Flow, Turbulence and Combustion最新文献

The influence of fuel Lewis number Le_F on the wall-normal variations of mean values of streamwise velocity component, temperature, and Reynolds stresses within turbulent boundary layers have been analysed using a Direct Numerical Simulation (DNS) database of oblique wall quenching of V-shaped premixed flames due to their interaction with inert isothermal walls in a turbulent channel flow configuration. The fuel Lewis numbers of Le_F = 0.6, 1.0 and 1.4 have been considered for the current analysis. It has been found that the flame starts to interact with the wall further upstream for a smaller value of Le_F due to an augmentation of the volume-integrated burning rate with a decrease in fuel Lewis number. The thermal expansion induced by heat release affects the variations of mean values of both streamwise velocity component and temperature in the wall-normal direction and they show significant deviations from the conventional log-law profiles. Recently proposed density-compensated modified wall functions for streamwise velocity and non-dimensional temperature, which were previously validated for head-on quenching configuration for unity Lewis number, are found to capture the corresponding behaviours extracted from DNS data for flame-wall interaction of V-shaped flames with non-unity Le_F. The wall-normal distributions of Reynolds stresses also deviate significantly from the corresponding non-reacting fully developed channel flow profiles during flame-wall interaction and Le_F affects the extent of this deviation. The anisotropy of Reynolds stresses has been found to increase with the progress of flame-wall interaction. The physical explanations for the observed mean velocity, mean temperature and Reynolds stress statistics have been provided and their implications for modelling of flame-wall interaction have been elaborated.

This study investigates preferential diffusion effects on the volume-integrated burning rate in turbulent premixed lean hydrogen/air flames, using 3D direct numerical simulations (DNS) to analyze flames at two global equivalence ratios and varying turbulence intensities. By examining the statistical distributions of local equivalence ratios, the analysis confirms pronounced preferential diffusion effects across all cases. Intenser turbulence tends to amplify these effects. Probability density functions (PDFs) of the local equivalence ratio further confirm stronger preferential diffusion at lower global equivalence ratios and significant sensitivity to the choice of reaction progress variable definitions, particularly between hydrogen-based and water-based definitions. To predict the impact of preferential diffusion, the volume-integrated burning rate is estimated by multiplying the laminar burning velocity for the local equivalence ratio with the probability density function of the local equivalence ratio distribution. The subsequent estimates are compared to corresponding values directly obtained from DNS data. Results show that, in cases where preferential diffusion effects are more pronounced, i.e. at lower global equivalence ratios and relatively higher turbulence intensities, the DNS-derived burning rates based on water and temperature progress variables are best approximated using burning rates computed from the local equivalence ratio field conditioned on positive mean curvature. In contrast, cases less affected by preferential diffusion yield burning rate per unit area values comparable to those in the unstretched laminar flames when evaluated using water- and temperature-based definitions. The findings suggest that the burning rate per unit flame area could be modeled using the laminar burning velocity corresponding to the local equivalence ratio, and presumed PDFs representing the distributions of the local equivalence ratio. The gamma (or beta) distribution has been found to reasonably approximate the PDF of the local equivalence ratio, which can be utilized for the modeling of the volume-integrated burning rate in premixed turbulent lean hydrogen/air flames.

To efficiently harness gas-driven liquid metal for two-phase magnetohydrodynamic (MHD) power generation, three types static mixers was installed normally at the front of the power generation channel. The mixing flow process of the liquid metal gallium (Ga) with the gas R245fa within the power generator was numerically studied using ANSYS software. The results indicate that under identical boundary conditions, the gas-liquid metal two-phase fluid exhibits stratified and annular flows after passing through the HEV and SK type mixers, respectively. In contrast, a churn flow is observed when the fluid passes through the SMV-type mixer. Behind the SMV-type mixer, the velocity of the liquid metal in the power generation channel is 8.65 times greater than at the mixer’s inlet, while the slip ratio of the two-phase fluid is only 1.3. The uniformity of the liquid metal in the power generation channel is 0.611, and the volume flow rates of the liquid metal were increased by 46.7% and 3.1% compared to the that of SK and HEV types, respectively. Additionally, the current density fluctuation in the power generation channel was observed to be more stable over time with the SMV-type mixer.

