Abstract:
Pavement Management System (PMS) is crucial in road network assessment and rehabilitation. One of the most significant and straightforward functional performance indices used in PMS is the International Roughness Index (IRI), which generally measures the road roughness of the pavement. Sophisticated road surface profiling
equipment is required to collect these IRI data. Due to the absence of such special equipment, the IRI prediction model serves as an alternative tool in pavement condition evaluation, enabling the assessment of the future performance of road pavements. Apart from surface distresses, initial IRI, age, traffic, and climatic conditions severely influence the IRI prediction model. This study aims to develop an efficient IRI prediction model focusing on different climate zones in Canada. Using Canada’s publicly available long-term pavement performance (LTPP) database, IRI prediction models were created using both multiple linear regression (MLR) and artificial neural networks (ANNs). These models considered factors, namely, initial IRI, age, traffic, surface distresses, and climatic zones determined through the clustering technique. The performance indicator reveals that the ANN model (R2 = 0.714) outperforms the MLR model (R2 = 0.661), and the claim was validated by testing both models. This research output encourages the agencies and stakeholders to confidently use this IRI prediction model for specific climatic conditions in their flexible pavement management at the network level.
Contributors: Barzegaran J, Khan S, Swarna S, Hossain K.