Algorithm for Two-Phase Facility Planning via Balanced Clustering and Integer Programming

Abstract

We present a solution for a two-phase facility planning scenario where in the first phase, there is some flexibility in determining where the locations of facilities (or sources) should fall. And in the second phase, new waypoints (or sinks) are added, but the location of the facilities are static. This solution applies the use of balanced clustering - using a modified K-Means approach, ensuring the cardinality of each group to be equal. Subsequently, it is followed by an integer programming solution, to solve the Hitchcock Transportation Problem. We show that the final solution can justifiably approximate the near optimal solution, and be a successful guide for facility planning in this specific scenario.