shutterstock_105896591.jpg
HRZZ_logo.png

Razvoj modela za procjenu ponašanja materijala temeljenih na    strojnom učenju / Development of machine learning-based models for materials behavior estimation
(MADEIRA)

 
Financiranje: Hrvatska zaklada za znanost - HRZZ-IP-2020-02-5764
Voditelj projekta: Prof. dr. sc. Robert Basan
Matična ustanova: Sveučilište u Rijeci, Tehnički fakultet, Rijeka, Hrvatska
Trajanje projekta: 2020-2024
Financiranje: Hrvatska Zaklada za znanost (HRZZ-IP-2020-02-5764)

Kontakt
Prof. dr. sc. Robert Basan
Sveučilište u Rijeci, Tehnički fakultet
Zavod za konstruiranje u strojarstvu
Vukovarska 58, 51000 Rijeka, Croatia
e. robert.basan@riteh.hr
t. +385 51 651 530
 

Sažetak projekta

Modeliranje ponašanja materijala ključan je dio suvremenog razvoja proizvoda te računalnih simulacija koje se provode u cilju čim boljeg iskorištenja materijala i smanjenja mase i troškova. Za uspješno modeliranje ponašanja materijala nužno je poznavanje njegovih značajki i parametara ponašanja. Njihovo eksperimentalno određivanje je najtočnije, ali je dugotrajno i skupo te se stoga nastoji izvoditi u najmanjoj mogućoj mjeri. Projektom MADEIRA cilja se na rješavanje nedostatka podataka o ponašanju materijala potrebnih za numeričke proračune i simulacije koji otežava šire prihvaćanje modeliranja materijala te uključivanje naprednih materijalnih modela u suvremeni proces razvoja proizvoda. Predloženo istraživanje usmjereno je razvoju modela za procjenu ponašanja i parametara materijala temeljenih na strojnom učenju kao rješenja za navedene probleme. Na osnovi prikupljenih rezultata istraživanja materijala i rezultata provedenih eksperimenata, sustavno će se istražiti, identificirati i mapirati kompleksni odnosi između svojstava/značajki materijala i njegovog mehaničkog ponašanja i pripadnih parametara na različitim razinama. Unaprijedit će se postojeći i razviti novi modeli procjene naprednih monotonih, cikličkih i zamornih parametara materijala te matematički modeli termalnih postupaka prerade metala i modeliranja njihovih mehaničkih svojstava nakon toplinske obrade. Istražit će se primjenjivost raznih metoda strojnog učenja te mogućnosti procjene parametara i naprednih konstitutivnih materijalnih modela koji se zbog kompleksnosti i velikog broja potrebnih parametara otežano i rijetko primjenjuju u praksi. Razvijene baze podataka i prediktivni modeli trebali bi doprinijeti široj primjeni modeliranja ponašanja materijala i računalnih simulacija, a uspostavljen metodološki okvir i principi primjene strojnog učenja mogli bi biti korisni i primjenjivi za razvoj prediktivnih modela i u drugim područjima primjene i istraživanja materijala.

 

Projektni tim

Sveučilište u Rijeci, Tehnički fakultet, Rijeka, Hrvatska
Prof. dr. sc. Robert Basan
Izv. prof. dr. sc. Dario Iljkić
Izv. prof. dr. sc. Tea Marohnić
Izv. prof. dr. sc. Sunčana Smokvina Hanza
Asist. Ela Marković

Univerza v Ljubljani, Fakulteta za strojništvo, Ljubljana, Slovenija
Doc. dr. Andrej Žerovnik

Principal investigator: Prof. D. Sc. Robert Basan
Host institution: University of Rijeka, Faculty of Engineering, Rijeka, Croatia
Project duration: 2020-2024
Financed by: Croatian Science Foundation (HRZZ-IP-2020-02-5764)

Contact
Prof. D. Sc. Robert Basan
University in Rijeka, Faculty of Engineering
Department of Mechanical engineering Design
Vukovarska 58, 51000 Rijeka, Croatia
e. robert.basan@riteh.hr
t. +385 51 651 530

 

