Data-Driven Construction Method of Material Mechanical Behavior Model

Round 1

Reviewer 1 Report

Dear Authors

Mechanical properties of materials are measured by certain factors, e.g. Young's modulus, tensile strength, Poisson's ratio, etc. They characterize the behavior of materials under the influence of stresses caused by external forces. The cause is stress - the effect is deformation. The relationship between stress and strain is Young's modulus E. According to Hooke's law, stress is proportional to strain and E modulus.

In the presented article, many terms have been mixed up, dependent variables have been mixed up with independent variables. The undertaken method of research is not appropriate. Moreover, Figures 1 and 3 are illegible. The performed experiments are not described properly.

The article is not suitable for publication in this form.

Best Regards

reviewer

Author Response

Thank you for your comments concerning our manuscript entitled “Data-driven Construction Method of Material Mechanical Behavior Model” (metals-1694701). Those comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our studies. We have studied comments carefully and have made correction which we hope meet with approval. Revised portion are marked in the paper. The major corrections in the paper and responses to your comments are in the attachment.

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Authors have proposed a data driven construction method of material mechanical behavior model to obtain the mechanical behavior response of the material under loading. This method uses Python script to drive Abaqus loop calculation to obtain data sets; and performs artificial intelligence training on data sets to realize model construction.  Taking the quasi-static tension of 9310 steel as an example; a material mechanical behavior model is constructed; and the accuracy of the prediction model is verified based on the experimental data. This manuscript seems interesting and might be considered after incorporation of major corrections Please find comments as below:

• Abstract should include the research gap/problem statement and motivation behind this research study. Best findings/novelty/originality should be highlighted. Future scope/recommendations would be added value.
• There is lack of critical discussion in literature review section. Advantages/disadvantages of this research study compared to previous studies should be clarified. I would like to recommend the authors some recent important articles: https://doi.org/10.3390/ma14112895; https://doi.org/10.3390/ma14185261
• Novelty/originality should be further improved. 
• Please improve the resolution/quality of figures. 
• Results should be supported/justified scientifically with valid data. 
• Conclusion should be restructured in view of considering the best finings of the study.
• Comprehensive proof read is essential throughout the manuscript as there are a lot if typo/grammatical errors. 

Author Response

Thank you for your comments concerning our manuscript entitled “Data-driven Construction Method of Material Mechanical Behavior Model” (metals-1694701). Those comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our studies. We have studied comments carefully and have made correction which we hope meet with approval. Revised portion are marked in the paper. The major corrections in the paper and responses to your comments are in the attachment.

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

This paper presents an interesting study on using machine learning (ML) methods for modeling material response. The developed method is shown to be close compared to experimental data. Although the study is worth pursuing and this paper is interesting, there are some major improvements that can be made to paper, and there are as follows:

(1) In the finite-element simulations, what type of kinematics are used for modeling? Small strain or finite strain? Does this issue matter ? It is an important issue when modeling large deformation, and for this case, how will the machine learning technique take into account such issues?

(2) A very important application of the ML model is to determine when fracture will occur and the modeling and simulation of fracture. To generate data to train the ML-based algorithm for fracture prediction, the simulations can be done within the Abaqus framework using a robust implementation of the element failure or deletion method, the simplest method for modeling fracture (see the papers of Reddy and co-workers [ Comput Methods Appl Mech Engrg (2019) vol.354 pp. 871-903; Mech Adv Mater Struct (2020) vol. 27 pp. 1085-1097 ]. Even if the modeling and prediction of fracture is not attempted in this work, the modeling of fracture is an important avenue for future work and also designers/engineerns, and this can be done within the Abaqus environment using the framework developed by the papers of Reddy and co-workers described above.

(3) Please show the ANN diagram along with the different layers, as this will help the visualization of the ML procedure.

(4) Can the ML framework developed by the authors be used to predict an independent set of experiments such as simple shear (tubular torsion) or combined tension-torsion? It is an interesting avenue for work, if not, future work. 

(5) In Page 4, the authors also reported the density of the material. Is this being used in their analysis? Does this matter for ML method, or are the authors using an Abaqus/Explicit simulation where the density must be specified? 

Author Response

Thank you for your comments concerning our manuscript entitled “Data-driven Construction Method of Material Mechanical Behavior Model” (metals-1694701). Those comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our studies. We have studied comments carefully and have made correction which we hope meet with approval. Revised portion are marked in the paper. The major corrections in the paper and responses to your comments are in the attachment.

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Dear Authors

I think that after the corrections and explanations have been made, the article can be published.

Best Regards

Reviewer

Author Response

Thank you for your comments concerning our manuscript entitled “Data-driven Construction Method of Material Mechanical Behavior Model” (metals-1694701). Those comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our studies. Thanks again for your comment.

Reviewer 2 Report

Authors have addressed comments properly. 

Author Response

Thank you for your comments concerning our manuscript entitled “Data-driven Construction Method of Material Mechanical Behavior Model” (metals-1694701). Those comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our studies. Thanks again for your comment.

Reviewer 3 Report

The authors have attempted to answer the queries raised by the Reviewer, but they have not fully addressed all the queries in a satisfactory manner.

Comment 1: The authors have commented that the Reviewer's suggestion is worth noting and a subject for future work. However, this can be addressed immediately by changing the options in the Abaqus analysis (as this point was not mentioned to be a future work). The query is IF the kinematics matter! So this point needs to be addressed in this work.

Comment 2: Since the ultimate goal is to determine how and when fracture occurs in any component, it would not be too difficult to run experiments in tension (which the authors have done) to fracture and collect data points for training the ML algorithm for fracture prediction. This can be simply using the element deletion option in Abaqus as well (it is very easy to do, and this will tremendously increase the impact of the work although the current implementation of the element deletion method in Abaqus is not as robust as the implementation mentioned in the papers cited in my previous review).

Comment 3: This point has been satisfactorily addressed by the authors. Thank you.

Comment 4: This can be the subject of future work.

Comment 5:  Thank you for clarifying that density is not used in the calculations since it is a static solution.

Author Response

Thank you for your comments concerning our manuscript entitled “Data-driven Construction Method of Material Mechanical Behavior Model” (metals-1694701). Those comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our studies. We have carefully studied the comments and provided explanations in the hope of approval. The response to your comments are as following.

Please see the attachment.

Author Response File: Author Response.pdf

Round 3

Reviewer 3 Report

Comment 1: The authors have mistaken the issue of KINEMATICS vs kinetics! My question is if the authors could show that the usage of finite-deformation KINEMATICS shows the same results as small-strain KINEMATICS. Not if the kinetics of the issue matter since the deformation rate is known to be low!

Moreover, my query did not involve asking the authors to perform simulations using Abaqus/Explicit. Following this, I am also not certain the goal of the authors in showing Figures 1 and 2 in their reply. What are they comparing these contour plots with ? 

Author Response

Thank you for your comments concerning our manuscript entitled “Data-driven Construction Method of Material Mechanical Behavior Model” (metals-1694701). We have carefully studied the comments and provided explanations in the hope of approval. Revised portion are marked in the paper. 

Please see the attachment.

Author Response File: Author Response.pdf