Concurrent modeling of causality, temporality, and systemic risk in supply chains: a hybrid approach

The uncertainty of operations for supply chain involved companies is becoming more complex with the growth of globalized business collaboration. A supply chain is a complex system with dynamic flows of capital, goods, information and people. The temporal fluctuations of politics, economics, nature and technology on a supply chain process may potentially cause disruptions to the whole system. The malfunctions of supply chain systems around the world cost companies billions of dollars and months of recovery time every year. Prevention and mitigation of supply chain disruption risks are crucial for companies to maintain their competitive advantage. However, the performance of mitigation plans may be unsatisfactory due to a myriad of interactive impact factors in supply chains under uncertainty. This situation requires a direct and concise tool to monitor and control supply chain risks concurrently. Various qualitative and quantitative risk analysis tools are introduced to unveil the myth of uncertainty. A Bayesian Belief Network (BBN) is one of the risk modeling approaches that provides a systemic conditional probabilistic view on risk analysis. A Dynamic Bayesian Network (DBN) offers a solution that enables temporal factors in a BBN, which is in consonance with the characteristics of time-sensitive risks in supply chains. Inputs and outputs of DBNs are probability values. Supply chain practitioners may have difficulty to concretize the values into practical operations immediately because supply chain performance is measured with actual units of money or inventory. System Dynamics (SD) is a simulation tool for modeling complex socio-technical systems in feedbacks, stocks and their flows. However, SD has limitations in simulating conditional probabilities within the dynamic flows. By utilizing the essence of DBN and SD, this dissertation proposes a Dynamic Flow Bayesian Network (DFBN) to offer a comprehensive methodology for supply chain risk analysis. An Optimized Dynamic Flow Bayesian Network (ODFBN) method is developed with modifications based on the DFBNs by incorporating multi-objective optimization, multi-pricing strategy and Value-at-Risk. By applying the concept of Supply Chain Network Equilibrium, an Equilibrated Dynamic Flow Bayesian Network (EDFBN) method is developed to balance the needs of each stage and maximize the profitability of the entire supply chain. In this dissertation, mathematical integration of the models is presented and application to a supply chain case study inspired by the real-world is also conducted. Finally, a prototypical executable interface for industrial implementation is developed.Ph.D.Includes bibliographical referencesby Ke Su