From ice hazards to system failures: analysis and prevention of catastrophic icebreaker escort accidents on the Northern Sea Route

1 Introduction

One of the most important milestones in the Russian Federation’s (RF) Arctic energy strategy was the commissioning of the first production line at the Yamal LNG facility on the Yamal Peninsula in December 2017. A total of three main production lines were commissioned in 2018, with the fourth Train-4¹ coming online in 2021, bringing the facility’s LNG production capacity to approximately 17.4 million tons (Mt) per year. The commissioning of Yamal LNG has not only increased Russia’s LNG export capacity but has also enabled regular LNG transportation along the NSR. Approximately 15 Arc7-class ice-strengthened LNG carriers built as part of the project are capable of year-round navigation, either with icebreaker support or independently in suitable ice conditions (Wan, 2025). This has transformed Sabetta Port into a key energy logistics hub along the NSR² (Figure 1). This development has directly impacted the total cargo volume on the NSR, and the cargo volume, which was approximately 7.5 Mt in 2016, has increased roughly fivefold between 2016 and 2024 as Yamal LNG approaches full capacity. LNG transportation has been the primary driver of this growth. Since Yamal LNG came online, the majority of cargo transported on the NSR has consisted of LNG. In 2024, LNG transportation reached a record high of 21.86 Mt, accounting for approximately 58% of the total NSR cargo (iMarine, 2025). Therefore, today the economic sustainability of NSR largely depends on Yamal LNG production.

On the other hand, attacks on LNG infrastructure in the Gulf region and emerging security risks in the Strait of Hormuz have made the geopolitical fragility of the global LNG supply chain more visible. These developments may lead major LNG importers, primarily China, to diversify not only their suppliers but also their transportation routes. In this context, RF Arctic LNG projects and the NSR could gain greater strategic importance as an alternative corridor to support energy supply security, especially after the Houthis declared a blockade in the Red Sea as of 5 August 2026 (The New York Times, 2026). If shipping companies perceive the route as unsafe or insurance premiums increase, they will seek alternative routes. In such a geopolitical environment, increased international demand for RF LNG would incentivise increased exports, particularly from Yamal LNG and, in the event of a future easing of sanctions, from Arctic LNG projects. This potential increase in export volume would naturally lead to an increase in the number of LNG transportation operations via the NSR.

However, this growth is not solely dependent on increased demand. Arctic LNG transportation is based on a specialised logistics system dependent on high-ice-class Arc7 LNG tankers and Russia’s nuclear icebreaker fleet³. The fact that even Arc7-class vessels may require icebreaker escort in certain sections, especially during winter months and in heavy ice conditions⁴, indicates that the increased number of voyages will also increase the demand for icebreaker support services. Therefore, the main reason for the increased need for escorts is not a shortage of Arc7 tankers, but rather the increased operational intensity on the NSR.

Moreover, delays in the construction of Arc7-class LNG tankers due to Western sanctions imposed on RF are limiting the addition of new vessels to the fleet and logistically restricting Russia’s Arctic LNG export capacity. Therefore, any future increase in demand can only translate into actual LNG transportation to the extent that it is compatible with existing transport capacity and icebreaker support capabilities. In other words, geopolitical risks originating from the Gulf increase the strategic importance of the NSR, and therefore, the development of Arctic resources requires a clearer understanding of the possibilities for using its transport corridors.

Although maritime accident research has extensively examined accident investigation and causal analysis, much of the existing literature remains focused on identifying the immediate and contributing causes of individual accidents. Such approaches provide important insights into how accidents occur, but they may provide limited understanding of the broader system conditions that allow multiple failures to align within complex maritime operations. In particular, greater attention is needed to the interaction between operational practices, human decision-making, organizational conditions, and environmental stressors rather than examining these factors in isolation. Therefore, there is a need for approaches that move beyond the identification of immediate accident causes and examine how latent organizational conditions, operational failures, human factors, and environmental stressors interact across the maritime safety system. Greater attention is needed to the interaction between operational practices, human decision-making, organizational conditions, and environmental stressors rather than examining these factors in isolation. This study aims to fill this gap.

Against this background, this study addresses the following research questions:

• RQ1: Which publicly identifiable catastrophic events associated with icebreaker-supported escort operations within the statutory boundaries of the NSR can be identified from the available documentary sources?
• RQ2: How can the documented organizational, operational, human, technical, and environmental factors associated with the Baikal–Vladivostok accident be interpreted through the Swiss Cheese Model, and what preventive and safety-governance strategies can be derived from this analysis?

