As a researcher focused on sustainable supply chains, my work primarily addresses the inefficiencies and strategic challenges embedded in the end-of-life management of traction battery packs. The rapid proliferation of electric vehicles has placed an unprecedented spotlight on how we manage the lifecycle of the traction battery pack, moving beyond the initial manufacturing phase to consider the substantial environmental and economic implications of its retirement. My research integrates two distinct yet complementary levers, namely policy mechanisms (with a focus on government reward-punishment systems) and technological enablement (specifically blockchain technology). This study is intended to optimize recycling decisions across different stakeholders, including manufacturers, consumers, and regulatory bodies, offering a comprehensive and quantifiable framework for improving the entire recycling ecosystem of the traction battery pack.
In Chapter 3 of my recent work, I delved into the realm of evolutionary game theory to model the strategic dynamics between governments, vehicle manufacturers, and consumers under both static and dynamic reward-punishment frameworks. This involved constructing a tripartite evolutionary game model where each stakeholder has distinct strategic options. The government can adopt either active or passive supervision, the vehicle manufacturer can choose an active or passive recycling strategy, and consumers can decide whether to return their used traction battery pack through formal or informal channels. The different payoff matrices stemming from these interactions form the foundation of my analysis, which seeks to uncover the trajectory of strategy evolution under state-controlled incentives. My exploration here builds upon the understanding that simply implementing a recycling policy is insufficient, as the extent of its effectiveness hinges on the complex interplay of the participants’ perceptions, reaction to policies, and their behavioral evolution, particularly given the various costs and benefits associated with the traction battery pack.
The formal starting point in my framework involves the formulation of an evolutionary game matrix, defining the notation used in the model. These parameters were carefully selected to mirror the nuances of traction battery pack management, including recycling technology levels, environmental preferences, and hidden benefits. A description of the key parameters is given in Table 1.
| Parameter | Description |
|---|---|
| Cm | Total cost of active recycling for the manufacturer |
| Cr | Setup cost related to recycling service stations |
| Ci | Technology investment cost for recycling |
| Ct | Transportation cost for collecting the traction battery pack |
| Im | Hidden benefits from active recycling, such as brand image and reputation |
| Imc | Maximum benefits from the sufficient recycling of the traction battery pack |
| α1, α2 | Active and passive recycling technology levels of the manufacturer |
| I1c, I2c | Consumer gains via formal and informal channels respectively |
| C1c, C2c | Consumer cost via formal and informal channels respectively |
| μ | Consumers’ environmental preference coefficient |
| L | Maximum environmental benefit obtained from choosing formal recycling channels for traction battery packs |
| Cg | Active supervision cost of the government |
| Rm | Reward from the government to the manufacturer for active recycling |
| Fm | Fine from the government for passive recycling of traction battery packs |
| β | The reduction ratio of consumer benefits from informal channels, implying crackdowns by active government supervision |
| θ | Profit share offered from reward given to the manufacturer, reaching the consumer |
| Re | Environmental gains achieved for the government |
| C1e, C2e | Environmental governance costs when the manufacturer chooses passive recycling but consumer selects the formal or informal channels |
Based on these definitions and considering the strategic choices (x, y, z) representing the probabilities of the manufacturer actively recycling, consumers using formal channels, and the government actively supervising, respectively, I developed the payoff matrix, as illustrated in Table 2. This matrix systematically outlines the accrued benefits and incurred costs for the three parties under the combined scenarios. For instance, when the manufacturer chooses active recycling while the consumer selects formal channels and the government adopts an active stance, the manufacturer’s payoff includes both direct profit from recovered materials, benefits from carbon reductions, and government rewards, minus the associated costs of recycling operations.
