Modeling and Analysis of Carbon Emissions Throughout Lifecycle of Electric Vehicles Considering Dynamic Carbon Emission Factors

Abstract

Amidst the global strategic transition towards low-carbon energy systems, electric vehicles (EVs) are pivotal for achieving deep decarbonization within the transportation sector. Consequently, enhancing the scientific rigor and precision of their life-cycle carbon footprint assessments is of paramount importance. Addressing the limitations of existing research, notably ambiguous assessment boundaries and the omission of dynamic coupling characteristics, this study develops a dynamic regional-level life-cycle carbon footprint assessment model for EVs that incorporates time-variant carbon emission factors. The methodology first delineates system boundaries based on established life-cycle assessment (LCA) principles, establishing a comprehensive analytical framework encompassing power battery production, vehicle manufacturing, operational use, and end-of-life recycling. Subsequently, inventory analysis is employed to model carbon emissions during the production and recycling phases. Crucially, for the operational phase, we introduce a novel source–load synergistic optimization approach integrating dynamic carbon intensity tracking. This is achieved by formulating a low-carbon dispatch model that accounts for power grid security constraints and the spatiotemporal distribution of EVs, thereby enabling the calculation of dynamic nodal carbon intensities and consequential EV emissions. Finally, data from these distinct stages are integrated to construct a holistic life-cycle carbon accounting system. Our results, based on a typical regional grid scenario, reveal that indirect carbon emissions during the operational phase contribute 75.1% of the total life-cycle emissions, substantially outweighing contributions from production (23.4%) and recycling (1.5%). This underscores the significant carbon mitigation leverage of the use phase and validates the efficacy of our dynamic carbon intensity model in improving the accuracy of regional-level EV carbon accounting.

1. Introduction

Global climate change, driven by anthropogenic greenhouse gas emissions, has established decarbonization as a global imperative. In response, China has formally committed to ambitious ‘carbon peak’ and ‘carbon neutrality’ targets (the ‘dual carbon’ goals) [1,2], integrating them into its overarching ecological civilization framework. Within this context, mitigating emissions from the transportation sector, a significant contributor to national carbon output, has garnered substantial attention. The rapid electrification of transport is central to this effort. By the end of 2024, China’s new energy vehicle (NEV) total reached 31.4 million units, with battery electric vehicles (BEVs) constituting a dominant 70.34% (22.09 million units). Furthermore, projections from the Ministry of Industry and Information Technology’s “New Energy Vehicle Industry Development Plan (2021–2035)” indicate that the EV fleet is poised for continued substantial growth [3]. Consequently, it is critically important to develop and refine robust methodologies for assessing the life-cycle carbon emissions of EVs. Such assessments are essential for accurately identifying key decarbonization levers within the electric mobility ecosystem, thereby supporting the achievement of stringent carbon reduction targets in the transportation sector and providing a scientific foundation for policies aimed at fostering cleaner and more sustainable low-carbon transportation systems.

Carbon emission quantification serves as an indispensable tool for evaluating the climate change impact of specific activities or products and is thus pivotal for achieving emission reduction targets and addressing the broader challenges of climate change. Prevailing carbon accounting methodologies primarily encompass three foundational approaches: the emission factor method, the mass balance method, and direct measurement. These are frequently integrated with extended techniques, such as Life-Cycle Assessment and dynamic modeling, to form comprehensive analytical frameworks. Consequently, diverse methodologies are adopted across various sectors for carbon emission assessment. For instance, Reference [4] employs a CO₂ emission factor approach, correlating the input of production materials during construction with the carbon emission factors associated with energy consumption in material processing and transportation, to calculate the life-cycle carbon emissions of road pavement construction. Addressing electricity sector emissions, Reference [5] introduces a graph computing-based framework to establish a multi-period, multi-regional system for tracking territorial carbon emission transmission, thereby quantifying electricity-related carbon emissions. In the building sector, Reference [6] leverages LCA principles in conjunction with Building Information Modeling (BIM) technology and OpenStudio energy simulation tools to assess and compare the carbon emissions of traditional and green residential buildings. Similarly, Reference [7] develops a carbon emission model for precast concrete composite slabs, focusing on their production and construction phases, by utilizing process-based inventory analysis. In the domain of wastewater treatment, Reference [8] applies both emission factor and mass balance methods to compare the carbon emission equivalents of three distinct biological nitrogen removal processes—nitrification/denitrification, anaerobic ammonia oxidation, and microalgal assimilation—specifically for treating rare earth tailings’ wastewater.

Specifically addressing research on dynamic carbon emission factors in power distribution networks, Reference [9], drawing upon the concept of power system power flow, introduces the principles of carbon emission flow and nodal carbon intensity, and establishes a theoretical framework for power system carbon emission flow analysis. Reference [10] presents a method for calculating nodal carbon intensity, which involves performing carbon flow calculations derived from conventional power system power flow analysis. Building upon the theory of power system carbon emission flow, Reference [11] quantitatively analyzes the transfer and distribution of carbon flows within the power grid. Distinct from traditional carbon flow calculations, Reference [12] introduces a Graph Convolutional Network (GCN)-based model; this model integrates neural networks to compute nodal carbon emission factors in distribution networks, leveraging potentially incomplete or uncertain measurement data. Reference [13] employs carbon flow tracing methods to determine nodal carbon intensities; based on these intensities, the authors propose a bi-level optimal dispatch strategy and formulate a corresponding bi-level optimization model for power systems. More recently, Reference [14] presented a T-Graphormer graph neural network-based model for the hourly prediction of power grid carbon emission factors, enabling the fine-grained temporal forecasting of grid carbon intensity.

