Abstract:
Effective design of transit routes and service frequencies can decrease the overall cost of providing transit services, which generally comprises passenger costs and operator costs. Usually route design is done with route scheduling and efficiency of the service is significantly depends on the route network. However, the route design problem is not straightforward due to its multi-objective nature, non-linearity and non-convexity of the objective function. The relation of the constraints and decision variables to the objective function are non-linear. Non-convexities are illustrated by the fact that more buses can be deployed without decreasing total travel time by changing the route length. Non-availability of reliable passenger demand data is another constraint.
At the first stage of this study, primary bus routes that operate between Divisional Secretariats Divisions were considered. Primary route network was developed based
on the passenger demand distribution in the existing system. An algorithm was developed to generate the route network with the condition that all nodes (zone centroids) have to be served at least by one bus route. In this algorithm, the maximum demand O-D pair in the demand matrix is considered first. Routes were assigned along the minimum path between those two zones assuming all the passengers would select the shortest travel path. Any inter-zonal travel demand between zones that falls within the route under consideration is also added to this route and taken off from the demand matrix. After generating the primary network, user is given the choice to fine-tune the network by using a set of algorithms for route merging, adding links and route sprouting. Fleet requirement, passenger transfers saving after each modification, revenue and operating costs per bus trip were calculated to evaluate route network.
To determine the passenger demand distribution, an Origin-Destination matrix was developed based on bus passenger interviews and available socio-economic information. Passenger demand distribution over the study area was obtained based
on a model calibrated using household & roadside interview data and travel costs between node pairs in the selected zones. It can be seen that most of the existing routes are operating in the high demand corridors. But there are new routes to be introduced to newly developed zones in the study area
