Structural Reliability Analysis Using Quadratic Polynomial Response Surface and Finite Element in MATLAB

Abstract

Many aging jacket platforms are being pushed for continued use beyond their design life due to advancement in oil extracting technology and economic reasons. Thus reassessment to determine the platform safety is vital. But no exact guide on methods to assess the safety of the aging platforms is available. Hence, development of reliability analysis methodologies is an active research area. Meanwhile, reassessment deals with numerous uncertainties especially in the load and resistance variables of a jacket platform. Response Surface methodology, a limit state approximation technique was deployed by many in many engineering fields to apprehend inherent uncertainty. Hence in this work, a reliability analysis methodology that combines simple response surface and finite element approach in MATLAB is adopted. The approach assumes a physical transfer function utilizing explicit multivariate expressions and random variables. Also, this avoids large number of finite element simulation required for any probabilistic analysis. The methodology developed allows for reliability analysis to be performed based on easy to program procedures. It also addresses the platform and environmental specific uncertainty variables to distinguish the distinctive characteristics of the cases. Upon execution of multiple finite element simulations, response of components and systems are formulated using quadratic polynomial response surface expressions, which eventually are utilized in the limit state functions. Later using numerical techniques in a separate computational routine the reliability indices are estimated. Utilizing the developed method, component and system level reliability indices are obtained. The developed methodology also has been verified with method currently available in the practice as well as with typical simulation method i.e. Monte Carlo Simulation.