Coupled Modeling of Vehicle Fleet Renewal Policies and Urban Environmental Corrosion: Dynamic Emission Trajectories and Infrastructure Coating Durability

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This article concerns modeling of vehicle fleet renewal policies. The authors analyze the coupling mechanism between vehicle replacement policies and urban environmental corrosion, following a logical sequence of data collection—model construction—empirical identification, and policy optimization. The authors address an important topic. In this article, they use analytical methods to find the optimal solution.

Detailed comments and questions.

1. There are no references to the authors' previous articles in the References. Do the authors have any previous work in this field? What topics have the authors covered so far?
2. Abstract and conclusion no. 2: The authors refer to the value of 1.5 years, but this value isn't present in this article. Please, clarify this issue.
3. The article is divided into a large number of subsections. The number of subsections makes it difficult to analyze the article's content smoothly. Please, reduce their number.
4. There are issues with the References (e.g., lines 166 and 188).
5. Fig. 6: Please, change the colour or move the captions in the figures. In their current form, they are illegible.
6. Lines 1032-1033: The authors provide information that during the preparation of this manuscript, the authors used [Gemini AI Tools] for the purposes of [Graphic Abstract]. However, there is no Graphic Abstract before the Abstract. What is the reason for this notation?
7. Fig. 1 and Fig. 7: Figure captions are not located below the figures.
8. Line 709: Please, explain the origin of the value 1.5 years.
9. Table 4: The table caption should be on the same page as the table.
10. Line 1151: What does [M] mean?
11. Please, standardize the spacing between lines (e.g. page 23 and 24).

Author Response

Comments 1: There are no references to the authors' previous articles in the References. Do the authors have any previous work in this field? What topics have the authors covered so far?

Response 1: Thank you for pointing this out. We agree that the previous version did not sufficiently reflect the authors’ previous research basis and methodological continuity. In response to this comment, we added two previous publications by the author team as References [15] and [25] in the revised manuscript. Revised manuscript text:： “In terms of frontier expansion, recent interdisciplinary studies have further extended LCA to cross-domain impact evaluation. Dan Li[15] systematically reviewed artificial intelligence applications in cervical cancer diagnosis, treatment, and prognostic assessment, demonstrating that AI-driven multimodal data fusion and deep learning can improve the accuracy of medical detection and outcome prediction by processing massive multi-source data. This paradigm of multimodal data integration and dynamic model optimization provides a methodological reference for improving the traditional static LCA framework, supporting the construction of dynamic emission accounting models that integrate vehicle operating data, environmental factors, and material degradation information.” “In terms of model performance optimization and dynamic mechanism design, Jingya Qi and Jun Zhang[25] conducted an experimental investigation on two-phase cooling in microchannels with different cross-section geometries, proving that structural shape optimization can simultaneously improve heat transfer efficiency, temperature uniformity, and operational stability under multi-source heterogeneous conditions. This conclusion implies that heterogeneous scenario adaptation and structural parameter optimization can be introduced into discrete choice models to enhance the fitting accuracy of consumer vehicle purchase and renewal behavior under differentiated policy incentives.”   These additions clarify that the present study builds upon the author team’s previous methodological experience in multi-source data integration, dynamic modeling, heterogeneous scenario analysis, and optimization methods, while extending these ideas to the coupled assessment of vehicle fleet renewal policies, life-cycle emissions, urban environmental corrosion, and infrastructure coating durability.

Comments 2: Abstract and conclusion no. 2: The authors refer to the value of 1.5 years, but this value isn't present in this article. Please, clarify this issue.

Response 2: Thank you for identifying this inconsistency. After rechecking the data, we found that “1.5 years” was a textual inconsistency. According to the revised lifespan assessment, the projected remaining lifespan under S0 is 9.6 years, while the projected lifespans under the policy scenarios decrease relative to S0. Therefore, the lifespan reduction should be expressed as 1.3–2.3 years, rather than 1.5–2.3 years. We corrected this value in the Highlights, Abstract, Results, and Conclusions, and added a sentence explaining how the lifespan penalty was calculated.

Revised manuscript text:

“Induced acidic stress prematurely cuts protective coating lifespans by 1.3 to 2.3 years.”

 “This localized acidification significantly accelerates chemical degradation, reducing the effective service life of infrastructure coatings by 1.3–2.3 years and necessitating premature, costly recoating.”

“Projections indicate that the elevated environmental corrosivity reduces the effective remaining lifespan of the coating system by approximately 1.3 to 2.3 years compared to the S0 baseline. The lifespan penalty is calculated by subtracting the projected remaining lifespan under each policy scenario from the anticipated baseline lifespan (S0).”

“This severe acidic stress violently breached the critical thickness threshold, resulting in a premature lifespan penalty of 1.3 to 2.3 years compared with the baseline.”

 

Comments 3: The article is divided into a large number of subsections. The number of subsections makes it difficult to analyze the article's content smoothly. Please, reduce their number.

