Digital Twin-Driven Low-Carbon Service Design and Modularization in Central Air Conditioning Ecosystems: A Multi-Criteria and Co-Intelligence Approach

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Digital Twin-Driven Low-Carbon Service Design and Modularization in Central Air Conditioning Ecosystems: A Multi-Criteria and Co-Intelligence Approach

The manuscript makes a timely contribution by proposing a Digital Twin-driven framework for low-carbon service design and modularization in Central Air Conditioning ecosystems, integrating demand analysis, co-intelligence modeling, and modular optimization. The case study in Zhuhai adds practical credibility. The work is well aligned with Sustainability but requires refinements in theoretical positioning, methodological justification, and integration of broader literature to strengthen its academic contribution.

1. The manuscript proposes co-intelligence modeling and DT-enabled modularization, yet its novelty needs clearer articulation. Beyond the PSS/SPSS literature, the authors should engage with adjacent domains such as optimization, pricing, and carbon reduction under uncertainty. This would demonstrate that the proposed framework not only advances service modularization but also builds on broader sustainability scholarship.
2. To enhance theoretical depth, I strongly recommend citing and discussing the following studies: i. Impact of extended warranty service on product pricing in online direct retailers' competitive market, ii. Evolution delayed decision game based on Carbon emission and capacity sharing in the Chinese market, iii. Optimal pricing and complex analysis for low-carbon apparel supply chains, iv. Pricing strategy and coordination of automobile manufacturers based on government intervention and carbon emission reduction, v. Incorporate robust optimization and demand defense for optimal planning of shared rental energy storage in multi-user industrial park, vi. Collaborative and effective scheduling of integrated energy systems with consideration of carbon restrictions, vii. Exploring the Impact of a District Sharing Strategy on Application Capacity and Carbon Emissions for Heating and Cooling with GSHP Systems, viii. A RL-based human behavior oriented optimal ventilation strategy for better energy efficiency and indoor air quality.
3. The use of rough-fuzzy DEMATEL, rough-fuzzy BWM, an improved Girvan–Newman algorithm, and IT2F-TOPSIS is technically rigorous, but the manuscript must justify why these methods are most appropriate for handling hybrid uncertainties. To strengthen this point, the authors could relate to: i. Circular economy solutions for net-zero carbon in China's construction sector: A strategic evaluation, ii. Uptake and adoption of sustainable energy technologies: Prioritizing strategies to overcome barriers in the construction industry by using an integrated AHP‐TOPSIS approach, iii. From waste to treasure: transforming construction waste into concrete masonry blocks: challenges and solutions for environmental sustainability in emerging economies.
4. The term “co-intelligence” is introduced as a novel element but lacks consistent theoretical grounding. It should be more explicitly linked to system synergy, knowledge co-creation, or collaborative intelligence frameworks to avoid being viewed as a relabeling of interdependency analysis.
5. While the DT enabling layer is well described, its role in enhancing multi-criteria decision-making is not fully explained. The manuscript should clarify how real-time DT data supports robustness in demand analysis, prioritization, and module evaluation.
6. The Zhuhai office building case study is informative but largely descriptive. Quantitative evidence (e.g., reductions in energy use, cost savings, carbon intensity improvements) would provide stronger validation.
7. The improved Girvan–Newman algorithm is proposed as an innovation, yet no comparative results are presented against the standard GN or other clustering algorithms. Demonstrating superior performance through modularity indices or efficiency metrics is essential.
8. Validation is limited to one office building. The authors should discuss adaptability to other building types (residential, hospitals, industrial parks) and policy contexts, which would reinforce the framework’s broader applicability.
9. The paper references China’s carbon neutrality goals but does not translate findings into concrete policy implications. Discussing potential integration into green building codes, CAC industry standards, or carbon trading mechanisms would strengthen its societal relevance.
10. Although multi-stakeholder needs are mentioned, the case study emphasizes technical aspects. Incorporating insights from facility managers, service providers, and end-users would demonstrate that the framework addresses both technical and social dimensions of value creation.
11. The uncertainty analysis focuses mainly on expert evaluations. Real-world operational uncertainties such as fluctuating energy prices, sensor failures, and weather variations should be acknowledged and linked to DT’s predictive capabilities.
12. The economic dimension is treated superficially. A cost–benefit or life-cycle cost analysis would help substantiate the claim that modularization yields not only carbon reductions but also tangible financial benefits.
13. The manuscript emphasizes system modularization but pays little attention to occupant-centered aspects such as comfort, satisfaction, and behavior. Integrating human-centered design principles would broaden the framework’s relevance.
14. The conclusion outlines general extensions but could be more specific. Testing the framework in a real-time DT platform, applying machine learning for predictive modularization, or extending to multi-building district energy systems would provide a sharper research roadmap.

 

Author Response

Comments 1: The manuscript proposes co-intelligence modeling and DT-enabled modularization, yet its novelty needs clearer articulation. Beyond the PSS/SPSS literature, the authors should engage with adjacent domains such as optimization, pricing, and carbon reduction under uncertainty. This would demonstrate that the proposed framework not only advances service modularization but also builds on broader sustainability scholarship.

Response 1:

We sincerely thank you for this insightful and constructive guidance. We fully agree that positioning our framework within the broader context of sustainability scholarship is crucial for clearly articulating its novelty and contributions.

