Posters

HPC summer school poster session

Date and place: October 2nd, 2015 at 16:30, Iperial College London, Huxley 341

PRISM: Platform for Research in Simulation Methods

Dr. Chris Cantwell (Aeronautics), Dr. Colin Cotter (Mathematics), Dr. Gerard Gorman (Earth Science and Engineering), Dr. David Ham (Computing and Mathematics), Prof. Paul Kelly (Computing), Prof. Christopher Pain (Earth Science and Engineering), Dr. Joaquim Peiro (Aeronautics), Dr. Matthew Piggott (Earth Science and Engineering), Prof. Spencer Sherwin (Aeronautics), Dr. Peter Vincent (Aeronautics)

The PRISM team are leading experts in developing finite element methods for industrial (multiphase flows and nuclear engineering, train aerodynamics, flow control, road and racing car aerodynamics, marine renewables), environmental (ocean modelling, numerical weather prediction, coastal engineering) and biomedical (cardiovascular flows, electrophysiology modelling in the heart, image registration) applications.
Kai Arulkumaran, Dr. Anil Bharath (Bioengineering)
Our team draws together expertise in designing, analysing and implementing sophisticated finite element methods, deploying these methods in a broad range of industrial, biomedical and environmental applications, and developing software tools that deliver portable parallel performance. They form the UK’s largest group of experts in developing finite element methods and software, with international reputations in high order methods, adaptivity, mesh generation. Combining these aspects allows simulations of highly complex multiscale problems that are otherwise inaccessible on available computational resources, leading to new capabilities across a broad range of engineering applications.

The  work of the PRISM team is focussed on agile synthesis of modelling techniques: flexibly selecting a combination of discretisation method, polynomial order, mesh, equation assembly algorithm, linear/nonlinear equation solver etc. to meet application-driven problem requirements and to match available hardware.

The Computational Methods Hub - education, training and support for PhD students from Centres for Doctoral Training

The Computational Methods Hub

The Computational Methods Hub provides education, training and support for computational science PhD students though direct course provision, through supporting excellence in computational science training in individual CDTs, and by providing ongoing support for CDT students over the course of their PhD.

The defect chemistry of thoria-ceria mixed oxides by DFT

N. Kuganathan, P. S. Ghosh, R. W Grimes, C.O. T. Galvin, A. Arya, B.K. Dutta and G. K. Dey

Fission gasses, such as Xe, formed during 0rmal reactor operation are accommodated initially at defect sites in the fuel lattice and are k0wn to have a deleterious effect on fuel performance, particularly at high levels of burnup. Using first-principles density functional theory, we calculate the three most stable structures of Scottky defects (SD) in ThO2, CeO2 and Th0.94Ce0.06O2 and accommodate a single Xe atom in the most favourable of these to investigate how Xe atom interacts with defects. Our simulations reproduce the experimental ThO2 and CeO2 bulk crystal structures well, establishing the quality of the pseudopotentials and basis sets used. Most favourable Schottky defect in all three structures is SD2. In larger supercells, SD3 is dominant as observed in the classical effective potential simulation. A single Xe atom is most stable when accommodated within a SD1 and is highly unlikely to occupy the interstitial site. This is because the SD provides a larger volume to accommodate a large Xe atom. Incorporation can be further facilitated by increasing the number of Schottky defects. Substitution of Ce for Th in ThO2 does 0t affect the overall trend observed in the pure (end-member) lattices due to the relatively small lattice distortions.

FGLab: Machine learning dashboard

Kai Arulkumaran, Dr. Anil Bharath (Bioengineering)

FGLab is a machine learning dashboard, designed to facilitate performing experiments. Experiment details and results are sent to a database, which allows analytics to be performed after their completion.

Parallel Direct Numerical Simulation of Three-Dimensional Two-Phase Flows

L. Kahouadji, O. K. Matar, D. Juric, J. Chergui & S. Shin

We present simulation results for several two-phase flow problems using a new solver for massively parallel simulations of fully three-dimensional multiphase flows. The code which we call BLUE is wholly written in Fortran 2003 and uses a domain decomposition strategy for parallelization with MPI.

The fluid interface tracking is based on a high fidelity hybrid Front-Tracking/Level Set algorithm for Lagrangian tracking of arbitrarily deformable phase interfaces including breakup and coalescence and a precise treatment of surface tension forces, interface advection and mass conservation. The code couples an implicit, incompressible Navier-Stokes solver with multigrid pressure solution to the Lagrangian tracking method for implementation on large-scale parallel computing architectures. The modular program structure allows for the application of the code to a wide variety of physical scenarios: free surface instabilities, flow of bubbles or drops with coalescence and breakup, droplet impact or flow around immersed solid objects for microchannel flows for example.