HIGH-FIDELITY EDDY-RESOLVING SIMULATIONS ON MULTI-CORE ACCELERATORS FOR MULTI-PHASE FLOWS IN CHEMICAL, ENERGY & TRANSPORT

The use of scale resolving simulations (SRS) for single phase flow applications has already shown dramatic accuracy benefits. The term SRS encompasses methods resolving a greater spectrum of turbulence e.g. large eddy simulation (LES), quasi-direct numerical simulation and hybrid methods e.g. detached eddy simulation (DES). The purpose of this work is to extend these methods for multi-phase applications. The use of SRS for single-phase turbulent flows is an area of fluids mechanics that has been widely studied for the past twenty years but SRS of multi-phase flows remains a very understudied area. The project will develop a massively parallel, high-order, fully implicit (temporal and spatial), multi-phase scale resolving methodology and perform simulations of (1) a representative aero-engine bearing chamber, (2) a representative transmission system gear and (3) a continuous chemical reactor. It will demonstrate the next generation of multi-phase high-fidelity flow simulations. We will exploit novel computing hardware through the extension and use of a state of the art fully implicit parallel library developed at the University of Oxford. The library, which enables 'future proofing' of CFD codes for modern hardware architectures, has been shown to give a 27x speedup on a GPU compared with the Intel Math Kernel Library tri-diagonal solver on a CPU. The research will be led by Dr. Richard Jefferson-Loveday, Assistant Professor in the department of Engineering at Nottingham University. It will be undertaken in collaboration with industrial partners MAHLE Powertrain, Rolls-Royce, ROMAX and GSK.

The project will have immediate substantial improvement for the design of:

(1) aero-engine bearing chambers
(2) wind turbine transmissions
(3) electric and hybrid vehicle transmissions
(4) continuous chemical reactors.

It will support right-first-time, high value design for multi-phase flow systems through the provision of far greater accuracy and realism through resolving a greater spectrum of turbulence and using modern computer hardware. The technology has great attraction as demonstrated by the letters of support from four industrial partners. The industrial partners will form a stakeholders' group to engage with the research right from the start.

The project will lead to developments for aero engine bearing chambers enabling more compact designs and higher operating speeds for smaller and more efficient engine cores. It will lead to higher efficiencies and reliabilities for wind turbine transmissions and considerable energy savings and reduced maintenance costs. The project will enable improved design methods for electric and hybrid vehicle transmissions enabling higher efficiencies and reduced space and weight through the use of high speed electric motors. At the same time it will also develop improved design for continuous chemical reactors leading to new environmentally friendly production routes through reduced solvent use and increased yields for a lower energy requirement. All of these developments will contribute to decreased reliance on fossil fuels and reduced CO2 emissions through improved design helping to safeguard future generations.

The project also has potential wider environmental and economic benefits for UK industries including oil, gas, nuclear, and marine and submarine transport supporting right first time design and construction and reducing long-term development costs and testing. All of these industries are under constant regulatory and financial pressure to develop processes that are environmentally and economically more efficient than current methodologies. Safety will be improved through more accurate predictive methods helping to prevent major accidents and disasters.