Heuristic approach for solving employee bus routes in a large-scale industrial factory

• Main Authors: Leksakul K., Smutkupt U., Jintawiwat R., Phongmoo S.
• Format: Journal
• Published: 2017
• Online Access: https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85014517582&origin=inward; http://cmuir.cmu.ac.th/jspui/handle/6653943832/40586
• Institution: Chiang Mai University

© 2017 Elsevier Ltd This paper compares different methods for solving a location-routing problem (LRP), using real-world data from the bus transport service for employees of a large-scale industrial factory in Thailand. We tested four AI (artificial intelligence) techniques Maximin, K-means, Fuzzy C-means, and Competitive Learning and two hybrids of these four K-means with Competitive Learning and K-means with Maximin to allocate the bus stops. The efficiency of the algorithms was compared, in terms of the quality of the solutions. The K-means with Maximin provided the best solution, as it minimized number of bus stop locations and employees’ total traveling distance while satisfied employee at maximum radius 1.73 km, compared to K-means with Competitive Learning, as the same number of bus stop it provided higher total traveling distance and maximum radius. The other non-hybrid techniques provided higher number of bus stop locations. We then used ant colony optimization (ACO) to determine the optimal routing between the 300–700 bus stops as allocated by K-means with Maximin. The optimal bus routing to transport the factory's 5000 plus employees required 134 buses (134 independent routes) covering 500 bus stops and traveling nearly 5000 km. While optimal, this routing was costly and created monitoring difficulties. To address these concerns, we constrained the number of bus routes; while this dramatically increased the total distance, it provided a more practical solution for the factory.