Review Reports - Short-Glass-Fiber Aspect Ratios in Polyamide-6 Composites: Homogenization and Deep Learning-Based Scanning Image-Microscope Segmentation Comparison

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

1. In paragraph 2.1, you should add a description of the materials, indicate their manufacturer and country of origin.

2. Authors It is worth making a separate paragraph in section 2, dedicated to the method of obtaining the samples. In particular, it is worth indicating at what temperature and pressure the samples were obtained instead of Figure 2, which is not informative at all.

3. Paragraph 2.3 is entitled Microscopy, but part of the paragraph is devoted to the methodology of fiber production. Although the process of microscopy is not sufficiently covered. In particular, it is not stated at what magnification the research was conducted. The fiber production process should be described in a separate paragraph.

4. In Figure 5 and 8, it is worth indicating the units of measurement of the indicators along the axes.

5. Table 1 should change the subheadings of the table. Fiber fractions are not given, while the names of the materials are indicated. Of course, from the name of the material you can understand the fiber content in it, but there is a discrepancy in the subheading of the table of its contents.

6. Table 2 is completely uninformative. The name of table 2 is incorrect. It is worth clarifying what statistics are shown. It is not clear why such a number of images were obtained for different samples.

7. In Figure 12, you need to specify the image magnification.

8. In Table 3, it is necessary to specify the units of measurement of the diameter of the fibers.

9. In paragraph 4, as well as in the Conclusions, it would be good to add suggestions on the practical application of the obtained results.

Author Response

1. In paragraph 2.1, you should add a description of the materials, indicate their manufacturer and country of origin.

Thank you for your recommendation, have added a description to section 2.1:

« In this study, composites reinforced with glass fibers with mass fractions of 15% (Armamid PA6 GF 15-1THSM [46]), 30% (Armamid PA6 GF 30-1T [47], Armamid PA6 GF 30-1THSM [48]) and 50% (Armamid PA6 GF 50-1 [49]) are considered. Materials are produced in Russia by R&P Polyplastic under the Armamid trademark.»

We’ve update injection molding process description, including temperature and pressure characteristics:

“The material was dried before the molding process was conducted at 85°C for 6 h. The injection molding of 200×150×4 mm plates was done at NegriBossi VE210-1700 machine with the following parameters: flow rate is 42 cm³/s, velocity to pressure switch over at 400 MPa, melt temperature is 250 °C, mold temperature is 100°C, exposure under pressure of 200 bar for 5 s”

Thank you for your recommendation, the pictures are taken at different magnifications, we have added microscope descripition and the microscopy parameters in the manuscript text:

“Scanning electron microscopy on the scanning electron microscope TESCAN with software VEGA with up to 10000x magnification”

«The photos of the fibers were taken at different magnifications from 68 to 215 times (predominantly from 100 to 200 times) for measuring the fiber length and from 442 to 1000 times (predominantly from 600 to 700 times) for measuring the fiber diameter; at a working distance of up to 8 to 15 mm; and at high voltage in 30 kV.»

4. In Figure 5 and 8, it is worth indicating the units of measurement of the indicators along the axes.

Thank you for your recommendation, we have added the missing units.

Thank you for your recommendation, we have corrected the table subheading to “Materials”

Thank you for your recommendation, we have add the explanation, changed the name of the table and expanded it.

7. In Figure 12, you need to specify the image magnification.

Thank you for your recommendation, the pictures are taken at different magnifications, we have added the ranges in the manuscript text. «The photos of the fibers were taken at different magnifications from 68 to 215 times (predominantly from 100 to 200 times) for measuring the fiber length and from 442 to 1000 times (predominantly from 600 to 700 times) for measuring the fiber diameter; at a working distance of up to 8 to 15 mm; and at high voltage in 30 kV.»

8. In Table 3, it is necessary to specify the units of measurement of the diameter of the fibers.

Thank you for your recommendation, we have added the missing units.

9. In paragraph 4, as well as in the Conclusions, it would be good to add suggestions on the practical application of the obtained results.

We add in Inroduction, Paragrap 4 word about application:

“The effective fiber length evaluation obtained in the work will allow, on the basis of a scanning electron microscope fiber length measurements, to make refinements to the parameters of mean-field homogenization models used both in calculations [43] and in the design [44] of structures made of composite materials.”

“There are many options for using neural networks in solving structural design problems – from acceleration of computations to generative design and post-processing [67]. In the current work, a neural network pipeline is used to conduct an experimental evaluation of the fiber length, refining the parameters of the mean field homogenization model of the material. By obtaining fiber length measurements and considering the presented effective fiber length correction, researchers can provide more accurate estimates of fiber aspect ratio, which makes a significant contribution [45] to the mean-field model's predictions of composite elasticity.”

As well it is mentioned in Conclusion:

“The obtained estimates can serve as a useful reference when calibrating other models of short-fiber-reinforced polymer materials.”

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have focused on presenting a methodology for determining fiber size distribution in injection molded composites. My questions and observations are as follows:

1) Please find attached a marked pdf file of the manuscript on which I have indicated a few corrections.

