Research in the Area of Adaptive Supply Chains in the ISE Department at RPI
In the ISE department at RPI, we are creating new frameworks for analyzing critical issues that will be faced by next-generation supply chains. Adaptive supply chains focus on the logistics of efficiently deploying finite resources to assemble, transport, sustain and distribute people and goods, thereby facilitating the fulfillment of demand associated with economic commerce, national defense, disaster response, and/or humanitarian aid. The current body of supply chain design and modeling research in this area focuses on life-cycle cost minimization under steady state conditions, sequential supply and demand management, and predictable asset and material values. This traditional approach is insufficient in modeling the challenges faced by current and next-generation supply chains where criteria related to risk management, information visibility, and resilience are emerging as critical issues.
Our research in this area has been funded by the National Science Foundation, Office of Naval Research, and industry. We are addressing problems in risk management in global supply chains, designing responsive seabased logistics systems, and supply chain restoration and resilience.
Risk Management in Global Supply Chains
Professor Xie's research group is developing theoretically rigorous and computationally efficient methodologies on dynamic risk management for global supply chains by using stochastic simulation. This research is motivated by our project on supply chain management collaborated with a biopharmaceutical company, Regeneron Pharmaceuticals, Inc., which is ranked as a Top 4 most innovative company worldwide by Forbes. In the biopharmaceutical industry, they must produce a product whose quality is reliable yet manufacture and supply it efficiently. There exist various challenges in these high-tech manufacturing industries including high-profit products, high uncertainty in the supply, production and demand, and rapid change in the technology and markets for end-products. Thus, an important question is how to construct reliable and cost-efficient supply chains and dynamically make operational decisions related to procurement, inventory, and production under various sources of uncertainty.
While global supply chains are impacted from many sources of uncertainty, modern data collection techniques, such as sensor technologies and bar codes, have resulted in the availability of rich data streams, which have the potential to provide detailed information on the status of various supply chain processes, including the real-time status of supply, production and customer demands. Therefore, the ability to efficiently extract relevant information from these data streams and incorporate it into real-time operational decision-making is critical for improving both supply chain reliability and cost efficiency.
For high-profit products, decision-makers are typically risk averse. In general, many challenges arise when trying to apply strictly analytical approaches to assess the risk performance of real complex supply chain systems. Since one of the most valuable features of stochastic simulation is its ability to characterize the risks inherent in complex stochastic systems, we use approaches in stochastic simulation to assess the system performance.
Therefore, based on the information obtained from real-world data and simulation experiments, we are developing a rigorous and efficient simulation-based prediction approach to support dynamic decision making for global supply chain risk management. The flow chart of our research is shown in Figure 1. At any time period t, given the historical data, we construct input models, which are the driving stochastic processes in the simulation experiments. These input models could include stochastic processes describing the supply, production and customer demand uncertainty. The underlying true input models are typically unknown and thus must be estimated using real-world data and/or expert opinion. The estimated input models are used to generate the samples for future time periods, e.g., future demands of the end-products. These samples are used to drive simulations in the experiments. Based on the simulation outputs, we could assess the system risk performance in the current and future time periods for a given decision policy. After that, we develop simulation-based dynamic optimization approaches to find the optimal policy and implement it. As the time evolves to next period t + 1 and new data arrives, we update our belief about the input models, and then repeat the procedure to find the optimal decision for next time period.
Our modeling framework is quite general and applicable to general stochastic systems, e.g., systems in manufacturing, finance, service industries and disaster response.
The Design of Responsive Seabased Logistics Systems
A military maritime logistics operation known as seabasing uses maritime platforms to transfers vital cargo stored on ships and rapidly delivers them ashore -- effectively conducting logistics operations from the sea. Strategically, seabasing provides flexibility in the ability to conduct a wide range of missions, including humanitarian aid distribution, crisis prevention, combat operations, and operational and tactical sustainment of military forces on the ground. Thus, the design of sea-based logistic delivery systems is critical to ensuring rapid transfer of these urgently needed deliveries. Sea-based logistics operate in a challenging environment and have unique mission characteristics. Thus, many logistics decisions are required to be made given imperfect visibility about the location, quantity, and expected delivery time of assets. These decisions include how should stowage, retrieval, loading, and unloading of supplies be conducted on ships to improve readiness and responsiveness?
ISE researchers are developing mathematical representations of the seabased logistics systems and processes in order to better understand, design, and operate them. This research includes (1) developing mathematical models that encompass the primary trade-offs in the system, (2) understanding structural properties and discovering solution approaches to solve the models, and (3) conducting computational experiments that use the developed models and approaches to provide policy recommendations and managerial insights. Specifically, researchers are developing descriptive models to characterize and to understand how and why cargo holds evolve from a highly organized state to a disorganized state. Given imperfect information about the location, quantity, and expected delivery requests, researchers are also developing prescriptive models to determine which items, and in what quantity, should be pre-staged on the flight deck. The developed models and algorithms are being used to quantify and evaluate military seabased logistic strategies for delivering emergent requests for tailored resupply packages in the presence of imperfect visibility. Ultimately, this research will lead to a better understanding of why seabased logistics operate in an uncertain environment, will quantity the impact on logistics performance of operating in a complex and uncertain environment, will analyze the trade-offs associated with different logistics system design and policy alternative, and will determine logistics strategies that support the transfer of vital sea-based resources to forces ashore that considers operating in an environment with imperfect visibility.
Supply Chain Restoration and Resilience
This research area focuses on mitigating the impact of disruptive events on supply chain operations through effective restoration strategies (i.e., how to recover from the disruptive event) and resilience strategies (for example, how to design a supply chain to minimize the impact of a disruptive event). Disruptions to commercial supply chains can have significant economic impacts; for example, Toyota had economic losses over $1 billion dollars and HP had losses over $700 million dollars due to the 2011 earthquake and tsunami in Japan. Although some economic loss from a disruptive event may be unavoidable, effective restoration efforts can limit the losses since demand within the supply chain will still occur during the disruptive event. Professor Sharkey is examining new models and algorithms for supply chain restoration efforts that specifically capture their unique aspects from traditional supply chain planning problems: the supply chain will be operational during the restoration efforts. Therefore, our models will be scheduling when and how the supply chain will change over time while still maintaining its operations.
Supply chain restoration and resilience is distinct from infrastructure resilience due to the lead times associated with production, the fact that customers may leave the system if their demand is unmet, and the different resilience strategies that can be implemented. Production lead times may mean that a disruption to a facility that produces a certain part for the final product of the supply chain may take some time to cascade up the chain to cause disruptions to the supply chain's ability to meet the demand for its final product. The impact of such a disruption could be mitigated by safety stocks throughout the supply chain; however, this strategy would increase cost. The fact that customers can potentially have their demand satisfied by another firm means that unmet demand after a disruptive event could have potentially devastating impacts to a supply chain since they could begin losing their customers. This issue could be mitigated by having emergency back-up contracts to fill the customers' unmet demand; again, though, this strategy would increase operational costs of the supply chain. Therefore, efforts to increase the resilience of the supply chain need to factor in the cost of such efforts and it becomes important to provide decision-makers with an understanding of the tradeoffs of supply chain resilience and the costs required to achieve such improvements. Professors Sharkey and Wallace are creating new supply chain resilience models in order to provide supply chain professionals with this understanding through collaborating directly with industry.