This research is focused on three distinct but interconnecting topics:1) investigation of physical and mechanical properties of ice rubble, 2) material modelling of ice rubble, 3) numerical simulation of the interaction between ice rubble structure interaction. large scatter in the values of physical and mechanical properties of ice rubble can be seen in the literature due to non-standardised methods. Therefore, important tests are identified, and their major findings are presented. The shear box and punch through tests are important methods for deriving ice rubble properties, yet the interpretation of the results of the test is not straightforward. Numerical simulation of these tests offers a unique way to gain insight into rubble behaviour and to calibrate a material model for ice rubble. The accurate modelling of the ice rubble is needed when designing marine structures in ice-covered waters.

The friction coefficient between ice-ice, ice-rubber and ice-steel was investigated by measuring contact forces in controlled laboratory experiments. The experiments revealed that surface topology, speed and temperature play important roles in determining the friction coefficient on ice surfaces. Given the similarities between cohesionless ice rubble and brash ice, like discrete nature, these two ice features can be explored together. A large-scale simple shear test and a pull-up test were conducted at Luleå harbour. The data collected over two test campaigns for large scale simple shear was analysed. An attempt was made to estimate, elastic properties (bulk and shear modulus), yield strength, kinematic and volumetric hardening parameters for continuous surface cap model (CSCM) based on large-scale test data.

The calibration of material model CECM was done by curve fitting of simulation data to tests. Three tests were simulated using the finite element method (shear-box test and punch through the test) and smoothed particle hydrodynamics (punch through and pull-up test). The automated optimization algorithms were used to find the admissible combination input parameters. The assumptions of isotropy and continuity of surfaces reduced the number of input parameters. Despite the mesh distortion issue, the simulation of punch through test post-peak behaviour was simulated correctly. In the simulation shear box, expansion due to shear stress in the rubble was observed. In the simulation, the results clearly demonstrated that the evolution of the deformation patterns was related to the load records in the experiments.

A seemingly promising numerical method i.e. smoothed particle hydrodynamics (SPH) was explored in this research. The limitation posed by mesh size in finite element formulation is removed in SPH formulation, yet this formulation takes advantage of the numerically robust underlying Lagrangian formulation. The plug formation in the simulation of punch through test and pull-up test was similar to that of test observations. Good agreement in the initial, peak and post-peak part of the load-displacement relationship was achieved by calibrating the material model parameter. It can be seen clearly while by comparing simulation results, that Mohr-Coulomb criteria are not fit to simulate the post-peak behaviour of ice rubble in the tests.

The Cohesive Element Method (CEM) was used to simulate two full-scale load events recorded at Lighthouse Norströmsgrund in the Gulf of Bothnia. Two load events were identified out of full-scale measurement campaigns, “Measurements of Structure in Ice” (STRICE), to simulate ice rubble stricture interaction process with CEM. a scaling formula, based on ice rubble porosity, is explored to estimate material properties such as elastic modulus, fracture toughness and fracture energy. The hydrodynamic effects of water were added to the numerical model by the spring dash-pot system which was developed during this work for the research on ice mechanics. The frictional forces were also part of the numerical model. A relative velocity based dynamic friction coefficient was introduced. The mass damping and stiffness damping were part of the numerical model for interaction between ice rubble structure interaction but were not fully investigated.