Nonlinear model updating algorithm for biaxial reinforced concrete constitutive models of shear walls

A novel biaxial constitutive model and a nonlinear model updating algorithm are proposed and implemented in ABAQUS software for simulating the reinforced concrete (RC) shear walls. First, the proposed model is established using the User Material (UMAT) subroutine to be compatible with the multi-layer shell elements and membrane elements with ABAQUS implicit solver. The modeling scheme, the biaxial and uniaxial constitutive models of concrete, and the uniaxial models of rebars are elaborately illustrated. Second, this research modifies five critical parameters in the developed model to consider the influence of initial cracking and softening of concrete and slip of longitudinal rebars. The modified parameters include the concrete stiffness reduction factor alpha(c) , the concrete compressive softening factor eta(c), the concrete shear softening strain gamma(u) , the steel stiffness reduction factor at initial tensile loading alpha(s) , and the steel stiffness reduction factor at peak load alpha(u) . Third, an optimization algorithm based on the discrete Frechet distance is proposed to quantify and minimize the difference between test results and finite element (FE) simulation results of load-displacement curves. Subsequently, a total of 24 RC shear walls are obtained from the literature, and the optimization method is adopted to obtain the most favorable material constitutive parameters based on the force versus lateral displacement curves. The comparison shows the proposed model with the optimization method predicts high accuracy simulation results for RC shear walls. Finally, five neural networks are trained to predict the material constitutive parameters using the 24 RC shear wall tests, and the accuracy of the neural networks is deemed satisfactory.