Hostname: page-component-7b7567c8c-7pwfn Total loading time: 0 Render date: 2026-11-16T23:26:36.260Z Has data issue: false hasContentIssue false

A global linearized framework for modelling shear dispersion and turbulent diffusion of passive scalar fluctuations

Published online by Cambridge University Press:  29 March 2021

Thomas Ludwig Kaiser*
Affiliation:
Laboratory of Flow Instabilities and Dynamics, TU Berlin, 10623 Berlin, Germany
Kilian Oberleithner
Affiliation:
Laboratory of Flow Instabilities and Dynamics, TU Berlin, 10623 Berlin, Germany
*
†Email address for correspondence: t.kaiser@tu-berlin.de

Abstract

In the field of gas-turbine engineering, entropy waves and fluctuations in fuel–air mixing are of significant importance. The impact of either mechanism on thermoacoustic stability of the engine and combustion noise considerably depends on how they are convected in the combustion chamber. In this work, a novel method is employed to analyse their convection. Both effects are modelled using a transport equation of a passive scalar linearized around the mean field. The linearized transport equation is discretized using finite elements. It is shown that turbulent passive scalar transport can be described by an eddy diffusivity in the linear framework. The method is furthermore validated against direct numerical simulation (DNS) of passive scalar transport in a turbulent channel flow. Taking the mean flow from the DNS as input, the method reproduces transport of periodic passive scalar fluctuations with high accuracy at negligible numerical expense. Previous studies investigated destructive interference of the passive scalar due to a non-uniform mean flow profile, a process termed mean flow shear dispersion. The method introduced in this study, however, allows us to additionally quantify the impact of molecular and turbulent diffusion. For the channel flow under investigation, mean flow shear dispersion is the dominant mechanism at low frequencies while, at higher frequencies, turbulent diffusion needs to be accounted for to reproduce the DNS results. Molecular diffusion, however, only has a minor effect on the overall convection in the turbulent channel flow.

Information

Type
JFM Papers
Copyright
© The Author(s), 2021. Published by Cambridge University Press
Figure 0

Figure 1.Sketch of the DNS domain of Morgans et al. (2013) and the strategy to measure the convection of passive scalar fluctuations via TF and IR.

Figure 1

Figure 2.Field of periodic passive scalar fluctuation for ${St}=0.5{\rm \pi}$ and ${St}=3{\rm \pi}$; The results in the first row are obtained analytically, using (4.1). The results in the second to fourth rows are obtained using FELiCS and are based on artificial diffusion, $D_{a}=0.001 D_{m}$, molecular diffusion, $D_{m}$, and the sum of molecular and turbulent diffusion, $D_{m}+D_{t}$, respectively.

Figure 2

Figure 3.TF (see (2.1)) based on DNS, the analytical model (4.1) and the linearized approach for various diffusion models; the grey area shows the range of results, if turbulent Prandtl numbers between ${Pr}_{t}=0.6$ and ${Pr}_{t}=1.2$ are applied.

Figure 3

Figure 4.Eddy viscosity based on the Boussinesq approximation, $\nu_{t,B}$, given by (3.12) and the resulting turbulent diffusivity, $D_{t,B}$, based on a turbulent Prandtl number of ${Pr}_{t}=0.9$. Furthermore, the eddy viscosity based on the Cess model, $\nu_{t,B}$, obtained by (3.11) is shown.

Figure 4

Figure 5.Comparison of IRs of various models to DNS data evaluated for $x_2=6.0 h$.

Figure 5

Figure 6.Comparison of FELiCS results transferred to time domain with DNS data.

Figure 6

Figure 7.Diffusion term of (3.14) split into molecular and turbulent contributions in the $x$- and $y$-directions.