Data-driven fractional subgrid-scale modeling for scalar turbulence: A nonlocal LES approach

Filtering the passive scalar transport equation in large-eddy simulation (LES) of turbulent gives rise to closure term corresponding unresolved flux. Understanding and respecting statistical features subgrid-scale (SGS) flux is a crucial point robustness predictability LES. In this work, we investigate intrinsic nonlocal behavior SGS through studying its two-point statistics obtained from filtered direct numerical (DNS) data for homogeneous isotropic turbulence (HIT). Presence long-range correlations true urges go beyond conventional local modeling approaches that fail predict non-Gaussian scalars. Here, propose an appropriate model microscopic motions by taking into account Boltzmann (FBTE) scalar. FBTE, approximate equilibrium distribution with ? -stable Lévy essentially incorporates power-law resemble observed Generic ensemble-averaging such FBTE lets us formulate continuum level appearing terms fractional-order Laplacian inherently nonlocal. Through data-driven approach, infer optimal version our using high-fidelity correlation function between gradient, sparse linear regression. priori test, yields promising performance reproducing probability (PDF) dissipation variance compared PDF DNS data. • We have aimed study conserved scalars flows seeking arising as equation. Due potential framework sources spatial nonlocality dynamics, managed flow. Using proper assumptions at kinetic level, derived resolved concentration. procedure being calibrated (in sense) form so it capable capturing embedded ground-truth