Logistics contributes significantly to the total global greenhouse gas (GHG) emissions so it is essential that traditional static cost-based modeling methods used for logistics network planning need to evolve beyond being environmentally oriented to account for continuous changing environmental and operational conditions. This paper describes an integrated, sustainable logistics network planning framework combining artificial intelligence (AI) with digital twin (DT) technologies utilizing multi-objective simulation and reinforcement learning (RL) algorithms to simultaneously minimize total logistics cost and total carbon dioxide (CO2) emissions. The use of IoT real-time data, AI-based optimization for routing and a DT that represents the existing physical logistics system enable dynamic decision making. A composite objective function, Z=αC+βE, is based on total cost (C) and total emissions (E) using a weight (α or β) that is adjustable. Network constraints are defined such that the selected demand, capacity, and routing align. The validation of the model will use emissions standards from DEFRA and IPCC. AI-DT integrated network modeling has been shown to reduce CO2 emissions by up to 35% while increasing total logistics cost by less than 6%. A series of sensitivity analyses were conducted on various scenarios by changing carbon tax or fuel prices to determine the robustness and scalability of the framework. This framework provides a practical quantitative methodology to achieve sustainable supply chain management in an era of increasingly demanding global sustainability objectives.

Keywords

Digital Twin (DT), Artificial Intelligence (AI), Low-Carbon Logistics, Reinforcement Learning, Emissions Optimization, Sustainable Logistics.