Abstract:

In this work we present a novel Heterogeneous Multiscale Method capable of representing strain field heterogeneities induced by evolution (and interaction) of localized failure mechanisms in massive structure, pertaining to micro scale (FPZ-fracture process zone), macro scale including softening (macro cracks) and non-local macro scale (bond-slip for long fiber reinforcement). The objective of Heterogeneous2 Multiscale Method is also to provide capabilities for quantifying the risk of premature localized failure through probability description of initial defects (microstructure heterogeneity) and uncertainty propagation through scales. The novel scientific concept pertains to multiscale formulation and solution of coupled nonlinear mechanics-probability problem replacing the standard homogenization approach that can only provide average (deterministic) properties of heterogeneous composites. This concept is of interdisciplinary nature with Mechanics (defining probability distribution) and Applied Mathematics (providing uncertainty propagation) combined in order to capture the influence of heterogeneities and fine scale defects on premature failure.
The most important challenge concerns the ability to provide the sound, probability-based explanation of size effect (with different failure modes observed for different size specimens and real structure built of the same composite materials). The proposed approach has the potential of changing the validation procedures for massive structures that are beyond the size suitable for testing at present. The other potential gains concern providing the Heterogeneous2 Multiscale Method that connects computations with design studies (optimization), testing (identification) and safety verification (monitoring) of massive composite structures.
The application domains are very broad and pertinent to important challenge on quantifying durability, life-time integrity and safety against failure of massive composite structures under extreme conditions. The illustrative applications come from the area of energy production systems, with both currently dominant nuclear (nuclear power plant) or renewable energy sources (wind turbines and hydro-turbines). Special attention is given to costly massive structures with ‘irreplaceable’ components, which are characterized by a number of different failure modes that require the most detailed description and interaction across the scales.

LECTURER_1
Lecturer:

Prof. Adnan Ibrahimbegovic
Chair for Computational Mechanics & Institut Universitaire de France
Sorbonne Universités / University of Technology Compiègne
E-mail: adnan.ibrahimbegovic@utc.fr
URL: http://www.utc.fr/~ibrahimb/

Mini-CV:

Adnan Ibrahimbegovic is known for his early works on structural mechanics (beams, plates and shells) and structural and multibody dynamics, as well as more recent works on continuum and discrete models for fracture, multiscale modeling of inelastic behavior and coupled/interaction problems based upon code-coupling approach. He is the author of close to 500 publications including 7 books and more than 150 refereed scientific papers.
His former doctoral students (31) and post-docs (5) are working at present mostly in research environments, either in France (at least one with each top engineering school in Paris: Ecole des Ponts Paris, Ecole Centrale Paris, ESTP; with well-known technology institutes: MIT, UT-Compiègne, INSA-Toulouse, IUT-Nantes; with French industry leaders: AREVA, EDF, Lafarge, Renault; with computer software developers: ALTAIR, SAMTECH) or abroad as university professors in 10 different countries (France, Mexico, Czech Republic, Slovenia, Vietnam, Algeria, Bosnia, Croatia, Hungary, Turkey).
He is also a very active participant in IACM sponsored activities, such as editor/reviewer for scientific journals, organizer of short courses for industry and academia, as well as mini-symposia and scientific conferences (Euromech on MBD in 2001, NATO ARWs in 2004, 2006 and 2008, ECCOMAS Thematic Conferences on `Multiscale Computations for Solids and Fluids’ in 2007 and in 2015, as well as joint IACM-IASS 7th International Conference on ‘Shells and Spatial Structures’ in 2012).
His international recognition was confirmed by several international awards: Alexander von Humboldt Research Award in `Technischen Mechanik’ in 2005 for Germany, IACM Fellow Award in 2006, Research Award in 2007 for Slovenia, Chair Claude Levy-Strauss in 2012 for USP Brazil, Chair Åsgard in 2013 for NTNU Norway, Chair TUBITAK for ITU Turkey and OCSE KAIST Professor in 2014 for South Korea. His recognition in France was confirmed by his promotion to `Classe Exceptionnelle’ (attributed by peers elected to the National Committee in Mechanics) in 2009, Chair for Computational Mechanics position in 2014 at TU Compiegne/ Sorbonne Universités, and ‘Membre Senior’ of IUF-Institut Universitaire de France in 2015.

