Grain structures generated by laser powder bed fusion (L-PBF) of the Ni-based superalloy Inconel 718 (IN718) are studied using Cellular Automaton modelling. A new hybrid methodology is developed to benefit from a full thermohydraulic simulations that predicts the shape of the consolidated track and the temperature field. The advantage is to reach unprecedented sizes of simulation domains while still taking advantage of a full numerical solution of the L-PBF process at the scale of the melt pool. The computed grain structure becomes available in a Representative Elementary Volume (REV) of an Additively Manufactured (AMed) specimen and can be studied statistically considering the distribution of crystallographic orientations and the grain density. The other advantage is the possibility to vary process and material parameters to approach industrial practice. For the first time, the strategy consisting in opposite bidirectional scanning of the powder with an additional 67 • rotation at each new layer is made available. It is compared with the standard opposite bidirectional monoaxial scanning strategy and the opposite bidirectional orthogonal biaxial scanning strategies with two perpendicular axes, all reproducing trends reported in experimental literature. The generation of virtual REVs of AMed microstructures by L-PBF for IN718 opens the way to study new processing parameters and coupling with models for the prediction of metallurgical properties.