To address the first question, open sources are examined thoroughly; for the second one, the Baikal–Vladivostok accident is used as an instrumental case, and the Swiss Cheese Model (SCM) is used to organise and interpret the documented interaction of latent organizational conditions, operational failures, human decision-making, and environmental stressors. Based on this analysis, the study develops prevention strategies focusing on risk-based escort planning, real-time decision support, human and organizational measures, and regulatory governance.

The remainder of this article is organised as follows. In Section 2, the background of icebreaker escort operations is presented. Section 3 gives a brief introduction to the traditional Arctic hazards and emerging operational risks. Section 4 presents materials and methods; Section 5 analyses the Baikal–Vladivostok accident as a case study; Section 6 presents the results and prevention strategies; and the study concludes with a conclusion section.

2 Background: icebreaker escort operations and emerging Arctic risks

With the increase in maritime activities in the Arctic region, ensuring operational safety has become vital for economic development (Fedi et al., 2020). A review of the existing literature reveals that risk analyses for Arctic maritime transport mostly focus on technical and physical parameters such as ice conditions, ship maneuverability, and collision probabilities (Khan et al., 2019). However, studies on human factors, organizational deficiencies, and governance mechanisms that lead to the failure of escort operations remain quite limited (Fedi et al., 2018).

In regions with challenging ice conditions, such as the NSR, icebreaker support is essential for vessels unable to navigate independently (Zhang et al., 2019). In the literature, icebreaker-supported operations are broadly categorised into four main areas: leading escort, convoy escort, breaking ship loose, and towing (Xu et al., 2022). Convoy operations consist of multiple vessels following a channel opened by an icebreaker and are the most common method preferred to enhance operational efficiency, especially in areas with increasing traffic density (Xu et al., 2022; Zhang et al., 2019). The most critical parameters in this process are a safe following distance to prevent collisions in case of sudden deceleration of the leading vessel and a safe speed at which ice loads do not damage the hull (Zhang et al., 2018). Studies show that as ice thickness increases, the distance between vessels shortens and speed decreases (Goerlandt et al., 2017).

Traditional literature attributes the main risks affecting Arctic navigation to physical environmental factors. Chief amongst these is besetting in ice, which occurs particularly because of ice compression (Fu et al., 2022). Additionally, low visibility, extremely low temperatures, engine failures, and hull damage due to ice loads are the most frequently reported events (Khan et al., 2014). Yan et al. (2026), in their study of incidents on the NSR, found that besetting in ice is particularly concentrated in the autumn season and in regions such as the East Siberian Sea (Yan et al., 2026). These physical hazards lead not only to operational delays but also to serious structural damage and environmental pollution risks (Goerlandt et al., 2017).

Particularly with the commissioning of the Yamal LNG project and the increase in energy-related destination transportation, the character of Arctic operations has changed (Moe, 2020). Today, the NSR has transformed from merely a transit route into an active industrial corridor with heavy traffic, complex maneuvers around LNG terminals, and the management of multiple vessel convoys (Gladkov, 2025). The risks emerging in this new era are socio-technical system crises rather than purely ice-navigation problems. New approaches in the literature emphasise that accidents stem not only from physical conditions but also from the interaction of the following factors:

• Communication and Coordination: Insufficient information sharing and language barriers between the icebreaker and the escorted vessel jeopardise safety (Zhang et al., 2019).
• Time Pressure and Operational Planning: Large-scale commercial investments such as LNG projects place significant time pressure on ship captains, increasing the risk of incorrect decision-making (Moe, 2020).
• Lack of Experience: The lack of experienced personnel in Arctic waters increases the likelihood of failure in emergency response operations (Fedi et al., 2018).
• Governance Gaps: Whilst risk indexing systems like the Polar Operational Limit Assessment Risk Indexing System⁵ (POLARIS) are effective in assessing ice conditions, they cannot fully encompass human errors and organizational inconsistencies (Fedi et al., 2018).

