| In the study of computational turbulence, the success of Large Eddy Simulation (LES) is largely determined by the quality of the sub-filter scale (SFS) model and the properties of the filter used to introduce resolved and unresolved length scales. Additionally, the SFS model and filter are strongly linked, in such a way that the filter process cannot be too contaminated with aliasing errors, as this would cause an inaccurate representation of the flow. One way to alleviate this issue is to use explicit filtering, so that better control over the filter may be achieved, and filter operator errors can be then controlled to a desired order of accuracy. One large advantage to using an explicit filter is that the mathematical definition of the filter may be exploited when considering various SFS models or even different LES techniques. Approximate deconvolution is a technique used in LES, which performs an inverse filtering operation to partly restore the original unfiltered solution. This thesis considers a form of structural modeling, known as approximate deconvolution, to perform a large-eddy simulation of homogeneous, isotropic turbulence. In particular, the discrete explicit filtering technique will be used to perform the deconvolution, and numerical results will show how the approximate solution may be used to perform LES. |
