Validation of a numerical model for prediction of out-of-p instability in ductile structural walls under concentric in-plane cyclic loading
Instability failure (also referred to as out-of-plane instability) has been observed in several experimental studies conducted on seismic performance of rectangular structural walls under in-plane loading. Observation of this failure pattern in some well-confined modern walls during the 2010 Chile and the 2011 Christchurch earthquakes has raised concerns about the reliability of current design code provisions. In this study, a numerical model composed of nonlinear shell-type finite elements was proposed and validated for seismic performance prediction and simulation of out-of-plane instability failure in rectangular walls. The plane sections are not enforced to remain plane in the planar direction in this type of model, and the in-plane axial-flexure-shear interaction can be simulated without requiring any empirical adjustment. The element used in the model (curved shell element) had integration points along the thickness unlike flat shell elements in which integration is performed in one plane only. This element was consequently able to capture the variation of strain along the thickness and simulate the deformation in the out-of-plane direction. Experimental results of cantilever wall specimens which failed in out-of-plane mode were used for verification of the adopted modeling and analysis approach. The numerical model was found to be able to predict the trend of initiation, increase, and recovery of out-of-plane deformation as well as the formation of out-of-plane instability that was observed during the tests. Development of this failure mechanism in the numerical model has been scrutinized using detailed response of the reinforcement and concrete elements positioned along the thickness of one of the specimens and at different stages of the failure mode. Also, the dependency of initiation and amount of the out-of-plane deformation on the maximum tensile strain developed in the longitudinal reinforcement during a specific loading cycle, and consequently all the parameters that influence its value (e.g., axial load, wall length, and cyclic loading protocol), as well as wall thickness has been confirmed by a set of parametric studies conducted on the models developed for one of the wall specimens.