Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper focuses on a key topic for the improvement of the industry 4.0 scenario. The authors propose a framework to integrate real-time monitoring and feedback control in FMS to improve quality indicators. The approach is clearly described and the paper and well structured. Nevertheless, some improvements are needed before considering for publication. Particularly, the description of the case study to scientifically support the outcomes must be deepened. Also, the relation between the presented results and the outputs from the physical tests should be specified. Finally, while the paper is generally well written, there are too many typos and writing errors.

Please consider the following comments:

Use of acronyms: they should be first stated in full and then used along the paper (e.g., Digital Twin in L and DT in L69, etc.).  Also, do you present a CT framework (abstract, introduction) or a CTF (discussion)?

LL61-64: References (e.g., [8]) are needed

L68-69: Citation of [6] is needed

Fig.1: It is not described; also, do you have permission for reproduction?

L110: Franciosa et al.

L114: D’Amico et al.

L126: Product Function/Feature. There is also a typo in the abstract of the reference.

Refs 15-18: It is not clear how they fully fit. Which are the cognitive enhancements?

L173: Johansen is a typo

Ref 27: The topic and the outcome are not clear

Section 2.2: How are the indicators selected? Types of data are retrieved, for example, from Wang et al. [22]. Any references?

Fig. 2; It is not mentioned within the text. Also, the flowchart does not match with the implementation in the result section.

L297: typo

L347: typo

L348 2 typos

Section 3: references for Unity engine? (e.g., L378, then 4.1.3)

3.2 lacks scientific support and can be further explained, also, provide more information about data integration.

Machine sensors: provide more details; which are the sensors and the measurements?*

Feedstock quality: provide more information; how do you measure purity and consistency?*

L425 typo

Product quality: which are the measurements?*

*eventually, they can be described in section 4.

Fig. 5: It is not clear. What is the scope and what are the differences between left and right machines and interfaces?

Section 4 must be improved. What is the case study? What are the machining operations? Provide sufficient details (e.g., geometry, functional features, work cycles, measurements, etc.).

Revise sect. 4 and provide precise description of outputs/data for Qf, Qm, Qp, Qi (here or in4.1, 4.2, 4.3, 4.4). What are soft foam and aluminium supposed for? What is consistency? How do you measure it? What is purity and how do you measure it? Which are the measurements on the feedstock? Which temperature is measured? What is vibration? Which time do you measure (Conversely, you mentioned a tool wear sensor)? How do you calculate processing quality? What are the inspection measurements? Which features, which tools, which accuracy, etc.? These questions would help us to understand the basis of the dataset further presented.

L481 typo

L481 Qtotal (t)

L494 Table 1

Tables 1-6 I would suggest introducing a column with sample n#

Tab.1 How is Qf calculated?

4.8, which are process (cutting) parameters? Eventually, a table could be useful

L498 (Qm)

L501 is it Qi?

L502 typo

Tab. 2 How is Qm calculated?

Tab. 2 format (bold)

L508 (Qp)

Tab. 3 How is Qp calculated?

Tab. 3 format (first column bold)

4.4 This must be deepened (see previous comments).

L518 (Qi)

Tab. 4 How is Qi calculated?

Tab. 4 format (first column bold)

Tab. 6 format (first column bold)

L573 typo

L574 typo

Fig. 6: X axis is missing

Tab 7: Check table format

L581: How do you define acceptable limits?

L583 typo

L597 OEE?

Section 4.8 should be also improved. Tab. 8,9: Where do the data come from? What the reported 10 cycles?

Tab 9: Check table format

L606 typo

L618 typo

Fig. 9 is not clear. Where do real-time data come from (see previous comments)? For example, how can you have real-time data inspection during machining? Also, what are the operations reported? (see the need for work cycles explanation highlighted). Sloting -> Slotting

Fig. 10: What is the calculation for the correlations?

Section 6: There are issues about the results reporting. Provide the necessary support with the key findings.

LL689-692: the description is misleading, better to report improvement in percentage for indicators such as the overall equipment efficiency and the scrap rate.

The same must be considered for the abstract, which can be revised considering the fulfilment of the suggestions.

 

 

 

 

 

 

 

 

 

Author Response

Reviewer 1(Revision in Yellow)

• the description of the case study to scientifically support the outcomes must be deepened.
• Also, the relation between the presented results and the outputs from the physical tests should be specified.

Comment: Use of acronyms: they should be first stated in full and then used along the paper (e.g., Digital Twin in L and DT in L69, etc.). Also, do you present a CT framework (abstract, introduction) or a CTF (discussion)?

Reply: I have fixed all the acronyms and typos through out the manuscript.

Comment: Fig.1: It is not described; also, do you have permission for reproduction?

