Next Article in Journal
Improving Corporate Environmental Performance Through Big Data Analytics Implementation: The Role of Industry Environment
Previous Article in Journal
Depopulation and the Development of Peri-Urban Green Areas of Large Cities: Lessons Learned from Romania
 
 
Article
Peer-Review Record

Machine Learning and Simulation for Efficiency and Sustainability in Container Terminals

Sustainability 2025, 17(7), 2927; https://doi.org/10.3390/su17072927
by Abderaouf Benghalia 1,*, Amani Ferdjallah 1, Mustapha Oudani 2 and Jaouad Boukachour 3
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Reviewer 3: Anonymous
Reviewer 4: Anonymous
Sustainability 2025, 17(7), 2927; https://doi.org/10.3390/su17072927
Submission received: 7 January 2025 / Revised: 25 February 2025 / Accepted: 26 February 2025 / Published: 26 March 2025
(This article belongs to the Section Sustainable Transportation)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

1. The manuscript looks like a project report. It's suggested to focus on methods or models.

2. It will be better if there is a schematic diagram to illustrate the relationship between the ML algorithm and the simulation.

3. Please specify the input/output parameters of the ML algorithm and provide a comparison between the calculated results and actual data.

4. It's suggested to explain the meanings of each parameter in Figure 4.

5. Figures in the manuscript are not very clear.

Author Response

Comments 1: The manuscript looks like a project report. It's suggested to focus on methods or models.

 

Thank you for pointing this out. We have significantly improved the manuscript based on the reviewers' comments. We started by enhancing the abstract to emphasize our methodological contributions and provide a stronger focus on methods and models. This includes reinforcing the connection to machine learning (ML) models and their specific application while highlighting how our approach stands out from existing studies:

"This study proposes a novel integration of machine learning and simulation, demonstrating its effectiveness in optimizing ship turnaround times and reducing carbon emissions."

In the general introduction, we now specify:

"In this work, we introduce a hybrid approach combining machine learning for predictive analysis and discrete-event simulation to optimize port operations. This approach allows the identification of key factors that can reduce handling costs and shorten the turnaround times of container ships."

In Section 2: Literature Review, we provide the following:

"The management of COâ‚‚ emissions in port terminals has become a central focus in achieving sustainability objectives, particularly as global trade continues to grow. The studies reviewed demonstrate that machine learning (ML) techniques are increasingly applied to tackle complex decision-making problems in port operations. ML algorithms excel in extracting actionable insights from vast amounts of data, enabling the prediction of critical variables such as ship turnaround times, container flow, and equipment utilization. These capabilities empower decision-makers to optimize resource allocation, reduce operational inefficiencies, and ultimately minimize carbon footprints.

However, while ML offers robust predictive capabilities, its integration with simulation techniques is essential to fully understand and enhance the performance of port systems. Simulation models complement ML by providing a dynamic framework for testing and validating operational scenarios under various constraints and uncertainties. This combination enables a holistic evaluation of strategies to balance efficiency with sustainability." In the general conclusion, we highlight the impact of our approach:

"Our integrated ML-simulation framework demonstrates a clear advantage in optimizing terminal operations, both in terms of reducing ship turnaround times and achieving sustainability goals." This improved version of the manuscript reflects a more methodologically focused and cohesive narrative, addressing the reviewers' valuable feedback.

 

Comments 2: It will be better if there is a schematic diagram to illustrate the relationship between the ML algorithm and the simulation.

Response 2: Agree. To address this comment, we have improved Figure 1 and provided a detailed description of the relationship between the ML algorithm and the simulation in Section 3.

 

Comments 3: Please specify the input/output parameters of the ML algorithm and provide a comparison between the calculated results and actual data..

 

Response 3: Agree. Agree.Thank you for pointing this out.

The input parameters of the algorithm include ship characteristics (ID, container type, dates, port calls), operational data (handling equipment used), and port characteristics (availability of berths, depth). As output, the model predicts the ship’s dwell time, enabling the determination of when berths will be freed. The comparison between the predicted and actual data is conducted using accuracy and the F1-score metrics to assess the model’s precision and reliability in an operational context. For a more detailed evaluation, the MAE (Mean Absolute Error) and RMSE (Root Mean Squared Error) metrics were added. The MAE, with a value of 0.055, indicates an average error of 1 hour and 20 minutes in predicting dwell times, which is suitable for operational planning. The RMSE of 0.14 suggests that larger errors (approximately 3 hours and 20 minutes) may occur, likely in specific cases. This indicates that the model is reliable for the majority of scenarios. These results, combined with consistent performance across the training and testing datasets, demonstrate the absence of overfitting and confirm the robustness of the model under operational conditions.

