Computational Frameworks for Probabilistic Performance-Based Wind Assessment of Envelope Systems of Engineered Buildings

Abstract

The adoption of performance-based wind engineering (PBWE) is rapidly becoming recognized as a fundamental step to reducing the economic losses induced by severe windstorms. A number of PBWE frameworks for the assessment of engineered building systems such as high-rise structures have been introduced. Although these frameworks have resulted in significant progress toward the efficient and effective estimation of performance within a PBWE setting, there is still a significant lack of frameworks that can holistically model the performance of the envelope systems of engineered buildings, notwithstanding how these systems are often critical to the overall performance of engineered buildings. The lack of research on this topic is mainly due to the high complexity of modeling the interdependent physical processes leading to damage. These processes include the 3D turbulent wind flow, wind-driven-rain and rainwater runoff, flying debris, internal/external wind pressures, structural dynamic responses, story drift/net pressure-induced envelope damages, debris impact-induced damages, and water ingress. As analytical solutions do not exist for this type of problem, efficient computational frameworks must be developed. To address this situation, this research presents a performance-based wind engineering framework that integrates system-level structure and envelope performance assessment. In particular, the external surface pressure is generated through a wind tunnel-enabled proper orthogonal decomposition (POD)-based non-Gaussian simulation framework while the stochastic internal pressures at envelope openings are modeled through a nonlinear coupled system of equations derived through the application of the unsteady-isentropic form of the Bernoulli equation and the principle of mass conservation. Linear modal analysis or high-fidelity nonlinear finite element methods (FEM) are used in modeling the dynamic structural responses as well as any subsequent damage. To predict the envelope component damage, suites of coupled fragility functions are derived to account for the effect of multiple demands acting simultaneously. As a concurrent hazard event, wind-driven rain is modeled in a separate computational fluid dynamics (CFD) domain where the mean wind flow is solved through a 3D Reynolds-averaged Navier-Stokes (RANS) equation with realizable emph{k}-epsilon model. Based on this mean wind flow, the wind-driven rain is solved through an Eulerian-multiphase model with turbulent dispersion considered. For efficiently estimating probabilistic performance metrics in terms of the total repair cost and amount of water ingress, the framework is further integrated into a conditional stochastic simulation framework where the uncertainties in the hazard inputs, such as wind speed, stochasticity of the aerodynamic loads, wind direction, and rainfall intensity are propagated to the system responses. Finally, the effect of the hazard duration on the performance metrics is investigated. In this respect, in place of the classic assumption of a nominal wind and rain hazard of 1-hour duration, a synthetic tropical cyclone approach is adopted in which the entire duration of the hurricane is simulated resulting in time-varying inputs of wind speed, wind direction, and rainfall intensity. A comprehensive comparison, in terms of a full range of probabilistic performance metrics, is carried out illustrating the limitations of current practice.