We conduct research into the mechanics and physics of solid earth processes on all scales (from nanometres to kilometres) using supercomputer simulations.

We carry out research within four themes:

  • Multiscale, multiphysics modelling: aims to couple the computation of several physical processes in a coherent way
  • Seismic and inversion research: investigates geophysics and rock physics to analyse the subsurface response to seismic or electromagnetic signals
  • Virtual near-wellbore: focuses on understanding the behaviour of fluids and particles in the region close to an oil, gas or coal-seam gas well
  • High-performance computing: looks at the appropriate use and configuration of supercomputers, parallel systems and graphics processing unit (GPU) programming to improve our capabilities in solving geoscience problems. The high-performance supercomputer Savanna is used by up to 100 scientists from within the UQ community and externally, with the following specifications:
    • 32 SGI ICE nodes with dual Intel Xeon E5462 2.80GHz (8 cores), 32GB RAM, 4x DDR Infiniband
    • 34 SGI nodes with dual Intel Xeon E5-2660v3 2.60GHz (20 cores), 128GB RAM, 4x FDR Infiniband
    • 4 NVIDIA Tesla K40m GPUs (2880 cores, 12GB RAM)
    • ~60 TB network attached storage, plus 68 TB local disk space on nodes.

Software packages

Within our high-performance computing research theme, we are also developing state-of-the art software for solving complex and large numerical simulations on highly parallel supercomputers and modern distributed user environments.

Through AuScope MNRF funding, the software packages have been made available for our school community under an Open Source licence.

The advanced computational technologies and simulation software under development is applicable to a wide range of industrial and environmental domains and provide a driver for innovation in the general area of simulation-assisted design, specifically in the sustainable energy, earth resources, mass mining and geotechnical sectors.

For more information on each tool, see the table below.

Escript

A python-based programming tool for mathematical modelling based on non-linear, time-dependent partial differential equations. It has been designed to give modellers an easy-to-use environment for developing and running complex and coupled models without accessing the underlying data structures directly. This approach leads to highly portable codes allowing the user to run a simulation on a desktop computer as well as highly parallel supercomputers with no changes to the program. Escript is suitable for rapid prototyping (for example for a student project or thesis) as well as for large software projects. It has successfully been used in a broad spectrum of applications including earth mantel convection, earthquakes, porous media flow, reactive transport, plate subduction, and tsunamis. Escript uses the finite element method (FEM). The code has been parallelized efficiently with MPI, OpenMP and hybrid mode. For more information, visit the Escript download page.

ESyS-Crustal

Software infrastructure for a deeper understanding and better description of interacting fault systems with potential applications in natural hazard forecasting and risk evaluation (earthquakes and tsunami generation forecasting), green energy exploitation (geothermal reservoir modelling), deep geological disposal (radioactive waste treatment and CO2 geological storage), groundwater modelling, minerals exploration and related environmental problems. ESyS-Crustal is designed for modern supercomputers and Linux-based multi-core desktop PCs.

The downunder tool-kit

An extension of the Escript software environment for building software for large-scale geophysical inversion of potential field, seismic and MT data. It makes use of the parallelization provide by Escript and therefore is particularly suitable for regional and continental scale inversion and joint inversion of data sets. Development is funded by the Australian Geophysical Observation System (AGOS). The Virtual Geophysical Laboratory is providing a portal for applying Escript-downunder online data sets. For more information, visit the Escript download page.

ESyS-Particle

A software package for particle-based numerical modelling. The software implements the Discrete Element Method (DEM), a widely used technique for modelling processes involving large deformations, granular flow and/or fragmentation. ESyS-Particle is designed for execution on parallel supercomputers, clusters or multicore PCs running a Linux-based OS. The C++ simulation engine implements spatial domain decomposition via the Message Passing Interface (MPI). A Python wrapper API provides flexibility in the design of numerical models, specification of modelling parameters and contact logic, and analysis of simulation data. ESyS-Particle has been utilised to simulate earthquake nucleation, comminution in shear cells, silo flow, rock fragmentation, and fault gouge evolution, to name but a few applications.

Projects

Project Duration
Improved spatial models of short-range permeability variations
ARC LIEF Collaborating/Partner Organisation Contributions
2015 to 2017
Non-linear geostatistics using copulas
QGC Pty Ltd
2015 to 2016
Uncertainty modelling with polynomial chaos expansion
QGC Pty Ltd
2015 to 2016
Multiscale, Multiphysics Modelling (M3)
ARC LIEF Collaborating/Partner Organisation Contributions
2015
Seismic and inversion research
ARC LIEF Collaborating/Partner Organisation Contributions
2015
Modelling non-seismic geophysics for CSG monitoring
Santos Ltd
2014 to 2016
Surat Geological Framework Model
Arrow Energy Pty Ltd
2014 to 2016
Development of a Virtual Near-Wellbore (WNWB) mathematical model for CSG
Santos Ltd
2013 to 2017
Simulation analysis and modelling UQ (SAM UQ)
National Collaborative Research Infrastructure Strategy (NCRIS)
2013 to 2016
Improvements in prestack seismic data processing for coal seam gas
ARC LIEF Collaborating/Partner Organisation Contributions
2013 to 2016
Construction and updating of Surat Cumulative Management Area geological model
UniQuest
2013

Contact us

To find out more about us, contact Dr Lutz Gross or Dr Suzanne Hurter.