Host Research Group
FR5_Energy, Fluids and Transfers Department - I2M LAB_Stéphane Glockner
Stéphane Glockner
+33 540003417
https://www.i2m.u-bordeaux.fr/
Group description
The group is composed of UBx, ENSAM, INP Bordeaux and CNRS researchers of I2M (Institute of Mechanics and Engineering of Bordeaux) and IMS (Integration: from Material to Systems) laboratories that works together around the theme “Multiphysics modelling – from model to advanced simulations” applied to industry 4.0. The industry of the future will rely on numerical technologies that should be able to design and manufacture innovative, complex, and optimized products, in terms of their properties, and be flexible and responsive to meet the various expectations. These challenges will require optimized experiments, data usage, and most importantly, virtual manufacturing processes that may replace tedious and costly legacy experimental campaigns. This long-term objective requires:
- Advanced multiphysics and multiscale modelling. Most of the applications discussed hereafter involve multiphysics and multiscale modelling approaches. Peculiar phenomena at the basis of some processes are still not well identified and/or modeled. Therefore, new modelling approaches are necessary and need to be adapted to the scale of observation/simulation in order to better characterize the properties of materials, their responses to conditions of use, and to control their manufacturing processes. The involved scale ranges from microscopic interfaces to macroscopic phenomena.
- Advanced simulations. They will represent complex physical phenomena from the micro scale even up to the manufacturing processes scale. They will be based on high performance computing as well as up-scaling technics. Moreover, in order to deal with some engineering problems, there is a need of reduced-order modelling and optimization techniques since computational resources are limited. These methods find their natural place in the design of hybrid and real time numerical manufacturing twins. We will also contribute to the development of numerical tools, repositories of skills and expertise, necessary to reach the axis goals. These tools claim to be general, transversal, performant, easy to use, with state-of-the-art models and numerical methods, and of great interest for the different scientific communities involved in the project.
The modelling of material manufacturing and properties is based not only upon fundamental mechanisms, but also upon their coupling and their peculiarities at the microscopic interface level. For instance, for multicomponent systems thermo-chemo-mechanical coupling governing transport of mass, energy and momentum is of relevant interest. For multiphase problems, fluid/fluid, fluid(s)/solid, solid/solid interfaces are encountered that lead to several modelling and numerical challenges. The understanding of interface thermodynamics should be improved, as well as their chemical and/or mechanical interactions. Some applications require a multiscale approach, from micro to macro, to take into account larger volume of materials. At the macroscopic scale, the equations are generally up-scaled to obtain a macroscale set of conservation equations complemented by various closure relations. The loss of information associated with these approaches needs to be reduced and a mastering of the comprehension of the physics at micro scale is imperative. Therefore, a main aim of this project is the development and application of new modelling approaches and frameworks for the modelling of reactive multiphase materials and processes at different scales, associated with computational strategies for numerical solutions of derived mathematical models.
The structured approach of this project will highlight some codes for the next decade to handle some of the multiphysics and multiscale numerical modelling expertise. The existing open-source code Notus (developed at I2M, https://notus-cfd.org) will provide the computational framework and constitute the solid background for numerical solutions of some of the developed multiphysics models. An example of the numerical difficulties is the treatment of the interfaces that bring their numerical difficulties (discontinuities of properties, mass transfer, wetting, phase change, etc.). Notus already disposes of several approaches such as Level-Set, Volume-of-Fluids, Moment-of-Fluids, Phase-fields as standard tools or under development. Part of the code has been proven to run on a petascale supercomputer on more than 100 000 processors in 2020 during a “great challenge” call of GENCI. Further developments, design and optimization will ensure the compatibility of Notus toward exascale computing. It gives us a glimpse of the major changes to multi-physics simulations as well as the ability to simulate large material volume at micro scale without any macro scale model as well as some processes as a whole. Results of such simulations can also facilitate definition of microscale-informed constitutive closure relationship needed in a macroscale formulation.
Some of the target applications need a research effort on reduced-order modelling that brings a paradigm shift thanks to the reduction of the computational time by several orders of magnitude. Others application need to rethink and to develop dedicated modelling strategies and robust design/optimization algorithms for the multi-scale topology optimization. These parts of this project will also be based on their own and original libraries or codes.
