The modelling of premixed combustion in the presence of a turbulent field in three-dimensional (3-D) confined scenarios was studied in this work, and applied to hydrogen combustion within ITER vacuum vessel (VV). Two different combustion approaches were tested with Large Eddy Simulation: a Flamelet Progress Variable (LES-FPV) and a Thickened Flame Model (LES-TFM). For the case of LES-TFM modelling, Dynamic Adaptive Chemistry (DAC) with a detailed kinetic mechanism for hydrogen combustion and in-situ adaptive tabulation (ISAT) methods were employed. Moreover, an adaptive meshing technique was used with the aim of tracking the flame front to ensure an adequate spatial resolution in this region. Experimental validation was performed to assess the ability of the different studied approaches to predict the flame burning speed, flame acceleration, and pressure evolution for lean H2-Air volume percent mixtures from 16 to 28 % propagating within a turbulent field. Results revealed that both approaches led to accurate predictions in terms of flame burning speed. When considering DAC and ISAT methods with detailed chemistry, LES-TFM model was found to be a cost-efficient solution. This model was used to analyse two loss of vacuum accident (LOVA) sequences within ITER VV. Results showed that turbulence might increase the flame burning speed by a factor of up to 3.5 for the case of big breaches (0.15 m2) but it would not affect in case of breaches of 0.02 m2. Besides, results showed that autoignition with 2 kg of H2 within the VV at 13.35 kPa might degenerate in detonation with average wall pressure levels ∼70 kPa.