Computational Simulation of Microflaw Detection in Carbon-Fiber-Reinforced Polymers

Round 1

Reviewer 1 Report (Previous Reviewer 5)

Dear Authors,

regarding the manuscript you have revised, it is intriguing and acceptable for publication in Electronics. I have no specific comment, except for improving the references, and reducing the self-citation.

Best Regards

Author Response

In the previous paper version, there were seven self-citations. That number was reduced to four, which are relevant papers to this study. Three citations, related to more general issues, were replaced.

Reviewer 2 Report (Previous Reviewer 3)

Authors presented a work on identification of microflaws in composite laminates. The work has been improved compared to the last submission, however, there are still a few concerns before accepting this work for MDPI.

Concerns:

1. As suggested by two reviewers, the statistical data is again not considered. The simulations are also very basic and test cases are limited.

Comments on location, size, and quantity of defects will add value to the manuscript. 

2. The orientation of rectangular defects can be of interest? Also, do authors think rather speaking in terms of width and height, aspect ratio can be of interest?

3. My comments on manufacturing were addressed but the residual void content was not mentioned.

4. Also please overlap Figure 6 a and 6b to clearly observe the differences.

5. Figure 7 name tags should be corrected. Figure 7c and 7d look the same. This can be because the area of defect is the same (?)

Other comments:

1. Matlab is not free!
2. Language should be improved. Several grammatical mistakes.

Author Response

Authors presented a work on identification of micro flaws in composite laminates. The work has been improved compared to the last submission, however, there are still a few concerns before accepting this work for MDPI.

Concerns:

1. As suggested by two reviewers, the statistical data is again not considered.

As answered before, the authors aimed with this work evaluating the defect detectability level in the micron range using ultrasound.

Since the defects studied in the present work are simulated, the ones with the same geometry will always produce equal reflected signals. There is no randomness in the signal behavior, so the statistical analysis is not suitable. Thus, more data including repetitive simulation will not provide additional information for the purpose of this study.

The simulations are also very basic and test cases are limited.

Regarding to the test cases, the authors focused on the most frequent defects that can appear in the studied structures, such as delaminations, inclusions and broken fibers.

The flaws representation was based on https://doi.org/10.1080/21642583.2017.1311240. The test cases were limited because we cannot represent air in the simulated medium, as explained in the manuscript, restricting the representation of some flaws like matrix crack, void, blisters or porosity. Anyway, as explained in the manuscript, defects involving air presence will be more evident with ultrasounds, since the acoustic impedance mismatch will be higher providing higher echo amplitudes.

The authors implemented the simulation procedure to answer the main question: “what is the minimum flaw size detectable in CFRP by ultrasounds?” The implemented procedure demonstrated suitable to answer this question.

Comments on location, size, and quantity of defects will add value to the manuscript. 

Respecting to simulation, the defects were considered at the sample center coincident with beam propagation axis. The size and quantity of defects are described in the text (line 118-129) and illustrated in figure 2.

2. The orientation of rectangular defects can be of interest?

Considering the delamination, as one of the most important defect in composites that is a result of interlaminar debondings, its evolution/formation is essentially parallel to fibers, and then the orientation is not relevant.

Also, do authors think rather speaking in terms of width and height, aspect ratio can be of interest?

From the point of view of detectability, the horizontal dimension (i.e. the dimension perpendicular to the direction of the propagation beam) of some flaws like delamination and broken fibres will be more relevant than the flaw height.

3. My comments on manufacturing were addressed but the residual void content was not mentioned.

The CFRP samples were manufactured with the procedures adopted for aeronautic industry, for which the maximum void content is approximately 1%, as mentioned for example in https://doi.org/10.1177/0021998318772152.

4. Also please overlap Figure 6 a and 6b to clearly observe the differences.

Done.

5. Figure 7 name tags should be corrected. Figure 7c and 7d look the same. This can be because the area of defect is the same (?)

Figure 7 was corrected.

Concerning to Figure 7c and 7d, they are very similar, but it is only a coincidence. In fact, at first sight it would be expected that the signal in 7d presented higher amplitude, however the reflection coefficient at the defect boundary is about one third of the one observed at the inclusion boundary.

Other comments:

1. Matlab is not free!

The reviewer is right. Corrected.

2. Language should be improved. Several grammatical mistakes.

The text was carefully read and some improvements were performed. Also the grammatical mistakes were corrected.

Reviewer 3 Report (Previous Reviewer 2)

This study intends to model and simulate micro-flaws in CFRP laminates, evaluating their detectability level. The work is well organized and meaningful to the practical engineering application. I recommend the publication of the manuscript after the following minor revisions are properly made:

1. Authors should give more information about the modelding process.

2. How did authors obtain the simulation parameters?

Author Response

This study intends to model and simulate micro-flaws in CFRP laminates, evaluating their detectability level. The work is well organized and meaningful to the practical engineering application. I recommend the publication of the manuscript after the following minor revisions are properly made:

1. Authors should give more information about the modelding process.

The k-Wave toolbox provides a relatively simple way of modeling ultrasound propagation, and in the modeling process we just follow the toolbox user’s manual, with the parameters specified in the section 2.1 of the manuscript. In lines 99 and 100 of the manuscript it were included the medium acoustic parameters, necessary for simulation.

The authors consider that the information provided is enough for reproducing the study, however a copy of the script developed for simulation can be included as supplementary material, if it is of relevance.

On the other hand, the computational grid was constructed using conventional Matlab functions for image processing and some additional functions from the k-Wave toolbox (like makeDisc). A copy of this script could also be submitted as supplementary material.