| Manufacturer | Consumer | Active Government Supervisor (prob z) | Passive Government Supervision (prob 1-z) |
|---|---|---|---|
| Active recycling (x) | Formal channel (y) | α1Imc + Im + Rm – Cr – Ci – Ct – θRm | α1Imc + Im – Cr – Ci – Ct |
| I1c + μL + θRm – C1c | I1c + μL – C1c | ||
| Informal channel (1-y) | Im + Rm – Cr – Ci – Ct | Im – Cr – Ci – Ct | |
| I2c – βI2c – C2c | I2c – C2c | ||
| Passive recycling (1-x) | Formal channel (y) | α2Imc – Fm | α2Imc |
| I1c – C1c | I1c – C1c | ||
| Informal channel (1-y) | – Fm | 0 | |
| I2c – βI2c – C2c | I2c – C2c |
From the payoff matrix, I derived the replicator dynamics equations which describe the evolutionary process of each strategy within the population. The replicator dynamic equation for the manufacturer selecting the active recycling strategy is given as follows (Eq. 1):
$$ F(x) = \frac{dx}{dt} = x(1-x)\left[ \alpha_1 I_{mc} – \alpha_2 I_{mc} + I_m – C_{r} – C_{i} – C_{t} + zR_m – F_m z + \theta R_m z y \right] $$
Similarly, the replicator dynamic equations for the consumers and the government are represented by Eq. 2 and Eq. 3, respectively.
$$ F(y) = \frac{dy}{dt} = y(1-y) \left[ I_{1c} – C_{1c} – I_{2c} + C_{2c} + \beta I_{2c} z + \mu L x – \theta R_m x z \right] $$
$$ F(z) = \frac{dz}{dt} = z(1-z) \left[ -C_g + F_m (1-x) – R_m x \right] $$
The equilibria of the system are determined by solving F(x) = 0, F(y) = 0, and F(z) = 0. Out of the various possible solutions, I focused on a set of eight pure strategy equilibria, which are essential for determining the evolutionary stability of the system. The Jacobian matrix of the system can be formulated, and the stability of each potential equilibrium point is assessed by evaluating the signs of the eigenvalues of the Jacobian matrix. The results from this stability analysis reveal several distinct system states. One specific state, \(E_5 = (1,1,0)\), which corresponds to {active recycling by manufacturer, formal channel by consumers, passive supervision by the government}, is identified as an ideal state. This is the desired equilibrium, whereby the market achieves a high-level of recycling efficiency as manufacturers are engaged proactively and consumers are participating appropriately, requiring less direct interference from the government.
Through this detailed equilibrium analysis, I established three potential realized scenarios. In the first scenario, where the cost of active supervision by the government \(C_g\) exceeds the fines collected \(F_m\), the rational response for the government is to remain passive, which eventually leads the system to a stable but least-optimal equilibrium state \(E_1 = (0,0,0)\). In the second scenario, the punishment surpassing the cost of active governmental supervision acts as a forcing factor. If the consumer’s net gain from informal channels is superior to that from formal channels under this governmental enforcement \(I_{2c}(1-\beta)-C_{2c} > I_{1c}-C_{1c}\), the trajectory pushes the system towards \(E_4 = (1,0,0)\), an undesirable outcome where only the manufacturer is active, indicating the policies were not enough to influence consumer behavior. In a third scenario, high consumer environmental preferences \(\mu > (C_{2c} + I_{1c} – C_{1c} – I_{2c})/L\) encourage consumer participation through formal channels. Then, if the manufacturer’s net gain from recycling a traction battery pack is sufficient to cover the cost difference when actively engaged, the system evolves towards the ideal equilibrium \(E_5 = (1,1,0)\).
My analysis of this static system, which assumes that government penalties and rewards are fixed, further highlighted the threshold effects of government interventions. For example, when I increased the level of fines \(F_m\) from 1.0 to 3.5 in a simulation, the system transitioned from the worst-case equilibrium E1 to the ideal state E5. The simulations revealed that the fine has a critical threshold point between 2.0 and 3.0 below which it fails to incentivize active manufacturer participation. While penalties were effective, their usefulness was limited beyond that threshold, and changes in rewards \(R_m\) proved to have a very limited impact on manufacturer decisions, merely slowing down the pace of degeneration to a passive strategy but failing to change the final outcome. Furthermore, the manufacturer’s technological capability \(\alpha_1\) revealed itself as a key internal driver, with improvements in recycling technology drastically boosting the system’s probability of reaching the ideal state. The environmental preference of consumers \(\mu\) was also crucial, demonstrating that higher concern for the environment directly drives formal recycling behaviors.
