In modern aero-engine combustors combustor tiles are used to protect the walls from the hot gases, the temperature of which is rising in new engines due to increasing pressure ratios. However, the amount of air used for wall cooling should be reduced to allow for maximal air flow rates through the fuel injector. This measure enables optimised lean combustion with lowest pollutant emission rates. This objective can be achieved by combining effusion cooling on the hot side with impingement cooling on the cold side of the tiles. This complex system needs to be simulated during design processes.

This project aimed to improve the predictive capabilities and decrease the uncertainties of current models regarding wall temperatures and thermal stresses. The model development was supported and the emerging method was validated by high-quality experimental data obtained from measurements on an engine-representative gas turbine combustor using Particle Image Velocimetry, Thermographic Phosphor Thermometry and Coherent anti-Stokes Raman Spectroscopy.

An iterative method was proposed which couples tabulated chemistry based CFD and finite element method (FEM) simulations. In the CFD calculations, previously ignored flame-wall interactions were considered by adjusting turbulence models and extending the tabulation method to non-adiabatic conditions. Results of highly resolved large eddy simulations were used to improve the computationally efficient RANS based techniques. The CFD calculations provided the convective heat transfer for the FEM simulations as a boundary condition. For an accurate prediction of the metal temperature – which is then fed back into the CFD part - and thermal stresses provided by the FEM, a probabilistic approach was applied. A Monte Carlo method with a meta-model was used to evaluate the thermal stochastic output improving the current state-of-the-art of thermal predictions.