Flow, Turbulence and Combustion最新文献

We perform Large-Eddy Simulations (LES) on the accelerating flow around streamwise-elongated rectangular cylinders with chord-to-depth ratios of 3:1 and 5:1 using Gaussian-type inflow accelerations of different intensities. The Reynolds numbers, defined with the freestream velocity and the crossflow dimension of the cylinder, range from Re = 17200 to Re = 65360. For both 3:1 and 5:1 rectangular cylinders the vortex shedding is characterized by constant-frequency time cells as observed in the literature for a square cylinder. For the 3:1 case, the Strouhal number variation range and the crossflow-force fluctuations within each time cell are the same for all cells. The results obtained under stationary inflow conditions for the rectangular 3:1 cylinder match well the statistical values computed in each time cell. On the other hand, for the 5:1 case, the cell-averaged recirculation region along the lateral side reduces in size during acceleration, leading to a narrower wake, decreased lift fluctuations, and higher Strouhal numbers. The shortening of the mean recirculation region with increasing Reynolds number for the 5:1 rectangular cylinder occurs at higher Reynolds numbers for accelerating inflows compared to stationary-inflow conditions. Finally, in agreement with what was observed for the square cylinder, for both considered aspect ratios the Strouhal number behaviors for accelerations of different severity collapse when plotted as a function of the Reynolds number.

Detailed modeling of combustion processes involving hydrogen poses challenges due to the high diffusivities of the light hydrogen molecule ((text{H}_2)) and radical (H) compared to other species. Thermodiffusion, also known as the Soret effect, describes the diffusive flux of species induced by gradients of temperature. The Soret effect becomes important if the fuel species is much lighter (or heavier) than the mean molar mass of the mixture. While accurate models for Soret diffusion exist, e.g. the multicomponent diffusion model, they are usually computationally expensive. In this work, modeling strategies for approximating Soret diffusion available in popular software packages as well as additional models from the literature are assessed in terms of their accuracy. Four methods for computing reduced collision integrals are compared and three formulations for the thermodiffusion coefficients are investigated for hydrogen and ammonia combustion. All tested approaches for computing collision integrals are found to yield good results. The approximate Soret diffusion model by Chapman and Cowling has shown the best prediction accuracy for typical hydrogen flames and ammonia/hydrogen blends when compared to the multicomponent diffusion model. Results are also compared to the model by Hirschfelder and Warnatz, implemented in the popular software packages Chemkin and STAR-CD, and the model by Bartlett and coworkers, which is available in Ansys Fluent, using different benchmark cases. This work shall serve as a review of implementation details of common models as well as a guideline for accurate and efficient Soret diffusion modeling in future hydrogen and ammonia combustion simulations.

This paper investigates the time-averaged and fluctuating aerodynamics of two slightly rough square-section prisms with rounded lateral edges of r/D = 0.16, positioned in-line at a centre-to-centre distance S/D = 4.0. For that purpose, distributions of the time-dependent surface pressures along both prisms’ mid-span cross-sections, the derived mean sectional pressure drag, lift, and pitch moment coefficients, as well as spanwise-integrated fluctuating fluid loads on the downstream prism and the frequency of the eddy shedding in its wake were measured simultaneously for Reynolds numbers between 100,000 and 7 million. Evaluation of the data and comparison with the results of an identical single prism revealed substantial changes of the flow over both prisms with Reynolds number for all studied incidence angles between ({0^ circ }) and ({45^ circ }) in the form of mutual aerodynamic influences due to proximity and wake-interference effects. For most studied flow parameters, a good agreement of the trends of the aerodynamic coefficients with incidence angle between the upstream and reference prism are obtained. Proximity effects are nevertheless clearly visible in the surface pressures, particularly at (alpha = 25.5{^ circ} ). Contrarily, wake-interference effects lead to a much lower and even negative drag on the downstream prism. The impingement of the shear layers coming from the upstream prism or of the eddies, formed in the gap between both prisms, dominates the aerodynamics of the downstream prism. This leads not only to transitions between the adjacent separation and wedge flow regimes, as well as between the co-shedding and reattachment flow states, but also triggers the vortex shedding processes between both prisms.

