Yi-Ju Chou

Modeling dilute sediment suspension using large-eddy simulation with a dynamic mixed model

Transport of suspended sediment in high Reynolds number channel flows [Re=O(600 000)] is simulated using large-eddy simulation along with a dynamic-mixed model (DMM). Because the modeled sediment concentration is low and the bulk Stokes’ number (Stb) is small during the simulation, the sediment concentration is calculated through the use of the Eulerian approach. In order to employ the DMM for the suspended sediment, we formulate a generalized bottom boundary condition in a finite-volume formulation that accounts for sediment flux from the bed without requiring specific details of the underlying turbulence model. This enables the use of the pickup function without requiring any assumptions about the behavior of the eddy viscosity. Using our new boundary condition, simulations indicate that the resolved component of the vertical flux is one order of magnitude greater than the resolved subfilter-scale flux, which is in turn one order of magnitude greater than the eddy-diffusive flux. Analysis of the behavio...

An Euler-Lagrange model for simulating fine particle suspension in liquid flows

This article presents a two-way coupled Euler-Lagrange model to simulate the suspension of fine particles in liquid flows. The goal is to develop a three-dimensional numerical model that is capable of replicating the detailed features of particle-laden turbulent flow. A two-phase fractional-step projection method is developed to ensure mixture incompressibility by solving a modified Poisson equation for pressure, which in turn affects the particle motion through the pressure gradient. An efficient particle-moving algorithm that exchanges particles between Eulerian meshes is developed that automatically retains the necessary particle information at each time step and does not require any Lagrangian particle tracking. A soft-sphere particle collision model is employed to avoid excessive particle overlap and to achieve the random closed packing limit for depositing particles. Since particles are all inherently localized in Eulerian meshes, efficient particle searches can be achieved when calculating particle-particle collision. This model is then used to simulate the gravitational settling of particles, and the results confirm the effect of mixture incompressibility and demonstrate that the model is capable of reaching the random closed packing limit. Numerical examples for the flow problems of particle-induced stratification are conducted, and the model is able to reveal the detailed features of particle-laden flows. Sensitivity on the grid resolution and deviations from the existing single-phase model results are also discussed.

Numerical study of particle-induced Rayleigh-Taylor instability: Effects of particle settling and entrainment

In this study, we investigate Rayleigh-Taylor instability in which the density stratification is caused by the suspension of particles in liquid flows using the conventional single-phase model and Euler-Lagrange (EL) two-phase model. The single-phase model is valid only when the particles are small and number densities are large, such that the continuum approximation applies. The present single-phase results show that the constant settling of the particle concentration restricts the lateral development of the vortex ring, which results in a decrease of the rising speed of the Rayleigh-Taylor bubbles. The EL model enables the investigation of particle-flow interaction and the influence of particle entrainment, resulting from local non-uniformity in the particle distribution. We compare bubble dynamics in the single-phase and EL cases, and our results show that the deviation between the two cases becomes more pronounced when the particle size increases. The main mechanism responsible for the deviation is particle entrainment, which can only be resolved in the EL model. We provide a theoretical argument for the small-scale local entrainment resulting from the local velocity shear and non-uniformity of the particle concentration. The theoretical argument is supported by numerical evidence. Energy budget analysis is also performed and shows that potential energy is released due to the interphase drag and buoyant effect. The buoyant effect, which results in the transformation of potential energy into kinetic energy and shear dissipation, plays a key role in settling enhancement. We also find that particle entrainment increases the shear dissipation, which in turn enhances the release of potential energy.

