| This thesis looks upon the application of constraint programming to solve fleet management problems under time constraints. These transportation problems, of great economic importance, have been the subject of considerable research from both academia and industry. There exist numerous variants of these problems but all have the common objective of servicing a set of customers with a set of routes, while respecting time considerations at each visit. However real life problems are becoming harder to solve since the number of complex constraints to satisfy is constantly growing.; Fortunately using constraint programming, a flexible and expressive optimization paradigm, allows easy modeling of most constraints encountered in real life problems. By combining this approach with known operations research techniques it is possible to devise efficient and flexible solution methods, that can handle most complex transportation problems.; The thesis is structured around three major themes, dealing with different aspects of time constrained fleet management problems. The first theme is the study of static fleet management problems with the combination of constraint programming and local search, in order to obtain good results on large size problems. The second will consider solving the same problems with an exact method, here a hybrid of constraint programming and column generation. Finally the real time fleet management problem will be considered, but more precisely a variant that is very hard to solve using only traditional methods. |