Reply: Yes, I have the permission and i have described the required information as well.

Comment: Refs 15-18: It is not clear how they fully fit. Which are the cognitive enhancements?

Reply:  Dear Reviewer, they don’t have cognitive capabilities, but they have integrated various advance technologies with Digital Twin. They are mentioned for their advance sensor application in DT framework. So, I have created another subsection for them.

Comment: Ref 27: The topic and the outcome are not clear

Reply:  The reference is further described.

Comment: Section 2.2: How are the indicators selected? Types of data are retrieved, for example, from Wang et al. [22]. Any references?

Reply: Section 2.2 is expanded to describe the indicators selection and as well as the data types of retrieval.

Comment: Fig. 2; It is not mentioned within the text. Also, the flowchart does not match with the implementation in the result section.

Reply: The figure is modified to represent the actual implementation of the system. And cited in the text.

Comment: Section 3: references for Unity engine? (e.g., L378, then 4.1.3)

Reply: Reference is added in both sections

Comment: 3.2 lacks scientific support and can be further explained, also, provide more information about data integration. Machine sensors: provide more details; which are the sensors and the measurements?* Feedstock quality: provide more information; how do you measure purity and consistency?

Reply: The necessary information is updated as instructed in section 3.2.

Comment: Product quality: which are the measurements?* *eventually, they can be described in section 4.

Reply: Section 3.2.4 is expanded as instructed.

Comment: Fig. 5: It is not clear. What is the scope and what are the differences between left and right machines and interfaces?

Reply: the description is added to the paragraph where the figure is cited.

Comment: Section 4 must be improved. What is the case study? What are the machining operations? Provide sufficient details (e.g., geometry, functional features, work cycles, measurements, etc.).

Reply: The details are added to section 4, as instructed.

Comment: Revise sect. 4 and provide precise description of outputs/data for Qf, Qm, Qp, Qi (here or in4.1, 4.2, 4.3, 4.4). What are soft foam and aluminium supposed for? What is consistency? How do you measure it? What is purity and how do you measure it? Which are the measurements on the feedstock? Which temperature is measured? What is vibration? Which time do you measure (Conversely, you mentioned a tool wear sensor)? How do you calculate processing quality? What are the inspection measurements? Which features, which tools, which accuracy, etc.? These questions would help us to understand the basis of the dataset further presented.

Reply: Section 4.1 , 4.2 , 4.3 and 4.4 are updated with information as instructed from the reviewer.

Comment: 4.8, which are process (cutting) parameters? Eventually, a table could be useful L498 (Qm).

Reply: The table is inserted as instructed by the Reviewer.

Comment: Fig. 6: X axis is missing

Reply: The x axis is fixed.

Comment: L581: How do you define acceptable limits?

Reply:  it is usually through industrial standards and application specific. Here the relative error is extremely small and justified.

Comment: Section 4.8 should be also improved. Tab. 8,9: Where do the data come from? What the reported 10 cycles?

Reply: the section is updated as per instructions.

Comment: Fig. 9 is not clear. Where do real-time data come from (see previous comments)? For example, how can you have real-time data inspection during machining? Also, what are the operations reported? (see the need for work cycles explanation highlighted). Sloting -> Slotting

Reply: the figure is fixed according to the reviewer’s comment.

Comment: Tab 9: Check table format

Reply: the table format is fixed.

Comment: Fig. 10: What is the calculation for the correlations?

Reply: IT is done through Pearson Correlation Coefficient.

Comment: Section 6: There are issues about the results reporting. Provide the necessary support with the key findings.LL689-692: the description is misleading, better to report improvement in percentage for indicators such as the overall equipment efficiency and the scrap rate.

Reply: It is included in the conclusion.

Comment: The same must be considered for the abstract, which can be revised considering the fulfilment of the suggestions.

Reply:   The abstract is also modified as per instructions.

Reviewer 2 Report

Comments and Suggestions for Authors

Thank you very much for your contribution, which should be partially improved.

Comments for author File: Comments.pdf

Comments on the Quality of English Language

Should be improved

Author Response

Reviewer 2 (Revisions in Turquoise )

We have adjusted the manuscript as indicated by the reviewers. We believe that the conclusion and the discussion sections covers all the relevant information as per our understanding. 

Reviewer 3 Report

Comments and Suggestions for Authors

57 "Although Digital Twin offers a valuable static or semi-dynamic model of the manufacturing process, but it lacks the predictive, adaptive and cognitive abilities." On the basis of what definition does such a statement appear? In the works, the use of digital twin is often identified as a solution that immediately implements AI. The distinction proposed by the authors should be supported by argumentation and a literature review. So as to demonstrate the literature basis that will directly justify the division adopted by the authors.

