HPC-based simulations, Engineering and Environment @ BSC

Barcelona BSC Campus Nord

Barcelona BSC Campus Nord


The registration to this course will open in September. Please, bring your own laptop. All the PATC courses at BSC are free of charge.

Course convener: Mariano Vazquez

Objectives:The objetive of this course is to give a panorama on the use of hpc-based computational mechanics in Engineering and Environment through the projects BSC are carrying on. This panorama includes the basics of what is behind the main tools: computational mechanics and parallelization.

Learning outcomes:The students who finish this course will be able to take active part in such projects both in academia or industry.

Level: For trainees with some theoretical and practical knowledge


Day 1

9.00 - 10.00:
Introduction to Computational Mechanics:
Mariano Vázquez
What is behind a simulation code? Main concepts. The Physical system and its Mathematical description
Discretization: divide and conquer Discretization: algorithms and codes


10.00 - 11.00:
The CompBioMed project: an example of engineering simulations in the biomedical industry
Mariano Vázquez
The European Center of Excellence in Computational Biomedicine is a joint effort to link engineers to medical research through the use of large-scale computational resources and efficient parallel programming. Biomedical research presents perhaps the most extreme and challenging simulation scenario for computational mechanics codes. In this talk we will describe how this project targets medical devices industry, pharma and clinical research.


11.00 - 13.00:

HPC and fluid-structure interaction
Juan Carlos Cajas - Beatriz Eguzkitza
We present a general coupling strategy for multi-physics problems. The basic idea is to have independent codes, one for each physical problem, and communicate the coupling variables using MPI. Special interest on Fluid Structure Interaction (FSI) problems. Usually, the multi-physics problems involve different space and time scales, which can lead to situations in which optimized algorithms for the individual problems are useless in the coupled one. Thus, different coupling algorithms and relaxation schemes are considered and tested. Cases for FSI problems in wind generation and bio-mechanics are considered.


14.00 - 16.00:
Parallel algorithms for Computational Mechanics:
Guillaume Houzeaux - Ricard Borrell
What is parallelization in a simulation code? Paradigms and scenarios.
Description of parallelization schemes. Parallel algebraic solvers and solving strategies.

16.00 - 18.00:
Computational Fluid Dynamics (CFD): turbulence, driving and sailing:
Herbert Owen - Oriol Lehmkuhl
CFD is one of the fields of Computational Mechanics where HPC and parallelization is more influential, due to the Physical complexity of the systems. This talk describes incompressible flow applications in automotive industry and yacht design. The Physical description includes turbulence modelling, free surface and floating rigid bodies.

Day 2:
9.00 - 11.00:
Introduction to Computational Solid Mechanics
Eva Casoni - Gerard Guillamet
This talk provides knowledge in all the technical aspects of Solid Mechanics analyses, which are in high demand in many industries.
Solid mechanicians focus on the deformation and failure of materials with a defined rest shape for real-world applications. A solid mechanician envision the final application and uses theoretical, experimental, numerical and computational tools to solve the problem.
A general view of the most used and useful approaches and constitutive theories applicable to the deformation and fracture of metals, composite and biological materials will be done in this talk, covering the general aspects of the modelling and solution approach.

11.00 - 13.00:
Introduction to mesh generation for simulation
Xevi Roca - Abel Gargallo
This course is a brief introduction to fundamental mesh generation approaches used in academic and commercial simulation. Mesh generation methods have succeeded in decomposing highly complicated domains by filling them with distributions of different types of elements such as triangles, quadrilaterals, hexahedra, tetrahedra, pyramids, and prisms. These meshing methods are used in a daily basis by computational engineers and scientists to obtain numerical predictions over complex geometrical configurations. The course introduces different types of: geometrical representations, meshing methods, element types, boundary approximations, quality measures, sizing approaches, and software packages. This is an introductory course intended to facilitate the election of the proper mesh generation methods for simulation.

14.00 - 16.00:
N-bodies Contact Detection and Resolution
Cristóbal Samaniego
The talk is divided in two main subjects:
First, the contact detection algorithm prevents interpenetration between bodies by estimating the time of collision. The algorithm includes efficient search methods to drastically reduce the number of operations when we estimate the time of collision between a pair of bodies.
Second, the contact resolution algorithm changes the velocity of the bodies in contact in order to prevent interpenetrations. This subject also includes methods to reduce the execution time. Also, other aspects of the n-bodies contact are described to improve and to have a more robust method to solve the interaction between rigid solids.