Response 3: Thank you for this helpful suggestion. We agree that the previous version contained too many subsections, which weakened the reading flow. We therefore reorganized the manuscript structure. In particular, the previous subsections 1.1.1 Real-World Context, 1.1.2 Theoretical Background, and 1.1.3 Research Significance were merged into a single Section 1.1 Research Background and Significance. This revision makes the Introduction more compact and improves the continuity from policy background to theoretical gap and research significance.

Revised manuscript text:

Against the backdrop of intensifying global climate change, the clean and low-carbon transition of the energy system has become an international consensus. As a core sector of fossil fuel consumption and carbon emissions, the low-carbon transition of the transportation industry is crucial to the energy revolution and climate govern-ance. Under China’s "Dual Carbon" strategic goals, deep decarbonization of transpor-tation has been integrated into the national green development framework. As the primary mobile pollution source in the transportation sector, vehicles have become the key to low-carbon transition through fleet optimization and technological upgrading. By the end of 2024, the number of NEVs in China had grown rapidly, yet the phase-out and replacement of high-emission old fuel vehicles remained slow. To accelerate the transition, the Chinese government has intensively issued vehicle trade-in subsidy policies and established a national implementation framework. However, two major challenges arise in practice: macro subsidies fail to precisely stimulate micro-level ve-hicle replacement behavior and may trigger unintended systemic consequences, mak-ing it difficult to achieve genuine net emission reductions. More importantly, acidic precursors such as SO2 and NOx from traffic emissions significantly increase urban atmospheric corrosivity, accelerate the aging of anti-corrosion coatings on bridges, guardrails and other infrastructure, shorten service life, and raise maintenance costs^[1]. Existing policy evaluations mostly focus on carbon emission reduction, ignoring the transmission chain of "vehicle replacement–emission change–atmospheric corrosion–coating durability", thus leaving blind spots in environmental and engineering design.

Theoretically, vehicle trade-in policies essentially represent the low-carbon re-structuring of the transportation socio-technical system driven by public policies^[2-4]. Current research has obvious limitations: traditional emission reduction assessments mostly adopt the tank-to-wheel one-dimensional model^[5], which only considers direct use-phase emissions and neglects life-cycle impacts and policy-behavior interactions; the mainstream LCA method^[6] relies on static parameters, failing to reflect the heter-ogeneity of consumer behavior and thus unable to link micro behavior with macro carbon accounting; although discrete choice models can characterize vehicle purchase decisions, they exclude environmental factors such as carbon emissions and corrosion effects from utility functions and are dominated by static analysis incapable of cap-turing long-term impacts. Furthermore, an integrated coupling framework of "policy–behavior–environment–material" has not yet been established, and relevant dynamic modeling studies remain scarce, leaving insufficient cross-system quantitative research. Therefore, an integrated model is urgently needed to fill this theoretical gap.

The theoretical significance of this study lies in establishing a coupling model among vehicle replacement policies, atmospheric corrosion, and infrastructure coating durability, breaking through the limitations of traditional methods, realizing endog-enous coupling of multiple factors, and enriching the theory of socio-technical system transition. The practical significance is to quantitatively identify the lag effects and emission inversion risks of policies, establish a quantitative relationship among emis-sions, corrosivity, and coating lifespan depletion, propose an optimal subsidy interval that balances carbon emission reduction and infrastructure sustainability, promote the shift from universal subsidies to differentiated and targeted incentives, and improve the comprehensive benefits of the low-carbon transition in transportation.

 

Comments 4: There are issues with the References (e.g., lines 166 and 188).

Response 4: Thank you for your careful reading. We checked and corrected the citation formatting throughout the manuscript. In particular, we standardized the in-text citation style and corrected concatenated or incorrectly formatted citations, such as changing “policies²³⁴” to “policies^[2-4]” and correcting similar formatting issues in the Literature Review.

Revised manuscript text:

“Theoretically, vehicle trade-in policies essentially represent the low-carbon restructuring of the transportation socio-technical system driven by public policies^[2-4].”

 

Comments 5: Fig. 6: Please, change the colour or move the captions in the figures. In their current form, they are illegible.

Response 5: Thank you for this suggestion. We revised Figure 6 to improve its readability and consistency. Specifically, we adjusted the color contrast, enlarged the legend and axis labels, and ensured that the figure caption is placed below the figure according to the journal’s formatting requirements. We also checked whether the figure content, caption, and in-text description were consistent.

 

Comments 6: Lines 1032-1033: The authors provide information that during the preparation of this manuscript, the authors used [Gemini AI Tools] for the purposes of [Graphic Abstract]. However, there is no Graphic Abstract before the Abstract. What is the reason for this notation?

Response 6: Thank you for pointing this out. We apologize for the confusion caused by this statement. The graphical abstract was prepared as a separate graphical abstract file. To avoid ambiguity, we added the graphical abstract to the revised submission materials and clarified the AI-use statement in the manuscript.

Revised manuscript text:

 

Figure 8. Graphical Abstract

 

Comments 7: Fig. 1 and Fig. 7: Figure captions are not located below the figures.

Response 7: Thank you for noting this formatting issue. We adjusted the placement of figure captions according to the journal style. The captions of Figure 1 and Figure 7 are now placed below the corresponding figures.

 

Comments 8: Line 709: Please, explain the origin of the value 1.5 years.

