Predictive numerical model for simulation of fiber-reinforced concrete beams in bending and inverse parametric analysis

Abstract

Determining material parameters is essential for understanding the physical properties of fiber-reinforced concrete (FRC), enhancing the design process, and optimizing costs. This paper presents a deterministic numerical model of FRC beams under three-point bending tests (TPBT), with an emphasis on the failure analysis. This model aims to replicate TPBT and address the challenge that comes with direct measurements of FRC parameters, especially in the context of failure mechanisms. It integrates analytical solutions and explicit parameters to describe the bending and failure behavior, while mitigating computational demands in parameter identification due to the complex heterogeneous nature of the material. Laboratory tests were conducted to record key parameters and served as a foundation for the model development and validation of its predictive capabilities. The generated simulated data from the predictive numerical model were then utilized as input for an inverse analysis based on the Levenberg–Marquardt method. The inverse analysis demonstrated high accuracy, with parameter convergence typically within 3 to 7 iterations and errors reduced to less than 2%. Model verification was performed by comparing the results obtained from the inverse analysis with the data obtained from laboratory tests, confirming the model’s effectiveness in accurately predicting FRC failure behavior.