The fabrication of complex geometries with uniform material properties in electron beam powder bed fusion (PBF-EB) remains a major challenge without universally applicable solutions. Local material properties in PBF-EB are determined by the local consolidation and solidification conditions, which emerge from the local thermal conditions and the spatio-temporal melt pool evolution. The local thermal conditions in PBF-EB and the resulting melt pool geometry are governed by the combination of the cumulative heating effect and the energy input of the current hatch line. Cumulative heating results from the superposition of the temperature fields from adjacent hatch lines and leads to elevated temperatures at subsequent hatch lines. The magnitude of the cumulative heating effect at each position of the hatch is determined by the energy input of previous hatch lines and their respective energy loss until the current hatch line, which is very sensitive to the local return time of the beam. In previous research, compensation strategies were developed with the objective to achieve homogeneous thermal conditions independent of the local scan length in complex geometries. The build up of the cumulative heating effect at the beginning of a new hatch segment, without prior hatch lines to build of the cumulative heating, is however not considered in any of the developed compensation strategies, resulting in the emergence of transient regions with locally different thermal conditions at the beginning of each hatch segment. Over the course of the transient regions, the contribution of the cumulative heating effect to the melt pool formation increases with each hatch line until a quasi-stationary state is reached, and the cumulative heating has reached a constant level. Transient regions emerge not only at the start of each hatch segment, but at each position in complex geometries without prior preheating and result in non-uniform consolidation and solidification conditions. This contribution introduces a numerical optimization scheme with the objective to minimize the extent of transient regions and provide homogeneous thermal conditions for melt pool formation over complex geometries. For this purpose, thermal simulations are coupled with a derivative free optimization algorithm to determine an optimal scanning strategy for line-based PBF-EB which is able to compensate the build up of the cumulative heating effect in cuboid specimen.