Publications and Projects
Iceberg has been used to support a diverse range of projects. Much of the work has resulted in the generation of publications and further funding for those projects.
The publications link on the RHS of this page provide a list of some of the publications generated with support from iceberg.
A small sample list of active projects and further links to these projects that take advantage of 'iceberg' are listed below:
Atomistic & Mesoscale Simulations for Radioactive Waste Disposal
Karl Travis, Department of Materials Science and Engineering
Low-end computing provision limits the radiation damage simulations to low recoil energies and to very short anneal times. Calculations involving several million atoms are needed for realistic simulations on candidate hosts for plutonium. The ability to model the complex fluid mechanics in a deep geological borehole, even with the simplification of smooth particle applied mechanics (SPAM) , still requires that multi-million particle simulations be conducted. Similarly, simulations of surface wear and crack propagation need high specifications HPC computer facilities
Biomineralization and Biomimetics
John Harding, Department of Materials Science and Engineering
Living systems have a remarkable ability to control the size, shape, structure and properties of crystals. Simulation is needed to elucidate the mechanisms that produce these outcomes, mapping out their domains of applicability and so guiding experiment. Nucleation, self-assembly and growth are long timescale processes that require multiscale models and high performance computation. Applications range from carbon capture (through understanding how marine organisms such as corals and coccoliths deposit calcium carbonate) through nanomaterials and biomaterials (which includes medical applications – bones and teeth are classic examples) to fouling (biofilms) and scaling (deposition of calcium carbonate in heat exchangers).
Computational Systems Biology
Mike Holcombe, Computing Science
As more is discovered about the structure, organisation and behaviour of cells, tissues, organisms and communities of biological systems the need to understand how all of these systems and phenomena work and interact in a holistic fashion becomes more urgent.The promise of being able to use the power of computers and of recent computational and mathematical modelling techniques to understand and predict important aspects of the behaviour of biological systems is an exciting and vitally important opportunity for medicine and biology.
The Computational Biology Group is at the forefront of this endeavour and is working extremely closely with experimental biologists and clinicians in building realistic and useful models of biological phenomena from the molecular level, to the cellular, tissue, organismal and social levels.
Computational modelling tools are developed and run on iceberg to perform computational experiments to understand and model the behaviour of biological systems.
Designing Nanomaterials for Energy Applications
Merlyne da Souza, Electrical Engineering
Ab initio density functional calculations using HPC facilities are essential to understand the detailed structure and electronic properties of nanoscale materials. Examples include charge transfer in carbon nanotubes, the performance of zinc oxide thin film transistors, the development of high-k dielectrics and defect engineering in semiconductors.
Large Scale Calculations for Aerospace & Energy Applications
Ning Qin , Mechanical Engineering
High performance computing is particularly useful for two areas: simulation of flow in and around complex geometries and the multi-disciplinary design optimization of wings, aero-engines and wind turbine blades. Enhanced computational power permits the use of large eddy simulations and detached eddy simulations. Applications include flow control for aircraft, drag/emission reduction, flow around high lift devices and landing gears, and noise
generation from airframes and aero-engines. While tools exist to carry out multi-disciplinary design optimization for a range of industrial problems, their practical application is hindered by available computing facilities. Investigation coupling high fidelity models in both fluid and structure can provides further understanding of the engineering problems, which can leads to innovative solutions for practical applications.
MHD Wave Coupling in the Solar Atmosphere
Robertus Erdelyi , Victor Fedun, School of Mathematics and Statistics
The excitation, time dependent dynamic evolution, nonlinear propagating (i.e. solitary) waves in open magnetic flux tubes, i.e. in the building blocks, of the compressible solar atmospheric stratified plasma is a major center of attention in modern solar, space and astrophysics. Proving the existence of such waves, and their leakage from the solar surface (photosphere) through the transition region into the solar corona is of fundamental importance in solar MHD research because of the energetic and diagnostics implications. We need to examine numerically the linear and non-linear propagation of magneto-acoustic and Alfvénic waves in flux tube embedded into the 2- or 3D realistic solar atmosphere. With much-needed forward modelling we serve as impetus for magneto-seismologic acquisition of the complex and dynamic solar atmosphere, that is a center of attention in recent (SOHO, STEREO, Hinode, SDO) an future (Solar Orbiter, Solar-C, HiRiSE) space missions.
Numerical Modelling of Magneto-helioseismology
Robertus Erdelyi, Victor Fedun, Mike Griffiths, School of Mathematics and Statistics
Code development for direct forward numerical modelling of the physical processes beneath the visible solar surface. Realistic numerical simulations allow us to provide revolutionary artificial data with the precision necessary to prove or disprove the physics derived from the ground (GONG) and space-based (MDI, HMI) solar sub-surface wave and oscillation observations. Also, the simulations provide a way to validate the interpretation of the observational data. This will lead to the ability to predict, on the basis of the observations (or taking the observational data as initial condition for the simulations), the large-scale characteristics of the solar activity and Space Weather.