Following your advice, we have thoroughly reflected on the manuscript's positioning and have systematically revised and enriched the Introduction and Literature Review sections. We are particularly grateful for the 11 invaluable references you recommended in your subsequent comments (Comment 2 and Comment 3). They have significantly broadened our perspective, enabling us to more comprehensively connect and dialogue our work with three key adjacent academic domains:

• Theoretical Level - Pricing, Carbon Reduction, and Government Intervention Strategies: We recognize that economic and policy instruments are core drivers of industrial transformation in the context of carbon neutrality. Relevant studies (e.g., Ma et al., 2022; Ma et al., 2021) have employed game theory models to explore optimal pricing and cooperation strategies for firms under carbon policy constraints. Our work extends this macro-level thinking to the building services sector, innovatively addressing how to translate high-level low-carbon economic objectives into micro-level, executable service modularization designs in a data-driven manner, thereby bridging the gap between theory and practice. (Pages 2, Lines 58-71)
• Technical Level - Optimization under Uncertainty and Energy System Scheduling: We have drawn upon advanced optimization methods for handling uncertainty in the energy sector (e.g., Wang et al., 2024; Meng et al., 2023). These studies provide powerful theoretical tools for the collaborative scheduling of complex energy systems, such as shared energy storage and district heating. Our framework contributes a novel application scenario to this field by leveraging Digital Twin (DT) technology to capture and predict uncertainties. Furthermore, we utilize our unique "co-intelligence" quantification model to achieve dynamic and robust optimization of the Central Air Conditioning (CAC) service ecosystem. (Pages 4, Lines 159-172)
• Application Domain - Sustainability Decision-Making in the Construction Industry: Your recommended studies on sustainability in the construction industry (e.g., Iqbal et al., 2021; Iqbal et al., 2025) have inspired us to recognize the vital role of systematic Multi-Criteria Decision-Making (MCDM) frameworks in designing sustainable solutions for complex physical systems like buildings. Our research not only adheres to this principle but also deepens it methodologically. Specifically, to address the unique "hybrid uncertainties" introduced by the DT environment, we have adopted more advanced rough-fuzzy methodologies, which constitutes a significant supplement and advancement to the research in this application domain. (Pages 7, Lines 294-299)

Through these in-depth revisions, we are confident that the revised manuscript now more clearly demonstrates that our proposed framework is not only an advancement in service modularization theory but also a meaningful contribution to the broader body of sustainability scholarship. Thank you once again for your invaluable guidance.

 

Comments 2: To enhance theoretical depth, I strongly recommend citing and discussing the following studies: i. Impact of extended warranty service on product pricing in online direct retailers' competitive market, ii. Evolution delayed decision game based on Carbon emission and capacity sharing in the Chinese market, iii. Optimal pricing and complex analysis for low-carbon apparel supply chains, iv. Pricing strategy and coordination of automobile manufacturers based on government intervention and carbon emission reduction, v. Incorporate robust optimization and demand defense for optimal planning of shared rental energy storage in multi-user industrial park, vi. Collaborative and effective scheduling of integrated energy systems with consideration of carbon restrictions, vii. Exploring the Impact of a District Sharing Strategy on Application Capacity and Carbon Emissions for Heating and Cooling with GSHP Systems, viii. A RL-based human behavior oriented optimal ventilation strategy for better energy efficiency and indoor air quality.

Response 2:

We are very grateful to you for recommending these high-quality studies to enhance the theoretical depth of our research. We have carefully studied all eight articles and recognize their high relevance to our work. As per your suggestion, we have integrated all of these references into the relevant sections of our revised manuscript, including the Introduction (Pages 2, Lines 58-71) and Literature Review (Pages 4, Lines 159-172).

Specifically, we have leveraged these studies as follows:

• References (i), (ii), (iii), and (iv) have helped us enrich the economic and strategic dimensions of our research background, demonstrating the importance of considering factors such as pricing, game theory, and government intervention under carbon constraints.
• References (v) and (vi) have provided us with advanced perspectives on system optimization under uncertainty, particularly in the areas of robust optimization and collaborative scheduling for energy systems. This has strongly supported the justification for our methodological choices.
• Reference (vii) illustrates the successful application of resource-sharing strategies in district energy systems, which aligns perfectly with our goal of promoting resource efficiency through service modularization.
• Reference (viii), as a case of applying a state-of-the-art optimization algorithm (Reinforcement Learning) in building energy efficiency, has offered valuable inspiration for the future research directions outlined in our paper.

By incorporating these works, we are confident that the theoretical foundation of our paper has been significantly strengthened, and our research's position within the relevant academic landscape is more clearly defined.

 

Comments 3: The use of rough-fuzzy DEMATEL, rough-fuzzy BWM, an improved Girvan–Newman algorithm, and IT2F-TOPSIS is technically rigorous, but the manuscript must justify why these methods are most appropriate for handling hybrid uncertainties. To strengthen this point, the authors could relate to: i. Circular economy solutions for net-zero carbon in China's construction sector: A strategic evaluation, ii. Uptake and adoption of sustainable energy technologies: Prioritizing strategies to overcome barriers in the construction industry by using an integrated AHP‐TOPSIS approach, iii. From waste to treasure: transforming construction waste into concrete masonry blocks: challenges and solutions for environmental sustainability in emerging economies.

Response 3:

Thank you for your valuable comment on the rigor of our methodology and for providing highly relevant references to strengthen our justification.

Following your excellent suggestion, we have added an in-depth discussion at the end of Section 2.4 in the Methodology section (Pages 7, Lines 294-310) to systematically justify why we selected the rough-fuzzy series of methods and IT2F-TOPSIS to handle the specific hybrid uncertainties inherent in our study.

Our justification is centered on the premise that the decision-making environment in a DT-driven service ecosystem is characterized by uncertainties that are both dual-natured and of a higher order, which traditional MCDM methods cannot fully capture:

• Duality: The process involves both the linguistic vagueness from individual expert judgments (addressed by Fuzzy Set Theory) and the cognitive uncertainty arising from information incompleteness in group decisions (addressed by Rough Set Theory). Our chosen combination of methods is capable of handling both types of uncertainty simultaneously.
• Higher-Order Nature: In the final module evaluation stage, significant disagreements among experts can make the uncertainty itself uncertain. Interval Type-2 Fuzzy Sets (IT2FS) are a powerful tool specifically designed to manage such higher-order uncertainties.

Furthermore, we have incorporated the references you recommended concerning sustainability in the construction industry (e.g., Iqbal et al., 2021; Iqbal et al., 2025a,2025b). These studies, which successfully apply methods like AHP-TOPSIS, validate the use of MCDM as an effective paradigm for complex decision-making in this domain. Building upon this, we argue that our research context is rendered more complex by the introduction of Digital Twins, thus necessitating a methodological upgrade. This positions our choice of methods as not only theoretically sound but also aligned with advanced practices in related application fields.

