Resource Benefit Evaluation of Lithium Recovery from New-Energy Vehicle Batteries

Abstract

With the popularity of new-energy vehicles, the recovery and reuse of lithium-ion battery (LIB) resources have become topics of great concern. This study explores the risks of the lithium resource chain in terms of supply–demand balance and lithium resource criticality. We propose a prediction algorithm for lithium production based on reverse-order MT-EGM-SD (metabolism–even grey model–system dynamics), upon which a system dynamics model for lithium resource recycling and reuse is constructed. We use dynamic simulation to evaluate the benefits of lithium resource recovery and the effects of different LIB recovery strategies. The results show that LIB recycling strategies, such as enhancing subsidy levels and strengthening public awareness initiatives, can significantly increase lithium resource recovery rates. From a medium- and long-term perspective, however, the technological progress strategy can greatly reduce lithium consumption intensity in the battery. Cascade use policy has significant economic benefits, but it delays the recycling of secondary raw materials. Under the joint strategy with the best resource efficiency (stringent government recycling regulations and significant advancements in battery production technology), the lithium supply–demand balance and the lithium resource recovery rate increase by 301.89% and 795.65%, respectively. Meanwhile, lithium resource chain risk, lithium criticality, and actual lithium demand decrease by 18.77%, 18.86%, and 75.11%, respectively.

1. Introduction

1.1. Motivation

There are concerns about the safe disposal of spent lithium-ion batteries (LIBs) used in new-energy vehicles (NEVs). The service life of LIBs is usually 8–10 years. If waste LIBs are not properly recycled, they will cause serious resource waste and pollution [1]. In addition, recycling metals from spent LIBs helps solve the problem of resource supply and demand [2]. Thus, recycling valuable metals from retired LIBs can address pollution while holding great significance for sustainable resource development [3].

Given its importance and high supply risk in the raw material supply chain, lithium is listed as a potential key material by the European Union and various countries [4,5]. China’s high demand for lithium resources is largely driven by the rapid development of emerging industries, such as NEVs and energy storage [6]. With the continued expansion of the LIB industry, concerns have been raised about the related supply and value chains [7]. NEV industry development has increased China’s demand for lithium, which could adversely affect the safety of China’s lithium supply chain in the future [8]. Given the difficulties of mining, there is an urgent need for a sustainable strategy to recover lithium from secondary resources with potential value [9]. Lithium resource recycling is considered the most effective way to reduce the demand for raw materials [10], improve resource efficiency [11], and stabilize the domestic lithium supply chain and economic situation. It is estimated that by 2030, the cumulative number of discarded LIBs will reach 121 billion, including over 522 kt of recyclable lithium resources [12]. By 2040, demand for lithium, cobalt, and nickel for NEV LIBs is expected to exceed the current production of raw materials, and the recovery potential of lithium and nickel will exceed half the demand for raw materials for LIBs [13]. Although the number of used LIBs has increased significantly, recycling technology and innovation have not developed at the same pace [14].

1.2. Literature Overview

Although some progress has been made, battery recycling technology still faces challenges in terms of efficiency, effectiveness, and environmental sustainability [15]. Due to the low recovery rate of lithium, studies have focused on technologies that could improve the recovery rate of valuable metals such as lithium [16]. Pyrometallurgy and hydrometallurgy are common methods for recovering LIBs [17]. However, these come with pollution problems, giving rise to proposals for green, cost-effective processes [18,19]. Natarajan et al. [20] proposed an emerging direct recycling technology that simplifies the process and surpasses traditional technologies in terms of energy savings and reduced carbon footprint, thus providing a promising solution. Saleem et al. [21] have shown that direct lithium extraction (DLE) can minimize Li losses in the recycling process.

In terms of the benefits of recycling LIBs, studies have focused on the driving factors in terms of stakeholders [22], the recycling potential and value of key materials [23], the economic benefits of recycling [24], and environmental benefits based on life-cycle assessments [25,26]. Chen et al. [27] proposed that recycling policies such as battery subsidies can significantly improve the economic and environmental benefits of recycling. Current sustainability transformations render certain minerals, such as lithium, ‘critical’ [5]. The long-term solution to avoid bottlenecks in LIB production is the creation of a circular economy by consolidating the LIB value chain with recycling [28].

The existing literature provides a theoretical foundation for this study, but several limitations remain: First, research on the resource benefits of lithium recovery from LIBs often lacks a systematic approach to comprehensively analyze and explore the causal relationships among various systems and variables. Second, current studies on the recycling of new-energy vehicle batteries are predominantly limited to static analyses, lacking a dynamic perspective. This results in an insufficient evaluation of the medium- and long-term impacts of recycling strategies. Third, existing research does not adequately address the synergistic effects of recycling policies, secondary utilization, and technological advancements on the criticality of lithium, failing to fully uncover their interaction mechanisms. Recycling lithium materials from LIBs offers economic, environmental, and resource benefits, helping to balance lithium resource supply and demand and mitigate lithium supply risks. In light of these gaps, this paper develops a system dynamics model for lithium resource recovery from the power batteries of new-energy vehicles. The main content and innovations are as follows:

This study proposes a lithium production prediction algorithm based on metabolism–even grey model–system dynamics (MT-EGM-SD). Taking China as an example, we accurately process and forecast data, construct a system dynamics (SD) model for the recycling and reuse of NEV LIBs, and explore whether the reuse and recycling of lithium is a sustainable choice. At the same time, to overcome the limitations of a single recycling policy, we introduce a strategy for improving LIB production technology, quantify the effect of LIB recycling and the technological progress of lithium resource supply and demand, as well as supply risk mitigation, simulate and analyze different strategies, find the optimal combination scheme, and make policy suggestions.

