1 October 2023 in Physics of Fluids
In this paper, a deep learning framework is proposed for predicting aerodynamic pressure distributions in general three-dimensional configurations. Based on the PointNet++ structure, the proposed framework extracts shape features based on the geometric representation of point cloud, outputs pressure coefficients corresponding to locations, and is able to accept inputs of point clouds with different resolutions. By PointNet++, we mean that local and global features of three-dimensional configurations could be effectively extracted through a multi-level feature extraction structure. A parametric approach is utilized to generate 2000 different space shuttle three-dimensional shapes, and their flows at the hypersonic speed are solved by computational fluid dynamics (CFD) as a dataset to support the training of the deep learning. Within the dataset, accurate predictions of pressure and aerodynamic forces are demonstrated for 400 unseen testing shapes. Out of the dataset, geometries that are tested for generalizability include slender, waverider, spaceplane, Apollo capsule, lifting body, and rocket. Remarkably, the framework is capable of predicting pressure distributions and aerodynamic forces for the unseen, independently designed geometries of various types in near-real-time and near-CFD accuracy, proving its excellent applicability to general three-dimensional configurations.
1 March 2023 in Physics of Fluids
High-enthalpy gas streams in a real engine or wind tunnel will contain some trace gases; however, this is often ignored in many studies. To investigate transverse fuel-jet mixing and combustion characteristics in a high-enthalpy inflow with trace-gas species, large-eddy simulations based on the HyShot II configuration have been adopted and verified. It was found that a trace amount of atomic oxygen (O) and nitrous oxide has a direct significance on ignition delay times and can, therefore, influence the overall flame distribution. In addition, the results show a greater sensitivity to the levels of O. The complex shock-wave system generated by the transverse jet is found to be a key factor in enhancing mixing and inducing combustion. It not only facilitates vortex generation by increasing the baroclinic term but also provides regions with high pressure and temperature, which accelerate the chemical-reaction rates for radical generation. The initial ignition locations, characterized by HO2 production rates, are mainly located in the low-speed region close to the injector, e.g., in the recirculation region, while OH is mainly formed downstream and accompanied by strong heat release. Based on the analysis of instantaneous contours and statistical results, the overall combustion was found to be in scramjet mode. The partially premixed flame dominates the combustor, where the combustion mainly lies in the flamelet regime, while the diffusion flame dominates the nozzle, where the combustion is spread over the flamelet regime, the broken flamelet regime, and the perfectly stirred reactor regime.
1 January 2024 in Physics of Fluids
Variable-fidelity surrogate models leverage low-fidelity data with low cost to assist in constructing high-precision models, thereby improving modeling efficiency. However, traditional machine learning methods require high correlation between low-precision and high-precision data. To address this issue, a variable-fidelity deep neural network surrogate model based on transfer learning (VDNN-TL) is proposed. VDNN-TL selects and retains information encapsulated in different fidelity data through transfer neural network layers, reducing the model's demand for data correlation and enhancing modeling robustness. Two case studies are used to simulate scenarios with poor data correlation, and the predictive accuracy of VDNN-TL is compared with that of traditional surrogate models (e.g., Kriging and Co-Kriging). The obtained results demonstrate that, under the same modeling cost, VDNN-TL achieves higher predictive accuracy. Furthermore, in waverider shape multidisciplinary design optimization practice, the application of VDNN-TL improves optimization efficiency by 98.9%. After optimization, the lift-to-drag ratio of the waverider increases by 7.86%, and the volume ratio increases by 26.2%. Moreover, the performance evaluation error of the model for both the initial and optimized configurations is less than 2%, further validating the accuracy and effectiveness of VDNN-TL.
2 August 2023 in Aerospace Science and Technology
1 April 2023 in Physics of Fluids
To explain the phenomenon observed in previous experiments of kerosene-ignition failure in scramjet combustors as the kerosene temperature increases, we numerically investigate the mixing characteristics of a kerosene jet injected into a cavity-based supersonic combustor at different injection temperatures by using a compressible two-phase flow large-eddy simulation based on the Eulerian–Lagrangian approach. The results indicate that, upon injecting kerosene at high temperatures, the flow field preceding the leading edge of the cavity is similar to a typical gas jet in supersonic crossflow. The wall counter-rotating vortex pair (CVP) develops more fully and eventually becomes the main vortex pair. This evolution of the wall CVP modifies the cavity shear layer and alters the local flow-field characteristics near the cavity. Upon injecting kerosene at high temperatures, its evaporation rate increases sharply and the cavity recirculation zone enlarges, which causes more kerosene vapor to be entrained into the cavity. Because the kerosene-vapor temperature is lower than that of the low-speed fluid in the cavity, a significant amount of kerosene vapor entering the cavity not only makes the mass fraction of kerosene in the cavity exceed the fuel stoichiometric mass fraction but also reduces the temperature in the cavity, which negatively impacts the ignition process. The ignition delay time is much longer when the injection temperature is high, which is consistent with the inability of the initial flame kernel to form in the experiment.
1 October 2023 in Physics of Fluids
The mixing and combustion characteristics of liquid kerosene spray in a cavity-based supersonic combustor under a Mach 2.52 inflow are experimentally and numerically investigated. The combustion mode transition of kerosene spray includes two aspects: 1. As the injection pressure increases, the combustion mode shows a transition of weak-intensive-transitional(weak to intensive)-weak-blowoff; 2. in the transitional combustion mode, there is a transition from weak to intensive combustion over time. In the weak combustion mode, the combustion chamber pressure increases slightly, and the flame is inside the cavity. In the intensive combustion mode, the pressure rises sharply from the isolation section and the flame is a typical cavity stabilized combustion mode. The weak combustion at low/high injection pressure is due to rich/lean fuel inside the cavity. In the transitional combustion mode, the flame on the downstream wall of the cavity gradually expands and joins with the flame inside the cavity, evolving from weak combustion to intensive combustion. The transition is accomplished in a very short time, on the order of milliseconds. This study deepens the understanding of the combustion mode of scramjet engines and provides a reference for the design of fuel injection schemes for scramjet engines.