Safety of LNG-Fuelled Cruise Ships in Comparative Risk Assessment

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This manuscript evaluates the safety of LNG-fueled cruise ships by comparing accident data from traditional cruise ships and LNG tankers, using the Formal Safety Assessment (FSA) methodology. The authors conclude that LNG is a safe alternative fuel and that the overall societal risk is within acceptable limits. This research is timely and addresses a critical issue in the maritime industry. However, to ensure the reliability of the findings and the validity of the conclusions, several important methodological and analytical issues must be addressed. Below are essential review comments that require resolution before the manuscript can be considered for publication.

1. The introduction highlights a crucial research gap concerning the absence of operational safety data for LNG-fuelled cruise ships and establishes a general research aim. However, the central research question lacks clear articulation and is somewhat implicit. To improve clarity and better direct the reader, the authors should explicitly express the primary research question in a concise, focused sentence rather than leaving it to be inferred from the discussion on safety equivalence.
2. The description of the Formal Safety Assessment (FSA) methodology in Section 1.2 is rather simplistic. Although the five standard stages of the FSA are outlined, the paper would greatly benefit from a more in-depth explanation of how each stage is specifically implemented in this study. This includes details such as the criteria for hazard identification, the management of limited LNG cruise ship data, and the integration of LNG carrier data into the risk models. Enhancing this section would improve both methodological transparency and reproducibility.
3. The current structure of the paper may lead to some confusion. Although Chapter 4 is titled “Methodology,” it actually presents specific analytical results from the FSA application, making it more suitable for the Results section. Similarly, Chapter 5 mainly contains interpretative content and would be better presented as a standalone “Discussion” section. Reorganizing the chapters to align with the conventional IMRaD structure would enhance clarity and improve the logical flow of the paper.
4. The figures, especially the event trees (Figures 7 and 8) and the F-N curves (Figures 9-12), are overly complex and lack clear labels and annotations, which makes them difficult to interpret. For example, it is challenging to follow the various branches and understand the conditional probabilities in the event trees without more explicit guidance. I would like the authors to redesign the figures to enhance clarity. This could include using color coding, clearer labels, and possibly breaking the event trees into smaller, more manageable sections.
5. The manuscript's conclusion asserts that LNG tankers and LNG cruise ships have demonstrated safety, but this claim is overly definitive, particularly given the limited data available for LNG cruise ships. Although the findings indicate a positive safety record, the small sample size prevents any conclusive statement. The authors should be more cautious in their final remarks and present them as preliminary findings. In other words, they need to moderate the certainty of their conclusions and acknowledge that the results are tentative due to data limitations.

Author Response

1. Comment: The introduction highlights a crucial research gap concerning the absence of operational safety data for LNG-fuelled cruise ships and establishes a general research aim. However, the central research question lacks clear articulation and is somewhat implicit. To improve clarity and better direct the reader, the authors should explicitly express the primary research question in a concise, focused sentence rather than leaving it to be inferred from the discussion on safety equivalence.

Response: Thank you for pointing out the important issue. The research focuses on the hazards of using LNG on large passenger ships. Specifically, we are concerned with LNG and its safety in the cruise industry. The research aims to demonstrate that LNG is safe, if not safer, than current energy sources in the maritime sector. The primary question of this research is: "Is an LNG-fuelled cruise ship safe enough to carry LNG and thousands of people, passengers or crew?" Risk factors will be addressed using the Formal Safety Assessment (FSA) methodology, which is commonly used by the International Maritime Organization (IMO) in preliminary risk assessments. The methodology will include a risk model for accidents which will be developed as a combination of risks for passenger ships and conventional LNG-powered ships (LNG tankers). Although societal risk is very important, no study has been conducted on societal risk, especially for LNG-fuelled cruise ships.

2. Comment: The description of the Formal Safety Assessment (FSA) methodology in Section 1.2 is rather simplistic. Although the five standard stages of the FSA are outlined, the paper would greatly benefit from a more in-depth explanation of how each stage is specifically implemented in this study. This includes details such as the criteria for hazard identification, the management of limited LNG cruise ship data, and the integration of LNG carrier data into the risk models. Enhancing this section would improve both methodological transparency and reproducibility.

