Fundamental understanding and modelling of turbulent premixed flame wall interaction :a direct numerical simulation based analysis

Abstract

This thesis focuses on fundamental physical understanding and modelling of turbulent premixed flame-wall interaction by using Direct Numerical Simulation (DNS) data. Three-dimensional compressible simulations of turbulent premixed flame-wall interaction have been carried out for head-on quenching (HOQ) of statistically planar flames by an isothermal inert wall and also for oblique quenching of a V-flame by two isothermal inert sidewalls (top and bottom walls). Simulations have been conducted for different values of Damköhler, Karlovitz and global Lewis numbers (i.e. Da, Ka and Le), and the chemical mechanism is simplified by a single-step Arrhenius type irreversible chemistry for the sake of computational economy in the interest of a detailed parametric analysis. The flame-wall interaction has been characterised in terms of wall heat flux magnitude and wall Peclet number (i.e. normalised wall normal distance). It has been found that the maximum wall heat flux magnitude decreases, whereas the minimum wall Peclet number (which quantifies the flame quenching distance) increases with increasing Lewis number in the case of laminar head-on quenching of planar flames. However, the minimum wall Peclet number for Le < 1.0 turbulent premixed flames has been to be smaller than the corresponding laminar value, whereas the minimum Peclet number in the case of turbulent flames with Le ≥ 1.0 remains comparable to the corresponding laminar values. It has been found that heat loss through the wall and flame quenching in the vicinity of the wall significantly affect dilatation rate distribution in the near-wall region, and has influences on the behaviours of the invariants of the velocity gradient tensor, which in turn influences statistical behaviours of flow topology and enstrophy distribution in the near-wall region. The statistical behaviours of vorticity and enstrophy transports in the near-wall region and the distribution of flow topologies within the flame, and their evolution with flame quenching have been analysed in detail using DNS data, and important fundamental physical insights have been gained regarding the flame-quenching processes associated with the flame-wall interaction. The DNS data has been explicitly Reynolds averaged to analyse the statistical behaviours of turbulent kinetic energy, scalar variance, turbulent scalar flux, FlameSurface Density (FSD) and scalar dissipation rate (SDR) and their transport in the near-wall regions. It has been found that existing closures of these quantities do not adequately capture their near-wall behaviours and in this thesis modifications to the existing closures have been proposed based on a-priori DNS analysis to account for the wall effects in such a manner that the modified closures perform well both near to and away from the wall. Furthermore, it has been found both FSD and SDR based conventional reaction rate closures do not adequately capture the mean reaction rate close to the wall, and the current analysis offers alternative reaction rate closure expressions both in the contexts of FSD and SDR based modelling approaches. Thus, the current thesis offers a unified modelling strategy for premixed flame-wall interaction in the context of Reynolds Averaged Navier-Stokes (RANS) simulations for the very first time. Finally, in order to validate the findings based on simple chemistry DNS, a limited number of DNS calculations of head-on quenching has been conducted using a multistep chemical mechanism for methane-air combustion. It has been found that the statistics of wall heat flux magnitude and wall Peclet number obtained from detailed chemistry simulations are in good qualitative and quantitative agreements with the corresponding results from simple chemistry DNS. However, detailed chemistry DNS reveals the presence of heat release at the wall during early stages of flame quenching, whereas heat release remains identically zero at the wall for simple chemistry DNS. In spite of this difference, an FSD based reaction rate closure which was proposed based on a-priori analysis of simple chemistry DNS has been found to work also for detailed chemistry DNS data without any modification. This provides the confidence in the models which have been proposed based on the analysis of simple chemistry DNS data.