A new predictive modeling approach is presented and the approach for localized failures in Illinois continuously reinforced concrete pavements (CRCP) is demonstrated. Some data retrieval guidelines from the illinois Pavement Feedback System data base are first presented. A preliminary data analysis was conducted to assist in data cleaning and assessing the variability of the data before the analysis was performed. Several modern regression techniques ("robust" and "nonparametric" regressions) were introduced in a proposed new predictive modeling approach. The proposed modeling approach was used to develop an improved model for localized failures in CRCP. The resulting model includes several variables such as cumulative ESALs, slab thickness, content and methods of the steel reinforcement, and base type for the prediction of CRCP failures. A sensitivity analysis was also performed to illustrate the effect of these variables on failures. Slab thickness and steel content are by far the most significant variables affecting failures. Crack spacing had no effect.
Relation:
Transportation Research Record: Journal of the Transportation Research Board 1505, pp.75-84
