Structural Reliability Analysis of Intact and Damaged Ships in A Collision Risk Analysis Perspective
MetadataVis full innførsel
- Institutt for marin teknikk 
Risk-based design is growing in importance due to the need of obtaining measures of the safety to aid making decisions regarding the optimal balance between safety and economy. Design against Accidental Limit State (ALS) is one aspect of risk-based design. Collision is one of the most probable types of accidental event. The aim of this thesis is to contribute to probabilistic assessment of hull girder failure in collision risk analysis. The procedures established and applied in this thesis may be applied in a wider context, e.g. in grounding risk analysis. The probability of collision-induced hull girder failure is dealt with by defining collision as a two-phase event, i.e. collision contact phase (Phase I) and post-collision phase (Phase II), for the struck vessel. Collision-induced hull girder loads are necessary to predict collision-induced hull girder failure in Phase I. Simplified external mechanics of collision are extended to consider collision-induced hull girder loads. Main challenge of this effort is to relate the radiation forces on part of the vessel to those on the whole vessel which is represented in terms of equivalent added mass. A simple method is proposed to solve it. The hull girder responses are also estimated by considering the ship to be flexible. A conventional modal superposition method as well as an alternative method is used. The focus of this thesis is the probability of collision-induced hull girder failure of struck vessels in Phase II, i.e. the probability of hull girder failure of vessels with collision-induced damages by using structural reliability methods. Relevant design collision scenarios are selected based on available information. Previous work on structural reliability of damaged vessels has mainly focused on scalarload process with extreme hull girder loads represented by those on intact vessels multiplied by load coefficients. This approach does not reflect the damage properties properly, which can be overcome by scenario-based assessment. Furthermore, vectorload processes should be considered in structural reliability analysis of damaged ships. First passage probability based on out-crossing rate, which is called the out-crossing rate method, can be used to perform reliability analysis of ships under vector-load processes. In this thesis, alternative methods are introduced to estimate out-crossing rate and they are compared and validated mutually for structural reliability analysis of a damaged oil tanker. Scenario-based reliability analyses are undertaken for a double hull oil tanker at full load condition with various collision-induced damages. Combined vertical and horizontal bending moments are considered and therefore the out-crossing rate method is used to predict the probability of hull girder failure. Collision scenarios are defined in terms of sea state, loading condition and damage properties etc. Ultimate strength of the damaged vessel is estimated by the Smith's method and is thus affected by damage height and vertical location. Damage length, penetration and longitudinal location determine which tanks are damaged and thus affect hull girder loads of damaged vessels. The sensitivity of the 3-hour's failure probability of damaged vessels to sea state is investigated. The sea state associated with collisions from the HARDER project and that for North Atlantic from DNV are considered. The sensitivity of the short-term failure probability of damaged vessels to damage is also studied. The statistical database for collision-induced damage in the HARDER project is adopted to assess the probability of damages. Finally, as a particular issue for damaged vessels, the effect of sloshing in tanks on motions, hull girder responses and thus the probability of hull girder failure is investigated. The linear multimodal approach is implemented to study sloshing effects based on the assumption that the damaged tanks are of rectangular shape.