Edificio C1, Campus Norte UPC

Email: onate@cimne.upc.edu; Tel: 34 932 057 016

Mr. Pere-Andreu Ubach de Fuentes, R&D Project Strategies

Email: ubach@cimne.upc.edu; Tel: 34 934 017 399

CIMNE was created in 1987 under the auspices of UNESCO by the autonomous government of Catalonia. It is the second oldest independent research center in Catalonia. CIMNE has its own juridical status as a consortium between the government of Catalonia and the University Polytechnic of Catalonia (UPC), and now Spain’s federal government will join the consortium. The headquarters of CIMNE are located at the department of Civil Engineering of UPC. CIMNE has also offices for training in 15 classrooms distributed around the world. CIMNE’s mission is three-fold: (1) Research, (2) Training activities, and (3) Technology transfer. Currently, more than 90% of its funds come from external sources, and its annual funding level is close to €10 million. There are currently 190 employees, and with the new national government funding, 40 more employees will be added. Specifically, about 45 employees are permanent research staff, 15 are administrative personnel, and the rest are researchers in training.

CIMNE has been one of the main research centers focusing on the development and application of finite element methods (FEM) in computational mechanics applications—in particular, stabilized FEM. A unique method developed at CIMNE is a hybrid FEM/particle approach that allows modeling of free surfaces and moving boundaries more easily; impressive results were shown for marine engineering applications. While the long-term focus has been structural and fluid mechanics, more recent work has been on electromagnetics, design and optimization, food engineering, and other multidisciplinary problems. The focus and strength of CIMNE researchers are multiphysics rather than multiscale applications. There is no particular emphasis on parallelization of their codes as they can currently fit up to 12 million FEM on a high-end PC; all of the codes used at CIMNE were developed in-house, including pre- and postprocessing, and many of their codes are now commercial software products. For example, GiD (www.gidhome.com) is a commercial PC-based pre/postprocessor with a very wide user base in industry and academia, and it is interfaced with many commercial codes, such as NASTRAN.

CIMNE is also focusing on verification and validation research, and currently it funds an experiment that will quantify forces on a concrete block by flow motion. This will be used as a benchmark for validating flow-structure interactions models that it has been developing. However, there is no particular emphasis on uncertainty quantification based on stochastic modeling techniques.

CIMNE has a very strong interactions and links with industry, with about half of its research and development projects funded by industry, e.g., aerospace, marine, civil, mechanical, biomedical, and metal forming companies. CIMNE researchers bid for about 150 proposals per year and in their 20 years’ history have had 141 European Commission projects, 217 national projects, and 452 industry projects. CIMNE is a shareholder in three companies, specializing in marine engineering, civil and structural engineering, and aeronautics and space applications.

CIMNE organizes courses and seminars related to the theory and application of numerical methods in engineering. The attendees are recent university graduates and also professionals. Since 1987 CIMNE has organized about 100 courses and 300 seminars. A new program is the Erasmus Mundus Master course, which is an international course for masters in computational mechanics for non-European students. In its first year at the time of the WTEC team’s visit (2008), the course had 30 students. One of the requirements is that the students spend equal time at least in two of the four different universities involved in the course (Barcelona, Stuttgart, Swansea, and Nantes).

CIMNE has developed a Web environment for distance learning education via the Internet. The Virtual Center of Continuing Education of CIMNE gathers information early on in a course to facilitate the registration process. Teachers can also follow the students and carry out different tutorials and exercises. This center hosts the Master Course in Numerical Methods in Engineering and other postgraduate courses of CIMNE (http://www.cimne.upc.edu/cdl).

CIMNE has been a leader in disseminating information on computational mechanics by organizing more than 80 national and international conferences; 13 of these conferences took place in 2007. CIMNE also publishes books, journals, monographs, scientific computing reports, and educational software related to theory and applications of numerical methods in engineering. In its first 20 years, CIMNE published 101 books, 15 educational software packages, 296 research reports, and 492 technical reports. CIMNE researchers publish about 60 papers per year.

