Multiobjective evolutionary algorithms for strategic deployment of resources in operational units

Large-scale infrastructure networks require frequent maintenance, often performed by a team of skilled engineers spread over a large area. The set of tasks allocated to an engineer can have a huge impact on overall efficiency, whether that be in terms of time taken to complete all tasks, staffing costs or environmental costs in terms of emissions. When required to efficiently allocate a set of geographically distributed tasks to a maintenance engineering workforce, one approach is to define working areas for which teams of engineers are responsible. Often a key obstacle to overcome when looking for solutions is ensuring a balance between multiple competing objectives. In this paper, we employ a number of multiobjective evolutionary algorithms to analyse a simulation model for a real-world workforce optimisation problem used by BT. We provide a detailed analysis of the class of problems to be solved, where the workforce and a set of service distribution points must be split into smaller working areas, referred to as operational units. As the choice of how many operational units to split a larger working area into is critical, some of the practical considerations that must be made when addressing such problems are highlighted. This research has allowed the planning team at BT to understand the unique complexities of the nature of the problems they face in different areas of the UK, particularly with respect to the choice of number of operational units, and has strengthened their ability to design operational units effectively. (C) 2019 The Authors. Published by Elsevier B.V.