We are confident that these revisions have significantly enhanced the rationale and persuasiveness of our methodological choices. We once again extend our sincere gratitude for the reviewer's insightful comments and for bringing the relevant literature to our attention, which have been invaluable in strengthening our work.

 

Comments 4: The term “co-intelligence” is introduced as a novel element but lacks consistent theoretical grounding. It should be more explicitly linked to system synergy, knowledge co-creation, or collaborative intelligence frameworks to avoid being viewed as a relabeling of interdependency analysis.

Response 4:

Thank you for this critical and valuable comment regarding the theoretical grounding of the term "co-intelligence." Your point is well-taken, and we recognize the importance of providing a more solid theoretical foundation for this core concept.

Following your guidance, a more explicit definition and elaboration of co-intelligence has been provided in the Introduction (Pages 3, Lines 98-102) and Methodology (Section 3.4) sections (Pages 13, Lines 473-478).

1. Establishing Theoretical Links: The revised text now explicitly states that the concept of co-intelligence is built upon established theoretical frameworks such as system synergy and collaborative intelligence. It is defined as the complex, multi-faceted, and synergistic interrelationships among service components (spanning both physical assets and cyber algorithms) within a DT-enabled cyber-physical system, all working towards a common low-carbon objective. (Pages 3, Lines 98-102)
2. Highlighting the Core Innovation: It is further clarified that co-intelligence is not merely a relabeling of interdependency analysis. Its core innovation lies in the transition from qualitative description to quantitative measurement. While traditional interdependency analysis often focuses on structural or process-based links, the proposed co-intelligence concept is operationalized through a three-dimensional quantitative model encompassing Resource Dependency, Directed Overlap in Processes, and Direct Functional Correlation. This mathematical characterization of complex synergistic relationships is a novel attempt and a necessary prerequisite for the subsequent network modeling and optimization algorithms (such as the improved Girvan-Newman algorithm) used in our framework. (Pages 13, Lines 473-478)

Through these revisions, we hope to have clearly demonstrated both the theoretical lineage of co-intelligence and its unique contribution at the quantitative analysis level, thereby addressing your concern.

 

Comments 5: While the DT enabling layer is well described, its role in enhancing multi-criteria decision-making is not fully explained. The manuscript should clarify how real-time DT data supports robustness in demand analysis, prioritization, and module evaluation.

Response 5:

Thank you for raising this critical point. We acknowledge that a clearer explanation of the specific role the DT plays in enhancing the Multi-Criteria Decision-Making (MCDM) process is essential for bridging the theory and practice of our framework.

As per your recommendation, we have substantially expanded and detailed Section 5.1 in the Discussion section to explicitly clarify how real-time DT data supports and enhances the robustness of the three key decision-making stages:

• In Demand Analysis and Prioritization: We have clarified that the DT's role transcends mere data acquisition. By integrating real-time operational data (e.g., energy consumption patterns, equipment status, occupancy levels), it transforms what would be a static, expert-based demand analysis into a dynamic, data-driven process. This objective data provides more reliable inputs for prioritization models like rough-fuzzy BWM, reducing biases from purely subjective judgments and thus enhancing the analysis's robustness. (Pages 23, Lines 795-812)
• In Module Evaluation: We have now emphasized the DT's crucial function as a dynamic simulation and validation engine. During the IT2FS-TOPSIS evaluation phase, instead of relying solely on linguistic expert ratings, alternative modular solutions (e.g., Solutions A1, A2, A3) can be virtually "deployed" and simulated within the DT environment. The platform can then predict the performance metrics of each solution under various future scenarios (e.g., projected energy savings, maintenance costs, indoor environmental quality). These predictive, quantitative metrics serve as powerful inputs for the IT2FS-TOPSIS decision matrix. This significantly enhances the objectivity and reliability of the evaluation, ensuring that the selected modular solution is more robust in real-world operation. (Pages 23, Lines 820-832)

We believe these specific clarifications now adequately address your concern and fully demonstrate the pivotal enabling role of DT technology within our framework.

 

Comments 6: The Zhuhai office building case study is informative but largely descriptive. Quantitative evidence (e.g., reductions in energy use, cost savings, carbon intensity improvements) would provide stronger validation.

Response 6:

We fully agree with you that clear quantitative evidence is the cornerstone of validating our framework's effectiveness. Thank you for pointing out this crucial weakness in our results presentation.

Following your advice, we have significantly revised the case study section to highlight and systematize its quantitative outcomes:

1. Elevating Key Results: We have extracted the most critical Key Performance Indicators (KPIs) from the case study—such as the 74.29% integrated building energy saving rate, an annual electricity saving of 1.173 million kWh, and an annual carbon reduction of 618.5 tCO2—and have explicitly stated them in the Abstract and at the end of the Introduction of the revised manuscript. This ensures that the practical impact of our work is immediately visible to the reader. (Pages 1, Lines 21-23)
2. Adding a Summary Table: Within the Case Study section (Section 4), we have introduced a new summary table (Table 8: Summary of Key Performance Indicators (KPIs)). This table provides a clear, side-by-side comparison of the project's design targets and the actual quantitative results achieved through our framework, including metrics like the total equivalent electricity consumption (optimized to 27.53 kWh/m²·a from a target of ≤35 kWh/m²·a). (Pages 22, Lines 785-786)
3. Strengthening In-text Discussion: We have rewritten the descriptive text accompanying the case study results to be more focused on interpreting and analyzing these quantitative data, rather than merely describing system functionalities. (Pages 22-23, Lines 780-793)

With these revisions, we are confident that the case study's validation power is substantially strengthened, providing robust, data-driven support for our framework's effectiveness and practical value.

 

Comments 7: The improved Girvan–Newman algorithm is proposed as an innovation, yet no comparative results are presented against the standard GN or other clustering algorithms. Demonstrating superior performance through modularity indices or efficiency metrics is essential.