2. Methods

2.1. Construction of Lithium Resource Recycling System

Based on feedback control theory and computer simulation technology, SD is a quantitative research method for studying complex socioeconomic systems [29]. Through modeling and simulation, it reveals the interactions between the components of a system, especially the effects of feedback loops, time delays, and nonlinear relationships on the behavior of the system. It has been applied to various topics, such as the coordinated development of urban agglomerations [30] and the coordinated development of the economy, resources, and the environment [31].

The research procedure of system dynamics is shown in Figure 1.

The system dynamics approach possesses several distinctive characteristics:

(1) Long-term temporal analysis capability: System dynamics is particularly effective for analyzing medium- and long-term problems characterized by time delays and dynamic feedback mechanisms, enabling the simulation of system behavior over extended periods. (2) Robustness in data-scarce environments: This methodology maintains its analytical value even with limited data availability, as it can effectively utilize qualitative analysis and model construction techniques. (3) Handling complex system behaviors: System dynamics excels in addressing high-order, nonlinear, and time-varying problems, effectively capturing complex nonlinear relationships and temporal variations within dynamic systems. (4) Flexible policy experimentation: The approach allows for the flexible configuration of control factors, enabling researchers to observe system behavior and state changes under various conditions, thereby facilitating policy impact analysis. (5) Scenario simulation capacity: Through comprehensive scenario analysis and policy combination simulations, it provides valuable insights into potential system behaviors, supporting decision-makers in evaluating various strategic alternatives. (6) Integrated analytical framework: System dynamics combines qualitative and quantitative methodologies, incorporating both causal loop analysis (qualitative) and policy simulation analysis (quantitative) within a unified framework.

The foundation for developing the system dynamics (SD) model for lithium resource recovery from new-energy vehicle power batteries lies in accurately identifying the system’s causal feedback relationships. This involves systematically analyzing the interaction mechanisms between the system’s components and their interconnections, which are visually represented through a causal loop diagram, as illustrated in Figure 2. During the modeling process, we use directed arrows (→) to indicate causal relationships:

A positive sign (“+”) denotes a reinforcing relationship, where both variables change in the same direction.

A negative sign (“−“) indicates a balancing relationship, where the variables change in opposite directions.

Due to the high risk of the lithium resource chain, the government will actively promote the battery recycling strategy, increase battery recycling efforts, and improve the level of lithium recovery and lithium resource recovery rate, so as to reduce the risk of the lithium resource chain. Through the improvement of battery production technology, enterprises can reduce the lithium consumption intensity in lithium-ion batteries, thereby reducing the demand for lithium in new batteries and achieving a better balance between lithium supply and demand.

The LIB resource recovery system is a complex system involving transportation, the environment, and resources. We use Vensim to build a resource benefit evaluation model for recovering lithium from NEV batteries and to analyze the interactions between variables. It includes three systems: the lithium market, NEV development, and the lithium resource recovery policy (Figure 3).

In the lithium demand system, we consider the demand for new LIBs brought by NEV growth and battery replacement, including ternary LIBs (NCMs) and lithium iron phosphate (LFP) batteries. Improvements in LIB production technologies can reduce the intensity of lithium consumption and reduce the growth of lithium demand in terms of adding and replacing batteries. Battery recovery policies can improve the level of lithium recovery. Recovered lithium can then be reused, thus reducing lithium demand.

For the risk assessment of the lithium resource chain, we select the factors of lithium supply–demand balance, the proportion of China’s lithium production, and lithium criticality. Lithium supply–demand balance and the proportion of China’s lithium production are negative indicators, and lithium criticality is a positive indicator (refer to the equation in Section 2.2 for details). The simulation time of the model is set to 2014–2030, and the step length is one year.

The model assumptions are as follows:

(1)

The model only considers the effect of the demand for NEV LIBs on supply and demand and lithium criticality, and does not consider lithium demand in other industries.

(2)

Lithium extracted from recycled LIBs can be reused without considering the quality of the recycled lithium.

The main equations of the flowchart are given below.

2.2. Lithium Resource Chain Risk Assessment System

(1) Lithium supply–demand balance = lithium supply/actual lithium demand.

The lithium supply–demand balance indicates the ability of domestic production to meet demand. We use the ratio of lithium supply to actual demand to express the degree of lithium supply–demand balance. When the lithium supply–demand ratio is greater than 1, the supply is surplus; otherwise, it is insufficient.

(2) Lithium supply risk.

Supply risk (SR) is based on the supply security of major suppliers and centralized governance, reflecting the ease of supply chain disruption. Factors causing SR include a lack of substitutes, the concentration of primary resource-producing countries, a low recovery rate, and poor governance in producing countries. These four factors are combined into an indicator. The SR formula is

SR = σ(1 − r)HHI_(WGI)

(1)

The SR index is the cubic product of the replaceability index (σ) of rare-metal lithium, the recovery rate in the life cycle (r), and the Herfindahl–Hirschman index (HHI_(WGI)). The HHI is obtained by weighting the worldwide governance indicators (WGI) of each producing country with the square of the production concentration (P_c) of each producing country as the weight, describing the production concentration and governance status of raw materials at the national level:

$${\mathit{HHI}}_{\left(\mathit{WGI}\right)}=\sum _c{\left(P_c\right)}^2\mathit{WGI}$$

(2)

The replaceability index (σ) indicates the difficulty of replacing lithium at the end-use stage. The higher the replaceability index of lithium resources in the final consumption link, the higher the supply chain risk. The replacement difficulty of mineral resources is divided into four levels, as shown in

Table 1. Replaceability index.

