Development of Vehicle Routing Problem (VRP) Models in Python to Optimize Last-Mile Distribution Routes in an SME
Year: 2026
Type: article
Abstract: Small and medium-sized enterprises (SMEs) that produce perishable food often plan last-mile distribution reactively, which raises fuel consumption, extends working days and forces vehicles to return to the plant for additional product. This study develops and evaluates vehicle routing problem (VRP) models implemented in Python for a family-owned cheese SME located in Apaxco de Ocampo, State of Mexico, which serves about 35 commercial destinations per day with more than 900 kg of product using two light trucks of 600 and 800 kg. The research followed the Design Thinking approach in five phases. The empathize and define phases characterized the current operation, three delivery corridors and the isolated fuel cost of each destination. The ideate phase formulated a heterogeneous capacitated VRP with working-day limits, multiple trips and order release times. The prototype phase implemented the model in Python using a Clarke-Wright savings heuristic, a simulated annealing metaheuristic and an exact mixed-integer verification solved with HiGHS. In the test phase, three scenarios were compared under identical orders, vehicles and assumptions: the current dispatching rules, a static scenario with all orders ready at 07:00 and a dynamic scenario with staggered production releases. Under the dynamic scenario, the model reduced the travelled distance from 382.0 to 285.4 km (25.3 %), daily fuel cost from MXN 1,293 to MXN 966, trips from six to four and minutes beyond the working day by 61.2 %. When production and dispatch were synchronized, distance fell to 152.8 km (60.0 %) with two trips and no overtime. A Monte Carlo experiment with 30 simulated days yielded a mean distance reduction of 35.1 %, and the exact verification confirmed the optimal sequence of every route. The results show that route optimization must be accompanied by the synchronization of production and dispatch.
Institution: Tecnológico Nacional de México, Campus Pachuca
Language: English
Subfield: Industrial and Manufacturing Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W7233370686
