Application of High-Order Large Eddy Simulation Turbomachinary

I am interested in the use of high order Large Eddy Simulation (LES) and its plication to turbomachinery. LES aims to model 80-90% of the turbulence on grid and, if it is correctly applied, it has can prove to be more informative that other industrially used schemes. The turbomachinery applications best suited to LES are jets and ribbed passages; 2 Reynolds number independent flows. Experimental testing these regimes is very expensive and if high order methods can be successfully applied the cost of testing could be greatly reduced. The main benefit of high order methods is that they are faster compared to industrially used codes for like-for-like error. This is due to their ability to use coarser meshes. The modelling of jets is particularly important as it will impact the predicted noise levels of the aircraft. This is something that is becoming more pertinent in Britain, especially with the proposed 3rd runway for Heathrow.
A further application suited to the use of LES is the modelling of low pressure turbine (LPT) blades. This requires the scheme to more accurately handle the viscous fluxes. There are several approaches that have been proposed and implemented to do this. However, the effect of these schemes on subsequent turbulent filtering, and the further effect of the positioning of high order points on filtering, has yet to be researched. This optimisation of the solution point quadrature will also feed back into accuracy of the high order Euler scheme, which may go onto benefit the work proposed in the first paragraph.
To make the schemes attractive to industry, I would also propose researching the efficient porting of this scheme for heterogeneous computing. This is becoming increasingly important with programs being pushed towards use on more complex supercomputer architectures that utilise both CPUs and GPUs. Also, from my experience, high order schemes require this kind of hybrid approach for best speed up.