Events

In supply chain risk management, the critical yet challenging problem of resilience investment, i.e., where to invest limited resources in the supply chain network to best mitigate disruptions, is readily addressed through the ever-increasing rich data nowadays. Collaborating with a globally leading supply chain management company, we characterize optimal strategies of resilience investment in generic tiered supply chain networks, in which capacitated nodes are prone to independent stochastic disruptions. First, we analytically examine investment strategies under extreme disruption probabilities. Managers should invest on nodes with highest capacities under rare disruptions, while on entire paths under frequent disruptions, for the purpose of maximizing the total material flows through the supply chain network. To further facilitate practical decision making, we characterize the investment function for nonreroutable ow as monotone supermodular and propose a greedy algorithm with guaranteed performance. Next, with a realistic distribution network and operational data from the company in collaboration, we adopt data-driven prescriptive analytics to generalize analytical insights to mid-level disruption probabilities under both nonreroutable and reroutable mechanisms.
We contribute to supply chain risk management theory and practice that, (1) the nodal resilience investment should not only consider individual characteristics such as node capacity, but also incorporate a path perspective that entails the node’s average path length and ow centrality, and (2) the investment focus between node and path is contingent on disruption probability and routing mechanism.