Implicit Large Eddy Simulations (ILES) are deployed to characterize the effect of swirling boundary conditions on vorticity transport as well as on the Lighthill’s tensor for a cold, supersonic aerospike nozzle jet. Four jets are simulated at a Nozzle Pressure Ratio (NPR) = 3, one jet without swirl and three jets with swirl numbers (mathcal{S} = {0.10,0.20,0.30}). Swirling boundary conditions lead to an increase in vorticity tilting and stretching downstream of the aerospike bluff body, potentially enhancing sound generation. An exact decomposition of the Lighthill’s tensor is undertaken and the magnitude of the obtained source terms is compared. The dominant acoustic source terms are amplified under swirling boundary conditions. Terms describing the interactions between dilatation fields and density gradients balance each other in shock regions. At higher swirl numbers, the convection of density gradients along the flow direction leads to an imbalance that contributes to increased sound generation. Finally, cross-correlations between the near-field pressure and individual source terms reveal that enstrophy correlates more strongly with the near-field acoustics at higher swirl numbers.

A scalar dissipation rate-based mean reaction rate closure modified for premixed flame-wall interaction, which was previously proposed based on a priori Direct Numerical Simulation (DNS) analysis, is implemented for Reynolds Averaged Navier-Stokes (RANS) simulations in two configurations. The first configuration is the oblique wall quenching of a V-shaped premixed flame in a turbulent channel flow, and the second configuration is the head-on quenching of a statistically planar flame in a turbulent boundary layer. To avoid uncertainties associated with wall functions in reacting flows, a low Reynolds number (k - varepsilon ) model that resolves the viscous sub-layer is used. Comparisons between RANS simulations using the modified closure and DNS data reveal satisfactory agreement for Favre mean streamwise velocity and Favre mean temperature. However, quantitative discrepancies are found in the predictions of the Favre-averaged reaction progress variable due to differences in mean reaction rate profiles between DNS and RANS results. This behaviour arises from differences in turbulence quantities (e.g. turbulent kinetic energy and dissipation rate) between RANS and DNS, leading to discrepancies in RANS predictions of the mean reaction rate and Favre-averaged scalar dissipation rate, despite these closures performing well in a priori analysis. Even with these discrepancies, the scalar dissipation-based mean reaction rate closure shows promise in predicting mean values of Favre-averaged streamwise velocity and non-dimensional temperature in premixed flame-wall interaction configurations.

The development of a hydrogen jet injected into quiescent argon was investigated in a temporal jet configuration via direct numerical simulations (DNS). A case of argon mixing in argon was used as the basis for comparison. Both systems were computed at jet Reynolds numbers of 5000 and 10,000. Attention was focused on the mechanism driving the mixing process, as well as the turbulent momentum and scalar transport. The physical properties of argon are very different from those of hydrogen (density ratio (≈20), kinematic viscosity ratio (≈0.1), and Lewis number ratio ((approx 3))), leading to significant differences between the two cases, in jet structure, instantaneous and mean profile characteristics. A common feature in all systems was the emergence of large quasi-two-dimensional rotating structures, responsible for the engulfment of surrounding fluid, which created elongated regions where most molecular mixing takes place, with one difference being faster mixing in the hydrogen cases. An a priori assessment of the classical gradient hypothesis for turbulent fluxes revealed that the turbulent Schmidt number ((mathrm{Sc}_t)) and C_µ are not constant in space nor time, with local values ranging from (0.2-1.4), and (0.6-1.1), respectively, contrasting with the constant values used in Reynolds-Averaged Navier-Stokes (RANS) modeling. Additionally, an evaluation of a two equation RANS model and a dynamic one-equation large eddy simulations (LES) model was made a posteriori by comparison of their predictions with the DNS results. Both approaches exhibited significant deviations from the DNS, primarily at the early stage, but relaxed to similar solutions as time progressed. The properties at the jet edge were less well predicted by the RANS model than by the LES model. This is attributed to both gradient diffusion modeling and the impact of a turbulent/non-turbulent interface. Possible model enhancements are discussed.

Graphical Abstract

To reveal the gas-water two phase flow characteristic in the anode porous transport layers (PTL) of Proton Exchange Membrane (PEM) electrolyzer, both experimental and numerical (twin numerical model) studies were carried out in this work. An optical visualization test setup was developed, high-speed camera was used for flow visualization, pressure sensor and flowmeter were used to quantify the pressure drop and mass flow rate characteristics of gas-liquid two-phase flows, and the pressure drop and flow regime in the anode free flow channel as well as the PTL were reveal under different gas-to-water ratio (GWR), incline angle, porous size and pores configuration. Besides, a twin numerical model was also developed, and the flow characteristics of gas-liquid two-phase flow were further revealed by simulation. Moreover, the bubble coverage α in the free flow channel and the water proportion in the catalytic interface were quantified, and the threshold GWR under different operation conditions were also determined, which provides insight for water management of PEM electrolyzer.

