A design methodology for bus transit route networks with coordinated operations

Previous approaches to bus transit network design focused on conventional service concepts that provide fixed route, fixed schedule, and uncoordinated systems, with the same vehicle size on all routes. As spatial trip patterns in most U.S. cities continue to evolve from a multiple origin, single destination pattern to a multiple origin, multiple destination pattern, conventional service concepts are no longer adequate to serve these new trip patterns. This dissertation incorporates three additional service design dimensions: route coordination, variable vehicle size, and demand responsive service, to better meet user needs and desired service levels.; The complex formulation and the combinatorial nature of the transit network design problem preclude solution by exact optimization models. A hybrid heuristic approach that relies on AI heuristics and search techniques and incorporates domain-specific human knowledge and expertise is developed. The overall approach has evolved from a design methodology developed by Baaj and Mahmassani (1991) for conventional transit systems. The solution approach incorporates a trip assignment model explicitly for timed-transfer (coordinated) transit systems, a frequency setting and vehicle sizing model, and a demand responsive service procedure for the integration of fixed-route and fixed-schedule service with demand responsive service.; The solution approach consists of four components. A route generation procedure constructs sets of bus routes corresponding to different service concepts and trade-offs between users and operators. A network evaluation procedure determines route frequencies and vehicle sizes and computes a variety of system performance measures reflecting user and operator costs. A transit center selection procedure identifies the set of transit centers to support the implementation of timed-transfer design and demand responsive service. A network improvement procedure applies modifications to the set of routes generated by the route generation procedure to improve performance in terms of the user's and operator's perspectives. The solution approach is tested with a benchmark problem and with data generated from the transit system of Austin, Texas.