3 Traditional Arctic hazards and emerging operational risks

Ships navigating in Arctic waters face a range of conventional hazards arising from extreme climatic conditions and the dynamic nature of sea ice, unlike offshore operations (Fu et al., 2022). These physical risks directly threaten the ship’s structural integrity, engine performance, and operational safety (Øyen and Kim, 2021; Yan et al., 2026). These fundamental risk factors, which directly threaten ship structure, engine performance, and operational continuity in Arctic maritime transport, are detailed under the following headings within the context of the dynamic nature of the Arctic environment (Yan et al., 2026):

• Ice Pressure: Ice entrapment or pressure is the static and dynamic force created on a ship’s hull by the accumulation of ice masses due to the effects of wind and currents (Boström and Österman, 2017; Goerlandt et al., 2017). Particularly during escort operations, high ice pressure causes the channel opened by the icebreaker to close rapidly. This forces a reduction in the following distance and increases the risk of collision. Severe pressure can cause the ship to become trapped in the ice (Montewka et al., 2015).
• Beset: The situation where a ship is trapped between ice floes to the point where it cannot move under its own power is defined as “besetment” (Fu et al., 2016; Xu et al., 2022). This usually occurs with high ice concentration (>80%) and thick ice sheets. A ship trapped in ice may drift uncontrollably with the ice, run aground, or risk capsizing due to increased lateral pressure.
• Low Visibility: Visibility in the Arctic is frequently restricted due to dense summer fogs, snowstorms, whiteouts, and long polar nights (Bellona, 2025; Valdez Banda et al., 2016). During the summer months, the probability of fog and low cloud ceilings can reach 70-80%, making the effective use of navigational aids such as radar and ARPA essential. Restricted visibility makes it difficult to detect ice floes and obstacles, increasing the risk of collisions and impacts (Khan et al., 2019).
• Machinery Failure: Extremely low temperatures place significant stress on ship machinery and auxiliary equipment (Øyen and Kim, 2021). An examination of incidents along the NSR reveals that machinery damage or failure is the most frequently reported event, accounting for 26.1% (Yan et al., 2026). Cold weather can affect fuel fluidity, cause icing of air intakes, and increase the brittleness of metal components, leading to power loss or complete mechanical failure (Øyen and Kim, 2021).
• Hull Damage: Hull damage caused by ice loads can range from sheet metal deformations and punctures to water ingress (Yan et al., 2026; Zhang et al., 2020). The impacts of ice ridges or ice fragments are particularly significant factors compromising structural integrity (Valdez Banda et al., 2015). Yan et al. (2026) report that 13.6% of incidents on the NSR are direct hull damage, often resulting from unsuitable speed choices for ice conditions (Yan et al., 2026).

These traditional hazards form the basis for proactive risk assessment and the implementation of systems like POLARIS in Arctic maritime transport (Fedi et al., 2018). The character of maritime activities in the Arctic has undergone a radical transformation with the commissioning of the Yamal LNG project and the increase in destination shipping. Today, NSR escort operations have evolved from being merely an ice navigation problem into a risk area where complex socio-technical systems are managed. The new operational risk factors that have emerged in this process are detailed below:

• Heavy Traffic: The commissioning of the Yamal LNG project at the end of 2017 caused an explosion in cargo volume and escort operations on the NSR. Although Russia targeted 80 Mt of cargo for 2024 in its strategic plans in 2020 (Skripnuk et al., 2020), the volume realised in 2024 remained at 37.9⁶ Mt due to the effect of sanctions. Whilst this figure reflects the total volume of cargo transport, LNG transport, the largest cargo item on the NSR, reached a record high of 21.86 Mt in 2024 and maintained its operational weight (Bellona, 2025).
• Convoy: The need for icebreaker escorts on the NSR necessitates convoy operations where a single icebreaker leads multiple vessels, especially when the number of icebreakers cannot keep up with the increasing traffic. Examining convoy dynamics in the Vilkitskii Strait, one of the narrowest and riskiest passages on the NSR, reveals that increased ship density within the convoy restricts the time of reaction to sudden slowdowns by the leading vessel, which results in traffic jams and increased risk of ship-to-ship collisions (Khan et al., 2019). Furthermore, risk models confirm that deviations from the channel opened by the icebreaker or failure to maintain a safe following distance can lead to the vessel becoming trapped in the ice and subsequently catastrophic collisions (Xu et al., 2022).
• LNG Terminal Traffic: Industrial expansion in the Arctic region has transformed ports such as Sabetta (Yamal LNG) and Utrenniy (Arctic LNG 2) into the most strategic and busiest logistics hubs on the NSR. These terminals have turned areas such as the Ob Bay and the Yenisei Bay into “industrial corridors” where ship traffic far exceeds Arctic standards (Yan et al., 2026). For example, the approach channel providing access to Sabetta Port is only 295 metres wide and 15.1 metres deep, severely limiting maneuvering space for giant Yamalmax-class LNG tankers approaching 300 metres in length (Bellona, 2025).
• Time Pressure: The commercial sustainability of large-scale energy projects such as Yamal LNG and the planned Arctic LNG-2 relies on extremely tight shipping schedules and just-in-time arrival at ports of destination (Moe, 2020). High-level political expectations, such as Russia’s target of 80 Mt of cargo volume annually by 2024 as stated in its strategic documents, place significant pressure on operational decisions (Skripnuk et al., 2020). The literature emphasises that such commercial and strategic goals can shift operational priority from safety to “commercial speed,” creating “high crew pressure” on ship personnel in convoy operations and increasing the likelihood of errors (Xu et al., 2022). Particularly in escort and convoy operations, the lack of time for personnel during the planning and implementation phases is identified as an organizational risk factor leading to decreased situational awareness and strain on safety standards (Valdez Banda et al., 2016).
• Coordination: Operational coordination between the icebreaker escorting the vessel and other stakeholders involved in the escort operation (Vessel Traffic Services, port authorities, etc.) is vital for safe passage. Today, Arctic escort operations are no longer defined as mere ice navigation problems, but as complex socio-technical systems requiring multi-layered coordination (Li et al., 2024). Studies of the Baikal-Vladivostok accident on the NSR in 2024 have demonstrated that a lack of coordination regarding safe following distance and speed selection, combined with inadequate operational planning, can lead to catastrophic consequences (Bellona, 2025). Furthermore, uncertainties in the governance structure on the NSR, such as the sharing of authority between different state agencies like Rosatom and the Ministry of Transport, and in operational coordination mechanisms, are cited in the literature as institutional risk factors for the system as a whole (Moe, 2020). Experts consider improving the quality of collaboration between bridge personnel and icebreaker masters, and developing a shared situational awareness amongst stakeholders, as a critical safety layer, especially in convoy and escort operations conducted in icy waters (Valdez Banda et al., 2016)
• Communication: Uninterrupted and clear information flow between an icebreaker and its escorting vessel is a cornerstone of operational safety. In the Arctic, communication is not only a technical infrastructure problem but also a significant human factor risk. A literature review reveals that inadequate English language skills are a major obstacle to communication between icebreakers and escorting vessels. Some seafarers’ vocabulary deficiencies and difficulties in constructing complex sentences lead to misunderstandings or a complete lack of understanding of instructions. Although Standard Marine Communication Phrases (SMCPs) include sections for ice operations, low adherence to these standards and the crew’s lack of familiarity with these patterns make it difficult to develop shared situational awareness in emergencies (Boström and Österman, 2017). Investigations into the Baikal-Vladivostok accident on the NSR in 2024 demonstrated that delayed communication and a lack of coordination between ships were key factors in the structural collapse during escort operations. Furthermore, the remoteness of the region and limited satellite coverage due to high latitudes make data sharing and continuous communication difficult (Fu et al., 2022). In this process, the use of technologies such as Automatic Identification Systems (AIS) is considered a critical supporting element that reduces the margin of error caused by language barriers or radio interference by enabling the electronic tracking of data such as speed and route without the need for verbal communication (Boström and Österman, 2017).

4 Materials and methods

The research consists of two phases. In the first phase, catastrophic safety incidents related to icebreaker escort operations within the NSR boundaries were searched through open sources. In the second phase, the Baikal–Vladivostok accident was examined as a detailed case study and evaluated within the framework of the Swiss Cheese Model (SCM).

4.1 Data collection, source reliability and cross-validation

The event-identification process followed four sequential steps. First, the incident dataset compiled by Yan et al. for the 2005–2024 period was used as the initial screening frame. Second, additional searches of publicly available official, institutional, academic, and maritime-industry sources were conducted to identify potentially relevant events reported up to August 2026. Third, candidate events were screened against three study-specific eligibility dimensions: geographical location within the statutory NSR boundaries, direct association with an icebreaker-supported escort operation, and the occurrence of at least one catastrophic consequence as operationally defined in this study. Finally, qualifying events were cross-validated using the source hierarchy described below, and cases for which the location, escort context, or severity could not be adequately verified were excluded. Accordingly, the resulting cases represent publicly identified qualifying events within the searched documentary record and should not be interpreted as a complete population of all escort-related accidents on the NSR.

