This thesis presents a numerical investigation into the combustion characteristics of hydrogen-enriched fuels using large-eddy simulation. Flame instabilities and fuel variability effects have been studied, which are some of the main concerns on applications of burning hydrogen mixtures in the power industry. The present work investigates these effects on burners of theoretical and practical interest. The role of instabilities is addressed in an annular flame with possible swirl motion, while the influence of fuel variability in the flame dynamics is described in an impinging configuration. The study shows that buoyancy plays an important role in the dynamics of hydrogen flames, promoting flow acceleration and creating large scale vortical structures. The azimuthal momentum introduced by the swirl motion leads to a large flame spreading and shortening. The effects of swirl induce the development of a toroidal recirculation zone, which enhances the mixing process and combustion. In addition, the swirling flames exhibit a vortex breakdown bubble that vanishes allowing the formation of a large central recirculation zone when the swirl number is increased. The fuel variability analysis shows that when the hydrogen content is increased in the fuel mixture, the flame becomes less vortical and wrinkled due to the more diffusive and less viscous flow field. Lower burning speeds and lower temperature values are found for hydrogen-leaner mixtures. Different finite-rate chemical kinetic mechanisms are also considered and their effects on the flame dynamics are addressed. These results show that some reduced models have a damping effect on turbulence due to the dissipation caused by an excess of heat-release and a poor prediction of the flame /turbulence interactions. The subgrid scale model implementation includes a one-equation model to compute the unresolved momentum transport, while two different approaches are considered to account for the subgrid scalar transport. These two models include a simple closure using the eddy diffusivity method, which is based on a gradient diffusion closure and the linear-eddy model, which is based on the one-dimensional turbulence theory. Comparison against experimental data is performed to validate the models for combustion applications. It is found that the gradient diffusi0l1 approach is not valid to represent the subgrid scalar transport for fuels containing hydrogen. The model predicts unphysical temperature fields suggesting that an alternative model must be used instead. Results using the linear-eddy model show a more accurate representation of the temperature field across the shear layers and better agreement with the experimental results.