61 Researchers and practitioners in response to these limitations, are exploring the next step of Digital Twin, that is "Cognitive Twin". Was the term Coginitve Twin proposed by the authors or does it result from the literature? What are the grounds for distinguishing such a formulation? Indicating the literature review [6], which is the only one to deal with the subject, seems to be an insufficient verification of the background for such a widely discussed concept. It is necessary here to conduct an in-depth analysis and demonstrate the legitimacy of the deadline.

107 The authors cite a number of articles where cognitive enhancements appear, but in fact they still talk about digital Twin. Are there studies that indicate, in addition to [6], the separation of Coginitive Twin as the next level for DT?

446 Historical data: how many production cycles are affected? To what extent do they allow for optimization?

 

467 How many test scenarios have been implemented? How many different test setups were performed?

Author Response

Reviewer 3(Revision in Pink)

Comment: L57 "Although Digital Twin offers a valuable static or semi-dynamic model of the manufacturing process, but it lacks the predictive, adaptive and cognitive abilities." On the basis of what definition does such a statement appear?

Reply: The support is shown through the literature review in highlighted lines 60 and 61.

Comment :In the works, the use of digital twin is often identified as a solution that immediately implements AI. The distinction proposed by the authors should be supported by argumentation and a literature review. So as to demonstrate the literature basis that will directly justify the division adopted by the authors. 61 Researchers and practitioners in response to these limitations, are exploring the next step of Digital Twin, that is "Cognitive Twin". Was the term Cognitive Twin proposed by the authors or does it result from the literature? What are the grounds for distinguishing such a formulation? Indicating the literature review [6], which is the only one to deal with the subject, seems to be an insufficient verification of the background for such a widely discussed concept. It is necessary here to conduct an in-depth analysis and demonstrate the legitimacy of the deadline. 107 The authors cite a number of articles where cognitive enhancements appear, but in fact they still talk about digital Twin. Are there studies that indicate, in addition to [6], the separation of Cognitive Twin as the next level for DT?

Reply:  the elaboration is added in line 62 to 64. And the 68 , 69. The citations are for reference.

Comment: 446 Historical data: how many production cycles are affected? To what extent do they allow for optimization?

Reply: the section is updated as shown in 501 to 504.

Comment: 467 How many test scenarios have been implemented? How many different test setups were performed?

Reply:   It is addressed in the updated section 4, Results.  

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The first review improved the literature review, the analysis of the results and the conclusion.

From a theoretical point of view, the proposed approach is very interesting.

Still, there is one relevant issue regarding the case study (section 3 and section 4). The replies were very vague, the description of the implementation is highly imprecise and therefore the experimentation appears unreliable.

Author Response

Revisions are Highlighted in Red

Comment: The first review improved the literature review, the analysis of the results and the conclusion. From a theoretical point of view, the proposed approach is very interesting.

Still, there is one relevant issue regarding the case study (section 3 and section 4). The replies were very vague, the description of the implementation is highly imprecise and therefore the experimentation appears unreliable.

Reply: The authors are deeply thankful to you for your suggestion to improve the manuscript further. The proposed work is inspired from Wang et al, [1]. His paper suggested four basic type of quality components namely feedstock, machine degradation, product processing and inspection quality. To further the framework, the authors used optimization techniques used by Feng et al [41] and Piotyr et al [42]. The optimization technique using Random Forest Method to extract weights in an equation/process also called feature selection based on importance. As Random Forest model is an ensemble method, it is best suited to find decision making and weight tuning in an equation.

In the light of the reviewer’s comment, we have now thoroughly revised section 3 and section 4. We have updated sections highlighted in the revised manuscript. We hope this revision will further clarify the questions around implementation.  

Author Response File: Author Response.pdf

Round 3

Reviewer 1 Report

Comments and Suggestions for Authors

The Authors have clearly improved the description of the implementation phases in rev 3.  

However, some parts introduced during rev 2 are not precise and details on machined components are missing, while the document is currently very long.  

However, the overall document has improved through revisions and the proposed framework is well described, so the paper can be published.

Author Response

Revisions are Highlighted in Voilet

Comment: The Authors have clearly improved the description of the implementation phases in rev 3.

However, some parts introduced during rev 2 are not precise and details on machined components are missing, while the document is currently very long. 

However, the overall document has improved through revisions and the proposed framework is well described, so the paper can be published.

Reply: The authors once again thank the reviewers and the editors for their valuable suggestions. We have now thoroughly revised the manuscript in the light of the reviewer’s comments. A key improvement includes the addition of detailed description of experimental processes undertaken by the FMS.  Details are in section 4.11.

Author Response File: Author Response.pdf