Numerical Test and Strength Prediction of Concrete Failure Process Based on RVM Algorithm
Round 1
Reviewer 1 Report
Dear Authors
Manuscript ID: Buildings-1973383
The manuscript structure presented by the authors is clear to the reader but some improvements are required (e.g., rewriting introduction, Experimental Program and Results and Discussion). Although the authors have tried to present a scientific explanation for the results, in some cases it is required a better and deeper interpretation. The final decision is to Reject the manuscript in the present form (but encourage the authors to resubmit an improved version of the manuscript).
Comments for author File: 
Comments.pdf
Author Response
Comment 1: This abstract was very long, and it wasn’t the results of manuscript.
Response: Thanks to the valuable comments of the reviewers, we re-summarized the abstract. And marked with red color.
Comment 2: There is no literature review in the introduction section. Authors must add a review to the relevant articles and note their possible deficiencies. Authors must note the importance of the research. Another important point that must be consider is to provide a review on the other works on this specific type of Artificial Neural network and RVM algorithm.
And also, the authors must be added some sentences for Novelty in this part.
Response: Thanks to the reviewer 's valuable comments, we have added about this part. And marked with red color.
Comment 3: I suggested the authors added some datum such as;
- Physical properties and XRF of cement
- Physical and mechanical of aggregates.
- How can you mix design?
I suggested the authors added a flowchart for clarified them how did they do?
I suggest that a comparison be made between the results of the rupture mode or failure shapes between laboratory samples and Abaqus models. Why is there no change in the amount of compressive strength when the thickness of the brick is increased?
Response: Thank you for the valuable comments of the reviewers, for the physical properties of cement we have discussed that there is not much need to add to the text. In the aspect of mixed design, we mainly consider the important influencing factors such as water cement ratio and substitution rate for experimental design, and also mention the problems in the manuscript. It is a good point to compare the fracture mode or failure shape between the supplementary laboratory sample and the Abaqus model. We will focus on some of the contents in the next article.
Comment 4: Before using the RVM method, statistical analysis of the datum and presentation of Simple regression (SR) between parameters should be done in this section. According to the use of the RVM method, the schematic of the model should be presented.
Response: Thanks to the reviewers ' valuable opinions, we have made a statistical overall description of each index, and we also describe the principle part of the RVM model.
Comment 5: References must be revised. In many cases volume, Issue, and DOI are absent.
Response: Thanks to the reviewers for their valuable comments, we have corrected the problems in the references.
Author Response File: 
Author Response.docx
Reviewer 2 Report
This study is based numerical test and strength prediction of concrete failure process based on RVM algorithm. The authors are expected to address the below given suggestions and make references to the various ideas brought about to improve the quality and readability of this manuscript before it can be considered for revaluation.
1. The authors should highlight clearly the novelty of this study.
2. What is the advantage of RVM algorithm considering the aim of the study.
3. For better clarity of the observed results add more statistical parameters i.e., Mean absolute error, F- value, etc. Besides, state what are the acceptable values or conditions to these statistical parameters? Basically, to answer the question 'how good is good' or 'how bad is bad'?
4. Figure 6: What do you mean by sample index is non-normal distribution. Add skewness analysis!
5. The literature review is quite low, the author should cite some more paper.
a. https://doi.org/10.1016/j.cemconcomp.2022.104414;
b. https://doi.org/10.1016/j.enbuild.2022.112280
6. Add more elucidated details pertaining to Fig. 7 for clear understanding. The provided detail is not sufficient.
7. Fig. 8: Keep the axis consistent with proper values and draw a clear linear line and double check the R2 value.
8. The authors have used different input parameters. Could author explain what is the sensitivity of these parameters towards the final outcome of the result. This point is critically important to understand the role of individual property.
Author Response
Comment 1: The authors should highlight clearly the novelty of this study.
Response: Thank the reviewers for their questions. We have added the description of this part in the introduction.
