Development and Adjustment of MDOT’s Pavement Condition Ratings (PCRs)
Type of DegreeMaster's Thesis
Civil and Environmental Engineering
MetadataShow full item record
Pavement condition assessment is a fundamental activity of every pavement management system (PMS). It determines the overall road network condition, as well as possible maintenance and rehabilitation activities (M&R) for restoring pavements’ structural and functional capacity. Before determining those required M&R procedures, the overall network condition is reported to upper management and legislature for a specific allocation of the available resources. The Mississippi Department of Transportation (MDOT) still use the pavement condition rating (PCR) for high-level reporting due to their comprehensive interpretation (0- 100 scale, comparable to a school examination). The current PCR calculation method is based on a point deduction process according to each distress type, severity, and extension. However, multiple changes such as variations in the pavement distresses accounted, multiple vendors in charge of condition assessment activities, and different sections being analyzed have occurred in the past decades. Therefore, MDOT requires a unified and straightforward PCR calculation process that comprises the predominant distresses for each pavement type. Consequently, the previous statement was considered as the main objective of the present study. PCR models were developed based on a provided database that included multiple pavement sections, their respective distresses, and PCR values from MDOT’s biannual pavement network assessment (1995 to 2020). A least squares procedure was used for determining the specific weights of each distress or performance indicator. These weight factors also helped adjust the predicted PCR values to historical PCR trends. Statistical analysis was performed to determine if there was an adequate agreement between predicted and historical PCRs, as well as appropriate accuracy for the developed models. The main results reveal that there is a strong to very strong agreement between the predicted and historical PCRs. Moreover, the developed models can accurately predict between 65% to 70% of the unseen data. As data become available further calibrations can be performed and models’ accuracy can be improved.