Response 8: Thank you for this comment. As explained in Response 2, the value “1.5 years” was a textual inconsistency. The corrected value is 1.3–2.3 years. We added an explanation of the calculation method: the lifespan penalty is calculated by subtracting the projected remaining lifespan under each policy scenario from the anticipated baseline lifespan under S0.

Revised manuscript text:

“Projections indicate that the elevated environmental corrosivity reduces the effective remaining lifespan of the coating system by approximately 1.3 to 2.3 years compared to the S0 baseline. The lifespan penalty is calculated by subtracting the projected remaining lifespan under each policy scenario from the anticipated baseline lifespan (S0).”

 

Comments 9: Table 4: The table caption should be on the same page as the table.

Response 9: Thank you for identifying this formatting problem. We adjusted the pagination and table layout to ensure that the table and its caption appear on the same page. We also checked the formatting of other tables to avoid similar layout problems.

 

Comments 10: Line 1151: What does [M] mean?

Response 10: Thank you for pointing this out. The marker “[M]” was derived from a reference classification style and was not appropriate for the journal’s reference format. We removed “[M]” and standardized the relevant reference entries according to the journal style.

 

Comments 11: Please, standardize the spacing between lines (e.g. page 23 and 24).

Response 11: Thank you for your careful observation. We reviewed the formatting of the entire manuscript and standardized the line spacing, figure caption placement, table pagination, and page breaks throughout the document.

 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript presents a coupled modeling framework linking vehicle trade-in subsidy policies to infrastructure coating degradation through dynamic Life Cycle Assessment (LCA), a Mixed Logit consumer choice model, and an Environmental Corrosion Risk Index (ECRI) integrated with Dose–Response Functions (DRF). The research topic is timely, the interdisciplinary scope is ambitious, and the identification of the “emission inversion” phenomenon under aggressive subsidy scenarios represents a novel and policy-relevant contribution. Overall, the manuscript demonstrates significant potential for contribution to the field of sustainable transportation policy and infrastructure durability assessment. However, several important issues related to methodological transparency, empirical justification of parameters, figure consistency, validation strategy, and manuscript presentation must be addressed before the work can be considered suitable for publication. Therefore, the recommendation is Major Revision.

 

1. The manuscript states that the coupled framework was implemented in MATLAB; however, no supplementary code, computational workflow, or reproducibility information is provided. In its current form, the numerical implementation cannot be independently verified. In addition, the procedures used to estimate the Mixed Logit model and integrate the Life-cycle Degradation Response (LDR) remain insufficiently described.

• Provide the MATLAB implementation, pseudocode, or a detailed computational workflow for the four-module coupling framework as supplementary material. If code sharing is not feasible, the authors should explicitly justify this in the Data Availability Statement.
• Clarify how the Mixed Logit integral in Eq. (2) was numerically evaluated. Specify whether simulated maximum likelihood estimation was employed and report the number and type of draws used (e.g., Halton, Sobol, pseudo-random).
• The stated-preference (SP) survey design is insufficiently documented. Please provide: example choice sets, attribute definitions and levels, questionnaire structure, respondent screening criteria, and details of how the “stratified random sampling” strategy was operationalized in the online survey environment.
• The numerical procedure used to integrate Eq. (6) over time should also be explicitly described, including time-step size, convergence criteria, and iteration logic between modules.

2. Several critical parameters are introduced without sufficient empirical justification or traceable sourcing.

• In Table 1, the “Standard Degradation Rate” (12.5 μm/year) and the “Acidic Sensitivity Coefficient” (α = 1.25) are central to all projected coating lifespan calculations. Although References [51,52] are cited, the manuscript does not explain whether these values were directly adopted, calibrated, or fitted to local environmental observations.
• The characterization factors used in Section 2.5.2 (CF_AP = 1.0 for SO₂ and 0.7 for NOₓ) should be explicitly linked to a recognized LCA methodology framework such as CML 2002, ReCiPe, or TRACI rather than being presented as model initialization constants.
• The projected coal-fired electricity share trajectory (65% in 2026 decreasing to 52% in 2030) requires citation to an official Chinese national energy policy or planning document.

3. There are multiple inconsistencies between figure captions, in-text descriptions, and the actual visual content presented. These discrepancies significantly affect readability and interpretation.

Examples include:

• Figure 4 is captioned as:
  “Spatiotemporal evolution of environmental stress drivers before and after the implementation of the optimal subsidy policy across administrative districts.”

However, the displayed figure appears to be a system-dynamics causal-loop diagram describing the carbon payback mechanism rather than a spatiotemporal visualization.

• Figure 6 is captioned as:
  “Projected trajectory of cumulative infrastructure maintenance costs and recoating frequencies under scenarios S0–S3.”

Yet the figure shown consists of spatial heatmaps of CO₂ and acidification potential distributions.

• The maintenance-cost trajectory figure referenced in the caption appears to be missing entirely.

The authors should comprehensively audit all figures, captions, numbering, and in-text references to ensure complete consistency throughout the manuscript.

4. Section 2.2.1 states that Beijing is used as the representative study region; however, much of the policy discussion and several conclusions are generalized to the national level.