2) There is a mention in the manuscript of possible changes to the fibers due to the stresses induced by the injection molding process. Specifically documenting this would have established a sound and necessary baseline for the subsequent studies of the fibers (post molding). I would therefore consider it essential to include size and aspect ratio data of the fibers added to the injection molding machine.

3) The tensile data clearly alludes to fiber orientation effects which are to be expected, especially in the molding of a thin plate. It would be helpful to include through-thickness sectional images to clearly present the flow induced alignment, and also see any boundary layer effects and they may directly relate to fiber length. I see this as an essential part of the study.

Comments for author File: Comments.pdf

Author Response

1) Please find attached a marked pdf file of the manuscript on which I have indicated a few corrections.

Thank you for your detailed recommendations. We have tried to take them into account, including working on the figures and specifying the experimental parameters such as tensile rate.

Thank you for the recommendation to conduct this additional study. We have completely repeated the process for materials pellets and have shown the shift in fiber length histograms during injection molding (Figure 14).

Thank you for your suggestion to include this issue in the work. We carried out an electron microscope study of polished sections (figure 11), and show that the fibers are mainly oriented along the molding direction.

Reviewer 3 Report

Comments and Suggestions for Authors

This paper investigates two separate ideas: 1) Using Deep-learning algorithm to characterise short fibres within the matrix, and 2) Using mean field homogenization theory to calculate the corresponding composite performance. This manuscript has great potential, but currently still requires some major improvements before it can be accepted for publication. Please find my comments below:

1) The introduction/literature review section is currently a bit weak. There are a few statements provided but with insufficient literature citations to support the arguments. I recommend this reference (https://doi.org/10.1016/j.jmrt.2024.06.089), which can be used in many places of the introduction, such as line 30, 42, 49, 55.

2) Can the authors provide more information about the glass fibre in section 2.1?

3) Section 2.5, please label Figure 5, what each line means. Also, why was the result of 50% GF-reinforced sample shown, when the section discusses about fibre-free sample?

4) Line 287, how did the model recognise the photo scale? How can the model differentiate between the scale versus the actual length of the fibres? Please provide an explanation.

5) MAJOR: I am not sure if this issue arises from poor writing or intended, but I did not see a clear link between the "Deep-learning fibre characterisation" and "model-based material properties prediction". There should be a strong link between these two themes, but currently I see that they are investigated separately. The results obtained from SEM should be used to train the Deep Learning model, which can then be used to predict the corresponding composite properties after. This revision must be performed to ensure the manuscript content and quality.

Author Response

We have expanded the literature review, including taking into account your recommendation. The work now includes 67 references.

2) Can the authors provide more information about the glass fibre in section 2.1?

We add in Section 2.1. more detailed description of used materials:

“In this study, composites reinforced with glass fibers with mass fractions of 15% (Armamid PA6 GF 15-1THSM [46]), 30% (Armamid PA6 GF 30-1T [47], Armamid PA6 GF 30-1THSM [48]) and 50% (Armamid PA6 GF 50-1 [49]) are considered. Materials are produced in Russia by R&P Polyplastic under the Armamid trademark.”

As well we added glass fiber parameters in Section 3.3.

“The obtained models most accurately describe the conducted strength experiment and reflect the properties of the composite materials. Glass fibers Young modulus is in the range of 60 GPa to 90 GPa [63]. In work [64], the elastic modulus range of the most popular E-glass fibers was estimated in the range of 72…76 GPa. In our work, we use Young's modulus = 72 GPa for glass fibers, and Poisson's coefficient of glass fibers is 0.22, which corresponds to the values for short-reinforced polyamide fibers in the Digimat MX system.”

3) Section 2.5, please label Figure 5, what each line means. Also, why was the result of 50% GF-reinforced sample shown, when the section discusses about fibre-free sample?

Thanks for the recomendation. We found a typo and corrected it:

“To calibrate the mean field homogenization composite material model parameters, tensile curve fitting was performed on the tensile test results.”

4) Line 287, how did the model recognise the photo scale? How can the model differentiate between the scale versus the actual length of the fibres? Please provide an explanation.

We’ve added the explanation with reference to Tesseract OCR Engine which we have used to solve this point:

“When calculating the fiber length and diameter, the pipeline recognizes the photo scale (using the Tesseract OCR Engine [66]) and calculates the necessary characteristics”.

We’ve add in the Discussion the more detailed description of place of deep-learning models in this study:

Reviewer 4 Report

Comments and Suggestions for Authors

The overall experimental design is appropriate and the results support the conclusions.

Comments on the Quality of English Language

The authors do need to recheck the English in the manuscript. For example, the "polyamide" is misspelled in the title and a few other places as "poliamide".

Author Response

The authors do need to recheck the English in the manuscript. For example, the "polyamide" is misspelled in the title and a few other places as "poliamide".

Thank you for your kind feedback. We have proofread the English language, including “poliamide” issue.

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

The authors have revised the manuscript well. The content alongside their discussions are now of good quality. I can now recommend paper acceptance.