Response: Thank you for drawing attention to the need to clarify. The first step will be data processing, the second step a formal safety assessment and the third step a safety evaluation.  As the purpose of this research is to assess and compare the safety of new LNG-fuelled passenger ships with that of conventionally fuelled ships, a formal safety assessment will provide a clear, generally applicable way to assess and compare risks. Accident data for LNG and passenger ships have been collected. Analysis of this data has identified the riskiest areas of exposure for passengers, crew and the shipping company. This analysis will generate event trees for the most common risks, as this is the only way to understand the theoretical outcomes of the different scenarios. These results will then be compared with data from current LNG passenger ship operations to determine whether additional safety measures are needed. Based on previous formal safety assessments, event trees will be generated for contact, collisions, fire/explosion and grounding. This research will define criteria for measuring safety by comparing the risk and reliability of LNG-fuelled passenger ships with conventionally fuelled passenger ships. Using the methods prescribed by the FSA this research will use an integrated approach to assess the overall safety of LNG use on large cruise ships.

3. Comment: The current structure of the paper may lead to some confusion. Although Chapter 4 is titled “Methodology,” it actually presents specific analytical results from the FSA application, making it more suitable for the Results section. Similarly, Chapter 5 mainly contains interpretative content and would be better presented as a standalone “Discussion” section. Reorganizing the chapters to align with the conventional IMRaD structure would enhance clarity and improve the logical flow of the paper.

Response: We agree with the reviewer. Chapter 4 is now shortened, keeping just the methodology explanation, and the results with the event trees interpretation were moved to Chapter 5, Results and Discussion.

4. Comment: The figures, especially the event trees (Figures 7 and 8) and the F-N curves (Figures 9-12), are overly complex and lack clear labels and annotations, which makes them difficult to interpret. For example, it is challenging to follow the various branches and understand the conditional probabilities in the event trees without more explicit guidance. I would like the authors to redesign the figures to enhance clarity. This could include using color coding, clearer labels, and possibly breaking the event trees into smaller, more manageable sections.

Response: We thank the reviewer for this valuable commentary, which has helped us to significantly improve the clarity and readability of the illustrations. In the revised manuscript, we have:

• We have added a new Figure 7, which is a simplified general event tree for the basic contact scenario. This figure focuses on the key escalation paths and is designed as a conceptual framework for risk assessment to improve accessibility and interpretability.
• Two detailed extracts from the event tree (for grounding and collision) have been retained, but with clearer labelling and a more structured layout that highlights the key events in the initial branches. During this process, we also identified and corrected an error in the calculation of the evacuation of passengers and crew.
• Improved the F-N curves (Figures 10–12) by standardising the style of the ALARP boundaries (dotted and dashed lines), clearly labelling the ALARP regions, and improving the visibility and readability of the curves.
• Consistent labelling and annotation in all figures to support comparability and ensure that readers can more easily follow the escalation paths and probability branches.

We believe that these changes have significantly improved the figures and strengthened the overall presentation of the analysis.

5. Comment: The manuscript's conclusion asserts that LNG tankers and LNG cruise ships have demonstrated safety, but this claim is overly definitive, particularly given the limited data available for LNG cruise ships. Although the findings indicate a positive safety record, the small sample size prevents any conclusive statement. The authors should be more cautious in their final remarks and present them as preliminary findings. In other words, they need to moderate the certainty of their conclusions and acknowledge that the results are tentative due to data limitations.

Response: Thank you for pointing that out. This limitation has now been included in the sentence with a note that the results are indicative (Conclusions).                                                                                            

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript addresses an important topic in maritime safety, examining the safety performance of LNG-fueled cruise ships using the IMO's Formal Safety Assessment (FSA) methodology. While the research addresses a relevant industry need, current limitations in event tree parameterization transparency, uncertainty quantification, and data quality handling compromise the credibility of the conclusions and prevent the work from meeting publication standards.

