Welcome to TCN CAE 2005 Conference

TCN CAE 2005

Invited speakers

The invited plenary speakers of the conference include (as of July 15th, 2005)

Xiao-Bo Chen, Departement de Recherche, Bureau Veritas S.A., Courbevoie, France

XB Chen is currently Head of Section Hydrodynamics in Research Department of Bureau Veritas, Professorship of Harbin Engineering University. He was graduated from Tsinghua University and awarded as Distinguished Graduating Student of TsingHua. He received Ph.D in hydrodynamics at Ecole Centrale de Nantes (France). Since 1991, he has worked on theoretical and applied wave-structure interactions. The principal contributor of the Seakeeping software HydroStar and author of more than 60 technical papers, he has also worked in IACS, ISO and ISSC organisations and involved in a number of EU R&D projects. Research interests: water waves, offshore and ship hydrodynamics.

Some applications of mathematical analysis in the study of water wave interaction with floating bodies

The interaction of water waves and floating bodies is modeled within the classical potential theory and analyzed by using the classical Fourier transform and Green's theorems. New results on the wave patterns generate by a pulsating source animated with a forward speed providing important insights into the complex features of water waves will be presented, as well as the recent innovative developments of wave diffraction and radiation including the new formulations of second-order wave loads, the multi-body interaction and the effect of liquid motion in tanks. These new developments have extended applications in offshore industry for the advanced hydrodynamic studies on LNG Carriers, FPSOs, barges, semi-submersibles and TLPs. Some results issued from several EC-funded and national research projects as well as those obtained in the classification activities of Bureau Veritas will be included.

Carlo Bottasso, Dipartimento di Ingegneria Aerospaziale, Politecnico di Milano

Carlo L. Bottasso is an Associate Professor of Flight Mechanics at the Dipartimento di Ingegneria Aerospaziale of the Politecnico di Milano, Milano, Italy, and an Adjunct Associate Professor of Aerospace Engineering at the D. Guggenheim School of Aerospace Engineering of the Georgia Institute of Technology, Atlanta, GA, USA. He received a Ph.D. in Aerospace Engineering from the Politecnico di Milano in 1993 and did Post-Doctoral work at the Rensselaer Polytechnic Institute. His current research interests are in the area of aero-servo-elasticity and flight mechanics of rotorcraft, multibody dynamics and model based control. He is the co-author of more than 60 peer-reviewed journal articles and 100 publications in conference proceedings.

Virtual Pilots for Maneuver Modeling and Simulation in Vehicle Dynamics

Sophisticated models of complex vehicle systems are nowadays available to industry through a variety of advanced multidisciplinary software tools. For example, fixed and rotary wing vehicle models account for interacting structural and aerodynamic fields, often coupled with hydraulic and electromechanical subsystem models and with control laws. Similarly, in the automotive field, car models account for detailed representations of the vehicle chassis, steering, engine and drive-line subsystems, including tire and road contact models.
The quick pace of the evolution of comprehensive vehicle models is accompanied by a similarly growing need to broaden the range of problems that can be addressed by simulation. For example, the forward dynamics simulation of the vehicle model response to assigned control inputs, a standard problem handled by all dynamic simulation environments, does not always answer all needs: a design engineer might in fact be more interested in flying a minimum time turn with a virtual model of a helicopter while not exceeding a maximum allowable load factor, or in driving a virtual car model on a wet turn with maximum exit speed without going into a spin.
To address these maneuvering modeling and simulation needs, virtual pilots are being developed to steer virtual vehicle models. In this lecture we describe model-based virtual pilots that, on the basis of a formal maneuver description, first plan an optimal path and then track it by driving the vehicle model along it. Both planning and tracking pilots are based on adaptive reduced models of the system and have the ability to learn, and hence improve their driving performance, as they steer the vehicle. In particular, we illustrate the use of this emerging technology in the context of the simulation of emergency take-off and landing procedures of rotorcraft.

