Buoyancy‐corrected k–ε models and large eddy simulation applied to a large axisymmetric helium plume

Abstract

SUMMARYThe present numerical study is focused on testing two different modeling approaches to simulate a large turbulent buoyant helium plume, in particular the near-field region. First, buoyancy-corrected k-models are applied in Reynolds-averaged Navier-Stokes (RANS) calculations, then large eddy simulation (LES) using a standard Smagorinsky model is examined. Good results are produced using the buoyancy-corrected models, in particular, excellent agreement is achieved for the radial profiles of the streamwise velocity. However, the predictions are very sensitive to the choice of the buoyancy constant, C 3 , in the models. The present LES calculations show that the puffing frequency is accurately predicted. Predictions for the time-averaged velocities are within experimental uncertainty at all locations. The predicted plume concentrations are in good agreement at the base of the plume, but the centerline values are overpredicted farther downstream. The higher-order statistics are best predicted with the finest mesh. A sensitivity analysis on grid refinement, values of the Smagorinsky constant and the Schmidt number are included.

Application of FDS and FireFOAM in Large Eddy Simulations of a Turbulent Buoyant Helium Plume

“…Overall, the quality of the results is comparable to results previously published in the literature [11,13] with other CFD packages, which is encouraging for the use of FireFOAM in future work for simulations of fire-induced flows. A recent, more extensive, numerical study with FireFOAM, [20], supports this claim.…”

“…By validating the code in a non-reacting plume, the effect of buoyancy-generated turbulence can be tested independent of the complexity introduced by turbulent combustion. Chung & Devaud et al [11] studied the near field of the helium plume with traditional eddyviscosity LES methods using FDS.…”

Large Eddy Simulations of CH4 Fire Plumes

“…Such a grid sensitivity study is presented at the end of the paper for completeness. In literature, P RI values ranging between 5-15 have proven to give satisfactory accuracy with an acceptable computational time for many fire scenarios [22], values up to 16 were used when simulating fire plume scenarios [38] while values up to 40 were reported to be sufficient for a 1-m in diameter turbulent buoyant helium plume [39]. In the present study, the P RI values were approximately between 84 (for Test #14)-102 (Test #17) which are sufficiently large enough to adequately capture the flow evolution and unsteady behavior of these fire plumes.…”

Experimental and numerical characterization of the flowfield in the large‐scale UW live fire research facility

“…The full details of the numerical code is described in [15], and here only its main features are briefly described. FDS has been previously used with success for a variety of reacting and non-reacting flows [16,17]. The filtering process in 545 LES introduces unclosed quantities that are not resolved, such as the subgrid Reynolds stress, the subgrid heat and mass flux, the combustion heat release rate and the radiation loss.…”