In summary, the emission factor method, which calculates carbon emissions by multiplying activity data with corresponding emission factors, is well-suited for macro-scale regional or sectoral accounting, though its accuracy is contingent upon the reliability of the input data. The mass balance method, leveraging the principle of mass conservation to track carbon flows, proves particularly effective for analyzing actual emissions from industrial processes such as chemical manufacturing and steel production. Direct measurement, exemplified by Continuous Emission Monitoring Systems (CEMSs), obtains precise emission data through direct monitoring, albeit typically at a higher cost. Furthermore, Life-Cycle Assessment offers a holistic perspective by encompassing emissions across the entire value chain of products or processes, such as in building construction and product manufacturing; the integration of tools like Building Information Modeling can further enhance its analytical depth. For complex scenarios, notably within power systems, dynamic carbon flow models are increasingly employed. These include approaches based on nodal carbon intensity, Graph Convolutional Networks, and predictive models like T-Graphormer, which often integrate power flow calculations and artificial intelligence (AI) techniques to enable real-time carbon tracking and inform optimization strategies. Collectively, these methodologies span a spectrum from static to dynamic and from macro-level to micro-level assessments, providing a robust toolkit to support informed decision-making for emission reduction across diverse scenarios.

In accordance with Life-Cycle Theory, the comprehensive lifespan of an electric vehicle is typically segmented into three principal phases: manufacturing, operational use, and end-of-life recycling. Significant scholarly attention has been directed towards assessing the carbon emissions associated with these life-cycle stages. For instance, Reference [15] conducts a comparative analysis of life-cycle greenhouse gas emissions from conventional internal combustion engine vehicles (ICEVs) and EVs in India, concluding that the higher emission profile of EVs, relative to ICEVs, is predominantly attributable to large-capacity battery components, emissions inherent in electricity generation, and energy-intensive manufacturing processes. Reference [16] employs a life-cycle approach to analyze carbon emissions across the pre-manufacturing, manufacturing, operational, and end-of-life stages of EVs, and discusses the environmental burdens—primarily energy consumption and emission levels—associated with lightweight materials and design concepts in EVs. Focusing on the European market, Reference [17] applies LCT to comparatively analyze the life-cycle environmental emission characteristics of EVs and conventional fossil fuel vehicles. Specific components and processes within the EV life cycle have also been investigated. Reference [18] utilizes a Stackelberg game model to investigate issues pertaining to power battery recycling and second-life (cascade) utilization. Reference [19], adhering to LCT and employing the GREET (Greenhouse gases, Regulated Emissions, and Energy use in Technologies) model, conducts a comprehensive assessment of EVs, considering multiple dimensions including energy consumption, pollutant emissions, environmental impact loads, and environmental costs. Reference [20] employs the carbon emission factor method to calculate the total life-cycle carbon emissions for a representative EV model. Further focusing on battery technologies, Reference [21] investigates, from an LCA perspective, the carbon emissions associated with common EV power battery chemistries, namely lithium iron phosphate (LFP), nickel–cobalt–manganese (NCM/NCA) ternary lithium, lithium manganese oxide (LMO), and lithium titanate (LTO) batteries. Similarly, Reference [22] utilizes a life-cycle model to perform a comprehensive LCA for ternary lithium and LFP batteries.

In summary, existing research on electric vehicle carbon emissions, predicated on Life-Cycle Assessment principles, has predominantly concentrated on quantifying emissions during the power battery manufacturing and end-of-life recycling phases. However, a systematic investigation into the dynamic characteristics of operational phase carbon emissions remains notably absent. Furthermore, prevalent carbon accounting models for the operational phase typically rely on static methodologies employing regional grid average emission factors. This simplification not only disregards the real-time influence of dynamic source–load interactions within the power system on carbon flow distribution but also fails to capture the spatiotemporal heterogeneity of marginal emission factors, which arise from fluctuating renewable energy generation and evolving unit commitment statuses. To address these limitations, this study develops a dynamic carbon emission assessment model that innovatively incorporates a source–load synergistic optimization mechanism. Specifically, for the operational phase modeling, we propose a low-carbon optimal dispatch strategy, grounded in carbon emission flow theory. This strategy explicitly accounts for unit commitment statuses, generation unit output characteristics, wind power integration constraints, and the flexible load profiles of EV charging and discharging, thereby enabling the dynamic tracking of carbon flow distribution across the generation, transmission (grid), and consumption (load) segments of the power system. Building upon this, dynamic nodal carbon emission factors are then utilized to precisely quantify EV operational emissions. Finally, by integrating these operational emission data with those from the manufacturing and recycling phases, the comprehensive life-cycle carbon footprint of EVs is determined.

The rest of the paper is structured as follows. Section 2 analyzes dual-settlement mechanisms and deviation penalty rules, proposing a two-stage optimization methodology. Section 3 develops the EHES participation model in energy-reserve markets with DR integration and establishes an EV charging response model considering economic and temporal factors, accompanied by the use of real-time scheduling strategies. Section 4 validates model effectiveness through numerical simulations; this is followed by the presentation of conclusions in Section 5.

2. Life-Cycle Carbon Emission Analysis Framework for Electric Vehicles

Based on Life-Cycle Theory, the assessment of electric vehicle carbon emissions is typically structured around four key stages, as illustrated in Figure 1: (1) raw material acquisition; (2) manufacturing and assembly; (3) operational use; and (4) end-of-life recycling and disposal. For the purposes of our modeling and discussion, we will sometimes group the first two stages into a comprehensive “production phase”.

