Network reconfiguration for supply chain risk mitigation

Risks and uncertainties around the globe have been evident as severe threats against reliable operations of supply chain networks. In order to survive the volatile conditions that characterize the current global environment, firms must prepare strategies that mitigate the effects of risks on their supply chain networks. This research proposes supply chain risk mitigation strategy based on supply chain network design under uncertainty and integrated advanced Information Technology (IT) architecture. The main objectives of this research include (1) understanding the impacts of risks on a supply chain network, (2) developing an approach to designing a robust supply chain network under risk, and (3) designing an information system architecture for a supply chain network under risk. In particular, we propose an architecture for network reconfiguration in a supply chain network operating under risk. Such a reconfigurable supply chain network has the ability to respond to risks as they arise and adjust its configuration accordingly to avoid catastrophic failure. To achieve a reconfigurable supply chain, three main components have been developed including an uncertain parameter database in which knowledge about the effects of risks is stored, a stochastic network optimization module that is responsible for determining long-term supply chain network configuration using a two-stage stochastic programming model, and a network configuration controller used to facilitate operations among supply chain nodes. Specifically, the information system of this reconfigurable supply chain network is based on the principle of Service-Oriented Architecture (SOA), such that heterogeneous nodes with diverse enterprise software platforms can communicate and integrate with each other seamlessly.;The first section focuses on obtaining knowledge about the effects of risk on a supply chain. This section proposes a Multi-Agent-base Simulation (MAS) model based on the well-known Gaia methodology and Unified Modeling Language (UML) standard and details it use in simulation experiments to gather knowledge about the network design parameters of the supply chain. In the second part, the knowledge obtained from the MAS model is used to formulate a two-stage stochastic programming model to redesign a supply chain network under risk in the stochastic network optimization module. The Sample Average Approximation (SAA) and L-Shaped decomposition methods are used to solve this supply chain network redesign problem efficiently. Additionally, three different sampling techniques are used to generate random numbers to the SAA method including Simple Random Sampling (SRS), Latin Hypercube Sampling (LHS), and Sobol' sequence technique. Results from numerical experiments showed that the LHS technique are the most efficient for solving small problems, whereas, Sobol sequence technique is more efficient for addressing moderate or large problems. In the third section, the network configuration obtained from the stochastic network optimization module is implemented on a reconfigurable supply chain network developed based on the SOA principle. The results from simulation experiments show that the network reconfiguration framework via the use of the SOA enables discoverability and interoperability in the supply chain and is capable of improving the supply chain performance in the presence of risk.