Measurements and LES calculations of turbulent premixed flame propagation past repeated obstacles
Introduction
Large Eddy Simulation (LES) is gradually replacing Reynolds-averaging (RANS) method of solving the Navier–Stokes equations to compute the structure of turbulent flames. Several recent works confirm the high fidelity nature of LES in predicting key characteristics of complex reacting flows including those of practical combustors [1], [2], [3], [4], [5]. The main attraction of LES lies in its ability to fully resolve features of the flow above a certain cut-off length scale hence, making it possible to compute transient dynamics; this being a clear advantage over RANS methods. The penalty, however, lies in the additional computational cost and the need to model the unresolved contributions, hence the issue of sub-grid-scale (SGS) modelling. This is, particularly, an important issue in LES modelling of turbulent premixed combustion given that chemical reaction occurs at the molecular level and hence needs to be modelled at the sub-grid scale.
A range of approaches to model combustion at the SGS are being pursued at varying degrees and relevance to the spectrum of turbulent combustion. The flamelet approach [6] was used by many researchers in the past in various forms [7], [8], [9] and, although limited to thin reaction zones, remains applicable to a wide range of applications. Recent developments of this approach involve flame generated manifolds (FGM) tabulated in terms of mixture fraction, reaction progress variable as well as other parameters such as a measure of flow strain [10]. Such formulations enable the application of flamelet modelling in premixed, non-premixed, as well as partially premixed flames. Two variations of the laminar flamelet approach are the flame surface density (FSD) where a transport equation for the FSD is solved [11] and the thickened flamelet model which has been applied successfully by Poinsot and co workers [12]. Recently, Di Sarli et al. [13], [14] demonstrated the importance of FSD based SGS model [1] to predict explosions in a vented chamber using LES. SGS modelling approaches that are seen as alternatives to flamelets include the filtered density functions (FDF) [15], the conditional moment closure [16] and the linear eddy model (LEM) [17]. Each of these approaches suffers from different limitations that are currently the subject of intense research. It is worth pointing that the combined LES/LEM is a truly multi-scale approach that is also receiving considerable attention.
In this paper the LES approach is used together with a recently developed dynamic flame surface density (DFSD) model [18], [19], [20] to compute turbulent premixed flames propagating in a laboratory scale combustion chamber containing a range of built-in solid obstructions. Earlier studies [20], [21] using the same DFSD model showed promising results in computing key characteristics of the propagating turbulent premixed flames but with only three selected configurations. In the present study, the main focus is to analyse the physics associated with flame–solids interactions and extend the calculations to a wide range of configurations to explore aspects such as the effects of location and number of the solid obstacles as well as area blockage ratio. The calculated results are validated against measurements taken from a novel experimental test facility [22], [23]. Eight different flow configurations are studied both experimentally and numerically. Results reported here also explore the effects of the resulting turbulence intensity on the structure of the reaction zone as well as the burning rate.
This paper is organised as follows: Section 2 briefly describes the experimental combustion chamber. Details of the newly developed SGS-DFSD model used in the LES calculations are outlined in Section 3. Numerical predictions for four groups of configurations are compared with available experimental data and reported in Section 4. Results are discussed highlighting the merits and drawbacks of the used model while discussing flame dynamics and behaviour in these groups of flow configurations. Finally general conclusions from the present investigation are summarised in Section 5.
Section snippets
Combustion chamber and test cases
The combustion chamber, shown schematically in Fig. 1a is only briefly described here and more details can be found elsewhere [22], [23]. It has internal dimensions of 50 × 50 × 250 mm giving an overall volume of 0.625 L. Up to three turbulent generating grids (also referred to as baffle plates or simply obstacles) may be placed in the chamber at 20 mm, 50 mm and 80 mm from the base. Each baffle plate consists of five strips, 4 mm wide, evenly separated by six gaps, 5 mm wide, thus creating an overall
Modelling and numerical issues
The governing equations and other numerical details associated with the LES model adopted in this paper are detailed elsewhere [18], [19], [20] and only a brief description is given here. A grid resolution of 90 × 90 × 336 (2.7 million cells) is adopted in the present calculations, as further refinement to 3.6 million cells shows no significant improvement in the results [19] for the present configuration. The filter width is calculated using a box filter, which is generally related to grid
Results and discussions
Results are presented in this section for the base flow configuration followed by those for the four groups of cases listed in Table 1. Calculations are compared with available averaged measurements which include pressure–time traces, mean and rms fluctuations of velocity, as well as high-speed video images of flame emission (2000 frames per second). It should be noted that the high-speed video images are not space-resolved and mark the leading edge of the propagating flame front. These are
Conclusions
Measurements and LES simulations have been carried out for propagating turbulent premixed flames in eight different flow configurations employing a newly developed SGS-DFSD model. A stagnant, stoichiometric propane/air mixture was used in the current investigation. All flow configurations were clustered into four groups based on the number and position of baffles, in order to understand the underlying complex mechanism of flame–flow–obstacle interactions. Main conclusions from the current study
References (37)
- et al.
Combust. Flame
(2002) - et al.
Proc. Combust. Inst.
(2005) - et al.
Exp. Therm. Fluid Sci. J.
(2006) - et al.
Combust. Flame
(2008) - et al.
Combust. Flame
(2001) - et al.
J. Hazard. Mater.
(2009) - et al.
Proc. Combust. Inst.
(1996) - et al.
Proc. Combust. Inst.
(1992) - et al.
Combust. Flame
(2009) - et al.
Proc. Combust. Inst.
(2002)
Combust. Flame
Combust. Flame
Proc. Combust. Inst.
Phys. Fluids
Ann. Rev. Fluid Mech.
Turbulent Reacting Flows
Phys. Fluids
Combust. Theory Model.
Cited by (58)
Experimental and numerical study on explosion behavior of hydrogen-air mixture in an obstructed closed chamber
2024, International Journal of Hydrogen EnergyLow Mach number lattice Boltzmann model for turbulent combustion: Flow in confined geometries
2023, Proceedings of the Combustion InstituteExperimental and numerical study of the influence of vent burst pressure on venting characteristic of hydrogen-air explosion
2022, Nuclear Engineering and Design