Quantitative assessment of nonlinear macro-models for global behavior and design of planar RC walls

Abstract

Performance-Based Seismic Design (PBSD) procedures require adequate estimation of the global cyclic force– displacement (F-D) behaviors of the primary lateral load resisting elements. As reinforced concrete (RC) walls constitute one of the most widely used lateral load elements in building construction, several nonlinear analytical modeling approaches have been developed over the past several decades to numerically simulate the hysteretic behavior of these structures. However, comparisons between these models have often provided few details on the model input parameters, have not used consistent model discretization, and have assessed hysteretic predictions more qualitatively than quantitatively. Accordingly, the objectives of this paper are to: 1) provide detailed information on the modeling parameters used to simulate the cyclic lateral F-D behavior of a set of planar RC walls with flexure-dominant as well as shear-dominant behaviors, 2) evaluate the sensitivity of the F-D curves to different wall model discretizations applied consistently between different models, and 3) propose and use a new quantitative approach to assess the hysteretic behaviors predicted by the numerical models. The considered models are PERFORM 3D, MVLEM, SFI-MVLEM, and BTM. To represent the information that is typically known during the design of a structure, the nonlinear modeling parameters were limited to the wall geometry, reinforcing layout, concrete compression strength, and reinforcing steel yield strength, with no calibration of the material behaviors to the reported test results. A total of 2,300 analyses were performed tosimulate four slender and four squat previously-tested RC walls. Results of the cyclic lateral F-D predictions were evaluated based on commonly used measures of effective stiffness, maximum strength, and ultimate displacement. Additionally, the hysteretic response of the numerical simulations was assessed by proposing, validating, and using the Modified Nash-Sutcliffe Efficiency (NSEm) metric. Comparisons between the computing times for the models were also made, as well as comparisons of the local nonlinear behavior of a slender wall. Results show that the discretization used in the PERFORM 3D, MVLEM, and SFI-MVLEM simulations did not significantly affect the predicted global response, whereas the discretization did affect the F-D curves from the BTM model. The values of NSEm agreed with qualitative evaluations of the F-D curves and were able to properly evaluate complex cyclic behaviors of RC walls including pinching and stiffness deterioration. The results of this investigation are expected to be used as modeling guidelines of RC walls when conducting PBSD.