The evaluation of the dynamic behavior of unreinforced masonry vaults due to an earthquake is still a challenge because of their complex geometries and bond patterns, the choice of appropriate material properties and modelling strategies. In this study the LNEC-3D shaking table tests on full-scale masonry cross vault specimens being part of the SERA Project Blind Prediction are represented within a AEM software and assessed with a dynamic analysis. The aim of this post-prediction is the validation of the employed material characteristics and the modelling approach. The Applied Element Method is based on the mechanical interaction between rigid bodies and zero-thickness nonlinear interface springs. First, the unreinforced specimen and then the strengthened one with a textile reinforced mortar have been assessed. For the masonry material a simplified version of the elastoplastic fracture model for simulating the cyclic cumulative damage of masonry elements subjected to uniaxial compression is used; the hysteretic constitutive law that governs the cyclic response of shear springs is based on a Mohr-Coulomb yielding criterion, where cohesion is set to zero right after reaching the maximum shear strength. The results of the dynamic nonlinear analyses have shown great consistency with the experimental results, highlighting the importance of the selected brick pattern and material model.