Abstract:

The mathematical modeling and virtual experimentation have been taken an importance similar to the material experimentation. In order to analyses a physical problem, the natural way to do that is to undertaken a material modeling and a virtual modeling (numerical and computational mathematics modeling). With these two types of modeling, virtual and material, the richness of information and the possibilities for complementarity characterize the new potential of problems solution. The ability to troubleshoot, the world community, in the past ten years, increased by two orders of magnitude, on the basis of this symbiosis between these two forms of experimentation. Virtual experimentation allows us to have access to additional information and even the information that would be very difficult to access for material testing. To do so, the efficiency and robustness of computational models is of crucial importance. Efficiency is essential, since it requires large amount of capacities of data storage and processing. The limit is the ability of parallel processing systems. It would be a mistake to expect that we can solve problems at the expense of brute force, that is, expect the machines to process computational models based on obsolete and inefficient methods. It is necessary to invest massively in modern methodologies, robust, efficient and reliable. It will be presented, in the present work, an efficient and robust methodology for solving problem of interdisciplinary and multi-scales natures in fluid mechanics.
The most frequent problems in computational fluid dynamics (CFD) involve some of the most challenging classes: multiphase reactive turbulent flows and fluid and structure interaction. The main physical characteristics of these problems are the presence of abrupt variations of physical properties and relationships of values between phases; the presence of complex geometries, deformable and mobile and the existence of chemical reactions involving hundreds of reactive components. In this context a methodology that encompasses the application of the following methods: dynamic mesh adaptativity for localized refinement; immersed boundary, for modeling complex and mobile geometries; subgrid turbulence modeling, which lends itself to Large Eddy Simulation of turbulence; robust numerical methods for problems with large variations of physical properties; robust linear systems solvers and parallelization for high performance processing. Examples of Application are presented in order to illustrate the robustness and efficiency for solving problems of great complexity, such as two-phase flows and flows with fluid-structure interaction. Results will be presented to the virtual experimentation of isolated bubbles, with the presentation of bifurcation between flow regimes. Virtual experimentation of two-phase flows with a population of bubbles and, finally, the virtual experimentation results for fluid and structure interaction problems will be presented.

Lecturer:

Prof. Aristeu da Silveira Neto
Federal University of Uberlândia – Faculty of Mechanical Engineering – Laboratory of Fluid Mechanics
e-mail: aristeus@mecanica.ufu.br
tel: 034 32394040 Ext. 605

Mini-CV:

Possui graduação em Engenharia Mecânica pela Universidade de Brasília (1983), mestrado em Engenharia Mecânica pela Universidade Federal de Santa Catarina (1985) e doutorado em Mecânica e Hidráulica pelo Instituto Nacional Politécnico de Grenoble (1991). Desenvolveu atividades de pós doutorado no CEA-Grenoble-Fr. em 1996. Atualmente é professor titular da Universidade Federal de Uberlândia. Tem experiência na área de Engenharia Mecânica, com ênfase em Mecânica dos Fluidos, atuando principalmente nos seguintes temas: Simulação Numérica e Escoamentos Turbulentos, Interação Fluido-Estrutura, Escoamentos Bifásicos, Modelagem da Turbulência.

Abstract:

Energy harvesting (EH), i.e. the process of extracting energy from the environment or from a surrounding system and converting it to useable electrical energy, is a prominent research topic, with many promising applications nowadays in the civil engineering field. Its areas of application currently focus to the powering small autonomous wireless sensors (thus eliminating the need for wires), in structural health monitoring and building automation applications. Regarding the latter, the prospect to implement autonomous sensors inside a building that monitor relevant parameters (temperature, humidity, chemical agent concentration etc.), and transmit intermittently data to a central unit is a recent and rapidly grown business, helped by the standardization of wireless (Wi-Fi) data transmission.
This study focuses on the numerical analysis and testing of a high efficiency Energy Harvesting device, based on piezoelectric materials, with possible applications for the sustainability of smart buildings, structures and infrastructures. The development of the device is supported by ESA (the European Space Agency) under a program for the space technology transfer.
The EH device, harvests the airflow inside Heating, Ventilation and Air Conditioning (HVAC) systems, using a piezoelectric component and an appropriate customizable aerodynamic appendix or fin that takes advantage of specific air flow effects (principally Vortex Shedding), and can be implemented for optimizing the energy consumption inside buildings.
In the present research, focus is given on different relevant modelling aspects, explored both using numerical methods (by means of FEM and CFD models) and in wind tunnel testing. In particular, different configurations for the piezoelectric bender (including rectangular, cylindrical and T-shaped) are modelled, tested and compared. The calibration of the numerical models, useful for the optimisation of the final design, and the electrical modelling and losses calculation for the EH circuit, are provided, and the effective energy harvesting potential of the working prototype device in laboratory conditions is assessed. Additional aspects relevant to the successful implementation of the research project are shown, including the final design of the device and the possible market impact.

