Numerical analysis on random wave-induced porous seabed response

Research on the response of random wave on offshore structures has received great deal of attention of many researchers and engineers in the design of marine structures. Most previous investigations have been limited to the regular waves. In this paper, based on Longuet–Higgins random wave theory and finite element method, a numerical model for random wave-induced seabed response is established. The seabed is treated as poroelastic medium and characterized by Biot’s partly dynamic equations (u–p model). The JONSWAP spectrum is adopted in Longuet–Higgins model, which is based on the cumulative superposition of linear diffraction solution. Based on the numerical results, the effects of random wave on seabed response are investigated by comparing with the corresponding Stokes wave and cnoidal wave. Then, a parametric study is conducted to examine the effect of wave and soil characteristic on the seabed.     

Reliability analysis of submarine slope considering the spatial variability of the sediment strength using random fields

The properties of marine sediments vary spatially, and the undrained shear strength of marine clay increases linearly with depth because of depositional processes and the effective overburden pressure. To evaluate the stability of submarine slope considering the spatial variability of soil strength, the random field discretized by the Karhunen-Loève expansion is combined with the limit equilibrium method to conduct reliability analysis. For simplicity, our physical model does not include many complexities such as the effects of excess pore water pressure on the stability of submarine slopes. Stability estimates of the infinite slope model, under both static and seismic loading, are made with three types of one-dimensional stationary or non-stationary random fields. The two-dimensional slope model is also analyzed, where the shear strength varies with the positions of the strips because of the discrete random-field function for the soil material. In submarine slope reliability analysis, the non-stationary random field of the linearly increasing soil strength is used, instead of the commonly used stationary one. To obtain the failure probability through Monte Carlo simulations, a novel response surface method based on Gaussian process regression is introduced to build the surrogate model. The computational efficiency is significantly increased, because there is a considerable reduction of calls of the deterministic analysis. Therefore, the proposed method makes the prediction of submarine landslides which are usually rare events with very small probabilities more efficient.