Project sumary

Materials modeling is a key part of modern product development and computer simulations which are performed in order to improve material utilisation and reduce weight and costs. The main prerequisite for successful materials modeling is knowledge of its behavior and properties. Their experimental determination is the most accurate, but it is long-lasting and expensive, and there is a trend to perform as few experiments as possible. The MADEIRA project addresses the lack of material data and behavior parameters needed for numerical simulations which is one of the major obstacles to wider use of advanced material models in the industry. Proposed research aims at the development of advanced machine learning-based models for estimation of material behavior and related parameters as a solution for mentioned problems. Based on results of materials research and data collected from published sources and results of own experiments, complex relationships between properties and behavior of the materials will be systematically analyzed, identified and mapped at different levels. Existing models for estimation of advanced monotonic, cyclic and fatigue parameters will be improved and new ones developed. Mathematical models of thermal processes of metals and prediction of their mechanical properties after heat treatment will be further improved. The applicability of different machine learning-based methods for estimation of parameters of advanced constitutive material models will be investigated which, due to the large number of required material parameters, are rarely applied in industrial practice. Developed database and predictive models should contribute to the wider application of material behavior modeling and computer simulations, and the established methodological framework and principles of machine learning may be useful and applicable for the development of predictive models in other fields of material research and application.

 

Project team

University in Rijeka, Faculty of Engineering, Rijeka, Croatia
Prof. D. Sc. Robert Basan
Assoc. Prof. D. Sc. Dario Iljkić
Assoc. Prof. D. Sc. Tea Marohnić
Assoc. Prof. D. Sc. Sunčana Smokvina Hanza
Assist. Ela Marković

University in Ljubljana, Faculty of Mechanical Engineering, Ljubljana, Slovenia
Assist. Prof. D. Sc. Andrej Žerovnik

Publikacije / Publications

Radovi u časopisima / Journal Papers

  • Basan, Robert ; Marohnić, Tea: A comprehensive evaluation of conventional methods for estimation of fatigue parameters of steels from their monotonic properties // International journal of fatigue, 183 (2024), 1; 108244, 19. (Q1 ; IF 6,0)
  • Marković, Ela; Basan, Robert; Srnec Novak, Jelena; Žerovnik, Andrej: Finite element analysis of unnotched and notched functionally graded steel specimens // Procedia structural integrity, 54 (2024), 156-163. (Q4)
  • Ivaničić, Daniel ; Marohnić, Tea ; Basan, Robert: Application of Digital Image Correlation method for verification of topology optimization of 3D printed load-bearing element // Procedia structural integrity, 51 (2023), 199-205. (Q4)
  • Marohnić, Tea; Basan, Robert; Marković, Ela: Estimation of Cyclic Stress–Strain Curves of Steels Based on Monotonic Properties Using Artificial Neural Networks // Materials, 16 (2023), 14; 1-18. (Q2 ; IF 3,4)
  • Basan, Robert; Marohnić, Tea; Božić, Željko: Application of material fatigue parameters estimation in analysis of rolling-sliding contact fatigue of gears // Procedia Structural Integrity, 46 (2023), 62-67. (Q4)
  • Basan, Robert; Marohnić, Tea; Marković, Ela: Evaluation of fatigue parameters estimation methods with regard to specific ranges of fatigue lives and relevant monotonic properties // Procedia Structural Integrity, 42 (2022), 655-662. (Q4)
  • Smoljan, Božo; Iljkić, Dario; Smokvina Hanza, Sunčana; Hajdek, Krunoslav: Mathematical Modelling of Isothermal Decomposition of Austenite in Steel // Metals, 11 (2021), 8; 1292, 14. (Q1 ; IF 2,351)
  • Smokvina Hanza, Sunčana; Marohnić, Tea; Iljkić, Dario; Basan, Robert: Artificial Neural Networks-Based Prediction of Hardness of Low-Alloy Steels Using Specific Jominy Distance // Metals, 11 (2021), 5; 714, 14. (Q1 ; IF 2,351)