 

Recommendation: Major Revision

 

1. The manuscript reports using S&P Sea-Web data with aggregate statistics (e.g., 537 accidents, 24 LNG cruise ships), but fails to provide specific search and screening protocols. This omission undermines the reproducibility and credibility of the findings.
2. While the manuscript presents accident frequencies (AF), return periods (RP = 1/AF), and historical trends, it lacks confidence intervals or statistical testing. Critical questions remain unanswered: Is the sample size adequate to detect meaningful differences between fuel types?
3. The manuscript repeatedly relies on "historical data or expert estimates" for branching probabilities and fatality percentages without providing complete probability matrices, expert panel composition, or judgment elicitation methods. Event tree outcomes are highly sensitive to these parameters, making their validation essential.
4. The authors acknowledge that LNG cruise ships have operated only since 2018, necessitating the use of LNG carrier data as reference with "modifications." However, the manuscript fails to systematically explain which parameters can be directly extrapolated versus those requiring adjustment, nor does it present sensitivity comparisons demonstrating "what if tanker data were not used."
5. Tables and text frequently show "Not known" or "location unspecified" entries, while the 24-ship LNG cruise sample remains extremely limited. These factors introduce estimation bias, yet the manuscript inadequately describes S&P Sea-Web data limitations or strategies for handling incomplete records.
6. The manuscript presents linear regression lines for accident frequency over time (Figures 2, 4) and makes trend interpretations without reporting R² values, p-values, or formal trend tests.
7. While bunkering is emphasized as a high-risk scenario, it remains unclear whether the model includes independent event trees for port bunkering operations and assessment of impacts on shore-side populations.

Author Response

1. Comment: The manuscript reports using S&P Sea-Web data with aggregate statistics (e.g., 537 accidents, 24 LNG cruise ships) but fails to provide specific search and screening protocols. This omission undermines the reproducibility and credibility of the findings.

Response: Thank you for this comment. The search was conducted by downloading reports on the size of the cruise ship and LNG tanker fleets and related accidents using maritime.ihs.com (S&P Maritime Portal), reviewing articles, IMO's Global Integrated Shipping Information System - (GISIS), European Marine Casualty Information Platform (EMCIP), USCG Marine Casualty & Pollution Data, MHIDAS, and internet news/industry websites (1.2 first paragraph).

 

2. Comment: While the manuscript presents accident frequencies (AF), return periods (RP = 1/AF), and historical trends, it lacks confidence intervals or statistical testing. Critical questions remain unanswered: Is the sample size adequate to detect meaningful differences between fuel types?

Response: We appreciate the reviewer’s critical comment regarding the adequacy of the sample size to detect meaningful differences between fuel types. In the revised manuscript, we explicitly address this issue by analysing the accident frequency normalised by fleet exposure (ship years) and performing a statistical rate comparison between LNG-fuelled and distillate-fuelled cruise ships. Between 2018 and 2024, LNG-powered ships had 71 ship years with three recorded accidents, while distillate-fuelled ships had 1,931 ship years with 88 accidents. The resulting accident frequencies were almost identical (0.042 vs. 0.046 accidents per ship-year), with a rate ratio of 0.93 and a wide 95% confidence interval (0.29–2.93). The p-value was not significant (0.90). These results indicate that the current sample size is insufficient to detect statistically significant differences between fuel types, given the very small number of LNG-related accidents. However, this limitation is unavoidable in the early phase of the introduction of LNG propulsion in cruise shipping. The available data suggest that LNG does not pose a greater accident risk than conventional fuels. However, we recognise that robust statistical confirmation requires a larger LNG fleet and longer observation periods. This has been emphasised in the revised discussion to make the limitations and the need for continuous monitoring transparent to the reader.

New text: To quantify the results, LNG-fuelled cruise ships have completed 71 ship years and experienced three accidents, which corresponds to an accident frequency rate of 0.042 per ship year. In the same period, distillate-fuelled ships completed 1,931 ship years and experienced 88 accidents, which corresponds to an accident frequency rate of 0.046 per ship year. A direct comparison of these rates yields a rate ratio of 0.93 with a wide 95% confidence interval (0.29–2.93) and a non-significant p-value of 0.90. This confirms that there was no statistically significant difference in accident frequency between LNG and distillate-fuelled cruise ships during the observation period. However, the large confidence interval illustrates the limitations of analysing small samples: With only three LNG-related accidents, the statistical power of the comparison is inherently weak, and any apparent differences are subject to considerable uncertainty. This is a typical challenge when assessing the safety performance of new technologies in their early introductory phase, when the exposure time (measured in ship years) is still limited. Nevertheless, the available data show that LNG propulsion does not lead to a higher frequency of accidents during operation than conventional fuels. However, caution is required when interpreting the results: Due to the small sample size, no clear conclusions can be drawn in favour of the higher safety of LNG or regarding possible hidden risks. From a broader perspective, this analysis emphasises the importance of continuous data collection and long-term monitoring. As the LNG-powered cruise fleet grows, statistical certainty will increase and allow a more reliable assessment of relative safety performance. Until then, current evidence suggests that LNG propulsion of cruise ships is at least as safe as conventional fuels, with no evidence of increased accident risk. However, much larger data sets are required for final validation (Chapter 3.3 after Table 4).

