Session P03: Reacting Flows: DNS and LES (3:10pm - 3:55pm CST)

Shock-induced Ignition in 2D Shock Turbulence Interaction

Shock induced ignition in the shock-turbulence interaction (STI) setting is studied with 2D direct numerical simulation (DNS) using finite-rate detailed chemistry. Four reactive STI simulations are conducted with initial Taylor microscale Reynolds number $Re_{\lambda}=(1250, 600)$ and turbulent Mach number $M_t = (0.1, 0.3)$ in the inflow turbulence at shock Mach number $M = 2$ with oxygen, hydrogen and argon mixture, and compared with laminar simulation at the same Mach number. Compressibility of the upstream turbulence leads to earlier ignition compared with laminar simulation, and the peak values of thermodynamic quantities at the flame front are found to be lower for the turbulent cases under consideration, owing to the partially premixed nature of the mixture. An analysis of the temporal evolution of the flame regime reveals that the reaction happens mostly in the thin reaction zone regime, which is characterized by a broadened preheat zone. Lower $M_t$ brings a slightly higher probability that the combustion happens in the regime of corrugated flamelets. The time evolution of the Takeno Flame Index (TFI) indicates that, as the flame propagates upstream, the combustion becomes increasingly non-premixed because of the larger variance of the species mass fractions.   

Combustion of droplets in turbulence

The combustion of droplets in turbulence involves complex, interacting phenomena. First, the droplets cluster in regions of low vorticity and modulate the velocity fluctuations of the gas. Second, the droplets release vapor and energy, which then mix with the gas. Third, the vapor reacts with the oxidizer creating diffusion flames, which surround individual droplets or droplet clusters, and premixed flames, which propagate through mixtures of vapor and oxidizer.\footnote{Jenny, Roekaerts, and Beishuizen, \emph{Progress in Energy and Combustion Science} \textbf{38}, 846-887 (2012).} \par These phenomena are investigated with direct numerical simulations. The gas phase is modeled with the low Mach number approximation, and the droplets are modeled as point droplets.\footnote{Weiss, Giddey, Meyer, and Jenny, \emph{Physics of Fluids} \textbf{32}, 073305 (2020).} The combustion of vapor and oxidizer is modeled with a one-step reaction mechanism. Droplet clusters are analyzed with Vorono\"{i} tessellations, and diffusion and premixed flames are identified with Takeno's flame index.\footnote{Yamashita, Shimada, and Takeno, \emph{26th Symposium on Combustion}, 27-34 (1996).} Simulations with different droplet diameters, droplet number densities and turbulence intensities are performed.   

Effect of flow topology on enstrophy production and scalar structures in chemically reacting compressible isotropic turbulence

Flow topology in hydrogen-air chemically reacting compressible isotropic turbulence is studied by using numerical simulations at turbulent Mach numbers ranging from 0.2 to 0.8. Various statistical properties of eight flow topologies based on the three invariants of velocity gradient tensor are investigated with a specific focus on the effect of flow topology on enstrophy production and scalar structures. It is found that the topologies unstable focus/compressing ($UFC$), unstable node/saddle/saddle ($UN/S/S$) and stable focus/stretching ($SFS$) are predominant flow patterns. The volume fractions of the topologies stable node/saddle/saddle ($SN/S/S$) and unstable node/unstable node/unstable node ($UN/UN/UN$) increase apparently at low turbulent Mach number $M_{t}$=0.2 due to strong heat release. The topologies $UN/S/S$ and $SFS$ have major contributions to the overall enstrophy production. Moreover, the curvatures of temperature isosurfaces are studied. It is found that during the reaction process, the magnitude of average Gauss curvature increases evidently at $M_{t}$=0.2 and 0.4, resulting a predominant saddle scalar surface geometry. The topologies $UN/S/S$ , $SFC$ and $UFS$ have major contributions to the increase of mean and Gauss curvatures.   

