Optimization of Chemical Kinetic Mechanisms and Numerical Simulations of Industrial Gas Turbine Burners

Doktorsavhandling, 2014

Increasing demands on environmentally friendly energy conversion indicate that pollutants need to be reduced. Computational Fluid Dynamics is efficiently used today in turbulent reaction flow simulations to gain a deeper understanding of the combustion process and consequently be able to improve combustor design and reduce emissions. This thesis aims to investigate recent developments in turbulent reacting flow in the context of engineering type numerical analysis tools, applied to the main problem area of swirl-stabilized flexi-fuel flames. Combustion simulations involve a strong coupling between kinetics, transport and turbulence and their interactions. Fully detailed kinetic mechanisms are expensive in terms of computer time when coupled with Computational Fluid Dynamics, and the complexity of combustion chemistry must hence be downsized. Chemistry look-up tables and reduced reaction mechanisms are two well-known methods used today. Compared to chemical look-up tables, the use of a global mechanism involving a limited number of species is of obvious practical interest, specifically when dealing with complex burner geometries in the industrial applications featuring multiple inlets, dilution with burnt gases in recirculation zones and heat loss. The proposed optimization methodology for global schemes is based on the 1D balance equations for species mass fractions and temperature, and also laminar flame speed involved in the reduced mechanisms correctly reproducing the detailed chemistry solution. The optimizations are performed using a set-up of different tools, such as CANTERA, CHEMKIN and modeFRONTIER. A new approach to sub-grid scale modeling is also suggested, which relies on the idea that effects and properties of the spatial filtering, inherent to Large Eddy Simulation solvers, could be directly accounted for when reducing chemistry and molecular transport properties. The optimized global schemes have been implemented in the CFD toolbox Ansys CFX and used in numerical simulations (RANS, URANS/LES and LES turbulence models) of two different partially premixed industrial gas turbine combustors and the test case of Sandia Flame D. The focus has been on grid generation and modelling of the CFD domain. Measured and predicted flow and flame dynamics, averaged temperature and species concentrations compare well for the different set-ups.