Solving Capacitated Vehicle Routing Problem with Multi-trips and Multi-products Using a Discrete Flower Pollination Algorithm

This study addresses the optimization of non-subsidized fuel (BBK) distribution in Sulawesi by formulating the problem as a Capacitated Vehicle Routing Problem with Multi-Trips and Multi-Products (CVRPMTMP). To tackle this complex routing challenge, a Discrete Flower Pollination Algorithm (DFPA) is developed and benchmarked against two well-established metaheuristics: Genetic Algorithm (GA) and Particle Swarm Optimization (PSO). The experimental results reveal that DFPA consistently outperforms GA and PSO in minimizing both total travel distance and distribution costs. Across six months of operational data, DFPA achieves up to 57% reductions in travel distance and more than 50% savings in distribution costs. The most significant improvement is observed in April, where DFPA demonstrates superior performance by effectively balancing global exploration with local refinement. Overall, the findings highlight the robustness and efficiency of DFPA in addressing multi-product, multi-trip routing problems within geographically constrained and archipelagic regions. This contribution provides valuable insights for logistics optimization in the energy sector, particularly in improving distribution efficiency and reducing operational costs.