Abstract : The design of traction systems is a complex task, which needs experts from various fields. Train manufacturers can manage this complexity and produce high performance rolling stock materials. However, any improvement in design methodology can lead to a competitive advantage in a global market.
This thesis focuses on the optimal design methodology of complex systems such as a railway traction system. The design process and tools are demonstrated via two applications: the design of a traction motor and the concurrent design of several key components.
The load cycle and transient thermal behaviour are essential in the design of a traction motor. The adaptation of a motor to its load cycle reduces significantly its mass. The multidisciplinary design optimization approach is used to manage interactions between various discipline models in the optimization process. The optimization time can be reduced through a task distribution and a parallel computing. The multi-objective design optimizations are also applied. Pareto fronts are obtained despite the difficulty in using the high fidelity but expensive in computation time such as Finite Element Analysis model.
The hierarchical decomposition approach: the Target Cascading method is applied to the traction system design problem. The system and components are designed simultaneously. This method is suitable for implementing the complex system optimal design process while respecting the product development structure of the company.