 

3. Comment: The manuscript repeatedly relies on "historical data or expert estimates" for branching probabilities and fatality percentages without providing complete probability matrices, expert panel composition, or judgment elicitation methods. Event tree outcomes are highly sensitive to these parameters, making their validation essential.

Response: We agree that the outcome from the event tree is highly sensitive. The approach to assigning probabilities to events in a branch can only be from past data, modeled if possible, or estimated. In comparison with MSC83/21/1 Formal Safety Assessment FSA – Liquefied Natural Gas (LNG) Carriers, the branches for the LNG part have been extended and improved. The probabilities in the first three branches in a tree are obtained from the cruise and LNG ships accident statistics from 1989 until today. The branches related to the LNG release are obtained from the review of FMECA, HAZID, and HAZOP reports of recent LNG ships, and our assessment of them. The fatality branch is mostly estimated but still based on a review of several ship accident reports.

Regarding the probability matrix, we try to do it, but it does not contribute to clarity. We can add the most relevant end states and the complete event tree in the appendix.

 

The expert panel is mentioned in a paper and consists of three authors, three marine officers, and two surveyors.

 

4. Comment: The authors acknowledge that LNG cruise ships have operated only since 2018, necessitating the use of LNG carrier data as reference with "modifications." However, the manuscript fails to systematically explain which parameters can be directly extrapolated versus those requiring adjustment, nor does it present sensitivity comparisons demonstrating "what if tanker data were not used."

Response: We appreciate the reviewer’s observation regarding the limitations of the S&P Sea-Web dataset. In the revised version of the manuscript, we have clarified that although this dataset is widely used for analysing maritime safety, it often contains incomplete entries, such as 'location not specified' or 'not known', particularly in relation to non-serious casualties and health-related incidents. This leads to some bias in the estimate, particularly when accidents are broken down by fuel type or operating context. We explicitly acknowledge that the dataset is incomplete and explain that we have therefore focused our analysis on normalised relative accident frequencies rather than detailed breakdowns of accident locations. In addition, the discussion emphasises that the LNG-fuelled cruise fleet remains extremely small (with only 24 ships in 2024) and that the limited number of recorded LNG accidents (three operational events) does not allow solid statistical conclusions to be drawn. We agree that more comprehensive and standardised reporting is essential and suggest this as a future area of research. This should be accompanied by continuous data collection and expert-based methods, such as the refinement of event trees, to reduce the uncertainties associated with incomplete records.

New text: As LNG-powered cruise ships have only been in operation since 2018, the available safety data is extremely limited. We have therefore relied on data from LNG tankers as a reference basis for the creation of the event tree and the estimation of the accident frequency. Specifically, the type of events in the event tree was taken from the LNG carriers' HAZID process. As the carrying and processing technology is cryogenic on both ships, the probabilities of failures are comparable. The probabilities of LNG leakage, cryogenic pipe damage, and vapour cloud ignition are used by the FSA for LNG tankers, but for some events, they are modified according to the expert team's opinion. Others, such as operational profiles, passenger exposure, and bunkering frequency, had to be adjusted to reflect cruise ship operations (Chapter 4, last paragraph).

 

5. Comment: Tables and text frequently show "Not known" or "location unspecified" entries, while the 24-ship LNG cruise sample remains extremely limited. These factors introduce estimation bias, yet the manuscript inadequately describes S&P Sea-Web data limitations or strategies for handling incomplete records.

