Supply Chain Modeling and Simulation Using Agents

Abstract

Good supply chain management is crucial for business success in today?s increasingly complex, global, and competitive business environment. Agent-based modeling and simulation (ABMS) is a natural fit to supply chains as it uses a bottom-up approach by modeling each supply chain entity as an agent which can interact with one another and response to changes based on its own interest. ABMS has been implemented to investigate, analyze, and diagnose supply chains. However, most of existing ABMS approaches are complex, and resulting models are hard coded and difficult for nontechnical users to understand, manipulate and analyze.

We propose a business process modeling notation (BPMN) based framework for modeling and simulation of integrated supply chains. BPMN is a widely recognized unified graphical modeling notation for business processes. A key advantage of BPMN is its ability to transform documentation of process flows to executable process model with simple notation. The proposed framework combines the advantages of ABMS and BPMN and it is validated by replicating an existing multisite specialty chemicals supply chain model built in MATLAB SIMULINK. The built BPMN-based model has a more natural representation of the chemical supply chain and faster simulation. Various scenarios also demonstrate that a BPMN-based supply chain model is easier to understand, manipulate, and has high level of scalability and flexibility.

The strict safety and environmental regulation on chemical storage and transportation, expensive purchasing, leasing and maintenance charge of tank fleet, and the serious consequences from tank cars shortage make tank fleet sizing become an essential part of chemical supply chain management. We build an agent-based simulation model of a multisite chemical supply chain through BPMN-based framework to address the tank fleet sizing problem. The simulation model explicitly takes into account of the independence of supply chain entities and their interactions across various supply chain operations such as replenishment planning and order assignment. Tank fleet is modeled as a set of objects that travel across the supply chain. The supply chain model is simulated with five tank fleet routing policies under different fleet sizes and various conditions. Optimal tank fleet routing policy and size are determined based on the comparison of the simulation results. In conclusion, BPMN-based supply chain modeling and simulation framework make it easier to design, model, simulate and manipulate agent-based model of supply chains and it has high level of scalability and flexibility. BPMN-based model serves as a qualitative and quantitative tool to support decision making in chemical supply chains including handling chemical supply chain disturbances and policy evaluation.