Comment 2: What is the advantage of RVM algorithm considering the aim of the study.
Response: Thank the reviewers for their questions. We have added the description of this part in the introduction. And in the model, we have mentioned the advantages of RVM algorithm.
Comment 3: For better clarity of the observed results add more statistical parameters i.e., Mean absolute error, F- value, etc. Besides, state what are the acceptable values or conditions to these statistical parameters? Basically, to answer the question 'how good is good' or 'how bad is bad'?
Response: Thank you for the reviewer 's questions, we added a description of this part of the article.
Comment 4: Figure 6: What do you mean by sample index is non-normal distribution. Add skewness analysis!
Response: Thank you for the reviewer 's questions, we added a description of this part of the article. And marked with red color.
Comment 5: The literature review is quite low, the author should cite some more paper.
- https://doi.org/10.1016/j.cemconcomp.2022.104414;
- https://doi.org/10.1016/j.enbuild.2022.112280
Response: Thank you for the reviewer 's questions, we have made appropriate references to two articles in.
Comment 6 Add more elucidated details pertaining to Fig. 7 for clear understanding. The provided detail is not sufficient.
Response: Thank you for the reviewer 's questions, we added a description of this part of the article.
Comment 7 Fig. 8: Keep the axis consistent with proper values and draw a clear linear line and double check the R2 value.
Response: Thank you for the reviewer 's questions, we added a description of this part of the article. The error representation of R2 value is corrected.
Comment 8 The authors have used different input parameters. Could author explain what is the sensitivity of these parameters towards the final outcome of the result. This point is critically important to understand the role of individual property.
Response: Thank you for the reviewer 's questions, the sensitivity between the parameters of each index, the author mainly from the Figure 7 correlation description to infer its contribution.
Author Response File: Author Response.docx
Reviewer 3 Report
In this manuscript the authors investigate strength prediction of the concrete failure based on a RVM algorithm considering as input a database obtained by simulation using a FEM model developed with the commercial software ABAQUS. The developed FEM model is a cube test specimen in compression.
C1) The gap research is not comprised in the Introduction text.
C2) Why did the authors decide to apply the RVM algorithm model using a simulated database? Supposing that, the mathematical model, i.e. FEM model, is feasible and quite realistic, the outcome, i.e. the numerical results for the compressive strength from the FEM model, have a certain degree of accuracy. Therefore, the input, i.e., the database, has a degree of prediction of the compressive strength with a certain margin of approximation.
C3) In Introduction, the made statements about concrete and the performed studies on it are trivial. Besides, the English vocabulary describing concrete and its characteristics is neither appropriate nor the common vocabulary used for this material. Therefore, the authors should read some basic common books about concrete and its properties.
C4) In the title, it is not mentioned correctly the material under investigation. The material under investigation seems to be concrete containing recycled aggregate as partial replacement of natural coarse aggregate. This type of concrete is called recycled aggregate concrete (RAC) not "recycled concrete" as the authors it mentioned many times in the text.
C5) The authors should read about RAC. The literature is very reach because the RAC have been studied heavily since 2000 onwards. In this respect, comparing ordinary concrete with RAC, for the same mix proportions, I am highly questionating whether the compressive strength of RAC is higher. If it is so, then the authors should have been referenced this important assertion from Introduction with published papers in high ranked journals devoted to the construction materials.
C6) Remarks on the experimentation part.
C6.1) As general remark, the experimentation part of the study is unsatisfactory.
C6.2) The text from line 123 to 129 is very confusing. It seems coming from another review.
C6.3) Table 1 is completely unsatisfactory in respect with the physical and chemical properties of Portland cement (PC). Besides, this kind of table is not justified for such study. The simple fact that the quite well know Portland cement PO42.5 is used in the concrete mixes is enough.
