On the Representativeness of the Cohesive Zone Model in the Simulation of the Delamination Problem
Round 1
Reviewer 1 Report
The paper presents nice numerical results and discussion about cohesive zone modelling referred to the composite structures. I appreciate this paper by marking minor revision. Before publication please consider the points listed below:
1. It will be great to discuss more in the introduction the problem of CZM models – present a few bilinear, exponential. Please, discuss advantages/disadvantages etc.
2. Missed references: line 426, 445, 456, 460-461, 463, 472, 505-506, 514-515, 526, 545, 548, 572-573, 604-605,
3. Add additional Figure with used CZM model – please mark a characteristic quantities used in numerical analysis
4. A valuable will be comparing with the experimental data
Author Response
Point 1: It will be great to discuss more in the introduction the problem of CZM models – present a few bilinear, exponential. Please, discuss advantages/disadvantages etc.
Response 1: Authors agree that supplementing the paper with additional information on CZM models as suggested by the Reviewer would benefit discussion in the paper. We suggest that in the context of the paper, this discussion may be better suited for section 2.3 (Definition of material behaviour), as it would help to explain the particular choice of the softening scheme adopted in the present study. The statement regarding use of various damage evolution laws has been added on page 8.
Point 2: Missed references: line 426, 445, 456, 460-461, 463, 472, 505-506, 514-515, 526, 545, 548, 572-573, 604-605,
Response 2: Authors regret that the manuscript as received by the Reviewer contained numerous referencing issues. These seem to have been caused by some inexplicable malfunction of software during the submission, which not only upset all the automated figure referencing starting from figure 8, but also resulted in truncation of some of the figure captions. In the revised version of manuscript, all the captions have been restored to their original form and all the missing references have been replaced.
Point 3: Add additional Figure with used CZM model – please mark a characteristic quantities used in numerical analysis
Response 3: Figure has been added as requested (Figure 6 in the revised version of the paper)
Point 4: A valuable will be comparing with the experimental data
Response 4: Authors agree that comparison with the experiment could have been a nice addition to the paper. However, conducting the experiments is time-consuming hence not practical at this stage. Furthermore, the FE results were compared with the analytical solution, which is in a sense a better indication of the accuracy of the FE simulation, as it is free from the experimental error.
Reviewer 2 Report
Firstly, this paper is very well written and structured. Very easy to follow! English scientific language is excellent!
Secondly, although this paper didn't develop any new methodologies for modeling delamination, it provides a comprehensive evaluation on the cohesive models used in ABAQUS. The sensitivity of predictions to mesh density, normal interface stiffness, and viscosity parameter were thoroughly studied. These results can potentially serve as user guidelines for researchers to select appropriate mesh density and other parameters of the cohesive models for modeling delamination of composite materials.
I would like to recommend this paper to be accepted. A few minor typos including the "reference links" need to be corrected before publication.
Author Response
Point 1: Firstly, this paper is very well written and structured. Very easy to follow! English scientific language is excellent!
Secondly, although this paper didn't develop any new methodologies for modeling delamination, it provides a comprehensive evaluation on the cohesive models used in ABAQUS. The sensitivity of predictions to mesh density, normal interface stiffness, and viscosity parameter were thoroughly studied. These results can potentially serve as user guidelines for researchers to select appropriate mesh density and other parameters of the cohesive models for modeling delamination of composite materials.
Response 1: The authors would like to thank the reviewer to appreciating the objectives of the work conducted and complimentary assessment of the quality of presentation
Point 2: I would like to recommend this paper to be accepted. A few minor typos including the "reference links" need to be corrected before publication.
Response 2: Authors regret that the manuscript as received by the Reviewer contained numerous referencing issues. These seem to have been caused by some inexplicable malfunction of software during the submission, which not only upset all the automated figure referencing starting from figure 8, but also resulted in truncation of the figure captions. In the revised version of manuscript, all the captions have been restored and all the missing references have been replaced.
Reviewer 3 Report
Report of the review of jcs-443149 for Journal of Composites Science, “On the representativeness of the cohesive zone model in the simulation of the delamination problem”
The manuscript presents the Finite Element simulation of delamination using cohesive zone modeling and calibrating some parameters.
Some comments are as follows:
1. The introduction section is very long and contains too much information. The literature review part can be shortened. Moreover, the novelty of the research is not clarified.
2. Figure 4 can be presented colorful. Different segments of the model are not detectable. In addition, Figure 4(b) is supposed to present the difference of the mesh density in two segments; however, it is not detectable.
3. Mesh convergency based on element size is too long, all the information is not necessary. The method is obvious and there is no novelty on it.
4. Many cross-references including references and figure numbers are missed through the whole draft.
5. Figure 9 is vague. Which part of the the model is presented? Where does delamination occur?
6. Overall, the presented research in the present form is not presenting any novelty or contribution. There are many papers investigated the effects of different parameters of cohesive elements. It is recommended to clearly explain about this issue.