Based on the realistic operational context, I extended my research to investigate a more adaptive counterpart: a dynamic reward-punishment mechanism. Instead of fixed values, these mechanisms adjust the penalties based on the government’s supervision probability (zvF_m^d =), proportional to regulatory actions, and adjust the reward based on the active recycling probability of the manufacturer (xRR_m^d =). Through re-analyzing the model with a dynamic penalty coefficient v, I discovered significant improvements in system stability, as illustrated by the numerical simulations. The dynamic mechanism substantially accelerates the system’s convergence to the desired equilibrium. In quantitative terms, the convergence time was markedly reduced, decreasing by about 60% compared to the static case. This highlights the critical advantage of implementing adaptive policies over static ones. The policy’s inherent flexibility to punish high violations with harsher penalties and reward high achievement with more considerable benefits produces more stable, efficient incentives within the market, facilitating a more sustainable model for suppressing informal recycling operations of traction battery packs and encouraging engagement.
Shifting focus from policy-driven behavioral evolution to the structural and operational configuration of the reverse supply chain, I next considered the decision-making processes among different recycling channel structures. This analysis scrutinizes how the traction battery pack recycling decisions are impacted by different economic and operational models. In my fourth chapter, I established a foundational model that presents the closed-loop supply chain setup without blockchain technology, comprising a traction battery pack manufacturer (leader), an electric vehicle manufacturer (follower), and consumers. Within this framework, I defined the total demand \(D\) as a function of retail price \(p\), the market potential, and, importantly, the positive impact of recycling efforts \(g\): \(D = \phi(1 + g) – \theta p\). On the recovery side, the volume recycled through channel \(j\) by entity \(i\) is represented by the linear function \(Q_{r} = a + b r_{j}^{i} – \delta r_{k}^{i} \), where \(a\) is the base quantity from voluntary consumer returns, \(b\) is the consumer sensitivity to the recycling price, and \(\delta\) is the competition intensity between different channels. Here, the vehicle manufacturer benefits from lower collection costs because they can exploit existing distribution networks, leading to \(c_{rV} < c_{rB}\). The net unit profit from a recycled traction battery pack is \(r = o_c – c + r_b\), representing the cost differential of using secondary materials and the value derived from second-life applications. A complete summary of these model parameters is outlined in Table 3.
| Symbol | Definition |
|---|---|
| \(\phi\) | Potential market demand |
| \(w\) | Wholesale price of traction battery packs |
| \(p\) | Retail price of electric vehicles |
| \(g\) | Recycling demand gain coefficient |
| \(\theta\) | Consumer price sensitivity coefficient |
| \(o_c\), \(c\) | Unit costs using original and recycled materials, respectively |
| \(r_b\) | Benefits from echelon or cascade utilization of a traction battery pack |
| \(a\) | Voluntary return quantity from consumers |
| \(b\) | Consumer sensitivity to recycling price of traction battery packs |
| \(r_{j}^{i}\) | Recycling price set by recycler \(j\) under model \(i\) |
| \(r\) | Net profit from recycling a unit traction battery pack |
| \(c_{r}\) | Fixed cost to recycle a unit of traction battery pack |
| \(\delta\) | Competition intensity coefficient among channels |
In this scenario without blockchain technology, I meticulously built and compared four distinct recycling models: Manufacturer Recycling (Model NB), Vehicle Manufacturer Recycling (Model NV), Hybrid Recycling from both manufacturer types (Model NBV), and an Alliance Recycling model (Model N(B+V)). Through a Stackelberg game where the traction battery pack manufacturer holds the leadership position, I derived the optimal equilibrium decisions for the wholesale price, retail price, and recycling prices under each of these distinct models.
