Hybrid Technique for Calibrating Network-Level Performance Models of Continuously Reinforced Concrete Pavements

Abstract

Pavement performance models exist in various forms to cater to the pavement management agencies’ needs and resources. Well-calibrated models are needed to accurately predict future pavement conditions and to forecast and prioritize confidently the future rehabilitation and maintenance expenditures. Statistical tools are commonly used to develop the performance models. These statistical models may be impractical or misleading if they do not consider experts’ opinions. This paper presents a hybrid technique where statistical tools and expert knowledge are combined for the calibration of pavement performance models. This technique was validated using historical pavement condition data for continuously reinforced concrete pavements (CRCP) from the Texas Department of Transportation’s pavement management information system (TxDOT-PMIS). The recalibrated CRCP performance models obtained with the hybrid technique represent an improvement when compared to the current models since they merge expert opinion and statistical analysis, which better reflect field observations regarding distress initiation, distress evolution rate, and maximum allowable amount of distress growth.