发表机构
German Aerospace Center (DLR)（德国航空航天中心）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究通过数值模拟检验机翼前缘湍流线性假设，发现快速畸变理论在含能尺度失效，压力-应变再分配导致局部湍流生成，并改进Amiet噪声预测。
AI 中文摘要
接近机翼前缘的湍流在最后几个前缘半径内受到拉伸和阻塞。快速畸变理论假设撞击前的流场是上游湍流的线性映像；我们直接检验了这一假设。我们在NACA 0012机翼上计算网格湍流，并在机头周围的三个全等曲面上采样，得到上游和撞击前测点之间的互谱张量。相对于无翼对照，线性谱在含能尺度（$κ= k\\,r_{\mathrm{LE}} \approx 0.3$）处已经下降了三倍，表明撞击前湍流的很大一部分是局部产生的，而非继承而来。同一位置也出现在物理空间中：各向异性不变量$III_b$在弦的最后几个百分点内从长椭球变为扁椭球，偏离了平面应变快速畸变预测。这一变化是由压力-应变再分配产生的，它将应变放大的壁法向应力排入其他分量；该项在对照中不存在，并在累积应变变得快速时开始活跃。在前缘半径十四倍的变化范围内，符号变化发生在$x \approx -1.5\\,r_{\mathrm{LE}}$处，且生成的湍流以展向为主。在二阶上，该映射仍然可用：当插入标准Amiet预测时，它改善了与实测前缘噪声的一致性，并且当仅提供前缘半径和自由流速度时，相同的系数可转移到独立的已发表案例中。
英文摘要
Turbulence approaching an aerofoil leading edge is strained and blocked in the final few nose radii. Rapid-distortion theory assumes that the pre-impact field is a linear image of the upstream turbulence; we test that assumption directly. Grid turbulence is computed over a NACA~0012 aerofoil and sampled on three congruent surfaces around the nose, giving the cross-spectral tensor between upstream and pre-impact stations. Relative to a no-aerofoil control, the linearity spectrum falls by a factor of three already at the energy-containing scales ($κ= k\,r_{\mathrm{LE}} \approx 0.3$), showing that a substantial part of the pre-impact turbulence is generated locally rather than inherited. The same location appears in physical space: the anisotropy invariant $III_b$ changes sign from prolate to oblate within the last few per cent of chord, departing the plane-strain rapid-distortion prediction. The change is produced by pressure--strain redistribution, which drains the strain-amplified wall-normal stress into the other components; the term is absent in the control and becomes active once the accumulated strain is rapid. Across a fourteen-fold range of nose radius the sign change occurs at $x \approx -1.5\,r_{\mathrm{LE}}$ and the generated turbulence is spanwise-dominated. At second order the mapping remains usable: when inserted into a standard Amiet prediction it improves agreement with measured leading-edge noise, and the same coefficients transfer to independent published cases when only nose radius and free-stream speed are supplied.
