Environmental Risk Mitigation via Deep Learning Modeling of Compressive Strength in Green Concrete Incorporating Incinerator Ash

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper investigates an interesting topic such as Environmental Risk Mitigation via Deep Learning Modeling of Compressive Strength in Green Concrete Incorporating Incinerator Ash. The methodology is not well described and there are many details missing. 

Introduction 

Citations need to be increased in order to draw a background of the topid.

Section 2

The authors needs to discuss the method they used: Convolutional Neural Networks (CNN) and Multivariate Optimization Algorithm (MVO).

Section 4

Reproducibility is fundamental in scientific papers: the authors need to give details of all the procedures they followed. This part needs to be significantly expanded. 

The criteria adopted in the selection of the five input parameters

is needed. 

 

Author Response

Introduction 

Citations need to be increased in order to draw a background of the topic.

R: Based on the esteemed reviewer's opinion, four updated articles have been added to the introduction section.

Section 2

The authors needs to discuss the method they used: Convolutional Neural Networks (CNN) and Multivariate Optimization Algorithm (MVO).

R: Based on the esteemed reviewer's comment, two paragraphs have been added to the methodology section. Additional details can be provided, but they have been omitted due to the high volume of the article. These details can be made available at the reviewer's discretion.

Section 4

Reproducibility is fundamental in scientific papers: the authors need to give details of all the procedures they followed. This part needs to be significantly expanded. 

R: Section 4 has been completed with more detailed information about the neural network and optimization algorithm. If needed, the coding can be provided.

The criteria adopted in the selection of the five input parameters is needed. 

R: The selection of input parameters for neural network modeling and optimization was explained in a new paragraph before table 2.

Thank you for the reviewer's kindness.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

This research paper presents a comprehensive study on the use of deep learning methods to model and predict the compressive strength of green concrete incorporating incinerator ash. The authors have conducted a thorough experimental program and employed advanced modeling techniques, including Convolutional Neural Networks (CNN) and Multi-Verse Optimization (MVO) algorithms. The paper demonstrates a good understanding of the subject matter and provides valuable insights into the potential of using incinerator ash in concrete production. However, there are several areas where the manuscript could be improved in terms of clarity, organization, and scientific rigor.

1. The abstract is too long and contains unnecessary details. It should be condensed to focus on the main objectives, methods, and key findings.
2. The introduction lacks a clear statement of the research gap and specific objectives. These should be explicitly stated to better frame the study.
3. The methodology section needs a more structured approach. Consider using subheadings to clearly delineate the experimental design, data collection, and modeling processes.
4. The description of the deep learning model architecture is inadequate. Provide more details on the CNN structure, number of layers, and hyperparameters used.
5. The rationale for the selection of the specific input parameters (CA, FA, AW, C, W) should be explained in more detail with reference to the relevant literature.
6. The results section would benefit from a more systematic presentation of the results, including statistical analyses to support the conclusions drawn.
7. The discussion of the results of the sensitivity analysis is somewhat unclear. Consider using a table or graph to summarize the relative sensitivities of different input parameters.
8. The conclusion section should be more concise, focusing on the main findings and their implications rather than repeating information from earlier sections.
9. The use of English throughout the paper needs improvement. There are numerous grammatical errors and awkward phrasing that should be corrected.
10. Figures and tables should be of higher quality and resolution. Some captions and legends are difficult to read.
11. The literature review in the introduction could be more comprehensive, including more recent studies on green concrete and machine learning applications in concrete technology.
12. The limitations of the study should be explicitly stated, along with suggestions for future research directions.
13. The paper would benefit from a brief discussion of the practical implications of using incinerator ash in concrete production, including potential challenges and benefits.
14. Consider adding a brief section on the environmental impact and sustainability aspects of using incinerator ash in concrete to strengthen the relevance of the study.

Comments on the Quality of English Language

Overall, while the English quality is sufficient for conveying the research content, some refinement would enhance its clarity and professionalism for publication in an international journal.

