Development and assessment of correlations for predicting stability limits of swirling flames

Abstract

Swirling flow systems are being used in many industrial processes like separation of particles in gas flows (cyclones), atomisation and spreading of liquids (water, oil) and fast mixing and high-intensity reaction in reactor systems as, e.g. stable and intense flames of swirl burners. The widespread use of swirl burners in the process and energy industries and, in particular, new concepts for the reduction of NO x-emissions (ultralean premixed combustion) raise the need for simple-to-use models for predicting lean stability limits of highly turbulent flames stabilized by internal recirculation. Based on recently published experimental data of the first author concerning the reaction structures of swirling flames operated near to the extinction limit, different methods for predicting lean blow-off limits have been developed and tested for different burner sizes and fuel gas compositions. The aim of the investigations was to find stabilization criteria that allow predictions of blow-off limits of highly turbulent recirculating flames without the requirement for expensive and time-consuming measurements in those flames. Several similarity criteria based on volumetric flowrates, burner size and material parameters of the cold gases, were found to be capable to predict stability limits of premixed and (in some cases) nonpremixed flames at varying swirl intensities, burner scales and fuel compositions. A previously developed numerical field model, combining a k,ε-model with a combined “assumed-shape Joint-PDF”/Eddy-Dissipation reaction model for the determination of the time mean reaction rates in turbulent flows was also tested for its potential for stability prediction. All the methods presented have specific advantages and limitations: the similarity criteria are restricted to geometrically similar systems, but they are easy to use, fairly precise and can take the detailled chemistry into account in an integral manner. The numerical field model necessitates large computational effort and is limited to very much simplified, global reaction mechanisms, but it offers the opportunity to make predictions for different burner geometries and swirl conditions.