The manufacturing phase of EVs is broadly categorized into component production and final vehicle assembly. Component production primarily encompasses the manufacturing of lithium-ion batteries, electric drive motors, body-related components, and various interior trim and functional parts. Final assembly involves processes such as body painting, body assembly, electric powertrain integration, interior fit-out, and final vehicle testing and commissioning. Throughout this phase, substantial direct carbon emissions arise from energy consumption during raw material acquisition and component manufacturing processes. Additionally, indirect carbon emissions are generated from the electricity consumed during assembly operations. Given that the power battery constitutes a critical and high-impact module of an EV, its production phase carbon emissions are non-negligible and warrant specific attention. Consequently, in assessing production phase emissions, this study distinctly considers the EV chassis and body separately from the power battery.

Lithium-ion power batteries have emerged as the core components of EV onboard energy storage systems, owing to their high energy density, compact structural design, and low self-discharge rate. Currently, predominant power battery chemistries include lithium–iron–phosphate batteries, ternary lithium batteries, lithium–manganese oxide batteries, and lithium–titanate batteries. This study will, therefore, exemplify the carbon emission calculations by focusing on these four prominent EV power battery types.

Although the operational use phase of EVs involves no direct tailpipe emissions, the indirect carbon emissions induced by electricity consumption for charging necessitate thorough upstream attribution analysis. These emissions primarily originate from two sources: (i) the carbon intensity of electricity generation, particularly in grids still heavily reliant on fossil fuels, and (ii) the additional generation required to compensate for energy losses within transmission and distribution networks, which contributes further to the carbon footprint. Acknowledging these emission characteristics, the present research develops a dynamic carbon accounting model predicated on vehicle–grid interaction principles. Currently, vehicle–grid interaction encompasses three primary modes: uncontrolled charging, smart charging, and bidirectional V2G power flow [23]. This study will calculate operational phase carbon emissions for EVs under each of these three vehicle–grid interaction scenarios.

The end-of-life recycling phase for EVs involves the systematic dismantling of decommissioned vehicles and their components for material recovery and reuse. The recycling of the EV body typically begins with the collection of decommissioned vehicles and their transportation to specialized dismantling centers. Here, they undergo disassembly, followed by the separate recovery and processing of constituent materials; for instance, metallic components can be remelted and repurposed, plastics are reprocessed into new materials, and glass is crushed for use in new glass products [24]. For computational tractability in this study, the analysis of body recycling primarily focuses on the remelting and reuse of metallic materials. Current management strategies for power batteries at their end of life predominantly include second-life (or cascade) use, hydrometallurgical recovery, direct physical recycling, pyrometallurgical recovery, and hybrid physical/chemical methods. This paper will consider the first three of these—second-life use, hydrometallurgical recovery, and direct physical recycling—as representative end-of-life management pathways for power batteries [25].

3. Modeling of Electric Vehicle Life-Cycle Carbon Emissions

3.1. Carbon Emission Model for the Production Phase

The model for quantifying carbon emissions during the electric vehicle production phase is developed based on the well-established emission factor method, a standard and widely accepted approach in Life-Cycle Assessment [26]. This method calculates emissions by multiplying activity data (e.g., mass of material or energy consumed) with corresponding emission factors. In our study, the specific parameters and application of this methodology for EV production are adopted from the comprehensive work by Kang et al. [20]. The model is expressed as follows:

$$C_{produce}=C_{material}+C_{assemble}$$

(1)

C_produce represents the total carbon emissions associated with the electric vehicle production phase, C_material denotes the carbon emissions attributable to the material acquisition phase, and C_assemble signifies the carbon emissions incurred during the manufacturing and assembly phase.

Carbon emissions attributable to the acquisition phase of key constituent materials for electric vehicles are formulated as follows:

$$C_{material1}={\displaystyle \sum _{i=1}^7{p}_{1i}{q}_{1i}}$$

(2)

C_material1 represents the carbon emissions from the acquisition phase of principal materials for the electric vehicle, p_1i denotes the mass of the i-th principal material used in the EV, and q_1i signifies the carbon emission factor associated with the i-th principal material for the EV.

The expression for carbon emissions attributable to the manufacturing and assembly phase of the electric vehicle (EV) body, excluding the power battery, is as follows:

$$C_{assemble1}={\displaystyle \sum _1^a{\displaystyle \sum _1^b{d}_{ij}{e}_j}}$$

(3)

C_assemble1 denotes the total carbon emissions generated during the manufacturing and assembly of the EV body, d_ij represents the quantity of the j-th energy type consumed by the i-th component of the EV body during this phase, and e_j is the carbon emission factor corresponding to the j-th energy type. In this study, ‘a’ represents the number of primary vehicle components considered, which is set to 7, and ‘b’ signifies the number of principal energy types consumed during manufacturing and assembly, which is set to 2.

For four typical types of power batteries, the expression for carbon emissions during their material acquisition phase is as follows:

$$C_{material2}={\displaystyle \sum _1^n{p}_{2i}{q}_{2i}}$$

(4)

C_material2 represents the carbon emissions attributable to the material acquisition phase of the electric vehicle power battery, p_2i denotes the mass of the i-th material acquired for the power battery, and q_2i signifies the carbon emission factor associated with the i-th material acquired for the power battery.

For these same four typical power batteries, the expression for calculating carbon emissions during their manufacturing and assembly phase is as follows:

$$C_{assemble2}={\displaystyle \sum _1^b{g}_j{h}_j}$$

(5)

C_assemble2 represents the carbon emissions generated during the manufacturing and assembly phase of the EV power battery, g_j denotes the quantity of the j-th energy type consumed during the power battery manufacturing and assembly phase, and h_j is the carbon emission factor corresponding to the j-th energy type consumed during power battery manufacturing and assembly. The parameter ‘b’ (set to 2 here) signifies the number of principal energy types consumed, consistent with its previous definition for the EV body assembly.

