Generating Near-Optimal Road Condition–Capacity Improvement Decisions Using Monte Carlo Simulations

Abstract

Road travel cost can be defined as a function of condition and volume-capacity factors. Asset managers intervene on heavily trafficked and poor condition roads based on criteria to optimize network travel and intervention (social) costs. These criteria may involve a trade-off between improving the road condition or capacity. Road performance is known through periodic inspection and stochastic modeling to estimate a deteriorated future condition. The predicted future condition and traffic growth rates change pavement section intervention (capacity or condition improvement) priority over time. The optimal road intervention choice can be determined using algorithms, including the greedy algorithm and Monte Carlo simulations. Greedy algorithms search through the entire sample space locally and stepwise to approximate global optima, whereas Monte Carlo simulations randomly sample candidate sections to generate more globally optimum interventions. This study proposes a road asset management model using Monte Carlo methods to optimally choose road network interventions considering condition and traffic changes over a planning horizon. The study includes an empirical application using real world data and compares the proposed Monte Carlo simulations approach to the greedy algorithm.