Pressure scaling of a reheat flame structure

Longitudinally staged gas turbine combustors have gained significant interest in power generation for their ability to achieve low emissions, high efficiency and fuel flexibility under a wide range of operational conditions. The combustion properties of the so-called reheat flame, stabilized in the second stage combustor, play an important role in achieving the desired operational characteristics. Recently, a 3D direct numerical simulation investigated stabilization of a reheat hydrogen flame at atmospheric pressure in order to characterize the modes of combustion. Since the operating pressure in industrial combustors is between 15 and 30 bar, this work builds upon and proceeds beyond the mentioned earlier effort to understand the pressure scaling effect on the stabilization of the reheat flame using 2D simulations. The computational domain consists of a mixing duct followed by a sudden area-change, into the combustion chamber. Preliminary results show that at higher pressures the flame stabilization location is increasingly sensitive to perturbations in the reactants' pressure/temperature, and can easily transition to an unstable state of combustion characterized by spatial oscillations. Further, the flame structure and the role of auto-ignition are quantified using CEMA.   

Spectral Element DNS of a Laboratory-Scale Non-Premixed Jet in Vitiated Turbulent Cross-Flow

This work presents a direct numerical simulation (DNS) study of a laboratory-scale non-premixed jet in vitiated turbulent cross-flow (JICF), using a highly scalable spectral element CFD code, Nek5000. The computational domain targeting an experimental configuration is larger than those in previous DNS studies, allowing for investigation of practical turbulent flow patterns. A non-reacting JICF DNS is first performed using the mixture-averaged molecular transport model to account for differential diffusion. Two-dimensional velocity and vorticity profiles from the simulation are compared with experimental measurements and show good agreement. Three-dimensional vortex structures and their formation mechanisms are analyzed in detail using the high-fidelity data. A reacting JICF DNS with a diluted hydrogen jet is then performed using a detailed chemical kinetic mechanism. The operator-splitting scheme along with an advanced chemistry solver are employed to accelerate the simulation. Predicted flame structures in both windward and leeward sides of the reacting jet are compared with experimental OH PLIF measurements. Effects of flames on the vortex structure are finally discussed by comparing the non-reacting and reacting JICF simulations.   

Direct Numerical Simulation of Multi-Injection Mixing and Ignition in Diesel Engine Environments

Multi-injection strategies are known to help reduce pollutant emissions and improve ignition reliability, among other benefits. Different parameters such as dwell time, injection duration and environment temperature govern the overall mixing and ignition sequence. While experiments provide valuable insight, a fundamental understanding on how multiple injections interact and mix, and how this influences low-temperature ignition and combustion of large alkanes is still lacking, especially at low-temperature conditions. The effect of ambient temperature will be highlighted for two Engine Combustion Network Spray A cases at 750~K and 900~K conditions, both at 60 atm and with 15\% O2 in the oxidizer, using Direct Numerical Simulations (DNS) of a simplified and downscaled configuration. At 900~K, the first injection ignites before the second injection starts. At 750~K, the first injection only provides low-temperature intermediate species that accelerate the ignition of the second injection. The cases are compared and differences highlighted in the overall ignition sequence, combustion modes (ignition versus flame propagation), displacement speed statistics, flame topology and modeling challenges.   

Direct Numerical Simulations of Blended Ammonia/Hydrogen/Nitrogen Premixed Flames in Intense Sheared Turbulence

Ammonia is being considered as an attractive carbon-free energy carrier. While hydrogen presents a promising carbon-free natural gas replacement, ammonia offers advantages in terms of storage and transport. Challenges, however, include neat ammonia not providing suitable flame properties (e.g., flame speed) and the generation of NOx pollutants. Ammonia flame properties can be adjusted by partial fuel cracking to provide an ammonia/hydrogen/nitrogen mixture. For ammonia/hydrogen/nitrogen blends, the amount of NOx released strongly depends on equivalence ratio. A fundamental understanding of turbulent flame properties and NOx generation mechanisms of such blends in turbulent conditions is still missing. We address this using Direct Numerical Simulations of temporally-evolving turbulent sheared flames at different equivalence ratios and pressures. In addition, we compare a baseline case to a case with natural gas (i.e., methane) at nominally similar conditions, highlighting differences in turbulent flame behavior. A statistical comparison of the cases is presented in terms of their displacement speeds, flame surface density statistics and chemical explosive mode analysis highlighting relevant chemical pathways and combustion modes.   