The Flame Spray Pyrolysis (FSP) process is a versatile and scalable method for controlled nanoparticle synthesis, with applications across various industrial sectors. FSP enables precise manipulation of nanoparticle properties, crucial for diverse applications. Carbon black (CB), important in emerging energy technologies like batteries and fuel cells, can be efficiently synthesized via FSP due to its controlled environment. Understanding CB formation is essential, given its impact on material properties. Computational Fluid Dynamics (CFD) simulations provide insights into nanoparticle formation and growth dynamics within FSP reactors, aiding in understanding process variables’ influence. This study models and analyzes CB nanoparticle formation within a specific enclosed FSP reactor with controlled coflow. The modeling approach is validated through a benchmarking ethylene sooting flame, and results are compared with existing experiments and previous models. The model accurately describes soot formation in the benchmarking case, providing reliable predictions of temperature, soot, and mean particle size. After validation, the model is extended to the FSP case. Two- and three-equation models describe soot and CB formation, with particle dynamics thoroughly discussed. The semi-empirical models assume spherical primary particles, and in the three-equation model, a population balance transport equation is solved for primary particle number density. Our investigation includes parametric sensitivity analysis, highlighting the significance of reliable model parameters, including the radiative effects of carbon particles. This work advances the understanding and predictive modeling of CB synthesis via FSP, promoting simpler alternative models compared to intricate quadrature-solved population balance approaches in the literature.

High-fidelity computational fluid dynamics (CFD) is widely used to understand turbulence and guide engineering design. While effective in predicting complex flow phenomena, CFD simulations at high Reynolds numbers require fine grids, resulting in prohibitive computational costs for parametric studies. To address this, we proposed a framework that uses machine learning (ML) to predict fine-grid results from coarse-grid simulations in a previous work. Coarsening the grid increases grid-induced error and affects turbulence prediction, necessitating a data-driven surrogate model to predict and correct these errors. A Random Forest (RF) regression was used to construct the surrogate model. The proposed framework was tested using a turbulent flow configuration consisting of an enclosed duct with a triangular bluff body acting as a blockage to the incoming flow. The chosen input features (IFs) were shown to be critical in predicting the turbulent flow field. In the current paper, we introduce further enhancements to the framework to allow it to be more robust in its prediction and application. These extensions also serve to reduce the computational cost of the approach without compromising on the accuracy. The proposed extensions include (i) adoption of Multivariate Random Forest (MRF) to replace the RF approach; (ii) identification and reduction of the IFs required for training and prediction using Variable IMportance Prediction (VIMP); (iii) predictions of flow field with changes in the bluff body configurations. The present paper aims to investigate the capability of the proposed extensions within the framework. We show that (i) the MRF allows for the accurate prediction of multiple outputs within one training instance but with a reduced computational cost relative to the RF approach. (ii) the impact of the IFs on the training can be understood via VIMP, and applying the MRF model with reduced IFs selected through VIMP does not cause any detriment to the accuracy of the prediction (iii) the extended framework trained with different bluff body configurations could be robustly applied to predict the flow field in an unseen configuration that is different from those trained. The predictive capability of the approach with these proposed extensions is demonstrated.

Scale-resolving simulations possess considerable benefits over modeled approaches because of their ability to access the underlying nonlinear fluid dynamics, and thus to predict not only the correct phenomenology, but also to generate insights on strategies to mitigate or eliminate undesirable features. The expense of resolving all pertinent turbulent scales becomes prohibitive however, as the size of the problem, typically measured by the Reynolds number based on a suitable set of reference parameters, becomes large, as is the case with flows of industrial interest such as full aircraft or their complex subsystems. This paper provides an assessment of scale-resolving methods, including some of the main benefits as well as barriers for use on large problems, together with a perspective on historical and recent trends that appear promising in the quest for routine industrial use. The factors that constitute the biggest hurdles to achieving acceptable wall-clock times and costs include meshing of complicated geometries, numerical schemes that are robust as well as accurate, suitable initial and boundary conditions, economical yet appropriate representation of near-wall turbulence, code parallelism, scalability and portability, and post-processing of the resulting big datasets. Considerations for these interrelated aspects are highlighted in the context of several 3D problems of increasing complexity, from wing sections without and with sweep, to aircraft wakes, propulsion subsystems, scramjet flowpaths and finally, full aircraft including empennages. Collectively, these examples feature the benefits of scale-resolving simulations. An illustrative approach that has reached a relatively high level of maturity using automatic mesh generation, a non-dissipative yet robust scheme, wall-modeling of turbulence, superior scalability and requiring little user intervention beyond providing the surface model, is used to demonstrate the potential of scale-resolving simulations for industry, achievable at modest cost and in reasonable wall-clock time.

