Supply chains are complex systems and stochastic in nature. Nowadays, logistics organizations are expected to be efficient, effective, and responsive while respecting other objectives such as sustainability and resilience. In this work, a multi-agent model is proposed for a multi-plant, multi-product supply chain network that supports an open network with n nodes (plants, retailers, etc.). Three replenishment policies are proposed with different criteria of selection. A multi-agent simulation tool was used to implement the proposed multi-agent model. Different scenarios and configurations, varying from static to dynamic, are defined and tested. The first objective of this work is to compare the performance of physical internet supply chain and classical supply chain networks using holding and transportation costs as key performance indicators. The second goal is to assess the performance of different replenishment policies for multi-plant, multi-product physical internet supply chain network. Experiment results validate the efficiency of the model to assess the performance of supply chain and to optimize the replenishment decisions.