Joint shear strength models for exterior RC beam-column connections exposed to biaxial and uniaxial cyclic loading

Joint shear strength models available in the literature can predict the shear strength for reinforced concrete (RC) beam-column connections exposed to uniaxial cyclic loading, while an accurate biaxial joint shear strength model is still lacking. Gene expression programming (GEP) is used in this research to develop uniaxial and biaxial joint shear strength models for exterior RC beam-to-column connections exposed to uniaxial and biaxial cyclic loading respectively. The GEP models are developed based on an experimental database available in the literature, where the models are randomly trained and tested. Uniaxial joint shear strength is also predicted using the ACI 352 and ASCE 41 formulations for connections exposed to uniaxial cyclic loading. The performance of the GEP models is statistically evaluated using the coefficient of determination R-squared. The R-squared values are 79%, 79%, 95% and 93% for the ACI, ASCE, uniaxial GEP and biaxial GEP models, respectively. The R-squared values of the GEP models are high, which confirms their accuracy and indicates that they are more fitting to the experimental results than the ACI and ASCE formulations.