In the Netherlands, there is a lack of skilled-technicians and at the same time the electricity grid needs enormous extensions due to the energy transition. As a result, optimising workforce schedules for mechanics is extremely important and complex.
We helped one of the major Distribution System Operators (DSO) in the Netherlands by creating an optimisation algorithm that finds for each mechanic the best set of tasks to execute. The main challenges in finding an optimal workforce schedule are that there are many objectives and operational requirements to consider. Previously, the schedules were created by hand which resulted in serious inefficiencies and took a lot of time of the planners.

Balancing many different objectives
One of the challenges we faced was to find the right balance for all KPI’s of a schedule. The DSO has three different and conflicting objectives. First, they would like the mechanics to be as productive as possible. An obvious example is that tasks with a higher business value are prioritised. Another aspect of this KPI was to ensure that the teams of mechanics that worked on a project of multiple tasks remained stable.
Second, the happiness of the mechanics is also an important factor because a high attrition rate will worsen the shortage of mechanics. The mechanic’s satisfaction is measured with a diverse set of objectives such as reducing travel time and having the preferred division of tasks. Finally, the DSO would also need a stimulus for using mechanics as a tutor to train new mechanics. By giving a score to all the KPIs that are related to these three objectives, the DSO can determine what schedule strikes the best balance for all these objectives.
Finding the best schedule
Obviously, there are also many operational requirements for a schedule to be feasible. Some tasks require a certain skill that not all mechanics have, some tasks need to be done synchronously, while other tasks can only be done when another task is finished. On top of that, the working times and holidays of mechanics needs to respected. When creating a schedule by hand for fifty mechanics it can already be very hard to find a schedule that meets all these requirements, let alone to find the best schedule for these conflicting objectives. We captured all relevant aspects in an optimisation model, but the problem’s complexity was too large to be solved with out-of-the-box algorithms. We have developed a specialised algorithm that splits the problem in a smart way in smaller problems and in that way finds very good solutions in a limited amount of time.
Although this model was applied in the energy sector, the mathematics and logic behind it can be applied in any context in which staff need to be assigned to tasks. Nowadays, the trade-off between productivity and employee satisfaction is becoming more and more important and to strike the balance decision support models can be extremely valuable.