Critical local damage scenarios for robustness assessment of irregular structures

Abstract

Modern-day architects tend to use singular and non-repeatable shapes for building structures. Currently, scenarios for robustness assessment of such structures are defined based on engineering judgement, which can result in designs that are insufficiently robust. This thesis focuses on development of a rigorous robustness assessment framework capable of reliably determining critical scenarios for irregular building structures. To reduce the computational demand associated with analysis of detailed numerical models of building structures undergoing robustness assessment, a novel shell element capable of effectively modelling reinforced concrete slabs under extreme loading conditions is developed. A comparative study conducted in this work demonstrates the computational efficiency of the new element without compromising the accuracy of the modelled response. To address the needs of robustness assessment in the early design stage, a gradient based methodology, founded on the general concepts of topology optimisation, is developed. This methodology is capable of determining critical zones of the considered structure efficiently and to obtain a good estimate of its robustness. The developed methodology is based on an enhancement of the robustness assessment framework developed at Imperial College London, enabling a rational evaluation of robustness for sudden local damage scenarios. To facilitate the convergence of the developed procedure, a simple yet reasonably accurate approach for estimating the sensitivity of the robustness measure to the changes in damage parameters is developed and verified. To provide a reliable framework for estimating robustness at the final stage of design, an efficient global approach based on a meta-modelling technique is developed within the robustness assessment framework. This approach benefits from the developed variant of multifidelity kriging algorithm. The proposed modification utilises a set of independent low fidelity models, obtained through the subdivision of the high-fidelity model into a set of a priori established components. Employing the developed approach these models can be assembled to construct a composite low fidelity model for the response of the whole structure. The developed framework is finally applied in a detailed case study to the robustness assessment of the Agora Garden Tower. This study successfully identified all of the critical scenarios as well as their respective robustness measures. The results of the robustness assessment indicated that concepts commonly accepted for the regular structures do not necessarily hold for specific types of irregular ones, which provides strong support to the original motivation for the conducted research.