Large Eddy Simulation in Complex Turbulent Flows with Coarse Resolution

Computational simulation of fluid flows is critical to engineering for a wide variety of technologies. Often, the utility of these simulations is limited by unreliable models for the effects of turbulence. The research proposed here is aimed at developing models for use in large eddy simulation that enable practical and reliable simulations of turbulence in complex flows. By developing truly reliable and broadly applicable turbulence models, the proposed research will have profound impacts in such fields as aeronautics, propulsion, power generation and wind energy, with the potential of enabling great improvements in technologies important to our country and society. The magnitude of this potential impact would be hard to overstate. In addition, by providing training for a graduate student and undergraduate summer interns in the development of mathematical models of complex physical systems, this project will contribute to the highly skilled workforce required to address a broad range of complex problems facing our country and the world. Large eddy simulation (LES), in which the largest scales of turbulence are simulated while the effects of smaller scales are modeled, is one of the most promising approaches to reliably predict the effects of turbulence. However, the resolution requirements for reliable LES with currently available models are generally impractical for complex turbulent flows of technological interest. In essence, for practical applications, it is necessary that the simulations be reliable for as coarse a resolution as possible, which generally violates the assumptions under which the models are formulated, rendering them invalid. New modeling approaches and tools that can address the challenges posed by coarsely resolved -LES have been developed. These include a model-splitting formulation to allow the subgrid models to fulfill different roles, discretization aware model formulations to address the effects of numerical discretization, and statistical models that allow the performance of a LES to be predicted and models to be optimized. The models and tools need to be generalized in several ways to enable reliable coarsely resolved-LES in complex flow applications, and such generalization is the subject of this proposal. Generalized models will be tested against direct numerical simulations, which require no modeling, and experimental data, to assess their reliability, limitations, and resolution requirements. The broader impact objectives of this project include the impact on a wide range of technologies enabled by reliable models for complex turbulent flows, and the training of a graduate student and undergraduate students in computational science and the modeling of complex physical systems. 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.