Single inertial particle statistics in turbulent flows from Lagrangian velocity models

We present the extension of a modeling technique for Lagrangian tracer particles [B. Viggiano et al., J. Fluid Mech. 900, A27 (2020)] which accounts for the effects of particle inertia. Thereby, the particle velocity for several Stokes numbers is modeled directly by a multilayered 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 indirectly 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. 7, N36 (2006)], our method reproduces the empirically observed sharp decrease of acceleration variance for increasing Stokes numbers. Furthermore, we discuss how our modeling approach could contribute to a better experimental characterization of inertial particle dynamics.