This study applies a Genetic Algorithm (GA) to optimize the Vehicle Routing Problem with Time Windows (VRPTW) for NAFTAL, Algeria’s national fuel distribution company. The model minimizes both fleet size and total travel distance while achieving high compliance (99.3%) with customer time constraints. Using operational data from the Constantine regional network one depot serving 148 service stations (149 nodes in total ), the GA achieved optimal solutions deploying 94 vehicles covering 15,415.63 km. Results demonstrated exceptional convergence stability (σ = 0.00 across 40 runs) and high computational efficiency (under 60 seconds per optimization run). Sensitivity analyses confirmed the robustness of the calibrated configuration, highlighting its reliability and scalability for real-world logistics. The proposed framework provides NAFTAL with a cost-effective, consistent, and practical decision-support tool for optimizing fuel-delivery operations. Future research will focus on integrating machine learning for demand prediction, extending the model to multi-product and heterogeneous-fleet routing, and enabling adaptive real-time optimization to support smart and sustainable logistics.