3.2. Carbon Emission Model for the Electric Vehicle Operational Phase

During the operational phase of electric vehicles, this study focuses on the distribution network level, employing nodal carbon intensities to calculate the associated carbon emissions. At this distribution network level, we propose an optimal dispatch strategy rooted in optimal power flow principles. This strategy aims to optimize power flow distribution, modulate the output of distributed generators, and strategically allocate the aggregated electric vehicle charging and discharging loads to various nodes within the distribution network. The optimization variables encompass the distributed generator output in each time interval, the quantity of power procured from the upstream transmission grid, and the number of electric vehicles engaged in charging or discharging at each node. The detailed solution methodology is illustrated in Figure 2.

3.2.1. Objective Function

The objective function in this study is formulated in such a way as to minimize the overall carbon emission intensity of the distribution network. This guides the regulation of distributed generator output, decision-making regarding power procurement from the upstream grid, and the allocation of electric vehicle charging and discharging loads. The objective function is presented as shown in Equation (6):

$$\min C{O}_{2,total}=P_{\mathrm{g}rid,t}\times C_t+{\displaystyle \sum _{i=1}^i{\displaystyle \sum _{t=1}^t{D}_i\times P_{DG,i,t}}}$$

(6)

P_grid,t represents the quantity of electricity procured from the upstream grid at time t, C_t denotes the carbon intensity of the electricity purchased from the upstream grid at time t, D_i signifies the carbon emission factor of the i-th distributed generator, and P_DG,i,t is the power output of the i-th distributed generator at time t.

It is important to clarify that wind energy is not treated as an isolated power source or the sole option for EV charging. As illustrated in the system topology (Figure 3) and formulated in the constraints, wind power is modeled as a grid-integrated renewable energy source. It operates in parallel with power procured from the upstream transmission grid and other dispatchable distributed generators. The optimization model co-optimizes all these generation sources to meet the system load, including the EV charging demand, while pursuing the objective of minimizing overall carbon emissions.

3.2.2. Constraints

The optimization model is subject to a set of physical and operational constraints for both the power grid and the electric vehicle fleet. The following power system operational constraints Equations (9)–(14) are standard formulations in economic dispatch and optimal power flow studies. They are presented here in a form consistent with the principles of carbon flow analysis, as established in foundational studies such as [9]. Equations (15)–(18), showing the constraints of the EV fleet, model the aggregate charging/discharging behavior and the battery state of the charge dynamics. These are standard models used in vehicle–grid integration research, as detailed in studies on integrated scheduling strategies [1].

(1)

Generator Constraints:

$$P_{\min }\le P_{DG}\le P_{\max }$$

(7)

P_min is the minimum power output of the distributed generator. P_max is the maximum power output of the distributed generator.

(2)

Wind Power Constraints:

$$0\le P_{WT}\le P_{WT\_f}$$

(8)

P_WT is the power output of the wind turbine generator. P_{WT_f} is the forecasted maximum power output of the wind turbine. This constraint defines the generation strategy for wind power within our optimization framework. Wind power is treated as a non-dispatchable, intermittent resource. The strategy is to maximize its utilization by accepting all available power up to the forecasted limit (P_{WT_f}) at each time interval. This approach, often referred to as a ‘must-take’ strategy for renewables, is common in power system dispatch strategies. The optimization algorithm then schedules the other dispatchable resources (grid power, distributed generators, and EV V2G) around this variable generation to achieve the overall low-carbon objective.

(3)

Grid Power Purchase Constraints:

$$100\le P_{\mathrm{grid}}\le {\displaystyle \sum _j{P}_{\mathrm{load},\mathrm{m}}}\ast 1.2$$

(9)

P_load,m represents the base load demand at node m.

(4)

Line Power Flow Constraints:

$$-2500\le P_{\mathrm{bra},\mathrm{j}}\le 2500$$

(10)

P_bra,j represents the active power flow on branch j.

(5)

PTDF (Power Transfer Distribution Factor) Constraints:

$$\begin{array}{l}-1\times {10}^{-4}\le P_{\mathrm{bra},t }-\\ PTDF\ast P_{inj,t}\le 1\times {10}^{-4}\end{array}$$

(11)

$$\begin{array}{c}{P}_{inj,t}=P_{grid,t}+P_{WT,t}+P_{DG,t}\\ \hspace{0.25em}\hspace{0.25em}\hspace{0.25em}\hspace{0.25em}\hspace{0.25em}\hspace{0.25em}+P_{EVd,t}-P_{load,t}-P_{EVc,t}\end{array}$$

(12)

P_inj,t represents the net power injection at the distribution network node at time t. P_EVd,t represents the total discharging power of electric vehicles at time t. P_EVc,t represents the total charging power of electric vehicles at time t.

(6)

Nodal Power Balance Constraints:

$$\begin{array}{l}-1\times {10}^{-4}\le P_{grid,t}+P_{WT,t}+P_{DG,t}\\ +P_{EVd,t}+P_{bra,to,t}-P_{load,t}-P_{EVc,t}\\ -P_{bra,from,t}\le 1\times {10}^{-4}\end{array}$$

(13)

P_bra,from,t represents the power flowing out from branches connected to the node at time t. P_bra,to,t represents the power flowing into branches connected to the node at time t.