2. How did authors obtain the simulation parameters?

The authors are not sure what this question refers to. The k-Wave has specific functions to introduce each one of the simulation parameters. Related to the computational grid it was answered in the question before. The acoustic parameters of the composite components were obtained from Kinsler et al., and Ono, as mentioned in the manuscript and presented in Table 1. The probe profile and excitation pulse were specified for mimicking our experimental system (it was also mentioned in the manuscript, section 2.1). No additional simulation parameters must be specified. We expect to have answered the reviewer question.

Round 2

Reviewer 2 Report (Previous Reviewer 3)

The authors resubmitted a partially modified manuscript for revision. They did account for a few of the reviewer's suggestions; however, the reviewer still feels that the work is quite rudimentary.  The suggestion for the statistical analysis was requested based on the different microstructures with similar defect placements!!! At least suggest this as your future work in the manuscript.

Author Response

As suggested, we include a final paragraph related to future work.

Author Response File: Author Response.pdf

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Review report on ‘Computational simulation of micro flaw detection in carbon fibre reinforced polymers’. The detailed comments are listed below:

1.      The introduction section is very weak and short. Add more references related to the current work and define the gap clearly.

2.      The quality of figures 4,6, and can be improved.

3.      Simulation details can be provided in more details.

4.      A better figure can be added for the experimental setup.

5.      The experimental setup is missing.

6.      Add the key bullet points in the conclusion section.

7.      Add few recent references:

 https://doi.org/10.1115/1.4050520

 https://doi.org/10.1115/1.4054100   

Reviewer 2 Report

This study presents an evaluation method for the micro flaws in carbon reinforced composite laminate by ultrasound. What’s more, four different types of very small defects were simulated by the Matlab k-Wave toolbox. The paper provides some interesting data but it still needs a minor revision:

1. In the second paragraph of the Introduction, line 41, “because their relative low cost” should be “because its relative low cost”.

2. In Figure 2, the scale between the lower water and the CFRP laminate is not clear.

3. In Figure 5, the test setup corresponding to figure 2 should be shown in detail.

4. In section 2.1, the introduction of modeling method should be more detailed.

5. In section 3, whether the amplitude of test and simulation should be in the same unit?

6. There are many grammatical errors in English sentences written by the authors, as well as spelling and typographical errors.

Reviewer 3 Report

The authors presented a numerical work on the detection of microflaws in CFRP composites. The idea and the work towards the goal were constructed in a good way however, the manuscript does not bring out the details that the authors would like to present. There are several modifications that need to be done for this manuscript to be accepted. In its current state, the manuscript cannot be accepted.

Firstly, the abstract needs to be modified more appropriately to show exactly is the novelty of the work.

The introduction is poorly written. There is no clear line of why is it necessary and what has been previously done and why the authors' approach can be better. This section should significantly be improved.

Line 63: Pseudo-random? But in figure 1, the fibers are arranged in a square pattern. The slight misalignments are not exactly representable to reality. Why haven't the authors adopted one of the well-known algorithms to generate the microstructures?  Also, a Vf=44% can be done with a brute force search algorithm which would be very simple to implement.

Also, was the reference geometry (distance between plies) constructed based on any of the previous microstructure observations? If so, please add the micrographs for comparison.

Figure 3 needs a color bar and proper labeling.

The experimental section is unclear. What was the composite type, how was it manufactured, what was the observation area and what were the defects present in it?

Line 122: filled with water? These defects are not exactly present in reality (at least in the case of non-marine structures). How is this selection justified?

Provide a table with acoustic properties.

The numerical geometry should be justified. It should be close to reality. Pores filled with oil and water in the middle of the composite plate are nonexistent. Also, how does it help the NDT technique? 

It is unclear how this toolbox or approach will be helpful. It also feels like the work is only conducted in 2D and not in 3D. How accurate will it be in 3D? and what would be the use? 

It would be interesting to work with actual images rather than using ideal geometries. 

The discussion is poorly constructed. There are no insights or perspectives on how this work will be used or the target audience. The work has not even considered the values or approaches in the literature to compare or discuss.

Reviewer 4 Report

This paper analyzed the behavior caused by defects through numerical analysis using Matlab as part of the CFRP non-destructive test. The analysis model was verified via experiments, and analysis was performed according to four defect conditions. It seems that this study can be used for CFRP non-destructive testing, however, more detail in analysis and consideration are needed for both experiments and simulations to be published in this journal.
In the case of the experiment, parameters about the experimental conditions and samples is insufficient, and only the result in the most basic condition are examined. So it is difficult to verify the consistency with the corresponding analysis model. Also, since the characteristics and internal structure of the an-isotropic CFRP sample are not confirmed in the test, it is necessary to verify whether the sample itself with more detail and compared with the analysis model in different directions and conditions. The analysis model also needs more explanation of the model and parameters . Due to its 2D conditions, boundary condition is significant variable, and the effect thereof must be considered. The four defect conditions are also too simplified based on 2D, and detailed explanations and validation for the models are necessary.

Reviewer 5 Report

Dear Authors,

I regret to inform you that the manuscript could not be accepted in the current form for publication. The article covers an interesting field such as the detectability of the flaw in composites structures. Nevertheless, the manuscript appears to be too short for an article. The reference list is very small, and the results do not provide the aims of the authors.

For example, after simulating the process, the authors write the following sentence: “Concerning to the detectability and signal to noise ratio, it was observed that both simulated and experimental results agreed well. This conclusion confirmed the simulator was able to mimic the experimental approach for defect detection and evaluate its detectability degree”.

This is not strictly correct. The goodness of the simulation model was tested only on the pristine composites (any statistical measurements were not cited), but not on the real damaged composites.

For the reasons mentioned above, in the reviewer’s opinion, the manuscript should not be accepted in this form but has a chance to be improved and extended with more data and analysis, even in a statistical way. The contribution of the authors remains of valid interest for further investigations.

Kind Regards.