The highly non-convex AC optimal power flow problem is known to scale very poorly with respect to the number of lines and buses. To achieve improved computational speed and scalability, we apply a distributed optimization algorithm, the so-called Augmented Lagrangian based Alternating Direction Inexact Newton (ALADIN) method. However, the question of how to optimally partition a power grid for use in distributed optimization remains open in the literature. In the present paper, we compare four partitioning strategies using the standard IEEE 9, 14, 30, 39, 57, 118, 300 bus models as the performance benchmark. To this end, we apply the graph partitioners KaFFPa and METIS as well as two spectral clustering methods to the aforementioned bus models. For larger grids, KaFFPa yields the best results on average.