(7)

Generator Ramping Constraints:

$$\begin{array}{c}-P_{DG, delta}\le P_{DG,t+1}-P_{DG,t}\\ \hspace{0.25em}\hspace{0.25em}\hspace{0.25em}\hspace{0.25em}\le P_{DG, delta}\end{array}$$

(14)

P_DG,delta represents the maximum allowable ramp rate (up or down) for the generator.

(8)

Basic Electric Vehicle Constraints:

$$\begin{array}{l}0\le Nc\le 50\\ 0\le Nd\le 50\end{array}$$

(15)

N_c represents the number of EVs charging at a specific node. N_d represents the number of EVs discharging at a specific node.

(9)

EV Charging/Discharging State of Charge (SOC) Constraints:

$$\begin{array}{l}{\mathrm{SOC}}_1={\mathrm{SOC}}_{\mathrm{init}}\\ +{\displaystyle \frac{({\mathrm{Nc}}_1{P}_c{\eta }_c-{\mathrm{Nd}}_1{P}_d/{\eta }_d)\mathsf{\Delta }t}{C}}\\ {\mathrm{SOC}}_t={\mathrm{SOC}}_{t-1}\\ +{\displaystyle \frac{({\mathrm{Nc}}_t{P}_c{\eta }_c-{\mathrm{Nd}}_t{P}_d/{\eta }_d)\mathsf{\Delta }t}{C}},\\ \forall t=2,\dots ,T\end{array}$$

(16)

$$\begin{array}{l}{\mathrm{SOC}}_{\min }\le {\mathrm{SOC}}_t\le {\mathrm{SOC}}_{\max },\\ \forall t=1,\dots ,T\end{array}$$

(17)

$$\begin{array}{l}{\mathrm{SOC}}_t+{\displaystyle \frac{{\mathrm{Nc}}_t{P}_c{\eta }_c\mathsf{\Delta }t}{C}}\le {\mathrm{SOC}}_{\max },\\ \forall t=1,\dots ,T\\ {\mathrm{SOC}}_t-{\displaystyle \frac{{\mathrm{Nd}}_t{P}_d\mathsf{\Delta }t}{{\eta }_{d}C}}\ge {\mathrm{SOC}}_{\min },\\ \forall t=1,\dots ,T\end{array}$$

(18)

SOC_t is the battery State of Charge at time period t. Nc_t is the number of charging instances during time period t. Nd_t is the number of discharging instances during time period t. SOC_init is the initial SOC of the EVs. η_c is the EV charging efficiency. η_d is the EV discharging efficiency.

3.3. Carbon Emission Model for the Recycling Phase

The recycling phase is disaggregated into two primary components: the recycling of the vehicle body and the recycling of the power battery. Consequently, the total carbon emissions for the recycling phase are formulated as follows:

$$C_{recycle}=C_{body}+C_{battery}$$

(19)

C_recycle represents the total carbon emissions from the EV recycling phase. C_body denotes the carbon emissions generated during the recycling of the EV body. C_battery signifies the carbon emissions associated with the recycling of the EV power battery.

The carbon emissions attributable to the recycling of the EV body are calculated as follows:

$$C_{body}=m_i{\displaystyle \sum _{i=1}^4{\displaystyle \sum _{j=1}^3{k}_i{p}_{ij}{q}_{ij}}}$$

(20)

m_i is the mass of the i-th type of metal in the EV body undergoing recycling. k_i is the recovery rate for the i-th type of metal during the EV body recycling process. p_ij is the quantity of the j-th type of energy consumed per unit mass of the i-th metal recovered. q_ij is the carbon emission factor for the j-th type of energy consumed during the processing of the i-th metal in EV body recycling.

The carbon emissions originating from the recycling of the EV power battery are determined by the following equation:

$$C_{battery}={\displaystyle \sum _{i=1}^3{y}_i{z}_i}$$

(21)

y_i represents the energy consumption associated with the i-th specific recycling pathway or technology employed for the EV power battery. z_i is the carbon emission factor corresponding to the energy consumed by the i-th recycling pathway for the EV power battery.

4. Example Analysis

4.1. Basic Data

To quantify carbon emissions, this study selects a representative electric vehicle model with a curb weight of 2098 kg and a residual mass of 1566 kg. The mass proportions of the primary materials constituting the electric vehicle body are detailed in

Table 1. Proportion of various raw materials in total mass of electric vehicles. The data are adapted from Kang et al. [20].

Material	Steel	Iron	Aluminum	Copper	Plastics	Glass	Rubber	Other
Mass Proportion (%)	65.2	6.0	6.9	4.3	10.6	2.9	1.9	2.6

. For the production phase, the carbon emission factors for the major raw materials that constitute the EV body are critical inputs. These values, which represent the cradle-to-gate emissions for producing 1 kg of each material, are presented in

Table 2. Carbon emission factors for major raw materials of the EV body. The data are adapted from Kang et al. [20].

Material	Steel	Iron	Aluminum	Copper	Plastics	Glass	Rubber
Carbon Emission Factor	6.82	0.402	15.8	18.38	4.136	1.241	2.758

. Similarly, for the end-of-life recycling phase, the analysis requires data on material recovery rates and the specific energy consumed during recycling processes. This recycling inventory, detailing the energy consumption (broken down by source) per kilogram of recycled material, is provided in

Table 3. Recycling inventory for the EV Body. The data are adapted from Kang et al. [20].

