Asset Management Insights Using Agent Models

Asset management is a systematic process to support strategic decision making for physical assets. Recognition of budget constraints, development and use of deterioration models, development of strategies and policies, and project selection are important elements of the process. More sophisticated asset management decision support systems also recognize the uncertainty inherent in the process and the challenges in allocating resources in a spatially and socially equitable manner over the extended periods of time that the assets are expected to provide service. However, little attention has been paid to capturing the complex interactions among decision makers. While asset management has been emerging as a discipline, agent-based modeling, a tool for exploring complex systems, has demonstrated the potential to model important interactions and heuristics that reflect the actual management process. By representing a network of pavement segments as an agent-based model, it is possible to examine the effects of agencies, politicians, user actions, deterioration, random failures and various policies on the performance of the system. This paper begins by defining the agents – the pavement segments, users, politicians, and engineers – and their interactions, and explains why such agent behaviors are not captured in typical pavement management systems and life cycle cost analyses. The prototype agent-based system is then described. The authors developed two prototypes, one in MATLAB, incorporating data drawn from the Highway Economics Requirements state version (HERS-ST), and one in Repast. The prototypes are discussed, and the paper concludes with directions for future work. Overall, the models demonstrate the potential value of life cycle cost analysis and the importance of planning for catastrophic failure.