zur Hauptseite

Institute for Advanced Simulation (IAS)

Navigation und Service


23.10.2013 13:30 Uhr
23.10.2013 15:00 Uhr
Jülich Supercomputing Centre, Hörsaal, Geb. 16.3, R. 006
1. Thema:
Combination of Stochastic and Numeric Modeling of Fibrous Microstructures of HT-PEFC Gas Diffusion Layers
Dieter Froning, IEK-3
In PEM-based fuel cells – where the HT-PEFC belongs to – the gas diffusion layer (GDL) connects the electrodes with the feeding channels. Efficient operation of fuel cells requires that the electrodes are sufficiently supplied by fuel resp. air from the channels. Reaction products must be transported away from the electrodes. Therefore mass transport of fluids is one major task for the GDL. The GDL itself is composed of materials based on carbon fibers, e.g. paper, woven and non-woven textiles. The geometry of the GDL is described by a 3D stochastic simulation model, with its parameters fitted from a 2D SEM image. Based on the stochastically generated microstructure, transport processes are numerically modeled by the Lattice Boltzmann technique. In real fuel cells, different local compressions influenced by the applied flow fields lead to different microstructures in regions under the land and under the channels. The effect of compression is discussed in detail. Transport processes are simulated in virtual structures to detect quantitative relationships between functionality and microstructure and to design virtual GDL materials with improved transport properties.
2. Thema:
Towards self-consistent models of brain networks: technology and neuroscience
Prof. Abigail Morrison, INM-6
Concerning its connectivity, the brain is different from any technology available. Each nerve cell receives inputs from on the order of 10,000 other neurons and sends output to the same number of units. Furthermore, only 50% of the connections are established locally within a cubic millimeter of brain tissue. The remaining synapses are established with nerve cells distributed across the whole brain. Despite recent progress in understanding the local dynamics, this connectivity of the brain has made it difficult to formulate self-consistent models at the resolution of neurons and synapses due to the large amount of computer memory which needs to be gathered and coordinated. In JINB33 we are developing the simulation technology for brain-scale simulations and the respective software infrastructure. The talk presents the results of the previous accounting period including a record simulation and discusses initial models of brain-scale networks. The work at Jülich is set into perspective with the challenges we are addressing with our partners in the European Human Brain Project (HBP).
Wednesday, 23 October 2013, 13:30 - 15:00
Jülich Supercomputing Centre, Hörsaal, building 16.3, room 006
Ankündigung als pdf:
 VSR Seminar on 23 October 2013 (PDF, 33 kB)

Alle Interessierten sind herzlich eingeladen.