Research IT

Enabling Award Winning Papers

Tom Flint from the Department of Materials was recently awarded the Donald Julius Groen prize for his outstanding paper from the Institution of Mechanical Engineers. Here he explains how the use of CSF was at the heart of his research.

My research involves developing mathematical frameworks that describe how multi-component and multi-phase substrates evolve subject to high energy density sources of heat; this generally involves the formulation of many, highly coupled, partial differential equations and converting these into forms that can be discretised numerically as these equations are far too complex to solve on paper. We need very powerful computers, such as the Computational Shared Facility (CSF), to solve these systems of equations numerically.

My work can be broadly divided into two research areas: thermal fluid dynamics, and microstructural modelling. In both areas, we have been doing some exciting work incorporating more complex physics into our mathematical modelling frameworks than has previously been considered.

One of the most exciting achievements, that myself and Dr Pratheek Shanthraj completed during the pandemic, was to develop a multi-component magneto-thermal-hydrodynamics framework that can describe how the momentum, thermal, compositional, and magnetic fields evolve in systems with large property gradients. We have validated this code up to Hartmann numbers of 10000 and reproduced the analytical solutions. For multi-phase problems we have re-produced experimentally observed phenomena.

We needed to develop this new approach because in arc welding, and many real-world processes, there are very large differences in the electrical conductivity and magnetic permeability across the interface (between the metallic substrate and gas/plasma in our arc welding case). This modelling framework is very powerful as now all the unknowns are either boundary conditions, or material properties (which may be a function of temperature). This allows us to replicate the process in the computer to optimise these processes at a fraction of the cost of doing the same thing experimentally.

These powerful mathematical frameworks, and associated numerical implementations in the form of computer codes, are not very useful without computational systems capable of running this software on many computational cores; so we can decompose our problem over many processors and solve complex problems.

The Donald Julius Groen award winning paper, ā€˜Magneto-hydrodynamics of multi-phase flows in heterogeneous systems with large property gradientsā€™ published in Scientific Reports, looked at electromagnetic and hydrodynamic evolution in systems with large property gradients and showed these approaches can be applied to arc-welding. The Donald Julius Groen prize is awarded by the Institute for Mechanical Engineering each year for outstanding papers or for outstanding achievements in the sphere of activity of the awarding Division or Group.

The CSF provided by Research IT, allows us to run our software on meaningful scenarios and generate results that are relevant. In our magneto-thermal-hydrodynamics work we used the HPC-pool on the CSF and distributed our most complex scenario, the arc-welding case, over 1024 cores and observe the development of complex physics.

Without the CSF, we simply would not be able to run our code on cases that demonstrate the generality and versatility of our code. We would be limited to running on perhaps 16 cores on a desktop that would take many months to complete the arc-welding simulation for example. The CSF is absolutely fundamental to all of my research and Iā€™m incredibly grateful to all of Research IT for running such an excellent resource.