Integrated tactical planning under uncertainty in the supply chain of divergent-type production systems

提出的研究方案探讨了面对不确定性的发散型生产系统中供应链战术规划的鲁棒性。供应链中的战术计划包括在包括供应商、生产单位和配送中心在内的网络中对采购、生产、分销和销售活动的综合计划。发散型生产系统的特点是一对多配方。这种生产系统通常面临不同类型的不确定性。传统的供应链规划工具是基于确定性优化模型的。在存在未来不确定事件的情况下，确定性方法提出的计划是不现实和不健壮的，并且与承诺的服务水平相比，通常会导致需求实现降低。采用稳健的规划工具，可显著改善服务水平、资源利用、采购及分销成本、环境影响等。本研究计划的主要目标是为存在不确定性的发散型生产系统的供应链开发高效稳健的战术规划模型，以便它们可以应用于工业。本研究的其他目标包括将所提出的鲁棒战术规划工具应用于1)软木供应链，其中我们面临原木特征的非同质性，以及2)在不确定条件下专注于回收报废产品的逆向供应链。采用随机规划和鲁棒优化方法对不确定条件下的供应链战术规划问题进行建模。本文还将提出基于精确和启发式方法的高效混合求解算法来求解所得到的模型。对于每个应用程序，我们的目标是分析和比较两种健壮的战术规划工具，包括所需的建模和计算工作、估计不确定性集的有效性、结果的质量和实施限制。虽然从经济和社会的角度来看，提出的强大的战术规划工具对加拿大工业很重要，但该研究计划也将推进随机优化的文献。

The proposed research program investigates the robustness in supply chain tactical planning in divergent type production systems faced with uncertainty. Tactical planning in a supply chain consists of integrated planning of procurement, production, distribution and sale activities in a network including suppliers, production units, and distribution centers. The divergent-type production systems are featured by the one-to-many recipes. Such production systems are faced usually with different types of uncertainties. The traditional supply chain planning tools are based on deterministic optimization models. The plans proposed by deterministic approaches are not realistic and robust in the presence of future uncertain events and will usually lead to decreased demand fulfillment comparing the promised service level. Adopting robust planning tools can lead to significant improvements in service level, resource utilization, procurement and distribution costs, environmental effects, etc. The main goal of this research program is to develop efficient robust tactical planning models for the supply chain of divergent-type production systems in the presence of uncertainties, such that they can be applied in industry. Other objectives of this research consist of application of the proposed robust tactical planning tool for 1) the softwood supply chain, where we are facing with the non-homogeneity in the characteristics of logs, and 2) in a reverse supply chain focused on the recovery of end-of-life products with uncertain conditions. Stochastic programming and robust optimization approaches will be adopted in order to model the supply chain tactical planning problem under uncertainty. Efficient hybrid solution algorithms based on exact and heuristic methods will also be proposed to solve the resulting models. For each application, we aim to analyze and compare the two robust tactical planning tools in terms of required modeling and computational effort, validity of estimated uncertainty sets, quality of results, and implementation limitations. Whilst the proposed robust tactical planning tools are important for Canadian industries from both economical and social points of view, this research program will also advance the literature of stochastic optimization.