Response 7:

Thank you for this insightful and critical comment. You are absolutely correct that a quantitative comparison against standard algorithms is the gold standard for validating the performance of a proposed innovation.

Given the time constraints for this revision, conducting a comprehensive comparative study involving multiple algorithms was not feasible. However, to address your concern and to strengthen the rationale for our approach, we have made significant clarifications in the Methodology (Section 3.5.1) and Discussion sections of the manuscript:

• Elucidating the Theoretical Superiority: We have now detailed the core distinction between our improved Girvan-Newman (GN) algorithm and the standard version in the methodology section. The standard GN algorithm is primarily driven by the network's topology (i.e., the number of paths), making it agnostic to the weights of the edges when calculating betweenness centrality. In our co-intelligence network, however, the edge weights represent meaningful functional synergy. Our key innovation is the incorporation of these weights into the shortest path calculation. This modification ensures that the algorithm prioritizes the removal of edges that might be structurally central but are functionally weak, thus yielding modules that are more functionally cohesive and logically sound. This transforms the algorithm from a purely topology-driven method to a semantically-driven one, which is theoretically more appropriate for our specific application. (Pages 14, Lines 533-544)
• Acknowledging Limitations and Outlining Future Work: We fully concur with your point and have now explicitly acknowledged in the Discussion section that the absence of a direct comparative performance analysis is a limitation of this study. We have identified a comprehensive performance analysis—comparing our improved GN algorithm against standard GN, the Louvain method, and other clustering algorithms using metrics such as modularity indices and Normalized Mutual Information (NMI)—as a key direction for our future research. (Pages 25, Lines 917-919)

 

Comments 8: Validation is limited to one office building. The authors should discuss adaptability to other building types (residential, hospitals, industrial parks) and policy contexts, which would reinforce the framework’s broader applicability.

Response 8:

Thank you for this excellent suggestion. We fully agree that discussing the framework's adaptability is crucial for reinforcing its broader applicability. While we had an existing subsection (Section 5.2 Generalizability of the Proposed Framework) on generalizability, your comment has prompted us to significantly refine and expand upon it to make the arguments more explicit and concrete.

Following your specific guidance, we have now further refined and expanded this section to make the arguments more explicit and concrete. Specifically, we have enriched the "Adaptable Service Design and Modularization Model" portion by providing detailed examples of how the framework's input parameters can be tailored for different building types. For instance, we now explicitly discuss how the value propositions and co-intelligence analysis would shift when applied to residential buildings (focusing on comfort), hospitals (prioritizing air quality and reliability), and industrial parks (emphasizing synergy with production loads). (Pages 24, Lines 849-856)

We believe these enhancements, inspired by your valuable feedback, make the discussion on generalizability more robust and directly address the points you raised.

 

Comments 9: The paper references China’s carbon neutrality goals but does not translate findings into concrete policy implications. Discussing potential integration into green building codes, CAC industry standards, or carbon trading mechanisms would strengthen its societal relevance.

Response 9:

Thank you once again for your forward-thinking suggestions. Linking our research findings to concrete policy implications indeed significantly strengthens the societal relevance of our paper.

As per your advice, we have added a new subsection titled "Policy Implications" to the Discussion section 5.3 (Pages 24-25, Lines 877-888). In this subsection, we discuss how our research can inform the development and implementation of relevant policies, and we have cited some of the highly relevant literature you previously recommended to support our arguments:

• Informing Green Building Codes: We argue that current green building codes are often based on static design metrics. Our framework demonstrates an evaluation methodology based on dynamic performance and system synergy. This suggests that future green building standards could incorporate more dynamic, lifecycle-based carbon performance indicators enabled by Digital Twins.
• Implications for CAC Industry Standards: Our modularization methodology provides a theoretical foundation for the CAC industry to develop standardized, interoperable service interfaces and modules. This could facilitate the industry's transition from selling products to providing flexible low-carbon service solutions, aligning with broader goals of government intervention and carbon reduction policies (Ma, Hou, et al., 2021).
• Integration with Carbon Trading Mechanisms: We discuss how the co-intelligence analysis within our framework could be integrated with carbon trading mechanisms. For example, by precisely quantifying the carbon reduction benefits of different service module configurations via the DT platform, our approach can provide a more reliable data foundation for enterprises participating in carbon markets. This would incentivize the adoption of more efficient low-carbon service solutions, a strategy consistent with promoting resource sharing and carbon reduction at the district level (Ma, Si, et al., 2022; Zhang et al., 2020).

We hope this discussion effectively addresses your concerns and demonstrates the potential societal value of our research.

 

Comments 10: Although multi-stakeholder needs are mentioned, the case study emphasizes technical aspects. Incorporating insights from facility managers, service providers, and end-users would demonstrate that the framework addresses both technical and social dimensions of value creation.

Response 10:

Thank you for this valuable comment. We fully agree that demonstrating how our framework integrates both technical and social dimensions is essential to showcase its comprehensive approach to value creation.

We have reflected on our original description of the case study and concede that it was overly focused on technical aspects, failing to adequately describe the multi-stakeholder engagement process. As per your recommendation, we have made significant additions to the Case Study section (Section 4.2.1 Low-Carbon Value Proposition and Service Demand Analysis) to clarify this process (Pages 18, Lines 673-685).

In the revised manuscript, we now explicitly describe how insights from various stakeholders were actively incorporated during the initial value proposition and demand analysis phase. Specifically, we have added descriptions of the following:

• We organized a series of workshops and in-depth interviews to engage deeply with building asset owners, facility managers, O&M personnel, and service providers.
• Through these interactions, we collected and integrated their diverse perspectives. For instance, facility managers emphasized system maintainability, operational efficiency, and ease of fault diagnosis. Service providers were concerned with service standardization and cost-effectiveness. End-users (tenants) focused on indoor comfort, health, and personalized control experiences.
• These multi-faceted needs from various social dimensions were systematically mapped into our three-dimensional demand identification model and subsequently quantified and prioritized using the rough-fuzzy methodology.