Response: We appreciate the reviewer’s observation regarding the limitations of the S&P Sea-Web dataset. In the revised version of the manuscript, we have clarified that although this dataset is widely used for analysing maritime safety, it often contains incomplete entries, such as 'location not specified' or 'not known', particularly in relation to non-serious casualties and health-related incidents. This leads to some bias in the estimate, particularly when accidents are broken down by fuel type or operating context. We explicitly acknowledge that the dataset is incomplete and explain that we have therefore focused our analysis on normalised relative accident frequencies rather than detailed breakdowns of accident locations. In addition, the discussion emphasises that the LNG-fuelled cruise fleet remains extremely small (with only 24 ships in 2024) and that the limited number of recorded LNG accidents (three operational events) does not allow solid statistical conclusions to be drawn. We agree that more comprehensive and standardised reporting is essential and suggest this as a future area of research. This should be accompanied by continuous data collection and expert-based methods, such as the refinement of event trees, to reduce the uncertainties associated with incomplete records.

New text: It should be noted that the S&P Sea-Web database, while widely used, contains incomplete records, with many accidents listed as “location unspecified” or “not known.” This limitation reduces the precision of detailed breakdowns and introduces estimation bias, particularly for LNG-fuelled ships with very few recorded events, which underscores the need for cautious interpretation and for continuous improvement of accident reporting systems (3.3. last paragraph).

 

6. Comment: The manuscript presents linear regression lines for accident frequency over time (Figures 2, 4) and makes trend interpretations without reporting R² values, p-values, or formal trend tests.

Response: We thank the reviewer for his valuable comments on the statistical treatment of accident frequency trends. In the revised manuscript, we have clarified our approach and included the results of the regression analyses together with their statistical parameters. Initial regressions on raw annual accident frequencies for cruise ships did not yield statistically meaningful results, as the data fluctuate widely due to the relatively small fleet size and the limited number of recorded accidents. In order to reduce these fluctuations and obtain a more robust measure of long-term developments, we have introduced an additional indicator: the rate of change in accident frequency. This is defined as the annual change in the average accident frequency per fleet year. The regression analysis of this indicator for cruise ships showed a statistically significant positive slope (R² = 0.84, p < 0.001), indicating a slight but steady increase in recorded accidents from year to year. This could be related to the rapid growth of the fleet and the operational complexity of the ever-increasing size of passenger ships. In contrast, the same analysis for LNG tankers showed a significant negative slope (R² = 0.76, p < 0.001), indicating a long-term decrease in accident frequency. This is most likely due to the earlier introduction of the fleet, the accumulated operational experience and the stricter safety standards. We believe that presenting this adjusted indicator alongside the raw data provides a more meaningful basis for interpretation, as it shows statistically significant differences in accident frequency trends between vessel types, considering the inherent limitations of the dataset.

New text: The calculated accident frequency (Figure 2) per ship year is shown with bars. While short-term fluctuations can be observed, the long-term trend shows a gradual increase in accident frequency. To reduce short-term fluctuations in the data, we have introduced an additional indicator, the so-called accident frequency rate of change, defined as the annual change in the average accident frequency per fleet year. This measure provides a smoothed representation of long-term developments and enables a more reliable trend analysis than the raw annual frequencies. The regression analysis of the rate of change in annual accident frequency for cruise ships shows a statistically significant positive slope (R² = 0.84, p < 0.001), indicating a slight but steady increase in recorded accidents over the observation period. Recently, however, there have been signs of stabilisation, possibly due to the effects of improved ship design, regulatory measures, automation, and risk management systems. Continued monitoring and analysis remain crucial to determine whether the safety measures introduced are effectively mitigating the risks of the growing cruise fleet (Before Figure 2).

 

7. Comment: While bunkering is emphasized as a high-risk scenario, it remains unclear whether the model includes independent event trees for port bunkering operations and assessment of impacts on shore-side populations.

Response: We thank the reviewer for his valuable comment. We agree that the bunkering of ships harbours particular risks, especially with regard to coastal populations. In the present study, we have highlighted bunkering as a high-risk scenario; however, the quantitative modelling was limited to onboard event trees involving passengers and crew. Independent event trees for bunkering in port and an explicit assessment of the impact on the population ashore were not included, mainly due to the lack of validated data on the frequency and consequences of such events. We have now clarified this limitation of scope in the manuscript and added a statement in the discussion recommending that future studies should develop harbour-specific event trees and risk models to capture the impact on shore-side communities. We have also added an additional note from one of the authors referring to the specific assessment of LNG bunkering using highly quantitative methods (CFD in an empirical dispersion model). Based on these calculated consequences and event statistics, we hope to be able to create a credible event tree in the near future.