CIMNE classrooms are physical spaces for cooperation in education, research, and technology development, created jointly by CIMNE and several other universities. They are located in Barcelona but also in other cities in Spain and around the world, including Mexico, Argentina, Colombia, Cuba, Chile, Brazil, Venezuela, and Iran.

Conclusions

CIMNE is a unique research and development center that has done a lot in its 20-year history to promote and foster both fundamental and technological advances associated with computational methods for engineering problems. Although its current emphasis is on multiphysics applications, CIMNE researchers still publish papers on fundamental mathematical topics, e.g., new formulations and error estimation and bounds in FEM, but also on optimization and design of complex systems, and they are branching out into new applications with great potential impact, e.g., food engineering and decision support systems. CIMNE’s education history and recent development of a four-campus MS program for international students as well as its distance learning and training activities are impressive and are having great impact in Spain, in Europe, and around the world.

Site: Ecole Polytechnique Fédérale de Lausanne (EPFL)
Institute of Analysis and Scientific Computing (IACS)

Department of Modeling and Scientific Computing (CMCS)

Department of Mathematics

SB, IACS-CMCS, Station 8

CH-1015 Lausanne, Switzerland

http://iacs.epfl.ch/index_e.html
(Remote site report) Politecnico di Milano (Polimi)
Laboratory for Modeling and Scientific Computing (MOX)

p.za Leonardo da Vinci 32

20133 Milan, Italy

http://mox.polimi.it/
Date Visited: February 23, 2008.
WTEC Attendees: S. Glotzer (report author), L. Petzold, C. Cooper, J. Warren
Hosts: Professor Alfio Quarteroni, Professor and Chair of Modeling and Scientific Computing at the Institute of Analysis and Scientific Computing of EPFL;
Professor of Numerical Analysis, Polimi, and Scientific Director, MOX
Email: alfiio.quarteroni@epfl.ch

Dr. Simone Deparis (post-doc), EPFL

Marko Discacciati (post-doc), EPFL

This site report is both a direct report of the system at the Ecole Polytechnique Fédéral de Lausanne (EPFL) Department of Mathematics and a remote site report for the Politecnico di Milano Laboratory for Modeling and Scientific Computing (Modellistica e Calcolo Scientifico, or MOX) in Italy, as Professor Quateroni holds positions at both these institutions. He has been Chair of Modeling and Scientific Computation (CMCS) at EPFL/CMCS since 1998, and director of MOX in Milan since 2002 (as well as Professor of Numerical Analysis at Polimi since 1989). Prof. Quarteroni was kind enough to prepare overviews of both institutions for the WTEC team that visited EPFL.

Professor Quarteroni studied mathematics at the University of Pavia, Italy, and at University of Paris VI.¹¹ In 1986 he became full professor at the Catholic University of Brescia, later professor in mathematics at the University of Minnesota at Minneapolis, and professor in numerical analysis at Politecnico di Milano. He was appointed full professor at EPFL in 1998. His team carried out the aerodynamic and hydrodynamic simulations for the optimization of Alinghi, the Swiss sailing yacht that won the last two America's Cup races in 2003 and 2007.

In the Swiss system (such as EPFL), the position of chair comes with tenure, while other faculty with up to four-year contracts work with professors. Even these temporary positions are considered to be good jumping-off points for careers in SBE&S by those the WTEC team members talked to. It is typical that scientists obtain their PhDs abroad but return to become professors at EPFL. The international outlook at EPFL is reflected in the makeup of its student body and faculty, as foreign nationals comprise one-third of the undergraduate students, two-thirds of the graduate students, and three-fifths of the professors. Indeed, EPFL has been hiring full professors from the United States in increasing numbers. Professors can count on having research support for several funded graduate students indefinitely. The excellent infrastructure is a strong attraction for researchers.