Author Response

The abstract is too long and contains unnecessary details. It should be condensed to focus on the main objectives, methods, and key findings.

R: The abstract has been rewritten and summarized.

2. The introduction lacks a clear statement of the research gap and specific objectives. These should be explicitly stated to better frame the study.

R: A full paragraph has been added to the introduction section to explain the research gap.

3. The methodology section needs a more structured approach. Consider using subheadings to clearly delineate the experimental design, data collection, and modeling processes.

R: The methodology section has been rewritten with several new subsections based on the opinions of the respected referee and is marked in red.

4. The description of the deep learning model architecture is inadequate. Provide more details on the CNN structure, number of layers, and hyperparameters used.

R: More details of the CNN neural network structures and MVO algorithm model are added in Sections 2.1 and 2.2 and Sections 4.1.1 and 4.1.2. Detailed parameters are also provided. In addition, the entire coding is available.

5. The rationale for the selection of the specific input parameters (CA, FA, AW, C, W) should be explained in more detail with reference to the relevant literature.

R: The selection of input parameters for neural network modeling and optimization was explained in a new paragraph before table 2.

6. The results section would benefit from a more systematic presentation of the results, including statistical analyses to support the conclusions drawn.

R: Due to the large size of the article, the statistical results of all models and the statistical comparison of the laboratory program have not been presented, but they are available and can be presented.

7. The discussion of the results of the sensitivity analysis is somewhat unclear. Consider using a table or graph to summarize the relative sensitivities of different input parameters.

R: Sensitivity analysis in this paper is performed to investigate the effect of input parameters on output and the method of performing it can also be presented in full. This process is completed based on the method of Lu et al. based on the relative and absolute derivatives of the output with respect to the inputs.

8. The conclusion section should be more concise, focusing on the main findings and their implications rather than repeating information from earlier sections.

R: The conclusion section has been rewritten and summarized based on the opinion of the esteemed referee.

9. The use of English throughout the paper needs improvement. There are numerous grammatical errors and awkward phrasing that should be corrected.

R: The English language of the article will be revised by the journal's correction and editing site ( https://www.mdpi.com/authors/english) after technical corrections based on the comments of the respected referees.

10. Figures and tables should be of higher quality and resolution. Some captions and legends are difficult to read.

R: The Figures of the article will be revised by the journal's correction and editing site ( https://www.mdpi.com/authors/english) after technical corrections based on the comments of the respected referees.

11. The literature review in the introduction could be more comprehensive, including more recent studies on green concrete and machine learning applications in concrete technology.

R: Based on the esteemed reviewer's opinion, four updated articles have been added to the introduction section.

12. The limitations of the study should be explicitly stated, along with suggestions for future research directions.

R: Research limitations are presented at the end of the introduction section and research suggestions are presented at the end of the conclusion section.

13. The paper would benefit from a brief discussion of the practical implications of using incinerator ash in concrete production, including potential challenges and benefits.

R: Based on the reviewer's comments, Section 3.1, entitled Benefits of Using Waste Incineration Ash in Concrete, has been added to the article.

14. Consider adding a brief section on the environmental impact and sustainability aspects of using incinerator ash in concrete to strengthen the relevance of the study.

R: This is also briefly presented in section 3.1.

Comments on the Quality of English Language

Overall, while the English quality is sufficient for conveying the research content, some refinement would enhance its clarity and professionalism for publication in an international journal.

R: Thanks to the kind comments of the respected referee, the article will be revised in accordance with the response to comment number 9 in English.

Thank you for the reviewer's kindness.

 

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

(Buildings-2025-001) The study effectively integrates deep learning with green concrete research. Strengthening experimental validation, refining sensitivity analysis, and discussing model limitations could enhance its reliability and practical application.

