The liquid resin infusion (LRI) process is continuously used for manufacturing large industrial parts across various sectors, yet it poses challenges due to its inherent variabilities. The MONOCLE research project, led by the Jules Verne French Institute and industrial partners, aims to develop a digital monitoring approach for complex parts during infusion. This initiative seeks to enhance decision-making during the process. Optimizing infusion necessitates simulations to save time and resources. To ensure reliability, numerical methods must be validated, particularly for thick and complex parts. Material characterization, especially permeability, is pivotal for accurate process modeling and defect prediction. This study establishes a physics-based model for thick part infusion, applying a numerical approach for permeability characterization validated by experiments via a monitoring system. Infusion experiments on 25 mm thick components were conducted, monitored by sensors. The software suite called PoroS developed at the GeM research institute was used to numerically predict the permeability of the material using the 3D fabric scans. The permeability tensor obtained using PoroS serves as one of the key inputs for the process simulations using PAM-RTM © software, validated against sensor data. The validated model facilitated numerical prediction for diverse scenarios, aiding decision-making for composite part monitoring. This integrated approach promises enhanced efficiency and reliability in LRI processes.