Systems Engineering, Virtual Fitting and Digital Twin in the design for assembly of tokamak machines

This research work presents methods and tools, complementary to the methodology currently followed in the ITER project, for the requirements management of large assemblies from their identification until the verification and validation phases. The main purpose is to improve their traceability along with the system lifecycle, identifying the key assembly characteristics from manufacturing to assembly. A requirements-based method is adopted to manage available knowledge, combined with the dimensional inspection activities. After having identified the key assembly requirements of the ITER toroidal magnet system, the main functional geometrical requirements, driving the alignment in their final position, are clearly identified. Then, the database for the collection of the full requirement set is clearly organized following the traceability rules and the model architecture is indeed structured and developed. Having defined the alignment strategy, the flow-down from assembly to subassembly level is performed and their impact is studied. Starting from geometrical data acquisition of the components, the as-built proposed tool enables to assess a set of assembly requirements verifying the impact of foreseen alignment strategies. Four virtual fitting pilot studies are presented to show how the use of the as built geometrical data are fundamental for the assessment of the assembly requirements in a defined alignment scenario. Through the virtual fitting it is possible to study the impact of the selected alignment on any other requirement with the aim to reduce the risk of misalignments verifying and validating the assembly strategy after manufacturing and before starting the assembly phase. For enabling the as-built based methods in the early design phase, when the as-built data are not available, a knowledge-based model for the generation of the variational geometry, called pseudo as-built, is developed. The first assembly model evaluates the toroidal gaps of the toroidal magnet system with the aim to test the feature instance generation on already manufactured components whose knowledge is high enough to introduce reasonable variation sources due to manufacturing and assembly. Robustness of the assembly model is tested and tuned with single and multiresponse verifications based on the desirability function for analysing the coherence of the results. This thesis work has been developed inside the Metrology and Reverse Engineering group at Fusion for Energy, Barcelona, Spain in close collaboration with the CREATE consortium.