Network Design Problem under Uncertainty

This paper proposes a bi-level model for network design problem (NDP) under demand uncertainty. Most of the conventional NDPs are carried out under deterministic condition. Only a few studies pay attention to NDPs under uncertainty in transportation networks. In this paper, the conventional NDP is extended to explicitly take into account the demand uncertainty, which refers to the fluctuations of day-to-day total number of vehicular trips between an origin-destination pair during the observational window. Under such uncertain condition, travel times are no longer deterministic and travelers have to consider risk-taking path choice behaviors. The corresponding reliability-based traffic assignment model is adopted as the lower-level problem of the bi-level model. The upper-level problem utilizes the expected total system travel time as the objective function for determining the optimal capacity enhancement scheme. An Iterative-Optimization-Assignment algorithm is developed to solve the bi-level model. In the numerical example, it is found that travelers' risk-taking behaviors have significant impact on the results of NDP under uncertain condition.