(658c) Reactive Optimization of Supply Chain Networks Under Disruptions

The supply chains of high value-added chemicals have more variability and uncertainty compared to commodity products since there are typically more steps for their production and distribution. The demand profiles are also different as the customized product portfolios have individual demand uncertainty from alignment to single customers instead of following the demand of a larger market. This means the variability of the final demand depends on each individual customer, which leads to a higher uncertainty in prediction and assessment of operation plans. Many fine chemicals require several processing steps that are performed at different locations around the globe. This feature results in a significant increase of transportation needs, which in turn increases potential disruptions in the supply chain logistics and asset management.

Given the above features, supply chains for chemicals are vulnerable to disruptions. These disruptions can lead to significant economic losses, including increased costs to meet customer orders, penalties associated with delayed orders, or lost revenue due to failure in fulfilling existing orders. Furthermore, existing customer satisfaction can be significantly reduced when an order is delivered late or cancelled altogether. Late or cancelled orders diminish customer satisfaction and diminish the status of the supplier. Using optimization models to respond to unplanned events in the supply chain can reduce the economic and operational impacts of the unplanned events. For example, rerouting current inventories, increasing production using raw material in stock, buying finished products from competitors or third parties, and managing existing order deadlines are some of the decisions that can be made to mitigate the impact of the unplanned events.

Research in the design and operation of supply chains under the threat of disruptions is a relatively recent area. The literature on the subject has been growing, with contributions coming from various disciplines such as simulations and network analysis [1][2]. Similarly, work has been done using the framework of stochastic programming to account for disruptions in the design of supply chains [3]. Another approach that is widely being considered for supply chain management, is the guaranteed service model, which assumes stationary normal demand and no back orders to obtain an optimal inventory policy [4]. Although work regarding the operation of supply chains under disruptions has been done for specific network topologies, extensions to general networks is still an active field of research [5].

The novelty of our proposed work comes from mitigating the impact of unplanned events on multi-product supply chains by minimizing the operational cost once the disruption occurs using a novel optimization model. Specifically, this can be achieved with a proposed multi-period Mixed-Integer Linear Programming (MILP) model that considers information about the material shipments, the possible routes, the plant production, and the current order schedule to deliver the best possible reaction plan for operation. The output of the model is a new schedule that shows how to react to given disruptions, while considering economic and customer satisfaction goals.

This model is able to generate a schedule for each internal shipment and decide corresponding amounts of material and select the routes. In particular, the model considers the different routes and their costs, allowing re-routing or use of airfreight if necessary. The model can determine how to manage the inventory to address the disruption, and further notify if there is a need to overproduce or buy products from competitors. Furthermore, the model can alert the company if a delivery will not be completed on time, the new possible delivery date, the new prices that can be suggested to the client, or if it is better for the company to cancel the order. Also, the model can distinguish the different priorities that each customer has for the company to reallocate the resources along the network accordingly. The objective of the model is not only how to advise the company on how to react to a disruption, but also how to recover the original operation point once the unplanned event is fully addressed. Note that the model integrates all the echelons in the supply and production network to develop centralized schedules that account for the entire operation in a unified way. In that sense, decisions are made in a manner that the overall optimality of the system is achieved as a whole, rather than locally and empirically.

In this work we consider three different type of disruptions that will negatively impact the overall operation of the network. The first type of disruption is a shortage in raw materials, in which the suppliers cannot deliver the original amount that the company requested for its scheduled operation. The second type of disruption is the equipment failure within the plant. In both cases, the production is reduced, affecting the current plant operation, and potentially delaying the deliveries of the existing customer orders. Finally, the third type of disruption is a logistics resource limitation, such as shortage in the availability of truck drivers. This category of unplanned events will directly affect internal shipping capacity.

We present several case studies of multi-echelon, multi-product supply chains that demonstrate the capabilities of the proposed optimization model under different disruption scenarios. These examples further illustrate the computational efficiency of the proposed multi-period MILP model, and they show how the proposed optimal schedules outperform the ones obtained using empirical approaches. Furthermore, we discuss possible extensions of the proposed model using multi-stage stochastic programming, and model predictive control integrated with supply chain operations [7].

References

[1] Kim, C. O., Jun, J., Baek, J. K., Smith, R. L., & Kim, Y. D. (2005). Adaptive inventory control models for supply chain management. The International Journal of Advanced Manufacturing Technology, 26(9), 1184-1192.

[2] Schmitt, A. J., & Singh, M. (2009). Quantifying supply chain disruption risk using Monte Carlo and discrete-event simulation. In Proceedings of the 2009 winter simulation conference (WSC) (pp. 1237-1248). IEEE

[4] Garcia-Herreros, P., Wassick, J. M., & Grossmann, I. E. (2014). Design of resilient supply chains with risk of facility disruptions. Industrial & Engineering Chemistry Research, 53(44), 17240-17251.

[5] Graves, S. C., & Willems, S. P. (2000). Optimizing strategic safety stock placement in supply chains. Manufacturing & Service Operations Management, 2(1), 68-83.

[6] Snyder, L. V., & Shen, Z. J. M. (2019). Fundamentals of supply chain theory. John Wiley & Sons.

[7] Perea-Lopez, E., Ydstie, B. E., & Grossmann, I. E. (2003). A model predictive control strategy for supply chain optimization. Computers & Chemical Engineering, 27(8-9), 1201-1218.