Two-stage optimization for planning logistics service networks

物流服务网络的结构是影响物流服务质量和盈利能力的重要因素。这些服务的典型网络结构是轴辐式网络。在这里，可以通过选择枢纽位置来实现捆绑效应，从而提高网络的整体效率。在以前的项目中，用于优化轴辐式网络的基本数学模型已经逐步扩展，以考虑车辆依赖的成本以及出货量中的不确定性。然而，这些模型还没有考虑到消费者的行为。这方面可以通过客户根据各自的价格在标准运输和快递运输之间进行选择来表达。快递运输与货物交付时间的服务承诺相关联，这可以通过仓库地点之间的直接运输来实现。然而，应该指出的是，快递减少了枢纽地点之间的货运量，从而对捆绑效应产生了负面影响。因此，本研究项目的中心目标是开发和求解一个充分和真实地反映客户行为的物流服务网络规划的双层优化模型。该模型的逻辑决策顺序如下：首先，确定快递和标准运输的价格。根据此定价，根据最终的客户决策，得出每个关系和发货方法的预期发货量。对于这些量，计算出运输网络中的最优路径。双层优化的一个关键方面是同时优化价格和路线决策，预期客户因定价而做出的决策。这里的目标是使整体利润最大化。这个优化问题的建模将逐步扩展，以考虑例如车辆周转和不确定性(特别是需求波动)，以及更现实的运输成本建模。此外，还将定期评估开发的优化方法对现实世界需求的适用性，并将对所提出的方法进行原型实现。运输物流研究所在快递、快递和包裹行业领域的流程知识，离散优化主席在数学建模和优化方法方面的专门知识，以及他们的长期合作，将使拟议项目得以成功开发。这里的目的是，除了发展交通网络规划决策支持的实用方法外，还在高度热门的双层优化研究领域提供一般的数学和方法论进展。

The structure of a logistics service network represents a significant influencing and success factor for the quality and profitability of logistics services. A typical network structure for these services are hub-and-spoke networks. Here, bundling effects can be achieved through the selection of hub locations, which can increase the overall efficiency of the network. The basic mathematical models for optimizing hub-and-spoke networks have been extended step by step in previous projects to integrate vehicle-dependent costs as well as uncertainties in the shipment volumes. However, the consumers' behavior has not yet been taken into account in these models. This aspect can be expressed by the customer choice between standard and express shipping, depending on the respective price. Express shipping is associated with a service promise for the delivery time of the shipments, which can be realized through direct transports between the depot locations. It should be noted, however, that express shipments reduce the volume of shipments between hub locations and thus have a negative impact on the bundling effects. Therefore, the central objective of this research project is to develop and solve a bilevel optimization model for logistics service network planning that adequately and realistically represents customer behavior. The logical decision sequence in this model is as follows: First, the prices for express and standard shipping are determined. From this pricing, the expected shipment volumes for each relation and shipping method are derived, based on the resulting customer decision. For these volumes, an optimal routing in the transport network is calculated. A key aspect of bilevel optimization is that the price and routing decisions are optimized simultaneously, anticipating the customer decisions resulting from the pricing. The goal here is to maximize the overall profit. The modeling of this optimization problem will be extended step by step to consider, for example, vehicle turnarounds and uncertainties (especially demand fluctuations), as well as a more realistic modeling of transportation costs. Also, the applicability of the developed optimization methods to real-world needs will be evaluated periodically and the proposed methods will be implemented prototypically. The process knowledge in the field of courier, express and parcel industry of the Institute of Transport Logistics and the expertise in mathematical modeling and optimization methods of the Chair of Discrete Optimization, as well as their longstanding cooperation, will allow a successful development of the proposed project. The aim here, in addition to the development of practical methods for decision support in the planning of transport networks, is to provide general mathematical and methodological advances in the highly topical research area of bilevel optimization.