Single inertial particle statistics in turbulent flows from Lagrangian velocity models

TL;DR

This paper extends a Lagrangian velocity model to include particle inertia effects, enabling stochastic simulation of inertial particle statistics in turbulence, and compares results with DNS and classical models.

Contribution

It introduces a multi-layered Ornstein-Uhlenbeck process for modeling inertial particle velocities, capturing preferential concentration effects at low Stokes numbers.

Findings

Model accurately reproduces second-order velocity structure functions.

Captures acceleration correlation functions and variance changes with Stokes number.

Shows improved agreement with DNS over traditional filtering methods.

Abstract

We present the extension of a modeling technique for Lagrangian tracer particles [B. Viggiano et al., J. Fluid Mech.(2020), vol. 900, A27] which accounts for the effects of particle inertia. Thereby, the particle velocity for several Stokes numbers is modeled directly by a multi-layered Ornstein-Uhlenbeck process and a comparison of key statistical quantities (second-order velocity structure function, acceleration correlation function, and root mean square acceleration) to expressions derived from Batchelor's model as well as to direct numerical simulations (DNS) is performed. In both approaches, Stokes' drag is treated by an approximate ``linear filter'' which replaces the particle position entering the fluid velocity field by the corresponding ideal tracer position. Effects of preferential concentration of inertial particles are taken into account in terms of an effective Stokes number that is determined from the zero crossing of the acceleration correlation function from DNS. This approximation thus allows the modeling of inertial particle statistics through stochastic methods and models for the Lagrangian velocity; the particle velocity is effectively decoupled from the particle position. In contrast to the ordinary filtering technique [Cencini et al., J. Turbul. (2006), 7, N36], our method captures the effects of preferential concentration of particles at low Stokes numbers which manifest themselves for instance by a sharp decrease of the acceleration variance for increasing Stokes numbers.