This MSc thesis was initiated as part of an interdisciplinary project researching post-disaster coastal reconstruction in the Tohoku region of Japan. On March 11, 2011, a highly destructive earthquake induced tsunami hit the north-eastern part of Honshu island, causing enormous damage to numerous small towns along the coast. Since 2011 various studies have been completed to better understand this historical event and explain the nature of tsunamis. However, some researchers have noted the inability of tsunami models, using only an earthquake as a source, to reproduce the timing, frequency and wave heights recorded along the Tohoku coast in March 2011. The differences are especially striking for the Sanriku coast – a relatively small area north of Sendai plain characterised by a ria-type coastline. D. R. Tappin et al. (2014) suggested that a secondary source is required to explain the deviating model results and locally higher tsunami waves, such as a submarine landslide. In low-lying coastal areas, tsunamis are an unavoidable hazard to human lives. A common way of addressing natural disasters of this scale is with timely warning and evacuation, which in turn requires appropriate techniques for prediction of an upcoming disaster. While earthquake induced tsunamis provide a sort of a warning signal, tsunamis generated by a submarine landslide can be much more difficult to predict. Landslide tsunamis are usually concentrated in local areas and do not propagate far, as they are more dispersive than earthquake tsunamis. However, the source of landslide tsunamis is often located on a continental slope near the coastal line. As a result, waves reach the shore in a very short period of time and run-up heights can reach extremely high levels (Harbitz et al., 2006). Thus, landslide generated tsunamis pose a serious hazard to densely populated coastal areas. Within the scope of this thesis, a number of experimental simulations on submarine slope failures were conducted in a so-called static liquefaction tank – a unique testing facility for large scale experiments at the Geo-Engineering Laboratory of TU Delft. To improve the understanding of tsunami generation, this study focuses on the moment of onset of a submarine slope failure and wave generation, linking the processes within failing soil mass to initial characteristics of waves. This study illustrates the connection between these two processes that are usually only considered individually from either a geotechnical or hydraulic perspective.

In this report, an experimental investigation of the influence of currents on submarine slope stability is presented for the Eastern Scheldt storm surge barrier case. Slope instabilities, including liquefaction failures, have been observed in the scour hole slopes near this barrier. The currents above the sloping bed are expected to initiate the liquefaction slope failures. The effects of the excess pore water pressure increase, that is generated by the currents, on the soil response are investigated by triaxial tests. Multiple excess pore water pressure rates are applied on a loose soil sample. The triaxial tests show that higher excess pore water pressure rates result in earlier developments of strains at lower stress ratios. This research is intended as a step towards a better understanding of the initiation mechanism of the liquefaction slope failures near the barrier.

The static liquefaction tank (SL-tank) is a unique large-scale facility that provides a means to simulate the conditions of a submerged soil slope under failure conditions. Static liquefaction (SL) may be defined as a significant loss of soil’s shear strength due to a substantial development of excess pore water pressure under monotonic loads. This complex soil behavior is commonly related to case studies such as submarine flow slides.
The main objective of this project is to use a finite element method (FEM) based on a Hypoplasticity (HP) framework to numerically simulate the experimental data obtained from the Static liquefaction tank facility of Delft University of Technology. A set of Elasto-Plastic (EP) constitutive models were chosen as an initial reference to simulate the SL-tank and afterwards compare their results to the HP framework. The numerical results were analyzed mainly in terms of excess pore water pressure and relating them to experimental data. Additionally, effective stresses, strains and stress paths were examined from the given model outputs.
A fine-cohesion less soil called Geba sand was used in the experimental procedure of the SL-tank, as well as in elements tests performed within this study. The model parameter determination and calibrations were performed by means of element tests, empirical correlations, theoretical formulations, and best-fits from experimental data. Soil behavior at low stresses is of fundamental importance for the performed experiments and numerical simulations in this work.
A potential instability behavior from the given numerical simulations was studied by means of an adopted instability line (IL). The IL criteria is a framework which is commonly illustrated in stress paths as a boundary that delimits a potential susceptibility to soil collapse. Element test data from fine-loose sands, as well as the numerical outputs from this work were used for estimating a potential IL applicable for the scope of this Master’s thesis project.
Results of this investigation showed clear limitations of the hypoplasticity constitutive laws in generating sufficient excess pore water pressures and deformations to trigger static liquefaction. Additionally, boundary effects in the assumed fixity conditions were a main potential issue regarding inaccurate results. Nevertheless, an enhanced model response (HP) was observed in comparison to Elasto-Plastic models.