BB2潜艇尾流与阻力预测的混合RANS-LES及WALE公式评估
Assessment of Hybrid RANS-LES and WALE Formulations for Wake and Resistance Prediction of the BB2 Submarine
AI总结：
本研究针对BB2潜艇高Re尾流与阻力预测，评估基于Taylor微尺度细化的混合RANS-LES及WALE模型，证明壁面建模瞬态模拟可更细致捕捉非对称涡和湍流脉动且保持阻力精度。
摘要：
潜艇水动力学在准确预测流动分离、尾流结构和阻力方面面临独特挑战，原因在于高雷诺数（Re）下复杂的几何形状和湍流行为。传统的雷诺平均Navier-Stokes（RANS）方法在解析近区和远区的非定常流动结构及湍流方面往往存在局限。为克服这些局限，分离涡模拟（DES）和大涡模拟（LES）等混合RANS-LES模型在捕捉近壁涡结构方面提供了更优性能。捕捉湍流涡和尾流结构对开展水动力噪声分析具有重要作用。对这些相干结构的详细解析和理解有助于降低水声特征，这对潜艇的隐身特性至关重要。基于既往研究，Breuer等人（2003）报告称，RANS未能捕捉非定常涡脱落，即使在三维模拟中也只能产生稳态结果；相比之下，DES和LES在不同网格分辨率下均成功解析了非对称脱落。Spalart（2009）报告称，对于高Re流动，DES比RANS或LES更有效，但其面临模糊网格和非单调网格细化行为相关的挑战。Liang与Xue（2014）发现，DES预测梢涡流场特性比RANS-SA更准确，并能捕捉复杂的三维涡结构。此外，Guilmineau等人（2018）证明IDDES模型可准确预测回流泡，并且在流动预测中与实验数据更吻合。Long等人（2021）证实了DDES模拟水翼和船用螺旋桨周围空化流动的能力。Lungu（2022）强调了混合IDDES-SST模型在DARPA潜艇模拟中的效率和精度。Zhang等人（2023）也指出，URANS难以解析小尺度湍流结构，而IDDES更适合预测舰船空气尾流非对称性等复杂现象。尽管如此，捕捉非定常性和湍流脉动尺度仍极具挑战性，因为单元尺寸要求必须适配所采用的每一种湍流模型。因此，作为Abidin等人（2024）前期工作的延续，本研究在Um=1.8235 m/s、Re为3.6×10^6的条件下开展高网格分辨率非定常模拟，以生成非对称尾流动力学和涡结构。当前研究扩展了方法学：依据潜艇特征长度（L、B和D），基于Taylor微尺度细化对网格和数值格式进行参数化，重点关注混合湍流模型和WALE，以考察其解析尾流区湍流的能力。瞬态模拟最初采用y+<5的壁面解析网格（76×10^6个单元），随后采用y+>30的壁面建模网格（56×10^6至74×10^6个单元）；与以往的稳态RANS模拟相比，这产生了明显不同且更详细的结果（涡和湍流脉动），同时未损害关注量（总阻力）的准确性。
英文摘要：
Submarine hydrodynamics presents unique challenges in accurately predicting flow separation, wake structure, and resistance due to complex geometry and turbulent behaviour at high Reynolds (Re) numbers. Traditional Reynolds-Averaged Navier-Stokes (RANS) approaches are often limited in resolving unsteady flow structures and turbulence in the near and far regions. To address these limitations, hybrid RANS-LES models such as Detached Eddy Simulation (DES) and Large Eddy Simulation (LES) offer improved performance in capturing near-wall vortical structures. The capturing of turbulent vortices and wake structures significantly contributes to conduct hydrodynamic noise analysis. Detailed resolution and understanding of these coherent structures help minimize hydroacoustic signatures, essential for submarines' stealth characteristics. Based on prior studies, Breuer et al. (2003) reported that RANS failed to capture unsteady vortex shedding, producing only steady results even in 3D simulations. In contrast, DES and LES successfully resolved asymmetric shedding across different grid resolutions. Spalart (2009) reported that DES is more effective than RANS or LES for high Re flows, although it suffers from challenges related to ambiguous grids and nonmonotonic grid refinement behaviour. Whereas Liang \& Xue (2014) found that DES predicts tip vortex flow characteristics more accurately than RANS-SA and can capture complex 3D vortex structures. In addition, Guilmineau et al. (2018) demonstrated that the IDDES model accurately predicts recirculation bubbles and aligns more closely with experimental data for flow prediction. Long et al. (2021) confirmed the capability of DDES in simulating cavitating flows around hydrofoils and marine propellers. Lungu (2022) highlighted the efficiency and accuracy of the hybrid IDDES-SST model in DARPA submarine simulations. Zhang et al. (2023) also noted that URANS struggles with resolving small-scale turbulence structures, whereas IDDES is better suited for predicting complex phenomena such as ship air wake asymmetry. Nevertheless, capturing the unsteadiness and turbulence fluctuations scales is extremely challenging because the cell size requirements should suit each turbulence model employed. Thus, as a continuation of the prior work of Abidin et al. (2024), the unsteady simulation with high mesh resolution at $U_m=1.8235$ m/s and Re of $3.6\times 10^6$ to generate the asymmetrical wake dynamic and vortical structure. The current research expands the methodology by parameterizing the meshes and numerical scheme based on Taylor microscale refinement with respect to the characteristic length of (L, B and D) of submarine, particularly focusing on hybrid turbulence model and WALE to observe the ability to resolve turbulence in the wake region. The transient simulations were performed initially using wall-resolved mesh (76x10^6 cells) at y^+ < 5 and then wall-modelled mesh (56x10^6 to 74x10^6 cells) at y^+ > 30, which produced notably different and more detailed results (vortices and turbulence fluctuation) than previous steady-state RANS simulations without risking the accuracy of quantity of interest (global resistance).
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