We are confident that with these specific additions, the manuscript now more clearly demonstrates how our framework effectively addresses both the technical and social dimensions of value creation in practice.

 

Comments 11: The uncertainty analysis focuses mainly on expert evaluations. Real-world operational uncertainties such as fluctuating energy prices, sensor failures, and weather variations should be acknowledged and linked to DT’s predictive capabilities.

Response 11:

Thank you for this insightful observation. We fully agree that a robust framework must acknowledge and be capable of addressing real-world operational uncertainties.

Following your recommendation, we have revised the manuscript to better articulate how our framework addresses these uncertainties. In the Methodology section 3.3 (Pages 11, Lines 441-442), we now explicitly state that our demand analysis model acknowledges real-world operational variables such as fluctuating energy prices and weather variations. Furthermore, in the Discussion section 5.1 (Pages 23, Lines 807-812), we have elaborated on how the DT's predictive capabilities are specifically designed to monitor, simulate, and ultimately mitigate the impact of these external volatilities, enabling proactive and adaptive system control. We believe this makes the connection between the problem of uncertainty and our proposed solution clearer.

 

Comments 12: The economic dimension is treated superficially. A cost–benefit or life-cycle cost analysis would help substantiate the claim that modularization yields not only carbon reductions but also tangible financial benefits.

Response 12:

Thank you for highlighting the need for a more thorough treatment of the economic dimension. We agree that demonstrating tangible financial benefits is crucial.

To address your point, we have strengthened the manuscript in two key areas. First, in the Methodology section 3.5.2, we have expanded the definition of our Economic Efficiency (EE) evaluation criterion to more explicitly emphasize its foundation in life-cycle cost principles (Pages 15, Lines 574-575). Second, we have added a new subsection on "5.3 Economic and Policy Implications" in the Discussion section (Pages 24, Lines 869-876). This new part leverages the quantitative energy-saving results from our case study to discuss the tangible financial benefits, thereby providing stronger evidence for the economic viability of our approach.

 

Comments 13: The manuscript emphasizes system modularization but pays little attention to occupant-centered aspects such as comfort, satisfaction, and behavior. Integrating human-centered design principles would broaden the framework’s relevance.

Response 13:

Thank you for drawing our attention to the importance of human-centered aspects. We acknowledge that the original manuscript could have better articulated the link between system modularization and the occupant experience.

To clarify this, we have made revisions in two main places. In the Methodology section 3.3 (Pages 11, Lines 433-435), we now explicitly state that the Social Benefits dimension of our value model is designed to encapsulate human-centered design principles, including occupant comfort, health, and satisfaction. Subsequently, in the Case Study section 4.3 (Pages 21-22, Lines 764-767), we elaborate on how the final selected modular solution, particularly the Smart Occupant Experience Module, directly materializes these principles by delivering adaptive environmental controls for end-users. This demonstrates how human-centered goals are embedded in our framework from initial design through to final implementation.

 

Comments 14: The conclusion outlines general extensions but could be more specific. Testing the framework in a real-time DT platform, applying machine learning for predictive modularization, or extending to multi-building district energy systems would provide a sharper research roadmap.

Response 14:

Thank you for providing such a sharp and specific research roadmap. Your suggestions—testing the framework on a real-time DT platform, applying machine learning for predictive modularization, and extending it to multi-building district energy systems—are both forward-thinking and highly insightful.

Based on your invaluable advice, we have completely restructured and enriched the " 7. Recommendations for Future Research " section (Pages 25, Lines 908-923). We have adopted your proposed directions as the core themes for our future research. The revised section now outlines a clear progression, beginning with real-time empirical validation, moving to methodological enhancement via machine learning, and culminating in the extension of the framework's scope to larger, district-level systems. We have also integrated other research avenues, such as a formal algorithmic comparison and the modeling of dynamic market conditions, into this new, more specific structure.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

File attached 

Comments for author File: Comments.pdf

Author Response

Comments 1: Abstract: The abstract is presented very well, but it could be improved by adding specific results to make it more attractive.

Response 1:

Thank you for your positive feedback and this excellent suggestion. We agree that adding specific quantitative results to the abstract makes our contribution more impactful. We have revised the abstract to include the key performance outcomes from our case study, such as the 74.29% integrated energy saving rate and the annual carbon reduction of 618.5 tCO2. We believe the abstract is now more attractive and compelling, as you recommended. (Pages 1, Lines 21-23)

 

Comments 2: Introduction: The introduction is well highlighted and covers all relevant aspects.

Response 2:

Thank you very much for your positive assessment of our Introduction. We appreciate your encouragement.

 

Comments 3: Literature Review: Include the latest literature from 2024 and 2025 to highlight the most recent research.

Response 3:

Thank you for this important suggestion. To ensure our research is contextualized within the very latest scholarly conversations, we have now included several recent publications from 2024 and 2025 in our Literature Review. These additions help to highlight the most current research trends and further sharpen the novelty of our work.

 

Comments 4: Research Methods: The research methods are written very well.

Response 4:

We are grateful for your positive feedback on our Research Methods section. Thank you.

 

Comments 5: Results: The results are presented clearly, but to make them more attractive, include at least one literature reference per section.

Response 5:

Thank you for this suggestion. We understand your point about making the results section more attractive by connecting our findings to the broader literature. While our Results section (Section 4) is dedicated to presenting our own empirical findings, we have ensured that in the Discussion section (Section 5), where we interpret these results, we now explicitly reference and compare our findings with those of prior studies. For example, when discussing the economic and policy implications of our results, we have cited relevant literature to contextualize our contributions (Pages 23, Lines 884 and Pages 24，Lines 887，888). We believe this approach achieves your goal of linking our results to existing research while maintaining the standard structure of a scientific paper.

 

Comments 6: Discussion: In the discussion, please provide references to the relevant figure numbers or table numbers.

