| Summary: | The Arctic is a sensitive environment and requires high standards for all operations. One of the main factors in developing safe Arctic operations is a reliable prediction of sea ice loads on structures. This thesis focuses on an ice failure process and on mechanisms and limits for the peak ice load values. The study relies on two-dimensional combined finite-discrete element method simulations of the failure process of level ice against an inclined marine structure. The complexity of the deterministic failure process forces to use statistical methods in the analysis of the output data. The used computer model is sensitive to initial conditions which allowed to create well-controlled ice load data sets. The work describes the evolution of the ice failure process by deriving the mean, the standard deviation, and the maximum loads from concurrent load outputs of different simulations within each data set. All these loads increased throughout the ice failure process, and the process did not reach a stationary stage. An extreme value analysis of the ice load data showed that ice loads could be limited by the ice crushing even if the structure has an inclined wall. Besides the failure process statistics, the work also examined peak ice loads. The Gumbel extreme value distribution fitted well on global peak load observations. The study also showed that scatter in the output data stem from the complex ice-structure interaction process itself. Due to high scatter in peak load observations, this type of analyses require large samples for observing small parameter effects which are typically caused by other parameters than the ice thickness. Despite challenges in regard to the high scatter in the process, this work presents a simple buckling model that normalized the peak load observations well. The buckling model connects the peak ice load observations to so-called force chains which have been argued to have an important role in the ice-structure interaction process. Arktisen alueen luonnonvarojen, kuten öljyn ja ...
|
|---|