Collapse reliability of high-rise reinforced concrete systems subject to fully non-straight/-stationary hurricane representations

Abstract

Performance-based wind engineering (PBWE) is becoming rapidly recognized as a necessary shift from common prescriptive wind design if the performance of wind-excited structures subject to a full range of wind intensities, including those that produce inelasticity and potential collapse, is to be fully understood and incorporated into design. Central to PBWE are advanced methodologies capable of not only efficiently propagating uncertainty but also explicitly assessing the inelastic response of the system. Notwithstanding the significant efforts dedicated to the efficient estimation of structural performance within PBWE settings, there is still a lack of frameworks that enable the modeling of the inelastic/collapse of structural systems to subject extreme wind hazards, especially with respect to long-duration, order of several hours, hurricanes. In response to this need, a reliability-based computational framework is proposed for efficiently investigating the inelastic/collapse behavior of the high-rise reinforced concrete (RC) systems under fully non-stationary/-straight hurricane events. The framework is based on integrating a high-fidelity fiber-based finite element modeling environment with a recently introduced non-straight/-stationary stochastic wind load model that is cable of fully capturing the continuously varying wind speed and direction that occur during the passage of a typical hurricane. To enable treatment of general uncertainty (i.e., uncertainties in the mechanical properties of the system, gravity loads, models themselves (epistemic uncertainties), and system geometry) during the estimation of probabilities associated with rare events (e.g., collapse), the finite element modeling environment and non-straight/-stationary hurricane model are embedded in a recently introduced stratified stochastic simulation scheme. To minimize the computational cost associated with running samples involving high-fidelity nonlinear structural models that are subject to long-duration stochastic loading a truncated non-straight/-stationary hurricane model is proposed. The computational framework is demonstrated on a 45-story RC archetype building located in New York, NY, for which the probabilities of failure associated with a full range of limit states, including collapse, are estimated that meet target coefficients of variation. The reliability of the system is subsequently investigated with the aim of shedding light on the ultimate performance of structures built in accordance with current practices. Failure mechanisms and reliabilities are evaluated for essential structural behavior, including rebar yielding, fracture, low-cycle fatigue as well as concrete crushing. This work contributes to the development of guidelines and standards for the next generation of PBWE.