The use of Computational Fluid Dynamics (CFD) to predict the acoustic signature of marine propellers operating in highly turbulent and non-uniform flow conditions has attracted considerable academic and industrial interest over the past decade. The negative effects of radiated noise from underwater maritime vessels and shipping activities on marine ecosystems and aquatic life are well recognized. This noise originates from various sources, most prominently propellers. This study aims to characterize the noise levels generated by marine vessels through high-fidelity numerical modeling of hydroacoustic phenomena. We employed Large Eddy Simulation (LES), the Schnerr–Sauer cavitation modeling approach, and the compressive volume-of-fluid (VOF) method to simulate cavitating flow over the propeller and predict the far-field radiated noise using the Ffowcs Williams-Hawkings (FW-H) hydroacoustic analogy. This study provides insights into the flow physics of noise generation due to wake structures—such as tip, root, trailing edge, and hub vortices—and cavitation patterns, including sheet, tip, and hub cavitation, over marine propellers operating upstream of a rudder. We characterized the instability of vortical structures in the wake of the marine propeller, including tip/hub vortex oscillation and instability, mutual-inductance instability (leapfrogging effect), elliptic instability, and primary and secondary wake grouping. Additionally, we examined sound levels related to propeller loading, cavitation development, periodic pulsating cavitation, and pressure fluctuations in both the near-field and far-field. Finally, the comparison between modeled and measured noise provided insights into the spectral contributions of propeller- and non-propeller-generated noise, helping to define the range of applicability for the assumption that propeller sources dominate overall noise emissions from the vessel.

The response of a hollow-cone n-heptane spray in a swirling airflow to forced acoustic disturbances is investigated using compressible Large-Eddy Simulations (LES) of the two-phase flow, with the dispersed fuel phase described in a Lagrangian framework. Phase-averaged numerical simulations with respect to the acoustic forcing are compared against phase-averaged phase-Doppler velocimetry (PDV) measurements of gaseous and droplet velocities, as well as phase-averaged measurements of droplet diameters over a forcing cycle. Liquid injection is modeled using the semi-empirical FIM-UR approach, a well-established framework for simulating fuel spray injection in gas turbines. Under steady injection conditions for both the gaseous and dispersed phases, simulations and experiments show excellent agreement in terms of gaseous velocity fields, droplet velocities, and droplet diameter distributions. In the acoustically forced regime, simulations also reproduce the gaseous-phase velocity response with good accuracy. Minor differences in negative velocity fluctuations at the burner outlet, attributed to pressure-drop mismatches across the air swirler, vanish downstream and can be mitigated by refining the mesh resolution in the air injection channels. For the dispersed fuel phase, however, droplet velocities are significantly overpredicted in the acoustically forced simulations. The difference arises because the simulations cannot fully reproduce the acoustically forced swirling flow, particularly in the internal recirculation zone, which is narrower than observed in experiments. Acoustic disturbances induce oscillations of the angle of hollow cone but also spatial and temporal fluctuations of the particle diameter and velocity distributions that are not accurately reproduced by the model. These findings highlight the critical importance of accurately modeling the unsteady response of the fuel injector in order to capture the complex swirling flow-spray-acoustic coupling. This coupling is shown to strongly influence the spray cone angle, as well as the distributions of droplet diameter and velocity.

Porous media are a promising technology to reduce turbulent boundary layer trailing edge noise. However, the fact that the porous material is grazed by turbulent flow on both sides makes its characterization not trivial. This paper describes the modifications resulting from the interaction between the grazing flows through the porous medium, defined as communication. To this end, lattice-Boltzmann simulations of two communicating turbulent channel flows separated by a fully resolved porous medium are carried out. The porous medium is realized as a 75% porous triply periodic minimal surface of type Schwarz’ P. Results are compared against the case with porous medium backed by a solid wall and the smooth wall channel flow. When communication between the two channel flows is allowed, spanwise coherent structures appear that are assimilated to a shear instability at a non-dimensional frequency of (St_t=0.02). Instantaneous flow through the porous medium is observed and is driven by a time-dependent pressure differential between the channels (with a zero mean and 7.8 Pa standard deviation). This leads to a decrease in energy in turbulent scales smaller than 2.5δ and for bulk scaled frequencies greater than (St_b=0.41). These flow modifications are not observed in the non-communicating case, with the wall preventing flow through, where the topology of the fluctuating statistics is similar to the smooth wall case. Finally, the drag is found to increase by over 200% with respect to the non-communicating case and 650% with respect to a smooth turbulent channel flow. The drag increase is found to be driven by the velocity fluctuations impinging on the porous topology. The communication does not follow the asymptotic drag relation for the same equivalent roughness, thus entering a different drag regime.