A Framework for Predictive Hybrid Models of Turbulence

先进技术系统的设计和运行依赖于使用可靠的计算机模拟来预测其行为的能力。许多重要的系统，包括汽车、航空、推进、发电和风能系统，都涉及复杂的流体流动。然而，这些系统中的流体流动大多是湍流的，这一事实阻碍了他们可靠的模拟。这意味着流体运动是混乱和不可预测的。目前还没有准确和广泛适用的模型来描述湍流对这种流动的影响。最近对复杂湍流进行计算机模拟的方法可以解决以前模型的几个基本局限性，并且它们无法产生对复杂湍流的准确模拟。因此，为了准确地描述复杂流体系统中的流体运动，必须改进湍流模型。这需要解决这一研究领域的几个突出挑战。新提议的建模框架旨在应对这些挑战，以实现对湍流的准确和可靠的计算机模拟。这种长期寻求的能力将允许开发更有能力和更有效的流体流动系统，如上面列出的领域。长期以来，人们已经认识到，雷诺平均Navier-Stokes(RANS)和大涡模拟(LES)湍流模型具有互补的优势和劣势，这表明它们混合起来可以产生更强大的模型。然而，之前的杂交技术被发现存在根本性的缺陷。一种新的混合建模方法消除了这些缺陷，并为本文提出的研究奠定了基础。这种方法可以考虑三个额外的湍流建模挑战，当解决这些挑战时，将产生高度可靠和广泛适用的混合RAN/LES模型。这些挑战是：已解决和未解决的湍流之间的能量交换，以允许已解决的波动的快速发展；消除由于复杂流体系统中通常需要的强非均匀分辨率而导致的大涡模拟中的大误差；以及，推广大涡模拟模型来解释复杂湍流混合模拟中不可避免地出现的未解决的湍流的各向异性。这些发展的结果将是强大的预测性混合RAN/LES模型，它将使涉及湍流流动的许多系统的技术进步成为可能。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

Design and operation of advanced technological systems rely on the ability to predict their behavior using reliable computer simulation. Many important systems including automotive, aeronautic, propulsion, power generation and wind energy systems involve complex fluid flows. However, their reliable simulation is hindered by the fact that the fluid flows in these systems are mostly turbulent. This means that the fluid motion is chaotic and unpredictable. There are currently no accurate and broadly applicable models to describe the effects of turbulence on such flows. Recent approaches to computer modeling of complex turbulent flows could address several fundamental limitations of the previous models and where they have failed to produce accurate simulations of complex turbulent flows. Improving turbulence models are therefore necessary for accurate description of the fluid motion in complex fluid systems. This requires addressing several outstanding challenges in this field of research. The new proposed modeling framework is aimed at addressing these challenges to enable accurate and reliable computer simulation of turbulent flows. This long- sought capability will allow the development of more capable and efficient fluid flow systems, like those in the areas listed above.It has long been recognized that Reynolds averaged Navier-Stokes (RANS) and large eddy simulation (LES) models of turbulence have complementary strengths and weaknesses suggesting their hybridization to produce a more capable model. However, previous hybridization techniques were found to have fundamental flaws. A new hybrid modeling approach eliminates these flaws and forms the basis for the research proposed here. This approach enables the consideration of three additional turbulence modeling challenges, which when addressed will result in highly reliable and broadly applicable hybrid RANS/LES models. These challenges are: the active exchange of energy between the resolved and unresolved turbulence to allow rapid development of resolved fluctuations; the elimination of large errors in LES that arise from the strongly inhomogeneous resolution that is usually necessary in complex fluid systems; and, the generalization of LES models to account for anisotropy of the unresolved turbulence which inevitably arises in hybrid simulations of complex turbulent flows. The result of these developments will be robust predictive hybrid RANS/LES models that will enable technological advances in many systems that involve turbulent fluid flow.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.