Numerical models for scour and liquefaction around object under currents and waves

Local scour and liquefaction are two of the most important processes which affect the interactions between fluid, object and sediment when an object (such as bridge pier, offshore foundation, etc.) is exposed to currents and waves. In the present study, numerical models are developed to understand these complicated processes. For the local scour process, two-dimensional and three-dimensional models are developed respectively. In the two-dimensional model, shallow water equations with finite volume method on unstructured mesh are used. The two-dimensional model uses the Godunov scheme and approximate Riemann solvers. Hydrodynamics and sediment transport equations are coupled and solved simultaneously. Asymptotic analysis of the system eigenvalues is given and the approximation is compared with the numerical results. The model developed in this thesis can deal with wetting and drying automatically. Discontinuity of the flow, such as a hydraulic jump, can be captured. For the three dimensional model, free water surface and automatic mesh deformation for the bed are incorporated in the model. The Reynolds Averaged Navier-Stokes (RANS) turbulence model is used to simulate the turbulent flow field. The turbulence model used is k-&epsis; Model. Two interfaces (water and air, water and sediment) present in the domain are captured with different approaches. The free surface of the flow is captured by Volume of Fluid (VOF) scheme which is an Eulerian approach. A new method for the VOF scheme is proposed to reduce the computational time while retaining relatively good accuracy. The water-sediment interface (bed) is captured with a moving mesh method which is a Lagrangian approach. Unlike the two-dimensional model, the flow field is coupled with sediment transport (both bed load and suspended load) using a quasi-steady approach. Numerical simulations are carried out and compared with experimental results. Good results are obtained with the proposed model. The flow field compares well with the experimental observations. Scour patterns are similar to the experimental data. Long computational time is needed for morphological simulation and parallel computation is used to accelerate this process. Three-dimensional model can capture the detailed flow structure around an object and predict the scour process more accurately. However, the two-dimensional model can be used as a fast assessment tool for large computational domains. For the liquefaction process, two different mechanisms (momentary and residual) are considered. For momentary liquefaction, a three-dimensional numerical model for the sea bed response under free surface water waves is constructed. Free water surface is modeled by volume of fluid (VOF) method and water waves are generated by numerical wave-maker boundary condition. An iterative numerical scheme is proposed to solve the Biot consolidation equation using a finite volume method (FVM). The coupling between water wave and sea bed is through both pressure and stress conditions on common boundaries. For residual liquefaction, the solutions to the one-dimensional model equation of the period-averaged pore pressure buildup are listed. The accumulation of pore pressure is modeled as the effect of the source term in the storage equation. Corrections to the solutions in the literature are provided. For deep soil condition, an asymptotic solution is proposed to estimate the pore pressure. A numerical model is also developed to solve the one-dimensional period-averaged pore pressure buildup equation. Good agreement between the results of numerical model and analytical model are found. These results also agree with the experiment data. A tentative step is also made to model the phase-resolved pore pressure. The basic idea of adding a source term to the governing equation is explored. The source term has the same form as that of the period-averaged residual pore pressure model. Test case shows that this model gives good results comparing to the one-dimensional period-averaged model