Passive scalar mixing and decay at finite correlation times in the Batchelor regime

• Published in: Journal of fluid mechanics 2017-08, Vol.824, p.785-817
• Main Authors: Aiyer, Aditya K., Subramanian, Kandaswamy, Bhat, Pallavi
• Format: Article
• Language: English
• Subjects: Computational fluid dynamics; Computer simulation; Correlation; Decay; Decay rate; Derivatives; Dye dispersion; Fluctuations; Fluid mechanics; Fluids; Formulas (mathematics); Initial conditions; Lagrangian current measurement; Mathematical models; Partial differential equations; Reynolds number; Scaling; Steady state; Velocity
• ISSN: 0022-1120; 1469-7645
• DOI: 10.1017/jfm.2017.364

An elegant model for passive scalar mixing and decay was given by Kraichnan (Phys. Fluids, vol. 11, 1968, pp. 945–953) assuming the velocity to be delta correlated in time. For realistic random flows this assumption becomes invalid. We generalize the Kraichnan model to include the effects of a finite correlation time, $\unicode[STIX]{x1D70F}$ , using renewing flows. The generalized evolution equation for the three-dimensional (3-D) passive scalar spectrum $\hat{M}(k,t)$ or its correlation function $M(r,t)$ , gives the Kraichnan equation when $\unicode[STIX]{x1D70F}\rightarrow 0$ , and extends it to the next order in $\unicode[STIX]{x1D70F}$ . It involves third- and fourth-order derivatives of $M$ or $\hat{M}$ (in the high $k$ limit). For small- $\unicode[STIX]{x1D70F}$ (or small Kubo number), it can be recast using the Landau–Lifshitz approach to one with at most second derivatives of $\hat{M}$ . We present both a scaling solution to this equation neglecting diffusion and a more exact solution including diffusive effects. To leading order in $\unicode[STIX]{x1D70F}$ , we first show that the steady state 1-D passive scalar spectrum, preserves the Batchelor (J. Fluid Mech., vol. 5, 1959, pp. 113–133) form, $E_{\unicode[STIX]{x1D703}}(k)\propto k^{-1}$ , in the viscous–convective limit, independent of $\unicode[STIX]{x1D70F}$ . This result can also be obtained in a general manner using Lagrangian methods. Interestingly, in the absence of sources, when passive scalar fluctuations decay, we show that the spectrum in the Batchelor regime at late times is of the form $E_{\unicode[STIX]{x1D703}}(k)\propto k^{1/2}$ and also independent of $\unicode[STIX]{x1D70F}$ . More generally, finite $\unicode[STIX]{x1D70F}$ does not qualitatively change the shape of the spectrum during decay. The decay rate is however reduced for finite $\unicode[STIX]{x1D70F}$ . We also present results from high resolution ( $1024^{3}$ ) direct numerical simulations of passive scalar mixing and decay. We find reasonable agreement with predictions of the Batchelor spectrum during steady state. The scalar spectrum during decay is however dependent on initial conditions. It agrees qualitatively with analytic predictions when power is dominantly in wavenumbers corresponding to the Batchelor regime, but is shallower when box-scale fluctuations dominate during decay.