Daniel Benoualid, Centre de Recherche Hutchinson S.A., Chalette-sur-Loing, France

Founded in 1855, Hutchinson company is today one of world leaders for industrial rubber processing with three main markets: Automotive Industry (Fluid Transfer, Car Body sealing, Transmission & Antivibration systems), Industry (products for aerospace, railways, Defense and Security…) and Consumer products (Baby & health care, Sponges…) In 1990, Daniel Benoualid has joined the Hutchinson group to develop and conduct the Scientific Computing Department at the Corporate Research Centre. With his team, he has developed expertise and specific simulation tools, promoting also, all around the group, in an industrial environment, the use of numerical simulation close to product and processes development. From 2005, Daniel Benoualid is acting as the Director of the Corporate Research Centre together with his numerical simulation activity.

Utilisation of HPC in an industrial environment

The role of FE simulations in engineering is not to be demonstrated any more. Big companies, and some of SME’s operating in highly competitive markets, have adopted simulations as a standard tool in the design process. However, classic approaches involving sequential codes and simple work-stations are not appropriate to tackle the challenging modelling problems in real industrial environment. High Performance Computing tools and, in particular, robust and general parallel FE simulations codes are required, as well as an inexpensive hardware platforms.
Hutchinson is a French industrial company producing rubber components for the automotive, aerospace, construction and consumer product industries. Behaviour of rubber components is highly non-linear, mainly due to large deformations and complex boundary conditions, in particular contact with friction. In fact, many of the new company products could not be designed, optimised, and manufactured in time, or even at all, without using FE simulation extensively. As a part of Hutchinson Research Centre in Chalette sur Loing, the Scientific Computing Department develops and uses an internal parallel simulation code base on the FETI domain decomposition technique. This technique is used to solve static and dynamic problems in time domain, and vibro-acoustic problems in frequency domain. The parallel computing methods have been developed in a closed collaboration with ONERA (a French research institution for aeronautical and aerospace application) in the framework of several EC-sponsored research projects. Currently, a cluster of more than 200 processors is used, under both Linux and Windows operating systems. In particular, using HPC, Hutchinson engineers cans simulate not only separate components, but also sub-systems and complete systems.
In this presentation, utilisation of HPC in Hutchinson will be described, together with examples of real industrial applications.

Peter Wriggers, Department of Civil Engineering and Geodesic Sciences, University of Hannover, Germany

Currently professor of Computational Mechanics at the University of Hannover. Past Dean of the Department of Civil Engineering, from 1990 to 1998 professor for Mechanics at TU Darmstadt. Vice President of the German Association for Computational Mechanics, Board Member of the German Association of Mechanics and Mathematics. Member of the Academy of Sciences and Literature in Mainz and Member of acatech, the German Convent of Engineering Science.

Editor-in-Chief of Computational Mechanics. Member of the editorial board of International Journal for Numerical Methods in Engineering, Computer Methods in Applied Mechanics and Engineering, Computer and Structures, Engineering Computations, Archives of Computational Methods in Engineering, Engineering with Computers.

Graduate from the University of Hannover, Germany, Ph D at the University of Hannover and habilitation at the University of Hannover. Research topics : contact problems, coupled problems, computational problems in material sciences, and nonlinear finite element methods for finite deformations.

Characterization and Simulation of Microheterogeneous Materials

An understanding of the constitutive behaviour of materials which are heterogeneous in the sense that particles are imbedded in matrix, such as concrete, can be achieved by experimental investigations, theoretical analysis and by numerical simulations.
While analytical results are of limited use and can only be applied to linear elastic problems, experimental techniques have been applied successfully since several decades for the determination of constitutive laws. Nowadays the use of numerical simulation techniques is quite young. This is due to the fact that the underlying three-dimensional models are quite complex and need a considerable computer power which only in the last decade has reached a state that is sufficient. The advantage of the numerical simulation is that on can look inside a probe during the loading process and also is able resolve fast processes on a different time scale.
A the multi-scale modelling process can be used for different materials one example is concrete. Within this process different three-dimensional mechanical models are applied on different length scales in order to describe the constitutive behaviour on such scale. The 3d-models are called Representative Volume Elements (RVE). Often the material on a specific length scale consists of different phases which have to be taken into account in order to characterize the material with sufficient accuracy. The RVE is then subjected to different loading conditions which lead to a material response. Based on these results a homogenization process can be initiated to describe the material behaviour averaged over the RVE. The resulting homogenized constitutive equation is then applied within the next scale to model the constituents of the RVE belonging to that scale.
In this contribution an overview with regard to the theoretical and numerical treatment of multi-scale modelling is presented and the developed methods are applied to a real heterogeneous material and compared with experimental data.