To identify catastrophic accidents associated with icebreaker-supported escort operations on the NSR, the study applied a hierarchical source-evaluation and cross-validation procedure rather than treating official records, academic studies, institutional sources, and news reports as evidence-equivalent.

First, the reported maritime incidents compiled by Yan et al. were reviewed as an initial incident-identification dataset. Yan et al. compiled 170 reported maritime incidents occurring along the NSR between 2005 and 2024 using multiple documentary sources and explicitly acknowledged that some incident records contained incomplete information. Accordingly, the dataset was used as a starting point for case identification rather than as the sole basis for determining the catastrophic nature or operational context of individual accidents. Accidents from 2024 to August 2026 were searched through different sources as well.

Second, potentially catastrophic cases were screened according to the severity criteria adopted in this study (IMO’s Formal Safety Assessment), with particular attention given to fatalities, serious injuries, loss or presumed loss of a vessel, and other consequences indicating a catastrophic outcome. Each candidate case was subsequently examined through cross-validation using available primary and secondary sources.

A source hierarchy was applied when accounts differed. Official accident investigation reports, although very limited, and records submitted to or available through IMO mechanisms were given the highest evidentiary weight, followed by official reports and accident records issued by national maritime or transport authorities and other governmental institutions. The IMO Global Integrated Shipping Information System (GISIS), which contains marine casualty and incident information (GISIS, n.d.), and publicly available sources. Primarily academic studies and institutional publications and secondarily contemporary maritime news reports were used as supplementary evidence, particularly for confirming the date, location, operational circumstances, and immediate consequences of incidents for which a complete official investigation report was not publicly accessible.

News sources were not treated as equivalent to official investigation findings. Their reliability was assessed according to source attribution, temporal proximity to the event, and consistency with independent reports. When different accounts conflicted, priority was given to the highest-ranking available source, particularly official investigation findings. Unresolved discrepancies were not interpreted as confirmed facts. Accordingly, the identification of cases was based not on a single database or media report, but on triangulation across multiple source categories with explicit prioritisation of official evidence. This approach was intended to reduce the risk of misclassification arising from incomplete accident records, inconsistent terminology, or differences between initial news reporting and subsequent official findings.

4.2 Maritime safety categorization and study-specific incident selection criteria

In IMO’s “Revised Guidelines for Formal Safety Assessment (FSA) for Use in the IMO Rule-Making Process” (IMO, 2018), Appendix 4 (Initial Ranking of Accident Scenarios), the severity index for a maritime safety issue is categorised into four categories:

• Minor (single or minor injuries to crew/passenger and local equipment damage on ship),
• Significant (multiple or severe injuries to crew/passenger and non-severe ship damage),
• Severe (single fatality or multiple severe injuries to crew/passenger and severe ship damage),
• Catastrophic (multiple fatalities to crew/passenger and total loss of a ship)

The IMO FSA classification is presented here as a general maritime-severity reference framework. However, the term “catastrophic” is operationalised more broadly for event inclusion in the present study and should not be interpreted as a direct application of the IMO FSA catastrophic category. The study-specific definition was adopted to capture exceptionally high-consequence escort events involving major loss of life, vessel loss, or severe environmental consequences. Accordingly, an event was classified as catastrophic for inclusion in this study when it met the geographical and operational criteria specified below and involved at least one of the following consequences:

• it occurred within the statutory boundaries of the NSR;
• it was directly associated with an icebreaker-supported escort operation; and
• it resulted in at least one catastrophic consequence, defined for this study as:
• multiple fatalities;
• loss, presumed loss, or abandonment of a vessel;
• severe environmental damage, or a documented major release of oil, hazardous substances, or radioactive material; or
• another consequence of comparable severity affecting human life, the vessel, or the marine environment.

