Artículo de revista
Modeling of reinforcement global buckling in RC elements
Fecha
2014Registro en:
Engineering Structures 59 (2014) 484–494
dx.doi.org/10.1016/j.engstruct.2013.11.015
Autor
Massone, Leonardo M.
López, Eduardo E.
Institución
Resumen
Buckling of longitudinal reinforcement in reinforced concrete (RC) columns or walls is commonly seen
with a length equal to the spacing between stirrups (local buckling), but experimental observations have
shown that the length of buckling can span a larger length, deforming several stirrups within the buckling
length (global buckling).
The behavior of the longitudinal reinforcement under compression resulting in global buckling is studied
in this work, based on a concentrated plasticity fiber model that considers four (4) plastic hinges. The
model was originally validated for local buckling and here is extended to global buckling by introducing
the effect of transversal reinforcement and expansion of the core concrete in the analysis. Modeling of the
forces from the stirrups acting on the longitudinal bar assumes that part of the force is transferred
directly to the expanding concrete core and the remaining force is balanced by internal stresses in the
longitudinal bar.
The bar buckling behavior is evaluated for different buckling length values and the length is chosen
such that it delivers the lowest maximum stress. The proposed model is validated by comparison of
the predicted buckling mode with experimental test results from the literature. The average error in
the mode prediction is 0.59 (about half the space between stirrups), which is a reasonably good value
considering that the database of tests used covers buckling modes from 1 to 7 (stirrup spacing). Besides of
providing good critical buckling length predictions, it allows obtaining the stress versus strain curve for
the overall buckling bar.
Analysis of three column specimens from the literature indicates that the overall stress versus strain
response can be obtained with a reasonable accuracy. Peak stress is obtained within an error of about
10% compared to the test result for a strain well represented by the model. The post-peak slope also gives
a good estimate for the degradation stage.