Multi-scale design optimization of electric vehicles by analytical target cascading: From battery cell level to marketing level

Abstract

The trend of vehicle electrification is rapidly growing and is likely to continue in the future. However, electric vehicles (EVs) with high vehicle performance, energy efficiency, and low cost are required to satisfy industry and customer needs. Design optimization is an important method that can significantly improve the performance of batteries and EVs. However, only a few studies integrate accurate full-order electrochemical models of lithiumion batteries into optimization frameworks of the battery module, pack, and EV design owing to complex modeling with huge computational costs. Multi-scale models with high efficiency and accuracy are required to provide optimal design, considering the overall system across scales. This study develops an efficient multi-scale design optimization framework for EVs, including four levels from the battery cell level to the marketing level. In the engineering domain, three levels of fundamental physics-based models are developed-an electrochemical cell, thermal module, and EV dynamics-to describe the performance of EVs. In the marketing domain, a utility model is used to maximize profit while considering customer preferences and costs. In addition, the vehicle model at the top level of the engineering domain is linked with the marketing model. Thus, all four model levels are hierarchically connected and can pass information. The multi-scale design optimization problem is solved by applying an analytical target cascading (ATC) method specialized for multi-level optimization. Through the ATC framework, optimization problems at each level can be independently solved while satisfying the targets from upper levels. Results show that each level successfully communicates with the others and carries out optimization to maximize profit while satisfying various constraints. ATC reduces the computational cost by 3% compared to the conventional all-in-one (AIO) approach. Parametric studies provide insights into how these design parameters, constraints, and operating conditions affect each other. To the best of our knowledge, this study is the first attempt toward developing multi-scale level modeling of EVs, which covers the marketing (very high business level) down to the cell (very detailed engineering level) while satisfying multi-level safety constraints.