I. Introduction
The application of rechargeable lithium-ion batteries (LIBs) has been greatly extended in recent years thanks to the blooming of electric vehicles (EVs) [1]–[3]. At the same time, enormous efforts have been made to develop battery power management systems [4] and thermal management systems [5], [6]. Researchers need to provide an optimal cooling system for the batteries' long-lasting energy and high-power supply performances [6]–[8]. Developing a comprehensive and advanced cooling system capable of optimizing battery layout design with active system control is of great signif-icance. However, most commercial battery cooling systems are designed separately regarding battery layout and electrical management. As a result, they can only produce suboptimal results with low cooling efficiency [9]. Motivated by the imperfections of existing battery management systems, we are targeting to explore the potential active cooling systems for battery packs with the lowest cooling cost. Control co-design is a well-studied field in mechanical design that aims to enhance design performances by leveraging effective control strategies. Among various control algorithms to regulate the system performance [10], applying proportional control to the coolant flow rate can efficiently achieve active cooling for batteries based on their temperatures. Since we set the inlet coolant temperature as a constant, the coolant pump energy consumption determines the cooling cost. In EVs, cell-to-pack (CTP) has become a more and more popular method. It omits the cell module assembly, so the volume utilization rate is improved by 15%-20% [11]. However, CTP also requires batteries to be inserted into nonstandard containers so that the space can be more efficiently utilized. Recent advances in additive manufacturing [12]–[14] have enabled the customized adjustment of battery layout for each EV since they have different containers for batteries. But the custom design of battery layout still needs a tremendous amount of effort if using the traditional experimental or simulation methods. In this study, we developed a data-driven approach to optimizing the battery layout in a specific pack to lower the cooling cost while keeping all the battery cells' temperatures in the high-performance range.