Simulation and modeling of non-canonical turbulent flows - Junlin Yuan, Michigan State University
Junlin Yuan, Assistant Professor, Department of Mechanical Engineering, Michigan State University
Abstract: The bulk of wall turbulence research has focused, perhaps disproportionately, on canonical flows along smooth flat plates with uniform freestream conditions. However, in engineering and environmental applications, such as flow around hydraulic turbine blades, navy platforms and in rivers, we see a wide range of dynamically complicated flow fields, affected by surface roughness, curvature, freestream pressure gradients, and unsteadiness. The consequence is that existing descriptions and models of turbulence have limited utility to design practice. My long term goal is to build essential physics into models, to enable a consistent description for turbulence across a wide range of flow complexities. The first part of the talk will be focused on understanding and modeling for rough-walled, equilibrium turbulence or non-equilibrium ones subjected to longitudinal pressure gradients. Using data from direct and large-eddy simulations (DNS and LES), I will show that roughness significantly modifies turbulence under strong spatial or temporal accelerations. I will also show an example of machine-learned modeling of hydrodynamic drag from roughness with arbitrary topography.
The second part of the talk is on using DNS to better understand river hyporheic exchange, a phenomenon of turbulent flow bounded by rough, permeable walls. In our understanding of riverine systems, a gap of knowledge exists in how pore-scale heterogeneities affect multiscale hydrologic and biogeochemical processes. I will show that dynamics at the scale of sediment grains and small roughness formed by uppermost-layer grains—typically ignored in existing predictive approaches—can be important for exchanges across a flat bed. Specifically, pore-resolved DNS of flows bounded by beds modeled as closely packed spheres were carried out. Results showed that bed roughness induces deep, multiscale subsurface flow paths that yield residence time distribution with a power-law tail. The main driver appears to be the interfacial pressure variation generated by roughness, a mechanism fundamentally similar to the effect of bedforms. Future work will investigate (i) pore-scale dynamics in transient or spatially varying river flows, (ii) reactive solute transport affected by pore-scale dynamics, and (iii) potential link with reduced-order transport models used for water management practice.
About the Speaker: Dr. Junlin Yuan is an assistant professor in Department of Mechanical Engineering at Michigan State University. She obtained both MS and PhD degrees (2015) from Queen's University, Canada. Her PhD research focused on simulation (DNS and LES) and modeling of turbulent flows in the context of rough hydraulic turbine blades. At MSU, she continued to use large-scale simulations to identify the dynamics of wall-bounded turbulence with various complexities, and to develop physics-based, data-driven models. Topics include turbulence responses to acceleration/deceleration, wall roughness, wall permeability, curvature, and rotation. Applications cover engineering, environmental and bio-locomotive topics. Her research group is currently funded by ONR, NSF and the industry.