Session ZC40: Reacting Flows: Instabilities

The Impact of Different Image Processing Techniques on the Modal Decomposition of Methane Jets Exposed to Acoustic Excitation

Currently, modal decomposition is a technique widely used in applications aimed at identifying and extracting dominant features of flows. The discovery of models based on the extraction of information from data is constantly evolving, transforming the process of modeling, predicting and controlling complex dynamic systems. A commonly used dataset consists of temporally evolving experimental digital images, which have a large degree of noise. The objective of this study is to explore the effect of various image processing methods on the dominant modes in an acoustically-coupled combustion problem extracted via Proper Orthogonal Decomposition (POD). Commonly used image processing approaches (median filters, block-matching and 3D filtering or BM3D, threshold filters, band-pass filters, Wavelet transforms, Fredholm transforms and Kalman filters) are investigated. For each approach, an optimal image enhancement parameter is found, which removes much of the noise and improves the experimental input data of the modal decomposition. Consequently, there is an increase in the probability of obtaining a greater total number of resulting modes, making it possible to define more precisely the dynamical behavior of acoustically forced diffusion flames.   

Acoustically Coupled Fuel Jet Combustion Near Pressure Nodes and Antinodes

This experimental study explores the combustion dynamics of single and coaxial laminar jet diffusion flames in the presence of standing acoustic disturbances. The flames are studied inside a closed cylindrical waveguide under atmospheric conditions, observed through high-speed visible imaging and analyzed using proper orthogonal decomposition (POD). Different burner geometries with varying annular-to-jet area ratios, tube wall thicknesses, Reynolds numbers, and velocity ratios are explored, in addition to varying applied acoustic conditions, including flames in the vicinity of a pressure node (PN) and a pressure antinode (PAN). Differing flame-acoustic coupling processes resulted from the alternative conditions, including sustained oscillatory combustion (SOC), multi-frequency, periodic lift-off and reattachment (PLOR), permanent flame lift-off (PFLO), and eventual flame blowoff (BO). The phase portraits extracted from POD mode coefficients capture distinct signatures associated with these transitions, which can aid in developing topology-based reduced order models (ROMs) for flame-acoustic coupling. Machine learning tools are developed to improve denoising of flame images, in addition to nonlinear decomposition methods to further reduce dimensionality, with promising implications for the advancement of nonlinear ROMs.   

Short-term prediction of a propagating flame in a Hele-Shaw cell

We numerically study the dynamic behavior of flame front fluctuations in a Hele-Shaw cell by solving a nonlinear evolution equation describing a downwardly propagating flame dynamics, including the short-term predictability of flame front fluctuations. Our recent study [Y. Nomi et al., Phys. Rev. E, vol. 103, p. 022218, 2021] has reported the two important findings under the negative normalized Rayleigh number conditions: (i) the randomness in flame front fluctuations significantly increases with the gravitational level, and (ii) the irregular formation of large-scale wrinkles driven by the Rayleigh-Taylor instability plays an important role in the formation of high-dimensional deterministic chaos. The effect of additive noise on the randomness in flame front dynamics and the cell size distribution of the wrinkles has been clarified in our subsequent study [Y. Nomi et al., Chaos, vol. 31, p. 123133, 2021]. The most interesting result in the present study is that the flame front fluctuations can be sufficiently predicted by a reservoir computing [T. Tokami et al., Phys. Rev. E, vol. 101, p. 042214, 2020]. In this presentation, we will discuss the relevance of the predictability nature to the identification of deterministic chaos.   

Rayleigh-Taylor Unstable Flames: Thick and Thin

Models for the behavior of Rayleigh-Taylor unstable flames typically treat the flame structure as infinitely thin, laminar-like, or turbulently thickened. In this talk, we report the analysis of a large parameter study of direct numerical simulations (DNS) of Rayleigh-Taylor unstable flames and show that none of these simplified models captures their beautifully complex structure. We show that the steepness of the reaction rate and the Prandtl number, parameters that are typically neglected in models, do affect the structure and speed of the flame in significant ways.   

Rayleigh-Taylor Unstable Flames: the Coupled Effect of Multiple Perturbations

In this talk, we explore the effect of the Rayleigh-Taylor instability on premixed flames. Using 2D direct numerical simulations (DNS) of Boussinesq flames with a model reaction rate, we show what happens when the flame is influenced by three types of perturbation: a large amplitude single-mode primary perturbation, a smaller amplitude single-mode secondary perturbation, and a multimode numerically-generated system perturbation. Early on, the evolution of the flame is dominated by the primary perturbation and the flame propagates as a metastable traveling wave in the form of bubbles separated by cusp-like spikes. However, the lifetime of this traveling wave depends on the properties of the secondary and system perturbations and on the strength of gravity. Once the traveling wave is destabilized, the bubbles rapidly grow to large scales. The level of organization of this process depends on the balance and interactions between the three types of perturbation. Finally, we compare this growth process to the growth of the mixing layer in non-reacting Rayleigh-Taylor.   