The recommended subsidy range (8,000–10,500 CNY) is presented as broadly applicable, despite the model being calibrated using Beijing-specific assumptions regarding: grid electricity mix, atmospheric pollutant baseline, fleet structure, and urban environmental conditions.

The manuscript should therefore either:

• explicitly limit conclusions to the Beijing case study throughout the paper, or
• provide sensitivity analyses demonstrating whether the identified policy window remains robust under different regional conditions (e.g., low-carbon provinces such as Yunnan versus coal-intensive regions such as Inner Mongolia).

5. The manuscript reports quantitative outputs such as ECRI values, lifespan penalties, and maintenance-cost projections; however, no validation procedure is presented.

At minimum, the authors should:

• compare the baseline scenario (S0) ECRI predictions against historical atmospheric corrosivity or environmental monitoring data from Beijing,
• discuss uncertainties associated with the DRF assumptions,
• and explain the limitations of applying a single acidic sensitivity coefficient (α) across heterogeneous urban microenvironments, particularly since Section 3.2.3 explicitly acknowledges spatial heterogeneity.

6. The manuscript requires substantial language revision.

Examples include:

• Abstract, lines 41–42:
  “Our results indicate that aggressive subsidies induce a transition to heavy NEVs …”

The sentence appears truncated in the manuscript.

• Section 1.1.2, line 98:
  “macro-policies234”

Reference formatting is incorrect and appears concatenated.

• Section 3.3.2, line 682:
  “To visualize the progressive progression”

This expression is redundant. Consider replacing with “progressive degradation” or “degradation progression.”

• Several paragraphs inconsistently switch between past and present tense.

 

7. Table 2 – Duplicate Log-Likelihood Entry

Table 2 reports “Log-likelihood” twice with two different values (−655.82 and −382.45) without clarification.

Please distinguish between:

• null log-likelihood (intercept-only model), and
• final model log-likelihood.

In addition, specify the type of pseudo-R² reported (e.g., McFadden’s ρ², Nagelkerke R²).

 

8. Table 3 – Sign Convention and Units

The column:
“Lifespan Penalty vs. S0 (Years)”
uses negative values (e.g., −0.7, −1.3, −2.3) to indicate lifespan reduction.

However, the Highlights section reports these penalties as positive quantities (“1.5–2.3 years”).

The sign convention should be standardized and explicitly clarified in a table footnote.

9. Table 4 – Interpretation of ΔAP Trends

The ΔAP values reported in Table 4 require clearer interpretation.

Specifically:

• the 10,500 CNY scenario shows ΔAP = −8.00%,
• whereas the 12,000 CNY scenario shows +25.0%,

which aligns with the proposed “emission inversion” narrative.

However, the 10,000 CNY scenario reports ΔAP = −18.00%, implying a substantial acidification reduction under the optimal policy range.

The manuscript should more carefully explain:

• the physical interpretation of these transitions, the threshold behavior triggering inversion.

 

10. Objective Function Parameters (Eq. 7)

The Pareto optimization framework combines Global Warming Potential (GWP) and maintenance cost using weighting coefficients ω₁ and ω₂; however, these values are never disclosed.

The manuscript must report:

• the numerical values of ω₁ and ω₂, the rationale for selecting them, and whether sensitivity analyses were performed with alternative weighting combinations.

Without this information, the Pareto front and the recommended subsidy interval cannot be independently reproduced.

 

 

Author Response

Comments 1: The manuscript states that the coupled framework was implemented in MATLAB; however, no supplementary code, computational workflow, or reproducibility information is provided. In its current form, the numerical implementation cannot be independently verified. In addition, the procedures used to estimate the Mixed Logit model and integrate the Life-cycle Degradation Response (LDR) remain insufficiently described. Provide the MATLAB implementation, pseudocode, or a detailed computational workflow for the four-module coupling framework as supplementary material. If code sharing is not feasible, the authors should explicitly justify this in the Data Availability Statement. Clarify how the Mixed Logit integral in Eq. (2) was numerically evaluated. Specify whether simulated maximum likelihood estimation was employed and report the number and type of draws used. The stated-preference (SP) survey design is insufficiently documented. Please provide: example choice sets, attribute definitions and levels, questionnaire structure, respondent screening criteria, and details of how the “stratified random sampling” strategy was operationalized in the online survey environment. The numerical procedure used to integrate Eq. (6) over time should also be explicitly described, including time-step size, convergence criteria, and iteration logic between modules