		Steel	Cast Iron	Copper	Aluminum
Recovery Rate (%)		90%	80%	90%	92%
Energy Consumption	Natural Gas	0.02	-	-	0.15
	Electricity	4.23	2.24	9.54	0.80
	Coal	-	8.24	-	-

. For the calculation of carbon emissions during both the production and recycling phases, parameters and methodologies are adopted from Reference [20]. The carbon emission factor for electricity consumed in these phases is based on China’s national average grid emission factor for the year 2021, 0.624 tCO₂/MWh specifically. The case study simulations and optimization analyses were implemented within the MATLAB R2021a environment, utilizing the Yalmip toolbox in conjunction with the Gurobi solver.

For the calculation of carbon emissions during the electric vehicle operational phase, this study evaluates four distinct scenarios:

Scenario 1: Calculation based on the theoretical driving range of a representative electric vehicle model, utilizing the average grid electricity carbon emission factor.

Scenario 2: Calculation assuming uncoordinated electric vehicle charging behavior, again using the average grid electricity carbon emission factor.

Scenario 3: Calculation based on optimally scheduling electric vehicle charging to minimize total load fluctuations on the distribution network, with emissions assessed using the average grid electricity carbon emission factor.

Scenario 4: Calculation based on optimally scheduling both electric vehicle charging and discharging with the objective of minimizing the overall carbon emission intensity of the distribution network. Emissions in this scenario are assessed using dynamic nodal carbon intensities.

For Scenario 4, the analysis is conducted using MATLAB. Initially, power flow calculations are performed to determine the steady-state power flow of the distribution network using the Power Transfer Distribution Factor matrix. Subsequently, nodal carbon intensities are derived from these power flow results. The IEEE 33-bus distribution test system is adopted as the network topology for the power system, a schematic of which is presented in Figure 3. Within this system, Node 1 is designated as the point for power exchange with the upstream transmission grid. Distributed generators are connected at Nodes 2, 4, and 15, while wind power generation units are integrated at Nodes 10 and 27.

4.2. Analysis of Carbon Emissions During the Production Phase

Based on the established model and calculation formulas for the electric vehicle body production phase, the carbon emissions attributable to the material acquisition stage are determined to be 10,771 kg, while emissions generated during the manufacturing and assembly stage of the vehicle body amount to 874.70 kg. Collectively, the total carbon emissions for the electric vehicle body production phase are calculated to be 11,645.70 kg.

Similarly, applying the model and calculation formulas for the electric vehicle power battery production phase, the carbon emissions during the manufacturing and assembly stage are found to be 1496.91 kg for lithium–iron–phosphate batteries, 1456.70 kg for ternary lithium batteries, 1916.94 kg for lithium–manganese oxide batteries, and 4244.98 kg for lithium titanate batteries. The carbon emissions associated with the material acquisition stage for the individual components of these four power battery types—namely LFP (lithium–iron–phosphate), NCM (nickel–cobalt–manganese), LMO (lithium–manganese oxide), and LTO (lithium titanate)—as well as their respective total material acquisition emissions, are detailed in Figure 4. Given that lithium–iron–phosphate batteries exhibit the lowest production phase carbon emissions among the assessed types and are extensively utilized, subsequent analyses in this study will focus on lithium–iron–phosphate batteries as the representative case. Consequently, by integrating the emissions from the material acquisition (as detailed in Figure 4) and the manufacturing and assembly stages, the total production phase carbon emissions for each of the specified power battery chemistries can be comprehensively determined.

In this study, it is assumed that electric vehicles equipped with the four distinct types of power batteries share an identical, representative vehicle body structure. Consequently, the carbon emissions generated during the component material production stage for the vehicle body are considered uniform across these four electric vehicle types, amounting to 10,771 kg. Therefore, incorporating emissions from power battery production, a lithium–iron–phosphate battery-equipped electric vehicle exhibited total production phase carbon emissions of 15,323.79 kg.

4.3. Analysis of Carbon Emissions During the Operational Phase

For the operational phase analysis, the following parameters are assumed for the electric vehicles: a power battery capacity of 40 kWh, a round-trip charging/discharging efficiency of 0.95, and a maximum charging/discharging power rate of 5 kW. The initial SOC for each electric vehicle power battery is assumed to be randomly distributed within the range of 0.2 to 0.8.

Scenario 1: In cities such as Shenzhen, electric vehicle models like the BYD E6 are extensively used as taxis. Many of these vehicles have achieved operational mileages exceeding 500,000 km, with some instances reaching 800,000 to 1,000,000 km (typically requiring one to two battery replacements). Taking the BYD E6 as a representative example and assuming a total operational lifespan mileage of 500,000 km, the total carbon emissions during the use phase are calculated to be 81,524.98 kg. This calculation utilizes the 2021 national average electricity carbon emission factor.

Scenario 2: This scenario assumes that electric vehicle user behavior mirrors that of conventional internal combustion engine vehicle users. Under this assumption, the daily driving mileage of privately owned electric vehicles is modeled to follow a log-normal distribution. Considering a total daily population of 200 electric vehicles within the analyzed distribution network area, the aggregated uncoordinated charging power profile for the electric vehicle fleet across a 24 h period is depicted in Figure 5.

Following the simulation in Scenario 2, the total charging demand for the 200 electric vehicles over a 24 h period is calculated to be 9576.90 kW. This equates to an average daily charging power demand of 47.89 kW per vehicle. Consequently, the average daily energy consumption per vehicle is determined to be 47.89 kWh. To estimate the lifetime operational carbon emissions, the typical warranty period of commercially available electric vehicles is used as a proxy for their operational lifespan. Assuming an operational lifespan of approximately 8 years, the total energy consumed throughout the electric vehicle’s operational phase is first estimated. Subsequently, using a carbon emission factor of 0.16 kg CO2-eq per MJ of electricity consumed, the annual carbon emissions for a single electric vehicle are calculated to be 9942.53 kg. Therefore, the total carbon emissions over the entire operational phase for one such vehicle amount to 79,540.24 kg.

