Tactical planning for dynamic technician routing and scheduling problems

Abstract

We consider the tactical planning aspect of a dynamic technician routing and scheduling problem with a time horizon of several days. In this problem study, some tasks are known beforehand, while others arrive dynamically. The tactical planning is to schedule the known tasks, such that we minimize the overall driving distance while ensuring short service times for the dynamic tasks. Without tactical planning, the known tasks would be spread throughout the whole area as they are scheduled based on a first-come-first-serve principle. In this study, we partition the area into disjoint slices covering the full plane and then assign the slices to individual work days. The partitioning of the area will be constructed with a balanced sweep algorithm using two approaches: The first is based on a parallel sweep-line following a given angle, while the second is based on a rotating sweep-line around a given origin. To guide the partitioning, we aim to minimize the degree of dynamism, which in our case is defined by the number of dynamic tasks, the distance from known tasks to dynamic tasks, and the tightness of the time windows. The tactical planning is evaluated using real-life and synthetic clustered data, where some of the tasks are known beforehand, and some are dynamic. We benchmark the various partitioning approaches against the default strategy of serving all known tasks on a first-come-first-serve basis. Computational results are reported showing around a 10% reduction in driving distance when using tactical planning. Furthermore, we show that up to 70% of the technicians can have a non-dynamic work day, without a significant increase in driving distance.