Jointed Plain Concrete Pavement Model Evaluation

Abstract

The mechanistic–empirical pavement design guide developed under NCHRP Project 1–37A employs mechanistic models to calculate pavement primary responses for predicting pavement performance. Therefore, it is imperative that the primary response models be evaluated in relation to their prediction accuracy. In this study, both theoretical accuracy verification and comparison with field measurement were employed to judge the prediction quality of three primary response models commonly used in concrete pavement analysis, namely, ISLAB2000, JSLAB2004, and EverFE2.22. The theoretical evaluation involved simulations, sensitivity trials, and comparisons of calculated slab stress, strain, and deflection responses of a number of hypothetical pavement systems. The field evaluation involved direct comparison of model-simulated responses against those measured from the SHRP Ohio test section. It was found that the agreement between analytical solutions and those obtained by the three analysis programs was affected by the load location, location of the desired response, and consideration of slab self-weight in the analysis. The study also showed that the three analysis models were fairly able to simulate field-observed responses by using both the multiple-step and one-step approaches. For the pavement section analyzed in this study, the average percentage relative deviation of peak predicted responses from corresponding measurements (deflection and strain) was estimated as 16.2%, 17.6%, and 22.9% for ISLAB2000, JSLAB2004, and EverFE2.22, respectively. Because the evaluation of the three programs was conducted against one pavement structure, additional verification is required through the use of pavement structures with different materials and layer thicknesses. Besides accuracy, the user's selection of a specific analysis program will depend on run time, ease of use, cost, and ability to incorporate the base and subbase layers of pavement structures.