Keywords
- Modeling
- Multiphysics
- Multiscale
- Optimization
- Reduced Order Modeling
- Simulation
- Material
- High Performance Computing
- Numerical methods
- Numerical code
Team Description
-
Sylvie Bordère (Principal Investigator RL1)
ORCID: 0000-0003-2372-8224
-
Alexandrine Gracia (Co-Principal Investigator RL1)
ORCID: 0000-0002-9566-7961
-
Stéphane Glockner (Principal Investigator RL2, Co-Principal Investigator RL1,3,8)
ORCID: 0000-0002-4554-4541
-
Antoine Lemoine (Co-Principal Investigator RL2)
ORCID: 0000-0002-1919-9350
-
Mejdi Azaiez (Principal Investigator RL3,7)
ORCID: 0000-0002-7886-4007
-
Giuseppe Sciumè (Principal Investigator RL4)
ORCID: 0000-0001-5231-7374
-
Yves Chemisky (Principal Investigator RL5)
ORCID: 0000-0002-8725-9554
-
Marco Montemurro (Principal Investigator RL6)
ORCID: 0000-0001-5688-3664
-
Arnaud Erriguible (Principal Investigator RL7)
ORCID: 0000-0003-2454-5307
Projects
-
SUPERFON (Sustainable SUPERcritical Processing of Fluorescent Organic Nanocrystals through complementary experimental and numerical approaches)
Pl: Arnaud Erriguible
Funding Agency*: NAT
Ongoing: yes
-
COFFA (Shape Design and Optimization for Additive Manufacturing)
Pl: Marco Montemuro
Funding Agency*: NAT
Ongoing: yes
-
KAM4AM (Fabrication additive assistée par la connaissance et l'intelligence artificielle)
Pl: Marco Montemuro
Funding Agency*: NAT
Ongoing: yes
-
Phasefield (Phase-field models, algorithms and simulations for multiphase complex fluids)
Pl: Mejdi Azaiez
Funding Agency*: NAT
Ongoing: yes
-
HYDILIC (Hydrothermal sintering: a low temperature process for the densification of ceramics)
Pl: Sylvie Bordère (modeling part)
Funding Agency*: NAT
Ongoing: yes
Publications
-
M. Azaïez,, T. Chacon Rebollo, S. Rubino, A cure for instabilities due to advection-dominance in POD solution to advection-diffusion-reaction equations., Journal of Computational Physics, 2021
10.1016/j.jcp.2020.109916 -
Le Maout, V, Alessandri, K, Gurchenkov, B, Bertin, H, Nassoy, P, Sciumè, G., Role of Mechanical Cues and Hypoxia on the Growth of Tumor Cells in Strong and Weak Confinement: a Dual in vitro-in silico Approach, SCIENCE ADVANCES, 2020
10.1126/sciadv.aaz7130 -
A. M. D. Jost and S. Glockner, Direct forcing immersed boundary methods: Improvements to the Ghost Node Method, Journal of Computational Physics, 2021
10.1016/j.jcp.2021.110371 -
S. Bordere, Glockner, S., Numerical modeling of diffusion-controlled phase transformation using the Darken method: Application to the dissolution / precipitation processes in materials, Computational Materials Science, 2021
10.1016/j.commatsci.2020.109944 -
E. Ben Romdhane, A. Guédon-Gracia, S. Pin, P. Roumanille, H. Frémont, Impact of crystalline orientation of lead-free solder joints on thermomechanical response and reliability of ball grid array components, Microelectronics Reliability, 2020
10.1016/j.microrel.2020.113812 -
G Costa, M Montemurro, J Pailhès, NURBS hyper-surfaces for 3D topology optimization problems, Mechanics of Advanced Materials and Structures, 2019
10.1080/15376494.2019.1582826
Research Lines
ADVANCED MATERIALS AND PROCESSES
- The direct simulation of the mass transfer at the interfaces, requiring consideration of the interface thermodynamics (property jumps), is a complex problem that is a source of modeling and numerical difficulties. The mechano-chemical coupling is a key issue in this transfer for both modelling and numerical simulation. The proposed modelling will be based on the unified governing equation for fluid flow and elastic solid deformation, and for mass transport, on the Darken methodology developed for diffusion in solids which was shown to be particularly adapted to deal with multi-component systems. This methodology is currently extended with Notus code to two-phase systems with the advantage to include several potentials. This original approach, coupled over the long term with exaflopic simulations of an arrangement of objects (grain, fiber, etc.) on the order of a million units, would make it possible to account for the heterogeneities specific to complex large-scale systems. Moreover, this strategy can help provide accurate interface-level information needed to improve macroscopic models. Numerous possible applications exist, from short to long term, including intermetallics growth during transient liquid phase bounding involved during the soldering processes, optimization of the dissolution / precipitation processes during metal deposition on carbon fibers by an innovative molten salt process, additive manufacturing, hydrothermal sintering modelled at the scale of a grain compact, optimization of the controlled dissolution of active ingredients of oral drugs by adding insoluble excipients.