 

 

Comments 4: It's suggested to explain the meanings of each parameter in Figure 4.

 

Response 4: We conducted a correlation analysis to better understand the relationships between the variables (Figure 4). In the matrix, the colors represent the intensity of the correlations: light beige indicates strong correlations, orange denotes moderate correlations, and black represents weak correlations. Each parameter corresponds to a variable in the dataset, such as port calls, arrival and departure dates, dwell time, container types (e.g., refriger-ated, sensitive), the number of 20TEUs/40TEUs (Twenty-foot Equivalent Units) unloaded by ship handling equipment. The values in the matrix measure the relationship between pairs of variables: a positive correlation (+1) means both variables increase together, a negative correlation (-1) indicates that as one variable increases, the other decreases, and a correlation close to 0 suggests no significant relationship. This analysis allowed us to identify key associations that are crucial for the model's performance. (Page 10)

 

Comments 5: Figures in the manuscript are not very clear.

Response 5: Thank you for pointing this out. We agree with this comment. We have improved the clarity of the figures throughout the manuscript.

 

 

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

The article is devoted to the description of the research conducted by the authors to improve the efficiency of the seaport, using the example of the port of Algeria, including in the context of sustainability.

The research topic is clearly relevant both in terms of potential economic efficiency of the proposed solutions and in terms of their compliance with the current trend to reduce emissions of CO2 and harmful substances into the atmosphere. The latter also determines the conformity of this paper to the journal's subject matter.

The article contains a sufficiently qualitative review of the current literature on topics related to the subject of the research.

The authors propose an original approach based on the combined application of machine learning and subsequent simulation modelling of port container terminal operation scenarios, in order to detect problem areas in the current operation process and to propose remedies for these problems.

The research materials are generally described in a reasonably good quality, except for some shortcomings as outlined below .

I believe that the work is relevant, may be of interest to the readers of the journal, and can be published in it after eliminating a number of shortcomings.

 

My remarks on the content and layout of the material in the article:

1. There are unfinished phrases in the text, namely, "(e.g., ...)" on lines 147 and 160

 

2. It is recommended to make a better Figure 2, which will be clearer when reading the text. If the text says north and south zone, it's probably best to orient the shot in a north-south direction.

 

3. The results of the work will look more serious if in Fig. 3 shows all the indicators considered, not just one of them. In addition, it should be specified exactly what each indicator is measured in (i.e. sign the axes).

 

4. Neural networks are mentioned in the text and in Table 1, but nowhere is the network architecture used in this study specified.

 

5. It is necessary to describe in more detail the machine learning problem that was solved in the paper. As it follows from the text, it is a classification type task, and the turnaround times was predicted. But above in the text it is stated that berth occupancy was predicted.

 

6. The abstract says ‘predictive accuracy of 0.9991’, but according to Table 1, this is precision, not accuracy 

 

7. In the text of the article, the benefits of the study for sustainable development and CO2 emission reduction could have been mentioned less frequently. The authors do it too often, probably to better show the connection between the study and the journal's subject matter.

 

8. On the other hand, if the authors are so insistent on reducing CO2 emissions, it would be logical to provide at least approximate quantitative indicators related to this. This would make the study more relevant to the journal's subject matter.

 

9. The paper does not sufficiently describe the relationship between the results obtained from machine learning and subsequent simulation. It should be written more clearly which values obtained through ML are used for simulation.

 

Author Response

 

Comments 1: There are unfinished phrases in the text, namely, "(e.g., ...)" on lines 147 and 160

Response 1: Thank you for your feedback. We agree. We have made the necessary corrections and improvements on page 4.

 

Comments 2: It is recommended to make a better Figure 2, which will be clearer when reading the text. If the text says north and south zone, it's probably best to orient the shot in a north-south direction.

Response 2: Agree. The figure has been improved.

 

Comments 3: The results of the work will look more serious if in Fig. 3 shows all the indicators considered, not just one of them. In addition, it should be specified exactly what each indicator is measured in (i.e. sign the axes).

Response 3:

The database contains detailed information on 4,171 vessels that docked between 2017 and 2022, structured across 16 variables. These indicators encompass a wide range of operational and logistical details, such as:

·         Port calls (number and nature of operations for each vessel),

·         Vessel details (size, type, and capacity),

·         Arrival and departure dates (timestamps),

·         Duration of stay (in days),

·         Berthing dock (specific quay assignment),

·         Container type (standard, oversized, etc.),

·         Cargo sensitivity (e.g., hazardous materials),

·         Presence of refrigerated containers (critical for certain cargo types).

