A Novel Specimen Produced by Additive Manufacturing for Pure Plane Strain Fatigue Crack Growth Studies

Round 1

Reviewer 1 Report

The authors proposed a cylindrical specimen with a central crack produced by additive manufacturing.  I agree the advantage of the specimen developed in this study. The manuscript was correctly modified, and I think this version is worth publishing.

Author Response

Dear Reviewer

Thank you for your positive appreciation.

Best regards

fernando antunes

Reviewer 2 Report

The subject paper has developed a very novel concept and useful test specimen for determining fatigue crack growth under vacuum (?) and under pure plane-strain conditions.  The authors think that the environment around the internal crack in the specimen is in a vacuum.  This may be the case, but the effects of vacuum on fatigue crack growth in the literature has shown much larger differences between pure vacuum and lab air (like an order of magnitude and not a factor of 2).  They should compare test data from a vacuum chamber with their data.  In addition, they have limited crack growth rate data—only two specimens—and the stress-intensity factor range versus rate shows a strange behaviour.  The reason(s) for this behaviour was not explained.  Was the behaviour due to the overloads used to mark the specimen surface and measure the crack size?  Was the behaviour due to the stress-intensity equation written in term of the net section stress?

The crack front is in pure plane-strain conditions, which produces essentially no crack closure conditions under constant amplitude loading, as shown by their finite-element analyses.  But the overloads (1.25, 1.75) did produce crack closure and the results would explain the phenomena of delayed retardation.  However, their tests had used a factor of 2 overload, so why did they not analyze the case of a factor of 2 overload?  Based on their 1.75 overload, the reviewer suspects that the factor of 2 overload would have not allowed the crack to grow along the same plane.  It did appear that the last few overloads in the tests caused severe crack front markings—out of plane crack growth (?).  Thus, the low crack-growth rates compared to the other test data could have been caused by the factor of 2 overloads.

One conclusion needs to be removed:  Page 14, line 433:  “The negative influence of the overloads, used for marking, on fatigue life is not relevant”.  Here the authors are stating that the factor of 2 overloads were not relevant, so what caused the large discrepancy in the stress intensity factor range versus crack growth rates compare to other test data.

A better method to monitor crack size versus cycles is required and more tests under the same loading and at higher load ratios (R) are required.  The revised paper requires major revision and further work is required to publish the paper.

Author Response

Please see the attached file

Author Response File: Author Response.pdf

Reviewer 3 Report

(Full disclosure, I was Reviewer 3 on the original submission)

This article is a resubmission of a manuscript which was rejected after review. The authors have done a detailed revision of the work, but the paper is still not ready for publication. There are still five major points I would raise with the revised version: 

1. The authors have written an extremely detailed, long, and technically sound response letter to the original reviewer reports. In some ways, the letter is more detailed, helpful, and clear than the manuscript itself. For some reason that I do not understand, the authors seem to be trying to keep this paper short and this is what is causing most of the confusion. There are a lot of important technical detailed discussed in the letter that are not in the paper. Why? There are no page limits or page charges for any MDPI journal. If I were the authors, I would simply copy-paste large sections of the reply letter (with the technical details) into the paper itself since it competently explains a lot of the technical issues. If the final paper is 25-30 pages long with all the details, that is totally fine. If they are developing a new tool, it makes sense to have a very long paper. I am satisfied with the technical answers to my original concerns but insist that the answers to these things actually go into the paper itself. 

2. The authors still need to actually demonstrate that their geometry is under pure plane strain. I do not mean to refer to the literature for definitions - I am looking for a mathematical proof from basic mechanics or an experimental study. 

3. The statement about the defect geometry being mostly irrelevant to the behavior of the tests is only true for very brittle materials and would not be at all true for many materials this new test could apply to. The authors must either:

(a) explicitly state that this assumption may be wrong for many materials and that it is claimed to be a valid assumption only in this case. And then discuss the sensitivity of the manufacturing process and how it will be difficult to accurately reproduce these samples. Therefore, the test is limited to a certain range of materials depending on their defect sensitivity. 

OR

(b) Go back to fracture mechanics theory for LEFM (elastic) and EPFM (plastic) and show a mathematical or experimental proof that the defect shape and orientation does not affect the outcome using this new geometry for a wide range of materials.    

4. In addition to these concerns, the authors DO need to address my last point in the original review about verification, validation, and certification in detail. It can be a conceptual discussion and plan for the future but it does need to be there for a new tool. 

