Implementation of the trending mechanistic-empirical pavement design guide for resilient modulus ( ) estimation, often requires the use of models. The present study proposes a multimodel approach for model performance… Click to show full abstract
Implementation of the trending mechanistic-empirical pavement design guide for resilient modulus ( ) estimation, often requires the use of models. The present study proposes a multimodel approach for model performance assessment amongst various stress-based models, in order to identify the existence of a confidence set of models. The multimodel approach involves the use of Akaike information criterion (AIC) and Schwartz Bayesian criterion (SBC), which are respectively based on Akaike’s information theory and Bayes factor approximation. The analysis was performed on data from long-term pavement performance database for fine and coarse-grained soils, including a parametric study of the effect of spatial variability of the soil. The results show that the use of AIC and SBC is robust for handling data, even in the presence of heteroscedasticity. Furthermore, a confidence model set can be proffered using the proposed multimodel approach, irrespective of spatial variability for the fine-grained soil. The proposed approach uses the AIC/SBC differences, 95% Kullback-Leibler confidence set for the Akaike weight and posterior probability, and stability of the model ranking.
               
Click one of the above tabs to view related content.