Design and evaluation of SMRF buildings - from mechanic-based to data-driven models (Passed)

Abstract

While the second-generation performance-based seismic design (PBSD) enables the structural engineers to target specific stakeholder-driven building performance objectives, it has not been widely adopted in practice. This is primarily due to the labor-intensiveness and high computational expense associated with performing iterative nonlinear response history analyses (NRHAs) and damage, loss, and downtime assessments. To address this challenge, a performance-based analytics-driven (PBAD) design framework is proposed as an alternative to the traditional PBSD methodology. Central to the PBAD design framework is the use of surrogate models to guide preliminary design iterations. The surrogate models, which serves as a statistical link between key design variables and seismic performance outcomes, can be used to target the optimal design region. Once the target region has been established, the design can be assessed, revised, and finalized using the mechanics-based approach that is used in the traditional design process. The expectation is that the PBAD design methodology will minimize the number of iterations and computational expense required to targetspecific stakeholder-driven performance objectives. To support the development and implementation of the PBAD design framework, an end-to-end Python-based computational platform, which could automate the process of conducting seismic design, constructing nonlinear structural models, and performing seismic response (both static and dynamic) simulation of steel moment resisting frames (SMRFs), is introduced. Using this platform, a comprehensive database that includes 621 SMRFs designed in accordance with modern building codes and standards, the corresponding nonlinear structural models, and their seismic responses under 240 ground motions. Based on this database, a set of surrogate models that are established using mechanics-based, hybrid, and purely data-driven approaches are developed and comparatively assessed.