plenary lectures, which will take place during the 29th KomPlasTech Conference.

• Data-Driven Deep Learning Evolutionary Algorithm EvoDN2: Development and Applications
  Nirupam Chakraborti
  Faculty of Mechanical Engineering, Czech Technical University in Prague
  
  Professor Nirupam Chakraborti was educated in India and in the USA, receiving his B.Met.E from Jadavpur University followed by an MS from New Mexico Tech, USA and subsequently PhC and PhD degrees from University of Washington, Seattle, USA. He lectured and conducted research in numerous universities worldwide. He is a former Docent of Åbo Akademi, Finland, former long term Visiting Professors of Florida International University, Miami, USA and POSTECH, Korea, and he also taught and conducted research at several other academic institutions in Austria, Brazil, Finland, Germany, Italy and the US. He was a Higher Academic Grade Professor at Indian Institute of Technology, Kharagpur and since 2022, after superannuating in India, he is working as a Visiting Professor of Czech Technical University in Prague. He is a Co-Editor of the prestigious journal Philosophical Magazine Letters published by Taylor and Francis. Internationally known for his pioneering work on evolutionary computation in the area of Metallurgy and Materials, Professor Chakraborti has been continuously rated among the top 2% highly cited world researchers in the Materials area, as per the reports published from Stanford University in the USA. Beside numerous journal articles, he is also author of a comprehensive book Data-driven Evolutionary Algorithms in Materials Technology published by CRC Press, USA/UK. His Plenary Lecture at KomPlasTech 2025 will involve a detailed description of the Evolutionary deep learning and optimization algorithm EvoDN2 developed in his group, along with several examples of its application in the material design and manufacturing domain as currently pursued by him and his students.
  More information can be found at the following web site
  
  

• Modeling and Numerical Simulation of the Gas Quenching Process of Pyrowear 53 Steel Gears
  Bartosz Iżowski
  Rzeszów University of Technology, Poland
  
  Bartosz Iżowski, PhD, is a Senior Materials and Stress Engineer in the Research and Development Department of one of the leading aerospace companies and an independent entrepreneur. With a career spanning over a decade, he specializes in leveraging advanced numerical methods to optimize manufacturing processes such as heat treatment, forging, and machining. Dr. Iżowski holds a B.Sc. and M.Sc. in Materials Science and Engineering from Warsaw University of Technology and a postgraduate diploma in Engineering of Casting and Metal Forming Processes from the Silesian University of Technology. He earned his PhD from the Faculty of Mechanical Engineering and Aeronautics at Rzeszów University of Technology, where his doctoral research focused on modeling quenching distortion during high-pressure gas quenching of gears made of Pyrowear 53 steel, designed for aerospace transmission applications. In his professional role, Dr. Iżowski has led projects addressing critical industry challenges, such as enhancing the durability of aerospace components, reducing material waste in forging operations, and optimizing heat treatment cycles for high-performance steels. He has successfully developed customized material and process models, implemented predictive tools for distortion and residual stress control, and contributed to the adoption of novel alloys and heat treatment methods for aerospace applications. Dr. Iżowski is a trusted consultant for companies seeking to improve operational workflows through numerical analysis, simulation-driven process optimizations, and tailored engineering solutions. His efforts have been instrumental in improving production efficiency, material performance, and product quality across various manufacturing sectors.
  More information can be found at the following web site
  
  

• Forays in Functional High-Entropy Ceramic Materials
  Cristian Ciobanu
  Department of Mechanical Engineering and in the Materials Science Program at Colorado School of Mines, US
  
  Cristian V. Ciobanu is a Professor in the Department of Mechanical Engineering and in the Materials Science Program at Colorado School of Mines. Prior to joining the School of Mines in 2004, he was a postdoctoral fellow in the Division of Engineering at Brown University (2001-2004). He holds degrees in Physics from University of Bucharest (B. Sc., 1995) and Ohio State University (M.S., 1998 and Ph.D., 2001). His research interests are in computational materials science and mechanics, specifically in structure-property relationships, nanoscale/nanomaterials problems, two-dimensional materials, materials for renewable energy applications, developments of evolutionary algorithms for computational materials design and optimization of atomic structures, self-organized nano and bio structures on crystal surfaces, among others. His research work in these areas has led to over 100 journal articles, two patents, and a book. Prof. Ciobanu is the current Chair of the Rocky Mountain Chapter of the American Vacuum Society, Fellow of the Royal Society of Chemistry, Fellow of the Institute of Physics; he is also a lifetime member of the American Physical Society (APS) and The Mineral, Metals and Materials Society (TMS). In addition to his teaching and research, Dr. Ciobanu carries out significant professional service in various editorial capacities for several journals (Materials Letters, Chinese Journal of Physics, Metals, Philosophical Magazine Letters, Materials Research Express), as well as technical paper and proposal reviewer and session chair/organizer for certain professional conferences.
  
