Digital twins for full-field monitoring in vibration control applications

Much of the success of a space mission relies on the quality of the tests that have been performed. It is crucial to have optimal test campaigns that accurately reproduce the harsh vibrating environment components endure. It is also critical to preserve them from unnecessary damage before launch. Virtual models, or digital twins, of testing benches and tested components can assist the design of tests while reducing the risks, development time, and overall costs. This work focuses on the creation of a digital twin of the SEREME multi-axis vibration platform, and its potential use for MIMO random control. The target platform allows the testing of small components and is excited by three electromechanical shakers in the plane and one in the vertical direction. This dissertation summarizes a six-month collaboration with Siemens Digital Industries Software. Starting from an uncorrelated FE model of the platform, preliminary modal analyses are run using Simcenter Nastran. This allows the extraction of modal properties, frequency response functions (FRFs), and characteristic matrices. The latter is then used to apply model order reduction methods (MOR) once the eigenequation is solved again in a dedicated framework. In parallel, a testing campaign is run on the real platform, and experimental accelerance FRFs are extracted using TestLab. The FE model is then updated using frequency domain data through an output error approach. An optimization routine helps achieve a model that mimics the behavior of the platform in the frequency range of interest. Frequency-domain correlation methods are used to quantify the quality of the updated model. A state-space model is then created and exported into Amesim. The system is then connected to the models of four electromechanical shakers. Inconsistencies in the computed frequency response functions are tackled through a feedback loop. As the resulting model experiences instabilities, the whole mechanical part is assembled, including actuators. The assembly relies on frequency-based substructuring methods. Consequently, a Craig-Bampton reduction is applied to be able to satisfy compatibility conditions at interface points. Once stability is achieved, simulated and experimental acceleration to voltage FRFs are compared. The final model manages to capture much of the dynamics of the platform. Major differences are investigated to get insights into necessary improvements and guide the integration of a tested component.