Stochastic optimization of disruption-driven supply chain network design with a new resilience metric

Abstract

The supply chain (SC) ability to return quickly and effectively to its initial condition or even a more desirable state after a disruption is critically important, and is defined as SC resilience. Nevertheless, it has not been sufficiently quantified in the related literature. This study provides a new metric to quantify the SC resilience by using the stochastic programming. Our metric measures the expected value of the SC's cost increase due to a possible disruption event during its recovery period. Based on this measure, we propose a two-stage stochastic program for the supply chain network design under disruption events that optimizes location, allocation, inventory and order-size decisions. The stochastic program is formulated using quadratic conic optimization, and the sample average approximation (SAA) method is employed to handle the large number of disruption scenarios. A comprehensive computational study is carried out to highlight the applicability of the presented metric, the computational tractability of the stochastic program, and the performance of the SAA. Several key managerial and practical insights are gained based on the computational results. This new metric captures the time and cost of the SC's recovery after disruption events contrast to most of previous studies and main impacts of these two aspects on design decisions are highlighted. Further, it is shown computationally that the increase of SC's capacity is not a suitable strategy for designing resilient SCs in some business environments.