Author Response
Point 1: The introduction section is very long and contains too much information. The literature review part can be shortened. Moreover, the novelty of the research is not clarified.
Response 1: Overall, the introduction section covers two main subjects of the paper, first one being verification of the representativeness of the delamination model, and the second one parameter choices in the model corresponding to a converged solution. Since the first subject is completely new hence requires some elaboration, we understand that the Reviewer suggests to shorten part covering to the second subject. The reason why it was included in the first place is that delamination modelling is perceived by many to be well-understood, but when one begins to work on it, he/she can find rather controversial accounts on choices of parameters in the model reported in countless publications on the subject. Furthermore, some of the recommendations regarding the parameter choices can even be erroneous, yet they are re-iterated from paper to paper. In fact, we encountered this issue ourselves, and in absence of clear guidelines it took us a substantial amount of time and effort to obtain a converged solution. Making such information available would undoubtedly benefit a wide range of users involved in finite element modelling of delamination problems. We believe it is appropriate to include in to introduction a brief review of most popular recommendations on parameter choices often encountered in the literature prior to providing guidelines of our own. As for the length of the introduction section, the judgement can be subjective sometimes. For instance, Reviewer 1 in his/her first point even suggested to include more contents in the introduction.
A statement further emphasising the novelty of research we claim was added in the end of introductory section.
Point 2: Figure 4 can be presented colorful. Different segments of the model are not detectable. In addition, Figure 4(b) is supposed to present the difference of the mesh density in two segments; however, it is not detectable.
Response 2: Authors agree that the variations in the mesh in FE models shown in Figure 4 are not apparent. This is due to very high mesh density, hence varying the colour of the figure will not make it more pronounced. We suggest as an alternative to increase the size of the figure 4(b) instead, which has been done in the revised version of the paper. Now the difference between the biased mesh on one side of delamination front and fine uniform mesh on another side can clearly be seen.
Point 3: Mesh convergency based on element size is too long, all the information is not necessary. The method is obvious and there is no novelty on it.
Response 3: While mesh convergence in general is indeed an old topic, the way it has been dealt with in the paper is new. The kind of information we present has not been put to public domain to date. In particular, there is no account available in the literature explaining in a systematic and the consistent manner all the implications of insufficient of mesh refinement. It has been shown in the paper that non-converged load-displacement curves can look completely “normal”, and there is no lack of users who, while aiming to reproduce the experimental data, would be satisfied with the modelling results as soon as the good agreement between the two has been reached, even though genuine convergence has not been obtained, hence the solution is not correct.
Furthermore, it has been shown that mesh demands even in the simple cases considered are very high. Presentation of mesh refinement case by case, where mesh is varied in one of three directions, will help the other users conducting mesh sensitivity studies to avoid unnecessary mesh refinement in certain directions. For instance, mode II delamination tends to be more sensitive to the sideways mesh density than that in the longitudinal direction as revealed in this paper but this has never been documented properly in the literature. This in fact was our main incentive for including the full mesh sensitivity study in the paper.
Point 4: Many cross-references including references and figure numbers are missed through the whole draft.
Response 4: Authors regret that the manuscript as received by the Reviewer contained numerous referencing issues. These seem to have been caused by some inexplicable malfunction of software during the submission, which not only upset all the automated figure referencing starting from figure 8, but also resulted in truncation of the figure captions. In the revised version of manuscript, all the captions have been restored and all the missing references have been replaced.
Point 5: Figure 9 is vague. Which part of the the model is presented? Where does delamination occur?
Response 5: Authors understand that the Reviewer refers to the Figure 9 in section 4.1 in the original version of manuscript, as due to software issues as explained in the response to Point 4, the paper ended up with two figures with the same number. To clarify the data presented in this figure, a schematic drawing was added to this figure, specifying what part of interface the figures shown refer to. In the revised version of the manuscript, it is Figure 18 on page …
Point 6: Overall, the presented research in the present form is not presenting any novelty or contribution. There are many papers investigated the effects of different parameters of cohesive elements. It is recommended to clearly explain about this issue.
Response 6: This has been addressed in response to Point 1 raised by the Reviewer. To reiterate the novelty of this paper, it is a reflection on the representativeness of the CZM for the simulation of the delamination problem, rather than a routine parametric study. Those who have direct experience of delamination modelling using CZM would agree on the challenges in obtaining fully converged solutions and being able to do so with confidence alone is indeed of great value. The authors also agree with the comment made by this Reviewer as well as the other Reviewers that there had been no lack of accounts describing the efforts of achieving converged solutions and consequently the use of CZM is more and more being taken for granted. However, can CZM really live up to this standard? There has been genuine lack of a sober reflection on the representativeness of CZM.
Round 2
Reviewer 3 Report
No comment!