PhD defence of Chengdan Xue

8 March 2022

Chengdan Xue defends his PhD in Computational Mechanics and Materials on March 15th, 22

Modeling of grain structure and hot cracking for arc welding processes

Chengdan Xue conducted his PhD work in 2MS team, under the supervision of Charles-André Gandin, Michel Bellet and Gildas Guillemot. Chengdan defends his PhD in Computational Mechanics and Materials on March 15th, 2022 in front of the following jury:

– Prof. Dominique Daloz, Université de Lorraine, Institut Jean Lamour

– MC Iryna Tomashchuk, Université de Bourgogne Franche-Comté, IUT Le Creusot

– Prof. Aude Simar, Université Catholique de Louvain

– Prof. Cyril Bordreuil, Université de Montpellier, LMGC

– IR. Pierre-Emile Lhuillier, EDF R&D

Abstract:

Welding is a permanent assembly process aimed at ensuring the continuity of the material. However, during the solidification stage, several types of defects such as hot cracking develop, leading to a decrease of the weld quality and part performances. Virtual grain microstructure would provide valuable information to enhance literature criterion on defects occurrence and to provide better understanding on their development.

In this work, the 3D Cellular Automaton – Finite Element (CAFE) method is applied to simulate the grain structure formation during Gas Tungsten Arc Welding (GTAW) process with or without added material for metal sheet or chamfer configurations. All simulations are dedicated to austenitic stainless steel welding. The simulated grain structure is compared with experimental observations developed in the present partnership of ANR project NEMESIS. New heat source model is also proposed and dedicated to welding in chamfer configuration. A good coherence is found regarding the grain texture (EBSD map and pole figures) once the corresponding thermal conditions (e.g. melt pool shape) are obtained.

Besides, a strain based hot cracking criterion is computed considering a GTAW process developed upon a metal sheet. Both cracking and non-cracking conditions are investigated regarding some experiments reported in the literature. Results of the welding simulations are coherent with these experimental observations to distinguish cases using criterion. Furthermore, the grain structure simulated by CAFE method is considered thereafter in order to investigate the influence of grain structure on hot cracking. Despite similar results as that without grain structure are obtained, a coherency with experimental observations is found.

Some discussions are thereafter proposed as a perspective for these activities, for example, validation of the WYSO hot cracking criterion with more welding configurations. In addition, the current ultrasonic Non Destructive Testing (NDT) used to detect and localize the hot cracking could also benefit from the reliable simulated grain structure to analyze and improve the NDT software performances.

Keywords: welding, cracking, solidification, multi-scale modelling, microstructure, CAFE approach

AFPM2022

10 February 2022

The Advanced Functional Polymers for Medicine, AFPM 2022, conference will be organized by the BIO Team of CEMEF on site in Sophia Antipolis from June 1st to June 3rd, 2022.

The purpose of the AFPM conference series is to strengthen the interactions within the community of chemists, material engineers, physicists, biologists and clinicians in the development of Advanced Functional Polymers for Medicine. The AFPM 2022 conference will offer delegates innovative and stimulating topics with a well-balanced programme of invited speakers and poster presentations.

Deadline for poster abstract submission: March 15th, 2022

> More details on the AFPM22 site

> Contact: Dr. Sijtze Buwalda, sijtze.buwalda@minesparis.psl.eu

PhD defence of Saoussen Ouhiba

9 February 2022

Saoussen Ouhiba defends her PhD in Computational Mechanics and Materials on Feb. 22nd, 22

Recrystallization of 6016 aluminum alloy during and after hot rolling

Saoussen Ouhiba conducted her PhD work in MSR team, under the supervision of Nathalie Bozzolo and Marc Bernacki. Saoussen defends her PhD in Computational Mechanics and Materials on Feb. 22nd, 2022 in front of the following jury:

– Prof. Knut Marthinsen, Norwegian University of Science and Technology, rapporteur

– MC. Myriam Dumont, Arts et Métiers Lille, rapporteur

– Prof. Franck Montheillet, Mines Saint Etienne, examinateur

Abstract:

The increasing need for weight reduction in automotive applications has led to paying increasing attention to aluminum alloys, especially to age-hardenable 6xxx (Al-Mg-Si) alloys. Although these alloys are known for their good formability, good corrosion resistance and sufficient strengthening potential, they are sometimes prone to the appearance of sheet surface defects. A better understanding of the microstructural evolutions occurring during the hot rolling process is needed to optimize the final properties of the material. Particularly, the interaction between precipitates and solutes with dynamic recovery, dynamic and post-dynamic recrystallization processes is studied. In this context, different initial precipitation states are generated prior to deformation and are subsequently submitted to hot compression tests. Microstructural analyses are performed using Scanning Electron Microscopy and crystal orientation mapping by EBSD. The factors possibly leading to the anisotropic development of coarse recrystallized grains during holding after deformation are also investigated through sequential annealing using a fast heating stage coupled to SEM. This is followed by 2D full field simulations based on the Level Set method in order to discuss the validity of specific assumptions regarding the anisotropic overgrowth of some recrystallized grains. Finally, the influence of different thermomechanical parameters on microstructure evolution is studied, determining thus the thermomechanical parameters promoting the formation of a finer final recrystallized grain size.

Keywords: aluminium alloys, hot rolling, recrystallization, coarse recrystallized grains, effect of precipitates/solutes

PhD defence of Junfeng Chen

21 January 2022

Convolutional neural networks for steady flow prediction around 2D obstacles

Junfeng Chen conducts his PhD work in the CFL team. He will defend his PhD in Computational Mathematics, High Performance Computing and Data, on January 27th, 22 in front of the following jury:

– Prof. Emmanuelle Abisset-Chavanne, Arts et Métiers ParisTech

– Prof. Anne Johannet, IMT Mines Alès

– Prof. Jean-Luc Harion, IMT Mines Telecom Lille Douai

– Prof. Elie Hachem, CEMEF Mines Paris, PhD Supervisor

– MA Frédéric Heymes, IMT Mines Alès, PhD Co-supervisor

– IR Jonathan Viquerat, CEMEF Mines Paris, PhD Advisor

Abstract :

Over the past few years, neural networks have arisen great interest in the computational fluid dynamics community, especially when used as surrogate models, either for flow reconstruction, turbulence modeling, or for the prediction of aerodynamic coefficients. This thesis considers using convolutional neural networks, a special category of neural networks designed for images, as surrogate models for steady flow prediction around 2D obstacles. The surrogate models are calibrated in the framework of data fitting, with the data set prepared by high-fidelity solvers to Navier-Stokes equations and projected onto cartesian grids. Once calibrated, the models show high accuracy in terms of velocity and pressure prediction, even around obstacles not seen during the calibration. In the next step, a new architecture of convolutional neural networks is proposed for anomaly detection and uncertainty quantification along with the steady flow prediction, making the surrogate model aware whether it is doing interpolation or extrapolation while doing prediction. With these methods, the user of a calibrated neural network can either decide whether to accept a prediction or not, or have a quantified estimation of the prediction error. The third contribution is to use graph convolutional neural networks as surrogate models to predict velocity and pressure on triangular meshes, which have significant advantages in geometry representation compared to cartesian grids. Thanks to the mesh refinement close to the solid interfaces, the graph-based model can give more accurate boundary layer prediction than traditional convolutional neural networks. The last part of this thesis considers integrating physical knowledge into the calibration of a graph convolutional neural network, which is calibrated by minimizing the residual of Navier-Stokes equations on a triangular mesh. The predicted velocity and pressure around a cylinder are of very high quality when compared to the results of high-fidelity numerical solvers. Being not in the framework of data fitting, this approach provides a novel solver to partial differential equations, and deserves more work on its convergence and computational cost.

Keywords: convolutional neural networks, surrogate models, surrogate models, finite element, uncertainty quantification, graph convolution

PhD defence of Karen Alvarado Vargas

12 January 2022

Karen Alvarado Vargas defends her PhD in Computational Mechanics and Materials on Jan. 20th, 22

Grain growth under the influence of the Smith-Zener pinning phenomenon with an evolution of the second phase particles: multiscale approach and application to nickel-base superalloys

Karen Alvarado Vargas conducted her PhD work in MSR team, under the supervision of Marc Bernacki and Nathalie Bozzolo. Karen defends her PhD in Computational Mechanics and Materials on Jan. 20th, 2022 in front of the following jury:

– Prof. Julien Bruchon, Ecole des Mines de Saint Etienne, SMS

– Prof. Michel Perez, INSA Lyon

– Prof. Alain Hazotte, Université de Lorraine, LEM3

– Prof. Olga Bylya, University of Strathclyde, Advanced Forming Research Centre

– Dr. Pascal De Micheli, Transvalor

Abstract:

In most polycrystalline nickel base superalloys, the grain size is controlled by second phase particles which pin the grain boundaries. The Smith-Zener model describes this physical interaction. Industrial forging processes involve hot deformation steps near the solvus temperature, where second phase particle dissolution occurs. Therefore, it is essential to understand and predict their evolution to properly control the grain size obtained after a specific subsolvus solution treatment and, in turn, the related material properties.

Three nickel base superalloys were studied (AD730, René 65, N19) through a series of isothermal treatments and detailed microstructural analyses based on SEM and EBSD techniques. More precisely, the primary 𝛾’ precipitates, particle fraction, size, morphology as well as the grain size evolution were analyzed. A temperature-time codependency equation was established to describe the evolution of primary 𝛾’ precipitates of each material using experimental data, the Thermo-Calc results, and the Johnson-Mehl-Avrami-Kolmogorov (JMAK) model.

Numerical simulations could then be performed using a full-field modeling framework for simulating grain growth (GG) phenomena, based on the Level Set (LS) method within a finite element context, with second phase particle evolving according to the previously established kinetic models. This numerical approach was preferred over the others found in the literature because it can consider substantial deformations, which opens the possibility of reproducing more realistic thermomechanical paths like the ones used in the metal forming industry, including both hot deformation and thermal treatment steps. In this work, a new formalism based on the LS to model GG mechanisms under the influence of the Smith-Zener pinning and able to consider precipitate evolution was proposed. The precipitates are represented using an additional LS function; a numerical treatment around the grain boundary in the vicinity of the precipitates is then applied to reproduce their pinning pressure correctly. Thus, considering the actual precipitate dissolution based on phenomenological laws, these simulations aim to predict grain size evolutions not only in terms of long term steady state, but also within the transient regime.

Keywords: Grain growth, Smith-Zener pinning, Level-Set, full-field simulations, 𝛾-𝛾’ superalloys, precipitate dissolution kinetics

PhD defence of Hazem Eldahshan

4 January 2022

Hazem Eldahshan defends his PhD in Computational Mechanics and Materials on Jan. 11, 22

3D ductile damage to fracture transition based on phase field and mesh adaptation: application to metal forming processes

Hazem Eldahshan conducted his PhD work in CSM team, under the supervision of Pierre-Olivier Bouchard and Daniel Pino Munoz. Hazem defends his PhD in Computational Mechanics and Materials on Jan. 11th, 2022 in front of the following jury:

– Laura De Lorenzis, Professor, ETH Zurich

– Carl Labergère, Professor, Université de Technologie de Troyes

– Jacques Besson, Directeur de recherche CNRS, Mines Paristech

– Yann Monerie, Professor, Université de Montpellier

– José Alves, Dr, Transvalor S.A.

– Etienne Perchat, Dr, Transvalor S.A.

Abstract:

This PhD contributes to the modeling of damage to fracture transition within a 3D parallel numerical framework based on the finite element method. The new contributions include: (i). a coupled phase field-damage formulation that is adapted to metal forming applications; (ii). Adaptive remeshing followed by the identification of the crack surface intersection with arbitrary mesh topologies based on the phase field evolution; (iii). fitting the crack surface within the mesh using local mesh partitioning operations; (iv). a nodal duplication strategy based on the local coloring algorithm in order to open the crack faces followed by a volume remeshing step.

The proposed framework offers a robust numerical tool for the modeling of damage to fracture transition in complex industrial processes such as: (i). the formation of internal chevron cracks during bar extrusion process; (ii). metal cutting simulations including bar shearing, blanking and piercing processes. Comparisons with the element deletion method show significant improvement in the quality of predicted sheared surfaces with the ability to accurately detect surface features such as burrs. In addition, both volume and energy are conserved during the simulation which resolves one of the main issues that appears with the element deletion method.

Keywords: Ductile fracture, Damage to fracture transition, Phase field model, Adaptative remeshing, Discrete crack propagation, Metal forming applications