This paper presents developments on modeling and simulation of decentralized decision making in supply chains. To simulate supply chain, multi-agent systems appear quite natural. However, the natural autonomy of agents and supply chain actors in their decision making questions the consistency of these decisions within the system. Decentralized decisions impact other actors, sometimes in a strong manner and with a risk of inconsistency. This paper focuses on supply chain planning and the associated decentralized decisions: inconsistency on such plans can lead to infeasibility. Starting from a given peer-to-peer negotiation process and its improved version, the purpose is to work with a serial chain of three actors, thus requiring several peer-to-peer discussions. The sequence and synchronization of these discussions, local decisions, and plan validations is not straight forward and the paper identifies 3 major cases. 2 of these cases and the two versions of the peer-to-peer negotiation process are simulated to understand their behavior through time. Generalization toward a n-echelon system is also discussed.