ECCOMAS'96 - New Trends in Scientific Computing

by Armel Guillet

The recent ECCOMAS (European Community on Computational Methods in Applied Sciences) conference, held in Paris 9-13 September 1996, was an undisputed success. The conference brought together nearly 700 participants, from 43 countries, who were able to take stock in the latest developments in scientific computation and its applications in telecommunications, transport, energy, environment, biology, chemistry, aerospace, etc.

A number of new trends which were highlighted included:

• Major developments in work on multiphase flows and free surface flows, the main applications involving combustion and the environment. Free surface flows make it possible, for example, to calculate sedimentation rates in bodies of water.
• The modelling of complex procedures in materials science represents another significant development which, though still requiring validation, has already yielded promising results. Modelling the impact of very high-speed projectiles on certain structures, or taking into account the crystalline structure of aluminium in order to define more accurately its behaviour after millions of loadings is thus becoming possible.
• The development of mathematical models and tools to describe multiscale phenomena. Such tools are used to describe heterogeneous or compound materials and plate and shell structures. The linking of numerical modelling with process optimization and control is becoming more widespread. Added to the improvements in modelling and optimization tools, this new approach makes it possible to go beyond the field of small laboratory mock-ups and to develop full-scale prototypes (active control of real flows, airfoil optimization, etc.). The problem of handling increasing complexity is solved by the appearance of robust algorithms, evolution strategies for example. These represent a key factor in competitiveness.

Two fields remaining particularly active include: the study of solvers and the development of numerical models. Solvers, used to approximate and solve equations, have already found their place in industry and lie at the heart of software packages used in design departments. However, problems of accuracy, validation of results or utilization on moving or adapted meshes remain critical. The sessions addressing these subjects were of great interest well attended.

Innovative numerical models are also being developed that allow distributed simulation with HPCN (High Performance Computer Network) tools which involve coarse grain parallel computers. Traditional applications, such as external aerodynamics and structure calculation, are increasingly considered in a multidisciplinary context both in research and development. The combined fluid-structures-electromagnetism approach in the aeronautics sector is one example.

Presentations and discussions held during the five days generated stronger ties between research and industry, which is the underlying basis of the conference. Opening ceremony participants Bernard Bigot (DGRT), Jacques-Louis Lions (Academy of Sciences), Bernard Larrouturou (President of INRIA), Guy Aubert (General Director of CNRS), David Gardner (British Aerospace), Arthuro Gareia Arroyo and Thierry van der Pyl (European Commission), Jacques Périaux (Dassault) were in agreement on the need for cooperation between research and industry. According to Michel Giraud, president of the Ile de France region, research is the key to competitiveness and future job creation. In this context, Jacques-Louis Lions highlighted, numerical modelling is a vital element. Research demands both medium-and long-term investment, together with industry-oriented develop-ments. Bernard Bigot, Serge Dassault and representatives of the European Commission also stressed this point.

While applied mechanics still have an important place in the industrial sector and play a major role in the field of application, scientific computation continues to have a growing impact on a number of new sectors (materials sciences, various chemistry applications, biology, immunology, image processing, etc.). The next Eccomas conferences, which will be held in Athens in 1998 and Barcelona in the year 2000, will continue to follow these trends.

Please contact:
Patrick Le Tallec - INRIA
Tel : +33 1 39 63 54 83
E-mail : Patrick.Le_Tallec@inria.fr