Search

Given the need to develop zero-carbon combustors for power and aircraft engine applications, $S_d$ of a turbulent premixed flame, especially for H$_2$-air, is of immediate interest. The present study investigates 3D DNS cases of premixed H$_2$-air turbulent flames at varied pressures for different $Re_t$ and $Ka$ with detailed chemistry to theoretically model $S_d$ at negative curvatures. Prior studies at atmospheric pressure showed $\widetilde{S_d}$ to be enhanced significantly over $S_L$ at large negative $\kappa$ due to flame-flame interactions. 1D simulations of an imploding cylindrical H$_2$-air laminar premixed flame used to represent the local flame surfaces undergoing flame-flame interaction in a turbulent flame at the corresponding pressure conditions are performed to understand the interaction dynamics. These simulations emphasized the transient nature of the flame structure during flame-flame interactions and enabled analytical modeling of $\widetilde{S_d}$ at these regions of extreme negative $\kappa$ of the 3D DNS. The JPDF of $\widetilde{S_d}$ and $\kappa$ and the corresponding conditional averages from 3D DNS showed a negative correlation between $\widetilde{S_d}$ and $\kappa$. The model successfully predicts the variation of $\langle\widetilde{S_d}|_{\kappa}\rangle$ with $\kappa$ for the regions on the flame surface with $\kappa\delta_L \! \ll \! -1$ at all pressures, with good accuracy. This shows the aforementioned configuration to be fruitful in representing local flame-flame interaction in 3D turbulent flames. Moreover, at $\kappa=0$, on average $\widetilde{S_d}$ can deviate from $S_L$, manifested by the internal flame structure, controlled by turbulence transport in the large $Ka$ regime. Thus, the correlation of $\langle\widetilde{S_d}\rangle/S_L$ with $\langle|\widehat{\nabla c}|_{c_0}\rangle$ at $\kappa =0$ is explored.

Chemical kinetics mechanisms are essential for understanding, analyzing, and simulating complex combustion phenomena. In this study, a Neural Ordinary Differential Equation (Neural ODE) framework is employed to optimize kinetics parameters of reaction mechanisms. Given experimental or high-cost simulated observations as training data, the proposed algorithm can optimally recover the hidden characteristics in the data. Different datasets of various sizes, types, and noise levels are tested. A classic toy problem of stiff Robertson ODE is first used to demonstrate the learning capability, efficiency, and robustness of the Neural ODE approach. A 41-species, 232-reactions JP-10 skeletal mechanism and a 34-species, 121-reactions n-heptane skeletal mechanism are then optimized with species' temporal profiles and ignition delay times, respectively. Results show that the proposed algorithm can optimize stiff chemical models with sufficient accuracy and efficiency. It is noted that the trained mechanism not only fits the data perfectly but also retains its physical interpretability, which can be further integrated and validated in practical turbulent combustion simulations., Comment: 8 pages, 9 figures

In this letter, the merging-vs-bouncing response of obliquely-oriented colliding jets under elevated and reduced gaseous environment pressures was experimentally examined. Experiments with water and n-tetradecane confirmed that the collision outcome transitions from merging to bouncing, and then to merging again, when the impact velocity was increased. This behavior which was previously reported for atmospheric pressure, has now also been observed at elevated and reduced pressures. New results also show that there exists a critical pressure (0.9 bar for tetradecane and 5 bar for water) below which increasing pressure promotes bouncing (expands the bouncing regime), while beyond this, merging is promoted (reduces the bouncing regime) instead. This leads to a non-monotonic influence of pressure on the non-coalescence outcomes of collisional jets, which was not previously reported. The study provides evidence of new behaviors in colliding jets at reduced and elevated pressures, which differs from well-studied droplet-droplet collisions., Comment: Corrected the abstract, results unchanged, a few minor edits based on reviews. This is the published version in Phy. Fluids

The gasification of multicomponent fuel drops is relevant in various energy-related technologies. An interesting phenomenon associated with this process is the self-induced explosion of the drop, producing a multitude of smaller secondary droplets, which promotes overall fuel atomization and, consequently, improves the combustion efficiency and reduces emissions of liquid-fueled engines. Here, we study a unique explosive gasification process of a tricomponent droplet consisting of water, ethanol, and oil ("ouzo"), by high-speed monitoring of the entire gasification event taking place in the well-controlled, levitated Leidenfrost state over a superheated plate. It is observed that the preferential evaporation of the most volatile component, ethanol, triggers nucleation of the oil microdroplets/nanodroplets in the remaining drop, which, consequently, becomes an opaque oil-in-water microemulsion. The tiny oil droplets subsequently coalesce into a large one, which, in turn, wraps around the remnant water. Because of the encapsulating oil layer, the droplet can no longer produce enough vapor for its levitation, and, thus, falls and contacts the superheated surface. The direct thermal contact leads to vapor bubble formation inside the drop and consequently drop explosion in the final stage., Comment: 8 pages, 5 figures

The inputs of deep neural network (DNN) from real-world data usually come with uncertainties. Yet, it is challenging to propagate the uncertainty in the input features to the DNN predictions at a low computational cost. This work employs a gradient-based subspace method and response surface technique to accelerate the uncertainty propagation in DNN. Specifically, the active subspace method is employed to identify the most important subspace in the input features using the gradient of the DNN output to the inputs. Then the response surface within that low-dimensional subspace can be efficiently built, and the uncertainty of the prediction can be acquired by evaluating the computationally cheap response surface instead of the DNN models. In addition, the subspace can help explain the adversarial examples. The approach is demonstrated in MNIST datasets with a convolutional neural network. Code is available at: https://github.com/jiweiqi/nnsubspace., Comment: Add link to github repo

