Sähkömekaanisten voimansiirtolinjojen simulointi: Systeemitason suorituskyky ja komponenttitason kunnonvalvonta

Electromechanical powertrains play a crucial role in modern transportation and industrial systems, driven by the increasing trends of electrification and sustainability. Computational simulation models serve as powerful tools for improving the design and analysis of such powertrains under various operating conditions. However, the development of effective simulation models involves balancing accuracy and computational efficiency, guided by practical constraints like experimental data-related challenges. In this thesis, I introduce scalable and adaptable simulation-driven methods for electromechanical powertrains, addressing two complementary challenges: system-level performance optimisation of electrified heavy-duty vehicles and component-level simulation condition monitoring (CM) of critical rotating machinery. Physics-based approaches are adopted throughout, combining validated dynamic models with realistic operational scenarios to generate high-quality datasets for both powertrain configuration design and diagnostic purposes. At the system level, I extended the Smart eFleet (SeF) toolbox to evaluate the performance of an eretrofitted heavy-duty truck. I prepared simulation modules for three distinct configurations across ten realistic driving cycles, created based on GPS data. The results demonstrated that retaining a multi-speed transmission significantly improves traction and gradeability under heavy-load conditions, with marginal effects on energy efficiency. At the component level, I adapted a multibody simulation (MBS) framework employing the surface geometry-based polygonal contact method (PCM) for rolling bearings and bevel gears. PCM parameters were calibrated against well-established Hertzian-based tools for healthy conditions, after which faulty geometries were applied. To prioritise signals that are less prone to external disturbances and potential for future CM, I acquired relative shaft displacements from the bearings and torsional vibrations from the gears. The simulated responses were carefully validated against experimental measurements. The results confirmed that PCM-based MBS models can replicate realistic fault responses for data-driven diagnostics purposes, including scenarios that are difficult or impossible to reproduce experimentally. The generated bearing dataset was then applied to train and test a machine learning-based fault classifier. The classification results exhibited a successful differentiation of bearing faults, whether located in the outer ring or inner ring, from healthy counterparts. This thesis delivers a scalable framework for system-level powertrain performance assessment and a validated methodology for integrating high-fidelity MBS contact models into component-level CM. Together, these advances improve performance optimisation, enable realistic fault simulation, and strengthen predictive maintenance capabilities in electromechanical powertrains.