Parametric and model uncertainties induced by reduced order chemical mechanisms for biogas combustion

This study investigates the impact of chemical kinetic uncertainties on biogas combustion using a Uncertainty Quantification (UQ)-based methodology. The results indicate that the variation of physicochemical properties introduced by composition variability introduces smaller uncertainties in the res...

Descripción completa

Detalles Bibliográficos
Autores: Freitas, Rodolfo S.M, Rochinha, Fernando A., Mira Martínez, Daniel, Jiang, Xi
Tipo de recurso: artículo
Fecha de publicación:2020
País:España
Institución:Universitat Politècnica de Catalunya (UPC)
Repositorio:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglés
OAI Identifier:oai:upcommons.upc.edu:2117/366613
Acceso en línea:https://hdl.handle.net/2117/366613
https://dx.doi.org/10.1016/j.ces.2020.115949
Access Level:acceso abierto
Palabra clave:Machine learning
Biogas combustion
Model error
Surrogate modeling
Uncertainty quantification
Combustibles gasosos
Simulació per ordinador
Àrees temàtiques de la UPC::Informàtica::Aplicacions de la informàtica::Aplicacions informàtiques a la física i l‘enginyeria
Descripción
Sumario:This study investigates the impact of chemical kinetic uncertainties on biogas combustion using a Uncertainty Quantification (UQ)-based methodology. The results indicate that the variation of physicochemical properties introduced by composition variability introduces smaller uncertainties in the resulting flame properties than the Arrhenius parameters involved in the kinetics used to describe the oxidation process. We demonstrate that the use of reduced mechanisms for methane-air oxidation could be a starting point to develop optimized schemes for biogas combustion. In that regard, we adopted an embedded discrepancy approach to understanding the limits of the use of a reduced mechanism for methane/air in this renewable fuel. This strategy provides a way to reduce systematically the cost of reaction kinetics in simulations, while quantifying the accuracy of predictions of important target quantities. Finally, we develop a surrogate model for biogas flame propagation using machine learning techniques to make feasible a broader UQ analysis.