As a key process that refreshes the interstellar medium, the dynamics and radiative properties of the supernova remnant (SNR) expansion front not only reflect the physical environment of the old interstellar medium (ISM) surrounding the supernova, but they also provide information about the refreshed ISM. However the expansion dynamics of SNRs cannot be simply explained by the conventional law of spherical shock wave propagation; on the other hand, the high energy radiation requires an additional electron acceleration mechanism in the shock front beyond thermal collision. We consider herein the detonation wave description of the SNR expansion, in which magnetic reconnection follows the shock front and transfers the SNR magnetic field energy to both fluid thermal energy and particle kinetic energy. The structure of the magnetic reconnection detonation (MRD) is identified based on scaling analysis in this paper. By applying the MRD description of the SNR expansion shock to the example of the Crab Nebula, this paper shows that the MRD description can explain both the accelerative expansion of the nebula as well as the origin of the luminous expanding shell., Comment: Accepted for publication in ApJ

Thermal-diffusional pulsation behaviors in planar as well as outwardly and inwardly propagating white dwarf carbon flames are systematically studied. In the 1D numerical simulation, the asymptotic degenerate equation of state and simplified one-step reaction rates for nuclear reactions are used to study the flame propagation and pulsation in white dwarfs. The numerical critical Zel'dovich numbers of planar flames at different densities ($\rho=2$, 3 and 4$\times 10^7$~g/cm$^3$) and of spherical flames (with curvature $c=$-0.01, 0, 0.01 and 0.05) at a particular density ($\rho=2\times 10^7$~g/cm$^3$) are presented. Flame front pulsation in different environmental densities and temperatures are obtained to form the regime diagram of pulsation, showing that carbon flames pulsate in the typical density of $2\times10^7~{\rm g/cm^3}$ and temperature of $0.6\times 10^9~{\rm K}$. While being stable at higher temperatures, at relatively lower temperatures the amplitude of the flame pulsation becomes larger. In outwardly propagating spherical flames the pulsation instability is enhanced and flames are also easier to quench due to pulsation at small radius, while the inwardly propagating flames are more stable., Comment: ApJ, 841, 21 (2017), 25 pages in arxiv version

The turbulent motion within molecular clouds is a key factor controlling star formation. Turbulence supports molecular cloud cores from evolving to gravitational collapse and hence sets a lower bound on the size of molecular cloud cores in which star formation can occur. On the other hand, without a continuous external energy source maintaining the turbulence, such as in molecular clouds, the turbulence decays with an energy dissipation time comparable to the dynamic timescale of clouds, which could change the size limits obtained from Jean's criterion by assuming constant turbulence intensities. Here we adopt scaling relations of physical variables in decaying turbulence to analyze its specific effects on the formation of stars. We find that the decay of turbulence provides an additional approach for Jeans' criterion to be achieved, after which gravitational infall governs the motion of the cloud core. This epoch of turbulence decay is defined as cloud core relaxation. The existence of cloud core relaxation provides a more complete understanding in the competition between turbulence and gravity on the dynamics of molecular cloud cores and star formation., Comment: 18 pages, 1 figure, accepted for publication in ApJ

It was demonstrated recently in Bychkov et al., Phys. Rev. Lett. 101 (2008) 164501, that the physical mechanism of flame acceleration in channels with obstacles is qualitatively different from the classical Shelkin mechanism. The new mechanism is much stronger, and is independent of the Reynolds number. The present study provides details of the theory and numerical modeling of the flame acceleration. It is shown theoretically and computationally that flame acceleration progresses noticeably faster in the axisymmetric cylindrical geometry as compared to the planar one, and that the acceleration rate reduces with increasing initial Mach number and thereby the gas compressibility. Furthermore, the velocity of the accelerating flame saturates to a constant value that is supersonic with respect to the wall. The saturation state can be correlated to the Chapman-Jouguet deflagration as well as the fast flames observed in experiments. The possibility of transition from deflagration to detonation in the obstructed channels is demonstrated., Comment: 12 pages

Spontaneous flame acceleration leading to explosion triggering in open tubes/channels due to wall friction was analytically and computationally studied. It was first demonstrated that the acceleration is effected when the thermal expansion across the flame exceeds a critical value depending on the combustion configuration. For the axisymmetric flame propagation in cylindrical tubes with both ends open, a theory of the initial (exponential) stage of flame acceleration in the quasi-isobaric limit was developed and substantiated by extensive numerical simulation of the hydrodynamics and combustion with an Arrhenius reaction. The dynamics of the flame shape, velocity, and acceleration rate, as well as the velocity profile ahead and behind the flame, have been determined., Comment: 41 pages; 11 figures

As a foundational element describing relativistic reacting waves of relevance to astrophysical phenomena, the Rankine-Hugoniot relations classifying the various propagation modes of detonation and deflagration are analyzed in the relativistic regime, with the results properly degenerating to the non-relativistic and highlyrelativistic limits. The existence of negative-pressure downstream flows is noted for relativistic shocks, which could be of interest in the understanding of the nature of dark energy. Entropy analysis for relativistic shock waves are also performed for relativistic fluids with different equations of state (EoS), denoting the existence of rarefaction shocks in fluids with adiabatic index \Gamma < 1 in their EoS. The analysis further shows that weak detonations and strong deflagrations, which are rare phenomena in terrestrial environments, are expected to exist more commonly in astrophysical systems because of the various endothermic reactions present therein. Additional topics of relevance to astrophysical phenomena are also discussed., Comment: 34 pages, 9 figures, accepted for publication in ApJ