A new framework for mixed-user dynamic traffic assignment considering delay and accessibility to information

We consider a transport network consisting of two classes of users, namely, selfish users whose objective is to minimize their individual travel times, and cooperative users whose objective is to maximize the aggregate throughput of their class, or equivalently, to minimize the total travel time of their class. The selfish users decide the paths from their respective origins to destinations depending on their knowledge of the traffic conditions prevailing up to the beginning of their journeys, and they do not alter their paths once their journeys start. The cooperative users, on the other hand, keep making the decisions of ‘the next link to take towards the destination’ throughout their journeys based on their current knowledge of the traffic conditions, which evolves with time and the availability of traffic information. We develop a new framework for the problem of informed mixed-user dynamic traffic assignment (IMUDTA) considering the delay and the degree of penetration of information. The proposed framework enables us to study the route choices made by the two classes of users and the resulting network performance. Due to its non-linear formulation with high complexity, we propose a novel incremental solution method to this problem where we present a transformation of this complex IMUDTA problem into a relaxed linear programming (LP) model. We show via numerical results that (i) appropriately postponing (intentionally or unintentionally) the access to information can improve the system performance, (ii) better information penetration increases the robustness of the network performance against the information delay, and (iii) the value of information (in terms of improving network performance) is reduced when the degree of information penetration is high. The applicability of our framework is demonstrated in a large-scale network example, and analyze the computational aspects and the impact of information on network performance with varying information delay and penetration.