16.00 - 18.00:
Introduction to numerical combustion
Daniel Mira
The energy market is leading towards cleaner solutions in order to reduce pollutant emissions from combustion systems.
Nowadays, numerical simulations have become an important tool to provide insight into the dynamics of flames as well as the overall performance of the entire combustion device. In particular, turbulent combustion is a complex phenomenon involving the interaction of chemical reactions and heat release with turbulent flow structures. This interaction leads to the development of a wide range of time and length scales, coupled to hundreds of species and reactions, so the requirements in HPC are an essential aspect to address this problem.
This session addresses some fundamental aspects of combustion modelling with emphasis on HPC and practical examples of gas turbines.

Day 3:
9.00 - 11.00:
Scientific visualization
Luz Calvo
The visual representation of scientific data has been a key component of science, advancing thanks to it or directly causing advances. Nowadays, the field of scientific visualization is growing fast, thanks to the technological explosion and a renewed interest of society in design and aesthetics. In this course we will survey the field of data visual representation, discuss about available tools, and touch on narrative topics that researchers can learn on their own to improve their graphical communication skills. We will explore elements of computer graphics, human-computer interaction, perceptual psychology and design in addition to data integrity to learn how to present this data to an observer in a way that yields insight and understanding.

11.00 - 13.00:
Biomechanics: Cardiac Computational Modelling
Jazmin Aguado-Sierra - Ruth Arís
From an engineering point of view, Biological systems are amongst the most complex Physical systems in Nature. Multiscale, multiphysics, great variability, large uncertainties, numerical issues, validation difficulties and extremely complex mathematical models are amongst the common features of computational biomechanics. Considering that all these problems usually show up altogether, the use of HPC-based simulations in biomechanics is a must.
In the BSC's CASE department, we focus in simulations at organ level. The "Alya Red Cardiac Computational Model" is a paradigmatic example, which will be deeply described in the talk.

14.00 - 15.00:
HPC Challenges in the Oil Industry
Mauricio Hanzich - Josep de la Puente
An introduction to the numerical methods involved in the modelling, migration and inversion of seismic and EM data for hydrocarbon exploration. Talk will include: why the geophysical exploration matters, what are the main challenges today and the future trends and how HPC is mandatory for many geophysical problems.
From methods and algorithms for geophysical exploration to HPC software on modern architectures. Talk will include: main issues to be tackled for HPC applications for Oil Industry, current programming models and paradigms for such applications and current state of HPC environments and future trends.

15.00 - 16.00:
Supercomputing for fusion energy applications
Xavier Saez
Future energy requirements set an unprecedented challenge for our
society. Fusion energy is uniquely placed to meet the growing energy
demand. In this talk illustrative examples of computer modeling in the
fusion energy field are discussed, with special emphasis in applications
requiring supercomputing resources.

16.00 - 17.00: meteo / atmospheric (arnau / matias)
Atmospheric transport modelling & High-resolution meteorological modelling using CFD
Matías Ávila - Arnau Folch
Example case 1: volcanic ash dispersal and civil aviation - Atmospheric transport models are used to simulate the dispersal of any substance in the atmosphere. Applications include dispersal of pollutants or air quality modelling, among several others. In paricular we focuss on volcanic ash dispersal and its impacts on civil aviation.
Example case: assessment of wind energy resources - CFD is the pivotal tool to increase the spatial resolution of mesoscale Numerical Weather Prediction Models. This talk describes how turbulent CFD models are used to assess winds and turbulence in the microscale, focussing on the evaluation of the wind resource for eolic energy.

17.00 - 18.00:
Atomistic simulations in material science

George Huhs - Stephan Mohr

Be it the design of lighter and more stable materials for a car, the development of longer lasting batteries for a smartphone, or improving the efficiency of solar cells -- the quest for novel materials is a key aspect in many technological branches. Material science is thus a very broad field, uniting numerous theoretical and experimental approaches. In this talk we will present methods based on a quantum mechanical description on the atomistic level. This approach gives insights into many characteristics of materials that might not even exist yet. To this extent complex equations must be solved, requiring large supercomputers for treating realistic problems. Recent developments push the frontiers of material science even further by applying big data analytics to collections of calculations. Finally we will present some examples and applications, demonstrating how these methodologies are valuable components in the toolchain for designing novel materials.

End of Course

The agenda of this meeting is empty