Three-dimensional wave-coupled hydrodynamics modeling in South San Francisco Bay

In this paper, we present a numerical model to simulate wind waves and hydrodynamics in the estuary. We employ the unstructured-grid SUNTANS model for hydrodynamics, and within this model we implement a spectral wave model which solves for transport of wave action density with the finite-volume formulation. Hydrodynamics is coupled to the wave field through the radiation stress. Based on the unstructured grid and finite-volume formulation of SUNTANS, the radiation stress is implemented in a way that directly calculates the divergence of transport of the wave-induced orbital velocity. A coupled hydrodynamics-wave simulation of San Francisco Bay is then performed. Through the input of wind forcing that is obtained from the reconstructed wind field, the model is capable of predicting wave heights that are in good agreement with the field measurements. We examine the importance of modeling sea bed dissipation in muddy shallow water environments by using a bottom friction model and a bed mud model with different mud layer thicknesses. Moreover, currents driven by wave shoaling and dissipation are investigated in the presence of abrupt bathymetric change. We find that spatially varying wave heights induced by spatially heterogeneous bottom mud dissipation produce wave-driven currents that are stronger than those induced by wave shoaling and can be of the same order as the tidal currents in shallow water. HighlightsAn unstructured-grid model is presented to simulate hydrodynamics under wind-wave and tidal forcing.Simulations are performed to study wave-coupled hydrodynamics in San Francisco Bay.Currents driven by waves are investigated in the presence of abrupt bathymetric change.

Three‐Dimensional Modeling of Fine Sediment Transport by Waves and Currents in a Shallow Estuary

A suspended sediment transport model is implemented in the unstructured-grid SUNTANS model and applied to study fine-grained sediment transport in South San Francisco Bay. The model enables calculation of suspension of bottom sediment based on combined forcing of tidal currents and wind waves. We show that accurate results can be obtained by employing two-size classes which are representative of microflocs and macroflocs in the Bay. A key finding of the paper is that the critical calibration parameter is the ratio of the erosion of the microflocs to macroflocs from the bed. Different values of this erosion ratio are needed on the shallow shoals and deeper channels because of the different nature of the sediment dynamics in these regions. Application of a spatially variable erosion ratio and critical shear stress for erosion is shown to accurately reproduce observed suspended sediment concentration at four-field sites located along a cross-channel transect. The results reveal a stark contrast between the behavior of the suspended sediment concentration on the shoals and in the deep channel. Waves are shown to resuspend sediments on the shoals, although tidal and wind-generated currents are needed to mix the thin wavedriven suspensions into the water column. The contribution to the suspended sediment concentration in the channel by transport from the shoals is similar in magnitude to that due to local resuspension. However, the local contribution is in phase with strong bottom currents which resuspend the sediments, while the contribution from the shoals peaks during low-water slack tide.

An Unstructured Immersed Boundary Method for Simulation of Flows over Bottom Topography

An immersed boundary method (IBM) for unstructured grids is presented for the simulation of flows over bottom topography. We implement the method in a nonhydrostatic, unstructured-grid code that solves the Navier-Stokes equations under the Boussinesq approximation. The equations are discretized on a staggered Cartesian z-level grid in the vertical plane and an unstructured triangular grid in the horizontal plane, and the solution to the hydrodynamics is based on a three dimensional semi-implicit free-surface approach together with the pressure-correction method for the non-hydrostatic pressure. In the simulation domain, the cells which contain the physical boundaries are treated as the ghost cells. Thus, each ghost cell itself has an immersed boundary, which is used to represent the presence of the physical boundaries based on the assumption that boundaries are piecewise linear within each cell. In the ghost cell, rather than directly imposing the desired boundary condition on the cell edge, which is the case for the z-level grids in the absence of an immersed boundary, the ghost-cell quantities are imposed at those edges to satisfy the desired boundary conditions at the piecewise-linear immersed boundaries. The ghost-cell quantities are obtained using linear extrapolations from the known interior points. Before solving the discrete governing equations, the spatial coefficients for the extrapolation are obtained based on geometries. Unlike previous studies on IBM, which impose the forcing term to make velocities to satisfy the desired boundary conditions, this study focuses on applying the immersed boundary method for solving the pressure field. The potential flow over a Gaussian hump is simulated using the proposed immersed boundary method and the results are compared with the simulation without IBM.