Milk-run systems are complex and dynamic. Several
types of materials and types of load units are provided
with a number of trains on a number of routes to various
points of use. Usually, the number of load units required
varies over time and is uncertain or known only a short
time in advance. Moreover, in most milk-run systems
driveways and resources, e.g. loading stations are used
by more than one train. This may result in dependencies
and blockages between individual trains, e.g. caused by
overtaking or stopping vehicles.
As milk-runs are most often used in just-in-time systems
with provision of small lot sizes in high frequency and
low buffers of inventory at the points of use, a reliable
and stable process is critical for success (Klug 2012).
Defects in the process, e.g. late delivery or delivery to
the wrong point of use may directly result in a
production standstill and therefore high excess cost.
Planners are challenged with the task of designing a
milk-run system where a defined service level of
provision is ensured and high efficiency (high capacity
and time utilization) is achieved at the same time