In this thesis document, we present a method too approximate vehicle routing costs on networks having certain geographical characteristics. The method presented here has a very short solution time and is easy to program, which permits us to rapidly evaluate possible changes to a complex distribution network. The method of approximation is based on techniques previously used to solve the traditional Bin Packing Problem.; In our study, we have shown that for the large network of a retailer operating in Canada, introducing flexibility in the delivery dates to stores can easily give 5% savings in some regions serviced by rail, representing millions of dollars for a large company. Despite lower savings in regions serviced by road, combining the results achieved here with the use of a commercial vehicle routing software looks promising but remains to be validated by the company in practice. |