CFD methods development and validation

Often this is the routine use of established CFD methods to support the design and analysis of experimental work. However, it also encompasses the development of new modelling techniques to allow the accurate simulation of the complex aerodynamic phenomena present in the combustion system. This is a significant challenge; a single modelling approach is not applicable across the broad range of flow physics. Consequently hybrid techniques are being developed which allow coupling of different turbulence models and flow solvers across different engine components.

Commercial and proprietary in-house codes are used for grid generation, analysis and post-processing, with a range of grids and turbulence models providing steady state or time dependent solutions. A 1956-core high performance PC cluster provides a parallel computing platform within the Department of Aeronautical and Automotive Engineering, and a larger 3000-core platform is available through the EPSRC University High Performance Computing (Midlands) service. The available computing power is therefore sufficient for large LES time dependent solutions to be run within reasonable time scales.

Methods have been developed within the UTC to simulate the primary breakup of liquid fuel. A hybrid Coupled Level Set Volume of Fluid (CLSVOF) method has been established which combines the liquid surface tracking accuracy of level set methods with the mass conservation of VOF. This allows accurate simulation of two-phase flows while reducing the number of computational cells.

Combustion models are also being developed which will allow Large Eddy Simulations (LES) to capture the complex combination of turbulence and chemistry which takes place in gas turbine combustors. Unlike many existing models these are being developed with the aim that they can be used with complex real-world, rather than idealised, geometry.

Other CFD work has studied the effect of numerically forcing combustion calculations in order to extract ‘flame describing functions’ and help understand unsteady heat release and combustion instabilities. Further work to understand and model the unsteady flow in the combustor has used pressure based compressible unsteady RANS (uRANS) techniques. These have the advantage of relatively low computational cost (compared to density based or LES codes) for this type of problem allowing them to be applied to realistic problems. Additionally, by developing appropriate acoustic boundary conditions it has been possible to obtain excellent agreement with experiments for the unsteady aero-acoustic response of both simple and complex geometries.