Resilience planning for Physical Internet enabled hyperconnected production-inventory-distribution systems

•PI-enabled resilient production-inventory-distribution planning is introduced.•A novel two-stage stochastic MILP model is developed.•A two-level heuristic is developed to solve the considered problem.•The performance of two-level heuristic, risk mitigation strategies are evaluated.•The behaviors of the PI in coping with various disruption risks are investigated. Based on the Internet of Things, standard coordination protocol, smart containers and other key foundations, the Physical Internet (PI) provides an interconnected, shared and adaptable logistics system, which has great potential to significantly increase the reliability and resilience of supply chains. In this paper, we assess the effectiveness of the PI for dealing with various disruptions in an integrated production-inventory-distribution system. To attain this, we constructed a two-stage stochastic programming model which captures the main characteristics of the PI and incorporates pre-event and post-event mitigation strategies in an integrated way. In the first stage (before disruptions), pre-event mitigation strategies are identified, while in addition to post-event mitigation strategies, the production, inventory, ordering and pickup and delivery route plans are determined in the second stage (after disruptions). Then a two-level heuristic algorithm that combines a three-phase heuristic and a fix-and-optimize procedure is developed to solve the model effectively. To evaluate the model, risk mitigation strategies and the two-level heuristic algorithm, extensive numerical experiments were conducted based on several realistic instances. The results demonstrate that the developed algorithm significantly outperforms the state-of-the-art solver in terms of solution quality and computational time. The results also indicate that tremendous cost and service level benefits can be achieved by adopting the proposed risk mitigation strategies. Compared with the traditional and horizontal collaborative systems, the PI-enabled system attains better performance in terms of the total cost and service levels. Finally, sensitivity analysis showed that the PI-enabled system has significant performance advantages when a range of parameter values were varied.