Dynamic control for in-plant milk-run systems with volatile transport demand

最近，厂内生产供应的运输系统通常被组织为厂内送奶系统。这些系统在一次巡演中向不同的使用点提供不同的材料，频率很高。在最先进的计划程序中，牛奶运输路线和出发时间是长期计划和固定的，假设物质需求持续不变，这一点早已为人所知。然而，在运行中的厂内牛奶运行系统中，材料需求在短期内是不稳定的，需求高峰导致昂贵的异常处理过程。该研究项目的目的是开发一种工厂内送奶系统的控制概念，根据当前的运输顺序和送奶系统的现状动态确定路线、发车时间和资源分配。与现有的送奶系统计划程序不同，几个计划决定必须落实到控制逻辑中。相应的决策算法没有发表，将作为这项研究的一部分进行开发和评估。这些算法必须不断地确定哪些运输订单要与哪种资源、在哪条路线上以及在什么时间服务。这一概念的目的是即使在运输需求在短期内不稳定的系统中也能实现稳定的系统运行，从而避免昂贵的异常处理程序。总体而言，与静态规划方案相比，由于根据实际运输订单进行资源分配，预计所需资源将会减少。此外，根据需求的长期变化重新规划路线和时刻表已经过时。此外，还将对动态方法最有益的典型用例提出建议。

Recently, transportation systems for in-plant production supply are often organized as in-plant milk-run systems. These systems deliver different materials to various points of use in a single tour with high frequency. In state-of-the-art planning procedures milk-run routes and departure times are planned and fixed long-term assuming constant material demand known well ahead. However, in operational in-plant milk-run systems material demand is volatile in the short term and demand peaks result in expensive exception handling processes. The objective of the research project is to develop a control concept for in-plant milk-run systems, determining routes, departure times and resource allocation dynamically based on current transportation orders and the milk-run systems current state. In contrast to existing planning procedures for milk-run systems, several planning decisions have to be implemented into the control logic. Corresponding decision algorithms are not published and will be developed and evaluated as part of this research. These algorithms continually have to determine, which transportation orders to serve with which resource, on which route and at what time. The aim of this concept is to enable stable system operations even in systems where transportation demand is volatile in the short term and thus avoid costly exception handling procedures. Overall, a reduction of necessary resources is expected in comparison to the static planning scenario due to the resource allocation based on actual transportation orders. Additionally, re-planning of routes and timetables based on long-term changes in demand is obsolete. Furthermore recommendations on typical use cases where the dynamic approach is most beneficial will be derived.