LECTURER_3
Lecturer:

Prof. Dr.-Ing. Franco Bontempi
Ordinario di Tecnica delle Costruzioni – Professor of Structural Analysis and Design
Facolta’ di Ingegneria Civile e Industriale – School of Civil and industrial Engineering
UNIVERSITA’ DEGLI STUDI DI ROMA LA SAPIENZA – UNIVERSITY OF ROME LA SAPIENZA
Via Eudossiana 18 – 00184 Roma – ITALIA
tel. +39-06-44585070, cell. +39-339-3956300
E-mail: franco.bontempi@uniroma1.it
URL: http://uniroma1.academia.edu/FrancoBontempi

Mini-CV:

Born 1963. Military duty 1989. Degree Civil Engineering 1988 and Ph.D. Structural Engineering 1993, Politecnico di Milano. Professor of Structural Analysis and Design at the School of Engineering of the University of Rome La Sapienza since 2000 (Structural Analysis and Design, Steel Constructions, Fire Structural Design). Research period at Harbin Institute of Technology, Univ. of Illinois Urbana-Champaign, TU Karlsruhe, TU Munich. Consultant for special structures, forensic engineering.

Abstract:

This lecture will explain how computational modelling can be used to support the development and performance optimisation of neonatal care devices such as incubators, radiant warmers and oxygen hoods. The talk will also discuss current research on the development of a computer model to simulate the treatment of encephalopathy hypoxic-ischemia in neonates, which involves a combination of thermoregulation and hemodynamic models. The treatment is based on therapeutic hypothermia, which can minimize sequels resulting from insufficient perfusion that may cause permanent brain cell damage.

LECTURER_4
Lecturer:

Prof. Luiz C. Wrobel
Director, Institute of Materials and Manufacturing
Brunel University London
Kingston Lane, Uxbridge UB8 3PH, UK
http: www.brunel.ac.uk
E-mail: luiz.wrobel@brunel.ac.uk
Tel.: +44(0)1895 266696

Mini-CV:

Professor Luiz Carlos Wrobel, BSc, MSc, PhD, CEng, FIMechE, CMath, FIMA, is Professor of Computational Mechanics and Director of the Institute of Materials and Manufacturing at Brunel University London. His expertise is in the field of computational modelling of engineering problems. He wrote/co-wrote four technical books in this field which have been translated to Russian, Japanese and Chinese. He also published over 150 journal papers on the development and application of advanced modelling and simulation techniques for solution of a number of engineering problems. His research interests range from thermo-mechanical problems to the modelling and optimisation of medical devices. His research has been funded by EPSRC, EU and several industrial companies. Prof Wrobel is a member of the Editorial Board of five major international journals and has given plenary and invited lectures at a number of international conferences.

Abstract:

This work addresses the enhancement of a hybrid MPI/OpenMP parallel finite element fluid solver using runtime dynamic load balance. The mesh partitioning into n MPI tasks is done a priori using METIS. In the present finite element context, METIS partitions the element graph of the initial mesh. The elements are therefore the vertices of the graph, which weight can be roughly estimated according to the number of integration points used for each type of elements. This weight is used by METIS to balance the weighted sizes of the partitions, while minimizing the interfaces between the different partitions. The interfaces are the geometrical entities where the different point-to-point exchanges take place during the matrix-vector products occurring during the iterative solvers iterations.
The starting point of this research stems from two facts. First, one cannot ensure a perfectly load balance of the mesh partition, because the load balance provided by METIS is not perfect and because the weights of the element graphs represent only a rough approximation of the real calculations. Second, should the load balance be theoretically perfect, the hardware will not respond in an homogeneous way and thus load imbalance will result. Assuming these two problems cannot be solved a priori, we have to find out a dynamical remedy at runtime.
The proposed solution consists in using the possibility to share resources between the different MPI tasks a the node level, thus obtaining a dynamically load balanced computation. Of course, in the limit where we rely exclusively on OpenMP at the node level, load balance will depend on the initial mesh partitioning. Thus nothing can be done. The solution therefore consists in placing several MPI tasks on the same node and let these tasks exploit the resources when these are available. We will show the performance of the strategy through the solution of large scale examples.