In the Italian system, a hierarchy of professors all work in a more coordinated fashion, with direction from the full professor. MOX was founded in 2002 with three faculty members, a budget of $100,000, and 120 freshmen. It created a new, now very popular, BS/MS curriculum in mathematical engineering, focused one-third on basic engineering, one-third on fundamental math, and one-third on applied math. Students who work on an externally funded project can generally obtain higher salaries. Funding in MOX comes, roughly equally, from industry, public funding agencies, private funding agencies, and public European projects.

Eight years ago, all countries in Europe participated in what is known as the Bologna reform; it is now a universal 3+2 (BS/MS equivalent) program. This uniformity allows students mobility throughout Europe. As the program is so good, the job prospects are excellent for individuals with master’s degrees, which has the unintended consequence of making talented PhDs much more difficult to obtain.

The CMCS projects focus on fluid flow problems with numerous areas of application. Medicine and water sports are primary interests.

At CMCS there is a project for extracting geometry from raw data, to get a 3D “map” of the arterial system. These reconstructions are then turned into a finite element structure that can provide the basis for detailed simulations of blood flow.

Another project is attempting a classification of aneurisms by their geometrical structure. A posteriori error analysis is difficult in this problem because of the complexity of fluid-structure interaction. Differences in the metabolic responses of different patients add a lot of uncertainty to the model. It has been found that the flow field is very sensitive to the geometry.

Particular attention is paid to difficulties in attaching different length and time scales (the so-called multiscale problem) to the cardio-vascular system. They have developed a 3D flow model, while the major arteries and veins are sensibly modeling in 1D, and currently the capillary network is treated as 0D; interfacing these different dimensionalities has proved nontrivial. The models have substantial similarities to electronic circuit models (not surprisingly as they are governed by underlying current conservation laws).

Perhaps the most spectacular modeling presented by Professor Quarteroni concerned the CMCS work in mathematical models for sport, specifically “Multiphysics in America’s Cup,” a tour-de-force in fluid flow modeling. This was a formal collaboration with the Alinghi designers, with substantial proprietary information being shared by the designers. Commercial codes were used, initially Fluent, then CFX, because of the perceived superiority of the included turbulence models. The implementation required a supercomputer. (In order to iterate the design with the developers, a turnaround time of 24 hours was required.) The simulations had approximately 20 million elements, with 162 million unknowns in the solution. The work was rewarded with two America’s Cup victories.

The MOX program has as its mission to develop, analyze, and apply numerical models for engineering problems and real-life applications, enhance the scientific cooperation with other departments of Polimi, foster the collaboration with industrial partners as well as national and international institutions through funded research projects, and to organize short courses open to industrial and academic researchers. The MOX research group, based in the Polimi Mathematics Department, is comprised of 3 full professors, 10 assistant professors, 2 associates, 4 post-docs, 4 post-grad research assistants, and 13 PhD students. There are over a dozen industrial partners, with closely integrated collaborative research on a number of topics.

MOX maintains a number of both fundamental and applied research areas. Fundamental research is being pursued in (i) Adaptive numerical methods and error control for PDEs; (ii) control and shape optimization; (iii) numerical methods for fluid dynamics; (iv) level set methods; and (v) fixed, hybrid, and discontinuous finite elements for PDEs. Collaborators include Massachusetts Institute of Technology, Virginia Tech, EPFL, University of Texas Austin, University of Maryland, University of Santa Fé (Argentina), University of Trento (Italy), INRIA-Roquencourt (France), and Imperial College (UK).

MOX is pursuing a number of applications, many of which are adressing the state-of-the-art in continuum modeling, both in algorithmic development and in the application of HPC to real-world engineering applications. One application that was discussed, Geophysical modeling of the evolution of sedimentary basins, required calculations that involve large space and long time scales, necessitating adaptive meshes, lagrange tracking of interfaces, topological changes, and methods for non-Newtonian fluids. This application required the development of production codes for use by the oil industry.

