Green AI in Logistics

Real-world evaluat ion of hybrid Green Al for sustainable and efficient smart supply chain distribution

This paper proposes a hybrid Green AI framework for achieving sustainable physical distribution in smart supply chains. The framework integrates real-world geospatial data, multi-objective optimization, and meta-inference algorithms. It aims to reduce transportation costs, delivery times, fuel consumption, and CO2 emissions while maintaining operational efficiency, in line with Sustainable Development Goals 11 and 13. The new MOIAC algorithm enhances ant colony optimization by incorporating environmental weighting into pheromone updates. Real-world validation utilizes Google Maps API data from 19 Egyptian cities, with demand modeled using a Zipf distribution (α = 0.9). OR tools serve as a high-fidelity proxy to simulate the performance of MOIAC and MOIPS under real-world conditions. The results show a 26.7% reduction in total distance, operating costs, and CO2 emissions compared to baseline methods, with MOIAC achieving the lowest average response time (120 ms). Comparisons with six algorithms—including Greedy, PSO, ACO, and Genetic-confirm the superiority of the proposed approach. This framework demonstrates how green Al and geospatial intelligence can contribute to theoretical optimization and practical logistics, providing a scalable, environmentally friendly, and operationally efficient solution for modern supply chain distribution.