Here, it is considered the network location-allocation p-median problem of establishing p facilities at some of n predefined points. In the p-median problem, the objective is to minimize the distance sum from each point to the nearest facility. This is a well-known problem in the field of discrete location problems, being NP-hard for general p, even when the network has a simple structure [5,6].
Integer linear programming has remained a standard technique to solve network location problems of moderate size. As the problem size increases, approaching the solution with linear programming becomes unpractical. For larger problems, specialized Lagrangean procedures such as Lagrangean relaxation with subgradient
optimization have proven to be robust. Lagrangean relaxation is a well-known procedure to approach combinatorial optimization problems. For a review of Lagrangean relaxation techniques and applications, see [7].