Value	Explain
0	No additional cost
0.3	Easy replacement and low cost
0.7	Replacement is difficult and costly
1	Irreplaceable

: 0 means no additional replacement cost and no replacement difficulty; 0.3 means the replacement cost is low and the difficulty is small; 0.7 means a higher replacement cost or greater performance loss, and replacement is more difficult; and 1 means replacement cannot be completed. Given its unique characteristics, lithium has few alternatives, most of which lead to a decline in product performance. Therefore, we set the replaceability index of lithium to 0.7.

WGI is an indicator developed by the World Bank [32] that reflects a country’s governance level and is applicable to different life-cycle stages of rare metals. It includes six subindicators: violence and responsibility, political stability, government effectiveness, corruption, legal system, and regulatory quality.

Table A1. WGI.

	China	Argentina	Australia	Brazil	Chile	Portugal	US	Zimbabwe
2014	5.95	5.75	1.79	5.07	2.73	3.11	2.54	7.66
2015	5.95	5.62	1.94	5.32	2.98	2.96	2.54	7.46
2016	5.87	5.02	1.90	5.34	3.14	3.00	2.55	7.48
2017	5.68	4.97	1.98	5.45	3.16	2.86	2.55	7.49
2018	5.65	4.98	1.91	5.56	3.15	2.91	2.57	7.41
2019	5.74	5.21	1.98	5.47	3.37	2.94	2.86	7.50
2020	5.62	5.30	2.07	5.46	3.43	3.02	3.09	7.52
2021	5.55	5.44	2.05	5.53	3.50	3.10	3.00	7.44
2022	5.67	5.46	2.02	5.58	3.57	3.09	3.01	7.39
2023	5.58	5.41	1.92	5.56	3.53	3.19	3.01	7.36

in Appendix B shows the specific data after normalization.

(3) Environmental risk of lithium.

Environmental risk (ER) measures the risk caused by the measures each producing country takes to protect the environment, which will reduce the supply of metal resources. The formula is

ER = σ(1 − r)HHI_(EPI)

(3)

We use the environmental performance index (EPI) released by Yale University to measure the level of environmental governance in lithium-producing countries. The greater the environmental damage caused by rare-metal exploitation, the stronger the restrictions on such exploitation in countries with better environmental governance, and the greater the SR [33].

Table A2. EPI.

	China	Argentina	Australia	Brazil	Chile	Portugal	US	Zimbabwe
2014	43.0	49.7	80.4	59.2	83.2	91.8	79.5	13.7
2015	55.3	39.1	80.3	56.6	83.5	89.6	83.2	24.7
2016	65.1	52.1	80.5	65.0	78.7	90.3	81.7	18.1
2017	62.0	58.1	74.0	55.7	80.6	87.0	79.3	29.5
2018	50.7	55.8	83.0	64.4	88.6	93.5	88.4	47.7
2019	39.7	58.6	89.8	69.8	86.9	89.8	81.9	26.3
2020	37.3	52.2	74.9	51.2	55.3	67.0	69.3	37.0
2021	32.9	46.7	67.5	47.4	51.0	68.3	60.2	41.6
2022	28.4	41.1	60.1	43.6	46.7	69.6	51.1	46.2
2023	35.4	47.0	63.1	53.0	49.6	61.9	57.2	51.6

in Appendix B shows the data. During calculation, it is normalized into a range of [1,2,3,4,5,6,7,8,9,10]. Similar to the calculation of SR, the calculation and data acquisition for the rare-metal substitutability index and recycling rate index are the same as in the dimension of SR. The HHI is the weighted sum of the EPI of each producing country with the square of the production concentration (P_c) of each producing country as the weight:

$${\mathit{HHI}}_{\left(\mathit{EPI}\right)}=\sum _c{\left(P_c\right)}^2\mathit{EPI}$$

(4)

(4) Lithium criticality = SR × 0.7 + ER × 0.3.

The critical assessment of lithium considers the two dimensions of SR and ER and is obtained by weighted summation according to the specific values of the two risks.

(5) Proportion of lithium production = lithium supply rate/global lithium production.

The proportion of lithium production is quantified by the ratio of China’s lithium production to global production.

(6) Lithium resource chain risk = EXP (0.4 × lithium criticality − 0.4 × the lithium supply − demand balance − 0.2 × proportion of lithium production).

Lithium resource chain risk considers three indicators: lithium criticality, the balance of supply and demand, and the proportion of production. The higher the lithium criticality, the higher the lithium resource chain risk. Supply–demand balance and the proportion of production are negative indicators. The higher the balance of supply and demand and the proportion of output, the lower the lithium resource chain risk.

2.3. Lithium Resource Demand Subsystem

(1) Proportion of NCM = WITHLOOKUP (time ([(2014, 0) − (2030, 1)], (2014, 0.15), (2015, 0.2), (2016, 0.28), (2017, 0.5), (2018, 0.63), (2019, 0.75), (2020, 0.8), (2021, 0.82), (2022, 0.84), (2023, 0.86), (2024, 0.88), (2025, 0.9), (2026, 0.91), (2027, 0.92), (2028, 0.93), (2029, 0.94), (2030, 0.95))).

China’s NEVs have experienced a transition from LFP to NCM [34]. The proportions of lithium manganese oxide and lithium cobalt oxide batteries are relatively low and are not considered. In the future, there is the possibility of changing to NCM LIBs. Therefore, we predict that the proportion of NCM LIBs will reach 95% in 2030.

Table 2. The distribution of different types of LIBs used in NEVs.

Type	2014	2015	2016	2017	2018	2019	2020	2021	2022	2023	2024	2025	2026	2027	2028	2029	2030
NCM	15%	20%	28%	50%	63%	75%	80%	82%	84%	86%	88%	90%	91%	92%	93%	94%	95%
LFP	85%	80%	72%	50%	37%	25%	20%	18%	16%	14%	12%	10%	9%	8%	7%	6%	5%

shows the proportion of different LIBs.

