Modeling materials at various scales: The solidification example

Michel Rappaz
Ecole Polytechnique Fédérale de Lausanne, Switzerland

Mackiw Lecturein Metallurgy

3:30pm - April 5, 2012
ETLC 1-001

Abstract:

Solidification of bronze components is one among the oldest technological activities of humankind, starting nearly 5'000 years ago. This "breakthrough" contributed to the development of the various civilizations. More recently in the 1950's, multi-scale solidification modeling also played a key role in the development of numerical methods applied to materials problems. The series of proceedings of the International Conferences Modelling of Casting and Welding started in 1980 (renamed Modelling of Casting, Welding and Advanced

Solidification Processing, MCWASP in 1990) shows very well the remarkable progresses that have been made in this area, thanks to Moore's law, to the optimization of standard numerical methods and to the development of new modelling tools.

Taking solidification as an example, the present lecture will show how information can be transferred among various scales, from the atomistic scale to the microstructure level, from the microstructure to the grain level and form the grain to the scale of a whole component. For example, the calculation of solid-liquid interfacial energies using Molecular Dynamics (MD) can be used in Phase-Field (PF) modelling of dendrite formation. Knowing the dendrite growth kinetics, calculations of grain growth, grain competition and texture development can be made using Cellular Automata (CA). Finally, CA can be coupled with Finite Element (FEM) or Finite Difference Methods (FDM) for the prediction of the grain structure of a whole casting or weld, taking into account heat and mass transfer.

Although modelling is nowadays a fantastic tool for materials scientists and engineers, this lecture will also emphasize that it must be based on sound experimental data and correlated with well-controlled experiments. To quote Samuel Karlin, mathematician (1924-2007), it must be kept in mind that "the purpose of models is not to fit the data, but to sharpen the questions".

Biography:

After a PhD in solid state physics (1979) at the Ecole Polytechnique Fédérale de Lausanne (EPFL) and a post-doc at Oak Ridge National Laboratory, Michel Rappaz joined the Laboratory of Physical Metallurgy of EPFL to start an activity on simulation of solidification in 1984. He was nominated Adjunct Professor in 1990 and Full Professor in 2003, time at which he took the lead of the laboratory, renamed then Computational Materials Laboratory. The main axis of his research is the connection between macroscopic aspects of solidification (heat and mass transfer) with microscopic aspects associated with microstructure and defect formation. Beside experimental investigations and validations, his laboratory (about 20 researchers/PhD students) has developed several new physical modelling tools (cellular automata for grain structure formation, granular models of mushy zone, inverse methods, porosity and hot tearing models, etc). Many of these developments are commercialised by a spin-off company founded by the laboratory in 1991 (Calcom SA), which has joined the French ESI group at the end of 2002. He initiated in 1992 an annual postgraduate course on solidification which has been attended by more than 600 participants from all over the world.

Michel Rappaz has received several awards: the Latsis price in 1990, the Mathweson award of TMS in 1994 and 1997, the Koerber foundation award with Profs Y. Brechet and M. Asbby in 1996, the Sainte-Claire Deville medal of the French Metallurgical Society, the Bruce Chalmers Award of TMS in 2002, the Mc Donald Memorial Lecture award of Canada in 2005. He is a highly-cited author of ISI, a fellow of ASM and IOP, a Honorary Professor of Queensland University and has co-authored about 250 papers, two books and several proceedings.