Large-eddy Simulation of Triangular-stabilized Lean Premixed Turbulent Flames: Quality and Error Assessment

Abstract

In this numerical study, an algebraic flame surface wrinkling (AFSW) reaction submodel based on the progress variable approach is implemented in the large-eddy simulation (LES) context and validated against the triangular stabilized bluff body flame configuration measurements i.e. in VOLVO test rig. The quantitative predictability of the AFSW model is analyzed in comparison with another well validated turbulent flame speed closure (TFC) combustion model in order to help assess the behaviour of the present model and to further help improve the understanding of the flow and flame dynamics. Characterization of non-reacting (or cold) and reacting flows are performed using various subgrid scale models for consistent grid size variation with 300,000 (coarse), 1.2 million (intermediate) and 2.4 million (fine) grid cells. For non-reacting flows at inlet velocity of 17 m/s and inlet temperature 288 K, coarse grid leads to over prediction of turbulence quantities due to low dissipation at the early stage of flow development behind the bluff body that convects downstream eventually polluting the resulting solution. The simulated results with the intermediate (and fine) grid for mean flow and turbulence quantities, and the vortex shedding frequency (f s ) closely match experimental data. For combusting flows for lean propane/air mixtures at 35 m/s and 600 K, the vortex 564 Flow Turbulence Combust (2012) 88:563-596 shedding frequency increase threefold compared with cold scenario. The predicted results of mean, rms velocities and reaction progress variable are generally in good agreement with experimental data. For the coarse grid the combustion predictions show a shorter recirculation region due to higher turbulent burning rate. Finally, both cold and reacting LES data are analyzed for uncertainty in the solution using two quality assessment techniques: two-grid estimator by Celik, and model and grid variation by Klein. For both approaches, the resolved turbulent kinetic energy is used to estimate the grid quality and error assessment. The quality assessment reveals that the cold flows are well resolved even on the intermediate mesh, while for the reacting flows even the fine mesh is locally not sufficient in the flamelet region. The Klein approach estimates that depending on the recirculation region in cold scenario both numerical and model errors rise near the bluff-body region, while in combusting flows these errors are significant behind the stabilizing point due to preheating of unburned mixture and reaction heat release. The total error mainly depends on the numerical error and the influence of model error is low for this configuration.

Reynolds-Averaged, Scale-Adaptive and Large-Eddy Simulations of Premixed Bluff-Body Combustion Using the Eddy Dissipation Concept

“…However, it is clearly seen that numerical solutions provided over-dissipative spectra in respect to the −5/3 slope. It is worth to notice that the present results were less dissipative than the LES data by Manickam et al [40] and reproduced the same trend as the measured spectrum by Sanquer et al [54].…”

“…The spectra obtained by LES of Manickam et al [40] and measured by Sanquer et al [54] were added to compare the present SAS results with others numerical solutions. A −5/3 slope is shown as well.…”

“…Such over-estimation could, possibly, be explained by setting of the Smagorinsky constant. Manickam et al [40] and Ma et al [36] have applied C s = 0.1 in their reactive simulations of the Volvo rig. Furthermore, as it was discussed by Ma et al [36], the Smagorinsky model is known to be sensitive to the level of heat release [11].…”

“…16 were normalized by the relevant vortex shedding frequency. For the sake of completeness, the reactive LES results by Manickam et al [40] were added, in spite of the fact that these results were calculated for the C2 case, as well as the measured spectrum by Sanquer et al [54]. The calculated spectra by LES and SAS models displayed a −5/3 region clearly.…”

“…Moreover, this case has been extensively used by other researchers to study turbulent premixed combustion (among of them [26], [36], [40], [51]). …”