Other applications of interest are in the life sciences, in particular, drug delivery from nano/microstructured materials and controlled drug delivery with application to drug-eluting stents (funded by Fondazione Cariplo and Istituto Italiano di Tecnologia, IIT). Additionally, a highly sophisticated analysis is being done of the complex fluid-dynamics of blood flow in the cardiovascular system.

An impressive collaboration with surgeons in London has evolved. The research involves multiscale simulations; it simulates the arterial vessels (not the veins yet) and is coupled with statistical analysis on hundreds of patients in hospitals. The surgeon can use the simulation to help decide which procedure to use. Professor Quarteroni observed that it, appropriately, takes a long effort to gain the surgeon’s trust. To initiate such a collaboration requires especially motivated doctors, and the modeler needs to prove his techniques. However, after a number of successes, he has the enviable position of having too many good opportunities, and must be very selective—lots of doctors want to work with him. Often the bridge between the modeler and the doctor is a bioengineer.

Students who are involved in industrial collaborations receive a lab course credit. Indeed, there is a large number of industrial collaborations:

• 
  ENI SpA, Divisione E&P: Modeling geological structure evolution in complex situations in 2D and 3D
  
• 
  Arena Italia SpA: Numerical models for the performance analysis of a swimmer
  
• 
  CTG Italcementi Group: Numerical simulation of an air-combustion gas mixer
  
• 
  Filippi Lido Srl: Mathematical and numerical models for the dynamics of rowing boats
  
• 
  RSI Tech: Mathematical modeling and numerical simulation of the propagation of a georadar signal in a hydraulic fracture
  
• 
  INDAM: Mathematical and numerical modeling of the electro-fluid-mechanics of the heart
  
• 
  MIUR–Cofin: Numerical models in fluid dynamics with application to the simulation of the cardiovascular system and the environment
  
• 
  IIT (Italian Institute of Technology): Nanobiotechnology models and methods for local drug delivery from nano/micro structured materials
  
• 
  Fondazione Cariplo: Development of mathematical and numerical models for optical devices based on nanostructured electrochemical produced materials; Mathematical modeling of drug release from drug eluting stents into the arterial wall.
  
• 
  Siemens, Fondazione Politecnico: Numerical and statistical methods for assessing the risk of rupture in cerebral aneurisms
  

The Polimi system allows for the development of a substantial code base. Multiple research foci, derived from industrial collaborations, medical research collaborations, as well as in-house motivated basic research interests, provide a critical mass of customers, and a large, motivated faculty team that maintains the code base. Substantial funding for these projects, including overhead, comes from industrial grants.

A number of software tools have been developed at MOX. They include, among several others,

• 
  LifeV (www.lifev.org), a finite element library in C++ developed in collaboration with EPFL and INRIA
  
• 
  Stratos, a finite element code for 3D free barotropic and baroclinic fields
  
• 
  Steam2D/Steam++/Steam3D, a suite of computer codes for the simulation of the evolution of sedimentary basins (non-Newtonian Stokes flow with interfaces)
  
• 
  Stratos-Wave, a simulation of boat dynamics by a variational inequality approach
  
• 
  Kimè, a simulation of the dynamics of a rowing scull
  
• 
  Glow1D/Glow2D, a simulation of the functioning of a glow plug for diesel engines
  
• 
  MM1, a section-averaged model for river flow and sediment transport developed in collaboration with CUDAM–University of Trento
  

MOX has managed the difficult trick of maintaining a substantial code-base by ensuring continuity of the software development. The group is large enough, with a number of senior members. Professors can increase their salary by up to 50% via industrial money.

