Incorporating Risk and Uncertainty into Infrastructure Asset Management Plans for Pavement Networks

Abstract

The risk in multiyear asset management plans for pavement networks stems from uncertainties in inputs to these plans. Although the literature is rich in probabilistic models and procedures for individual components of asset management systems (e.g., probabilistic lifecycle cost analysis and probabilistic performance prediction models), risk and uncertainty have not yet been fully and explicitly incorporated into asset management plans. In this paper, performance risk is expressed as the probability of failing to sustain an average Condition Score greater than or equal to 70 (out of 100) throughout the planning horizon and as the probability of not being able to maintain 90% of the network in what could be classified as a good or better condition. This paper provides a methodology for incorporating risk assessment into asset management plans by accounting for uncertainties in key inputs to these plans with specific application to pavements. The inputs used include current pavement condition, available maintenance and rehabilitation funds, unit cost of maintenance and rehabilitation (M&R) treatments, and performance predictions. The uncertainties in these inputs are modeled as probability distributions, and the corresponding risks are evaluated using a Monte Carlo simulation. Maintenance and rehabilitation projects at network level are selected annually optimizing the benefit-cost ratio. This methodology was applied to a portion of the pavement network in Texas to demonstrate the practicality of the proposed approach. This pilot evaluation suggests that uncertainties in performance predictions and M&R unit costs have the greatest effect on the network’s performance risks in a multiyear planning period. This methodology also helps to evaluate the sensitivity of the overall risk to the sources of uncertainty. The incorporation of traffic uncertainty and monetarization of the benefit and consequently the risk can improve the methodology.