In the data preprocessing phase of our study, we focus specifically on the "Stay" criterion, which represents the time spent by ships at the port. This criterion is calculated based on the ships' arrival and departure dates and serves as a key indicator for assessing and optimizing the operational performance of terminals. The predictions made with machine learning (ML) algorithms are related to this criterion, allowing for more accurate estimation of ships' stay times. We performed thorough preprocessing on the entire dataset and its various indicators. However, in order to avoid cluttering the article, we chose not to display all the box plot figures. Details regarding this preprocessing and statistical analyses can be found on pages 8 and 9 of the article.

- Concerning the accuracy of the axes, we have made a modification in Figure 3 and in the corresponding text.

 

Comments 4: Neural networks are mentioned in the text and in Table 1, but nowhere is the network architecture used in this study specified.

Response 4: Thank you for your comment. We have addressed this point and clarified the network architecture used in the manuscript as follows: We implemented several classification models, including Decision Tree, K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Random Forest, and Artificial Neural Networks (ANN). The architecture of the implemented neural network is designed to optimize both convergence and performance. The network consists of a single hidden layer with 200 neurons, specified by the parameter hidden_layer_sizes=(200). The activation function used is ReLU (Rectified Linear Unit), chosen for its efficiency in modeling non-linearities. The optimization algorithm used is Adam, known for its speed and stability. The maximum number of training iterations is set to 300 to maintain a good balance between accuracy and computation time. Additionally, a stopping criterion based on internal tolerance has been integrated to halt training if convergence is reached before the iteration limit is met.

 

 

Comments 5: It is necessary to describe in more detail the machine learning problem that was solved in the paper. As it follows from the text, it is a classification type task, and the turnaround times was predicted. But above in the text it is stated that berth occupancy was predicted.

Response 5:

Thank you for your constructive feedback. Through the various comments received, we have enriched our article by providing more details on the machine learning problem we addressed. We have clarified that the objective of our study is to predict ship dwell time, which includes both waiting time and time spent at the berths. Thus, while the comment refers to a classification problem, our actual goal is to estimate the total dwell time of ships, encompassing both of these components. We have updated the article to better reflect this clarification.

 

Comments 6: The abstract says ‘predictive accuracy of 0.9991’, but according to Table 1, this is precision, not accuracy.

Response 6: thanks for this note. We have replaced it in the summary.

 

Comments 7: In the text of the article, the benefits of the study for sustainable development and CO2 emission reduction could have been mentioned less frequently. The authors do it too often, probably to better show the connection between the study and the journal's subject matter.

Response 7: Indeed, we have taken your comment into consideration and have revised our manuscript.

 

 

Comments 8: On the other hand, if the authors are so insistent on reducing CO2 emissions, it would be logical to provide at least approximate quantitative indicators related to this. This would make the study more relevant to the journal's subject matter.

Response 8:

We thank you for this relevant comment. Improvements and clarifications have been made to the analysis of the results on page 20 to make the study more relevant.

 

Comments 9: The paper does not sufficiently describe the relationship between the results obtained from machine learning and subsequent simulation. It should be written more clearly which values obtained through ML are used for simulation.

Response 9: Thank you for your comment. We have provided further details regarding the nature of the data obtained through prediction, which feed into our simulation system. See pages 7 and 13. The ML models predict ship turnaround times, which include waiting times and handling times at the quay. These predicted values, covering a full year of operation, are used as input data in the simulation system. Additionally, we have modeled the key entities of the terminal, such as ships, quay cranes, trucks, and storage areas, with particular focus on quay cranes due to their high operating costs and significant contribution to COâ‚‚ emissions. By integrating the ship turnaround times predicted by the AI into the simulation, we can assess the terminal's capacity to handle varying traffic flows and optimize both operational efficiency and emissions reduction strategies.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

1.       The in-text citations have to be modified into correct formats to meet academic requirements.

For example, line 154 “the authors of [19] proposed an Artificial Neural Network model, considering crane productivity, number of cranes, and operating delays. [20] outlined an ensemble methodological……”.

Line 167 “the authors of [22] created a hyper-heuristic model utilizing deep reinforcement learning”.

 

2.       There are a lot of typos in the manuscript, please check through and correct them.

For example, line 160 “…where temporal-related properties (e.g., ...) of vessels are considered…”

Line 521 “…port efficiency and CO2 emissions…”, please mind the lower case letter.

3.       In the current manuscript, little supporting data were found in the introduction and literature review section. For example, line 32 “Optimizing these processes not only boosts port productivity but also supports sustainability by lowering fuel consumption, cutting equipment energy use, and reducing carbon emissions.” What articles proved these viewpoint? Are there any available quantitative evidences to support these ideas? Please refer to more good-quality references.

