Capacitated K-Means Clustering Algorithm for Mobile Depot Allocation

Document Type

Conference Proceeding

Publication Date

1-1-2022

Abstract

A fundamental challenge in last-mile delivery is to assign orders from main depots often located at the periphery of cities to different temporary transshipment points (aka mobile depots), each with limited capacity. In this study, we propose a novel capacitated K-means clustering algorithm which assigns customers' orders to different clusters considering the location and the number of orders from each customer. In the traditional K-means clustering method, the only criteria considered is distance but, in this study, the clustering algorithm is modified considering customer demands and mobile depot capacity. The optimal location for the mobile depots is chosen from a possible set of mobile depots. In capacitated K-means clustering algorithm, a given set of customer orders are assigned to k disjoint clusters. We propose a new cost function considering mobile depot capacities, customer locations, and their order demands. We also developed an integer programming model to compare the results of the capacitated K-means clustering method to the integer programming model. The goal is to minimize total travel distance considering capacity constraints.

Publication Source (Journal or Book title)

IISE Annual Conference and Expo 2022

This document is currently not available here.

Share

COinS