- Capillary ascension is a well-known physiochemical process when a wetting liquid colonizes the open and interconnected pores of porous ceramics. It is crucial in various important industrial sectors such as the development of composites with metal or ceramic. Besides experimental approaches, there is a need to develop numerical models to simulate different infiltration scenarios corresponding to various pore morphologies and fluid properties. The challenging long term goal of this task is to propose a numerical code capable of reproducing, at the pore network and at an interface scale, different infiltration scenarios corresponding to various pore morphologies and fluid properties. To accurately simulate the triple line we first propose the development of a numerical framework for the geometrical representation of the fluid/fluid and fluid/solid interfaces and their interactions. We typically face multi-material problems with the presence of 3 materials in a single grid cell. To address this fundamentally geometric problem we will rely on the coupling of multi-material Level-Set or Moment-of-Fluid approaches with the Immersed Boundary Method. Secondly, we will focus on the physics-based modeling of the triple line motion within the previously defined framework. As in 3D CFD simulations it is currently not possible to resolve the smallest scale phenomena, which is on the order of nanometers, we will follow a macro-micro approach based on the combination of the Generalized Navier Boundary Conditions and the Cox Law, Although this approach has been successfully used to simulate succesfully different stages of the capillary rise in a tube it needs to be extended to complex geometries for which special attention must be paid to the rapid geometric curvature contour changes that are present in real geometries. Finally, in order to represent a sufficiently large material volume, the proposed developments will be pursued in the high performance computing framework of Notus code.
- Many industrial sectors, such as pharmaceutical or civil and military aeronautics, use this process in order to obtain mixtures that are as homogeneous as possible. The RAM type mixer is characterized by the use of a non-intrusive method based on the rapid oscillation of the system. The objective of this task is to contribute in the field and to realize a numerical simulation code of the RAM mixer. The tool will be an aid to the dimensioning and design of the most homogeneous mixture possible to ensure its effectiveness and to optimize the technological solutions that use it. Due to the nature of the problem, the numerical modeling of the different scales will be one of the components to be realized and then implemented in the Notus code. The direct simulation of the mixture by vibration is complex and requires the choice of a very small-time step imposed by the vibration frequency, which makes the simulations very costly. One solution is to implement a technique for separating scales between the fast times related to vibration and the long times related to fluid dynamics. The second strategy consists in derive and implement a reliable scale model using data and model reduction techniques. In this RL we plan to follow the direct simulation approach that will be helpful to validate the averaged models, and to consider the particle transport and its interaction with the fluid. This complete approach that combines modelling, numerical methods, and high performance computing will allow us to be able to perform peta/exaflopic simulations of the full process and become a leader in the domain. In the pharmaceutical industry, obtaining of a homogeneous mixture of the active substance and the excipient will ensure the manufacture of large quantities of drugs while maintaining stability in their composition. We aim on the long term to take advantage of the tools developed in this task in order to adapt and orient them to the pharmaceutical application. Several issues will need to be revisited and solutions improved. The project involves several international collaborations. Among them, we mention the team of Prof. Jie Shen Perdue University (USA) and Prof. Lybimova Tatyana Perm University (Russia).
- Macroscale multiphase models are typically founded on the Mixture Theory (MT). Despite its large use, MT is not quite exhaustive and general to model multiphase materials with hierarchical structure in which considered phases may consist of several species. To address such modeling challenge this task aims at the development of a general mathematical framework founded on Thermodynamically Constrained Averaging Theory, TCAT, which is a modern rigorous approach for modelling of reactive multiphase porous medium system. TCAT have two main advantages with respect to MT: i) by formally averaging the microscale equations up to the macroscale scale (instead of postulating them directly at the larger scale, as done in MT), larger scale variables are expressed precisely in terms of microscale precursors; ii) each phase may be constituted by several species. This second advantage allows us to make a rigorous distinction between: intra-phase transfer of mass/momentum (between species within the same phase) and inter-phase transfer of mass/momentum (from one phase to another one); advective and diffusive transport of species.