C6.4) Very likely the authors should have been introduced a large informative table regarding the composition parameters of the cement mixes. Information about the recycled aggregate, source, type of crushing, percentage of hardened cement paste bonded to the natural aggregate particle, absorbtion, density, etc. This kind of data about the used recycled aggregate should have been mentioned in order to increase credibility of the study.
C6.5) The number of casted and tested test specimens is completely unsatisfactory in order to get accurate data about strength and subsequently to be used for the calibration of the FEM model. I think that a much better idea would have been the taking over a very well done experimental part about RAC from a well cited paper published in a well ranked journal and developing the RVM model based on it.
C6.6) Many phrases about the behavior of concrete in compression are not accurate. It seems to be created by individuals without a real experience in testing concrete. For concrete, the authors should have been known that there is a strain corresponding to the peak stress not peak strength as it is written in the manuscript and an ultimate strain. The first is close to 2%o percents and the second is at least 3.5%o. It should have been underlined about what concrete deformation is taking about. In compression concrete is not suddenly destroyed, without any warning. The brittleness is much higher in tension comparing to compression. Therefore, it is largely recognized that the failure in compression of concrete is progressive, not so brittle, particularly for concrete of low and medium strength like the concrete in this manuscript.
C6.7) Within this study, it cannot be seen the utility of testing concrete at intermediary ages, lower than 28 days. Perhaps the authors thought checking the FEM model for these ages, too. For this study, it is highly questionable because it is largely known that there are cements with low, normal and high rate of hydration.
A big missing of the experimental part is the missing of the stress-strain diagrams in compression for developed concrete. It can be asserted that when a numerical model was developed aiming the behavior of a certain material, concrete, masonry, etc., a paramount information used for calibration of the model is the stress-strain diagram of the tested material. For the authors, in this manuscript the main task within experimental part should have been the getting the stress-strain diagram in compression providing the strength and deformation properties of concrete as follows, tangent and secant modulus, peak stress and the corresponding strain.
C7) The modeling part is quite week. A very simplistic FEM model was developed by the authors using a commercial software. I presume that the authors should have been known until now that the modelling of the ordinary concrete behavior is a very complex issue and it is highly based on Fracture Mechanics approaches developed for cvasi-brittle materials. Therefore, a question is raising about the validity of the author's FEM model for the behavior of concrete. Have been previously published the model in a well ranked journal devoted to Fracture Mechanics or Behavior of Materials using numerical methods ? If yes, then the authors should reference their study or a similar one containing such model in Abaqus that was largely admitted as a good FEM model able modeling the entire behavior of concrete from stress equal to zero up to failure.
In order to underline the previous remark, in the next phrases, I am going to quote basic information from a very well know author Michael D. Kotsovos that have written the book Finite-Element Modelling of Structural Concrete.
"It is generally accepted that the cause of fracture and failure of concrete is the proliferation of flaws or micro-cracks which exist within the body of the material even before the application of load. These flaws are attributable to a number of causes; the main ones are:
a. Discontinuities in the cement paste matrix resulting from its complex morphology (such flaws range in size from a number of angstrom units at the gel-lattice level, to several microns (or above) for isolated or continuous capillary pores).
b. Voids caused by shrinkage or thermal movements due to incompatibility between the properties of the various phases present in concrete.
c. Discontinuities at the boundary between the aggregate particles and the paste or mortar matrix caused by segregation.
d. Voids present in concrete as a result of incomplete compaction.
Regarding with the FEM model, therefore, a good lecture is the mentioned book. It shouldn't forget that the behavior of RAC under loading is slightly different "
There are presented 3 conclusions which are mostly general. They are very less focused on the findings. The text of the conclusions is too long because contains information which have been already presented in the manuscript.
As a conclusion, the authors attempted to comprise in one manuscript three distinct investigations, experimental, FEM modeling and creating a RVM predicting model for the compressive strength. They used in the mentioned order the outcome of one investigation as data base for the next investigation in order to predict compressive strength of recycled aggregate concrete.