Response 1: Thank you for this comprehensive and important comment. We agree that the methodological transparency and reproducibility of the coupled framework needed to be strengthened. We revised the manuscript and supplementary materials accordingly. First, we added a computational workflow and pseudocode for the four-module coupling framework in the Supplementary Material and updated the Data Availability Statement. Second, we clarified the numerical evaluation of the Mixed Logit integral in Eq. (2). The model was solved using Simulated Maximum Likelihood Estimation (SMLE) with 500 Halton draws. Revised manuscript text: “Since the multidimensional integral in Equation (2) lacks a closed-form solution, this study employs the Simulated Maximum Likelihood Estimation (SMLE) method to evaluate the choice probabilities. Specifically, 500 Halton draws were utilized to approximate the integral, which ensures superior computational efficiency and estimation stability compared to standard pseudo-random sampling. By aggregating these micro-probabilities across the population over time, the dynamic trajectory of the regional vehicle fleet structure is explicitly generated.” Third, we expanded the SP survey description, including questionnaire structure, respondent screening criteria, attribute definitions, choice-set reconstruction, and stratified random sampling. Revised manuscript text: “The dynamic evolution of a regional vehicle fleet fundamentally determines the atmospheric emission burden of corrosion-related precursors, while this macro-level evolutionary process is deeply shaped by consumers’ vehicle replacement decisions at the micro level. To accurately capture and quantify heterogeneous consumer preferences under different policy interventions, this study designed and implemented a micro-level behavioral preference questionnaire targeting urban private vehicle owners. The questionnaire was systematically organized into five core sections. The first section collected respondents’ demographic and socioeconomic information. The second section recorded existing vehicle characteristics, including vehicle age, emission standard, estimated residual value, and maintenance costs, thereby establishing a baseline for replacement decisions. The third section focused on travel and vehicle-use patterns, with particular attention to annual mileage, major travel purposes, and the availability of fixed parking spaces and charging conditions. The fourth section examined replacement intention and policy response, serving as the experimental module for extracting the key dependent variable. The fifth section measured cognitive and subjective evaluation variables, including respondents’ perceived daily operating cost of new energy vehicles, range anxiety, and trust in government policy.” “To effectively initialize the utility function of the mixed Logit model, the questionnaire extracted respondents’ core preference attributes through a multidimensional decomposition of decision scenarios. The specific attributes and corresponding levels are presented in Table 1. For subsidy intensity, which constitutes the central policy lever, several policy gradients were designed, including no subsidy, CNY 3,000, CNY 5,000, and CNY 10,000.” “In the actual survey implementation, to reduce respondents’ cognitive burden and avoid extreme choice bias, the questionnaire did not simply present an overly complex set of fully enumerated alternatives. Instead, based on the multidimensional logical structure described above, the respondents’ discrete choice set was logically reconstructed during the data processing stage for use in the Mixed Logit model.” “To ensure that the micro-level behavioural sample could support a cross-scale and high-resolution projection of macro-level vehicle fleet evolution, this study adopted a rigorous operationalized stratified random sampling strategy across China.” “The questionnaire was distributed online. To ensure data quality, concealed logical consistency checks were embedded within the survey. During data cleaning, responses that failed the logical checks, exhibited abnormal completion times, such as durations shorter than a reasonable reading-time threshold, or contained extensive missing values in core variables were strictly excluded. Ultimately, 597 high-quality valid responses were obtained and retained.” Fourth, we clarified the time integration of Eq. (6). The model uses a fixed annual time step, Δt = 1 year, and follows a sequential, unidirectional coupling logic among modules.

Comments 2: Several critical parameters are introduced without sufficient empirical justification or traceable sourcing. In Table 1, the “Standard Degradation Rate” (12.5 μm/year) and the “Acidic Sensitivity Coefficient” (α = 1.25) are central to all projected coating lifespan calculations. Although References [51,52] are cited, the manuscript does not explain whether these values were directly adopted, calibrated, or fitted to local environmental observations. The characterization factors used in Section 2.5.2 (CF_AP = 1.0 for SO2 and 0.7 for NOx) should be explicitly linked to a recognized LCA methodology framework such as CML 2002, ReCiPe, or TRACI rather than being presented as model initialization constants. The projected coal-fired electricity share trajectory (65% in 2026 decreasing to 52% in 2030) requires citation to an official Chinese national energy policy or planning document.

Response 2: Thank you for this valuable suggestion. We agree that the original manuscript did not sufficiently explain the empirical basis and traceability of several key parameters. We revised the parameter description accordingly. First, we clarified that the coating degradation parameters were initialized using established DRF values in corrosion engineering and calibrated with localized environmental monitoring data. Revised manuscript text:

“To translate fleet transitions into coating degradation risks, the physical environment and material-related parameters were initialized by integrating established databases, regional environmental monitoring records, and engineering specification standards. The baseline tailpipe emission factors for SO₂ and NO_x emitted by ICEVs were calibrated using the latest national vehicle emission inventory. For NEVs, upstream grid emission factors were initialized based on the regional power grid’s energy mix, particularly the proportion of coal-fired power. Life-cycle AP parameters were derived from the GREET model and were further localized to reflect the industrial characteristics of the target region. The initial atmospheric corrosivity state was determined using historical environmental monitoring data from the regional ecological environment bureau, including annual average concentrations of SO₂ and NO_x.”

Second, we linked the acidification characterization factors to the ReCiPe 2016 midpoint impact assessment methodology.

Revised manuscript text:

“In this study, these parameters are explicitly mapped onto the internationally standardized ReCiPe 2016 midpoint impact assessment methodology (hierarchical perspective), thereby guaranteeing methodological compatibility with mainstream lifecycle frameworks.”

Third, we added an official policy/planning basis for the coal-fired electricity share trajectory and clarified the assumed decline from 65% in 2026 to 52% in 2030.