Direct Numerical Simulation of Conventional and Alternative Jet Fuel Mixing and Ignition in Compression Ignition Engine Environments

Unmanned Aerial Systems (UAS) powered by compression ignition engines are increasingly required to operate with a diverse range of fuels, from conventional jet fuels to bio-derived fuels (e.g., alcohol-to-jet fuel and ethanol). This poses challenges for reliable UAS operation at high-altitude, low-temperature conditions as different fuels exhibit very different ignition and combustion characteristics. Detailed mixing and ignition at UAS conditions are studied with different fuels by performing Direct Numerical Simulations (DNS) based on fuel injection and ignition experiments conducted at the Army Research Laboratory. The DNS results are compared with the experiments to ensure consistency in the global behavior. While the experiments provide invaluable insight into the dynamics of mixing and ignition processes, descriptions of small scale features are unattainable from experiments. DNS complements experiments by providing a wealth of data to understand turbulence-chemistry interactions and chemical pathways controlling ignition. Reference quasi one-dimensional counterflow simulations are also presented providing a broader parametric sweep to complement DNS and experimental findings.   

Parametric study of sub-grid-scale flame--turbulence interactions in jet flames using a simplified kinetic model

Thermal expansion effects in turbulent premixed flames are known to invalidate gradient-diffusion assumptions of conventional eddy-viscosity-type turbulence models. The resulting counter-gradient transport can be accounted for in the Reynolds--Averaged Navier--Stokes (RANS) context by augmenting conventional dissipative models with thin-flame thermal-expansion source terms, but associated challenges in Large-Eddy Simulation (LES) are significant due to the dynamical nature of sub-grid-scale flame--turbulence interactions. Of particular concern is the active-cascade regime (moderate Karlovitz number; high Damk\"{o}hler number), in which large-scale shear production and small-scale thermal expansion compete to determine the local filter-scale energy dynamics. Using a single-step Arrhenius kinetic model, Reynolds- and Damk\"{o}hler-number-parameterized Direct Numerical Simulation (DNS) databases of turbulent premixed jet flames are obtained in the low- to moderate-Karlovitz-number regime. Competing effects of large- and small-scale production are analyzed using resolved and sub-grid-scale turbulence energy and scalar variance budgets, and modeling approaches are proposed.   

Dynamically-dominant Subfilter-scale content for application to LES of Turbulence-Flame Interactions in Premixed Turbulent Combustion

In premixed turbulent combustion with strong turbulence, reaction-rate dynamics and heat release concentrate near thin front-like ``reaction zone'' regions with characteristic thickness at diffusion scales that are largely unresolved in large-eddy simulation (LES). Particularly problematic is the prediction of resolved-scale (RS) evolution of the third-order chemical nonlinearities associated with the local generation of species concentrations and thermal energy and second-order advective nonlinearities driven by space-time evolution of momentum within strong turbulence localized to sheet-like reaction zones that are largely unresolved by the LES grid. Using DNS of the interaction between a flame and arrays of rectilinear vortices, we describe the extraction of ``dynamically dominant'' subfilter-scale (SFS) structure near reaction zone sheets and the encapsulation of that structure within physics-based mathematical forms. We quantify the potential increase in accuracy in the application of these ``basis functions'' within the chemical and advective nonlinearities in the evolution of RS momentum, species concentration and thermal energy within the framework of large-eddy simulation. \textit{Supported by AFOSR.}