Session M2: Reacting Flows: DNS

DNS of cross sections of reacting jets with anisotropic turbulence forcing

Direct numerical simulations (DNS) of high Karlovitz number flames have been performed extensively in an inflow/outflow configuration, but in the absence of mean shear. Without a mean shear to sustain turbulence, the turbulent kinetic energy decays in the domain. Hence, a linear turbulence forcing has been used in previous simulations to emulate the missing shear effects. The turbulence forcing utilised in the current study is physically and mathematically consistent with the mean flow of an axisymmetric turbulent jet and consequently anisotropic, as opposed to the isotropic linear forcing used by Lundgren et al. Previous simulations used constant values for the forcing, calculated at either the jet axis or the middle of the mixing layer. The current study improves upon it by focusing on a cross section of the jet, where the forcing matrix changes with radial distance, as the mean flow changes with radial distance. Both non-reacting and reacting flows are considered, and the velocity fluctuations are calculated and compared with that of experimental turbulent jets. The flame structure, speed and surface area are observed for the reacting case and compared with results from previous simulations and experiments of turbulent jet flames.   

Burning velocity and flame surface area in high Karlovitz number flames

Accurate knowledge of the burning velocity of turbulent flames is of importance for many combustion devices. For low Karlovitz number flames, Damkohler proposed that the ratio of turbulent to laminar flame speed is proportional to the ratio of turbulent to laminar flame surface area. In recent DNS studies, it has been observed that Damkolher's scaling for low Karlovitz number flames still holds for high Karlovitz number flames. However, recent experimental studies have reported notable differences between global burning velocities and flame surface area measurements. In this work, the numerical and experimental results are further analyzed to explain the apparent contradiction. Emphasis is placed on identifying and quantifying potential experimental limitations at high Karlovitz numbers. More specifically, experimental flame surface measurements typically use binarized PLIF images. These images are two-dimensional and their resolution is limited by that of the PLIF system. The implications of using a two-dimensional iso-contour and the effects of the image resolution are assessed through post-processing of DNS datasets. Furthermore, the effects of integral length scale, Karlovitz number, and differential diffusion on the flame surface area are considered separately.   

A Structure Function Analysis of Intermittency and Universality in Turbulent Premixed Flames

In recent years, turbulence-flame interactions have been explored using both spectral and physical space approaches, but to date there remains limited understanding of how turbulence spectra and scales of motion vary throughout the flame structure (i.e., through premixed flamelets). We shed light on this question using a recently developed structure function analysis to analyze a range of premixed reacting flows. High-resolution direct numerical simulations (DNS) allow for a flame-normal structure function to be calculated locally, capturing the effect of heat release on the turbulent scales of motion and energy spectrum. We use this analysis to examine whether a universal relation for such structure functions can be derived through non-dimensionalization that is fully consistent throughout the flame, or whether different flame regions (e.g., the preheat and reaction zones) obey different laws. The same techniques are applied to a set of turbulent premixed flames with varying chemical kinetic mechanisms to further investigate universality across fuel types. The resulting implications for intermittency and universality in premixed reacting flows are discussed.   

The evolution of the flame surface in turbulent premixed jet flames at high Reynolds number

A set of direct numerical simulations of turbulent premixed flames in a spatially developing turbulent slot burner at four Reynolds number is presented. This configuration is of interest since it displays turbulent production by mean shear as in real combustion devices. The gas phase hydrodynamics are modeled with the reactive, unsteady Navier-Stokes equations in the low Mach number limit, with finite-rate chemistry consisting of 16 species and 73 reactions. For the highest jet Reynolds number of 22 $\times$ 10$^3$, 22 Billion grid points are employed. The jet consists of a lean methane/air mixture at 4 atm and preheated to 800 K. The analysis of stretch statistics shows that the mean total stretch is close to zero. Mean stretch decreases moving downstream from positive to negative values, suggesting a formation of surface area in the near field and destruction at the tip of the flame; the mean contribution of the tangential strain term is positive, while the mean contribution of the propagative term is always negative. Positive values of stretch are due to the tangential strain rate term, while large negative values are associated with the propagative term. Increasing Reynolds number is found to decrease the correlation between stretch and the single contributions.   

Turbulence-flame interaction in high Reynolds number premixed combustion

A set of four Direct Numerical Simulations of premixed flames is performed to investigate the effects of the Reynolds number on the interactions between the dynamics of the flame and turbulence. The Reynolds number based of the jet velocity and width ranges from 2800 to 22400, while the Karlovitz number is approximately 40 and is maintained constant for all the simulations. A 16 species skeletal mechanism for methane-air flames is employed and grids with up to 22 Billion grid points are used. Different aspects are investigated, including the flame structure and flame speed, the local statistics of scalar gradients, and the scaling of the fundamental turbulence properties with the Reynolds number. It is found that the structure of the flame, e.g., statistics of heat release rate conditioned on temperature, is very close to that observed in a one-dimensional flamelet and does not vary significantly with the Reynolds number. The analysis of the Reynolds number scaling of the Kolmogorov and integral scales reveals that the influence of combustion and heat release on the smallest scales of turbulence decreases as the Reynolds number increase.   