Revised manuscript text:

“The simulation incorporates the evolving energy mix of the regional power grid to calculate the indirect AP. Based on the regional energy transition roadmap, the proportion of coal-fired power is assumed to decrease from 65% in 2026 to 52% in 2030[54]. However, the model accounts for the marginal emission intensity of peak-load power generation triggered by the simultaneous charging of newly adopted NEVs. The characterization factors for environmental acidification are strictly derived from the ReCiPe 2016 midpoint approach, initialized as 1.0 for SO₂ and 0.7 for NO_x.”

Fourth, we clarified the empirical basis of the degradation rate and acidic sensitivity coefficient.

Revised manuscript text:

“The material response module is initialized with parameters representing a standard C4-grade (High Corrosivity) urban environment protection system. The specific parameters for the simulated coating system (Epoxy Zinc-Rich Primer + Polyurethane Topcoat) are detailed in Table 1. It is important to emphasize that these baseline parameters—specifically the standard degradation rate and the acidic sensitivity coefficient—are not purely theoretical constants or arbitrary assumptions. Instead, they are rigorously calibrated by mapping the standard atmospheric corrosivity classification principles of the ISO 9223 framework onto historical, multi-year localized environmental and corrosion monitoring data.”

 

Comments 3: There are multiple inconsistencies between figure captions, in-text descriptions, and the actual visual content presented. These discrepancies significantly affect readability and interpretation.

Examples include:

Figure 4 is captioned as: “Spatiotemporal evolution of environmental stress drivers before and after the implementation of the optimal subsidy policy across administrative districts.”

However, the displayed figure appears to be a system-dynamics causal-loop diagram describing the carbon payback mechanism rather than a spatiotemporal visualization.

Figure 6 is captioned as: “Projected trajectory of cumulative infrastructure maintenance costs and recoating frequencies under scenarios S0–S3.”

Yet the figure shown consists of spatial heatmaps of CO2 and acidification potential distributions.

The maintenance-cost trajectory figure referenced in the caption appears to be missing entirely.

The authors should comprehensively audit all figures, captions, numbering, and in-text references to ensure complete consistency throughout the manuscript.

Response 3: Thank you for pointing out these important inconsistencies. We sincerely apologize for the mismatch between the figures, captions, and in-text descriptions in the previous version. We conducted a comprehensive audit of all figures, captions, numbering, and cross-references.

Figure 4 was revised so that the visual content and caption consistently describe the spatial distribution and environmental stress drivers. Figure 6 was also corrected to present the maintenance-cost trajectory corresponding to the caption. We further checked all in-text figure references to ensure consistency.

Revised manuscript text:

  “Figure 4. Spatiotemporal evolution of environmental stress drivers before and after the implementation of the optimal subsidy policy across administrative districts.

The intensified emission clusters conceptually mirror the localized accumulation of acidic precursors, highlighting high-risk zones for infrastructure coating degradation within the urban network.”

“Figure 6. Projected trajectory of cumulative infrastructure maintenance costs and recoating frequencies under scenarios S0-S3

Figure 6 illustrates the long-term economic consequences of policy-induced material degradation through a cumulative maintenance cost trajectory. Under the baseline and moderate scenarios (S0-S2), the infrastructure maintains a stable protective cycle with predictable fiscal requirements. However, the high-stimulus scenario (S3) triggers a significant temporal compression of the maintenance intervals.”

 

Comments 4: Section 2.2.1 states that Beijing is used as the representative study region; however, much of the policy discussion and several conclusions are generalized to the national level.

The recommended subsidy range (8,000–10,500 CNY) is presented as broadly applicable, despite the model being calibrated using Beijing-specific assumptions regarding grid electricity mix, atmospheric pollutant baseline, fleet structure, and environmental conditions.

The manuscript should therefore either explicitly limit conclusions to the Beijing case study throughout the paper, or provide sensitivity analyses demonstrating whether the identified policy window remains robust under different regional conditions.

Response 4: Thank you for this important comment. We agree that the relationship between the representative case and broader policy implications needed to be clarified. In the revised manuscript, we clarified that Beijing is used as a representative benchmark case for high-density metropolitan transportation environments, while the spatial interpretation of the results should not be directly extrapolated to all regions in China. We also strengthened the discussion of regional heterogeneity, especially differences in grid structure, usage intensity, and environmental exposure.

Revised manuscript text:

“The logical boundary of this research is defined as a ‘Policy–Environment–Material’ closed loop. Spatially, the study focuses on the national urban transportation network, with Beijing used as a representative benchmark case for high-density metropolitan transportation environments where infrastructure is directly exposed to heavy traffic-related emissions. Therefore, the spatial interpretation of the results should be understood as national-scale implications for similar high-density urban areas rather than a direct extrapolation to all regions in China.”

“In regions dominated by coal-fired power, vehicle replacement policies centered on ‘replacing oil with electricity’ show significant systemic vulnerability. In areas with a high proportion of clean energy, the low-carbon level of the power grid can effectively support emission reductions across the entire lifecycle of new energy vehicles, resulting in strong system resilience; conversely, in regions dependent on coal-fired power, the high-carbon structure of the grid significantly undermines the emission reduction advantages of new energy vehicles during the usage phase.”

