文摘
A continuous approximation (CA) model is proposed for the fleet composition problem in rectangular grid networks. The model extends Jabali et al.’s (Transp Res Part B 46(10):1591–1606, 2012) methodology for radial networks. In the model, delivery points are assumed to be uniformly distributed in a square-shaped service region. The region is partitioned into zones, each zone is allocated to one vehicle, and each vehicle has to visit all the delivery points within its zone. The problem involves finding the optimal fleet of vehicles to minimize the total fleet acquisition costs and travel costs. The CA model is compared to a well-known column generation heuristic. Although the two models have similar results, the CA model is much faster with a computation time of less than 1 s for all experiments. Sensitivity analysis is performed on different parameters. Results show that the largest available vehicle is commonly filled to capacity and is used in the mid-section of the service region. Moreover, increasing the time limit constraint has a step-wise impact on the fleet composition.