A unified degrading shear strength model for rectangular and circular solid/hollow RC columns

Reliable quantification of shear strength in the plastic hinge regions of reinforced concrete (RC) columns is crucial for seismic design. However, existing shear strength models often rely on the linear superposition of multiple shear-resisting contributions, an approach that has limitations in representing the evolution of stirrup stresses, the contribution of longitudinal reinforcement, and the effects of cross-sectional shape. This study compiled experimental data from 258 rectangular and circular solid/hollow RC columns that failed in shear or flexure–shear modes and evaluated the applicability of 12 existing shear strength models. Based on the smeared-crack concept and Mohr’s circle, a mechanics-informed unified shear strength model was developed to represent flexure–shear damage evolution while accounting for the effects of concrete, transverse reinforcement, axial force, longitudinal reinforcement, shear-span ratio, and displacement ductility. The model was calibrated using the compiled database and independently validated for the column categories with sufficient data. The results show that the existing models exhibit considerable variability in prediction accuracy across different column types. Among them, the Sezen model provides the best overall balance between accuracy and scatter, with an overall mean predicted-to-experimental strength ratio of 1.009 and a coefficient of variation (CV) of 0.230, although it slightly underestimates the shear strength of circular solid columns. The proposed model provides a common formulation for rectangular and circular solid/hollow RC columns and achieves improved prediction accuracy and lower scatter, with an overall mean predicted-to-experimental strength ratio of 0.999 and a CV of 0.171. When a shear strength reduction factor of 0.8 is applied, the proposed model achieves an overall empirical safety rate of 91.1%, while the statistically evaluated subgroups exhibit empirical safety rates ranging from 87.1% to 95.4%, with most values close to or above 90%. These results indicate a favorable balance among prediction accuracy, scatter, and safety performance.