4.       Please improve the resolution of the graphs.

Comments for author File: Comments.pdf

Author Response

Comments 1: The in-text citations have to be modified into correct formats to meet academic requirements. For example, line 154 “the authors of [19] proposed an Artificial Neural Network model, considering crane productivity, number of cranes, and operating delays. [20] outlined an ensemble methodological……”. Line 167 “the authors of [22] created a hyper-heuristic model utilizing deep reinforcement learning”.

Response 1: We sincerely thank you for your constructive comment. All necessary modifications have been made according to the citation and formatting guidelines.

 

Comments 2: There are a lot of typos in the manuscript, please check through and correct them. For example, line 160 “…where temporal-related properties (e.g., ...) of vessels are considered…” Line 521 “…port efficiency and CO2 emissions…”, please mind the lower case letter.

Response 2: Thank you for your comments. We have taken all your observations into account and have corrected the typos as well as the points you mentioned.  

 

 

Comments 3:  In the current manuscript, little supporting data were found in the introduction and literature review section. For example, line 32 “Optimizing these processes not only boosts port productivity but also supports sustainability by lowering fuel consumption, cutting equipment energy use, and reducing carbon emissions.” What articles proved these viewpoint? Are there any available quantitative evidences to support these ideas? Please refer to more good-quality references.

Response 3: Thank you for your feedback. We have taken your suggestions into account and made substantial improvements to the writing, particularly by enriching the relevant section with pertinent references.

 

 

Comments 4: Please improve the resolution of the graphs.

Response 4: Agree, we have worked on improving the quality of the figures in the new version of the article.

 

 

 

Author Response File: Author Response.docx

Reviewer 4 Report

Comments and Suggestions for Authors

The study presents an innovative integration of machine learning (ML) and discrete-event simulation to enhance the efficiency and sustainability of container terminal operations, with a particular focus on reducing ship turnaround times and carbon emissions. The empirical validation using data from the Algiers Port Container Terminal is commendable, and the high predictive accuracy achieved (0.9991) is noteworthy. However, there are several areas where the manuscript could be improved to strengthen its contribution to the field:

 

  1. The manuscript would benefit from a more detailed explanation of the specific ML algorithms used, their parameter settings, and the rationale behind their selection. 
  2. The simulation model's assumptions and constraints should be more explicitly stated to ensure reproducibility.
  3. The current figures' resolution is rather low. Please fix this issue. 
  4. The manuscript seems to include too many short sections, which can disrupt the flow of ideas. It is recommended to simplify the structure by merging some of the sections to ensure a more coherent and subject-wise flow.
  5. The manuscript does not provide a comprehensive comparison of the proposed model with the state-of-art models. 
  6. The authors should attention to minor grammatical errors and formatting inconsistencies throughout the manuscript.
  7. The study would be enhanced by a more thorough comparison with existing approaches in the literature.
  8. The dataset used for training and testing the ML models is briefly mentioned, but a more comprehensive description, including data preprocessing steps, feature selection, and potential biases, would be valuable.
  9. The authors should check the abbreviation in “The Container Division of the Algiers Port Authority (EPAL)” to ensure consistency and clarity. 
  10. Please change the term "accuracy" to "precision" in the abstract based on the results of Table 1. The sentence should be, “Port Container Terminal, achieving an exceptionally high predictive precision of 0.9991.”

Author Response

Comments 1: The manuscript would benefit from a more detailed explanation of the specific ML algorithms used, their parameter settings, and the rationale behind their selection.

Response 1: Thank you for your valuable comment regarding the need for a detailed explanation of the machine learning algorithms and their configurations. In response, we have added more details to our manuscript, particularly on pages 7 and 8, to provide a comprehensive description of the algorithms used, their parameter settings, and the rationale behind their selection.

 

Comments 2: The simulation model's assumptions and constraints should be more explicitly stated to ensure reproducibility.

Response 2: Thank you for highlighting the need to further clarify the assumptions and constraints of the simulation model. In response, we have added a description in Section 4: Simulation and Performance Evaluation, on page 13.

 

Comments 3: The current figures' resolution is rather low. Please fix this issue.

Response 3: Indeed, we have worked on improving the quality of the figures to ensure optimal resolution.

 

Comments 4: The manuscript seems to include too many short sections, which can disrupt the flow of ideas. It is recommended to simplify the structure by merging some of the sections to ensure a more coherent and subject-wise flow.

Response 4: In the new version of the article, we have revised the overall structure, taking into account all the comments from the different reviewers.

