Sea ice ridge keel punch through experiments: model experiments and numerical modeling with discrete and combined finite-discrete element methods

Simulations and laboratory-scale experiments were performed to study sea ice ridge keel punch through experiments and ice rubble behavior in them. The punch through experiments are an important method for deriving ice rubble properties, yet the interpretation of the experiment results is far from straightforward. Anyhow, accurate interpretation is important for the material modeling of ice rubble, and further for an accurate modeling of the ice rubble related problems needed when designing marine structures in ice covered waters. The experimental work was performed using rubble consisting of plastic blocks. Plastic blocks made it possible to study of experimental method itself, because the interpretation of the experiment results was simplified from that of the experiments with ice rubble. The effect of indentor velocity was experimentally studied, as earlier laboratory-scale punch through experiments have suggested that the ice rubble sheer strength depends on the loading rate. The loading rate dependency of the rubble shear strength was found to be likely related to the experimental set-up rather than to the rubble material. Simulations of punch through experiments were performed using a three dimensional discrete element method and a two dimensional combined finite discrete element method. In both of these methods, rubble is modelled as a discontinuum. The methods were developed during this work for the research on ice mechanics. Simulations of punch through experiments on non-cohesive rubble were performed using the three dimensional discrete numerical model. A technique for modeling freeze bonds was developed within the framework of two dimensional combined finite discrete element method simulations. The simulations helped to provide insight on the analysis of the punch through experiment results. The simulation results clearly showed that the evolution of the deformation patterns was related to the load records in the experiments. In the case of partly consolidated rubble, the initial failure patterns of the rubble were observed to be related to the measured maximum force values. Furthermore, the behavior that has been earlier understood to be the result of rubble material softening was in fact shown to be due to changes in the rubble geometry during the experiment. The discontinuous modeling of rubble showed that physical phenomena could potentially be rendered out from the more traditional continuum models. These phenomena included, for example, the importance of tensile freeze failures in the failure process of partly consolidated rubble.