Scenario 3: This scenario assumes that 200 electric vehicles within the distribution network area participate in a coordinated charging scheme. The resulting optimized charging load profile for the electric vehicle fleet over a 24 h period, alongside the conventional load of the distribution network and the combined total network load, is illustrated in Figure 6.

Following the optimization, the total energy demand for the 200 participating electric vehicles over a 24 h period is calculated to be 6744.47 kWh. This corresponds to an average daily energy consumption of 33.72 kWh per vehicle. Based on this daily energy consumption and using the previously defined carbon emission factor (0.16 kg CO₂-eq per MJ, or its kWh equivalent), the annual carbon emissions for a single electric vehicle under this coordinated charging strategy are calculated to be 7000.60 kg. Consequently, the total carbon emissions over the entire operational lifespan (assumed to be 8 years) for one such vehicle are estimated to be approximately 56,005.0 kg.

As depicted in Figure 6, by implementing an ordered charging strategy to manage the charging schedules of electric vehicles, these vehicles are preferentially charged during off-peak electricity demand periods and avoid charging during peak demand hours. This approach effectively reduces the peak-to-valley difference in the distribution network’s load profile and mitigates load fluctuations, thereby enhancing the operational security and stability of the distribution network.

Scenario 4: In this scenario, the objective is to minimize the overall carbon emission intensity of the distribution network. This is achieved by concurrently optimizing the power output of distributed generators, the quantity of electricity purchased from the upstream transmission grid, and the number of electric vehicles charging or discharging at each node.

Following this optimization, the hourly number of electric vehicles engaged in charging and discharging at each node are depicted in Figure 7 and Figure 8, respectively. The simulation, conducted on the IEEE 33-bus test system, yields the nodal carbon intensity distribution across the network, as illustrated in Figure 9. Figure 9 visualizes the spatiotemporal distribution of nodal carbon intensity, revealing significant variations both by time of day and by network location. Notably, carbon intensity is lowest (blue) during off-peak hours and at nodes near renewable generation (e.g., Nodes 10 and 27), while it peaks (yellow and red) during high-demand periods at nodes supplied by higher-carbon sources. This detailed, dynamic information is precisely what our model leverages to strategically schedule EV charging and discharging, thereby validating our approach of using dynamic carbon signals to minimize emissions.

Based on these optimized operations and the calculated dynamic nodal carbon intensities, the average daily carbon emissions specifically attributable to the charging activities of a single electric vehicle are determined to be 14.46 kg. Extrapolating from this daily figure, the annual carbon emissions for one electric vehicle under this strategy amount to 5277.90 kg. Consequently, the total carbon emissions over the entire operational lifespan (assumed to be 8 years) for one such vehicle are estimated to be approximately 42,223.20 kg.

To validate the effectiveness of our proposed optimization framework and quantify its emission reduction potential, we designed four distinct scenarios for comparison. Scenario 2 represents the baseline case of uncoordinated charging, reflecting a lack of optimization. In contrast, Scenarios 3 and 4 demonstrate the performance of our optimization model under different strategies. This comparative analysis serves to prove that our approach provides tangible optimization benefits beyond simple emission calculations. A comparative summary of the electric vehicle use-phase carbon emissions under all four scenarios is presented in

Table 4. Carbon emissions during the use of electric vehicles in four scenarios.

Scenario	Operational Phase Carbon Emissions (kg CO2-eq)
Scenario 1	81,524.98
Scenario 2	79,540.24
Scenario 3	56,005.40
Scenario 4	42,223.20

.

As indicated in Table 2, Scenario 1 yields the highest carbon emissions during the electric vehicle’s operational phase. However, the emission calculation for this scenario solely considers electric vehicle charging, without incorporating V2G functionalities or the nuances of user charging behaviors. Furthermore, it exclusively employs the national average electricity carbon emission factor, neglecting the influence of dynamic carbon emission factors. These limitations arguably reduce the conclusiveness of the results obtained under Scenario 1 for representing typical or optimized operational impacts.

Comparing the optimized scenarios, Scenario 3 demonstrates a 29.6% reduction in total carbon emissions relative to the uncoordinated charging of Scenario 2. More substantially, Scenario 4 achieves a 46.9% reduction in total carbon emissions compared to Scenario 2. These findings underscore the significant contributions of both coordinated electric vehicle charging and V2G participation to carbon emission mitigation. Notably, the results suggest that V2G integration offers a more pronounced carbon reduction effect.

4.4. Analysis of Carbon Emissions During the Recycling Phase

For this study, the total recyclable mass of the electric vehicle is assumed to be 1566 kg. The material composition of this recyclable mass includes steel at 65.2%, cast iron at 6.0%, aluminum at 6.9%, and copper at 4.3%. The carbon emissions generated from the recovery of these various metals during the electric vehicle body recycling process are detailed in

Table 5. Carbon emissions generated by recycling various metals.

Metal Type	Steel	Cast Iron	Copper	Aluminum
Carbon Emissions (kg CO2-eq)	615.72	79.25	91.35	13.83

.

Based on these data, the total carbon emissions attributable to the electric vehicle body recycling phase are calculated to be 800.15 kg. Notably, the recovery of steel accounts for the largest portion of these emissions, primarily due to its high mass proportion in the vehicle body.