Site: Ecole Polytechnique Fédérale de Lausanne (EPFL), Blue Brain Project

SV-BMC / AA-B118

CH-1015 Lausanne, Switzerland

http://bluebrain.epfl.ch/
Date Visited: February 26, 2008
WTEC Attendees: L. Petzold (report author), S. Glotzer, J. Warren, C. Cooper, V. Benokraitis
Hosts: Dr. Robert Bishop, EPFL, Chairman of Advisory Board to Blue Brain Project
Email: bob.bishop@epfl.ch

Dr. Sean Hill, IBM, Blue Brain Project Manager for Computational Neuroscience

Dr. Felix Schurmann, EPFL, Blue Brain Project Manager

In June 2005, IBM and the Brain Mind Institute at the Ecole Polytechnique Fédérale de Lausanne (EPFL) in Switzerland announced a plan to create a digital 3D replica of the brain. Named after the IBM Blue Gene supercomputer it relies on, the Blue Brain Project has the goal of modeling, in every detail, the cellular infrastructure and electrophysiological interactions within the cerebral neocortex, which represents about 80% of the brain and is believed to house cognitive functions such as language and conscious thought. Professor Henry Markram, who founded the Brain Mind Institute, directs the Blue Brain Project. Dr. Felix Schurmann manages the project and its international team of 35 scientists.

The Blue Brain Project was started by Henry Markram, a biologist, motivated by his experimental biology work. The Neocortex takes up 80% of the human brain. The project started out to reverse-engineer the neocortex. This involved collecting experimental data for different types of cells, types of electrical behavior, and types of connectivity. The model includes 10,000 neurons, 340 types of neurons, 200 types of ion channels, and 30 million connections. This is a detailed multiscale model, related to variables as measured in a laboratory. It is a huge system of ordinary differential equations, modeling electrochemical synapses and ion channels. The main objective of the model is to capture the electrical behavior. The model has been implemented in such a way that it is easy to modify and supplement. The current capability is a faithful in silico replica at the cellular level of a neocortical column of a young rat. A goal is to capture and explain electrical, morphological, synaptic, and plasticity diversity observed in the lab. Building a database of experimental results is a major part of this effort.

Some of the long-term objectives of this effort are to model a normal brain and a diseased brain to identify what is going wrong, and to evaluate emergent network phenomena. A demonstration was given in which the model replicated slice experiments that exhibit network-scale phenomena. The simulation showed low-frequency cortically generated oscillations emerging from changing extracellular potassium. In this way, it may someday be possible to do in silico pharmacology experiments.

The Blue Brain Project models the 10,000 neuron cells, involving approximately 400 compartments per neuron, with a cable-equation, Hodgkin-Huxley model. There are 30 million dynamic synapses. The computing platform used is a 4-rack blue gene L with a 22.4 TFlop peak and 2048 processors. The SGI Prism Extreme is used for visualization. The project features an impressive integration of experiment, modeling, high-performance computing, and visualization.

Site: Eni SpA

Via Felice Maritano 26

20097 San Donato, Milan, Italy

http://www.eni.it/en_IT/home.html
Date Visited: February 28, 2008
WTEC Attendees: A. Deshmukh (report author), G. Karniadakis, G. Lewison, P. Westmoreland
Hosts: Dr Francesco Frigerio, Refining & Marketing Division

Tel: +39-02-52036542; Fax: +39-02-52036354

Email: francesco.frigerio@eni.it

Ing. Alfredo Battistelli, Environmental Remediation and Land Conservation

Snamprogetti SpA

Tel: +39-07-21881467; Fax: +39-07-21881984

Email: alfredo.battistelli@snamprogetti.eni.it

Dr Luciano Montanari, Technical Manager

Physical Chemistry Department, Refining & Marketing Division

Tel: +39-02-52046546; Fax: +39-02-52036347

Email: luciano.montanari@eni.it

Ing. Fabrizio Podenzani, Development of Technology, Process Engineering

Refining & Marketing Division

Tel: +39-02-52056717; Fax: +39-02-52036116

Email: fabrizio.podenzani@eni.it