*5.2. Multidisciplinary Optimization*

It is necessary to capture the effects of each energy domain on the dynamics of the other domains to optimize a system. For example, when analyzing extended-range electric vehicles, the generator and internal combustion engine coupling should be optimized to provide the energy needed to charge the batteries and increase the vehicle's autonomy.

EREV power trains reduce design space for the remainder of the physical system and increase the complexity of the control. The coupling (dependency) among the parameters of the physical system (e.g., topology) and the control parameters transforms the problem into a multi-level problem, as depicted in Figure 14. If solved sequentially, it is by definition sub-optimal [124]. Therefore, the physical system and the control should be designed in an integrated manner to obtain an optimal system. Due to the oversized dimensions of the design space, computer simulations of dynamical systems have become more important as a preliminary step to building prototypes, e.g., for different architectures and component sizes. Computer simulations significantly speed up the control synthesis of a given design and topology. However, even with computer systems, finding the optimal vehicle design that provides the best control performance is typically intractable. It is not feasible (cost or time-wise), given design space, to build all possible vehicles and evaluate which configuration and parameters provide the best control performance.
