Flame surface density modelling for the large eddy simulation of turbulent premixed flames

Large Eddy Simulation (LES) has become an increasingly useful tool for the prediction of turbulent reactive flows with the increasing availability of cheaper and faster computing power. In the context of premixed combustion, LES encounters the challenge of resolving the flame thickness, which is nor...

Full description

Bibliographic Details
Main Author: Ma, Terence Kwai Kin
Other Authors: Van Wachem, Berend ; Marquis, Andrew
Published: Imperial College London 2013
Subjects:
621
Online Access:http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.616834
Description
Summary:Large Eddy Simulation (LES) has become an increasingly useful tool for the prediction of turbulent reactive flows with the increasing availability of cheaper and faster computing power. In the context of premixed combustion, LES encounters the challenge of resolving the flame thickness, which is normally smaller than the filter width used in typical engineering applications. This thesis considers the Flame Surface Density (FSD) approach to provide closure to the filtered LES reaction rate. The FSD can either be modelled algebraically (FSDA) or determined through a transport equation (FSDT) and both approaches are investigated in the LES of three different test cases. The first case explores the response of different FSDA models towards changes in turbulence levels, and compares the instantaneous flame structures and reaction rates predicted by FSDA and FSDT methods. The remaining cases examine the LES of two turbulent premixed burners. A relatively large range of FSDA models are tested under the same operating conditions for the first time, and the LES-FSDT equation is applied to premixed flames that involve a higher level of geometric complexity than earlier work. Generally, the results show that the performance of some FSDA models are inconsistent between the two premixed burners, suggesting that the models may operate optimally under different turbulent conditions. By contrast, the consistently good agreement of the FSDT results with experiments suggests that the method has much potential in the LES modelling of turbulent premixed flames. However, the improved FSDT predictions were dependent on the value of the model constant within the sub-grid curvature model, and the value yielded an additional dependency on filter width. For these reasons as well as for the higher computational expense, the effective use of FSDT requires further development, while the application of the FSDA models remains a viable alternative to the FSDT approach.