发表机构
University of Massachusetts Amherst（马萨诸塞大学阿默斯特分校）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文通过理论分析和DSMC、R13、R26数值实验，证明稀薄空腔流中热通量无法唯一确定四阶封闭态，且R26在捕捉环流和热通量方向时优于R13，差异主要存在于壁面与角落层。
AI 中文摘要
冷到热的热传递是稀薄空腔流动中一个显著的非平衡特征，常被用来判断连续介质封闭模型是否捕捉到了高阶输运。我们考察了这种一致性实际上确立了何种结论。精确的热通量平衡表明，四阶贡献仅通过复合张量(A_{ij}=R^{\mathrm{cl}}*{ij}+Δδ*{ij}/3)的散度进入。一个紧支撑的Airy构造产生了一个无限维的对称无散扰动族，该扰动族保持通量侧壁面迹线。二维物理空间、三维速度空间（2D3V）的描述也使得横向分量(A_{zz})未被观测到。因此，热通量无法唯一确定潜在的四阶状态。我们使用直接模拟蒙特卡洛（DSMC）系综以及独立的正则化13矩（R13）和正则化26矩（R26）解在两个稀薄度水平下测试了其后果。逆梯度输运和主导张量通道在网格和粒子数变化下保持不变，而局部标量四阶矩和逆梯度区域的空间组织则更为敏感。一个有限粒子修正的四阶矩审计表明，在较高稀薄度下，解析的R26与DSMC复合张量差异中超过95%位于离散散度零空间中，而投影采样噪声仅隐藏了其约一半的能量。R26比R13更准确地捕捉了环流和热通量方向，但剩余差异集中在壁面和角落层。
英文摘要
Cold-to-hot heat transfer is a conspicuous non-equilibrium feature of rarefied cavity flows and is often used to judge whether a continuum closure captures higher-order transport. We examine what such agreement actually establishes. The exact heat-flux balance shows that the fourth-order contribution enters only through the divergence of the composite tensor (A_{ij}=R^{\mathrm{cl}}*{ij}+Δδ*{ij}/3). A compactly supported Airy construction yields an infinite-dimensional family of symmetric divergence-free perturbations that preserve the flux-side wall trace. A two-dimensional-physical, three-dimensional-velocity (2D3V) description also leaves the transverse component (A_{zz}) unobserved. Heat flux therefore cannot uniquely identify the underlying fourth-order state. We test the consequences using direct simulation Monte Carlo (DSMC) ensembles and independent regularized 13-moment (R13) and regularized 26-moment (R26) solutions at two rarefaction levels. Counter-gradient transport and the dominant tensorial channel persist under changes in grid and particle number, whereas the local scalar fourth moment and the spatial organization of the counter-gradient region are more sensitive. A finite-particle-corrected fourth-moment audit shows that more than 95% of the resolved R26--DSMC composite-tensor discrepancy at the higher rarefaction lies in the discrete divergence-null space, whereas projected sampling noise hides only about half of its energy. R26 captures the circulation and heat-flux direction more closely than R13, yet the remaining differences concentrate in the wall and corner layers.
