**5. System Optimization**

As mentioned before, EREVs have complex architectures that contain multidisciplinary technologies. Since the EREV performance can be affected by many multidisciplinary, interrelated factors, computer simulations are the most critical technology with which to optimize performance and reduction costs. Additionally, EREV simulations can help manufactures to minimize prototyping costs and rapidly evaluate their concepts. Since an EREV's system consists of various subsystems clustered together by mechanical, electrical, control, and thermal links, the simulation should be a parameterized mixed-signal one. Hence, optimization is at the system level, at which there are many tradeoffs among various subsystem criteria. The preferred system criteria generally involve numerous iterative processes. In summary, the system-level simulation and optimization of EREVs should consider the following key issues.


#### *5.1. Controller Optimization for Plant*

In an EREV, a plant could be an ICE, an electric motor, or a battery. Different strategies can optimize the plant and its controller: sequential, iterative, bi-level, and simultaneous strategies [123]. Sequential optimization often leads to non-optimal system designs due to plant/controller optimization coupling. Iterative plant/controller optimization strategies attempt to improve the initial design by first improving the plant design without compromising control performance and optimizing the controller design without compromising plant performance. In a bi-level plant/controller optimization strategy, two nested optimization loops are used. The outer loop optimizes the scalar-substituted objective function by changing only the plant's design. The role of the inner loop is to generate the optimal controller for each plant selected by the outer loop. The simultaneous strategy can be mathematically and computationally challenging for several reasons. The simultaneous plant/controller optimization problem is a hybrid static/variational problem. Even when the plant and controller optimization subproblems are convex, the collaborative problem is not guaranteed to be convex. Figure 14 shows the strategies for plant/controller optimization.

**Figure 14.** Strategies for plant/controller optimization.
