Use of Synthetic Data in Maritime Applications for the Problem of Steam Turbine Exergy Analysis
Round 1
Reviewer 1 Report
The manuscript entitles “Use of Synthetic Data in Maritime Applications for the Problem of Steam Turbine Exergy Analysis”
This short paper extends a previous study by the same authors. This short paper study and its results are interesting, with proper explanation. However, scientifically many aspects need to be considered. This manuscript may be published with a few minor revisions.
General comments:
- It is always good to verify the method to consider at least another data set for maritime applications, depending on data availability.
- Regenerate Figure-6; a few words must be clarified on panels (a) and (b).
- Table-3 needs to be rewritten, with the same column number in each row. Put “--” or leave it blank if no values are available.
- Line 438 “.....less than 2/3rds ……”, rewrite it in words.
Author Response
Respected reviewer,
Thank you for your revision of our manuscript. Please find answers to your comments in the attachment. We have marked the changes made to the manuscript according to your comments using cyan text color.
Kindest regards,
the authors
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This paper presents the Use of Synthetic Data in Maritime Applications for the Problem of Steam Turbine Exergy Analysis. I have some comments for the revised version: 1. the results in the abstract can be improved to be more precise. 2. The introduction can be improved by providing the importance of ML, what is the current situation and then the gap of the work. 3. The auhtors can divide between description of the physical model and ML model. 4. The auhtors can provide a schematic diagram of the process of all simulation 5. Please differ between . and , in numbers. "." is for decimal and "," is for thousands 6. Conclusion can be improved by providing more detailed description of the work
Author Response
Respected reviewer,
Thank you for your revision of our manuscript. Please find answers to your comments in the attachment. We have marked the changes made to the manuscript according to your comments using red text color.
Kindest regards,
the authors
Author Response File: Author Response.pdf
Reviewer 3 Report
Dear Authors,
The manuscript is a good effort in introducing synthetic data based models in early stage prediction of performance in marine applications. The problem statement is clear and the proposed method has been described sufficiently.
I have some concerns as addressed below and if appropriate explanations are added in the revised draft, it can be accepted.
1) What is the motivation to use Copula, TVAE and CTGAN as opposed to any other method? Please explain why these were chosen specifically.
2) In MLP, how did you arrive at 6 number of hidden layers as the limit? In other words, what if it had 8 hidden layers?
3) Also, please add a table showing the computing resources used to train and how much compute time was involved in each of the methods for the readers to get a better understanding of the resource requirement.
4) Need an explanation for the hyperparameters chosen in Table 1 for MLP and Table 2 for XGB. What parametric sensitivity study was conducted to choose these?
5) For better readability can you have the same limits for MAE, MSE for both MLP and XGB comparison in all your figures (2-7)?
6) What happens if I give more than 100 points of data but very sparse. How does the model behave? In other words, could you show how densely packed or sparse are your real data points?
Need some minor edits.
Author Response
Respected reviewer,
Thank you for your revision of our manuscript. Please find answers to your comments in the attachment. We have marked the changes made to the manuscript according to your comments using green text color.
Kindest regards,
the authors
Author Response File: Author Response.pdf
Round 2
Reviewer 3 Report
All concerns have been answered.
Read through once.
