This research addresses optimizing production, inventory, location, and routing in a multi-level supply chain. In the studied supply chain, products are first sent to several distribution centers (DC) after production. Next, these products are delivered to several companies. Companies are divided into two independent groups considering sharing logistics resources. In this regard, vehicle routing optimization is applied for each group. For this purpose, a mixed-integer linear programming model has been developed. This model simultaneously minimizes total supply chain costs and total negative environmental impacts. A multi-objective dragonfly algorithm (MODA) has been developed to solve this model. Then, the implementation results are compared with the Non-Dominated Sorting Genetic Algorithm and Epsilon Constraint Method (EPC). Comparing the solution methods based on numerous test problems shows that the proposed MODA algorithm can be examined differently. The MODA algorithm has been able to have a significant advantage in providing solutions close to ideal points. Moreover, this algorithm has provided the most effective Pareto solutions in different test problems. However, the NSGA-II algorithm performs better regarding the spread of non-dominated solutions.
