E-CAM Workshop on Particle-Based Models and HPC @ CSC, Finland

Europe/Helsinki
Training room Dogmi, 1st floor (CSC - IT Centre for Science)

Training room Dogmi, 1st floor

CSC - IT Centre for Science

Life Science Center, Keilaranta 14, Espoo, Finland
Description

Workshop Objectives

The goal of the workshop is to gather together scientists interested the cutting-edge methodology of approaches from the Discrete Element Method to Lattice Boltzmann simulations to coupled particle-fluid codes. The program consists of talks by experts highlighting recent progress, including both case studies of scientific problems and method development for HPC, contributed talks by participants, and a session of hands-on-training on 2-3 example codes using the CSC facilities. This three day afternoon-morning workshop is equivalent to two full days of program.

Programme

Please see the detailed workshop programme under "Timetable" menu.

 

Speakers and organizers:

 

Language:  English
Price:          Free of charge

Support
    • 12:00 12:15
      Registration Training room Dogmi, 1st floor

      Training room Dogmi, 1st floor

      CSC - IT Centre for Science

    • 12:15 13:50
      Welcome Lunch 1h 35m Restaurant, Life Science Center Keilaniemi

      Restaurant, Life Science Center Keilaniemi

    • 13:50 14:00
      Workshop Opening Training room Dogmi, 1st floor

      Training room Dogmi, 1st floor

      CSC - IT Centre for Science

      Life Science Center, Keilaranta 14, Espoo, Finland
    • 14:00 14:40
      Afternoon Session I: Ice mechanics with DEM / Arttu Polojärvi (Aalto University) Training room Dogmi, 1st floor

      Training room Dogmi, 1st floor

      CSC - IT Centre for Science

      Life Science Center, Keilaranta 14, Espoo, Finland
    • 14:40 15:20
      Afternoon Session I: Fracture models with DEM / Jan Åström (CSC) CSC's training room Dogmi

      CSC's training room Dogmi

      CSC - IT Centre for Science

      Life Science Center, Keilaranta 14, Espoo, Finland
    • 15:20 15:40
      Coffee break 20m CSC's training lobby

      CSC's training lobby

      CSC - IT Centre for Science

      Life Science Center, Keilaranta 14, Espoo, Finland
    • 15:40 17:00
      Afternoon Session II: Granular media with DEM / Vanessa Magnanimo (University of Twente) CSC's training room Dogmi

      CSC's training room Dogmi

      CSC - IT Centre for Science

      Life Science Center, Keilaranta 14, Espoo, Finland
    • 17:00 17:10
      Closing the Day 1st Training room Dogmi, 1st floor

      Training room Dogmi, 1st floor

      CSC - IT Centre for Science

      Life Science Center, Keilaranta 14, Espoo, Finland
    • 09:00 10:20
      Morning Session I: Contributed talks CSC's training room Dogmi

      CSC's training room Dogmi

      1) B. Muite: Numerical investigations of Convection; 2) I. Honkonen: Particle-assisted magnetohydrodynamics; 3) K. Khakalo: Coarsening and mechanics in the bubble model for wet foams; 4) M. Alcanzare: Controlled propulsion and steering of magnetic nanohelices; 5) E-L Laine: Problems associated with 3D modelling of rock fracturing – some examples of FEMDEM applications.

    • 10:20 10:40
      Coffee break 20m CSC's training lobby

      CSC's training lobby

    • 10:40 12:00
      Morning Session II: 1. Lattice Boltzmann method - algorithm, implementation and applications / Keijo Mattila (University of Jyväskylä) 2. Lattice Boltzmann implementation on GPUs / Fredrik Robertsen (CSC) CSC's training room Dogmi

      CSC's training room Dogmi

      The lattice-Bolztmann method (LBM) is based on a statistical description for the dynamics of underlying microscopic particles. That is, the method relies on evolution equations which represent discrete counterparts for particular model Boltzmann equations. Due to the kinetic theory roots, LBM is commonly classified as a mesoscopic method: it is well suited for the simulation of many complex fluid flow phenomena involving, e.g., multiphase and multicomponent fluids, flow in porous media, and particle suspensions. In this presentation, the history and background of LBM is first discussed, together with some application examples, followed by a short introduction to the theoretical basis of the method, including an overview of the discretization procedure, and finally concluding with the implementation aspects.

