A dual toll pricing is a conceptual policy in which policy maker imposes toll on both hazardous materials (hazmat) vehicles as well as regular vehicles for using populated road segments to mitigate a risk of hazmat transportation. It intends to separate the hazmat traffic flow from the regular traffic flow via controlling the dual toll. In order to design the dual toll pricing policy on a highly realistic road network environment and detailed human behaviors, an extended Belief-Desire-Intention (BDI) framework is employed to mimic human decision behaviors in great detail. The proposed approach is implemented in AnyLogic agent based simulation software with using a traffic data of Albany, NY. Also, search algorithms in OptQuest are used to determine the optimum dual toll pricing policy which results in the minimum risk and travel cost based on the simulation results. The result reveals the effectiveness of the proposed approach in devising a reli-able policy under the realistic road network conditions.
Hazardous material transportation simulation
Currently, government agencies in North America and Europe are trying to mitigate risk associated with the hazmat transportation on a road network by separating the hazmat traffic flow from normal traf-fic flow, especially high-density traffic flows. An example of this is the ban on trucks carrying non-radioactive hazmat on certain road segments (e.g. Texas Department of Transportation 2009 has a list of prohibited roads for Texas). In the literature, these types of policies are categorized into network-design (ND) policy. Significant research works have been done in this field to minimize hazmat transportation risk (e.g., Kara and Verter 2004). The ND policy is very effective to restrict hazmat traffic from highly dense regular traffic flow. However, the main limitation of the ND policy is not to consider the carriers’ priorities such as a travel cost and wastes the usability of certain road segments. Moreover, only restricting certain road segments sometimes cannot rationally adjust the hazmat flows to areas with less risk (Wang et al. 2011).
Accordingly, an alternative policy tool, toll-setting policy (TS) which is a more flexible restriction policy for hazmat shipments than the ND policy, was proposed by Marcotte et al. (2009) to deter hazmat carriers from using certain road segments via toll pricing. Unlike the ND policy, the TS policy allows a government agency to consider the drivers’ and carriers’ priorities such as drivers’ preference to avoid the road which has a high risk, and carrier companies try to achieve more profit by choosing the least cost path that reduces an operational cost (Pécheux et al. 2004), simultaneously. Recently, Wang et al. (2011) suggested a dual toll pricing framework to control both regular and hazmat traffic flows for the public safety since hazmat carriers are not only a part of the hazmat traffic flows, but also belong to, and cannot be separated from the regular traffic flows. By separating the hazmat traffic from the heavy-congestion regular traffic flows, the dual toll policy tries to mitigate severe accident risks and to avoid peak-time traffic congestions. However, the existing research works have some unrealistic assumptions, for example, individual driver’s behavior (e.g. route planning behavior) is the same as others and the driver has perfect information of the current status of the network. Thus, in order to design and evaluate a more reliable dual toll pricing policy, this paper adopts the extended Belief-Desire-Intention (BDI) framework which is one of the well-known models to mimic a human decision behavior (Lee, Son, and Jin 2010). Since the extended BDI framework is able to illustrate a human reasoning process based on perceived information in a greater detail, it has been successfully implemented in various fields such as crowd management system and manufacturing system. In this paper, this framework will demonstrate an individual driver’s route choice behavior with imperfect information, which is implemented in the agent based simulation model.