A Substantiated Risk Assessment of a Jacket Structure

Abstract

SUMMARY We demonstrate a rational approach to structural reliability based on 'state of the art ' models. The risk of failure was calculated for a jacket structure in the Gulf of Mexico subjected to environmental loading. This was done both for one particular extreme hurricane that the structure survived and also for the long-term loading. Statistics of extreme loads were generated from a hindcast study. The fluid loading was calculated using well validated theories for the kinematics of extreme random waves and currents. Non-1inear pushover analysis was used to evaluate the ultimate strength of the structure. The likelihood of structural failure was predicted by calculating the probability that the environmental load would exceed the resistance of the structure. Physical uncertainty in the extreme environmental load and the resistance of the structure were included. These results and those from a previous analysis of a damaged structure match well with observations, corroborating the approach and its wider application. INTRODUCTION The reliability of an offshore structure is the probability that it will survive a given time span. The calculation of reliability is fraught with difficulties. One major obstacle is the ubiquitous application of conventional design practice. This has developed on the basis of historical experience (of getting it right, or sometimes wrong with catastrophic effects). There is significant modelling uncertainty, that is 'human ignorance', in conventional deterministic methods. In order to predict a 'true' failure probability, the physical uncertainties in the processes should be the major factor in the problem formulation: modelling uncertainties must be minimised such that they do not dominate the results. Rational and realistic models must be available for all the processes (such as wave kinematics and structural system behaviour) and all their interactions must be accounted for. Only then can a realistic estimate of the probability of failure be made. To date in the offshore industry, reliability analysis methods have been employed mainly in a comparative and qualitative sense. Bea [1] applied them to set environmental design criteria. Factors of safety in APTs proposed LRFD design code have been calibrated using reliability techniques [2, 3]. However, large uncertainties related to modelling errors had to be accounted for in the problem formulation. The prediction of failure rate relied on engineering judgment and calibration against survival experience (Marshall and Bea [4]). The results are often interpreted as notional and unsuitable for direct comparison with actuarial risks [5, 6]. The present work is distinguished from much of the previous work by its utilisation of realistic, highly validated models for the prediction of extreme loading and the calculation of ystem failure strength and the consequent minimisation of modelling uncertainty. In the following sections the method of evaluating structural reliability is outlined, identifying the important aspects. Essentially it builds upon a framework similar to that proposed by Bea [1] for calculating structural reliability, and applied for notional risk calculation by Anderson, Silbert and Lloyd [6].