Investigation of Turbulent Hydrogen Premixed Flame Topologies at Different Combustion Regimes Using Computational Singular Perturbation

The investigation of turbulent flames at higher Reynolds and Karlovitz numbers has been gaining research interest, due to the advances in the computational power that has facilitated the use of direct numerical simulations (DNS). One of the additional challenges associated with highly turbulent premixed flames is the difficulties in identifying the turbulent flame topologies as the flame structures become severely corrugated or even disrupted by the small scale turbulent eddies. In these conditions, the conventional methods using a scalar iso-surface may lead to uncertainties in describing the flame front dynamics. In this study, the computational singular perturbation (CSP) is utilized as an automated tool to identify the flame front topologies based on the dynamical time scales and eigenvalues. In particular, the tangential stretch rate (TSR) approach, an extended generalized method to depict the dynamics of chemical and transport processes, is used for the flame front identification. The CSP/TSR approach and tools are used to compare the flame fronts of two turbulent H2/air premixed flames and to identify their similarities/differences, from a dynamical point of view. The results for two different combustion regimes are analyzed and compared.   

DNS of spherically expanding turbulent premixed flames in homogeneous isotropic turbulence

Spherically expanding turbulent flames of lean methane air premixed mixture are studied at elevated pressures and temperatures. A methane/air mixture with an equivalence ratio of $0.7$ is considered at $4$ atm with an unburnt temperature of $800$ K. Direct numerical simulations of spherically expanding flames in decaying homogeneous isotropic turbulence are conducted. The simulations have Reynolds numbers based on the Taylor microscale varying between 60 and 90 and grids with up to 8 billion points. The observed enhancement of the burning rate is linked to the increase in flame surface area through turbulent wrinkling. The Surface Density Function formalism is used to quantify the evolution of the flame surface area. Specifically, the role of turbulent stretching on surface generation and that of diffusion on destruction are assessed.   

Lagrangian Enstrophy Dynamics in Highly Turbulent Premixed Flames

Turbulent combustion is a multi-scale and multi-physics problem depending upon both chemical and fluid dynamic processes. These processes are often examined using an Eulerian framework, but recently the Lagrangian framework, a long-time tool in non-reacting flow research, has become increasingly common for the study of turbulent combustion. The two analysis frameworks are in fact equivalent, with the only difference being a change in reference frame. In this study, a Lagrangian fluid parcel tracking algorithm is used to analyze the enstrophy (i.e., vorticity magnitude) dynamics in turbulent premixed reacting flows. The analysis of vorticity dynamics in the premixed flame case is based on data from a three dimensional direct numerical simulation of a premixed stoichiometric hydrogen-air flame in an unconfined domain. Enstrophy budget terms are tracked along Lagrangian trajectories as fluid parcels travel through the flame, with particular focus on understanding the dynamical causes of turbulence variations through the flame preheat and reaction zones with respect to both the fluid parcel and the flame. Additionally, the ability of trajectories to completely sample the flame is discussed.   

DNS of Multicomponent Spray Combustion

To examine effects of multicomponent transportation fuels on the evaporation and combustion, direct numerical simulations of turbulent spray flames are performed under consideration of preferential evaporation and combustion of a multicomponent surrogate fuel. Model approximations for the description of the liquid and gas phase are discussed to make the problem computationally tractable. Simulations were performed to examine effects of droplet size, diffusion models, global equivalence ratio and turbulent intensity. These findings are compared against results from previous work on multicomponent combustion in homogeneous environments. Relevant combustion regimes for multicomponent spray-flames will be presented by introducing additional spatial and time-scales arising from the competition between evaporation, turbulence and reaction chemistry.   

Near-wall flame propagation characteristics

Understanding near wall flame propagation behavior is important for turbulent combustion modelling including flame---wall interaction, but also could be useful as a robust experimental technique for the measurement of laminar flame speed. Previous studies have measured time-resolved CH fluorescence, based on which flame displacement speed is estimated as a function of time or distance from the wall. The present study is the sequence of the experimental work and is based on direct numerical simulations (DNS) (i) to investigate near wall flame behavior in the context of flame displacement speed at various definition in a detailed manner, and (ii) to explore the possibilities of utilizing these characteristics as an experimental technique as a substitute or complement for a conventional experimental technique such as a double kernel method.