STOCHASTIC RESPONSE ANALYSIS OF OFFSHORE STRUCTURES

The first part develops a general method for estimating the system reliability of offshore structures using the full distribution method and introduces a probabilistic definition of structural redundancy. Structures are treated as systems of structural components. The probability of structural failure is then evaluated by examining a limited number of significant sequences of member failures that produce structural collapse. The structure examined is an indeterminate deep offshore truss under fully developed sea conditions. This reliability analysis and redundancy definition will form an important basis for design and integrity assessment of offshore structures.; In the second part, a dynamic response analysis of the same offshore tower is performed in the frequency domain and the equations of motion are solved using equivalent linearization techniques. Two methods of reliability analysis are considered: (a) a statistical fatigue damage analysis using an industry model to demonstrate the influence of the design stress level on the fatigue reliability and (b) a first-passage failure probability with periodic inspections. A crack growth analysis is performed using fracture mechanics to estimate propagated crack size and corresponding residual strength under random service loading. It is shown that there is always a limit in the number of inspections beyond which no significant improvement can be achieved.; The third part presents a method of system identification of nonlinear multi-degree-of-freedom structural systems. To identify the structural coefficient matrices appearing in the equations of motion of an offshore structure, the original nonlinear system is approximated by an equivalent linear system. Estimation of the structural parameters of this linear system is made by Kalman filtering techniques based on excitation and response time histories of the actual nonlinear system. The results obtained by this equivalent linearization technique have been compared with those derived from the time domain analysis of the original nonlinear system and the corresponding results of another equivalent linear system developed by conventional methods used in the second part. The analysis reveals that the proposed linear system offers an efficient alternative to conventional linearization methods and produces reliable values of the structural parameters.