(2) Lithium consumption intensity (NCM) = 0.2337 × ((1 − technological progress) ^{(time − 2014)}).

Referring to the initial lithium consumption intensity in 2014 [34] and assuming the average annual progress rate of battery production technology remains unchanged, the relationship between lithium consumption intensity and technological progress can be obtained.

(3) Retirement amount of NCM = IF THEN ELSE (time < 2019, DELAY1I (growth of NEVs × proportion of NCM, 8, 0), IF THEN ELSE (time < 2021, DELAY1I (growth of NEVs × proportion of NCM, 9, 0), IF THEN ELSE (time < 2026, DELAY1I (growth of NEVs × proportion of NCM, 10, 0), DELAY1I (growth of NEVs × proportion of NCM, 12, 0)))).

We refer to Zheng et al. [35] for the life span of different battery types. The “Energy Saving and NEV Technology Roadmap” proposed that the full life cycle of LIBs would reach 10 years by 2020, 12 years by 2025, and 15 years by 2030. We set the future development of battery life accordingly, as shown in

Table 3. Service life setting of LIBs.

Type	2014–2018	2019–2020	2021–2025	2026–2030
NCM	8 years	9 years	10 years	12 years
LFP	10 years	11 years	12 years	15 years

:

Other parameter equations can be found in Appendix A.

2.4. Lithium Production Prediction Algorithm Based on Reverse-Order MT-EGM-SD

To accurately predict the parameter values in the simulation period, we construct a prediction method based on reverse-order MT-EGM-SD. First, we use the smoothness test in grey theory to process the data. Then, we use the metabolic grey prediction (MT-EGM) model to predict the data. Finally, the prediction function of lithium production is constructed. Figure 4 shows the algorithm block diagram.

The calculation process of the global lithium production data prediction algorithm is as follows:

Step 1: Data collection and processing.

Write the original data in sequence form: $X^{(0)}=(x^{(0)}(1),x^{(0)}(2),\cdots ,x^{(0)}(n)).$

Taking global lithium production data as an example, we collect data from 2014 to 2022 according to the statistical yearbooks [36].

Table A3. Lithium production data.

	China	Argentina	Australia	Brazil	Chile	Portugal	US	Zimbabwe	Total World
2014	2.3	3.2	12.4	0.2	10.8	0.3	0.9	0.9	31.0
2015	2.0	3.6	11.9	0.1	9.8	0.3	0.9	0.9	29.5
2016	2.3	5.8	14.0	0.2	13.6	0.4	0.9	1.0	38.2
2017	6.8	5.7	21.3	0.3	14.2	0.8	0.9	0.8	50.9
2018	7.1	6.4	57.0	1.0	17.0	1.2	0.9	1.6	95.2
2019	10.8	6.3	45.0	2.2	19.2	0.9	0.9	1.2	86.9
2020	13.3	5.9	39.7	1.4	21.6	0.3	0.9	0.4	83.6
2021	14.0	6.0	55.3	1.7	28.3	0.9	0.9	0.7	107.9
2022	19.0	6.2	61.0	2.2	39.0	0.6	0.9	0.8	130.3
2023	33.0	9.6	86.0	4.9	56.6	0.4	0.6	3.4	198.0

in Appendix B shows the data.

$X^{(0)}=(x^{(0)}(1),x^{(0)}(2),\cdots ,x^{(0)}(9))$= (31.0, 29.5, 38.2, 50.9, 95.1, 86.9, 83.7, 107.9, 130.4).

Step 2: Quasi-smoothness test of sequences.

Set $X=(x(1),x(2),\cdots x(n)),x(k)\ge 0,k=1,2,\cdots n$. Then,

$$\rho (k)={\displaystyle \frac{x(k)}{{\displaystyle \sum _{i=1}^{k-1}x(i)}}},k=2,3,\cdots n.$$

(5)

is called the smoothness ratio of sequence $X$ [37].

We use the smoothing ratio based on the value of the elements in sequence $X$ to investigate its change characteristics. That is, we use the ratio ρ(k) of the kth data $x(k)$ in the sequence and the sum ${\displaystyle \sum _{i=1}^{k-1}x(i)}$ of the previous $k-1$ data to investigate whether the data in sequence $X$ change smoothly.

If the sequence $X=(x(1),x(2),\cdots x(n)),x(k)\ge 0$ satisfies the following conditions, then $X$ is called a quasi-smooth sequence:

$$A(k)={\displaystyle \frac{\rho (k+1)}{\rho (k)}}<1,k=2,3,\cdots n-1.$$

(6)

(1) $\rho (k)\in [0,\epsilon ],k=3,4,\cdots ,n.$

(2) ε < 0.5.

Whether the quasi-smoothness condition is satisfied is an important criterion for testing whether a grey system model can be established for a sequence. Therefore, we calculate the smoothness ratio $\rho (k)$ and $A(k)$ of $X^{(0)}$ before establishing the grey prediction model;

Table 4. Smoothness ratio of $X^{(0)}$.

Time	${\mathit{X}}^{(0)}$	$\mathit{\rho }(\mathit{k})$	$\mathit{A}(\mathit{k})$
2014	31.0
2015	29.5	0.952	0.664
2016	38.2	0.632	0.816
2017	50.9	0.516	1.233
2018	95.1	0.636	0.559
2019	86.9	0.355	0.711
2020	83.7	0.252	1.029
2021	107.9	0.260	0.959
2022	130.4	0.249

shows the results.

$\rho (\mathrm{k})=(\rho (2),\rho (3),\cdots \rho (9))$ = (0.952, 0.632, 0.516, 0.636, 0.355, 0.252, 0.260, 0.249);

$A(k)=(A(2),A(3),\cdots A(8))$ = (0.664, 0.816, 1.233, 0.559, 0.711, 1.029, 0.959).