5. I still do not think this paper qualifies as a research article (it should be a technical note or tutorial paper). However, I would consider it an acceptable research article if the authors add the following three items to the paper on revision: 

(a) Very detailed and technically sound answers to points #3 and #4 IN the paper (not the response letter)

(b) A specific comparison between this method and the traditional methods in terms of mechanics, outcome, error sources, etc. I still think the vacuum testing claim for this sample is weak and recommend not putting it in there until more extensive tests have been done - however, this is a minor concerns and will not affect my accept/reject recommendation after revision. 

(c) Expanded discussion section

Author Response

Please see the attached file.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The subject paper has developed a very novel concept and useful test specimen for determining fatigue crack growth under vacuum (?) and under pure plane-strain conditions.  But the authors think that the environment around the internal crack in the specimen is in a vacuum.  This may be the case, but the effects of vacuum on fatigue crack propagation in the literature has shown much large differences between pure vacuum and lab air (like an order of magnitude and not a factor of 2).  They should compare test data from a vacuum chamber with their data.  In addition, they have limited crack growth rate data—only two specimens—and the stress-intensity factor range versus rate shows a strange behaviour.  The reason(s) for this behaviour was not explained.  Was the behaviour due to the overloads used to mark the specimen surface and measure the crack size?  Was the behaviour due to the stress-intensity equation written in term of the net section stress?

The crack front is in pure plane-strain conditions, which produces essentially no crack closure conditions under constant amplitude loading, as shown by their finite-element analyses.  But the overloads (1.25, 1.75) did produce crack closure and the results would explain the phenomena of delayed retardation.  However, their tests had used a factor of 2 overload, so why did they not analyze the case of a factor of 2 overload?  Based on their 1.75 overload, the reviewer suspects that the factor of 2 overload would have not allowed the crack to grow along the same plane.  It did appear that the last few overloads in the tests caused severe crack front markings—out of plane crack growth (?).  Thus, the low crack-growth rates compared to the other test data could have been caused by the factor of 2 overloads.

One conclusion needs to be removed:  Page 14, line 433:  “The negative influence of the overloads, used for marking, on fatigue life is not relevant”.  Here the authors are stating that the factor of 2 overloads were not relevant, so what caused the large discrepancy in the stress intensity factor range versus crack growth rates compare to other test data.

A better method to monitor crack size versus cycles is required and more tests under the same loading and at higher load ratios (R) are required.  Further work is required to publish the paper.

Author Response

Dear reviewer

Please find our comments in the attached document.

We hope that the paper can now be accepted.

yours sincerely

fernando antunes 

Author Response File: Author Response.pdf

Reviewer 3 Report

The authors have done a lot of work on this paper and have addressed most of the issues I raised in my review reports for both submissions of this paper. I acknowledge that the remaining issues (as argued in the authors' reply letter) may be subject to good-faith disagreements or interpretations within this rapidly developing field and that one paper cannot cover every possible question.

From here, I commend the authors on the amount of work they have put into this manuscript and will defer to the editor on whether the paper is acceptable for publication as it is. If the editor is satisfied with the current state of the work (after several revisions), I recommend acceptance.

I will just mention to the authors that some things in the paper, while not wrong, could be a bit controversial and there may be some conversation around this paper in the future. The authors should continue working on this method/problem and refine it more in the future. 

Author Response

Dear reviewer

Thank you for your comments, which promoted the increase of the global quality of the paper.

We intend to use this geometry in future works, therefore we believe that we will identify and solve the problems involved. In fact, we are now producing more specimens.

best regards

fernando antunes

Round 3

Reviewer 2 Report

The revised paper is suitable for publication.

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

The authors proposed a cylindrical specimen with a central crack produced by additive manufacturing. They asserted that the specimen was effective to investigate the effect of vacuum on crack growth rate without expensive vacuum equipment. I agree the advantage of the specimen developed in this study. However, I cannot understand what the authors want to say in this paper. The purpose and the direction are unclear, and the manuscript includes serious mistakes resulting from insufficient polishing. In my opinion, therefore, this manuscript does not meet the requirement of this journal for publication from the viewpoint of quality and reliability. I would like you to check the following comments. 

 

1. The authors mentioned that FCG life in vacuum is higher than in air in P6. In Figure 5, however, the opposite results are found. I think the symbols used in the figure are not correct.

 

2. The text in the last sentence in P7 was cut off. This sentence may be related with the concluding sentence, so I cannot understand the summary of this paragraph.

 

3. In the introduction, the authors mentioned that the main objective of this paper is to propose a specimen geometry which permits pure plane strain along the whole crack front. However, no clear explanation was made regarding the amount of plain strain achieved by the proposed specimen. Especially, in P7, the authors compared the crack growth rate between the data using CT specimens and that using the specimen proposed. I think this point is very important, however, no clear reason for the difference was concluded. So, I cannot understand what the authors want to say in this manuscript.