    • 12:00 13:30
      Lunch 1h 30m Restaurant, Life Science Center Keilaniemi

      Restaurant, Life Science Center Keilaniemi

    • 13:30 14:30
      Afternoon Session: Cell-division on GPUs / Jan Westerholm CSC's training room Dogmi

      CSC's training room Dogmi

      We simulate the behaviour of tissue growth by approximating
      cells as C_180 fullerenes capable of cell division. The fullerenes
      consist of carbon atoms connected by classical springs and an internal pressure
      pushing the carbon atoms out. The carbon atoms interact through
      a short range repulsion and a long range attraction. The emphasis of
      this presentation is on the design of an efficient GPU implentation in CUDA,
      capable of simulating over 200000 carbon atoms/1000 cells on one GPU.

    • 14:30 14:50
      Coffee break 20m CSC's training lobby

      CSC's training lobby

    • 14:50 17:00
      Tutorial: Hands-on with GPUs CSC's training room Dogmi

      CSC's training room Dogmi

    • 17:00 19:00
      Get Together party CSC's training lobby, 1st floor

      CSC's training lobby, 1st floor

      CSC - IT Centre for Science

      Life Science Center, Keilaranta 14, Espoo, Finland
    • 09:00 10:20
      Morning Session I: Contributed talks CSC's training room Dogmi

      CSC's training room Dogmi

      1) J. Koivisto: Friction controls submerged granular hoppers; 2) J. Farmer: A multiscale code for Raman amplification in plasma; 3) J. Kwak: Numerical simulations of pedestrian flow; 4) Y. Gong: Basal friction evolution and crevasse distribution during the surge in Basin-3 Austfonna ice-cap.

    • 10:20 10:40
      Coffee break 20m CSC's training lobby

      CSC's training lobby

    • 10:40 12:00
      Morning Session II: Fracture with DEM / Ferenc Kun (University of Debreceen) CSC's training room Dogmi

      CSC's training room Dogmi

      Discrete element modelling of the compressive failure of porous rocks
      Ferenc Kun (Department of Theoretical Physics, University of Debrecen)

      Understanding the processes that lead to catastrophic failure of porous granular media is an important problem in a wide variety of applications, notably in Earth science and engineering. Such failure is often preceded by detectable acoustic emissions which may be used to forecast the impending catastrophic event. Under compressive loading in the vicinity of failure localization emerges such that cracking events get concentrated in a damage band where eventually a macroscopic crack develops and the system falls apart. For a comprehensive understanding of the spatial structure of damage and of the statistics and dynamical features of acoustic emissions computer simulation of realistic models is indispensable.

      Here we investigate the scaling properties of the sources of crackling noise and the spatial structure of damage in a discrete element model of porous granular materials [1,2,3]. Simulations are performed to investigate the strain controlled uni-axial compression of cylindrical sand stone samples. We show that in our DEM framework cracking avalanches can be identified and the source size, energy, and duration can all be quantified for an individual event. The statistics of single event quantities are all characterized by power law distributions over a broad range of scales. The waiting time also depends on event size: after large events one has to wait longer for the next one, indicating a degree of memory of the associated transient stress relaxation [1,2].

      Close to failure damage localizes in a narrow shear band containing a large number of poorly-sorted fragments with properties similar to those of natural and experimental faults. We determined the position and orientation of the central fault plane, the width of the shear band and the spatial and mass distribution of fragments. The relative width of the shear band decreases as a power law of the system size and the probability distribution of the angle of the central fault plane converges to around 30 degrees. The mass of fragments is power law distributed, with an exponent close to that inferred for experimental and natural fault gouges. The fragments are in general angular, with a clear self-affine geometry [3].

      [1] F. Kun, I. Varga, S. Lennartz-Sassinek, and I. G. Main, Phys. Rev. E 88, 062207 (2013).
      [2] F. Kun, I. Varga, S. Lennartz-Sassinek, and I. G. Main, Phys. Rev. Lett. 112, 065501 (2014).
      [3] G. Pal, Z. Janosi, F. Kun, and I.G. Main, Physical Review E 94, 053003 (2016).

    • 12:00 12:20
      Workshop Closing CSC's training room Dogmi

      CSC's training room Dogmi