Obviously, $X^{(0)}$ does not meet the quasi-smoothness test of the sequence, and the data need to be smoothed.

Step 3: Smoothness processing of sequences.

(1) Establish the inverse sequence of the original sequence:

$$X^{(1)}=(x^{(0)}(9),x^{(0)}(8),\cdots x^{(0)}(1)).$$

(2) For the inverse sequence X⁽¹⁾, if $x_k$ is not used as the smooth ratio, but $x_{k-1}$ and $x_{k+1}$ are smooth ratios, $x_k$ is averaged. We let

$$x_k=1/2(x_{k-1}+x_{k+1}).$$

(7)

(3) For the inverse sequence X⁽¹⁾, if there is $x_k$ with continuous n ≥ 2 that does not meet the quasi-smooth sequence condition, then the MT-EGM model is established for the first n – m + 1 data that meet the smooth sequence condition to predict the data in the next few periods.

As A(7) = 1.029 > 1, A(6) = 0.711 < 1, A(8) = 0.959 < 1, X⁽¹⁾(7) is averaged, and X⁽¹⁾(7) = 1/2(x⁽¹⁾(6) + x⁽¹⁾(8)) = (86.9 + 107.9)/2 = 97.4.

$\rho (3),\rho (4),\rho (5)>0.5$ is continuous. Therefore, the metabolic EGM model is established for the reverse sequence X₁⁽⁰⁾ = (x⁽⁰⁾(9), x⁽⁰⁾(8), x⁽⁰⁾(7), x⁽⁰⁾(6)).

Step 4: Reverse EGM model.

The EGM model is established for the reverse sequence X⁽¹⁾, satisfying the smoothness condition X₁⁽⁰⁾ = (x⁽⁰⁾(9), x⁽⁰⁾(8), x⁽⁰⁾(7), x⁽⁰⁾(6)) = 130.4, 107.9, 97.4, 86.9.

(1) Calculate the 1-AGO generation sequence of sequence X⁽¹⁾: 130.40, 238.30, 335.70, 422.60;

(2) Calculate the nearest mean generating sequence of the 1-AGO generating sequence: 184.35, 287.00, 379.15;

(3) Calculate the grey model development coefficient a and grey action quantity b: a = 0.1, b = 127.9;

(4) Find the time response formula:

$${\widehat{X}}_1^{(0)}=(1-e^a)(x^{(0)}(9)-{\displaystyle \frac{b}{a}})e^{-a(k-1)}=(1-e^{0.1})(130.4-{\displaystyle \frac{127.9}{0.1}})e^{-0.1(k-1)}$$

(8)

Calculate the simulation sequence according to the time response formula: $\widehat{X}$ = 130.4, 108.0, 96.9, 87.0.

The relative error of simulation is 0.2%.

Step 5: Metabolic prediction model.

(1) Use the above model to predict the data: $x_1^{(0)}(5)$ = 78.2;

(2) Add new information $x_1^{(0)}(5)$ to X⁽¹⁾, remove the oldest information x⁽⁰⁾(9), and obtain $X_1^{(1)},X_1^{(1)}=(x^{(0)}(8),x^{(0)}(7),x^{(0)}(6),x^{(0)}(5))$;

(3) Repeat the above steps until $x_1^{(0)}(1)$ is predicted;

(4) Add the predicted data to the original inverse sequence $X_1^{(0)}$ to obtain a new sequence:

$$\begin{gathered} X_1^{(2)}=(x^{(0)}(9),x^{(0)}(8),\cdots x^{(0)}(6),x_1^{(0)}(5),x_1^{(0)}(4),\cdots x_1^{(0)}(1)=130.4,107.9, \\ 97.4,86.9,78.2,69.9,62.8,56.2,50.4. \end{gathered}$$

Step 6: Construct and test the EGM model.

Restore the reverse sequence X₁⁽²⁾ and establish the EGM model for model verification. If the simulation accuracy-type test passes, data prediction in the simulation period will be carried out. If it fails, we return to the second step.

Table 5. Model test results.

Time	Original EGM Model			Reverse-Order MT-EGM Model
	Actual Value	Simulation Value	Relative Error	Actual Value	Simulation Value	Relative Error
2014	31.0	31.0	0.00%	50.4	50.4	0.00%
2015	29.5	40.9	38.64%	56.2	54.6	2.85%
2016	38.2	48.4	26.70%	62.8	61.5	2.07%
2017	50.9	57.2	12.38%	69.9	69.3	0.86%
2018	95.1	67.7	28.81%	78.2	78.0	0.26%
2019	86.9	80.1	7.83%	86.9	87.9	1.15%
2020	83.7	94.7	13.14%	97.4	99.0	1.64%
2021	107.9	112.0	3.80%	107.9	111.4	3.24%
2022	130.4	132.5	1.61%	130.4	125.5	3.76%

shows the model test results.

(1) The restore sequence is

$$X_2=(x_2(1),x_2(2),\cdots x_2(9))=50.4,56.2,62.8,69.9,78.2,86.9,97.4,107.9,130.4.$$

(2) The simulation sequence is

$${\widehat{X}}_2=({\widehat{x}}_2(1),{\widehat{x}}_2(2),\cdots {\widehat{x}}_2(9))=50.4,54.6,61.5,69.3,78.0,87.9,99.0,111.4,125.5.$$

We can see in Table 5 that compared with the original EGM model, the reverse-order MT-EGM model retains the original data to the greatest extent, the simulation accuracy of the data is greatly improved, and all are within 5%. Thus, the simulation effect of the model is good, and it can predict the data in the simulation period.

Step 7: Prediction of simulation data and construction of SD function.

