A GRASP-Tabu Heuristic Approach to Territory Design for Pickup and Delivery Operations for Large-Scale Instances

Table 1

Instance characteristics.

Type	Characteristics
Symmetric	Unrealistic instance but used to measure the performance of the heuristic for larger instances. Optimal solutions consist of “symmetric” districts: districts with the same workload content and the same shape. To generate the instance, a district is created and rotated to form the rest of the districts symmetrically. Workload content is also equal for all the districts. Euclidean distances are computed even if the points are not directly connected to guarantee symmetry.
Semisymmetric	Very similar to the symmetric instance, but Euclidean distances are computed only for those points connected by an edge. For the rest a shortest path distance is computed, and, hence, there is no guarantee that an optimal solution corresponds to a symmetric configuration.
Asymmetric	More general instances consist of vertices uniformly distributed over a plane using Euclidean distances for those vertices connected by an edge and shortest path distances for the rest.
Urban	Instances that consist of vertices distributed over a region that resembles the structure of the metropolitan area of Monterrey, Mexico.
Parcel	Instances generated with GPS data of a representative workday provided by the parcel company. This data set is a full size real instance.