Yan et al.’s study, which found that 14 icebreakers were involved in accidents amongst the reported events along the NSR (Yan et al., 2026) between 2005 and 2024, was initially utilised to identify the icebreaker-related accidents along the NSR. Next, open sources were searched for accidents that were reported after 2024. Considering the above taxonomy and applying the study-specific eligibility criteria, two publicly identified events were found to qualify for inclusion as catastrophic escort-related events within the searched documentary record when searching for incidents during escort operations in the NSR. The first one is the fire that broke out in the nuclear icebreaker Vaigach in December 2011 in the east of the Kara Gate, whilst she was on her way from Dudinka to Murmansk escorting a vessel, resulting in the death of two crew members and one with a serious wound who was evacuated by helicopter to the mainland (Pettersen, 2011). The other is the Baikal-Vladivostok accident, which is the case study of this study.

4.3 Data analysis

The collected data were analysed descriptively to identify the characteristics of the selected catastrophic icebreaker-escort accident and the factors documented in the official investigation materials. The analysis focused on the accident sequence, operational circumstances, human and organizational factors, technical conditions, and environmental context. Documentary findings were subsequently examined using a structured analytical framework based on the SCM.

4.4 Analytical framework and coding procedure

The SCM was applied as an interpretative framework for systematically organizing the documented safety weaknesses and system conditions identified in the accident investigation rather than as an independent method for establishing causal relationships beyond those identified in the official investigation. A structured evidence-based coding procedure distinguished official factual and causal findings from evidence-based author interpretations and author-derived preventive implications. Direct accident mechanisms, contributing operational factors, latent organizational conditions, and environmental contextual stressors were analysed separately. Factors not explicitly supported by documentary evidence were not treated as established causes, whilst ambiguous evidence was either qualified as an interpretation or excluded from causal assessment. Analytical classification and coding framework for the Baikal–Vladivostok accident in Table 1.

Table 1. Analytical classification and coding framework for the Baikal–Vladivostok accident.

In the table, E1–E6 are based on accident causes and findings explicitly stated in the investigation report. E7 is a preventive implication derived from official advice; E8–E9 are analytical classifications taken from the event’s narrative and technical context; and E10 is explicitly separated as an author-derived preventive implication. This separation prevents the official findings from being confused with the author’s interpretations.

4.5 Translation and analytical reliability

The official Russian-language investigation report, which was publicly accessible when retrieved by the author in June 2026 but is no longer accessible through the original online source, is provided as Supplementary Material S1. A certified English translation of the report is provided as Supplementary Material S2 and was used as the primary working document for the qualitative analysis. The translated version was used as the primary working document for the qualitative analysis. To minimise the risk of misinterpretation arising from specialised Russian maritime terminology, key passages used in the analysis were checked against the corresponding original-language text where necessary, particularly those relating to the sequence of events, technical conditions, communication, ice conditions, operational decisions, and the official findings of the investigation.

The analysis was conducted using a structured evidence-based coding procedure. Each analytical statement was recorded together with its source location in the investigation report and classified as either: (i) an official factual finding, (ii) an official conclusion or causal finding, (iii) an evidence-based author interpretation, or (iv) an author-derived preventive implication. Ambiguous or insufficiently documented evidence was not treated as an established causal finding and was instead classified as uncertain or excluded from the final analytical interpretation.

4.6 Limitations of the study

This study has several limitations that should be considered when interpreting the findings. First, selection bias cannot be completely excluded because the study relies on publicly available and identifiable accident records. Incidents that were not publicly reported, were insufficiently documented, or were unavailable through accessible official or secondary sources may therefore be absent from the analytical sample.

Second, reporting bias may affect the evidence base because the level of detail available for individual incidents varies substantially across official reports, academic sources, institutional records, and news reports.

Third, outcome heterogeneity is an additional limitation. The catastrophic-event classification incorporates severe outcomes such as fatalities and total vessel loss, but these outcomes may arise through different operational mechanisms. Consequently, the cases should not be interpreted as homogeneous manifestations of a single accident process.

Fourth, the small number of catastrophic escort cases identified for the study period introduces substantial small-event uncertainty. The limited case count restricts the ability to infer stable patterns or statistically generalisable relationships amongst the identified defence-layer deficiencies.

Fifth, translation risk cannot be eliminated completely. Although the official Russian-language investigation report was translated into English by a certified sworn translator and key passages were checked against the original text, specialised maritime, legal, and technical terminology may still introduce subtle differences in interpretation.

