Numerical modelling of turbulent premixed combustion for gas turbine conditions with incorporation of molecular transport effects

Design of combustion systems with increased efficiency and reduced fuel consumption under controlled pollutant emissions is mandatory due to the fast depleting trend of the fossil fuel reserves, and environmental concerns. Premixed turbulent high pressure combustion is a practically viable option to tackle these issues, especially in relation with gas turbine combustion. The central theme of this research work is the numerical investigation of the molecular transport effects and the dynamics of turbulent premixed high-pressure flames. These elements of premixed turbulent combustion are exhaustively studied on five different flame configurations of varied degree of complexity, ranging from a simple Bunsen-like burner to an industrial gas turbine combustor.

The focus of this thesis is diversified on three subjects.

Firstly, the behaviour of various turbulent premixed combustion models for the variation of pressure and fuel types with a broad set of simple Bunsen-like flames are numerically tested, where the flow and turbulence field has a relatively simple structure and is calculated with the Reynolds averaged Navier-Stokes (RANS) approach. It is found that several of the existing reaction models are insensitive to the effects of pressure and fuel type. Therefore, a new reaction model is developed, being based on an Algebraic Flame Surface Wrinkling relation (AFSW model), which can describe well the broad set of over 100 Bunsen flame data. The fuel influence is modelled for several hydrocarbon fuels with a Lewis number effect, which shows that molecular transport effects are of importance even for high turbulence conditions. The AFSW model shows remarkable workability also for the other flame configurations, including the gas turbine combustors for pressure variation up to 32 bar. In a set of calculations of a gasturbine burner, it is found that the flame dynamics in conjunction with the vortex breakdown point is sensitive to the Lewis number (i.e., for fuel type). As an alternative reaction model, also the Lindstedt-Váos model is extended in a similar way with a pressure-term and the Lewis number, being described in the appendix.

Secondly, the applicability of the AFSW reaction model is tested in conjunction with more elaborate turbulence models, based on the time dependent large-eddy simulation (LES). Here, the AFSW reaction model was incorporated as a subgrid scale (sgs) reaction closure and was tested for three sgs turbulence models. Validation is done successfully against experimentally measured flame brush thickness and mean flame position on those flame configurations, where the turbulent flow pattern is rather complex with recirculation and swirl. This approach allowed for the first time the calculation and explanation of experimentally observed dual-flame instability of a specific gas-turbine burner.

Thirdly, a preliminary study is started to incorporate the possibility of hydrogen blended methane-air flames, which is of importance as a possible future fuel component, e.g., in the frame of reduced CO2 emissions. As the molecular weight and with that the diffusivity of hydrogen differs significantly from that of other fuels, this is a non-trivial challenge for any reaction model. In an analytical analysis and with limited computations in the RANS context, it is found that the AFSW model is insensitive to the preferential molecular diffusion effects, occurring here. As an outlook a submodel for the chemical time scale is proposed, based on a leading point concept of critically curved laminar flames. Further studies of this new approach are necessary for thorough validation.