Stochastic Modeling of Geometric Imperfections in Aboveground Storage Tanks for Probabilistic Buckling Capacity Estimation

Abstract

This study develops a methodology to model random geometric imperfections in welded cylindrical aboveground storage tanks (ASTs). Precise modeling of the imperfections in ASTs is important for quantitative assessment of buckling strength of ASTs under various external loads, such as wind and storm surge, because buckling phenomenon is highly sensitive to stochastic imperfections. Even though buckling of ASTs due to extreme winds during hurricanes has been reported as a common failure mode, existing literature lacks models for random geometric imperfections to facilitate probabilistic buckling capacity estimation of ASTs under such external loadings. Current studies on buckling behavior of ASTs use deterministic imperfection models, neglecting the inherent uncertainties present in the imperfections and the response of ASTs. Usually, studies use eigen or buckling mode shapes to depict imperfections which are most detrimental to the buckling strength of ASTs; furthermore, the amplitude of these imperfections is also random but is often deterministically chosen as a multiple of the shell thickness. Such imperfection modeling methods may lead to conservative buckling load estimates and are unable to facilitate probabilistic capacity estimation, which is important for reliability assessment or future calibration of existing design guidelines per the load resistance factor design (LRFD) philosophy. Therefore, this study develops a stochastic model for global imperfections in welded cylindrical ASTs using a two-dimensional Fourier series expression. The coefficients of the Fourier series are modeled as random variables, which are based on imperfection data measured from similar structures in Germany and Australia. The proposed imperfection modeling scheme is applied to different sample tanks to assess the influence of imperfections on the wind buckling capacity. The results show presence of significant uncertainty in buckling capacity. Furthermore, useful insights are gained on the effects of imperfection magnitude on the mean and variance of the capacity. In the future, observations from this study can be used to probabilistically determine the knockdown factors for buckling design of ASTs or to support regional risk assessment of existing portfolios of ASTs.