This research is highly questionable because the results feeding the model are obtained in a very debatable manner.
Author Response
Comment 1: The gap research is not comprised in the Introduction text
Response: Thank you for the reviewer 's questions, we added a description of this part of the article. The shortcomings of existing research are supplemented.
Comment 2: Why did the authors decide to apply the RVM algorithm model using a simulated database? Supposing that, the mathematical model, i.e. FEM model, is feasible and quite realistic, the outcome, i.e. the numerical results for the compressive strength from the FEM model, have a certain degree of accuracy. Therefore, the input, i.e., the database, has a degree of prediction of the compressive strength with a certain margin of approximation.
Response: Thank you for the reviewer 's questions, we added a description of this part of the article. Why did the author decide to use the simulated database to apply the RVM algorithm model? The main reason for this problem is also mentioned in the introduction, mainly to solve the problem of long data acquisition cycle.
Comment 3: In Introduction, the made statements about concrete and the performed studies on it are trivial. Besides, the English vocabulary describing concrete and its characteristics is neither appropriate nor the common vocabulary used for this material. Therefore, the authors should read some basic common books about concrete and its properties.
Response: Thanks to the reviewer 's question, we have partially replaced the potential non-professional vocabulary.
Comment 4: In the title, it is not mentioned correctly the material under investigation. The material under investigation seems to be concrete containing recycled aggregate as partial replacement of natural coarse aggregate. This type of concrete is called recycled aggregate concrete (RAC) not "recycled concrete" as the authors it mentioned many times in the text.
Response: Thanks to the reviewer for pointing out the problems in the article to be revised.
Comment 5: The authors should read about RAC. The literature is very reach because the RAC have been studied heavily since 2000 onwards. In this respect, comparing ordinary concrete with RAC, for the same mix proportions, I am highly questionating whether the compressive strength of RAC is higher. If it is so, then the authors should have been referenced this important assertion from Introduction with published papers in high ranked journals devoted to the construction materials.
Response: Thanks to the reviewer pointed out that when the water cement ratio is the same, the greater the strength of RAC the higher the strength of RAC.
Comment 6 Remarks on the experimentation part.
C6.1) As general remark, the experimentation part of the study is unsatisfactory.
C6.2) The text from line 123 to 129 is very confusing. It seems coming from another review.
C6.3) Table 1 is completely unsatisfactory in respect with the physical and chemical properties of Portland cement (PC). Besides, this kind of table is not justified for such study. The simple fact that the quite well know Portland cement PO42.5 is used in the concrete mixes is enough.
C6.4) Very likely the authors should have been introduced a large informative table regarding the composition parameters of the cement mixes. Information about the recycled aggregate, source, type of crushing, percentage of hardened cement paste bonded to the natural aggregate particle, absorbtion, density, etc. This kind of data about the used recycled aggregate should have been mentioned in order to increase credibility of the study.
C6.5) The number of casted and tested test specimens is completely unsatisfactory in order to get accurate data about strength and subsequently to be used for the calibration of the FEM model. I think that a much better idea would have been the taking over a very well-done experimental part about RAC from a well cited paper published in a well ranked journal and developing the RVM model based on it.
C6.6) Many phrases about the behavior of concrete in compression are not accurate. It seems to be created by individuals without a real experience in testing concrete. For concrete, the authors should have been known that there is a strain corresponding to the peak stress not peak strength as it is written in the manuscript and an ultimate strain. The first is close to 2%o percents and the second is at least 3.5%o. It should have been underlined about what concrete deformation is taking about. In compression concrete is not suddenly destroyed, without any warning. The brittleness is much higher in tension comparing to compression. Therefore, it is largely recognized that the failure in compression of concrete is progressive, not so brittle, particularly for concrete of low and medium strength like the concrete in this manuscript.
