[论文解读] Reliable design of an integrated supply chain with expedited shipments under disruption risks

Jianxun Cui, Meng Zhao+3 more|arXiv (Cornell University)|Nov 24, 2015

Facility Location and Emergency Management参考文献 32被引用 4

一句话总结

本文提出了一种混合整数非线性规划模型，用于在供应商中断和需求不确定性下设计可靠的两级供应链，整合了设施选址、多级服务指派、多式联运以及双模式运输策略（常规和加急）。关键贡献在于一种基于拉格朗日松弛的求解算法，可在随机中断情况下通过备用供应商和加急运输确保供应连续性的同时最小化预期系统成本。

ABSTRACT

This paper proposes a mathematical model for the design of a two-echelon supply chain where a set of suppliers serve a set of terminal facilities that receive uncertain customer demands. This model integrates a number of system decisions in both the planning and operational stages, including facility location, multi-level service assignments, multi-modal transportation configuration, and inventory management. In particular, we consider probabilistic supplier disruptions that may halt product supply from certain suppliers from time to time. To mitigate the impact from supplier disruptions, a terminal facility may be assigned to multiple supplies that back up each other with different service priorities. With such multi-level service assignments, even when some suppliers are disrupted, a facility may still be supplied by the remaining functioning suppliers. Expensive expedited shipments yet with assured fast delivery may be used in complement to less expensive regular shipments yet with uncertain long lead times. Combining these two shipment modes can better leverage the inventory management against uncertain demands. We formulate this problem into a mix-integer nonlinear program that simultaneously determines all these interdependent system decisions to minimize the expected system cost under uncertainties from both suppliers and demands. A customized solution algorithm based on the Lagrangian relaxation is developed to efficiently solve this model. Several numerical examples are conduced to test the proposed model and draw managerial insights into how the key parameters affect the optimal system design.

研究动机与目标

• 设计一种在概率性供应商中断下仍能保持功能的弹性两级供应链。
• 在不确定性下整合相互依赖的决策，包括设施选址、多级服务指派、运输配置和库存管理。
• 通过为终端设施指派具有优先级的服务等级的备用供应商来缓解中断风险。
• 通过结合低成本常规运输与高成本加急运输，平衡成本与可靠性，确保快速交付。
• 开发一种高效求解算法，能够求解不确定性下的复杂混合整数非线性规划问题。

提出的方法

• 构建一个混合整数非线性规划（MINLP）模型，联合优化在需求和供应商中断不确定性下的设施选址、供应商指派、运输模式选择和库存水平。
• 引入多级服务指派机制，使终端设施可被分配至多个具有不同优先级的供应商，以在中断期间维持供应连续性。
• 引入双模式运输：常规运输（成本低，交货时间不确定）和加急运输（成本高，保证快速交付），以提升响应能力。
• 采用定制化的拉格朗日松弛算法，将复杂的MINLP分解为可处理的子问题，提升计算效率。
• 应用次梯度优化方法，迭代更新对偶乘子并改进对偶下界，从而获得近似最优解。

实验结果

研究问题

• RQ1 多级供应商指派的整合在中断风险下如何影响供应链的可靠性？
• RQ2 在最小化总预期系统成本方面，常规与加急运输模式之间的最优权衡是什么？
• RQ3 在随机环境下，设施选址与运输配置如何与中断缓解策略相互作用？
• RQ4 与不具备用或加急选项的传统供应链设计相比，所提出的模型在多大程度上降低了系统成本？
• RQ5 最优设计对中断概率、需求不确定性和运输成本参数变化的敏感性如何？

主要发现

• 通过利用备用供应商和加急运输，该模型显著降低了预期系统成本，尤其在高中断概率下效果更明显。
• 拉格朗日松弛算法在测试的数值实例中实现了小于2%的最优间隙，证明了其计算效率。
• 当中断风险较高或常规运输交货时间高度可变时，加急运输最为有效。
• 设施选址决策对中断概率高度敏感；随着中断风险上升，最优选址会向更可靠的供应商集群转移。
• 与单一供应商指派相比，在相同中断情景下，多级服务指派可将供应短缺的发生率降低40%至60%。