 

4. The authors think that the environment around the internal crack of this specimen is a vacuum. However, the gas molecules must be included in the space of internal defect under additive manufacturing process. Much more careful descriptions should be done regarding the effect of vacuum on crack growth rate.

 

5. I think the overload used in this study has a considerable retardation of crack growth rate. To declare the merit of this specimen, the authors should explain the effect of overload quantitatively.

Reviewer 2 Report

The subject paper has developed a very novel concept and useful test specimen for determining fatigue crack growth under vacuum and under pure plane-strain conditions, and the paper should be published.  However, the current version is rejected.  There are several areas in the paper that need to be corrected and improved.  The major issues are that some figures (Figs. 5 and 6) appear to be mislabeled, and the da/dN vs ΔK data for vacuum and air (Fig 6) tends to level off in rate instead of increasing with ΔK.  This may be caused by the stress intensity factor (SIF) equation written in terms of the net-section stress.  Reviewer is not questioning the SIF equation because the SIF can be written in terms of net-section stress but using net-section stress is unusual.

(1) General:  Authors use CT and MT for compact and middle-crack tension specimens.  It is common practice to use C(T) and M(T).

(2) Page 2:  Statement “Besides, there are researchers arguing that PICC does not exist (particularly for plane strain conditions), even suggesting that the plastic wake is responsible for crack opening and not for crack closure [24-26].”

Reviewer agrees that crack closure should not exist under plane-strain conditions, but crack closure does occur under plane-stress conditions.  For structural applications, most cracks grow from free surfaces where stress concentration are higher than interior locations and the material is under plane-stress conditions.

The reviewer is very excited about the new novel test specimen being proposed in this paper.  It would be very important for the authors (in the future) to conduct tests under various R (P_min/P_max) to show da/dN vs ΔK behaviour under “pure” plane-strain conditions.  Would these tests show an R effect on the da/dN vs ΔK data?

The authors also need to pursue the measurement of load vs displacement during loading and unloading to provide experimental evidence on the lack or presence of crack closure.

(3) Page 5, Section 3:  The authors should avoid the used of trade names for the equipment used, such as “Dartec”, “(TestResources, Minneapolis, USA)” and “Leica DM4000 M LED”.  Stating that they used “a servo hydraulic (test) machine” and an “optical microscope” is satisfactory.

(4) Page 6/7 (Fig 5):  The statement “… FCG life in vacuum is higher than in air. In other words, for the same crack extension, the number of cycles to achieve the crack extension is higher in vacuum, which means that the FCG rate in vacuum is lower than in air.”  However, Figure 5 shows that the crack grows faster in vacuum.  Is Figure 5 mislabeled?

(5) Page 7 (Eqn 1-3):   The SIF equation is written in terms of the net-section stress.  The reviewer has not checked the equation and the equation may be correct, but it is unusual to use net-section stress.  Normally, the remote stress is used.

(6) Page 8 (Fig 6):  The reviewer is concerned with the da/dN vs ΔK test data.  One would not expect the rates to tend to level off.  Is there a problem with the SIF equation or is there residual stresses in the test specimen?

(7) Conclusions:

“a pure plane stress state exists along all crack front.”  Statement should be “pure plane strain”

“the crack closure level is relatively small or does not exist at all.”  Subject paper did NOT present test evidence on the crack closure level.  It is only an opinion.   The conclusion must be removed.

“since there is no crack closure or this is limited, the negative influence of the over-loads, used for marking, on fatigue life is less relevant.”  Subject paper did NOT present test evidence on the crack closure level during and after over-loads.  It is only an opinion.   The conclusion must be removed.

(8) Ref 37:  Need more information, such as publisher or publishing organization.

(9) Figures:  Need to improve figure captions, some are too brief, such as “da/dN-ΔK results” and “a-N curves”.

Reviewer 3 Report

In this work, the authors develop and present a novel specimen geometry for studying pure plane strain fatigue crack growth in SLM-processed titanium alloy. The work presented is interesting and is definitely something unique in the field. However, I do not think the contribution of this work is intellectual as required for a research article - the article is clearly written as a technical note or tutorial in its content and scope (this is not a criticism of the work). I also have some other major concerns with the presented study and the paper itself, as outlined below. I believe that the work can be revised into a useful paper, but the amount of time and work required is likely beyond that which would be reasonable for a major revision. Therefore, I cannot support acceptance of the paper as it stands, but I highly encourage the authors to rework it to address my comments and resubmit to this journal as a technical note or tutorial. I will be willing to review the resubmitted version if asked by the handling editor.