 

Comments 5: The manuscript reports quantitative outputs such as ECRI values, lifespan penalties, and maintenance-cost projections; however, no validation procedure is presented.

At minimum, the authors should:

compare the baseline scenario (S0) ECRI predictions against historical atmospheric corrosivity or environmental monitoring data from Beijing,

discuss uncertainties associated with the DRF assumptions,

and explain the limitations of applying a single acidic sensitivity coefficient (α) across heterogeneous urban microenvironments, particularly since Section 3.2.3 explicitly acknowledges spatial heterogeneity.

Response 5: Thank you for this constructive suggestion. We agree that the credibility of the coupled model depends on a clearer validation and uncertainty discussion. We therefore added a new subsection entitled Section 3.4 Model Validation and Methodological Uncertainties.

In this section, we compare the baseline scenario with historical environmental and material exposure baselines, discuss uncertainty in the DRF assumptions, and clarify the limitation of applying a single acidic sensitivity coefficient across heterogeneous urban microenvironments.

Revised manuscript text:

“To establish the credibility of the multi-module coupled framework, a hindcasting validation of the Baseline scenario (S0) was performed against historical national environmental baselines and materials exposure data. Although projecting long-term macro-policy effects over a multi-year horizon (2026–2030) inherently involves systemic uncertainties, the simulated near-term trajectories of Acidification Potential and coating degradation kinetics show robust alignment with historical empirical benchmarks.”

“In real-world engineering infrastructure, atmospheric corrosion kinetics are highly non-linear and sensitive to localized micro-climates, wind fields, local boundary layer aerodynamics, and sheltering effects. Consequently, using a single α may under- or over-estimate the localized lifespan depletion rates under intense emission hotspots. To overcome this limitation, future expansions of this framework could couple the current macro-system model with micro-scale Computational Fluid Dynamics (CFD) simulations to map the precise fluid-structure chemical interactions and micro-environmental concentration gradients on specific critical infrastructure surfaces.”

 

Comments 6: The manuscript requires substantial language revision.

Examples include:

Abstract, lines 41–42: “Our results indicate that aggressive subsidies induce a transition to heavy NEVs ...” The sentence appears truncated in the manuscript.

Section 1.1.2, line 98: “macro-policies234” Reference formatting is incorrect and appears concatenated.

Section 3.3.2, line 682: “To visualize the progressive progression” This expression is redundant. Consider replacing with “progressive degradation” or “degradation progression.”

Several paragraphs inconsistently switch between past and present tense.

Response 6: Thank you for your careful reading and helpful language suggestions. We revised the language throughout the manuscript. Specifically, we corrected the truncated sentence in the Abstract, standardized citation formatting, removed redundant wording, and adjusted tense usage.

Revised manuscript text:

“Our results indicate that aggressive subsidies induce a transition to heavy NEVs, triggering an ‘emission inversion’ that spikes upstream grid acidic emissions. This localized acidification significantly accelerates chemical degradation, reducing the effective service life of infrastructure coatings by 1.3–2.3 years and necessitating premature, costly recoating.”

“Theoretically, vehicle trade-in policies essentially represent the low-carbon restructuring of the transportation socio-technical system driven by public policies^[2-4].”

“To visualize the progressive degradation of material failure, the dynamic reduction in coating thickness was simulated over the critical transition window from 2026 to 2030.”

 

Comments 7: Table 2 – Duplicate Log-Likelihood Entry

Table 2 reports “Log-likelihood” twice with two different values (–655.82 and –382.45) without clarification.

Please distinguish between:

null log-likelihood (intercept-only model), and

final model log-likelihood.

In addition, specify the type of pseudo-R² reported (e.g., McFadden’s ρ², Nagelkerke R²).

Response 7: Thank you for pointing out this table-formatting and terminology issue. We revised the table (now Table 3 in the revised manuscript) to distinguish between the Null Log-likelihood of the intercept-only model and the Final Log-likelihood of the estimated model. We also clarified that the reported pseudo-R² is McFadden’s pseudo-R².

Variables	Coefficient (β)	Std. Error	z-value	p-value
Alternative Specific Constant: ICEV	1.142	0.215	5.31	<0.001
Alternative Specific Constant: NEV	0.875	0.243	3.6	<0.001
Subsidy Intensity	1.583	0.142	11.15	<0.001
Operating Cost	-0.046	0.007	-6.57	<0.001
Charging Convenience	0.812	0.118	6.88	<0.001
Battery Capacity & Weight Index	0.435	0.082	5.3	<0.001
Random Parameters
SD of Subsidy Intensity	0.924	0.176	5.25	<0.001
SD of Operating Cost	0.018	0.005	3.6	0.001
Model Fit Statistics
Number of Respondents	597
NullLog-likelihood	-655.82
FinalLog-likelihood	-382.45
Pseudo R2（McFadden's）	0.416

 

Comments 8: Table 3 – Sign Convention and Units

The column: “Lifespan Penalty vs. S0 (Years)” uses negative values (e.g., –0.7, –1.3, –2.3) to indicate lifespan reduction.