New text: Although LNG bunkering was identified as a high-risk scenario, the present event tree modelling was confined to shipboard consequences for passengers and crew. Independent event trees for port bunkering operations and the associated impacts on shore-side populations were not developed within this study. This limitation reflects both the scarcity of validated accident frequency data for port operations and the variability of port layouts and safety distances. Future research should explicitly extend the modelling to cover bunkering operations in port environments, including potential domino effects and exposure of shore-side populations, so as to provide a more comprehensive safety assessment. A study has been conducted on LNG bunkering risk in Gerbec et al. [61] (Chapter 4, second-to-last paragraph).

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

This article focuses on the safety of LNG cruise ships. With the Formal Safety Assessment (FSA) methodology recommended by the IMO as its core framework, it systematically conducts research on the safety assessment of LNG-powered cruise ships, aiming to provide a scientific basis for the cruise industry to adopt LNG as an alternative fuel. However, certain parts of the contents should be further optimized to enhance readability and academic depth

（1）After the abbreviation is defined for the first time, the entire text should consistently use that abbreviated form, such as Formal Safety Assessment (FSA) in the manuscript.

（2）Some long sentences contain redundant expressions; it is recommended to split them appropriately to improve readability.

（3）The literature review is insufficiently comprehensive, lacking a systematic analysis of recent studies on alternative fuel safety, cruise ship risk management, and LNG refueling safety. It is recommended to broaden the scope of literature coverage and conduct thorough comparisons with existing research findings. The author may refer to the following references: https://doi.org/10.1016/j.enconman.2025.119641; https://doi.org/10.1016/j.oceaneng.2025.121898

（4）The terms “LNG-fueled cruise ships” and “LNG cruise ships” are used interchangeably in the manuscript. It is recommended to standardize the terminology.

（5）It is recommended to supplement the conclusion with a clear discussion of the research limitations and propose more specific future research directions based on these limitations.

（6）The cost-benefit analysis in the FSA remains at the framework level and lacks specific quantitative assessments.

（7）The conclusion section is overly general. While the study concludes that LNG-fueled cruise ships offer higher safety, it fails to discuss the impact of limitations, such as limited operational data and simplified model assumptions on these findings. It is recommended to address these aspects and propose more specific directions for future research.

Author Response

1. Comment: After the abbreviation is defined for the first time, the entire text should consistently use that abbreviated form, such as Formal Safety Assessment (FSA) in the manuscript.

Response: Thank you for pointing this out. The entire text now consistently uses the abbreviated form, such as FSA, in the manuscript.

 

2. Comment: Some long sentences contain redundant expressions; it is recommended to split them appropriately to improve readability.

Response: We are grateful for the reviewer’s feedback on expressions. The manuscript has now been reviewed and edited by a professional editor who is a native English speaker and well-versed in maritime topics to further tighten sentence structures, standardize terminology, and improve clarity in line with the reviewer’s valuable suggestions.

 

3. Comment: The literature review is insufficiently comprehensive, lacking a systematic analysis of recent studies on alternative fuel safety, cruise ship risk management, and LNG refueling safety. It is recommended to broaden the scope of literature coverage and conduct thorough comparisons with existing research findings. The author may refer to the following references:

https://doi.org/10.1016/j.enconman.2025.119641; https://doi.org/10.1016/j.oceaneng.2025.121898

 

Response: We thank the reviewer for pointing out the inadequacies of the original literature review. In the revised version, we have significantly improved this section by going beyond a simple listing of references. Each cited paper is now accompanied by a brief description of its main contribution in relation to the topics of alternative fuel safety, cruise ship risk management and LNG bunkering safety. In addition, the proposed sources have been included and critically reviewed, and their relevance to our study has been emphasised. This broader and more systematic literature coverage strengthens the contextual basis of our research and ensures that our findings are discussed in the light of the latest developments in the field (Chapter 1.1. Literature review, last paragraph).

 

4. Comment: The terms “LNG-fueled cruise ships” and “LNG cruise ships” are used interchangeably in the manuscript. It is recommended to standardize the terminology.

