Transit Network Assignment, Simulation and Frequency Setting: Integrated Approaches and Large Scale Application

Transit systems in the United States are facing the problem of under-utilization in small urban areas and the opposite problem of crowding in large urban areas. The solution to both of these problems lies in better allocation of the limited resources. In order to achieve that, one needs tools that would re-allocate the resources optimally or at least closer to optimum. Furthermore, developing these tools requires a better understanding of traveler behavior, transit operations and the interaction between the two, and this needs to be done in a computationally efficient manner to obtain solutions for large-scale transit networks.;The allocation of resources translates to network re-design at the planning level and to frequency setting at the strategic level. This dissertation focuses on the strategic level, where the purpose is to develop integrated approaches to transit network modeling and frequency setting that is applicable to large scale networks. First, a stand-alone transit frequency setting platform under a fixed route design is developed. A detailed problem formulation that captures the use of different service patterns for a given route; the spatial and temporal variation of ridership elasticities with respect to headways; and the vehicle load constraints at every stop of the system is developed for an entire day. Mathematically, optimal solutions are achieved. However, the shortcoming of this formulation is that it does not capture the complicated interactions happening in the network.;To capture the interaction between the supply and demand in a deeper fashion, first, a multi-modal transit least cost hyperpath algorithm is developed. This time and approach-dependent algorithm, which by itself can serve as a network evaluation tool, is used as part of a model that assigns travelers to the transit vehicles, walking and biking paths. Furthermore, a transit assignment-simulation tool is developed that both simulates the travelers and the vehicles to provide very detailed evaluations of an assignment instance coupled with a gap-driven solution approach to reach fast convergence for the assignment problem. Finally, an assignment based frequency setting algorithm is developed to overcome the shortcomings of the stand-alone frequency setting algorithm.