1. The manuscript emphasizes the role of incinerator ash but lacks a comparative analysis with other green concrete additives. How does incinerator ash compare to fly ash or silica fume in similar conditions?
2. The statement "MATLAB 9.5 (2018b) was employed for coding, training, and evaluation" does not specify hyperparameter tuning methods. Could further details on optimization strategies improve reproducibility?
3. "The regression coefficient (R) of 90% reflects the deep learning model’s accuracy..."—consider providing confidence intervals for this metric to clarify its statistical robustness.
4. The methodology section does not discuss data partitioning ratios for training, validation, and testing. Were standard practices such as 70-15-15 or k-fold cross-validation used?
5. The impact of water content on compressive strength is well-documented, yet the sensitivity analysis suggests unexpected trends. Does this align with known concrete mix behavior?
6. The research effectively applies CNN and MVO, but a discussion of alternative machine learning models (e.g., Random Forest, XGBoost) for predictive strength modeling would add depth.
7. The introduction cites global environmental impacts of cement but does not provide regional data on Iran’s specific concrete consumption and waste management. Would this context strengthen the study's motivation?
8. "The error index reveals an average error of 0.14"—is this RMSE or MSE? Clarifying the error metric would improve interpretability for readers unfamiliar with the methodology.

 

Author Response

The manuscript emphasizes the role of incinerator ash but lacks a comparative analysis with other green concrete additives. How does incinerator ash compare to fly ash or silica fume in similar conditions?

R: A comparison with other additives for green concrete has not been made due to the very high volume of the article and the main goal of the article, which is to model the compressive strength of green concrete with incinerator ash using deep learning. Of course, comparisons with previous research on the use of incinerator ash in concrete have been made, but due to the high volume of the article, they were not included. If deemed appropriate by the honorable reviewer, they can be provided. The comparison chart is presented below.

As can be seen in the figure above, in the study by Nataraja et al. 2023, the effect of incinerator ash was initially positive at lower percentages but later turned negative. This finding is consistent with the results of the present study on high-grade concrete and indicates that the laboratory program results are acceptable.

The statement "MATLAB 9.5 (2018b) was employed for coding, training, and evaluation" does not specify hyperparameter tuning methods. Could further details on optimization strategies improve reproducibility?

R: More details of the CNN neural network structures and MVO algorithm model are added in Sections 2.1 and 2.2 and Sections 4.1.1 and 4.1.2. Detailed parameters are also provided. In addition, the entire coding is available.

"The regression coefficient (R) of 90% reflects the deep learning model’s accuracy..."—consider providing confidence intervals for this metric to clarify its statistical robustness.

R: According to the honorable reviewer's opinion, Smith's criteria for determining the accuracy of the deep learning method have been added below Figure 7.

The methodology section does not discuss data partitioning ratios for training, validation, and testing. Were standard practices such as 70-15-15 or k-fold cross-validation used?

R: The data partitioning method and its division into training, evaluation, and testing sets have been added at the end of section 4.1.2.

The impact of water content on compressive strength is well-documented, yet the sensitivity analysis suggests unexpected trends. Does this align with known concrete mix behavior?

R: The results of the sensitivity analysis method have been mentioned without the slightest change, and the effect of the water content is in second place after the percentage of cement.

The research effectively applies CNN and MVO, but a discussion of alternative machine learning models (e.g., Random Forest, XGBoost) for predictive strength modeling would add depth.

R: The honorable reviewer's opinion is completely correct, but due to the high volume of the article, which includes the laboratory phase, deep learning method, and sensitivity analysis, this comparison has not been made. However, it will probably be fully conducted in the deep learning phase in another article and will be submitted.

The introduction cites global environmental impacts of cement but does not provide regional data on Iran’s specific concrete consumption and waste management. Would this context strengthen the study's motivation?

R: The information requested by the honorable reviewer has been briefly presented in section 3.1.

"The error index reveals an average error of 0.14"—is this RMSE or MSE? Clarifying the error metric would improve interpretability for readers unfamiliar with the methodology.

R: Thank you to the honorable reviewer. This was a very subtle point, and it has been corrected.

Thank you for the reviewer's kindness.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors improved the manuscript that may be accepted for publication.