*4.2. Results*

A disassembly strategy in this work consists of three decisions: (1) the optimal disassembly sequence, (2) the optimal circular economy strategy for each component, and (3) the optimal disassembly depth, which represents the stopping point of the disassembly process. These three decisions for both defined scenarios are shown in Figure 11. The used parameters for the initialization of the implemented genetic algorithm are listed Table 2. In the first scenario, the result is an incomplete disassembly with a disassembly depth of 90.63%. We see that the disassembly stops immediately after the modules are removed. In this case, two modules are reused, and two modules are recycled. This represents the best possible route due to the condition constraints. However, the remaining four modules are repurposed, although two modules could be remanufactured. This is because remanufacturing is not economically feasible in our use case. In the first scenario, an economic profit of EUR 553 can be achieved (see Figure 12). In the second scenario, it can be increased by 3.16% to EUR 570.5, although the disassembly costs are higher due to the complete disassembly. This is due to the fact that the upper and lower casing shells, as well as the cooling plates, are reused, and thus higher revenues can be realized. It is worth mentioning here that although the modules are the most valuable components on a battery, other parts contribute in a significant way to increasing the profitability of the disassembly. In the course of our research, we spoke with an EVB recycler in Germany. He said that recycling some EVB variants is particularly attractive because of massive busbars made of copper. Figure 12 shows the evolution of the objective function and the disassembly depth over the 100 generations. Thereby, the algorithm converges after few generations (<20), which means that the optimization can be terminated earlier and thus performed faster.

#### *4.3. Discussion*

Optimizing disassembly strategies for EVBs shows a key role in making the circularity of these systems more efficient by targeting higher priorities from the waste hierarchy presented in Figure 6. Here, selecting the optimal circular economy strategies at the component level must be considered as part of the disassembly process planning, since the chosen strategy influences the disassembly techniques and thus the disassembly times and costs.

Disassembly planning and optimization are becoming increasingly complex due to several factors: first, disassembly is subject to many uncertainties, which makes disassembly planning an adaptive and iterative process. Second, disassembly can be performed using different modes (sequential/parallel, complete/incomplete, destructive/non-destructive, automated/manual). Third, on the one hand, disassembly planning is a data-intensive process, and on the other hand, it must be ensured that disassembly is executed even when data are lacking. Fourth, there are multiple variables to optimize the defined objective function, such as the number of tool and direction changes and disassembly times, which depend on other factors such as the joining methods, the disassembly techniques, and the accessibility of the parts. Finally, disassembly strategies do not only consist of a disassembly sequence but also include other decisions, such as the disassembly depth and circular economy strategies for the different components. These three decisions are considered in our proposed adaptive disassembly planner with an integrated disassembly strategy optimizer.

**Table 2.** Used parameters for the genetic algorithm.


**Figure 11.** Optimal disassembly strategies consisting of three decisions (1—optimal disassembly sequence, 2—optimal disassembly depth, and 3—optimal circular economy strategies at the component level): (**a**) upper and lower housing shells and the cooling plates are in a bad condition and have to be recycled; (**b**) upper and lower housing shells and the cooling plates are in perfect condition and can be assigned to every circular economy strategy.

However, in the context of this paper, we only addressed sequential disassembly, since the aim of our current research is to develop an automated disassembly solution for battery packs down to the module level using a robot arm as a single manipulator. Our proposed disassembly strategy optimization method still needs to be extended to the following aspects: (1) including planning methods for cooperative disassembly, which can be applied by using at least two manipulators in fully automated disassembly solutions or by employing a human–machine collaboration, (2) taking into account the tool and direction changes, as they definitely influence the disassembly time, and (3) integrating adaptive methods for updating the condition and precedence constraints in case of complications during the disassembly process or when destructive disassembly steps are used. Furthermore, the adaptive planner should consider further factors, such as the configuration and the availability of the stations. On the one hand, the stations in a disassembly factory may have to be designed differently to be able to disassemble different battery variants and are therefore not suitable for carrying out all disassembly strategies and, on the other hand, a high-capacity utilization should be achieved. This means that EVBs have to be assigned to stations that cannot perform the best possible disassembly strategy in some cases. However,

this measure can significantly improve capacity utilization and consequently contribute to establishing highly automated and flexible disassembly factories in the near future, which will become more and more profitable with increasing return volumes. In the literature, there are no concepts for highly flexible disassembly factories for EVBs. In the following publications, we will present several future layouts for disassembly factories under consideration of the presented building blocks of an automated disassembly station in section 0 and show potential challenges for the adaptive disassembly planner with respect to the proposed layouts.

**Figure 12.** Model results: (**a**,**b**) optimal economic profit and disassembly depth for scenario 1; (**c**,**d**) same results for scenario 2.

Lastly, optimization of disassembly strategies, often described as disassembly sequence planning (DSP) in several literature sources, should be addressed in the product design phase. This will clearly contribute to achieving fully automated, cost-effective, and environmentally efficient disassembly for battery systems in the automotive sector. In particular, the modules, as the most valuable components in the battery, should be removable after only a few disassembly steps. This is obviously not the case for the battery considered in this paper.

#### **5. Conclusions**

An adaptive disassembly planner with an integrated disassembly strategy optimizer for electric vehicle batteries is presented in this paper. It serves to adaptively plan disassembly strategies and optimize them using heuristic optimization algorithms. A disassembly strategy consists of three decisions about the optimal disassembly sequence, disassembly depth, and circular economy strategy for each component. The disassembly strategy optimizer is implemented using a modified genetic algorithm and tested on a selected battery. Thereby, two condition scenarios were considered. In both scenarios, all modules

are removed. The disassembly of the remaining components depends on their subsequent route. The presented optimization method is computationally efficient and can be further improved by applying a convergence termination condition. The introduced disassembly planning method can be used at the end-of-life phase to plan the disassembly depending on components' state and market conditions. Furthermore, our approach is also suitable for use in the begin-of-life stage to ensure the guidelines of "design for disassembly" in the design stage. Nowadays, there is a need for action in both application cases because, first, disassembly processes are mainly carried out based on experience, and second, battery treatment at the end-of-life phase is hardly considered when designing these systems.

**Author Contributions:** Conceptualization, S.B.; methodology, S.B.; software, S.B. and F.P.R.; writing original draft preparation, S.B.; writing—review and editing, M.W., K.P.B. and F.P.R.; visualization, S.B.; project administration, M.W.; funding acquisition, M.W. and K.P.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** The authors wish to thank the Ministry of the Environment, Climate Protection and the Energy Sector Baden-Wuerttemberg for funding this work under the funding code L7520101 as part of the accompanying research of the project "DeMoBat". The financial support is gratefully acknowledged.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available within the article.

**Conflicts of Interest:** The authors declare no conflict of interest.
