The need to design of efficient combustion systems with minimal emissions of pollutants has led to the development of large detailed reaction mechanisms for combustion involving hundreds of chemical species reacting in a complex network of thousands of elementary reactions. Incorporating such a detailed reaction mechanism into multidimensional simulations is practically impossible. Different methodologies have been proposed for the reduction of detailed mechanisms. In the present work, model reduction approaches based on timescale separation and thermodynamic analysis are revisited, introduced, validated and used.