 

Comments 5: The manuscript does not provide a comprehensive comparison of the proposed model with the state-of-art models.

Response 5: We thank you for your remark regarding the need to clarify our contribution in relation to the state of the art. In response, we have enhanced the description in the article, particularly on page 6, to better outline our positioning with respect to existing works and to specify our contribution, which is aligned with operational and strategic challenges while addressing sustainability requirements, particularly the reduction of CO2 emissions.

 

Comments 6: The authors should attention to minor grammatical errors and formatting inconsistencies throughout the manuscript.

Response 6: We thank you for your remark regarding the grammatical errors and formatting inconsistencies in the manuscript. In response, we conducted a thorough review of the entire article to improve its quality. We have also taken into account all the reviewers' comments on this aspect to ensure a clear and consistent presentation of the content.

 

Comments 7: The study would be enhanced by a more thorough comparison with existing approaches in the literature.

Response 7: We thank you for your suggestion to strengthen the study by including a comparison with existing approaches in the literature. In response, we have added a comparative analysis to the article, incorporating the various remarks received, including an evaluation of state-of-the-art approaches relevant to our field of study.

 

Comments 8: The study would be enhanced by a more thorough comparison with existing approaches in the literature.

Response 8: We acknowledge the importance of your remark and have made several improvements in response. We have provided a more detailed explanation of the dataset used, including the preprocessing steps. Additionally, we have clarified the parameters and elaborated on the validation of the various steps carried out in our methodology.

 

Comments 9: The authors should check the abbreviation in “The Container Division of the Algiers Port Authority (EPAL)” to ensure consistency and clarity.

Response 9: We thank you for this observation. We have reviewed the use of the abbreviation EPAL throughout the manuscript.

 

Comments 10: Please change the term "accuracy" to "precision" in the abstract based on the results of Table 1. The sentence should be, “Port Container Terminal, achieving an exceptionally high predictive precision of 0.9991.”  

Response 10: Thank you for pointing out this detail. We have corrected this error in the abstract.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript looks like a project report. It's suggested to focus on methods or models.

It's suggested to explain the meanings of each parameter in Figure 4.

Author Response

Comments 1: The manuscript looks like a project report. It's suggested to focus on methods or models.

We thank you for your feedback and constructive comments, which have allowed us to improve our manuscript. We have taken your suggestions into account and made several significant improvements to our article. In particular, we have strengthened the orientation of our work towards a methods and models-based approach, by integrating a thorough and justified state of the art.

We appreciate your comments, which have contributed to enhancing the scientific quality of our work. However, if a specific point requires further revision, we would be happy to make adjustments based on your clarifications.

 

Comments 2: It's suggested to explain the meanings of each parameter in Figure 4.

Response 2: Agree. Thank you for your feedback. We have now added explanations for each parameter in Figure 4 to improve clarity.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Authors have edited the manuscript according to the comments and suggestions. However, there are still many points to be improved.

  1.  There are still a few typos in the manuscript, please check through and correct them. For example, line 31 “…can also contribute to the reduction of CO2 emissions..…”, please mind the lower case letter.
  2. Please cite proper references in the introduction section. Few references is found in three  paragraphs from line 44-line 77, which makes the statements not convincing.
  3. The English could be improved to more clearly express the research.
Comments on the Quality of English Language

 The English could be improved to more clearly express the research.

Author Response

Comments 1: There are still a few typos in the manuscript, please check through and correct them. For example, line 31 “…can also contribute to the reduction of CO2 emissions..…”, please mind the lower case letter.

Response 1: Thank you for your careful review. We have revised the manuscript completely based on this comment and the errors have been corrected.

 

Comments 2: Please cite proper references in the introduction section. Few references is found in three  paragraphs from line 44-line 77, which makes the statements not convincing.

Response 2: Thank you for your suggestion. We have carefully revised the introduction section and added appropriate references to better support our statements.

 

 

Comments 3: The English could be improved to more clearly express the research.

Response 3: Thank you for your feedback. Significant improvements have been made to the writing throughout the manuscript. However, if you have specific suggestions regarding certain parts of the article, we would be happy to refine them further.

 

Author Response File: Author Response.pdf

Round 3

Reviewer 1 Report

Comments and Suggestions for Authors

Agree

Author Response

Comments 1: Agree.

Response 1: We would like to sincerely thank you for your valuable feedback and for accepting our previous responses regarding our manuscript. Your comments have been highly beneficial in improving the quality of our work, and we truly appreciate your time and effort in reviewing our submission.

Author Response File: Author Response.pdf

Back to TopTop