Internal blast loading of submerged floating tunnels in concrete

Master thesis in Civil and Environmental Engineering. NTNT - Norwegian University of Science and Technology. Department of Structural Engineering.A submerged floating tunnel (SFT) has been proposed for crossing the fjords which today are operated by ferries along the Norwegian highway E39. One concern with a potential SFT, which probably would be built using reinforced concrete, is if an explosion is to go off inside it, either accidentally or intentionally. Since fullscale experimental testing is out of the question, one must resort to numerical analyses and component tests. Uniaxial compression tests were performed for concrete cubes and digital image correlation (DIC) analyses of the tests provided satisfactory results. The tests were then simulated using both the concrete damaged plasticity (CDP) model in ABAQUS and the Karagozian & Case (K&C) model in LS-DYNA and both were found to provide adequate results. The CDP parameters were obtained by scaling previous results and the model displayed pathological mesh dependency. The K&C model proved simpler to use as the only necessary input was the concrete strength. However, for LS-DYNA it was found that an unnaturally low friction coefficient was needed and that the model displayed unphysical post-peak behavior. Concrete pipes were subjected to blast loads by using C4 charges. The charge placement clearly affected the failure and the effects of confinement and scaled distance were evident. Increasing the wall thickness and adding reinforcement proved to be effective design measures with regard to blasts. For both ABAQUS and LS-DYNA, Lagrangian analyses overpredicted the damage, despite underestimating the blast. Eulerian analyses of the blast were performed in ABAQUS but underestimated the pressure. Lastly, coupled Eulerian-Lagrangian (CEL) analyses were performed in ABAQUS, but the analyses eventually stagnated, there was pressure leakage, the pressure-time curves fluctuated excessively, the pressure was underestimated, and the damage of the pipe was overpredicted. For both the Eulerian and CEL analyses deciding on a proper time step scaling factor proved challenging. Furthermore, methods of incorporating stochastic behavior for concrete were investigated for simulating both the compression and blast tests. For the pipes, these methods reduced the amount of damage.Statens vegvesen Region øs