  More information can be found at the following web site
  
  

• Numerical predictions of fracture in amorphous materials: Current achievements and challenges
  Sebastian Pfaller
  Institute of Applied Mechanics, Friedrich-Alexander-Universität Erlangen-Nürnberg, Germany
  
  Sebastian Pfaller is a university lecturer at Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU). He achieved is doctorate (degree Dr.-Ing.) in 2015 with the doctoral thesis entitled "Multiscale Simulation of Polymers". He habilitated on "Discrete and Continuous Methods for Modelling and Simulation of Polymeric Materials" in 2021 and is head of the Capriccio group, which he established at the Institute of Applied Mechanics at FAU. His research interests comprise interdisciplinary and scale bridging simulations and their applications. This includes particle-based descriptions at atomistic and molecular resolution, material characterisation and development of continuum mechanical constitutive laws, which forms the basis for multiscale description of deformation and failure processes in various kinds of amorphous materials like polymers and inorganic glasses as well as the understanding of interphases and interfaces in composites. He has acquired various third-party funding, among others for projects in the interdisciplinary research training group “Fracture across Scales” and in the collaborative international research project “BIO ART” on bio-sourced polymers. Following his strong interdisciplinary interests, he has collaborations with national and international experts from chemistry, physics, mathematics, materials science, and engineering science, is (co-)organiser of various (mini)symposia at international conferences, and is active as reviewer for international funding agencies as well as scientific journals. Sebastian Pfaller will present current research results on fracture simulations of amorphous materials like inorganic glasses and polymers. In this context, he will introduce the Capriccio method as a concurrent technique to couple a continuum to particle-based regions, which has been designed for amorphous thermoplastics and is also applied to thermosetting polymers and silica glasses. He will discuss current achievements with regard to its capabilities for quantitative predictions of fracture mechanical quantities as well as challenges in view of length and time scales involved in such simulations.
  
  More information can be found at the following web site
  
  

• Recent Advances in Robust Variational Physics-Informed Neural Networks
  Sergio Rojas Hernandez
  School of Mathematics, Monash University
  
  Dr. Sergio Rojas is a Senior Lecturer in the Applied and Computational Mathematics section of the School of Mathematics at Monash University, Australia. His research focuses on numerical analysis, scientific computing, and mathematical modeling, with emphasis on developing and analyzing residual minimization-based numerical methods for solving complex Partial Differential Equations. These methodologies integrate state-of-the-art approaches such as Finite Element, Discontinuous Galerkin, Hybridizable Discontinuous Galerkin, Minimum-Residual, and Variational Physics-Informed Neural Networks methods. Before joining Monash, Dr. Rojas was a Lecturer at Pontificia Universidad Católica de Valparaíso (PUCV), Chile, and a Research Associate at Curtin University, Australia. He earned his PhD and MSc in Engineering Sciences from Pontificia Universidad Católica de Chile, where he specialized in analytical and semi-analytical methods for electromagnetic problems. Additionally, he holds a Master's in Mathematics from the University of Pavia, Italy, and a Bachelor's in Mathematics from PUCV.
  In early 2024, we introduced RVPINNs, a robust extension of the Variational Physics-Informed Neural Networks (VPINNs) method. In RVPINNs, the loss functional is formulated using a Petrov-Galerkin-type variational approach, where the trial space consists of a (Deep) Neural Network and the test space is a finite-dimensional vector space. Unlike standard VPINNs, RVPINNs minimise a loss based on the discrete dual norm of the residual, providing a reliable estimator of the approximation error in the energy norm under the assumption of a local Fortin operator. This ensures more accurate approximations of the partial differential equations governing experimental data. However, a key challenge of RVPINNs is the need to invert a Gram matrix at each Neural Network nonlinear solver step, making the method computationally expensive if the variational formulation and discrete test space are not carefully selected.
  In this talk, we will present recent advances in RVPINNs, focusing on adaptive strategies for test space selection and variational formulations that enable the construction of block-diagonal Grammatrices, significantly accelerating training while maintaining accuracy. We will also demonstrate the effectiveness and robustness of our approach through tailored numerical experiments, confirming theoretical error estimates and highlighting substantial improvements in computational efficiency.
  More information can be found at the following web site