Analysis of interpolation schemes for the accurate estimation of energy spectrum in Lagrangian methods

Abstract

Direct evaluation of energy spectra in purely Lagrangian meshless methods is a challenging task. On the other hand, improvement of turbulence modeling in a Lagrangian framework relies strongly on our ability to estimate energy spectra up to the maximally resolved wavenumber. In this paper we compare different strategies to extract energy spectra from a velocity field defined on a scattered set of points. Spectra can be directly evaluated from irregularly distributed sample by using Discrete Fourier Transform (DFT) and their regularized versions. Alternative procedures require a preliminary interpolation into a grid where, on a second stage, a Fourier analysis can be performed. As a last approach a Moving Least Squares (MLSs) technique for preliminary interpolation is investigated and the results are discussed. Although exhibiting good accuracy in the low-moderate wavenumber window, the first two strategies introduce unacceptable large errors in the near-maximal-resolved wavenumber, where the highest accuracy is often required. Here we propose a second-order Moving Least Squares (MLSs) scheme as an optimal tool that allows us to reproduce precisely the energy spectrum over the entire wavenumber window. We discuss the importance of this result with respect to the development of accurate turbulence models for purely Lagrangian meshless methods.