- Starting from the scale where phases constituting the multiphase system of interest can be clearly identified, conservation equations of mass energy and momentum as well as thermodynamic relationships and entropy balances will be established for each species in each phase constituting the system. Such equations will be up-scaled following TCAT procedure to obtain a macroscale set of conservation equations and a simplified entropy inequality (SEI). The SEI will provide guidance about functional dependencies of the constitutive relationships needed to close the mathematical model. Constitutive relationships will be qualitatively and quantitatively defined thanks to numerical upscaling of direct microscale simulations (performed with Notus) of multiphase flow within a suitable representative elementary volume (REV). An adaptive combined CutFEM/XFEM/level-set framework will be developed in collaboration with Legato team (legato-team.eu, University of Luxembourg) for suitable treatment of moving interface problems.
- TCAT models are currently developed at I2M for specific applications in transport oncophysics and experimental-numerical concrete multiphysics in collaboration with University of Grenoble Alpes. The envisaged general computational framework will extend applicability of TCAT to a larger range of problems, spanning from material, civil and environment engineering, to the biomedical industry.
- New architected materials and composites are the future of materials in engineering applications due to their extraordinary ability to tailor mechanical properties and obtain optimal light weight and high strength and/or stiffness features. The design of such materials require a very accurate numerical simulation of their multi-physics response, but also local fields to evaluate their durability. A major challenge of efficient and accurate multiphysic simulations is the generation of realistic representative primitives/volumes of material microstructure. This is a tedious task since it has to be adapted to (i) characterization methods; (ii) multiphysics simulation analysis softwares; (iii) analysis, post-treatment and optimization methods. Idealized geometric representation should be optimized to match real microstructures geometry. In the other hand, direct 3D-reconstruction micro-computed tomography suffer from the exact recognition of phases and complex geometries. The ambitious goal of this task is to gather and fuse those methods: The improvement of 3D-reconstruction of micro-computed tomography will allow the recognition of complex objects, using machine learning classification methods. Coupled with algorithms for 3D virtual materials generation, the goal is to reconstruct an artefact free virtual material, which is an accurate representation of the microstructure. State of the art geometrical and 3D representation open-source libraries will be used (CGAL, Opencascade). I2M laboratories and LCTS have started to share their expertise in the generation of composite materials microstructures and geometry reconstruction (short fiber reinforcements and CMC, respectively). This synergy is a strong indicator of success of this task, and allow to envision long term objectives.
- Standard modelling strategies of lightweight cellular materials, and the related design/optimization tools, are not adapted to the new possibilities offered by modern additive manufacturing (AM) technology. To go beyond these limitations some medium-term challenges must be faced.
- An efficient global/local multi-scale modelling strategy and mathematical formalism will be developed to assess, in a reasonable time, physical responses at the different scales and the related coupling effects (collaboration with University of Pisa and University of Bologna).
- The Topology Optimization (TO) algorithm SANTO (SIMP And NURBS for Topology Optimization) will be generalized to include different homogenization schemes and different physics. SANTO combines the Solid Isotropic Material with Penalization (SIMP) scheme with the Non-Uniform Rational Basis Spline (NURBS) hyper-surfaces (collaboration with CEA-CESTA, University of Florence, University of Rome).
- A general and automatic surface reconstruction strategy of the boundary of the optimized solutions resulting from the TO phase is still lacking. The methods available today are fully automatic only for 2D problems, whilst only semi-automatic surface reconstruction strategies can be found in the literature. Surface reconstruction is articulated around mapping and fitting phases. The problem of finding a suitable mapping can be smartly formulated by using non-Euclidean geometry concepts and tools. Eventually, the uncertainty on the basic primitive surfaces/curves constituting the boundary of the optimized topology could be included in the surface fitting problem by exploiting the features of the PolitoCAT software, developed at the I2M, which is based on polytopes for tolerance analyses.
- Further interesting long-term challenges could be faced:
- Material and geometrical non-linearity can be integrated into the formulation of the multiscale TO problem (including anisotropy).
- The technological constraints of geometrical nature relate to the AM process could be integrated into the multi-scale TO problem in order to obtain optimized and fabricable lightweight cellular materials (in collaboration with G-SCOP and PIMM laboratories).
