Laboratory Indentation Testing Of Polycrystalline Ice An Investigation Of Fracture PDF Download

Cohesive zone model is a promising technique for simulating fracture processes in brittle ice. In this work it is applied to investigate the fracture behavior of polycrystalline cylindrical samples under uniaxial loading conditions, four-point beam bending, and L-shaped beam bending. In each case, the simulation results are compared with the corresponding experimental data that was collected by other researchers. The model is based on the implicit finite element method combined with Park-Paulino-Roesler formulation for cohesive potential and includes an adaptive time stepping scheme, which takes into account the rate of damage and failure of cohesive zones. The benefit of the implicit scheme is that it allows larger time steps than explicit integration. Material properties and model parameters are calibrated using available experimental data for freshwater ice and sea ice samples. For polycrystalline ice, granular geometry is generated and cohesive zones are inserted between grains. Simulations are performed for samples with different grain sizes, and the resulting stress-strain and damage accumulation curves are recorded. Investigation of the dependency between the grain size and fracture strength shows a strengthening effect that is consistent with experimental results. The proposed framework is also applied to simulate the dynamic fracture processes in Lshaped beams of sea ice, in which case the cohesive zones are inserted between the elements of the mesh. Evolution of the stress distribution on the surface of the beam is modeled for the duration of the loading process, showing how it changes with progressive accumulation of damage in the material, as well as the development of cracks. An analytical formula is derived for estimating the breaking force based on the dimensions of the beam and the ice strength. Experimental data obtained from the 2014-2016 tests are re-evaluated with the aid of this new analysis. The computation is implemented efficiently with GPU acceleration, allowing to handle geometries with higher resolution than would be possible otherwise. Several technical contributions are described in detail including GPU-accelerated FEM implementation, an efficient way of creation of sparse matrix structure, and comparison of different unloading/reloading relations when using an implicit integration scheme. A mechanism for collision response allows modeling the interaction of fragmented material. To evaluate the collision forces, an algorithm for computing first and second point-triangle distance derivatives was developed. The source code is made available as open-source.