Response 6:

Thank you for pointing out this oversight. You are absolutely right that referencing the specific figures and tables makes the discussion much clearer and easier to follow. We have gone through the Discussion Section 5 and have added explicit references (e.g., " as illustrated in Figure 3", "as shown in Table 8") wherever we discuss specific findings, ensuring a clear link between our interpretations and the data presented.

 

Comments 7: Conclusion: Conclude a separate section.

Response 7:

Thank you for this clear and helpful suggestion. Following your advice, we have restructured the end of the manuscript. The original combined 'Conclusions and Future Work' section has now been separated, and Section 6 is now a dedicated "Conclusions" section, focusing solely on summarizing the key findings of our research (Pages 25, Lines 890-908). We agree that this separation improves the clarity and impact of the paper's conclusion.

 

Comments 8: Recommendations: Instead of using Future Work, use the heading Recommendations

Response 8:

Thank you for this excellent suggestion on the terminology. We have implemented this change precisely as you recommended. The section outlining our future research directions, which now follows the new "Conclusions" section, has been retitled as "Section 7: Recommendations for Future Research. (Pages 25, Lines 909-924)" We believe this heading better reflects the specific and actionable research paths we now propose.

We would like to thank you again for all your valuable suggestions. We particularly appreciate your positive evaluation and encouragement of our work, as well as your clear guidance on the paper's structure and presentation. Your feedback has been very helpful in improving the overall quality and readability of the manuscript.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

The title of the article corresponds to the area of ​​research, and the Abstract reflects the content of the paper. The structure of the manuscript as a whole corresponds to a scientific article. However, after reading it, I consider it necessary to note that:

1. The article presents a description of a large and interesting work done by the authors, all stages of which are described in detail. But, unfortunately, the authors' article turned out to be more journalistic than scientific. It presents the product result well, but I did not see any scientific content.
2. In the Literature Review section, the authors limit themselves to general phrases describing the ideas and results of other researchers, but there is no specific critical analysis of them.
3. I understand that the big work done by the authors requires a detailed description, however, in the main third and fourth sections of the manuscript, the scientific ideas and scientific results that the authors would like to convey to the readers are lost behind the voluminous text. I suggest that the authors reduce the volume of the manuscript and present it in a more specific form.
4. The fifth section, Discussion, and the sixth section, Conclusions and Further Work, largely repeat what was presented in the main part of the manuscript. It seems to me that the material in these two sections should also be made more specific, and the volume of text in them can be shortened.

Thus, in my opinion, the paper in this form may be interesting to readers, but its content has no value for publication in a highly rated scientific journal. I suggest that the authors substantially revise the manuscript.

Author Response

Comments 1: The article presents a description of a large and interesting work done by the authors, all stages of which are described in detail. But, unfortunately, the authors' article turned out to be more journalistic than scientific. It presents the product result well, but I did not see any scientific content.

Response 1:

Thank you for this frank and incredibly valuable feedback. We have taken your critique very seriously and understand that the original manuscript's presentation obscured its scientific contributions.

To address this, we have undertaken a fundamental revision of the paper to shift its focus from a descriptive narrative to a rigorous scientific exposition. Key changes include:

• Highlighting Quantitative Achievements: We have extracted the key quantitative results (e.g., 74.29% energy saving rate, 618.5 tCO2 annual reduction) and placed them prominently in the Abstract to immediately showcase the measurable outcomes of our framework (Pages 1, Lines 21-23). A new summary table (Table 8) has also been added to the Case Study section for stronger validation (Pages 22, Line 780).
• Strengthening Theoretical Grounding: The Introduction and Literature Review have been substantially rewritten to critically analyze and integrate our work with broader academic scholarship in sustainability, including optimization under uncertainty and carbon-constrained economics, thereby strengthening its scientific context. (Pages 2, Lines 58-71; Page 4, Lines 159-172; Page 5, Lines 223-230; and Page 7, Lines 294-310)

We are confident that these changes have fundamentally transformed the manuscript from a journalistic presentation into a focused scientific paper that clearly articulates its methodological innovations and empirical contributions.

 

Comments 2: In the Literature Review section, the authors limit themselves to general phrases describing the ideas and results of other researchers, but there is no specific critical analysis of them.

Response 2:

Thank you for pointing out this significant weakness in our literature review. You are correct that the original version lacked critical depth.

Following your advice, we have completely overhauled Section 2 (Literature Review). Instead of merely summarizing previous work, in the revised section, we now critically analyze the state-of-the-art in relevant domains to precisely identify the research gaps that our framework aims to fill. This new approach provides a much stronger analytical foundation for our research and directly demonstrates how our work contributes to and extends the existing body of scientific knowledge. (Page 4, Lines 159-172; Page 6, Table 1; and Page 7, Lines 294-310)

 

Comments 3: I understand that the big work done by the authors requires a detailed description, however, in the main third and fourth sections of the manuscript, the scientific ideas and scientific results that the authors would like to convey to the readers are lost behind the voluminous text. I suggest that the authors reduce the volume of the manuscript and present it in a more specific form.

Response 3:

Thank you for this precise and actionable suggestion. We agree completely that the core scientific ideas in Sections 3 and 4 were diluted by excessive descriptive text.

As you recommended, we have significantly reduced the volume of the manuscript and sharpened the focus of Section 3 (Methodology) and Section 4 (Case Study). Our streamlining strategy involved removing procedural redundancies in both the text and tables, and shifting the narrative to highlight the core scientific contributions:

• In Section 3 (Methodology), the step-by-step descriptions for demand analysis (3.3) Pages 11-12, Lines 430-446) and module evaluation (3.5.2) (Page 15, Lines 566-575) were rewritten into concise, integrated paragraphs focusing on the scientific rationale. Furthermore, to improve the flow, the detailed quantification scale tables (Tables 1-3) for our co-intelligence model have been moved to Appendix A. (Page 13, Lines 473-493)
• In Section 4 (Case Study), the narrative was shifted from describing the case to interpreting the results. Lengthy descriptive passages about the building were removed, and the text now concentrates on analyzing the scientific insights. Similarly, the detailed IT2 Fuzzy Decision Matrix (the original Table 9) has been moved to Appendix B to streamline the presentation of the final evaluation. (Page 18-21, Lines 673-733)

We are confident that these comprehensive revisions—condensing text, moving detailed tables to appendices, and focusing the narrative on scientific insights—have made Sections 3 and 4 more specific and significantly shorter, allowing the core ideas and results to emerge much more clearly, as you wisely advised.