Major comments 

1. As previously stated, this is a valid study that could be useful after some additional work but certainly does not qualify as a research article since there is no new conceptual contribution or new way of thinking about problems - it is a new tool being introduced. It should be a technical note/brief or tutorial article. These are still valid journal papers and better represent the actual contribution of the work and its potential impact on the field. I often cite tech notes and tutorials as much or more than research articles anyway. 

2. The title is very misleading and does not represent the paper at all. The specimen developed is useful for a specific process and material, as demonstrated by the authors. This is not a universal testing specimen. Just looking at your data, this specimen would not work with more brittle materials (i.e., any titanium alloy except the one used). 

3. The claim that vacuum studies can be done using this specimen without a vacuum system since the "crack front does not contact the air" is not convincingly shown. It is claimed logically (and it could make sense) but any microcracks or porosity in the build would make this not true. Unless this claim is actually tested rigorously and can be shown to be true, it is probably best not to claim it. Anyone who wants to do vacuum tests most likely already has the required equipment available and so this claim adds very minimal value. 

4. Your specimens are going to be notoriously hard to manufacture consistently, quickly, and in bulk. Many of the new testing standards and specimens with similar claims to this study are done to make testing faster and more reliable. For this to work, a non-destructive evaluation method must also be developed to ensure that the internal features are consistent. It is not as much of a problem for TI-64 (the material used) since it is relatively ductile but for many materials a tiny variance in the size or angle could have a large impact on the testing reliability and repeatability. In addition, how do the authors ensure that there is no powder inside of the specimen? 

5. It is not shown well enough why these cylindrical specimens provide superior fatigue testing results to the classic CT and MT specimens. The CT and MT tests have 50+ years of development behind them and are the basis of most of the standards we used now. How are these better? Besides possibly being cheaper and easier to manufacture using AM? 

6. The instability of crack front geometry is discussed at several points in this paper, but its impact on the study at hand was not clearly shown by the authors. The data shows that the crack front was probably unusually stable for this kind of test - in the revised version of this paper, this is probably something the authors should consider presenting as a clear benefit and why this method may be useful or superior to other methods. 

7. How did the authors actually establish that the specimen geometry satisfies the requirements for plane strain condition? This is not trivial to show, especially with an additively manufactured material. It requires several rounds of experimental results in addition to calculating the stress/strain intensity factors, etc. Calculating K_I is not enough to show this convincingly.  

8. There is almost no discussion of powder quality, which could have a large impact on the usefulness of this work

9. Line 153-155: After the heat treating cycle to relieve residual stresses, were the samples inspected for cracks, etc? Were any samples rejected/replaced at this point? 

10. Section 4: I do not understand the need for the FEA model. There is not enough discussion provided about it or what it is trying to show. 

11. Conclusions: All of these bullet points need to have caveats or more detailed explanation. The claims are too strong for the presented results. 

12. There is little discussion of standardization, certification, verification, and validation concerns with this method. When proposing a new method, some discussion of these is essential. 

Minor concerns 

13. The first four authors have the same affiliation, so condensing this will make the paper look nicer

14. There are many, many technical terms and a lot of jargon used in this paper. While the authors did a good job defining everything, it can get a little confusing. I recommend adding a nomenclature table. 

15. There are a large number of self citations to the authors in the reference list, with one author taking the majority of these. While it is not automatically bad to cite the authors several times, it is kind of hard to justify in a paper like this and should be carefully reviewed. 

Good luck to the authors and I wish you a successful and quick revision cycle

Reviewer 4 Report

The following comments are given by the reviewer (answers/discussions should be preferably included in the revised version of the paper):

- Abstract: The most significant results should be included in the abstract, e.g. how much is “a great influence of atmosphere”?

- Section 2: Specimen design: How accurate could the real specimens be manufactured? Please provide more information about the reproducibility of the specimens. It is of course fine to produce such specimens; however, if the reproducibility is weak, the outcome of several tests may be majorly affected. How can a quality assurance procedure may look like to proof the reproducibility?

- Section 3: Overloads were applied for markings. As they may affect crack growth (e.g. plasticity induced retardation effects), how was this considered?

- Section 4: More detailed information regarding the finite element model/analysis should be given. This analysis may not be appropriate located within the Discussion section as it shows further investigations and results. It may be shifted to a section dealing with numerical analysis.

- Conclusions: Should be supported with more quantitative results to highlight the outcome.