Patrick LeTallec, Département de Mécanique Ecole Polytechnique, Palaiseau, France

Currently professor of Computational Mechanics at Ecole Polytechnique. Past Vice President for Academic Research at Ecole Polytechnique, past President of the French Society of Applied and Industrial Mathematics, and professor of Applied Mathematics at University of Paris Dauphine from 1988 to 1999. Member of the editorial board of International Journal for Numerical Methods in Engineering, Computer Methods in Applied Mechanics and Engineering, Computer and Structures.

Graduate from Ecole Polytechnique in Paris France, Ph D at the University of Texas at Austin, and thèse d'état at University of Paris. Research topics : domain decomposition techniques, coupled problems, and structural problems in large deformation.

Computational challenges and multiscale modelling in structural mechanics

Patrick LeTallec, Département de Mécanique Ecole Polytechnique, Palaiseau, France
Numerical modelling has been very popular and successful in the past years, and is now a key part in the design loop. Recent emphasis has been put on biomechanics, dynamics, or in model coupling, as we will illustrate during the conference. But this success is challenged in four directions :
- relevance, in particular because the model limitations of the different commercial codes are not well understood by the users,
- complexity, because of technological developments involving local refinements of the design,
- reliability because of the uncertainty on key data,
- in depth understanding of the local mechanisms governing the response of the system under study.
Multiscale modelling is a good answer to these challenges. The talk will try to develop this point by giving an up-to-date -description of the macroscopic approach, explaining the potential gain brought by the understanding of the microscopic scale , and presenting the forthcoming challenge of matching macroscopic modelling and micromechanics. One of them is to derive in reasonable time reliable macroscopic averages of the response of the underlying microstructures. The talk will be illustrated by examples from industry, biology or material science.

Christian Bucher, Institute of Structural Mechanics, Bauhaus-University Weimar, Germany

Christian Bucher is a full professor of Structural Mechanics in the faculty of Civil Engineering of Bauhaus-University Weimar, Germany since 1994.
He received a diploma in Civil Engineering and a PhD in Structural Mechanics from the University of Innsbruck, Austria in 1982 and 1986, respectively.
He has been visiting professor with the universities of Boca Raton/Florida, Tokyo/Japan, Boulder/Colorado and Waterloo/Canada.
Research interests focus on stochastic structural mechanics with emphasis on dynamics, system identification, and reliability-based optimization. He is the co-author of more than 140 publications.

Robustness Analysis in Structural Optimization

Virtual Product development aims at establishing the fundamental behaviour of new or updated products before carrying our real hardware tests. In this context, feasibility decisions can be based on purely computational results. It is therefore essential that the main features of the product are captured with sufficient accuracy. In many cases, the virtual development is applied to structures or structural elements which should perform in an optimal way.
Structural optimization typically aims at high performance levels for a clearly specified set of conditions. Unfortunately, this goal can usually be achieved only by sacrificing robustness of the design. This implies a high sensitivity with respect to unforeseen stochastic situations or unavoidable random manufacturing tolerances. The influence of such imperfections on the response can be analyzed e.g. by probability based methods. This type of analysis is frequently termed robustness analysis. While robustness analysis can be performed quite independently it appears to be most beneficial to carry out optimization such that robustness requirements are automatically included in the optimization process. This can be achieved by introducing additional constraint conditions or appropriate modifications of the objective function.
An example for such a design concept is reliability-based optimization based on the notion of the failure probability. This is most appropriate for high-risk structures such as e.g. power-generating facilites. Alternatively, simpler stochastic measures such as variances or standard deviations might be more appropriate for the design of low-risk structural elements which are frequently found e.g. in the automotive industry. Currently, this variance-based type of approach appears to be tractable from the viewpoint of the numerical effort required.
The lecture discusses the basic requirement for robust optimization and attempts to outline pros and cons of different approaches to the solution of this problem.
Several numerical examples including industrial applications highlight the potential of such an approach to optimization.