Gas turbine combustors commonly feature swirling flows. The swirl is usually characterized by the swirler geometry. The development of swirl-stabilized burners includes the experimental assessment of the resulting flow field and the quantification of the swirl, e.g. through Laser Doppler Anemometry (LDA). LDA accurately acquires flow velocity components in a small probing volume. Nevertheless, the measurement quality represents a trade-off between invested measurement time and spatial resolution. In this work, the potential of Physics-Informed Neural Networks (PINNs) to assimilate flow fields from sparse LDA measurements is investigated. A novel burner is employed in which the swirl is fluidically adjustable from a non-swirled jet to a fully swirled flow through a secondary air flow injection. Data is acquired through LDA within spatial measurement grids at multiple axial distances from the swirler. Assuming symmetry, axial, tangential, and radial velocity components are determined. A PINN is subsequently trained with the acquired data, creating a continuous and differentiable flow field representation by evaluating RANS equations. By systematically reducing the training data while evaluating the physical validity of the reconstructed field, a minimum training data requirement is identified. As a result, for three operating conditions, the flow field is adequately characterized by a minimum of measurement points.

Modeling the combustion of liquid oxygen (LOx) and methane (CH4) under subcritical conditions remains challenging due to the complex interactions between two-phase flow, atomization, and combustion processes. This study uses Large Eddy Simulation (LES) with a diffuse interface method to investigate the behavior of a LOx/GCH4 single-injector rocket combustor. The proposed multifluid approach captures phase transition phenomena while maintaining computational efficiency. Numerical results are compared against experimental data, highlighting the model ability to predict flow features, such as the wall pressure distribution and wall heat fluxes. This study emphasizes the importance of accounting for the liquid core, or the dense phase, within the Eulerian framework, rather than relying on Lagrangian injection models, resulting in enhanced predictions of flame topology and heat flux distributions. Although the model exhibits good agreement with experimental measurements, it underestimates heat flux by approximately 10% at the end of the domain, likely due to limitations in the chemical kinetics model. These results show that the diffuse interface method is a promising tool for the simulation of subcritical liquid rocket combustion.

We present a data-driven approach to Reynolds-averaged Navier-Stokes (RANS) turbulence closure modelling in magnetohydrodynamic (MHD) flows. In these flows the magnetic field interacting with the conductive fluid induces unconventional turbulence states such as quasi two-dimensional (2D) turbulence, and turbulence suppression, which are poorly represented by standard Boussinesq models. Our data-driven approach uses time-averaged Large Eddy Simulation (LES) data of annular pipe flows, at different Hartmann numbers, to derive corrections for the (k)-(omega) SST model. Correction fields are obtained by injecting time averaged LES fields into the MHD RANS equations, and examining the remaining residuals. The correction to the Reynolds-stress anisotropy is approximated with a modified Tensor Basis Neural Network (TBNN). We extend the generalised eddy hypothesis with a traceless antisymmetric tensor representation of the Lorentz force to obtain MHD flow features, thus keeping Galilean and frame invariance while including MHD effects in the turbulence model. The resulting data-driven models are shown to reduce errors in the mean flow, and to generalise to annular flow cases with different Hartmann numbers from those of the training cases.

Inside the engine combustor, addition of the coolant from the wall makes the physics of flame-wall interaction (FWI) even more complex. Considering the application relevance, wall heat flux is analyzed and modeled. Under various flow conditions, the model predictions satisfactorily match the direct numerical simulation (DNS) results. The effects of coolant on the entrained flame and head-on flame are clearly different. Statistics of the near-wall flame orientation and curvature are sensitive to the coolant blowing ratio (BR). The entrained flame is more likely to be swept away, while the head-on flame is more stable. Both the model and simulation indicate consistently that an increase in BR, although quantitatively small, will greatly reduce the wall heat flux induced by the head-on flame. In contrast, the change of wall heat flux induced by the entrained flame is much smaller. Since most of the near-wall flame is head-on, the BR effect is significant. Additionally, in an a priori large eddy simulation (LES) study, the model predictions show better consistency with DNS, in comparison with the most commonly used turbulence sub-grid models.