Development of Rapid Solutions for Prediction of Critical Continuously Reinforced Concrete Pavement Stresses

Abstract

A rapid solution is presented for predicting critical tensile stresses on the top surface of continuously reinforced concrete pavements (CRCP). These tensile stresses are responsible for the development of CRCP punchouts and have to be considered in a mechanistic-based design procedure. The solution is based on a combination of the neural network (rapid solution) and finite element (numerical analysis) techniques. This approach combines the convenience and computational efficiency of neural network solutions with the flexibility and power of the finite element analysis. Such a combination is quite efficient for analyzing damage accumulation in CRCP, which requires predicting portland cement concrete tensile stresses for a large number of loading and site condition combinations. The procedure for stress prediction is based on the finite element model developed with ISLAB2000. The neural network has been trained with the results from ISLAB2000. A concept developed specifically for this study—the equivalent CRCP structure—was used extensively to reduce the number of independent parameters of the neural net work and speed up its training. The proposed rapid solution provides a good match of the ISLAB2000 stress values for a small fraction of the computation cost. This makes the rapid solution CRCP a natural choice for analyzing CRCP stresses and for inclusion in a mechanistic-empirical CRCP design procedure.