The objective of infrastructure management is to provide optimal maintenance, rehabilitation and replacement (MR&R) policies for a system of facilities over a planning horizon. While most approaches in the literature have studied the decision-making process as a finite resource allocation problem, the impact of construction activities on the road network is often not accounted for. The state-of-the-art Markov decision process (MDP)-based optimization approaches in infrastructure management, while optimal for solving budget allocation problems, become internally inconsistent upon introducing network constraints. In comparison, approximate dynamic programming (ADP) enables solving complex problem formulations by using simulation techniques and lower dimension value function approximations. In this paper, an ADP framework is proposed, wherein capacity losses due to construction activities are subjected to an agency-defined network capacity threshold. A parametric study is conducted on a stylized network configuration to infer the impact of network-based constraints on the decision-making process.
The objective of transportation infrastructure management is to provide optimal maintenance, rehabilitation and replacement (MR&R) policies for a system of facilities (roads, bridges, tunnels, etc.) over a planning horizon. While most approaches in the literature have studied it as a problem of optimal allocation of limited financial resources, the interdependence between facilities, as introduced by a unifying network configuration, is often not accounted for. The implementation of MR&R activities on road networks can result in significant delays to travelers due to loss in network capacity, detours, etc. According to one estimate, more than 60 million vehicles per hour per day of capacity were lost due to work zone activity on the National Highway System over a two week period in the United States in 2001 (Wunderlich and Hardesty, 2003). Given that the impact of scheduling work zones, especially in saturated traffic flow conditions, can be severe, it is important to systematically address and incorporate these user concerns within the decision-making process.
The recognition of an over-arching network configuration introduces several challenges, as well as opportunities, for system-level MR&R decision-making. In this paper, network-induced inter-facility interactions are examined in the context of structural interdependence among facilities, wherein MR&R activities on individual facilities leads to a cumulative effect on the capacity of the network.
Existing system-level MR&R decision-making paradigms do not adequately account for the impacts of construction activities on the road traffic. Consequently, the measures taken by agencies to mitigate travel time increases occur at the project-level, i.e., once the MR&R activities have already been determined. However, in order to effectively address these user concerns at the system-level, the network configuration can provide insights into determining how each facility affects the system performance (such as the capacity or connectivity of the network).
For instance, consider a system comprising of four road segments. Without explicitly identifying the individual pavements within the network, it is impossible to gauge the impact of the proposed maintenance activities. However, it can be seen that if all the facilities in the system are arranged in series, as shown in Fig. 1a, then each facility is critical for the functioning of the network. As a result, a partial/complete road closure during peak hours of traffic will adversely affect the traffic. In comparison, if all the road segments are in parallel (Fig. 1b), the network exhibits a very high level of redundancy. Consequently, potential road closures can be accommodated by rerouting traffic through parallel routes. A more realistic network would perhaps comprise of links in both series and parallel, as shown in Fig. 1c. Hence, in order to better mitigate the impact of construction activities on road users, the identification of optimal system-level MR&R policies should capture the relation between the road segments in a systematic manner.
Full-size image (13 K)
Fig. 1.
Different types of network configurations.
In this paper, the state-of-the-art MDP-based approaches were extended to incorporate network-based considerations into MR&R decision-making. The structural interdependence problem was motivated to account for the adverse impact of construction activities on the road network. The results indicate that randomized policies, while ideal for modeling budget allocation problems, do not adequately capture these network-based constraints. In comparison, ADP is able to provide better results by learning the value function parameters using simulation techniques.
In the context of infrastructure management, the use of approximate dynamic programming is a recent phenomenon. While previous research established that MR&R decision-making problems can be modeled using ADP for budget allocation problems, it is shown here that ADP is beneficial for modeling problems with complex inter-facility dynamics.
The future work includes addressing some of the limitations associated with a capacity-based framework, and scaling up the network size so as to better assess the computational performance of ADP. Other problems relevant to the applicability of ADP in infrastructure management include accounting for economic interdependence between individual facilities, wherein insights from the parallel machine replacement literature about the optimal policy structure can be helpful for developing suitable basis functions.