For instance, in Model NB, where the traction battery pack manufacturer delegates the recycling responsibility directly to itself, the maximization problem is constructed to find optimal decisions through backward induction. The optimization problem faced by each player is described by the profit functions \(\Pi_B^{N_B}\) and \(\Pi_V^{N_B}\). By applying the necessary conditions, such as setting the first-order derivatives to zero, I derived optimal values. I obtained the optimal wholesale price \(w^{N_B}\), optimal retail price \(p^{N_B}\), and optimal recycling price \(r_{B}^{N_B}\). These equilibrium values were found to be sensitive to the value of recycling \(r\), the base recycling \(a\), and various cost coefficients.
$$ w^{N_B} = \frac{c + g(1-\phi) – \theta\phi}{2\theta} $$
$$ p^{N_B} = \frac{3c + 3g(1-\phi) – \theta\phi}{4\theta} $$
$$ r_{B}^{N_B} = \frac{a + b r + \epsilon E_r – b c_{rB}}{2b} $$
Similarly, for the vehicle manufacturer recycling model (NV), the vehicle manufacturer becomes the entity setting the consumer-facing recycling price based on the transfer price determined by the battery manufacturer. Stackelberg game analysis yields the optimal value for the recycling price \(r_{V}^{N_V}\) as:
$$ r_{V}^{N_V} = \frac{a + 3b r – b c_{rV} – 3b c_{rv} – b\epsilon E_r}{4b} $$
In the hybrid model (NBV), both the traction battery pack manufacturer and the vehicle manufacturer compete in the recycling market, leading to differentiated recycling solutions for each channel, denoted as \(r_{B}^{N_{BV}}\) and \(r_{V}^{N_{BV}}\). In contrast, the alliance model (N(B+V)) is solved by treating both enterprises as one integrated entity, maximizing the combined profit. This centralization eliminates double marginalization from the collaborative effort, resulting in the highest total supply chain profitability, while also allowing the alliance to lower the retail price \(p^{N_{(B+V)}}\) compared to the other structures. The numerical case study on these models yielded the outcomes that can be seen in Table 4.
| Metric | NB | NV | NBV | N(B+V) |
|---|---|---|---|---|
| Supply chain profit | 4,933,474,747 | 4,859,473,348 | 5,000,859,104 | 6,540,679,996 |
| Sales volume | 49,700 | 49,700 | 49,700 | 99,400 |
| Recycling volume | 18,228 | 9,150 | 19,225 | 20,050 |
| Recycling rate | 36.68% | 18.41% | 38.68% | 20.17% |
The numerical analysis presented in Table 4 reveals important structural insights. The alliance model, by functioning as a centralized entity, yields the highest supply chain profit by eliminating double marginalization and integrating resources. Conversely, the hybrid recycling model, which facilitates competition between two independent recyclers, resulted in the highest recycling rate for traction battery packs, attributed to price and service competition expanding the collection volume. This is a classic conflict in reverse supply chain design. The vehicle manufacturer-only model consistently performed the weakest across all performance indicators in the benchmark scenario. Interestingly, while the alliance model maximizes profits and sales, its single-channel alliance recycling yields lower recycling rates than the hybrid model, showing that strong profit outcomes do not necessarily equal maximum collection efficiency when channel competition is suppressed.
In my fifth chapter, I introduced the critical layer of digital trust into this baseline. Blockchain technology enables transparency, traceability, and immutability in the recycling process. The demand function in the forward chain evolves to \(D = \phi(1+g) + k\lambda – \theta p\). Here, \(\lambda\) represents the level of blockchain technology adoption and \(k\) is the trust coefficient reflecting consumers’ sensitivity to the technology. The quantity of returned traction battery packs is correspondingly enhanced by the blockchain effect as \(Q_{r} = a + b r_{j}^{i} – \delta r_{k}^{i} + k \lambda \). Investment in this technology entails a significant cost, structured in my model as a quadratic function \(C(\lambda) = \frac{1}{2} A\lambda^2\) , where \(A\) is the cost coefficient. This cost is shared between the manufacturer and the vehicle manufacturer with proportions \(t\) and \(1-t\), respectively, reflecting their collaboration in technological adoption.
