*2.4. Related Literature*

For an overview of related literature on planning dynamic inductive charging infrastructure, we refer to Jang [22], Majhi et al. [23] and Yatnalkar and Narman [24]. They provide a comprehensive review of research articles and pilot projects. Most of these projects consider implementing a dynamic charging infrastructure for public bus systems. Many papers also consider battery capacity in the planning. Jang et al. [25] formulate a model in which the battery size is determined in addition to the placement of the charging infrastructure. The SOC is also considered in the model. The route is divided into segments that are either fully equipped with an ITU or not equipped at all. Ko and Jang [17], on the other hand, consider the route continuously. However, the model presented for this purpose is nonlinear. Both papers consider only one bus route at a time. In contrast, Hwang et al. [18] describe a multi-route environment. The model presented here has the special feature that the vehicles' capacities can be different. Liu and Song [26] consider stochastic

elements in their model formulation for planning a charging infrastructure. They first present a deterministic model and then a model with uncertain travel time and energy consumption. The reason why particularly public buses are considered is that the tours are known in advance. The papers often consider so-called closed environments. That means systems where the influence of traffic and other external factors is small. In this case, the tour's energy consumption and intake can be easily determined. Some papers consider other environments, e.g., Schwerdfeger et al. [27] looks at planning a charging infrastructure for highways.

Only Helber et al. [7] and Broihan et al. [8] considered the optimal placement of wireless charging infrastructures on airport aprons. Helber et al. [7] studied the characteristics of dynamic inductive charging on airport aprons and introduced the first mathematical optimization model for planning such an infrastructure. Broihan et al. [8] presented a reformulation and extension of this model considering multiple vehicle types and a service level restriction. In a numerical study, they analyzed test instances based on real airport aprons. For most of the instances, they could not prove optimality within a time limit of seven days with a standard solver. The resulting optimality gaps ranged from 8% to 15%. However, with the analyzed instances, they could not identify the reasons behind high computation times and the problem properties that lead to those. It is exactly this open question that we address in this paper.