Radovi na konferencijama / Conference Papers

  • Iljkić, Dario; Smokvina Hanza, Sunčana; Smoljan, Božo; Kvrgić, Dario: Mathematical modelling of tempered hardness of quenched steel // MTECH 2023 - 4. međunarodna konferencija o materijalima : toplinska obrada i inženjerstvo površina, materijali i tribologija, korozija i ispitivanje materijala. Zagreb: Hrvatsko društvo za toplinsku obradu i inženjerstvo površina, 2023. str. 114-121
  • Marković, Ela; Basan, Robert; Srnec Novak, Jelena; Žerovnik, Andrej: Development of a finite element model of functionally graded material steel specimens // Proceedings of the 44th International Conference on Materials Mechanics. Darmstadt, Njemačka: Institut für Stahlbau und Werkstoffmechanik der Technischen Universität Darmstadt, 2023. str. 23-30
  • Iljkić, Dario; Smokvina Hanza, Sunčana; Smoljan, Božo: Computer simulation of quenching and tempering of steel // Proceedings of the 44th International Conference on Materials Mechanics. Darmstadt, Njemačka: Institut für Stahlbau und Werkstoffmechanik der Technischen Universität Darmstadt, 2023. str. 41-54
  • Basan, Robert; Marohnić, Tea; Marković, Ela: An overview of design-relevant nonferrous alloys and estimation of their fatigue behavior // Proceedings of the 44th International Conference on Materials Mechanics. Darmstadt, Njemačka: Institut für Stahlbau und Werkstoffmechanik der Technischen Universität Darmstadt, 2023. str. 63-68
  • Iljkić, Dario; Smokvina Hanza, Sunčana; Smoljan, Božo; Kvrgić, Dario: Prediction of quenched and tempered steel hardness using carbon equivalent (CE) // 32nd International Conference on Metallurgy and Materials / Brno, Czech Republic, 2023.
  • Marohnić, Tea; Basan, Robert: Application of ANNs for Estimation of Material Behavior – Learning From Small Datasets // Proceedings of the 43rd International Conference on Materials Mechanics / Vormwald, Michael (ur.). Sani, Greece: Technische Universitat Darmstadt - Institut fur Stahlbau und Werkstoffmechanik, 2022. str. 109-114
  • Basan, Robert; Marohnić, Tea: Estimation of fatigue behavior of high strength steels // Proceedings of the 43rd International Conference on Materials Mechanics / Vormwald, Michael (ur.). Sani, Greece: Technische Universitat Darmstadt - Institut fur Stahlbau und Werkstoffmechanik, 2022. str. 33-37
  • Iljkić, Dario; Smokvina Hanza, Sunčana; Smoljan, Božo; Pomenić, Loreta; Štic, Lovro; Liverić, Lovro: Load capacity and corrosion behavior of quenched and tempered steel 42CrMo4 and cast steel GS- 42CrMo4 // 19th International Foundrymen Conference - Humans - Valuable Resource for Foundry Industry Development Split, Hrvatska, 2021.