Furthermore, this analysis assumes a lithium–iron–phosphate power battery with a capacity of 40 kWh. It is calculated that if this battery undergoes a typical cascade utilization pathway followed by material recovery at its end-of-life period, the associated carbon emissions for this recycling phase amount to 51.78 kg.

4.5. Analysis of Total Life-Cycle Carbon Emissions for Electric Vehicles

Aggregating the emissions from all life-cycle stages, the total life-cycle carbon emissions for the electric vehicle can be determined. A summary of these emissions across the different operational scenarios is presented in

Table 6. Carbon emissions throughout entire lifecycle of electric vehicles.

Life-Cycle Stage			Carbon Emissions (kg CO2-eq)
Electric vehicle life-cycle carbon emissions	Production phase	Vehicle body	11,645.70
		Power battery	1496.91
	Operational phase	Uncoordinated charging	79,540.24
		Coordinated charging	56,005.40
		V2G	42,223.20
	Recycling phase	Vehicle body	800.15
		Power battery	51.78
	Total life-cycle emissions (with V2G operation)		56,217.74

. For the purpose of calculating a representative “Total Life-Cycle Carbon Emissions” figure that reflects optimized operation, the value corresponding to the V2G operational scenario (Scenario 4) is used for determining the operational phase’s contribution to the grand total.

As detailed in Table 6, the operational phase is the dominant contributor to the electric vehicle’s total life-cycle carbon emissions across all scenarios considered. Specifically, operational emissions amount to 79,540.24 kg for uncoordinated charging (Scenario 2), 56,005.40 kg for coordinated charging aimed at load fluctuation minimization (Scenario 3), and 42,223.20 kg for the V2G strategy focused on network carbon intensity minimization (Scenario 4). Conversely, the recycling phase contributes the least to the life-cycle emissions, totaling 851.93 kg (comprising 800.15 kg for body recycling and 51.78 kg for lithium–iron–phosphate battery recycling). These findings clearly indicate that the operational phase possesses the most substantial potential for carbon emission mitigation in the life cycle of electric vehicles.

5. Conclusions

This study employed a life-cycle assessment framework, segmenting the electric vehicle’s lifespan into production, operational, and recycling phases. For the production and recycling phases, carbon emission models were developed using an inventory analysis approach. During the operational phase, an electric vehicle carbon emission model was established, aiming to minimize the overall carbon emission intensity of the distribution network while considering the spatiotemporal distribution characteristics of electric vehicles. Based on the comprehensive life-cycle carbon emission calculations derived from our case studies, the following key conclusions are drawn:

(1)

Within the total life cycle, when operating under a Vehicle-to-Grid (V2G) strategy, the operational phase accounts for the predominant share of carbon emissions (75.1%), substantially exceeding contributions from the production phase (23.4%) and the recycling phase (1.5%). This highlights how efforts to enhance the carbon reduction efficacy of electric vehicles should primarily concentrate on optimizing vehicle–grid interactions and dispatch strategies during the operational phase.

(2)

During the operational phase, the adoption of a Vehicle-to-Grid strategy demonstrates substantial carbon mitigation benefits, achieving emission reductions of 46.9% and 24.6% compared to uncoordinated charging and coordinated (load fluctuation minimization) charging scenarios, respectively. Such strategies not only facilitate greater integration and utilization of renewable energy sources but also markedly decrease the overall life-cycle carbon footprint of electric vehicles.

(3)

The core contribution of this study is the development of an integrated framework that dynamically links grid-level carbon flows with a comprehensive LCA. While we do not compare our optimization model against other specific algorithms from the literature, our scenario-based analysis provides clear evidence of its value. By comparing the optimized V2G strategy (Scenario 4) against the uncoordinated charging baseline (Scenario 2)—a proxy for conventional static LCA approaches—we demonstrate a substantial 46.9% reduction in operational phase carbon emissions. This finding proves that our proposed source–load synergistic optimization is not merely a calculation tool but an effective strategy for emission mitigation, highlighting the importance of dynamic modeling in unlocking the full decarbonization potential of EVs.

Our study’s primary novelty lies not in identifying the dominance of the operational phase, but in the dynamic methodology used to analyze it. By capturing the spatiotemporal heterogeneity of carbon intensity—a factor static models using single average emission factors cannot establish—our approach moves beyond mere calculation. It provides a more accurate assessment and, crucially, unlocks actionable optimization strategies that are otherwise invisible, demonstrating clear advancement compared to prior methods.

It is important to note that the 46.9% emission reduction identified represents a theoretical optimum. Achieving this in practice is contingent upon overcoming significant real-world barriers to V2G adoption, including infrastructure costs, policy development, and user acceptance. Our findings therefore underscore the substantial decarbonization potential of V2G, motivating further research and policy support to address these challenges.

While this study provides a comprehensive framework, certain limitations point to avenues for future research. For instance, our model assumes constant battery capacity and efficiency over the vehicle’s lifespan. We concur with the reviewer’s insightful recommendation that real-world battery degradation affects energy capacity and round-trip efficiency, influencing both optimal charging strategies and lifetime carbon emissions in turn. Future work should incorporate a detailed battery aging model to capture these long-term dynamics, providing a more nuanced analysis of how EV carbon footprints evolve over their entire operational life. Furthermore, while the current optimization was solved using a commercial solver to ensure optimality for this offline analysis, real-time implementation would necessitate more computationally efficient methods. As the reviewer aptly suggests, the integration of advanced control algorithms, such as genetic algorithms or other metaheuristics, could be explored to find near-optimal solutions with significantly reduced computational time. This would make the proposed low-carbon dispatch strategy more viable for online, real-world applications and represents a promising path for future development.