Performance based formal optimized seismic design of steel moment resisting frames. (15th July 2020)

• Record Type: Journal Article
• Authors: Idels, O.; Lavan, O.
• Is Part Of: Computers & structures. Volume 235(2020)
• Publication Date: 2020-07-15
• Subjects: Non-linear time history analysis -- Seismic design -- Irregular structures -- Discrete material optimization -- Structural optimization -- Moment resisting frame; Structural engineering -- Data processing -- Periodicals; Electronic data processing -- Structures, Theory of -- Periodicals; 624.171
• DOI: 10.1016/j.compstruc.2020.106269
• Languages: English
• ISSNs: 0045-7949

Abstract

Highlights: An efficient methodology for the seismic optimization of steel MRFs is presented. Non-linear time history analysis is utilized to evaluate structural responses. This is done, for the first time, using an efficient gradient-based approach. The adjoint method is employed to derive the sensitivities in an efficient manner. The examples reveal the efficiency of the proposed approach. Abstract: In this paper an efficient optimized seismic design of steel moment resisting frames is presented. Non-linear time history analysis is utilized to evaluate structural responses under seismic excitations. The goal is to optimize the cross-section properties of all elements comprising the frame, in order to achieve the minimal structural steel volume as a representation of the cost. Performance based design constraints are formulated to limit the inter-story drifts for desired levels. As efficiency is of concern, a gradient based optimization is adopted. The design variables are formulated as continuous and along with the discrete material optimization (DMO) function, standard steel table sections can be attained. The adjoint analytical method is employed to derive the sensitivities in an efficient manner. Numerical examples demonstrate the effectiveness of the methodology. Even in cases of irregular, real-scale, moment resisting frames – optimized desired discrete designs have been achieved in a reasonable computational effort.