Advanced Phenomenological Hysteresis Model for RC Columns Incorporating Pinching

This study develops empirical equations for the cyclic deterioration parameters of an advanced phenomenological hysteretic model implemented in OpenSees. A dataset comprising 150 experimental tests on both conforming and nonconforming RC columns was utilized to derive these equations via multiple linear regression with backward elimination. To ensure robustness, an advanced regression framework based on a machine learning approach was developed to conduct high-volume simulations, iterating over multiple functional forms for input and target variables. The resulting model explicitly incorporates unloading stiffness deterioration and pinching effects, simulating the hysteretic behaviour of RC columns more accurately than existing models.