(1) Using the EGM model to predict the data in the simulation period, global lithium production data for the next eight years are predicted:

$$\begin{gathered} X^{(0)}=(x^{(0)}(10),x^{(0)}(11),\cdots x^{(0)}(17))=141.3,159.2,179.3,201.9,227.4, \\ 256.1,288.4,324.8; \end{gathered}$$

(2) Construct the schedule function of global lithium production based on MT-EGM-SD.

Global lithium production (2014–2030) can be obtained from the above steps. Therefore, the schedule function of global lithium production can be established as follows:

Global lithium production = WITH LOOKUP (time ([(2014, 0) − (2030, 400,000)], (2014, 31,000), (2015, 29,500), (2016, 38,200), (2017, 50,900), (2018, 95,100), (2019, 86,900), (2020, 83,700), (2021, 107,900), (2022, 130,400), (2023, 141,300), (2024, 159,200), (2025, 179,300), (2026, 201,900), (2027, 227,400), (2028, 256,100), (2029, 288,400), (2030, 324,800))).

The calculation process for EPI and WGI is similar.

2.5. Model Validation

We confirm the applicability and rationality of the model using the historical value test method, which judges the effectiveness of a model based on the difference between the simulated value of system variables and historical data. We select the number of NEVs from 2014 to 2023 to test the relative error of the model.

Table 6. Model check (unit: 10,000 vehicles).

Time	Actual Value	Simulation Value	Relative Error
2014	22	22.0	0.00%
2015	41	41.9	2.24%
2016	91	90.6	0.47%
2017	153	152.1	0.59%
2018	261	259.1	0.74%
2019	381	378.1	0.76%
2020	492	488.8	0.66%
2021	784	778.1	0.75%
2022	1310	1298.3	0.90%
2023	2041	2020.8	0.99%
2024	3140	3098.5	1.32%

shows the results. We can see that the relative error of the simulation value of the NEV ownership data is within 5%. Thus, the model can accurately describe the basic status of the research system and be used to simulate lithium recovery from NEV batteries in the next stage.

3. Results and Discussion

3.1. Effect and Limitations of LIB Recycling Strategy

We change the government’s recycling efforts and set LIB recycling rates to 0.1, 0.5, and 0.9 under the scenarios of low, medium, and high policy effort.

Figure 5 shows the change trend simulation diagram of the main variables under the single LIB recycling strategy. We can see in Figure 5a,b that under the effect of the separate battery recovery policy, the lithium supply–demand balance improves and lithium resource chain risk decreases, but the overall supply–demand balance shows a fluctuating downward trend. The closer to the end of the simulation, the smaller the effect of the battery recovery policy. In Figure 5c–e, lithium SR shows an upward trend over time, while ER shows a slow downward trend. Lithium criticality tends to be stable for a long time under the joint action of SR and ER, and criticality decreases with increases in recovery rate. Figure 5 shows that the lithium resource recovery rate increases significantly under the effect of the recovery policy. Overall, the battery recovery policy can reduce the supply risk and ER of lithium and thus reduce lithium criticality. In the later stage of simulation, lithium criticality is the main factor affecting lithium resource chain risk.

By the end of the simulation period (2030), the high recovery effort increases the lithium supply and demand balance by 23.58% compared with the low recovery effort, reducing lithium resource chain risk by 7% (

Table 7. Effect of the single recycling policy (2030).

Variable	Low Recovery	High Recovery	Degree of Change
Lithium supply–demand balance	0.106	0.131	23.58%
Lithium resource chain risk	1.271	1.182	−7.00%
SR	0.766	0.623	−18.67%
ER	1.021	0.83	−18.71%
Lithium criticality	0.843	0.685	−18.74%
Lithium resource recovery rate	0.023	0.205	791.30%
Lithium supply–demand balance	0.106	0.131	23.58%

). The SR, ER, and criticality of lithium are reduced by more than 18%, and the lithium resource recovery rate is increased by 791.30%. At the end of the simulation period, the recovery rate of lithium resources reached more than 20%.

3.2. Nature and Long-Term Effects of the Technological Progress Strategy

Set the technological progress rate to 0.01, 0.05, and 0.1 under the scenarios of low, medium, and high LIB production technological progress, respectively (Figure 6). We can see in Figure 6a–c that under the influence of the separate technological progress policies, the lithium supply–demand balance significantly improves, and the degree of improvement gradually becomes significant over time. Lithium criticality has hardly changed. The risk reduction in the lithium resource chain is mainly attributable to the lithium supply–demand balance, mainly because the improvement in LIB production technology can reduce lithium consumption intensity, as shown in Figure 6e,f, thereby reducing actual lithium demand, as shown in Figure 6d.

We can see that by the end of the simulation, under the effect of technological progress, the lithium supply–demand balance reaches 0.348, which is an increase of 228.30% compared with the low technology state, and the degree of change is large (

Table 8. Effect of the single technological progress policy (2030).

Variable	Low Technological Progress	High Technological Progress	Degree of Change
Lithium supply–demand balance	0.106	0.348	228.30%
Lithium resource chain risk	1.271	1.153	−9.28%
Lithium criticality	0.843	0.842	−0.12%
Lithium consumption intensity (NCM)	0.199	0.043	−78.39%
Lithium consumption intensity (LFP)	0.133	0.029	−78.20%
Actual lithium demand	3,680,700	1,124,600	−69.45%

). The lithium resource chain risk decreases from 1.271 to 1.153, with a change of 9.28%. Lithium criticality is almost unchanged, indicating that battery technology improvement can effectively improve the balance between the supply and demand of lithium, thereby reducing lithium resource chain risk. Technological progress mainly reduces actual lithium demand by reducing LIB consumption intensity. In 2030, the lithium consumption intensity of NCM and LFP will be reduced to 0.043 kg/kwh and 0.029 kg/kwh, respectively, a decrease of more than 78%. Actual lithium demand decreases by 2.5561 × 10⁶ t, a change of 69.45%.

