Departments of Transportation regularly evaluate the condition of pavements through visual inspections, nondestructive evaluations, image recognition models and learning algorithms. The above methodologies, though efficient, have drawn attention due to their subjective errors, uncertainties, noise effects and overfitting. To improve on the outcomes of the shallow learning models already used in pavement crack prediction, this paper reports on an investigation of the use of recursive partitioning and artificial neural networks (ANN; deep learning frameworks) in predicting the crack rating of pavements. Explanatory variables such as the average daily traffic and truck factor, roadway functional class, asphalt thickness, and pavement condition time series data are employed in the model formulation. Overall, it is observed that the recursive partitioning (regression tree – R² > 0.8 and classification tree – R² > 0.6) and ANN (continuous response – R² > 0.8 and categorical response – R² > 0.6) are compelling machine learning models for the prediction of the crack ratings based on their goodness-of-fit statistics, mean absolute deviation (MAD < 0.4) and the root mean square errors (RMSE between 0.30 and 0.65).

There is a growing demand to preserve transportation infrastructure utilizing limited funds, and the modeling of flexible pavement deterioration has become an integral component of any pavement preservation model. Markov chains have been used to model the performance of pavements in various pavement management systems (PMSs). The Markov property may be considered restrictive when modeling the deterioration of transportation assets, primarily because of the “memoryless” property and assumption of exponential distribution for sojourn times in the condition states. This paper outlines a semi-Markov model for modeling pavement deterioration in which the sojourn time in each condition state is assumed to follow a Weibull distribution and, thus, is more flexible than the traditional Markov chain model. The semi-Markov model does not possess the memoryless property if the sojourn time distribution is not exponential. Monte Carlo simulations are generated for the deterioration of flexible pavements over t...

The comprehension of network-level consequences resulting from disruptive events is a main gray area in the evaluation of transportation network resilience at the regional level. Explaining hazard impacts on regional network infrastructures and identifying significantly affected areas are important for communicating the need for building resilient infrastructure. This paper presents a framework for assessing the regional network resilience by leveraging scenario-based traffic modeling and GIS techniques. High-impact-zone location identification metrics were developed and implemented in preliminarily identifying areas affected by bridge closures. Resilience was estimated, and an index developed by utilizing practical functionality metrics based on vehicle distance and hours traveled. These are illustrated for the Tampa Bay, Florida, area. Findings for 10 bridge closure scenarios and recovery schemas indicated significant regional resilience losses. The I-275 bridge closure indicated the highest functional loss to the regional network: the aggregated resilience index below 0.5 reflects severe network performance deficit and mobility limitations.

This paper presents an investigation of the Markov property underlying the stochastic deterioration models for highway bridges, including transition probabilities between the condition states. Using historical data of sojourn times for the ‘decay’ (no improvement intervention) deterioration, hazard functions were developed and ‘instantaneous’ 1 year transition probabilities estimated for the sojourn times in the condition states, for various bridge categories, by type of material and roadway carried. The rate of transition out of each state was found to be not constant relative to time, as assumed for Markov chain models, but rather, increasing with the time spent in the state. Best-fit distributions of the sojourn times were determined to not be exponential (Markov chain), but Weibull, with the parameters established using the maximum likelihood estimate (MLE) method. A semi-Markov model of the bridge deterioration process was formulated and developed, including kernels of transition probabilities and time-based matrices of multi-step transition probability functions.