| This dissertation deals with the development of aircraft loading planning and optimization models in order to satisfy demands with an heterogeneous aircraft fleet such as the Hercules CC130 used by Canadian Air Force. We propose the use of a more realistic model than those found in the literature. Our proposed model accounts for the position of items in the aircraft, the diversity of each load, the different pick-up locations. Further extensions are presented as special cases and can be treated by this model, such as the transportation of passengers, the consideration of shipping priority, the case of items with non uniform mass density and the constraints of precedence. We consider the center of gravity (CG) of each item and we study the CG of the load/aircraft as well, in other words, the balancing of the aircraft. The problem is solved in two steps: the resolution of the problem with only one pick-up location and one delivery and the problem with several pick-ups and one delivery.;To guarantee the flight security and to save on fuel, the CG of the aircraft must fall within a predefined range. This particularity makes the modeling more difficult and presents a risk for suboptimal solution if all potential plans are not considered. We have studied this balancing problem along the longitudinal axis of the aircraft.;Transportation of items located in various geographical points is studied next. The unified formulation model is used and extensions have been made to account for cities and the cost of transport between them. Two possibilities of loading are proposed in the model. In the first case, the items are loaded according to the order of city visits and no position arrangement is possible in intermediate city. For the second case, arrangements are allowed. This later is less restricted than the former and more realizable plans were considered. On the other way, it implies forbidden plans. Each formulation is solved by a column generation algorithm. The master problem is composed of one or many demand constraints for items of the same type. The sub-problem structure remains the same: a shortest path problem with resource constraints. Since the size of each instance is large, and to find a good solution, we propose exact algorithm and heuristic to solve them. (Abstract shortened by UMI.)... |