Response: Thank you for this valuable suggestion. This is now standardized throughout the manuscript.

 

5. Comment: It is recommended to supplement the conclusion with a clear discussion of the research limitations and propose more specific future research directions based on these limitations.

Response: We note the reviewer’s comment that the conclusion should be strengthened by emphasising the limitations of the study and possible avenues for future research. In the revised manuscript, the conclusion has been expanded to explicitly recognise the main limitation of the study: the extremely low number of reported accidents involving LNG-fuelled cruise ships since 2018. This limits the statistical power of the comparative analyses. On this basis, we suggest that future research should focus on continued data collection and systematic monitoring of LNG-fuelled cruise ship operations. As the fleet grows, more robust statistical analyses will be possible. In addition, refinement of the event tree methodology with the support of expert panels could enable a deeper understanding of accident scenarios and consequences. These steps will ensure that the study not only assesses the current situation but also provides the basis for improved long-term safety assessments.

New text: (revised Conclusions).

6. Comment: The cost-benefit analysis in the FSA remains at the framework level and lacks specific quantitative assessments.

Response: We appreciate the reviewer’s comment. In this paper, cost-benefit analysis (CBA) was indeed treated at the framework level in accordance with the IMO’s FSA structure. However, due to the very limited operational and economic data available for LNG-fuelled cruise ships, a full quantitative cost-benefit analysis of risk control measures could not be developed at this stage. Instead, the focus of our analysis was on risk identification, event tree modelling and societal/individual risk criteria, which form the necessary basis for future CBAs.

New text: Although the framework for the formal safety assessment (FSA) includes the cost-benefit analysis (CBA) as one of the most important steps, it was not addressed at this stage of the study. The limited availability of reliable economic and operational data for LNG cruise ships prevents a meaningful monetisation of risk control options at this stage. Instead, emphasis has been placed on risk identification, event tree modelling and the assessment of risk acceptance criteria, which will form the basis for future quantitative CBAs once more empirical data on the costs and benefits of LNG-specific mitigation measures will be available (Methodology).

7. Comment: The conclusion section is overly general. While the study concludes that LNG-fueled cruise ships offer higher safety, it fails to discuss the impact of limitations, such as limited operational data and simplified model assumptions on these findings. It is recommended to address these aspects and propose more specific directions for future research.

Response: We thank the reviewer for this constructive comment. We agree that the conclusions were too general. In the revised version, we now explicitly point out the limitations of our study, including the limited operational data and the simplified assumptions used in the modelling of the event tree. We have also clarified that these factors may affect the accuracy of the results, and we suggest specific directions for future research, such as the collection of long-term operational and incident data for LNG cruise ships, the refinement of the probabilistic modelling framework, and the development of a full quantitative cost-benefit analysis of mitigation measures. These additions are intended to provide a clearer outlook for future work.

New text: To emphasise the safety potential of LNG cruise ships, it is important to recognise the limitations of the present study. The analysis relied on cruise ship and LNG cruise ship accidents statistical data, limited operational data, and several simplifying assumptions in the modelling of event trees, which may affect the accuracy of the quantified risk levels. These limitations mean that the results are indicative rather than definitive. Future research should therefore focus on collecting long-term operational and incident data from LNG-fuelled cruise ships, refining the probabilistic models with more detailed failure and human reliability data, and extending the FSA to include a fully quantitative cost-benefit analysis of LNG-specific risk control options. These efforts will enable more robust validation of current results and support evidence-based regulatory decisions (Conclusions, last paragraph).

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Thank you for your thorough revisions. I have carefully considered your insightful feedback, and I believe the manuscript has been greatly improved as a result. With these enhancements, it is now in a stronger position for publication.

Reviewer 2 Report

Comments and Suggestions for Authors

Thank you for your thorough response to the previous round of reviewer comments and for the careful revisions made to the manuscript.

The overall quality and readability of the manuscript have been significantly improved. I believe this revised manuscript now meets the publication standards of the Journal of Marine Science and Engineering (JMSE).

This research provides a valuable risk assessment of LNG-fueled cruise ship safety and makes a positive contribution to the field of maritime safety.

I have no further revision suggestions for the manuscript. 

Reviewer 3 Report

Comments and Suggestions for Authors

The authors have revised the paper according to my suggestions.