Hubert Lobo, Matereality LLC, Ithaca, NY, USA

Hubert Lobo is the founder of DatapointLabs, a testing laboratory specialized in generating representative properties for CAE. He has been working in the area of material characterization and modeling for CAE since 1986, first with the Cornell Injection Molding Program and then with C-Mold, where he developed their material modeling test protocols.
His company, DatapointLabs is the primary source of material properties for structural analysis, process simulations such as injection molding, extrusion, and blow molding offering analysis ready material models for over a dozen major CAE programs. Matereality is his new company bringing tools to manage the growing pool of material property data available for virtual product development.
In 2002, the Society of Plastics Engineers honored Mr. Lobo as one of its youngest Fellows, recognizing his pioneering work in quantification of material behavior for CAE. Mr. Lobo actively participates in testing standardization efforts within the US and internationally at ISO. He has a Masters degree in Engineering from Cornell University.

Quantification of Material Behavior for Simulation

Material modeling is a key aspect of successful CAE with non-linear materials such as plastics, rubber and foams. After the right material model is selected, the properties of these complex materials need to be determined with precision and in conformance with the needs of the CAE application. Properties needed vary depending on the task at hand and the material model parameters are often difficult to obtain. It is therefore quite a complex matter to prepare material models for CAE.
In the finite element analysis (FEA) of plastics, analysts are now migrating to non-linear and multi-physics applications. Successful implementation of these simulations require the understanding and use of a variety of material models for elastic-plastic behavior, rate dependency, temperature dependency, creep, stress relaxation and fatigue. We examine these effects and look at limitations of existing material models for plastics.
FEA of rubber is typically handled using hyperelastic models. The challenge here is to obtain the right kind of data and to transform it into one of several available material models. Issues discussed will include the Mullens effect, strain range effects and problems related to visco-elastic behavior.
Foam modeling is complex because each class of foams has a particular kind of behavior. The discussion will focus on the different types of foams and the kind of material models most appropriate for each. The particular case of foams in impact situations, including non-linear rate dependency, will be discussed.
Process simulation including injection mold analysis, thermoforming and extrusion simulation require sophisticated are non-linear, non-isothermal material models. Pros and cons of some of these modeling strategies will be presented.
Finally, we consider the challenges that need to be addressed in the future as simulation progresses to higher levels of sophistication: the need for multi-dimensional material modeling and most crucially, the desire to digitally store this complex expensive information so it is always available to those who need it and not lost over time.

Solke Bruin,Department of Chemical Engineering and Chemistry , Eindhoven University of Technology, The Netherlands

Born in Indonesia 1940 in Pematang Siantar (Sumatera). Graduated from the Agricultural University in the Netherlands in 1965. Finalized his PhD in 1969 (cum laude). Postdocs at the Technical University Eindhoven (Prof.Dr.Ir. H.A.C. Thijssen) and the USDA s Western Regional Research Center in Berkeley (California, USA) in 1970.
Joined the Royal Shell Research Laboratories in Amsterdam (KSLA) in January 1971 in the Chemical Engineering Group (Equipment Engineering) and worked on distillation, liquid extraction and other separation processes.
In 1974 appointed full-professor in Process Engineering at the Agricultural University in the Netherlands. Teaching of (bio)chemical engineering principles to students in food technology, environmental engineering and biotechnology. Research activities included drying of food materials (calculation methods for dehydration with strongly concentration dependent diffusion coefficients of water), adsorption of chemicals from aqueous solutions to active carbon and modeling post harvest physiology during storage of agricultural products. He was Department Head of the Food Science department from 1977 to end 1980, when he joined Unilever Research in Vlaardingen (The Netherlands) as divisional manager of the Process Engineering Group.
In Unilever Research he fulfilled various senior positions in research management ( strategic research programme formulation and programme execution). In his latest position he was responsible for Exploratory Research in the Foods Processing area with research units in both the Colworth House- (UK) and Vlaardingen (NL) Research Laboratories and member of the Management Committee of Unilever Research Vlaardingen in the Netherlands. He retired from Unilever in May 2001. In September 2003 he was appointed extra-ordinary  Hoogewerff Foundation professor at the Technical University Eindhoven in the Netherlands to start a new chair  Product-driven Process Technology in the Chemical Engineering Faculty of that University.

