Computational modeling of stably stratified, turbulent shear flows

The present thesis is motivated by the desire to understand and to predict the effects of stable stratification on turbulent flow and passive scalar dispersion. A methodology which enables the simulation of stably stratified, turbulent boundary-layer flows in a flexible solver has been established. Data from highly resolved large-eddy simulations of channel flow and direct numerical simulation of free turbulent shear flow, originating from Kelvin-Helmholtz instabilities, is used to investigate the effects of stable stratification on turbulent shear flow and scalar dispersion. Changes in turbulence dynamics and structures have been of particular interest, as these areas are largely unexplored in the existing literature. The present work suggests that, in channel flow, three distinct regions of the channel can be identified; the shear region closest to the wall, the transition region, and the buoyancy region in the center of the channel. It has been established that the most obvious effects of imposed stable stratification are reduced vertical turbulent fluctuations and transport, as well as a related increase in turbulence anisotropy. Furthermore, the results reveal important changes in the structural state of the turbulence. The “compression” of vertical structures can be quantified by the structure dimensionality tensor, and this effect is demonstrated to be significant in some cases. The transition region in channel flow is strongly homogenized by increased stratification, which suggests an increased decoupling between the inner shear region and outer buoyancy region. From the free-shear flow data, it appears that the Reynolds stress anisotropy increase up to a certain level of stratification, corresponding to a gradient Richardson number of approximately 0.4. For stronger stratification, however, the turbulence anisotropy is reduced, most likely caused by relaminarization. Moreover, analysis of the free-shear turbulence dynamics suggests that the shear production term removes energy from the turbulent field near the edge of the shear layer, whereas the buoyancy destruction term is an instigator of turbulence. It is shown that this is related to the observed turbulent flux reversal, i.e. a change in sign of turbulent shear stress, . It is also found that, for the Reynolds numbers considered presently, the viscous dissipation rate is highly anisotropic. Another interesting feature of free-shear flow is how the imposition of strong stable stratification seems to mimic the non-local pressure effects encountered in the vicinity of impenetrable walls. However, the kinematic blocking effect of walls does not seem to be emulated by the imposed stable stratification, despite the overall reduction of vertical velocity fluctuations. Turbulence models employed in stably stratified flows ought to include non-local information to incorporate important effects of stratification. Due to changes in the turbulence field, passive scalar transport is altered significantly by stable stratification. In channel flow, the effect of stratification is stronger on scalar releases in the buoyancy region. Vertical scalar flux is reduced, leading to inhibited vertical dispersion and higher peak concentrations downstream than in the neutral case. For example, four boundary-layer lengths downstream, peak concentration was more than 50 % higher for Rit = 240 compared to neutral flow. Whereas peak mean concentration downstream decays exponentially in the neutral case, this is not the case under imposed stratification. As a predictive tool, eddy-diffusivity models for scalar transport therefore need to take stratification into account