Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Thank you for the opportunity to read the paper and share my comments with the Authors.

I'm not sure I received the correct version of the paper for review. The submitted version of the paper included comments regarding the incorrect numbering of some figures.

I have doubts about the wording in the title. Are the Authors optimizing profits or maximizing profits? Furthermore, if the profit (over 900 units) is approximately three times higher than the costs (approximately 300 units), doesn't this indicate an advantage for the seller over the buyer (Figure 16). Is this a monopoly? Please explain!

The paper lacks presentation of the historical data on which the research was based, as well as data related to the natural gas market. What has the demand for natural gas been like over the past few years? As the Authors rightly pointed out, many factors influence the demand for natural gas. I assume that natural gas consumption depends on the season (air temperature), the day of the week (weekdays, holidays – these influence natural gas consumption in industry), i.e., factors that determine seasonality in the natural gas market. Demand and supply in the natural gas market are also influenced by macroeconomic factors (e.g., the development of renewable energy sources such as photovoltaics) and political factors (e.g., Russia's aggression against Ukraine). Presenting the data and explaining their variability will provide a good introduction to the topic for the reader.

The paper lacks a clearly defined objective(s) (preferably in a separate paragraph). Please consider providing specific objectives and/or research hypothesis(es).

In my opinion, in addition to MAE and MAPE, other ex-post errors, such as ME and MPE, should be presented. These errors allow for an assessment of whether the forecasts are not subject to systematic overestimation or underestimation errors. It is also worth presenting RMSE, as this is one of the more easily interpreted ex-post errors.

Please also clarify the forecast horizon for which ex-post errors were calculated. Was it one period (day) in advance?

Detailed comments:

- units are missing in some tables and figures; please provide them;

- figures 13-15 – please adjust the Y-axis range so that the differences between the compared data are visible;

- In my opinion, not all table and figure titles accurately reflect their content, e.g., figures 13-15.

In my opinion, the paper requires changes and additions. The most important comments are presented above.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

Research on Profit Optimization Method of the Natural Gas Industry Chain Based on
Demand Forecasting
Fangkai Shen , Zhaoming Yang ,Zhiwei Zhao, Jianqin Zheng, Yuantao Zhang, Hongying Li a,
and Huai Su
Summary
The manuscript proposes an integrated approach for optimizing the natural gas industry’s
overall profit (considering the supply chain), coupling a machine-learning-based demand
forecasting model with an economic optimization framework. Three forecasting models:
XGBoost, Random Forest, and SVR, are compared, with SVR identified as the most accurate
predictor of short-term natural gas demand. The predicted demand is then incorporated into a
nonlinear optimization model solved using Gurobi, aiming to maximize the overall economic
efficiency of the gas supply chain. The model’s performance is validated using a case study
involving multiple supply sources, storage tanks, LNG terminals, and consumers.
The study is ambitious in integrating prediction and optimization but exhibits several
methodological, presentation, and conceptual issues that need substantial revision before
publication.
Major Points
1. Incomplete Treatment of Robust Optimization and Uncertainty
The introduction promises a “robust optimization” framework, yet no robust or
uncertainty-handling approach is presented later in the paper. There is no quantification
of uncertainty in machine learning predictions (e.g.,confidence intervals), nor any robust
optimization model is solved toevaluate the optimization outcomes. This is a major
conceptual gap between the paper’s stated objectives and its implementation.
2. Chronological and Data Integrity Issues
The paper mentions the “intersection of available data” to create uniform time windows
(June 2022–August 2024). However, the authors do not explain how temporal continuity
is preserved or how potential lag/lead relationships are managed after truncation. This is
especially important when modelling storage units within the supply chain since these
units have a notion of ‘memory’ associated with them.
3. Inaccurate or Incomplete Model Representations
The schematic of XGBoost (Figure 3) is inaccurate, from the figure it appears that
several XGBoost models are combined to obtain the forecasts, however in reality several
weak decision trees are used to obtain accurate forecasts where the inaccuracies of
trees from previous iterations are corrected in later iterations.
4. Incomplete Mass Balance and Storage Treatment
Although the optimization section defines flow constraints, there is no clear mass
balancevalidation across storage or LNG terminals. An additional stock variable has to
keep track of available inventory and demand fulfillment through storage units.
5. Mismatch Between Title and Actual Content: The current title, “Research on Profit
Optimization Method of the Natural Gas Industry Chain Based on Demand Forecasting,”
is overly generic and does not reflect the study’s specific focus or methods. The paper
primarily applies short-term machine-learning-based demand forecasting (using
XGBoost, RF, and SVR) integrated with a deterministic economic optimization
model, rather than developing a new “profit optimization method.” A clearer and more
accurate title would specify both the methodological tools and scope (e.g., “Integration of
Machine Learning Demand Forecasting and Economic Optimization for the Natural Gas
Supply Chain”).
Minor Points
1. Referencing Style and Formatting
The referencing format is inconsistent: missing spaces after brackets and incorrect
referencing style. References should conform to the Energies journal’s MDPI style.
2. Ambiguity Between “Storage” and “LNG Terminal”
The paper does not clearly differentiate the functions of these two components, both are
treated as generic “nodes” with injection and withdrawal flows.
3. Figure Numbering and Caption Errors
Figures 6–8 are missing, and Figures 2–4 are duplicated. Figure captions do not
correspond with the text references, making it difficult to verify model comparison results.
These presentation errors severely limit technical clarity.
4. Language and Typographical Issues
Frequent minor errors appear (e.g., “intersectionof,” “balamnce,” “representationof”),
suggesting a lack of proofreading. Grammatical editing and spacing correction are
necessary for readability and professionalism.
Overall:
While the integration of machine learning and optimization for gas supply-chain profit
maximization is timely and relevant, the current manuscript requires significant methodological
and editorial revisions. At the current stage, I am afraid I would recommend rejection of this
manuscript.

