Simultaneous Decision Making of Optimal Toll Levels and Locations in a Multi-Class Network Equilibrium: Genetic Algorithm Approach

The purpose of this thesis is to explore bi-level genetic algorithm (GA) based optimization models to make decisions simultaneously for the second-best optimal toll locations and toll levels. The upper-level subprogram is to minimize the total travel time (system cost). The lower-level subprogram is a user equilibrium problem where all users try to find the route that minimizes their own travel cost (or time). The demand is assumed to be fixed and given a priority. First, two different versions of GA based solution procedures are developed and applied to an example Sioux Falls network assuming homogeneous road users in the network. This kind of problem is referred to as a single-class optimization problem. However, in reality heterogeneous road users exist. As such, the two GA options are compared with one another and the preferred GA option is further applied to the network consisting of multi-class users with different value of times (VOTs). Another heuristic approach is also considered to determine toll rates only on the most congested links for both single-class and multi-class scenarios. Such heuristic toll rates are compared with the combined solution of optimal location and toll rates to demonstrate the most congested links in a network may not be considered as intuitive candidates for optimal toll locations