| OCR Text |
Show Paper No. 19 PREDICTION OF FLAME SURFACE DENSITY SHAPE IN VARIOUS COMBUSTORS : Comparison with experimental measurements M.H. SENNOUN* and A. CHARETTE Universite du Quebec a Chicoutimi 555, bould de I'Universite, Chicoutimi, G7H 2B1, Quebec, Canada ABSTRACT The scope of this study is to show several applications of a new turbulent combustion model in various geometries. The model is described in Ref [19]. In this model the flame surface density (FSD) prediction is the most important parameter, because all the rest (fuel consumption rate, temperature, heat release rate, etc..) depends on this quantity. A good prediction of flame shape and position is the key of the success of this kind of model. In this paper w e show, for various combustors, the position where combustion takes place, the shape of the reaction zone which represents the interface between fresh gases and hot combustion products. In all these combustors, we have carried out numerical simulation in two-dimensionnal geometries for premixed propane-air turbulent flames. The results and comparisons are presented and commented. INTRODUCTION All the actual efforts in the turbulent combustion modelling using the flamelet approach and the flame surface density (FSD) concept are focused on the modelling of the source and sink terms of the F S D governing equation. Expressions for these source terms are assembled in Duclos et al. [5]. This reference also contains a comparison of a set of F S D models in the case of one-dimensional premixed turbulent flame propagation in frozen turbulence. In all these models the mean consumption rate of fuel (the mass of fuel burnt per unit time per unit volume) is expressed as : CJJU = p^UiX' , where po is the density of the fuel in the fresh gases, U L the mean consumption speed along the flame front and E the flame surface per unit volume (m2/m3). The quantities U L and £ have to be modelled, and this leads to various degrees of complexity in flamelet models. Originally developed by Marble and Broadwell [11] for non premixed flames, the Coherent Flame Model (CFM) describes the turbulent reaction field as a collection of laminar flame elements. These elements are convected and distorted by the turbulent motion but retain an identifiable structure (the flamelets remain 'coherent', i.e. 'organized'). The F S D description has been extended to premixed flow configurations by Candel et al. [3]. There are several approaches to the flamelet description. However they all share the following ingredients : 1) A laminar flamelet submodel providing the structure and properties of the strained reactive elements; 2) A description of the turbulent flow comprising mass average equations describing the mean flow variables and the main species mass fractions and relevant closure equaUons; 3) A set of rules which couple the flamelet submodel to the turbulent flow description. In the Coherent Flame Model ( C F M ) (Marble and Broadwell [11] and Candel et al. [3]), a balance equation for the flame surface density (FSD), accounting for transport, diffusion, producuon and destruction of flame area, plays the role of an interface between the local flamelet model and the global mean turbulent flow model. The F S D balance equation takes into account transport, diffusion, producdon and destruction of flame area, and it is cast in the following general form : {transport} = {turbulent diffusion} + {production} - {destruction}. Initially devised on the basis of some intuitive arguments, the balance equation of flame surface is actually derived from basic principles (Candel and Poinsot [2], and Pope [17]). The balance equation for the flame area starts from the basic equation for a material surface per unit volume E . Replacing the flame surface density E by pS( and introducing the mass averaged decomposition of the different variables, one gets : d(pSj) | d(pujSt) = a (p.. asf\ p D dt dxi dxi \CTSJ dxi) corresponding author. 1 |
