Scholarlink

<p>This study addresses significant prediction inaccuracies in aviation fuel centrifugal pumps under low-flow conditions, where strong rotational and separated flows dominate. A Very Large-Eddy Simulation-Reynolds-Averaged Navier–Stokes (VLES-RANS) hybrid modeling approach is used to evaluate the applicability of turbulence models across external performance, internal flow characteristics, and computational cost. Traditional Reynolds-Averaged Navier–Stokes (RANS) models, limited by time-averaging, fail to resolve secondary flows and vortex breakdown induced by intensified centrifugal forces under 0.2<i>Q</i><sub><i>d</i></sub>–0.4<i>Q</i><sub><i>d</i></sub> conditions. In contrast, the Very Large-Eddy Simulation-Shear Stress Transport k-ω (VLES-SST k-ω) model reduces the maximum head deviation from 2.76% to 0.32% and the efficiency deviation from 6.45% to 3.32% across 0.2<i>Q</i><sub><i>d</i></sub>–0.4<i>Q</i><sub><i>d</i></sub> flow rates, significantly improving small-flow performance prediction. For complex flow features, the hybrid model outperforms RANS in resolving entropy production, pressure distributions, and vortex structures, achieving 90.4% agreement in velocity components with Large-Eddy Simulation (LES) results. At 0.2<i>Q</i><sub><i>d</i></sub> and 0.4<i>Q</i><sub><i>d</i></sub>, pressure fluctuation amplitude and frequency match 80% of those from LES results, enabling precise quantification of irreversible losses and momentum transfer. In the RANS-VLES interface region, the hybrid models show good transition characteristics. Computationally, it reduces runtime by 30–50% compared to LES, with costs at low-flow rates decreasing by 55%—effectively balancing accuracy and efficiency effectively. By resolving large-scale vortices in the main flow while maintaining near-wall computational efficiency, the VLES-SST k-ω model overcomes RANS limitations. It represents an optimal choice for simulating high-Reynolds-number flows in complex rotating machinery, such as aviation fuel pumps, particularly under challenging low-flow operating conditions.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

High-fidelity numerical prediction of aviation fuel pumps based on the VLES-RANS hybrid algorithm

  • 期刊论文

  • 发表日期：

    2026年1月6日

查看全文
  • Mingzhen Xiao,
  • Yibin Li,
  • Jianghao Hu,
  • Bo Wang,
  • Jingbo Chen,
  • Xin Liu

摘要

This study addresses significant prediction inaccuracies in aviation fuel centrifugal pumps under low-flow conditions, where strong rotational and separated flows dominate. A Very Large-Eddy Simulation-Reynolds-Averaged Navier–Stokes (VLES-RANS) hybrid modeling approach is used to evaluate the applicability of turbulence models across external performance, internal flow characteristics, and computational cost. Traditional Reynolds-Averaged Navier–Stokes (RANS) models, limited by time-averaging, fail to resolve secondary flows and vortex breakdown induced by intensified centrifugal forces under 0.2Qd–0.4Qd conditions. In contrast, the Very Large-Eddy Simulation-Shear Stress Transport k-ω (VLES-SST k-ω) model reduces the maximum head deviation from 2.76% to 0.32% and the efficiency deviation from 6.45% to 3.32% across 0.2Qd–0.4Qd flow rates, significantly improving small-flow performance prediction. For complex flow features, the hybrid model outperforms RANS in resolving entropy production, pressure distributions, and vortex structures, achieving 90.4% agreement in velocity components with Large-Eddy Simulation (LES) results. At 0.2Qd and 0.4Qd, pressure fluctuation amplitude and frequency match 80% of those from LES results, enabling precise quantification of irreversible losses and momentum transfer. In the RANS-VLES interface region, the hybrid models show good transition characteristics. Computationally, it reduces runtime by 30–50% compared to LES, with costs at low-flow rates decreasing by 55%—effectively balancing accuracy and efficiency effectively. By resolving large-scale vortices in the main flow while maintaining near-wall computational efficiency, the VLES-SST k-ω model overcomes RANS limitations. It represents an optimal choice for simulating high-Reynolds-number flows in complex rotating machinery, such as aviation fuel pumps, particularly under challenging low-flow operating conditions.