Computer modeling in product design: opportunities and limitations

Solke Bruin,Department of Chemical Engineering and Chemistry , Eindhoven University of Technology, The Netherlands
A brief overview is given of the contributions computer modeling of manufacturing processes for Fast Moving Consumer Products (FMCP) has given to increase product and process understanding. In addition to modeling of the manufacturing process steps themselves, an important contribution to optimum product design is the modeling of what happens during use of consumer products is stressed in the paper. Some examples will be given related to food products, household products personal care products.
Some lines of future contributions are given like trends in description of phase equilibria, linking time and length scales, food informatics, and some computational flow dynamics (CFD) examples
Despite the rapid and sustained increase in computer power there are a number of limitations and rate limiting factors one needs to be aware of in particular in foodprocess engineering. These will also be discussed briefly in the paper.

Maurizio Angelillo, laboratorio di biomeccanica, dipartimento di Ingegneria Civile, Università di Salerno

Maurizio Angelillo is full professor of  Scienza delle costruzioni at the University of Salerno, school of Engineering since 2002.
He obtained his degree in 1977 at the University of Naples and the M.S. in mechanics in 1983 at the University of Minnesota, where he also was visiting professor in 1996 and 1997.
My field of interest is the modeling, analysis and computation of solid materials and structures with particular emphasis on nonlinear and failure processes in solids. My current research interests range from mesh adaptation and simulation of fracture in structures to biomechanics. Some of the research areas of application which are of particular interest to my group are:
- New numerical tools for description of masonry structural behavior.
- Minimization of nonconvex forms of energy over variable triangulations for simulation of fracture propagation.
- The mechanical description of the heart fiber structure.
- The study of impact loads on the brain material.

Numerical models of the human head for the prediction of injury and death by impact

Biomechanics research at the University of Salerno, in the field of respiratory mechanics, began in 1988. Recently (2002) a project concerning the study of head injury has been started. The project is conducted jointly with the Department of Neurosurgery of the  Università Cattolica in Rome. The purpose of the project is ultimately to improve protection against head injury, our specific interest being directed toward an improved design of motorcycle helmets.
One of the activity of these years has been the development of a computer model capable of predicting the extent and the location of the injury in the brain due to different kinds of impact.
In the construction of such model there are several difficulties.
The most crucial step was represented by the correct mechanical simulation of the different bulk and interface materials composing the head. Such materials are very different in stiffness, some of them are highly non linear in their response and can behave as solids or fluids depending on the loading rate.
The second difficult was to replicate as accurately as possible the geometrical complexity of the brain architecture. Indeed a reasonable compromise must be reached in order to represent with sufficient detail the regional and material differences observed in the head materials, still producing a 3D model that numerical codes could digest.
Because of the large size of the model, a fast computer is needed to solve the problem in a reasonable amount of time: reducing the computational effort is important since automotive safety engineers need to analyze a large number of head impacts thus making CPU times of over ten hours unfeasible.
Our model uses an existing software program which solves problems in stress analysis under fast dynamics conditions using the finite element method. The model consists of a unique data set to simulate head impact. The data set consists of a geometric mesh obtained through surface generation from MRI sections and subsequent 3D meshing. The mesh describes all of the components of the head: the scalp, the skull, the cerebral spinal fluid, the membranes, the brain hemispheres, the falx separating the two hemispheres, the ventricles.
Testing and validation of the program is currently underway. So far a large variety of impact situations, some of which are hard to replicate experimentally, has been simulated showing reasonable accuracy in predicting the location and severity of injury when the model results are compared with data from MRI of patients for which the approximate impact conditions are known