In this work, we present preliminary results of our research on the construction of an agent-based simulation framework suitable to support the analysis of complex supply chain interactions as the one required for the performance assessment in collaborative supply chains. In particular we focus in the modeling of dynamic interactions through agent-to-agent message communication avoiding predefined supply chain network structures. For defining the internal structure of agents we explore the application of the SCOR reference model to bring a business process perspective and adopt the requirement of making explicit separation of the execution and control and decision making processes.