Flow instabilities, mixing, and combustion in porous solid-fluid domains

Many natural and industrial processes involve mixed porous-fluid domains where multiple physics interact over disparate lengths and time scales, such as the combustion of multi-species biofuels. Although many modeling studies so far, have concentrated on detailed physics within the single fluid or porous phase, there rarely are reports of a comprehensive model considering both phases. In this work, we numerically study different scenarios involving porous-fluid regions using a single-domain approach. The model considers the compressible Navier-Stokes/Darcy-Forchheimer equation and two separate energy conservation equations for the heat transfer in the solid matrix and the interstitial fluid in the porous region, along with the multi-species transport and detailed chemical kinetics. We validate the method against a set of experimental and direct numerical simulation results including the thermal decomposition of biomass particles. We then investigate the thermally- and buoyancy-driven fluid instabilities (Rayleigh-Bénard like) through three-dimensional large-eddy simulation (LES) and Reynolds-averaged Navier-Stokes (RANS) modeling. We focus on the interfacial porosity variation and configurational effects in turbulence mixing and its implications for ignition and fire spread.   

Stochastic simulations of reacting flows under non-equilibrium conditions in the continuum regime

Direct Simulation Monte Carlo (DSMC), generally utilized for simulating rarefied flows, is used for simulating flows in the continuum regime. This method provides an advantage of simulating flows at a molecular level and hence, treating microscopic phenomena such as molecular transport processes and internal energy exchange through individual collisions. A non-reacting case with Kelvin-Helmholtz instabilities developing in a shear layer under thermal non-equilibrium conditions is simulated and the results are compared with Direct Numerical Simulation (DNS). It is found that DSMC captures the overall flow characteristics accurately even in the continuum regime. Subsequently, cases involving reacting flows are explored with DSMC coupled with traditional models for chemical reactions (namely TCE and QK models). Simulations of equilibrium one-dimensional hydrogen-air deflagration and detonation are performed using two different reaction mechanisms. Both mechanisms produce good results for deflagration compared against freely propagating premixed flame structure. However, for detonations, DSMC does not capture the induction zone accurately, especially when using the QK model in its current form. This discrepancy could arise from the lack of treatment of state-specific reaction cross-sections under non-equilibrium flow conditions. Efforts are under way to include in DSMC the reaction cross-sections obtained from Quasi Classical Trajectory (QCT) calculations.   

Numerical simulation of boundary layer flashback of hydrogen-air premixed flame using an extended FGM method

Numerical simulation is beneficial for the development of combustion devices, but large-scale combustion simulations require the use of a combustion model to reduce the computational cost. Flamelet-generated manifold (FGM) method has been widely used as an effective combustion model. However, the conventional FGM method cannot fully consider the effects of preferential diffusion and flame stretch, which have a large influence particularly on hydrogen flames. In this study, the FGM method is extended to explicitly consider the preferential diffusion, flame stretch, and non-adiabatic effects. The extended FGM method is applied to a boundary layer flashback simulation of hydrogen-air premixed flame under the condition of an unburnt temperature of 750 K, an ambient pressure of 1 atm, and an equivalence ratio of 0.5. The results show that the extended FGM method improves prediction accuracy on the flame propagation speed and distributions of physical properties compared with the conventional FGM method.   

Thermodiffusive instabilities and nitrogen oxides in lean ammonia/hydrogen/nitrogen-air laminar premixed flames

Though a promising zero-carbon hydrogen carrying fuel, ammonia suffers from poor combustion properties. One way to improve the combustion properties of ammonia is to partially crack the fuel into a mixture of ammonia, hydrogen, and nitrogen. Under fuel-lean conditions, pure hydrogen flames are thermodiffusively unstable, which has implications for not only flame propagation but also the formation of nitrogen oxides. However, much less is known about the onset of thermodiffusive instabilities when hydrogen is part of a multi-component fuel also containing ammonia. In this work, detailed two-dimensional simulations of laminar premixed flames are conducted to understand the onset of thermodiffusive instabilities in laminar premixed flames of ammonia/hydrogen/nitrogen mixtures and air. The degree of ammonia cracking, that is, the relative amount of ammonia to hydrogen/nitrogen, is varied to understand the onset of thermodiffusive instabilities as a function of fuel composition. Additionally, the influence of the fuel mixture and thermodiffusive instabilities on the formation of nitrogen oxides is analyzed. This understanding is critically required for ammonia to be a practical zero-carbon alternative fuel.   

Numerical investigation of combustion instability and flashback of lean-premixed low-swirl hydrogen jet flame

As an eco-friendly combustion method, lean-premixed hydrogen combustion is attracting attention since lean-premixed hydrogen combustion has the advantage of significantly reducing NOx emissions and emitting no CO2. However, the lean condition is particularly prone to combustion instabilities (CI) and flashback, which can cause severe damage to combustors. In this study, the effects of the equivalence ratio on the CI and flashback of a lean-premixed hydrogen jet flame in a low-swirl combustor (LSC) are investigated using Large-eddy Simulation employing Dynamic Smagorinsky Model and Dynamically thickened flame model as a turbulence and turbulent combustion models, respectively. The results show that the amplitude of the pressure oscillations increases significantly when the spatially averaged temperature inside the combustor reaches a certain temperature, and that flashback tends to take place under a certain combustion condition.