Experimental and Numerical Study of a Turbulent Recirculation Zone with Combustion

Abstract

Recirculation zones are widely used for the stabilization of combustion in flows with a large velocity. The main example of this is the case of afterburner devices for turbojet engines, where the turbulent flame is stabilized by bluff bodies behind which recirculation zones occur. The prediction of the general shape of the flow field and of the temperatue field of these recirculating flows is of primary importance for the prediction of the "stability domain" of the burners. Two properties of the recirculation zone are of particular interest for combustion stabilization: its volume and the mass flow rate exchanged with the external flow. These two characteristics control the residence times of the fluid particles within the recirculation zone, which has to be large enough to ensure a good stabilization of the combustion.Without combustion, the prediction of turbulent separated flows behind obstacles is not simple. On one hand the flow is governed by elliptical equations, and on the other hand the turbulence is very complex, non homogeneous, non isotropic and with large curvature of the streamlines. The first attempts to predict by a numerical calculation and a turbulence model such a recirculation zone pointed out some difficulties: a simple and global characterristic, like the recirculation length, has been found always (about 20%) ). It seems, however, that an improvement of the k -e turbulence model taking into account the curvature of the streamlines leads to more satisfactory agreement [2]. Such a prediction, even imperfect in its details, particularly concerning the components of the turbulence kinetic energy, can be of large interest for engineering purposes.What is the situation for recirculation zones behind bluff bodies in combusting flows? It appears that their calculation, including already known turbulence models as well as already known turbulent combustion models, is fully possible. In fact, many examples of such a calculation are existing in the literature ([3] for instance); much more complex flow field have been attacked for numerical prediction also (see [4]). But in simple cases where detailed experimental results are available, the situation is not at all clear. For instance, concerning the recirculation zone behind bluff bodies, it is not clear at the present time whether the combustion increases the size of this zone, or decreases it, with respect to the non burning cases. The first experiments of Winterfeld [5] and Clare et al. [6] have shown an increase of this length, while the ones of Grindley reported in [3], and Pitz and Daily [7] show the opposite. The studies of Taylor and Whitelaw [8], and Fuji and Egushi [9] show both an increase of the length.Indeed, it is easily understandable that combustion can have opposite influences: i) in the case of a confined duct, the combustion behind the bluff body leads to an increase of the gas specific volume, which, expectedly, produces a flow acceleration, but, less expectedly, compresses the recirculation zone upstream and decreases its len...

Numerical simulation of a backward-facing step combustor using reynolds-averaged navier–stokes / extended partially stirred reactor model

“…The geometry of the §ow problem was chosen in accordance to the experimental setup of ONERA [19,22] of the reference frame X = Y = 0 is ¦xed at the nose of the step. Figure 2 represents schematically the main features of the §ow.…”

“…The proposed RANS/EPaSR model is validated against experimental database created at ONERA for an airmethane premixed §ame stabilized by a backwardfacing step combustor data [22,23]. The RANS/EPaSR model is compared with a few contemporary RANS combustion models such as QL RCM, PFTC no TCI, and PFTC β-PDF.…”

“…It is seen that the mean reattachment of a separated §ow behind the step numerical results depends on the used turbulence model. Table 2 presents the comparison of the calculated mean reattachment lengths X r /h obtained with di¨erent turbulence models and the experimental data [22,23] (the error of estimation of X r is about ±0.2h). One can conclude that the kl turbulence model gives the best agreement with the experimental data.…”

“…6) and QL RCM and PFTC (no TCI) models (see Fig. 7) with experimental data [22,23]. In the case of the EPaSR model, the calculations were performed with the di¨erent de¦nitions of the micromixing time τ * : Figure 6 Pro¦les of Favre-averaged temperature "…”

Automatic Hybrid RANS/LES Strategy for Industrial CFD

“…The first test case is a geometrically 2D backward facing step with an upper wall experimentally studied by Moreau et al [8]. The geometrical features and measures of the computational domain used in this study are detailed in Fig.…”