 

Comments 4: The fifth section, Discussion, and the sixth section, Conclusions and Further Work, largely repeat what was presented in the main part of the manuscript. It seems to me that the material in these two sections should also be made more specific, and the volume of text in them can be shortened.

Response 4:

Thank you for this critical and highly constructive feedback. We have carefully re-examined our manuscript and agree that the original Discussion and Conclusion sections contained significant redundancies and lacked specificity.

Following your advice, we have undertaken a substantial revision and restructuring of these final sections to make them more concise, specific, and impactful:

• Streamlining the Discussion Section (Section 5): We have made significant cuts to shorten the volume of this section. Specifically, the entire original subsection 5.1 ("Key Findings and Practical Implications") has been removed, as we recognized that it largely repeated findings and concepts already presented in the Introduction and Methodology sections. The remaining subsections have been renumbered and refined. This major deletion directly addresses your concern about repetition.
• Enhancing Specificity in the Discussion: While shortening the overall length, we have simultaneously increased the specificity by adding new, targeted subsections. For example, we have introduced Section 5.3 (Economic and Policy Implications). These new sections do not repeat previous content but instead offer fresh, specific interpretations of our findings' broader applicability and relevance, directly addressing your call for more specific material. (Page 24-25, Lines 865-889)
• Refining the Conclusions and Recommendations: The original "Conclusions and Future Work" section has been split into two new, distinct sections: Section 6 (Conclusions) and Section 7 (Recommendations for Future Research). The new Conclusions section has been rewritten to be a concise summary of our core contributions, removing any repetitive methodological descriptions. The new Recommendations section is now highly specific, outlining a concrete research roadmap with actionable steps, moving away from the general statements of the previous version. (Page 25, Lines 890-924)

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

My Comments can be found in the attached document.

Comments for author File: Comments.pdf

Author Response

Comments 1: The abstract is comprehensive, but it is a little bit long, especially when explaining the methods. Please make it more concise by putting more focus on the novelty, contributions, and findings.

Response 1:

Thank you for this excellent suggestion. We agree that the previous abstract was overly detailed in its methodological description, which may have diluted the core highlights of our research. Our intention was to showcase the methodological rigor, but we understand that this came at the expense of conciseness.

Following your advice, we have carefully revised the abstract to make it more concise while retaining the core novelty, contributions, and findings (Page 1, Lines 10-25). Specifically, our revisions focused on:

• Condensing the Methodological Narrative: Instead of explaining the purpose of each method in detail, we now mention the key innovative methods more directly. For example, we removed descriptive phrases like "informed by TBL..." and "for achieving flexible, re-usable...".
• Highlighting Core Innovations: The revised text now places a clearer emphasis on the most novel contributions, ensuring that key terms such as the "Energy-Scenario-Intelligence model," "quantitative co-intelligence relationship analysis," "improved Girvan-Newman algorithm," and "Interval Type-2 Fuzzy TOPSIS" are preserved but presented in a more streamlined manner.
• Accentuating the Key Findings: We have retained and highlighted the key quantitative findings from the case study (e.g., the 74.29% integrated energy saving rate) as the core evidence of the framework's effectiveness.

We are confident that the revised abstract now strikes a better balance, effectively communicating the scientific contributions and key results in a more concise and impactful way, as you recommended.

 

Comments 2: “Lines 137–147”: The integration of TBL with SPSS is well explained, but slightly repetitive with earlier mentions in the Introduction. Please consider condensing.

Response 2:

Thank you for this sharp observation. Upon review, we agree that the explanation of the Triple Bottom Line (TBL) concept in Section 2.1 was indeed repetitive of earlier mentions in the Introduction.

As you suggested, we have now condensed this passage. The revised version avoids redefining the TBL concept and instead more concisely links the design of low-carbon SPSS directly to the principles of the TBL framework. We are confident that this change removes the redundancy and improves the overall flow of the manuscript. (Page 4, Lines 144-146)

 

Comments 3: “Lines 154–190”: The Digital Twin section is well-developed but could be improved by citing additional building energy management studies beyond HVAC, such as lighting, occupant comfort, and smart grids. This would expand the scope of the discussion.

Response 3:

Thank you for this constructive suggestion. We fully agree that expanding our discussion beyond HVAC to the broader field of building energy management effectively showcases the versatility and potential of Digital Twin (DT) technology.

Following your guidance, we have now enriched Section 2.2 (Digital Twin in Product-Service Systems and Building Applications) of the literature review. We have added citations and a brief discussion on the application of DTs in areas such as smart lighting, occupant comfort modeling, and smart grid integration. (Page 4-5, Lines 187-192)

We are confident that these additions, as you suggested, have successfully expanded the scope of this section and further strengthen the argument for DT as a core enabling technology for sustainable building operations.

 

Comments 4: “Lines 191-216”: the discussion regarding MCDM could be improved for more readability by adding a summary table comparing methods (AHP, ANP, DEMATEL, BWM) with strengths and weaknesses.

Response 4:

This is an excellent suggestion, thank you. We fully agree that a summary table would greatly enhance the readability and clarity of our discussion on MCDM methods.

As per your guidance, we have now added a new table (Table 1: Comparison of Key MCDM Methods) in Section 2.3 of the Literature Review. This table systematically compares the key methods mentioned—AHP, ANP, DEMATEL, and BWM—based on their Strengths and Weaknesses with respect to handling dependencies, data requirements, decision complexity, and capability to handle uncertainty. (Page 6, Line 240)

We have also referenced this new table in the text and used its insights to reinforce why we chose DEMATEL and BWM for different stages of our research, and ultimately, why advanced approaches like rough-fuzzy methods were necessary to overcome the limitations of these traditional tools. We are confident this revision significantly improves the clarity and professionalism of this section.