C6.7) Within this study, it cannot be seen the utility of testing concrete at intermediary ages, lower than 28 days. Perhaps the authors thought checking the FEM model for these ages, too. For this study, it is highly questionable because it is largely known that there are cements with low, normal and high rate of hydration.
A big missing of the experimental part is the missing of the stress-strain diagrams in compression for developed concrete. It can be asserted that when a numerical model was developed aiming the behavior of a certain material, concrete, masonry, etc., a paramount information used for calibration of the model is the stress-strain diagram of the tested material. For the authors, in this manuscript the main task within experimental part should have been the getting the stress-strain diagram in compression providing the strength and deformation properties of concrete as follows, tangent and secant modulus, peak stress and the corresponding strain.
Response: Thanks to the reviewers ' questions, we will further supplement and enrich the experimental part in the next high-level article. We have re-summarized the content of lines 123 to 129, which is a reference to some other articles. We finally retained Table 1. And in the article added some about RAC related content. In this study, we did not see the utility of testing concrete at the intermediate age (less than 28 days). An important reason for this is that based on existing research, if its seven-day strength is predicted to be too low, RAC is still more inclined to compare 28-day strength with ordinary concrete strength. The experimental part of this article does have a stress-strain diagram. The main reason is that our research focuses on a research method, such as obtaining a database through simulation methods and combining machine learning methods to achieve the prediction of RAC strength.
Comment 7 The modeling part is quite week. A very simplistic FEM model was developed by the authors using a commercial software. I presume that the authors should have been known until now that the modelling of the ordinary concrete behavior is a very complex issue and it is highly based on Fracture Mechanics approaches developed for cvasi-brittle materials. Therefore, a question is raising about the validity of the author's FEM model for the behavior of concrete. Have been previously published the model in a well ranked journal devoted to Fracture Mechanics or Behavior of Materials using numerical methods? If yes, then the authors should reference their study or a similar one containing such model in Abaqus that was largely admitted as a good FEM model able modeling the entire behavior of concrete from stress equal to zero up to failure.
In order to underline the previous remark, in the next phrases, I am going to quote basic information from a very well know author Michael D. Kotsovos that have written the book Finite-Element Modelling of Structural Concrete.
"It is generally accepted that the cause of fracture and failure of concrete is the proliferation of flaws or micro-cracks which exist within the body of the material even before the application of load. These flaws are attributable to a number of causes; the main ones are:
- Discontinuities in the cement paste matrix resulting from its complex morphology (such flaws range in size from a number of angstrom units at the gel-lattice level, to several microns (or above) for isolated or continuous capillary pores).
- Voids caused by shrinkage or thermal movements due to incompatibility between the properties of the various phases present in concrete.
- Discontinuities at the boundary between the aggregate particles and the paste or mortar matrix caused by segregation.
- Voids present in concrete as a result of incomplete compaction.
Regarding with the FEM model, therefore, a good lecture is the mentioned book. It shouldn't forget that the behavior of RAC under loading is slightly different "
There are presented 3 conclusions which are mostly general. They are very less focused on the findings. The text of the conclusions is too long because contains information which have been already presented in the manuscript.
As a conclusion, the authors attempted to comprise in one manuscript three distinct investigations, experimental, FEM modeling and creating a RVM predicting model for the compressive strength. They used in the mentioned order the outcome of one investigation as data base for the next investigation in order to predict compressive strength of recycled aggregate concrete.
This research is highly questionable because the results feeding the model are obtained in a very debatable manner.
Response: Thanks to the reviewer 's questions, we added the content of the book mentioned by the reviewer. For Abaqus software, it is a more appropriate software for simulating concrete. The constitutive model of concrete included in ABAQUS software is mainly Brittle Cracking Model, Smeared Cracking Model and Plasticity Damage Model. And the revised part is marked with red color.
Author Response File: Author Response.docx
Reviewer 4 Report
The topic is very important and the results are promising. However, there is one major issue. The comparisons are restricted to only one type of recycled aggregate from one source.