With this extended setup, I re-evaluated the equilibrium strategies for the four analogous recycling models under blockchain enablement, titled Model YB, Model YV, Model YBV, and Model Y(B+V). The manufacturers’ optimization problems now include the blockchain investment terms. For example, under the alliance recycling model with blockchain, denoted as Y(B+V), the joint optimization objective is shown in Eq. 4, and it considers the additional demand gained through the trust-building function of blockchain.
$$ \Pi_{BV}^{Y_{(B+V)}} = (p – c – g + \phi)( \phi(1+g)+ k\lambda – \theta p) + (r – c_{r(B+V)})( a + b r_{BV} + k\lambda) – \frac{1}{2} A\lambda^2 $$
Solving the Stackelberg game for the various modes, I derived the optimal decisions, which include the level of technology investment \(\lambda\). For instance, the optimal blockchain investment level in the alliance-recycling mode is expressed by a mathematically complex function \(\lambda^{Y_{(B+V)}}\), which is dependent on the trust coefficient, consumer sensitivities, costs, and marginal value from recycling. The complexity of these equilibrium solutions shows how deeply intertwined technological investment, pricing decisions, and market response become under such systems.
A comparative numerical evaluation between the no-blockchain and blockchain-enabled scenarios produced results as seen in Table 5. I calculated the profit and recycling performance metrics for various recycling models under both technology contexts.
| Recycling Model | Supply Chain Profit (No Blockchain) | Supply Chain Profit (With Blockchain) | Recycling Rate (No Blockchain) | Recycling Rate (With Blockchain) |
|---|---|---|---|---|
| Manufacturer (B) | 4,933,474,747 | 4,998,974,780 | 36.68% | 38.53% |
| Vehicle Manufacturer (V) | 4,859,473,348 | 4,916,297,388 | 18.41% | 19.15% |
| Hybrid (BV) | 5,000,859,104 | 5,108,024,171 | 38.68% | 43.42% |
| Alliance ((B+V)) | 6,540,679,996 | 6,684,186,668 | 20.17% | 23.02% |
The analysis confirmed that blockchain technology significantly boosts both profitability and recycling rates across all four distinct models. However, its effect is not uniform. Blockchain helps to decrease the optimal recycling price because the trust enabled by complete information lessens the consumer’s reliance on high prices to offset the risk of information asymmetry. For example, under blockchain, the average optimal recycling price in hybrid mode drops by 1.36%, helping firms lower their operating costs and build a stable supply channel. In contrast, the final product price increases because product traceability increases consumer confidence in its performance and used-car residual value. The alliance model continued to offer the largest total supply-chain profit, and the hybrid model saw the most significant percentage increase in recycling rate (from 38.68% to 43.42%), showing that blockchain aids cooperative channel competition by making it more transparent. More critically, even with lower consumer-facing recycling prices, both the volume and rate of recycling increased, solidifying the notion that consumer trust originated from blockchain can be regarded as a major facilitator for channel engagement and quality-based competition.
To unravel the influence of factors on the blockchain implementation and overall firm performance, I further performed a sensitivity analysis, systematically varying key parameter values. I investigated the impact of several parameters, including the cost-sharing ratio, price elasticity, consumer responsiveness to recycling, and the trust level in blockchain technology, on profitability and recovery performance. The analysis corroborated several essential findings. For instance, Table 6 illustrates the nuanced impact of altering the consumer blockchain trust coefficient \(k\) under differing values, here focusing on the resulting supply-chain profit and recycling rate for different models.
