Railroads constitute an important component of the North American intermodal transportation network as they move containers long distances inland. They face difficult large-scale decision-making problems, and in this work, we introduce such a problem: the tactical Intermodal Blocking and Railcar Fleet-Management problem. Taking train schedule and related capacity as an input, we aim to build a cost-effective plan that satisfies demand expressed as numbers of containers of different types to move. It corresponds to a service network design problem with fleet management decisions: the design decisions are associated with the selection of extra train services, and the selection of so-called railcar blocks. In addition, we manage a heterogenous fleet of railcars that need to be matched to the demand characteristics. We formulate the problem as a mixed-integer linear program, and it cannot be solved within a ten-hour time limit with a general-purpose solver. We propose an exact solution approach based on iterative fixing of variables which allows us to solve real large-scale instances from our industrial partner, the Canadian National Railway company, that is one of the largest railroads in North America. The results show that solving this problem can lead to important gains in terms of cost-reduction and increased demand satisfaction.

Multilayered networks are used to model systems with multiple interacting components, each being represented by a network layer. More than two layers may be involved. The ``multi-layer network" term is generally associated with complex multilayered problem settings, frequently encountered in planning transportation and telecommunication systems, where an arc in a given layer is defined with respect to a set of arcs in another layer. Multi-layer network design simultaneously designs all layer-specific networks to satisfy at minimum overall cost a given set of origin-destination demands. Extending the already-difficult network design problem class, multi-layer network design presents additional modelling and algorithmic challenges arising from the connectivity relations and requirements characterizing the design, flow, and attribute decision variables associated with the arcs of interconnected layers.
We recall and enhance the basic definitions and formulations, and introduce new ones for richer multi-layer networks, with more than two layers and connectivity relations involving several layers simultaneously. We illustrate these concepts through multi-layer service network design models for tactical planning of consolidation-based freight transportation carriers.

During this seminar, we embark on an in-depth discussion of the freight transportation system, emphasizing a multi-tier hierarchical hub network integrating air, rail, and highway transportation modes. This structure encompasses primary hubs, secondary hubs, and service nodes interconnected to meet time-sensitive demands. Our main focus lies in developing an optimization model for the integrated hub location and scheduled service network design problem, catering to strategic and tactical planning levels. Additionally, we introduce a novel decomposition-based metaheuristic algorithm to effectively address the complexities of this integrated system. Through this method, we aim to break down the original problem into manageable subproblems, thereby enhancing efficiency and solution quality. For each subproblem, we do exploration and exploitation phases to find near-optimal solutions. Our results empower decision-makers to optimize hub network design and allocate capacity configurations, thereby enhancing service efficiency and ensuring cost-effectiveness.

This study focuses on a profit maximization variant of the multi-commodity network design problem with location decisions and incompatible commodities. A novelty of the problem is the restriction of commodity flows for safety and security reasons, as commodities have a predetermined list of other commodities that cannot be shipped together. Therefore, decisions regarding serving commodities should be made simultaneously with design and flow decisions as demand selection is affected by them. The problem, therefore, is to design a network by selecting both nodes and arcs that maximizes the total profit including the revenue obtained from selecting demands, design cost (for the selection of nodes and arcs), and the flow costs to transport the commodities. In this research, we assume that commodities are transferred through transshipment locations, and direct shipment between the origin and destination of a commodity is not allowed. The shipment between transshipment nodes is allowed to use several arcs. Finally, an extensive set of computational experiments is conducted on a set of instances from the literature.