LECTURER_5
Lecturer:

Prof. Guillaume Houzeaux
Dpt. Computer Applications in Science and Engineering
Barcelona Supercomputing Center (BSC-CNS)
Edificio NEXUS II – Office 3A, c) Jordi Girona 29
08034 Barcelona, Spain
Tel: +34 93 413 7781
Skype user: guillaume_houzeaux_bsc
http: www.bsc.es/about-bsc/staff-directory/houzeaux-guillaume
ResearcherID: D-4950-2012
e-mail: guillaume.houzeaux@bsc.es

Mini-CV:

Guillaume Houzeaux studied physics at the Université de Montréal, Canada. After his Batchelor studies, he joined Prof. Habashi’s CFD group at Concordia University, Montréal, to start working in CFD and turbulence modelling, and obtained a Master of Applied Science in 1995. Afterwards he moved to Barcelona and carried out a PhD in Domain Decomposition methods applied to CFD at the Technical University of Catalonia, Barcelona, Spain. He obtained the PhD degree in 2002 (ECCOMAS Award for the Best Ph.D Thesis of 2002 on Computational Methods in Applied Sciences and Engineering) and during two years worked at the International Center for Numerical Methods in Engineering (CIMNE) in the Building, Energy and Environment group. In 2004 he obtained a ‘Ramon y Cajal’ Spanish government contract to study indoor ventilation in buildings.
Since 2005, GH is a research group leader at the Computer Applications in Science and Engineering (CASE) Department of the Barcelona Supercomputing Center – Centro Nacional de Supercomputación de España (BSC-CNS).
In BSC, GH (together with Mariano Vázquez) co-leads a group of around 30 researchers whose main mission is to develop the multi-physics parallel simulation platform Alya. This code, part of the European PRACE benchmark suite, is an HCP-based tool adapted to run efficiently in large-scale parallel computers. This involves Physical modelling, Mathematical algorithms and code development and optimization, all with the strong constraint of efficient use of parallel resources. This multidisciplinary group includes post-docs, PhD students and programming engineers.
GH’s main research lines are all in the field of HPC and Computational Mechanics, such as: incompressible flows and stabilization issues, low-Mach regimes, computational biomechanics (particularly the respiratory system), domain decomposition methods, multi-physics coupling, and parallelisation of algorithms on distributed memory supercomputers.

Abstract:

In mechanics, as well as in various other fields, variational formulations offer many advantages compared to classical formulations directly based on partial differential equations. For conservative mechanical systems, Hamilton’s principle in dynamics, or the Minimum Potential Energy principle in statics, for example, have been widely used for a long time, and they are at the basis of most modern numerical approximation methods (finite elements probably being the most prominent one today). Similarly, variational principles can be derived for dissipative problems (e.g. Maximum Dissipation Principle in perfect plasticity).
For more complex problems, such as elasto-(visco-)plasticity, it has been shown that variational formulations can be derived from the optimization of energetic path integrals. Upon time discretisation, this approach leads to Incremental Variational Formulations of boundary-value problems and associated constitutive updates. This approach is applicable to all constitutive models entering the framework of Generalized Standard Materials (e.g. elasto-plasticity, visco-elasticity, at small or large strains). Such variational formulation of continuous and incremental boundary-value problems was later extended to coupled thermo-mechanical problems. In this incremental setting, it is then possible to recover many interesting mathematical features of elasticity problems, like symmetry, direct links between convexity (or lack thereof) and stability, … bringing significant advantages in terms of analysis and numerical methods.
In this presentation, we will show how this optimal path variational framework can be applied to finite strains thermo-elasto-visco-plasticity and thermo-visco-elasticity. In those cases, boundary-value problems (including conduction, transient thermal effects, and mixed boundary conditions) take the form of a saddle-point optimization problem. Approximate solutions to this problem can then be obtained by applying a Galerkin Finite Element approach, although more sophisticated hybrid approaches can be used to tackle volumetric locking for example. We will also discuss how this variational structure can be exploited to derive various monolithic and staggered algorithms to solve coupled field problems, as well as adaptive multi-mesh methods.

LECTURER_6
Lecturer:

Prof. Laurent Stainier
École Central de Nantes
Institut de Recherche en Génie Civil et Mécanique
Equipe Structures et Simulations
E-mail: Laurent.Stainier@ec-nantes.fr
Tel : +33 (0)2 40 37 25 86
Fax: +33 (0)2 40 37 25 73
http: www.ec-nantes.fr/stainier-laurent-93780.kjsp

Mini-CV:

Laurent Stainier received his Ph.D. degree in 1996 from the University of Liège, in Belgium (where he also obtained his Master in Aerospace Engineering in 1992). After post-doctoral positions at University of California, San Diego, and California Institute of Technology, Pasadena, he became in 2001 FNRS Research Associate (Chercheur Qualifié) with the Belgian National Science Foundation, at the Aerospace and Mechanics department of the School of Engineering, University of Liège. In 2008, he took a Professor position at Ecole Centrale Nantes, in France, and joined the Research Institute of Civil and Mechanical Engineering (GeM). Since 2012, he serves as Director of GeM. Prof. Stainier research interests revolve around mathematical modelling and numerical simulation of non-linear, dissipative, physical systems encountered in mechanical and aerospace engineering, with a specific emphasis on multi-scale, multi-physics aspects.