Third M.I.T. Conference on
Computational Fluid and Solid Mechanics
June 14–17, 2005


As mentioned previously in the Prefaces of the two earlier MIT Conferences, there are now most exciting and important research tasks in computational mechanics. In the Preface of the Second MIT Conference, these research tasks are described and detailed by summarizing eight key challenges. Of course, these challenges are for the next decades and hence the Preface of the Second Conference is still very much applicable and is repeated here in the following text.

The exciting and most important research tasks have come about because it appears possible to reach a new level of mathematical modeling and numerical solution that will lead to a much deeper understanding of nature and to great improvements in engineering designs.

This new level of mathematical modeling and numerical solution does not merely involve the analysis of a single medium but must encompass the solution of multiphysics problems involving fluids, solids, their interactions, chemical and electro-magnetic effects; must involve multi-scale phenomena from the molecular to the macroscopic scales; must include uncertainties in the given data and solution results; and, in engineering, must focus on the optimization of designs for the complete life spans of the systems.

Based on these thoughts, we can identify the following eight key challenges for research and development in computational mechanics.1

Challenge 1. The automatic solution of mathematical models. Many advances have already been made in the development of numerical procedures, and notably the finite element method, to automatically solve mathematical models for a given accuracy. However, there are many further advances needed in the meshing algorithms, in discretization schemes to have uniformly optimal schemes, in error measures, and so on, including actual practical implementations utilizing advances in hardware. A further step is to automatically create mathematical models, hierarchically, and establish error measures on these models.

Challenge 2. Effective numerical schemes for fluid flows. Numerous publications exist on the numerical solution of fluid flows, but the numerical methods proposed are far from satisfactory. ‘Ideal’ solution schemes would be much more predictive, reliable and effective. Clearly, major advances are still possible.

Challenge 3. The development of an effective mesh-free numerical solution method. While much research effort has been expended on the development of meshless methods, only a few proposed techniques are truly meshless and these are not yet sufficiently effective. A reliable, general and efficient mesh-free method for solids and fluids will greatly advance the field of analysis and surely such a method can be developed.

Challenge 4. The development of numerical procedures for multiphysics problems. One major area is given by fluid flows, including heat transfer, chemical and electromagnetic effects, fully coupled to structures. Of course, there are in addition many other multiphysics areas involving thermo-mechanical, electro-mechanical, chemical and other coupling effects. Advances have been made for simulations in these fields but significant further progress can be accomplished.

Challenge 5. The development of numerical procedures for multi-scale problems. Many devices and phenomena in engineering and the sciences involve multiple scales. The spanning of scales in analyses of engineering designs, notably those using nano-technology, in analyses of bio-medical applications, material modeling, and problems in the earth sciences, to name just a few, provides an exciting challenge.

Challenge 6. The modeling of uncertainties. The purpose of an analysis is to model nature, as represented in a new design or an already existing system. However, invariably there are uncertainties and these ideally would directly be included in many analyses. This will surely be possible.

Challenge 7. The analysis of complete life cycles of systems. At present, largely, only the initial design of a system is analyzed and optimized. There is a need to extensively develop ‘virtual laboratories’ in which complete life cycles of systems are optimized. Such simulations can be used to greatly advance engineering designs.

Challenge 8. Education. The powerful tools for analysis are only of value if they are used with sound engineering and scientific judgment. This judgment must be created by a strong, basic and exciting education in the Universities and ongoing, life-long education in practice.

The objectives of the bi-annual MIT Conferences on Computational Fluid and Solid Mechanics are to meet and build upon these challenges by bringing together the leading researchers and practitioners of computational mechanics in the world, from Academia and Industry, and by helping young researchers to enter the field.

We are once again very grateful to the sponsors of the Conference for providing the financial and intellectual support to attract speakers and bring together Industry and Academia. In the spirit of helping young researchers, as for the two earlier MIT Conferences, fellowships have been awarded to about 100 young researchers for travel, lodging, and Conference expenses, and, in addition, Conference fees have been waived for all students.

The papers presented at the Conference and published in this book represent, in various areas, the state of the art in the field. As for the two earlier MIT Conferences, in addition to the presentations based on the papers published in this book, state-of-the-art presentations based on short one-page Abstracts will also be given. The presentations at the Conference have been largely attracted by the Session Organizers. We are grateful for their efforts and look forward to our continued cooperation to fulfill the very exciting objectives of the MIT Conferences on Computational Fluid and Solid Mechanics.


1 Taken from ‘Key Challenges at 40 Years After’, a presentation by K.J. Bathe at the UC Berkeley Symposium in honour of R.W. Clough and J. Penzien, May 2002. R.W. Clough, a doctoral Graduate of MIT and Professor of UC Berkeley, coined the name ‘finite element method’ about 40 years ago.