Multiperiod network improvement model

As traffic demand increases over time, improvements to existing transportation networks must be considered for enhancing efficiency, capacity, or both. Because of limited resources even justifiable projects may have to be implemented gradually. The selection and timing of improvement projects are very important to ensure the most cost-effective investment plan. Conducting this task for transportation networks is particularly challenging since the project effects tend to be inherently interdependent. By inadequately estimating project impacts during intermediate periods most existing methods tend to generate inappropriate improvement plans. The study developed a multiperiod network design problem model for the dynamic investment problem. A branch-and-bound algorithm was designed to determine the best project combinations and schedules. An artificial neural network model was used for estimating multiperiod user costs. The proposed model can efficiently handle the interdependencies among projects and demand changes in each period. This method can be used for programming various transportation network improvements or transformations.