The authors are suggested to collect experimental results from at least 2 more previously published papers to show the comparison between the predicted and experimental results.
Presently, there are too few data points to make cogent conclusions.
Another issue is the use of technical English language in the manuscript. The manuscript can be improved by using more quantitative and technical results in the introduction.
Author Response
Comment 1: The topic is very important and the results are promising. However, there is one major issue. The comparisons are restricted to only one type of recycled aggregate from one source.
The authors are suggested to collect experimental results from at least 2 more previously published papers to show the comparison between the predicted and experimental results.
Presently, there are too few data points to make cogent conclusions.
Another issue is the use of technical English language in the manuscript. The manuscript can be improved by using more quantitative and technical results in the introduction
Response: Thanks to the reviewers pointed out that the problem of small amount of test is the problem of the article. The purpose of this manuscript is to propose a new database construction and RAC uniaxial compressive strength test method. In the next study, we use the wonderful ideas proposed by the reviewers.
Author Response File: Author Response.docx
Reviewer 5 Report
This paper gives an investigation on the numerical test and strength prediction of concrete failure process based on RVM algorithm. Generally, this topic is interesting, and thus, this paper can be considered after a major revision. The following comment should be addressed in the revision version. (1) More than three references should be not cited in one sentence, such as “[8-14]” and “[25-31]”. The introduction section should be rewritten. (2) More references in recent 3 year should be added, the following two references may be helpful for this paper. (a) Characterization of sustainable mortar containing high-quality recycled manufactured sand crushed from recycled coarse aggregate (b) Hysteresis and damping properties of steel and polypropylene fiber reinforced recycled aggregate concrete under uniaxial low-cycle loadings (3) The author should clear the objective and innovation of this work, which is very important. (4) There is no meaning for Figure. 1, which should be deleted from this paper. (5) More numerical parameters should be supplied in the revised version, and the error bar should be added in test value group in Fig. 4. (6) The shortage of this work should be further highlighted at the end of this paper.Author Response
Comment 1: More than three references should be not cited in one sentence, such as “[8-14]” and “[25-31]”. The introduction section should be rewritten.
Response: Thank the reviewer for pointing out the problem for the citation. These descriptive words are indeed based on the summary of these articles, so I suggest that this part of the quote should be retained, but we have re-refined the introduction.
Comment 2: More references in recent 3 year should be added, the following two references may be helpful for this paper. (a) Characterization of sustainable mortar containing high-quality recycled manufactured sand crushed from recycled coarse aggregate (b) Hysteresis and damping properties of steel and polypropylene fiber reinforced recycled aggregate concrete under uniaxial low-cycle loadings
Response: Thanks to the reviewers ' valuable opinions, we added descriptive text and references about this part.
Comment 3: The author should clear the objective and innovation of this work, which is very important
Response: Thanks to the reviewer's valuable suggestions, the article has added descriptive discourse on this part of the content.
Comment 4: There is no meaning for Figure. 1, which should be deleted from this paper.
Response: Thanks to the reviewer's valuable comments, Figure 1 is still recommended to be preserved after consideration.
Comment 5: More numerical parameters should be supplied in the revised version, and the error bar should be added in test value group in Fig. 4.
Response: Thanks to the reviewers for their valuable comments, we have added a parametric representation of the error curve in Figure 4.
Comment 6 The shortage of this work should be further highlighted at the end of this paper.
Response: Thank you to the reviewers for their valuable comments, and we add at the end the shortcomings regarding the current work.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
Dear Authors
Manuscript ID: Buildings-1973383
Title: “Numerical Test and Strength Prediction of Concrete Failure Process Based on RVM Algorithm”
The final decision is to accept the manuscript in this present form.
Reviewer 2 Report
The authors have addressed my comments. The paper is acceptable in its current form.
Reviewer 4 Report
The authors have responded to the comments from the reviewer.
Reviewer 5 Report
This paper can be published at the current form.