3.3. Negative Effect of the Strategy for Improving Reuse Rate on Resource Efficiency

We set the low, medium, and high LIB reuse rates to 0.1, 0.3, and 0.6, respectively. Figure 7 presents a simulation diagram of the change trend of the main variables under a single strategy to improve the reuse rate. As shown in Figure 7a, the single policy of improving the reuse rate has little effect on lithium resource chain risk. Its main effect is to increase the cascade use of LIBs, thus delaying the recovery process of lithium; so, lithium recovery is reduced at the same time, as shown in Figure 7c,d. The lithium resource recovery rate decreases, as shown in Figure 7b.

As shown in

Table 9. Effect of the single promotion and reuse policy (2030).

Variable	Low Reuse Rate	High Reuse Rate	Degree of Change
Lithium resource chain risk	1.226	1.247	1.71%
Lithium resource recovery rate	0.114	0.072	−36.84%
Cascade use	31,873	191,124	499.64%
Lithium recovery	430,574	271,069	−37.04%

, lithium resource chain risk increases by 1.71% in 2030, with little change. With the improvement in the reuse rate, the cascade use of lithium increases by almost 500%, reaching 1.91124 × 10⁵ t, while the recovery of lithium decreases by 1.59505 × 10⁵ t, and the lithium resource recovery rate decreases by 36.84%. Although the LIB reuse policy slows down the recovery efficiency of lithium resources and has a certain negative effect on the balance of lithium supply and demand and lithium resource chain risk, it improves the cascade use of lithium and has certain economic benefits. We do not further explore the economic benefits of cascade use.

3.4. Joint Strategy

Table 10. Joint scheme design.

Strategy	Battery Recovery Rate	Technological Progress	Reuse Rate
Current	L	L	L
Scheme 1	H	L	L
Scheme 2	H	H	L
Scheme 3	H	H	H

shows the design of the joint strategy, in which the initial state is one of a low battery recovery rate, low technological progress, and low reuse rate. Scheme 1 is the policy for improving the battery recovery rate, Scheme 2 is the strategy for improving the battery recovery rate and technological progress, and Scheme 3 is the strategy for a high battery recovery rate, high technological progress, and high reuse rate.

Figure 8 presents a simulation diagram of the effect of the joint strategy on the main variables. In Figure 8a, the change from curve 2 to curve 3 is the most obvious, and in Figure 8c, the change from curve 1 to curve 2 is the most obvious. This indicates that technological progress has a greater effect on the lithium supply–demand balance, while battery recovery has a greater effect on lithium criticality. Therefore, in the simulation diagram of lithium resource chain risk in Figure 8b, the changes from curve 1 to curve 2 are mainly affected by the lithium criticality changes, and the changes from curve 2 to curve 3 are mainly affected by the lithium supply–demand balance. Overall, curve 3 has the largest change compared with curve 1, indicating that the combined strategy of improving battery recovery and technology level in scheme 2 has the greatest effect on lithium resource chain risk.

Figure 8d shows a simulation diagram of actual lithium demand. We can see that actual lithium demand is mainly affected by technological progress, and the effect of other strategies is small. Curve 4 in Figure 8 shows that the strategy of improving the reuse rate has a certain negative effect on the resource benefits analyzed in this study, including lithium resource chain risk, the balance between lithium supply and demand, and lithium criticality.

Table 11. Joint strategy effect (2030).

Variable	Current	Scheme 1	Change	Scheme 2	Change	Scheme 2	Change
Lithium supply–demand balance	0.106	0.131	23.58%	0.426	301.89%	0.392	269.81%
Lithium resource chain risk	1.156	1.089	−5.80%	0.939	−18.77%	0.972	−15.92%
Lithium criticality	0.843	0.685	−18.74%	0.684	−18.86%	0.743	−11.86%
Actual lithium demand	3,697,860	3,014,310	−18.49%	920,576	−75.11%	998,467	−73.00%
Lithium resource recovery rate	0.023	0.205	791.30%	0.206	795.65%	0.138	500.00%

shows the statistics for the effect of the combined strategy on each variable. As shown in Scheme 2, it has the greatest effect on each variable relative to the initial state: the lithium supply–demand balance increases from 0.106 to 0.426, lithium resource chain risk decreases from 1.156 to 0.939, and lithium criticality decreases from 0.843 to 0.684, with changes of 301.89%, 18.77%, and 18.86%, respectively. This indicates that the joint strategy of battery recovery policy and technological progress has the greatest effect on the lithium supply–demand balance, lithium criticality, and lithium resource chain risk. Actual lithium demand decreases from 3.69786 × 10⁶ to 9.20576 × 10⁵ t, and the lithium resource recovery rate increases from 0.023 to 0.206, changes of 75.11% and 795.65%, respectively. Scheme 3 reduces the effect of scheme 2 relative to the initial state. Although secondary use makes full use of battery capacity and has certain economic benefits, it lags behind in terms of resource benefits.

3.5. Discussion

Previous studies have found that recycling valuable metals from waste batteries can increase supply and alleviate the supply–demand gap [38]. The difference in our study is that the effect of the LIB recycling policy on the lithium supply–demand balance first increases and then decreases, and the long-term effect is limited. Some studies propose that using recycling as a tool to reduce the demand for main raw materials will become important from 2035 to 2050. They suggest that the recycling of valuable metals is not economically feasible at present, but it might reduce the risk of supply shortages in the long run [39]. The reason for this difference could be related to different considerations of the development prospects of NEVs. Under different NEV penetration rates, the demand for important metal materials will also be different. The simulation period of this study extends up to 2030, without considering the longer-term development scenario of NEVs. Next, even under the high LIB recycling policy, the development of vehicle electrification will be limited by lithium storage capacity [40], which is an issue that needs further consideration in the future.

