Thermo-Plastic Carbon Fibre Reinforced Polymer (TP-CFRP) materials are extensively researched for aerospace applications due to their unique properties. A key advantage of these materials is their ability to be remelted, offering benefits such as novel joining methods and recyclability. Unlike traditional bonding techniques using fasteners or adhesives, thermoplastic resin can be melted, resulting in a bond that, after cooling down, has properties similar to those of the original material. One technique for heating the material is induction, which generates an alternating electromagnetic field via a coil, causing the conductive carbon fibres to heat up. Induction welding is a contactless and rapid joining method; however, it is sensitive to numerous process and material parameters, making it challenging to control and predict its behaviour, particularly for Uni-Directional (UD) TP-CFRP. To overcome this, numerical modelling of the induction welding process is crucial for understanding and predicting the behaviour of TP-CFRP, ultimately optimising the manufacturing process. The induction heating methodology, established in previous work [1-3] for UD-TP-CFRP, has been applied in this study.This methodology includes a Machine Learning (ML) decoupled electro-magnetic and thermal analysis in Abaqus, accounting for dynamic coil movement. The methodology is then applied to the welding of a thick aerospace demonstrator comprising a skin and a L-stiffener. The primary challenge lies in achieving a high interface temperature without overheating other component areas [4]. In this work the induction heating modelling approach has been adapted to accommodate all relevant components of the weld setup (including heatsinks), enabling accurate temperature predictions. The developed numerical methodology can be employed to predict and optimise the induction welding manufacturing process, thereby reducing the need for trial-and-error approaches and facilitating the production of high-quality welds for complex UD TP-CFRP components on the first attempt.