| Trust \(k\) | Profit of YB | Profit of Y(B+V) | Recycling Rate of YBV | Recycling Rate of Y(B+V) |
|---|---|---|---|---|
| 2 | 4,947,733,297 | 6,639,589,771 | 39.94% | 22.62% |
| 4 | 4,988,277,985 | 6,684,186,668 | 43.42% | 23.02% |
| 6 | 5,083,716,123 | 6,755,490,573 | 50.19% | 23.69% |
The sensitivity analysis proved that the consumer trust coefficient \(k\) acts as the primary catalyst in the relationship between technology investment and returns. My results demonstrate that as consumer trust in blockchain increases, the optimal investment level \(\lambda\) also escalates, positively impacting product pricing and total supply-chain profitability. For instance, in the hybrid and alliance models, an increase in trust from k=2 to k=6 resulted in substantial profit increases, emphasizing that blockchain investments need to be coupled with consumer education to maximize their positive impact. The recycling rate, in the hybrid model where channel competition is present, showed especially strong responsiveness to the trust coefficient. This validates the point that customers who value verifiable sustainability might prefer to recycle traction battery packs through a formal, transparent system even without being incentivized by high prices, creating both cost and efficiency advantages for recyclers.
These findings have profound managerial and policy implications. First and foremost, my research strongly suggests that governments should shift from static, one-size-fits-all policies to dynamic and self-adjusting regulatory mechanisms that are contingent on the actual state of the recycling ecosystem. By integrating penalties for violations and rewards for proactive behavior that update over time, policies can more effectively steer the entire system of traction battery packs toward a stable equilibrium. Second, policy must give pronounced consideration to technology level and consumer awareness as both act as internal motivating factors that drive sustainability alignment. To trigger real change, the government should encourage cost-effective technological advancements within the industry and launch campaigns to raise the public’s awareness of the positive externalities of properly recycling a traction battery pack.
From an operational perspective, my results can guide firm-level decision-making on the selection of suitable collection schemes for traction battery packs. For a firm whose paramount objective is achieving high profit margins and resource integration, alliance recycling emerges as a highly attractive strategy, granting greater coordination and consolidation of both forward and reverse resources. Conversely, when the primary objective is to maximize the collection rate of traction battery packs, the dual-channel, hybrid structure is more likely to succeed, particularly when endowed with blockchain, because the well-informed, transparent channel competition encourages participants to put forward better services and value propositions. It is essential, however, for enterprises to understand the combination of these factors. The use of blockchain in a collaborative alliance mode will produce much higher profit than open-market competitors, yet a hybrid mode when enhanced by blockchain allows for higher circulation rates for the battery packs to be captured for circular value recovery. The technology investments’ outcome is thus optimized when strategies are aligned to the intended sustainability and economic outcomes while taking the consumer’s digital trust into account. The allocation of blockchain implementation cost should also be equitable to ensure all partners see profit improvements, thereby fostering voluntary collaborations in the system.
I recognize that this modeling framework has certain limitations. In the theoretical setting, I did not systematically integrate other environmental policies, such as carbon taxes or cap-and-trade, although these mechanisms will likely exert a strong influence on the economics of a traction battery pack in a future decarbonizing economy. The principle of bounded rationality was not fully explored in this model, as it considers a risk-neutral environment and does not accommodate for deviations based on fairness concerns or other cognitive biases. In my subsequent work, I intend to extend and enrich the current body of research by focusing on multi-tier networks of recyclers, incorporating more varied entities such as third-party and echelon utilization firms, and investigating the interactions between carbon regulations and reward-punishment mechanisms. This would enable a more comprehensive view of how different interventions could enhance the overall circularity of the traction battery pack supply chain under different policy and informational structures. Specifically, I wish to introduce complex, heterogeneous consumers and decision-makers within a behavioral operations framework to more realistically calibrate the power of trust and policy and design robust management strategies for a future where volumes of retired traction battery packs will be soaring.
In conclusion, my study offers dual and counterbalancing approaches for optimizing traction battery pack recycling decisions, utilizing dynamic policy adjustments and technological empowerment to unravel the strategic issues of asymmetric information, policy limitations, and low market trust. The dynamic reward-punishment policy acts as a catalyst for positive behavioral changes, while blockchain provides a robust platform where that behavioral movement can turn into high-efficiency operation by direct, decentralized trust. Through the integrated insights gained from both the evolutionary game model and the closed-loop supply chain analysis, it becomes evident that effective recycling of the traction battery pack is not merely an issue of economics, but also one of synchronized design between technical enablers and adaptive institutions, which calls for holistic management from all stakeholders in the value chain.