Golroudbary [41] studied the sustainability of recycling key materials for LIBs in the two aspects of energy consumption and greenhouse gas emissions based on SD. They suggested that recycling is conducive to solving the shortage of key materials and severe environmental pollution [42], but there is high energy consumption. Studies have also noted that recycling a ton of lithium will emit 5–6 t of CO₂ [43]. This study confirms the resource benefits of LIB recycling, but the environmental benefits need to be further explored. One of the biggest challenges in battery recycling today is the environmental impact of the recycling process [44], and under the general recycling system, it is difficult for the amount of critical materials recycled to meet the large increase in resource demand [45], which is why this study proposes a strategy for technological progress in battery production. According to the existing material flow analysis, the reuse of LIBs delays the recycling of secondary raw materials [46]. This might cause SRs, which is the same result as that found in the policy research on improving the recycling rate in Section 3.3. Furthermore, this model assumes a perfect substitution of primary materials with recycled materials, which may overestimate the associated environmental benefits [47]. Therefore, trade-offs need to be made between the economic value and environmental and resource performance of lithium recovery [48].

Unlike this article, [49] points to advances in batteries to improve recyclability and refers to designing a battery that is easier to recycle. Some suggest that compared with recycling technology, alternative technology breakthroughs can better alleviate the problem of insufficient supply in the market [50]. However, it is not easy to find alternative materials; thus, this study proposes the strategy of technological progress in LIB production, reducing the consumption intensity of LIBs, and reducing the demand for lithium from the source. This could be a better long-term strategy.

4. Conclusions

This study develops a system dynamics (SD) model to evaluate the resource benefits of lithium-ion battery (LIB) recovery from new-energy vehicles (NEVs). The model incorporates the impacts of LIB recovery processes, advancements in LIB production technology, and LIB recycling and reuse strategies on three key factors: lithium supply chain risk, lithium supply–demand balance, and lithium criticality. The results show the following:

(1) LIB recycling policies can improve the lithium resource recovery rate. By 2030, the lithium resource recovery rate could reach 20.5% under a high LIB recycling policy. Recycling can reduce lithium resource supply chain risk by influencing lithium criticality, and the effect on the lithium supply–demand balance shows a trend of first increasing and then decreasing.

(2) Improving LIB production technology can reduce lithium consumption intensity and reduce actual lithium demand resources by 69.45%. By 2030, the demand for lithium will be 2.5561 × 10⁶ t. Technological progress greatly affects the lithium supply–demand balance but has little effect on lithium criticality. Lithium supply chain risk decreases by 9.28%, mainly affected by the balance of lithium supply and demand.

(3) The cascade use policy for LIB recycling can improve the cascade use, reaching 1.91124 × 10⁵ t at the end of the simulation, showing a certain economic benefit. At the same time, lithium recovery is reduced by 1.59505 × 10⁵ t, the lithium resource recovery rate is reduced by 36.84%, and lithium resource supply chain risk is increased by 1.71%, with reverse effects in terms of both economic and resource benefits.

(4) Regarding the resource benefit of recycling key metal lithium, the joint strategy of improving LIB recovery and LIB production technology is the best solution. It can improve the lithium supply–demand balance and reduce lithium criticality and resource supply chain risk. At the end of the simulation (2030), the supply–demand balance and the lithium resource recovery rate increase by 301.89% and 795.65%, respectively, while lithium resource chain risk, criticality, and actual lithium demand decrease by 18.77%, 18.86%, and 75.11%, respectively.

Currently, while battery recycling for new-energy vehicles has garnered significant attention, the recycling system remains underdeveloped and faces numerous challenges. Firstly, the recycling channels are fragmented, with a substantial number of used batteries entering informal channels. These small-scale operations often lack professional recycling technologies and environmental protection equipment, not only resulting in low resource recovery rates, but also exacerbating environmental pollution issues. Secondly, the absence of standardized recycling protocols poses a significant challenge. Disparities in technical requirements and quality inspection standards among different enterprises have become a major obstacle to the industry’s standardized development.

Given these challenges, it has become imperative to conduct in-depth research on new-energy vehicle battery recycling strategies. The development of an efficient, environmentally friendly, and sustainable battery recycling system is crucial for improving recycling standards. This initiative not only contributes to environmental protection, but also serves as a vital approach to resource conservation and recycling. Moreover, it represents a core element in promoting the sustainable development of the new-energy vehicle industry.

Based on our findings, the government should strengthen LIB recycling efforts, establish an effective waste LIB recycling system, and form a government-dominated pattern. In LIB recycling, it is necessary to clarify corporate responsibilities and increase publicity to raise public awareness of environmental protection. Again, technological innovation in LIB production should be encouraged. With the popularity of NEVs, the demand for lithium and other scarce metal resources has increased. On the one hand, battery manufacturers should seek alternative materials and a wider range of battery technology combinations; on the other hand, considering the difficulty of finding alternative materials in the short term, enterprises should continue to innovate LIB production technologies, improve the energy density and durability of LIBs, and reduce lithium consumption intensity per unit of battery capacity. At the same time, it is necessary to promote the cascade use of scrapped LIBs, make full use of battery capacity, and relieve the pressure of LIB recycling. Although cascade use might delay the recovery process for valuable metal lithium, it should be supported and promoted because of its social and economic benefits.

This study mainly considers the resource benefits of recycling valuable metal lithium from waste LIBs. However, the rare metals in batteries include not only lithium, but also beryllium and rubidium. Furthermore, the detailed economic, environmental, and social benefits, as well as the financial development [51] of recycling are outside the scope of this study, and are suggested as key focuses of future works.