Sixth, the qualitative SCM analysis was conducted by a single researcher and therefore involves a degree of coder subjectivity. Formal intercoder reliability testing was not possible. To mitigate this limitation, an evidence matrix was used to link each analytical interpretation to the corresponding passage of the official investigation report and to distinguish official findings from author-derived interpretations.

Finally, the detailed SCM analysis is based on a single case study and therefore has limited generalisability. The findings should be interpreted as analytically informative rather than statistically representative of Arctic escort operations as a whole. Further multi-case studies using comparable investigation data would be required to assess whether the identified defence-layer weaknesses recur across different vessels, ice regimes, operators, and geographical sectors.

5 Case study analysis: Baikal–Vladivostok accident

In the second phase of the research, the Baikal–Vladivostok accident, which occurred in June 2024 and has an accessible investigation report (Rostransnadzor, 2024), was selected as a case study. The reasons for selecting this accident are as follows:

• It occurred during an icebreaker escort operation within the boundaries of the NSR,
• It resulted in loss of life and total loss of a ship,
• Operational, environmental, technical, and organizational factors played a combined
• role,
• It represents current risks encountered in Arctic escort operations,
• The investigation report was available from publicly available sources.

The phases of the accident process were reconstructed chronologically; the factors influencing the event were analysed using the SCM.

5.1 SCM coding framework

The Swiss Cheese analysis was conducted using a structured hybrid deductive–inductive coding approach. The analytical framework was established before the detailed interpretation of the accident sequence. The analytical framework was established before the detailed interpretation of the accident sequence. Five defence layers were defined deductively based on the multi-layered defence concept of the SCM and the characteristics of maritime escort operations: (1) governance and regulatory safeguards, (2) planning and preparedness, (3) operational control, (4) human and communication factors, and (5) technical safeguards. Environmental and ice conditions were classified separately as external hazards and operational-contextual stressors rather than as defence layers.

The specific weaknesses identified within these categories were subsequently derived inductively from the documentary evidence relating to the Baikal–Vladivostok accident. Thus, the analytical categories were established independently of individual accident outcomes, whereas the specific failures assigned to each category were based on the available evidence.

To reduce retrospective fitting, each identified factor was coded using a predefined analytical protocol specifying: (i) the relevant defence layer or contextual category, (ii) the operational definition of that layer or category, (iii) the identified failure, weakness, hazard, or contextual condition, (iv) supporting documentary evidence, (v) whether the factor represented an official finding or an author-derived interpretation, (vi) whether it constituted an active failure or latent condition, and (vii) its assessed role in the accident sequence. Factors that could not be supported by documentary evidence were not treated as established causal findings.

To enhance the transparency and dependability of the qualitative analysis, analytical classification and coding framework were developed, as shown in Table 1. Each analytical statement was linked to identifiable documentary evidence from the official investigation report and classified according to its evidentiary status. The coding framework distinguished between official factual findings, official causal findings, official recommendations, and author-derived interpretations. Factors not explicitly established in the investigation report were not presented as official causes. Ambiguous evidence was either qualified as an interpretation or excluded from causal assessment.

5.2 Analysis of the case study with the SCM

The SCM, developed by James Reason (Reason, 2000), was used as an interpretative framework to examine how weaknesses in multiple safety barriers contributed to the accident sequence. In this study, the defence system comprises five layers: governance and regulatory safeguards, planning and preparedness, operational control, human and communication factors, and technical safeguards. Environmental and ice conditions are treated separately as external hazards and contextual stressors that exert pressure on these defence layers. The SCM provides a systematic approach to understanding and mitigating risks within complex systems (Shabani et al., 2024). The SCM metaphorically represents failures as “holes” in slices of cheese, with each slice representing a defence layer against potential hazards (Wegman, 2021). When holes in different defence layers align in the same direction, the danger propagates throughout the system, leading to an accident (Reason, 2000; Shabani et al., 2024).

One of the key contributions of the SCM is the distinction between active errors and latent conditions. Active errors refer to direct errors made by operators or ship personnel; latent conditions encompass long-standing, unaddressed systemic weaknesses such as organizational structure, safety culture, training deficiencies, procedural shortcomings, maintenance operations, or managerial decisions. Often, the primary contributing factor in accident occurrence is not active faults but rather these latent organizational conditions (Reason, 2000; Shabani et al., 2024).