Comments on the Quality of English Language

See review file

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

1. Adjust the article to the requirements of the journal format. In line 45, the number 1 should be written as [1], etc. References to literature in the text should be in square brackets.
2. There is no source given under the figures and tables. Under each table and figure, the following should be written: Source: ........ and the source should be entered here. If it is your own research, it should be written: Source: own research.
3. On page 15 of the article, the figure numbering is incorrect - Figure 2. The correct spelling is Figure 6. 
4. On page 16 of the article, the figure numbering is incorrect. The correct spelling is Figure 7 and Figure 8.
5. Tables 2, 3, 4, 5, and 6 are very poorly described. There are no conclusions.
6. Item 24 in the References does not include the year of publication.

 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

paper's objectives (subsection 1.3) and other corrections have significantly improved the paper's appearance compared to the pre-review period.

Line 421 – why is it stated: "... on three metrics..."? Please explain or correct it.

Some figures (e.g., figures with ex-post errors) are missing a description of the Y-axis units. Figure 8, with the coefficient of determination, in my opinion, should have a maximum value on the Y-axis of 1.

I also suggest paying attention to the Journal's editorial requirements and making corrections, for example, appropriately formatting the source descriptions under the figures.

After corrections, the paper can be published, in my opinion.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

1. The source is still missing, this time - figure number: 9,10,11.
2. Item 24 in the References does not include the year of publication. - still not there. In response to the comments, the authors wrote: "Thanks for the reviewer's kindly comment. In response to the key issues you raised, we suggest using Google Scholar to conduct the search by article title, which will enable you to directly access the relevant articles." I will leave it without comment.

Given the fact that it was not corrected correctly, I believe that the article in this version should not be published.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 3

Reviewer 1 Report

Comments and Suggestions for Authors

I have no further critical comments about the paper. I accept the changes introduced in the manuscript by Authors.

Reviewer 3 Report

Comments and Suggestions for Authors

All the reviewer's repeated comments were taken into account by the authors.