ENERGY EFFICIENCY
- Effective thermal energy storage represents one of the competitive advantages concentrated solar power systems have to deal with the natural intermittency of solar energy. This Task aims at exploring, both from the full and reduced order numerical point of view, novel phase change materials (PCM) for thermal energy storage. Efforts will concentrate on developing a reduced order flow solver able to model these new PCM composites. Additional research will be dedicated to further understand peritectic transitions and their potential as new thermal energy storage systems. The proposed project will create an accelerated flow solver for complex flows that will rely on reduced order models (ROMs). Two types of ROM will be considered for this particular problem. At first, a posteriori methods, such as proper orthogonal decomposition based methods, will be created. The advantage of this approach is that the solutions from any program (Notus for our project) (or experiment) may be used for creation of the ROM; this allows black box programs to be used. At the same time, a ROM based on a priori methods would also be very useful in cases where the equations within the software could be modified. A priori methods, such as proper generalized decomposition, do not rely on previously calculated information but rather create the model during flow solution. These methods have been studied over the past few years. Instead of calculating a database of solutions and using these solutions to create the ROM, the a priori method would create the ROM during calculation of the first solution. All subsequent solutions would then be calculated by a simple linear recombination that would be practically instantaneous. The outcome of the task is twofold: an accelerated flow solver for complex flows, such as those found in PCM-infused matrices and in free surface flows with sedimentation, and a multi-parameters optimization tool for process design. At the end of the first conception phase of a numerical simulation tool for the implementation virtual materials optimized for energy storage, we plan on the long term to collaborate with our partners for the design of a prototype. First by testing our algorithms on the basis of phase change materials existing in the industry, then by proposing new topological and structural material. The project involves several international collaborations. Among them, we mention Prof. G. Rozza from SISSA Italy and Prof. T. Chacon from Sevilla university Span. Prof Chacon is candidate for an international chair of Bordeaux Idex.
SUSTAINABLE MANUFACTURING
- The use of sustainable supercritical technologies should allow to reduce and control the size of produced materials. Furthermore, to go through fundamental mechanisms to processes and to realize the promise of an efficient “green” process, it is essential to access large production scales. One major originality of the approach developed by the involved team is to propose the same level of description for the simulation in microfluidic device as in the larger reactor. Advanced petascale simulations with Notus code are currently used as a modern tool of the process design (reactor, injector design, and operating conditions) to investigate two scaling options, i.e. scaling-up (large semi-continuous reactor) vs. numbering-up (parallel microfluidic reactors). Following this strategy, we will first focus on hydrothermal materials synthesis in supercritical environment. It will be necessary to add to the existing modeling the thermo-compressible effects and the chemical reactivity, and also new development toward exascale simulations. Recent works of the team published in Science Advances highlighted that a combination of supercritical water and molten salts could be significantly beneficial for supercritical hydrothermal synthesis. As a result, the ultimate ambition will be to propose numerical simulations capable taking into account this two-phase medium and managing the heat and mass transfer at the interface. Collaboration with Tohoky Universtiy will be developed on these topics.
This project is connected to one of the six work packages of the emerging “Impulse Network” project of University of Bordeaux BEST 4.0 “Making better to live better” that intends to offer completely original solutions to complex scientific issues around the design, manufacture and control of materials and processes in a wide variety of industrial applications. The project aims to develop and promote multiphysical and multiscale models and methods throughout the chain going from intelligent sensors to actuators, in order to optimize the industrial systems in real-time. Therefore, the use of intensive computing means (exaflopic computations) will be intelligently associated with upscaling methods, model reduction and machine learning methods. The research units are attached to departments of UB. The I2M, IMS, LABRI, IMB and Estia Recherche units constitute the Engineering and Digital Sciences (SIN) department, which brings together all skills in the fields of mechanics, electronics and control, computer science and mathematics. The LCTS, LOF, ICMCB, ISM and CRPP units are attached to the Department of Materials and Radiation Sciences (SMR) which aims to bring together skills in chemistry and physics.
BEST4.0 offers original methodological developments which have potential applications in the medium and long term in different sectors of industry. Many industrial groups have expressed their interest in the scientific objectives of the project and some have expressed the desire to follow the progress of the project. All of these industrials have already collaborated with the project's research units, notably through CIFRE financed PhDs or even post-doc: SAFRAN (additive manufacturing, composite materials, on-board systems, aeronautical structures and drones, aeronautical maintenance), CNES (materials for satellites, space maintenance), CEA (CESTA, Ripault) energy, high temperature materials), LETI (on-board electronics, sensors, IoT), LECTRA (cutting of flexible materials, communicating processes, software), VALOREM (renewable energies, energy transition, composite materials), DASSAULT (aeronautics, embedded systems, advanced design software), ARIANE GROUP (materials for space launch vehicles, advanced manufacturing processes), SEIV-ALCEN (manufacturing and maintenance of complex structures).
We will be open to new collaboration with the University of the Basque Country.