Empirical-Markovian approach for estimating the flexible pavement structural capacity: Caltrans method as a case study

• Main Author: Khaled A. Abaza
• Format: Article
• Language: English
• Published: Elsevier 2021-06-01
• Series: International Journal of Transportation Science and Technology
• Subjects: Flexible pavement; Pavement design; Structural capacity; Pavement rehabilitation; Markovian processes
• Online Access: http://www.sciencedirect.com/science/article/pii/S2046043021000010

An Empirical-Markovian approach is proposed to estimate the pavement structural capacity as a function of key stochastic and design parameters. The key stochastic parameters are the initial and terminal deterioration transition probabilities typically estimated from pavement distress records. These two transition probabilities have a major impact on the pavement performance trend predicted using Markovian processes. In addition, typical pavement design factors are included as related to traffic loadings, materials properties, and climate conditions. In particular, two distinct Empirical-Markovian models are developed to estimate the pavement structural capacity in terms of relative strength indicators such as the structural number (SN) and gravel equivalent (GE). The first model can estimate the structural capacity based on the initial transition probability and relevant design parameters, while the second one deploys the terminal transition probability along with other design parameters. The recommendation is to use the higher of the two structural capacity values estimated for a particular pavement project. The sample models presented based on the California Department of Transportation (Caltrans) design method have resulted in good model fittings as demonstrated by the various deployed statistics and error analysis, thus indicating the usefulness of the proposed Empirical-Markovian approach in estimating the pavement structural capacity for rehabilitation and design purposes. The model exponents can be obtained from solving a linear system of equations using data from a small project sample or solving a multi-variable linear regression model when a large road sample is available. The latter case provided sample generalized models with statistics indicating their high significance.