This study aims to improve safety and decision support for complex periodic structures by developing an efficient and robust numerical modelling method. Finite element modelling of these structures raises challenges related to the size of the models. To remedy this, we exploit the periodicity assumption, which considerably reduces the complexity of the calculations. By focusing on a single basic cell, we can study the vibratory behaviour of the entire structure while keeping sufficient accuracy. Characterising vibration behaviour is essential for assessing the reliability of structures regarding different failure scenarios (e.g. fatigue, shock, etc.). Realistic numerical modelling coupled with probabilistic analysis can be used to quantify the probability of failure and identify critical zones. We compare the results obtained with Monte Carlo method to validate the robustness of our approach and to highlight the major gain in calculation time that it offers.