• Citer

Optimisation of chemical mechanisms for MILD conditions

par Bertolino, Andrea ;Fürst, Magnus ;Parente, Alessandro ; [et al.]
Référence (6-8/05/2019: Aachen), Seventeenth International Conference on Numerical combustion (NC19)
Publication Publié, 2019-05-06

Abstract de conférence

• Accès en ligne
• Détails
• Contenu
• Statistiques

Résumé :

Moderate or Intense Low-oxygen Dilution (MILD) combustion is among the best new technologies for clean and efficient combustion in industrial applications. While large Damköhler numbers indicate mixing controlled flames, flameless combustion is characterized by unitary Damköhler numbers, and the presence of a relevant amount of diluents. Under these conditions, an overlapping between mixing and chemical time scales occurs. For this reason, modelling chemistry with a detailed mechanism is needed. Unfortunately, existing kinetic mechanisms accomplish a non-accurate estimation for this particular regime, even for fuels with a simple and well-known chemistry, such as hydrogen and syngas. Experimental data-driven optimisation for kinetic mechanisms is nowadays a promising technique to solve this problem.This work addresses the following questions: how would the choice of the kinetic mechanism impact its optimisation, if the latter is carried out using common experimental targets? Would the parameters to be improved similar? Would their values converge to similar optimal combinations? The most impactful reactions will be identified using local sensitivity analysis. Then, a local brute force sensitivity analysis on their Arrhenius parameters will help discarding automatically those parameters that are of minor relevance. Finally, an optimization based on evolutionary algorithm will be performed, coupling Dakota and OpenSMOKE++, which minimizes the mean squared error on the overall dataset to find the optimal parameter combination.