Osteoporotic Fracture Risk by Finite Element Analysis of Medical Scans
Dr Lang Yang, Department of Human Metabolism
Osteoporosis-related vertebral and hip fractures are major health problems in the UK. This project aims to develop patient-specific finite element models of the femur and vertebra from the patients´ medical bone scans and apply the models in large clinical studies to assess their ability to estimate the bone strength and to predict of fracture risk.
Quantum Dynamics of Small Molecules
Antony Maijer, Department of Chemistry
iceberg cluster allow us to generate accurate potential energy surfaces for 4 and 5-atom systems and run quantum dynamics calculations on such surfaces. The results of these calculations can be used to compare directly to sophisticated experiments done in other groups. Molecular cross sections and reaction rates would be generated, which can be used in subsequent modelling of atmospheric processes. These would be world-leading state-of-the-art calculations, which would have no equivalent elsewhere, making us a very attractive partner for collaborations between experimental and theoretical groups.
New Perspectives in Particle Physics
Dan Tovey, Department of Physics and Astronomy
The University of Sheffield ATLAS group in the Department of Physics and Astronomy is seeking evidence for Supersymmetry and the Higgs boson at the Large Hadron Collider at CERN. The work of the group is crucially dependent on access to high throughput computing for analysis of the expected 3 peta-bytes of data generated per year and for performing essential supporting high statistics Monte Carlo simulations. Increased local computing power enables us to analyse and understand the latest data more quickly and hence increases our ability to generate the most exciting results in competition with other UK and worldwide groups. It also buys us access to, and influence within, the national (GridPP) and international (LCG) organisations which coordinate Grid-based LHC data processing. This gives us a larger stake in the increasing spin-offs from particle physics Grid computing developments in areas such as bio-medical informatics and commercial data-mining.
Simulation of Complex Flows
Shuisheng He, Mechanical Engineering
High performance computing is used to conduct fundamental research on turbulence as well as solving industrial flow and heat transfer problems involving complex geometries. Various Computational Fluid Dynamics (CFD) methodologies have been employed, including, Reynolds Averaged Navier Stoke (RANS), Direct Numerical Simulation (DNS)/Large Eddy Simulation (LES) and Lattice Boltzmann Method (LBM). Examples of projects include (i) study of unsteady turbulent flow over rough surfaces using DNS (EPSRC); (ii) simulation of migration of carbon dioxide in rocks under conditions of saline aquifers in support of Carbon Capture and Storage (CCS) using LBM (EPSRC) and (iii) study of effect of cross flow in nuclear reactors using RANS (EDF).
Simulations of Electrical Activation in Whole Hearts
Richard Clayton, Rod Smallwood , Dept of Computer Science
The heart is an electromechanical pump, where electrical activation initiates and co-ordinates mechanical contraction. Integrative and multiscale computational models of heart tissue are important because they enable us to study how cellular (or subcellular) phenomena such as a gene mutation can predispose an individual (i.e a whole heart) to a lethal failure of normal electrical activation. There are several current trends within the heart modelling community that require high performance computing; one is towards models that couple electrical activation and continuum mechanics, another is towards simulations based on individual hearts (human and animal), and a third is towards projects that combine detailed data from experiments and simulations. All of these areas are both data and compute intensive.
Star Formation – a Hot Issue in Astrophysics
Simon Goodwin, Department of Physics and Astronomy
Understanding star formation is a major issue. The big questions include: why do stars form as binary or triple systems and in clusters of hundreds or thousands of stars? Why do stars have the mass distribution that they do? How do star clusters evolve? Under what conditions are planets formed? To address these problems requires multiple large scale calculations involving astrophysical fluid dynamics and gravitational N-body dynamics. These are both extremely computationally expensive, typical simulations take several weeks with currently available clusters. High performance computing will make a major difference to our ability to understand how stars form and evolve.
State-of-the-art speech and Language Technology
Thomas Hain, Department of Computing Science
The use of discriminative training techniques in speaker adaptation is a project running on iceberg.
The project is part of a larger project which applies state-of-the-art speech and language technology to the task of recognising spontaneous speech in meetings. More specifically, it examines the impact of discriminative acoustic model estimation upon the performance of the speech recogniser in this scenario.
WUN Medieval Projects - Virtual Vellum
Michael Meredith, Michael Pidd , Humanities Research Institute
Providing home for the Virtual Vellum Project which is an e-Science demonstrator project that has been funded by EPSRC/JISC/Arts & Humanities e-Science Initiative and the UK e Science Core Programme with the aim of promoting and demonstrating the use of technology within arts and humanities research.