This study deals with the problem of railroad network design problem through link capacity expansions with a focus on freight transportation. It is intended to reach a certain level of network throughput in terms of freight demand, while holding the total expansion costs at the minimum level. Given that all railroad network links can be considered as potential candidates for expansion, in large-scale the problem will be computationally intractable with an NP-Hard structure. To tackle such a problem, a bi-level structure is proposed and implemented in this study. In the lower level algorithm, the freight traffic equilibrium within the network links is modeled, while the upper level algorithm targets the minimization of the total expansion costs over the network. In the lower level, to address the freight traffic equilibrium, an incremental assignment algorithm is taken into account. In this algorithm, small fractions of freight demand are repeatedly assigned and added to the network links. A network link is considered as blocked and removed from the network topology when its freight traffic flow reaches the corresponding capacity. The incremental assignment proceeds to further iterations until either the entire demand matrix is assigned or the whole network locked with respect to its blocked links. In the upper level, in an iterative structure, the algorithm attempts to increase the freight throughput over the network by iteratively expanding network links. Using the lower level algorithm as an evaluation tool, the algorithm first performs an incremental assignment, obtains the blocked links of the network, and finally, in a greedy feature, selects the most cost-effective blocked links to be expanded. The entire process of evaluating the network throughput (i.e. incremental assignment), specifying blocked links (i.e. links leading to the network lock), and expanding the most cost-effective blocked links is repeated until the algorithm reaches a certain level of throu
