Seismic performance of older R/C frame structures accounting for infills-induced shear failure of columns

Highlights

• Seismic performance of older infilled frames is assessed including shear failures.

• Nonlinear pushover and IDA analyses are performed at different limit states.

• Nonlinear shear in columns highly increases vulnerability of fully infilled frames.

• Columns shear failure with weak infills results in a more ductile behavior.

• Open bays location alter the global mechanism from ductile to more brittle.

Abstract

Older infilled frame R/C structures have several deficiencies as compared to seismic code-conforming structures, inherent to the behavior of the structural elements and interaction between structural and nonstructural components, which should be accounted for an accurate performance assessment. In order to investigate the importance of incorporating the contribution of the infills (typically considered as non-structural members) and their interaction with non-ductile columns in a building structural model, a comprehensive performance based earthquake assessment over a range of hazard levels is carried out through Nonlinear static and Incremental Dynamic Analyses of a 2D frame model incorporating: (1) fiber section axial force–moment interaction, (2) cracking and crushing of the infills and (3) infills-induced shear failure of columns. The analyses are performed on three different common structural configurations: (A) bare frame, (B) uniformly infilled frame, and (C) partially infilled frame (pilotis frame). Results show the importance of correctly accounting for shear failure of columns when modeling the infill–frame interaction, and the influence of the inelastic shear law, opening placement, type of infills, and choice of infill constitutive model.

Introduction

Prior to the development of seismic building codes of practice most structures were designed to resist gravity loads only. The result is that a large portion of today building stock has inadequately detailed structural members and, in regions of medium and high seismicity, these structures represent a high risk.

There are several complex mechanical aspects inherent to the building stock of older infilled frame reinforced concrete frames, such as the interaction between frame and infill panels [1], [2], the presence of unreinforced joints [3], possible bond-slip or anchorage slip near the joints [4], [5], [6], the axial–flexural interaction and flexure–shear interactions in frame components [7], [8]. These mechanisms can significantly affect the local response of buildings and, if not accounted for in an accurate manner, compromise the accuracy of the global seismic performance assessment.

Common causes for the poor performance of older infilled reinforced concrete frame structures are the presence of a soft story and captive columns [9]. However, several major earthquakes have also shown that severe damage and collapse may also occur in the case of uniformly infilled frames with shear deficient columns [10].

Collapse assessment of structures having the key deficiencies described above can be performed using simplified yet reliable modeling approaches for nonlinear analysis. The most common modeling techniques for infilled frames can be divided into two main classes: (1) Local or Micro models, and (2) Simplified or Macro-models. Micro-models, that rely on the infill discretization into 2D fine elements, can accurately capture the behavior at the local level, the cracking patterns, the ultimate load and the collapse mechanisms. However, the inherent computational effort [11], [12], makes them unpractical for practicing structural engineers, whereas Macro-models, that typically rely on the so-called bidiagonal strut models, are more often chosen as a trade-off between accuracy and ease of use. These simplified models [13], [14], [15] retain a reasonable precision in describing the salient features of the infill failure and of the infill–frame interaction, provided appropriate geometry and constitutive laws are used.

Following several years of research, infill models are now available in several structural engineering softwares that can capture different failure mechanisms, such as diagonal cracking, corner crushing, and horizontal sliding of panels [16], [17] and may extend to consider out-of-plane mechanisms [18], [19].

Critical modeling aspects for accurately simulating the interaction of the panels with the surrounding frame are: the infill type, the opening location and opening percentage [20], the infill constitutive model selection [21], [22], [15], and, most importantly, the shear failure of the concrete columns interacting with the unreinforced masonry infills. Assessing the importance of these mechanisms and parameters is a key issue for creating reliable structural models for the performance assessment of non ductile R/C frames.

Several works show why and how inelastic shear mechanisms should be modeled in RC members and in particular in poorly detailed columns, in terms of shear strength, degradation and cyclic shear rule [23], [24], [25], [26], [27], [28], [29], [30], [8]. A shear strength model and a phenomenological macro model of shear failure were proposed by Elwood [31], and more recently by Baradaran Shoraka et al. [32].

On the other hand, there are limited studies on the infill induced shear failure of columns in R/C frame structures [33], [34], [35]. Among these, D’Ayala et al. [33], proposed an approach based on detailed micro models, whose application shows good correlation with experimental results and captures the actual shear failure mechanism of the columns. However, the computational effort of this approach is still unpractical for application to entire structures.

The infill macro modeling technique proposed by Celarec and Dolsek [34], within the toolbox by Dolšek [36], is based on an iterative pushover analysis scheme intended to capture the shear failure of columns. This method effectively reproduces the global level response, however it neglects the redundancy and force redistribution that lead to the progressive collapse of the structure. Burton and Deierlein [35], study the collapse simulation considering infills–frame interaction, infills-induced shear failure in the columns and foundation flexibility effects on the overall frame response. This work assumes that the lateral thrust transferred to the column is not greater than 25% of the total infill strut force. Results from this work refer to solid unreinforced masonry walls typical of structures built in California the 1920s, whereas in other regions infill panels are made of softer hollow clay bricks.

The shear strength models and the shear behavior macro-models by Elwood [31], and Baradaran Shoraka et al. [32], have been more recently employed for collapse assessment in presence of infill–frame interaction [37] with lumped plasticity elements and strong brick infills modeled with equivalent struts.

The objective of the research presented in this paper is to carry out a comprehensive evaluation of the impact of shear failure of columns on the building performance accounting for the different types of components/mechanisms, and common structural layouts. The different nonlinear models, are incorporated into an advanced computational finite element model accounting for: (1) the axial force–bending moment column response and interaction, modeled through fiber-based sections, (2) the interaction between hollow clay infills and frame through concentric and eccentric bidiagonal struts, (3) the shear response, including shear failure and post-peak degradation in poorly detailed columns modeled with elastic and inelastic cyclic shear rules. Results from the plane frame configuration analyzed herein can be easily extended to other configurations in height.

For the purpose of this study, a trade-off between accuracy and simplicity of use in presence of masonry infills, is the model used by Marini and Spacone [8], where the shear deformations are uncoupled from flexural and axial effects at the section level, but are coupled at the element level where equilibrium between shear forces and bending moments is enforced. The model relative simplicity makes it well suited for global collapse assessment, in particular in conjunction with fiber section-based frame elements.

At the global structural assessment level, the static pushover procedure based on the N2 Method demand determination [38], [39] is a widely used nonlinear method of analysis for seismic performance assessment of buildings. The inherent limitations of the N2 method in capturing the inelastic demand due to the overall strength degradation typical of infilled frames can be overcome using the extended N2 method, Dolšek and Fajfar [40], and through the more robust Nonlinear Time History and Incremental Dynamic Analysis (IDA) procedure [41].

The evaluation is carried out with reference to three configurations, namely the Bare Frame model (BF), the Uniformly Infilled Frame model (UIF), and the Partially Infilled Frame (PIF) model. These are representative of the most typical layouts of building structures, where interior frames are usually non-infilled, and exterior frames present full or partial infilling. The bare frame model is most commonly adopted in design assumptions by practicing engineers, and comparison of results on different configurations can bring insight into the impact of realistic models of infills and shear mechanisms.

Parametric sensitivity is investigated with respect to location of openings, infill type, and uncertainty inherent to the selection of the equivalent strut constitutive model.

The performance of the different configurations is evaluated over a range of 16 hazard levels and is checked for the code-mandated reference limit states. Results on all configurations are presented in terms of IDA curves, demand statistics and fragility curves, and compared with capacity curves and pushover-based demand estimates.