However, the Highlights section reports these penalties as positive quantities (“1.5–2.3 years”).

The sign convention should be standardized and explicitly clarified in a table footnote.

Response 8: Thank you for this important comment. We agree that the previous sign convention could lead to confusion. We standardized the lifespan penalty as a positive absolute value representing the magnitude of lifespan reduction relative to S0. Negative signs were removed where necessary, and the Highlights, Abstract, Results, and Conclusions were revised consistently. We also added an explanation of how the lifespan penalty was calculated.

Revised manuscript text:

“Induced acidic stress prematurely cuts protective coating lifespans by 1.3 to 2.3 years.”

“Projections indicate that the elevated environmental corrosivity reduces the effective remaining lifespan of the coating system by approximately 1.3 to 2.3 years compared to the S0 baseline. The lifespan penalty is calculated by subtracting the projected remaining lifespan under each policy scenario from the anticipated baseline lifespan (S0).”

“Note: Coating lifespan penalty is reported as the absolute reduction in coating service life relative to the S0 baseline. Positive values indicate the magnitude of lifespan loss rather than a numerical increase in service life.”

 

Comments 9: Table 4 – Interpretation of ΔAP Trends

The ΔAP values reported in Table 4 require clearer interpretation.

Specifically:

the 10,500 CNY scenario shows ΔAP = –8.00%,

whereas the 12,000 CNY scenario shows +25.0%,

which aligns with the proposed “emission inversion” narrative.

However, the 10,000 CNY scenario reports ΔAP = –18.00%, implying a substantial acidification reduction under the optimal policy range.

The manuscript should more carefully explain:

the physical interpretation of these transitions,

the threshold behavior triggering inversion.

Response 9:

Thank you for this valuable suggestion. We expanded the physical explanation of the ΔAP transition and clarified the threshold behavior behind the emission inversion.

The revised manuscript explains that moderate subsidy levels reduce older high-emission ICEVs and therefore lower acidification potential. However, once subsidy intensity crosses the critical threshold, consumer preference shifts disproportionately toward Heavy/Oversized NEVs with large battery capacities. This increases electricity demand and forces reliance on marginal coal-fired peak-shaving units, thereby increasing upstream SO₂ and NO_x emissions and triggering emission inversion.

Revised manuscript text:

“This abrupt transition from a robust mitigation regime at 10,500 CNY to a severe emission spike at 12,000 CNY mathematically highlights a distinct non-linear threshold effect within the coupled transportation-energy system. Mechanistically, this localized tipping point is governed by the structural coupling of micro-consumer utilities and macro-grid dynamics. When financial incentives cross the critical 10,500 CNY boundary and scale up to 12,000 CNY, the monetary dividend over-compensates for the purchasing premium of higher-end vehicle segments. This triggers an irrational consumer market boom and a disproportionate behavioral shift toward Heavy/Oversized NEVs equipped with excessively large battery capacities and high operational energy footprints.”

“On the energy supply side, the simultaneous charging demands from this concentrated heavy vehicle fleet inflict a sharp, synchronized surge on the regional power grid. To maintain system stability during these sudden peak-load regimes, the utility network is forced to rapidly deploy marginal, low-efficiency coal-fired peak-shaving units rather than relying on baseload clean energy. Because these marginal thermal units emit sulfur dioxide and nitrogen oxides at rates exponentially higher than stabilized baseload generation, the upstream environmental burden violently flips, inducing a heavy regional atmospheric acidification penalty that offsets the local tailpipe carbon reduction dividends.”

 

Comments 10: Objective Function Parameters (Eq. 7)

The Pareto optimization framework combines Global Warming Potential (GWP) and maintenance cost using weighting coefficients ω1 and ω2; however, these values are never disclosed.

The manuscript must report:

the numerical values of ω1 and ω2,

the rationale for selecting them,

and whether sensitivity analyses were performed with alternative weighting combinations.

Without this information, the Pareto front and the recommended subsidy interval cannot be independently reproduced.

Response 10: Thank you for this important comment. We agree that the objective function parameters should be explicitly reported to improve reproducibility. We revised the description of Eq. (7) and clarified the values and rationale for the weighting coefficients.

In the baseline simulation, the weights were set as ω1 = 0.5 and ω2 = 0.5, representing an equal-weight policy paradigm that assigns equivalent importance to environmental decarbonization and infrastructure maintenance sustainability. We also clarified that these weights are not fixed empirical constants, but adjustable policy parameters for sensitivity expansion under different regional governance priorities.

Revised manuscript text:

“It is imperative to note that these weights are not static empirical fixtures but rather flexible policy levers designed to capture shifting regional governance priorities. For instance, a sensitivity expansion of this framework reveals that in highly industrialized or ecologically vulnerable zones, ω1 can be upscaled to emphasize immediate emission containment. Conversely, in regions bound by tight municipal fiscal constraints, ω2 can be augmented to prioritize the minimization of premature infrastructure degradation expenditures. Disclosing this parameter flexibility ensures that the optimization model remains highly adaptable to differentiated real-world decision scenarios.”

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript has been revised satisfactorily, and recommend it for acceptance.