5 years ago

chapter 5 turbulent diffusion flames - FedOA

chapter 5 turbulent diffusion flames - FedOA

Flame Fuel Fig. 2.2

Flame Fuel Fig. 2.2 Turbulent jet diffusion flame configuration. Due to this complexity, many existing turbulent combustion models have been based on a simplified view of chemistry and diffusion processes. The three main classes of numerical tools usually employed in combustion research [49]: Direct Numerical Simulations (DNS), Reynolds Averaged Navier–Stokes (RANS) computations and Large-Eddy Simulations (LES). The DNS approach consists in solving exactly all the physical spatial and time-scales embedded in the representative flow equations, without any model for turbulence. This method generally requires prohibitive numerical costs. RANS equations are restricted to a description of the mean flow field. Mean transport equations are obtained by averaging in time or over ensembles the instantaneous balance equations. This operation yields unclosed quantities, representative of the turbulent fluctuations, that must be modeled [49]. The RANS approach is presently the only one really suited for the simulation of practical configurations, but its accuracy is limited and it cannot be used in a predictive manner, but only to identify major trends. Large Eddy Simulations can be seen as an intermediate between DNS and RANS. In LES, the largest structures of the flow field are explicitly computed like in DNS whereas the effects of 34

small-scale structures are modeled. LES simulations are thus, by construction, more expensive than RANS calculations, but faster than corresponding DNS. The balance equations for LES are obtained by filtering spatially the instantaneous balance equations. LES are already widely used for non-reacting flow simulations but are still at an early stage for turbulent combustion. Note that, for all three numerical modeling approaches, a correct description of the behaviour of turbulence requires three-dimensional (3D) calculations, and therefore high computing times. This constraint can be released for RANS in the presence of statistically homogeneous directions. This ideal case is, however, not often found in industrial configurations. Furthermore, all modelling works in turbulent jet diffusion flames need to be studied experimentally to provide data such as combustion by-product concentrations. This will assist in the model’s development and validation. During the thesis two burner configuration have been realized to obtain turbulent jet diffusion flames of ethylene and methane, Fig. 2.3. 50 mm Fuel 30 mm 4 mm (a) (b) Fig. 2.3 Burner configuration for turbulent jet diffusion flames (a), turbulent jet diffusion flame of ethylene (b). 35

DNS of Turbulent Nonpremixed Ethylene Flames
Heat release rate measurement in turbulent flames