Sažeci i prezentacije na konferencijama / Conference Abstracts and Presentations

  • Marković, Ela; Basan, Robert; Srnec Novak, Jelena; Žerovnik, Andrej: Finite element analysis of unnotched and notched functionally graded steel specimens // Book of abstracts of the 5th International Conference on Structural Integrity. Portugal: The International Conference on Structural Integrity, 2023. str. 60-60
  • Marohnić, Tea; Marković, Ela; Basan, Robert: Study of Relevance of Alloying Elements in Estimating Cyclic Yield Stress and Ramberg-Osgood Parameters of Steels // Book of abstracts of the 5th International Conference on Structural Integrity. Portugal: The International Conference on Structural Integrity, 2023. str. 76-76
  • Marohnić, Tea; Basan, Robert; Marković, Ela; Smokvina Hanza, Sunčana: Statistical analysis of parameters and behavior of quenched and tempered steels // e-book of Abstracts of the 7th International Conference on Engineering Against Failure ICEAF VII. Grčka: The International Conference of Engineering Against Failure., 2023. str. 19-19
  • Basan, Robert; Marohnić, Tea; Marković, Ela; Iljkić, Dario: A study of methods for fatigue parameters and behavior estimation of quenched and tempered steels // e-book of Abstracts of the 7th International Conference on Engineering Against Failure ICEAF VII. Grčka: The International Conference of Engineering Against Failure., 2023. str. 3-3
  • Basan, Robert; Marohnić, Tea; Marković, Ela: An overview of design-relevant nonferrous alloys and estimation of their fatigue behavior // 44th International Conference on Materials Mechanics / Vormwald, Michael (ur.). Cres, Croatia: Technische Universitat Darmstadt - Institut fur Stahlbau und Werkstoffmechanik, 2023.
  • Iljkić, Dario; Smokvina Hanza, Sunčana; Smoljan, Božo: Computer simulation of quenching and tempering of steel // 44th International Conference on Materials Mechanics / Vormwald, Michael (ur.). Cres, Croatia: Technische Universitat Darmstadt - Institut fur Stahlbau und Werkstoffmechanik, 2023.
  • Srnec Novak, Jelena; De Bona, Francesco; Benasciutti, Denis: On the modeling of cyclic plasticity of metals with a new isotropic model // 44th International Conference on Materials Mechanics / Vormwald, Michael (ur.). Cres, Croatia: Technische Universitat Darmstadt - Institut fur Stahlbau und Werkstoffmechanik, 2023.
  • Marković, Ela; Basan, Robert; Srnec Novak, Jelena; Žerovnik, Andrej: Development of a finite element model of functionally graded material steel specimens // 44th International Conference on Materials Mechanics / Vormwald, Michael (ur.). Cres, Croatia: Technische Universitat Darmstadt - Institut fur Stahlbau und Werkstoffmechanik, 2023.
  • Marković, Ela; Marohnić, Tea; Basan, Robert: Application of machine learning models for estimation of material properties // International Doctoral Conference "Digitalisation in science and society" Book of Abstracts / Ljubljana, Slovenija, 2022. str. 13-13
  • Bjelobradić, Sanja; Basan, Robert; Marković, Ela; Marohnić, Tea DIC-based experimental analysis of plate specimens with central holes loaded in tension // 6th Annual PhD Conference on Engineering and Technology „My first conference 2022“ Book of abstracts / Dugonjić Jovančević, Sanja ; Sulovsky, Tea ; Tadić, Andrea (ur.). Rijeka, 2022. str. 31-31
  • Marković, Ela; Marohnić, Tea; Basan, Robert Application of feature selection techniques in assessing variables relevant for estimation of materials parameters and behavior // 6th Annual PhD Conference on Engineering and Technology „My first conference 2022“ Book of abstracts / Dugonjić Jovančević, Sanja ; Sulovsky, Tea ; Tadić, Andrea (ur.). Rijeka, 2022. str. 38-38
  • Basan, Robert; Marohnić, Tea; Marković, Ela: Evaluation of fatigue parameters estimation methods with regard to specific ranges of fatigue lives and relevant monotonic properties // ECF23-Book of Abstracts / Moreira, Pedro ; Reis, Luis (ur.). Funchal, Portugal: ESIS Publishing House (ESIS-PH), 2022. str. 224-224
  • Marohnić, Tea; Basan, Robert: Estimation of materials' parameters of strain-life fatigue behavior using empirical and artificial neural networks based approach // 26th International Conference on Fracture and Structural Integrity Torino, Italija, 2021.
  • Basan, Robert; Marohnić, Tea: On the applicability of cyclic and fatigue parameters estimation methods to high strength steels // 6th International Virtual Conference of Engineering Against Failure, E-book of Abstracts / Pantelakis, Spiros ; Lampeas, George ; Tserpes, Konstantinos (ur.). Patras, Grčka: University of Patras, 2021. str. 203-203
  • Basan, Robert; Marohnić, Tea; Božić, Željko: Application of material fatigue parameters estimation in analysis of rolling-sliding contact fatigue of gears // 5th International Conference on Structural Integrity and Durability, ICSID 2021, Book of Abstracts / Božić, Željko (ur.). Zagreb: University of Zagreb Faculty of Mechanical Engineering and Naval Architecture, 2021. str. 56
  • Ivaničić, Daniel; Marohnić, Tea; Basan, Robert: Application of digital image correlation method for verification of topology optimization of 3D printed load-bearing element // 5th International Conference on Structural Integrity and Durability, ICSID 2021, Book of Abstracts / Božić, Željko (ur.). Zagreb: University of Zagreb Faculty of Mechanical Engineering and Naval Architecture, 2021. str. 57

Poster projekta / Project Poster

MADEIRA-poster.PNG