Logistics planning under uncertainty

当今社会的主要挑战之一在于如何管理资源的可持续利用，以及如何将资源转化和分配为对环境影响小、社会和经济影响大的商品和服务。价值链管理领域对许多公司的改进做出了重大贡献。这一发展的一个关键部分是基于公司核心业务系统中使用的决策支持系统(DSS)。新的信息和决策技术使分析性规划工具的这一发展成为可能，而业务研究对此至关重要。另一个重要贡献是信息和数据的可用性。今天，各种信息系统，包括公共和商业数据库、传感器和地理信息系统，产生了大量的信息，被称为“大数据”。然而，这些数据中有很大一部分是关于过去的活动，而规划则需要关于未来的数据。例如，由于价格、需求和行驶时间的不确定性，DSS所需的许多数据都不是确切知道的。尽管如此，大多数(如果不是全部)物流决策支持系统都是基于确定性和解耦的规划模型。这一缺陷给后勤业务带来了许多规划和业务挑战，因为必须实施某些预期和应对措施，以处理不确定性。虽然有一些方法可用于不确定情况下的规划，但它们通常不适用于实际的物流规划。这些措施包括稳健优化、随机规划以及金融和运营对冲。在这项提案中，我们的目标是在考虑不确定性的情况下，为价值链物流开发有效的方法。结果将导致更强大和更稳定的解决方案，减少需要重新规划的需要，并减少陷入需要昂贵的纠正行动的困难境地的风险，以及更少的紧急管理，即只在紧急问题出现时解决它们。所有这些都导致了成本的降低和对环境的负面影响，增加了灵活性和公平性，最后但并非最不重要的是，我们集体资源的可持续利用。

One of the key challenges in today's society resides in managing a sustainable usage of the resources as well as of their transformation and distribution into goods and services with both low environmental impact and high positive social and economic impacts. The area of value chain management has had a major contribution for improvements at many companies. A critical part of this development is based on Decision Support System (DSSs) that are used in companies' core business systems. This development of analytical planning tools has been made possible by new information and decision-making technologies, for which operational research is critical. Another important contribution is the availability of information and data. Today, a massive amount of information, termed "big data", is generated by various information systems including public and commercial databases, sensors and geographical information systems. However, much of this data is about past activities, while planning requires data about the future. Much of the data needed by DSSs is not known exactly due to, for example, price, demand, and driving time uncertainties. Despite that, most (if not all) DSSs for logistics are based on deterministic and decoupled planning models. This drawback creates number of planning and operational challenges for logistics operations as certain anticipation and counter measures must be implemented in order to handle uncertainty. Although there are approaches available for planning under uncertainty, they are generally not adapted for practical logistic planning. These include robust optimization, stochastic programming and financial and operational hedging. In this proposal, we aim to develop efficient approaches for value chain logistics taking into account uncertainty. The results will lead to more robust and stable solutions with less need of re-planning, and have reduced risk to be stuck with a difficult position which requires expensive corrective actions, and less emergency management i.e. only solve urgent problems as they appear. All this lead to reduced costs and negative environmental impact, increased flexibility and fairness, and last but not least sustainable use of our collective resources.