 

Comments 5: “Lines 272–336”: Figures 1 & 3 are very detailed. Add 1–2 sentences explaining how each figure directly contributes to solving the research problem.

Response 5:

Thank you for this valuable suggestion on improving the clarity of our figures. We agree that adding a concluding sentence to explain the core contribution of these complex diagrams helps the reader to grasp their purpose more quickly.

As per your guidance, we have now added 1-2 explanatory sentences for both Figure 1 and Figure 3 in the main text.

• For Figure 1, we have added a sentence where it is referenced, clarifying that the overall framework diagram's primary role is to "systematically transform macroscopic low-carbon objectives into executable, modular service solutions." (Page 8, Lines 322-324)
• For Figure 3, we have likewise added a clarification, stating that the figure's core contribution is to illustrate "a data- and model-driven closed-loop optimization mechanism, showing how the Digital Twin enables a continuous sense-analyze-decide-actuate feedback loop between the physical and cyber worlds." (Page 10, Lines 382-385)

We are confident that these added explanations will allow readers to more clearly understand the specific role each figure plays in solving our central research problem.

 

Comments 6: “Lines 499–543”: The improved Girvan-Newman algorithm is a good contribution, but the novelty could be more explicitly emphasized (how exactly it differs from the traditional GN algorithm).

Response 6:

Thank you for this precise and valuable suggestion. We fully agree that the novelty of our improved Girvan-Newman (GN) algorithm needed to be more explicitly emphasized.

Following your guidance, we have rewritten and expanded the description of the algorithm in the Methodology section (Section 3.5.1) (Page 14, Lines 533-544). The revised text now clearly articulates the exact difference from the traditional GN algorithm along one core dimension: the driving logic.

• Traditional GN is Topology-Driven: We explain that the standard GN algorithm is purely topology-driven, as it calculates betweenness centrality based on the number of shortest paths without considering edge weights. This makes it suboptimal for networks where edge weights carry significant physical meaning, such as the synergistic relationships in our co-intelligence network.
• Our Improved Algorithm is Semantically-Driven: We now emphasize that our core novelty is in adapting the algorithm to be semantically-driven. This is achieved by incorporating the "co-intelligence" value as the inverse of the edge weight in the shortest path calculations. This means that edges with stronger synergistic relationships are treated as "shorter" paths. This fundamental change ensures that the algorithm prioritizes the preservation of functionally cohesive links, rather than just structurally important ones, during its iterative removal process.

By drawing this clear contrast between a topology-driven and a semantically-driven approach, we are confident that we have now explicitly highlighted how exactly it differs and have fully addressed your concern.

 

Comments 7: “Tables 6-10”: consider keeping simplified versions of the tables in the main manuscript and moving the detailed ones to the appendix.

Response 7:

Thank you for this valuable suggestion regarding the structure and readability of our tables. We agree that moving overly detailed tables to the appendix can significantly improve the flow of the main manuscript.

Following your advice, we have carefully reviewed all relevant tables. To enhance readability and streamline the methodological sections, we have moved Table 9 (IT2 Fuzzy Decision Matrix), which contains complex Interval Type-2 Fuzzy Numbers, to Appendix B. This allows readers to focus on the key analysis in the main text while still providing the full data for reference. (Page 20, Lines 720-726)

Furthermore, building upon the spirit of your suggestion for improved conciseness, and in line with our previous revisions to streamline the methodology, we have also moved Tables 1, 2, and 3 (Quantification Scales for Co-Intelligence Relationships) to Appendix A. These tables, while crucial for method transparency, detail specific quantification scales that are better suited for an appendix, allowing the main text of Section 3.4 to focus on the conceptual contribution of the co-intelligence model. (Page 13, Lines 490-493)

We are confident that these changes significantly enhance the manuscript's readability and logical flow, as you suggested, by optimizing the placement of detailed data.

 

Comments 8: Please expand the “future work” in the conclusion section by addressing scalability, integration with renewable energy, and barriers to industrial adoption.

Response 8:

Thank you for your excellent suggestion to expand the "future work" section. We agree that addressing specific future challenges would provide a clearer roadmap. As per your advice, we have completely restructured this section (now Section 7: Recommendations for Future Research) to be more specific. The new recommendations now explicitly address key themes, including scalability (by extending to multi-building systems), integration with renewable energy, and barriers to industrial adoption (such as data governance and the need for real-world validation). (Page 25, Lines 909-924)

 

Comments 9: Some sentences are overly long; please consider splitting them into two sentences if possible. This would increase readability. It’s currently very brief.

Response 9:

Thank you for this valuable advice on sentence structure and readability. We agree that splitting overly long and complex sentences is crucial for improving the overall clarity of the manuscript.

Following your guidance, we have now performed a comprehensive language polish and edit throughout the entire manuscript. We paid special attention to identifying sentences containing multiple clauses or complex logic and have split them into two or more shorter, clearer sentences where appropriate. This effort was applied across all sections, particularly in the Methodology and Discussion, where we aimed to ensure that each core point is now conveyed with more direct and accessible language.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Accepted

Author Response

Comments and Suggestions for Authors

Accepted

Thank you for your profoundly helpful and insightful comments.

Reviewer 3 Report

Comments and Suggestions for Authors

The authors provided responses to my comments and made the necessary additions to the paper. I think that in this form the article can be recommended for publication.

Author Response

Comments and Suggestions for Authors

The authors provided responses to my comments and made the necessary additions to the paper. I think that in this form the article can be recommended for publication.

Thank you for your profoundly helpful and insightful comments.

Reviewer 4 Report

Comments and Suggestions for Authors

My comments are properly addressed in the revised version. Thanks!

Author Response

Comments and Suggestions for Authors

My comments are properly addressed in the revised version. Thanks!

 

Thank you for your profoundly helpful and insightful comments.