Advancing pavement management systems using a two-tier probabilistic performance modeling approach for improved condition forecasting and budgeting analysis

Abstract

Pavement management systems (PMSs) are essential tools for transportation agencies to develop maintenance and rehabilitation programs for their pavement networks. These systems use prediction models to forecast future pavement conditions, which can be deterministic, probabilistic, or increasing, based on artificial intelligence (AI). While deterministic models are common, they do not account for uncertainties in pavement performance forecasting. This paper presents a two-tier probabilistic modeling approach to advance PMSs by addressing variability while improving prediction accuracy. The first tier develops probabilistic-based pavement performance curves (PBPPCs) using statistical percentiles to represent pavement sections with high, medium, or low performance. The second tier applies a dynamic calibration method that adjusts performance curves for individual pavement sections using data from condition field inspections. A comparative analysis between deterministic budget estimates and those derived from the two-tier probabilistic model revealed substantial differences, reinforcing the value of incorporating uncertainty into the modeling process. By integrating this probabilistic modeling approach into PMSs, agencies can make more informed decisions, improve budget allocation strategies, and promote proactive asset management practices to maintain pavement networks in a state of good repair.