A comprehensive literature review by Shabani et al. (2024) reveals that the SCM is widely used in many fields, including healthcare, aviation, nuclear energy, transportation, and industrial risk management. The researchers note that the model is an effective tool, particularly in complex systems, for visualizing risks, identifying failure chains, and developing preventative safety barriers. However, they emphasise that the model should not be considered a static construct; it yields more effective results when considered alongside organizational culture, leadership, communication, and safety management practices (Shabani et al., 2024).

Fukuoka and Furusho conducted a significant study on the application of SCM in the maritime sector (Fukuoka and Furusho, 2016). By examining 84 serious maritime accidents, the researchers considered safety management systems (SMS) and operational risk management as two fundamental defence layers of the SCM. In that study, ten latent conditions were identified using the SHEL (Software-Hardware-Environment-Liveware) model, and the effects of these conditions on different maritime accidents were analysed. In most accidents, the most common latent condition was identified as an indefinite operator condition. However, researchers emphasise that different latent conditions can become dominant in different types of accidents, and therefore each type of accident has its own unique chain of failures (Fukuoka and Furusho, 2016).

In this study, the SCM was used as an interpretative framework to organise and examine documented weaknesses across the accident sequence. It was not used to establish independent causal relationships or to prove that the identified factors, individually or collectively, caused the accident.

The analysis distinguishes between documented direct mechanisms, contributing operational actions, latent organizational conditions, contextual environmental stressors, and author-derived preventive implications. Within the SCM framework, these documented factors are organised across multiple safety layers to examine how operational and organizational weaknesses coexisted with severe environmental conditions during the accident sequence. The SCM is therefore used to provide a systems-oriented interpretation of the documented evidence rather than to establish causal relationships beyond those identified in the official investigation report.

5.3 Operational context

On 4 June 2024, the Russian-flagged tugboat Baikal was assigned to recover the drifting barge Dikson in the Yenisei Bay of the Kara Sea after the barge had broken loose due to ice movement. The operation was conducted to provide icebreaking assistance and, if necessary, clear a navigation channel around the drifting barge under the escort of the icebreaker Vladivostok, which was instructed by the Marine Operations Headquarters/GlavSevmorput.

The operation took place under severe Arctic ice conditions. Ice concentration reached 9/10, with fast ice thickness ranging from 1.5 to 2.0 m. The weather was characterised by northerly winds of approximately 12–13 m/s, overcast skies, and visibility of approximately 7 nautical miles. Although the tugboat was certified for escorted navigation in medium and heavy ice conditions, the prevailing ice environment was highly demanding, requiring continuous coordination between the escorting icebreaker and the escorted vessel.

The escorted vessel, Baikal, was a 40 m long Arc4-class tugboat built in Finland in 1982 with a main engine output of 2,574 kW. The escort vessel, Vladivostok, commissioned in 2015, was an Icebreaker 7-class diesel-electric icebreaker with a propulsion power of 17,424 kW. The substantial difference in propulsion capability and manoeuvrability between the two vessels significantly influenced the escort operation, particularly after entering open water where the icebreaker accelerated considerably faster than the tugboat.

5.4 Sequence of events

• T0 – Escort operation begins: At approximately 18:30 LT, communication between the two vessels was established on VHF Channel 74. The Chief Officer of the icebreaker instructed the Master of the Baikal to follow the icebreaker whilst maintaining a safe distance. However, no specific escort distance or convoy speed was formally agreed before departure.
• T1 – Increasing separation during transit: As the convoy progressed through heavy ice, differences in vessel speed and manoeuvring capability caused the separation distance to increase steadily. Whilst the tugboat initially followed the icebreaker by approximately three cables, the gap gradually widened. After entering open water, the icebreaker increased its speed to approximately 14.9 knots, whereas the tugboat maintained only 5.2 knots, resulting in a separation of approximately 17.2 cables (1.7 nautical miles). Simultaneously, communication between the vessels became limited, and no active monitoring of the separation distance was maintained.
• T2 – Ice channel closure and compression: At approximately 18:52 LT, the tugboat entered the navigation lane previously opened by the icebreaker. However, the drifting fast ice rapidly closed the channel behind the icebreaker. The Baikal became trapped between converging ice fields and was subjected to intense ice compression. The vessel experienced successive rolls to starboard and port before rapidly developing a starboard heel that increased to approximately 90 degrees.
• T3 – Loss of stability and abandonment: Recognizing the critical situation, the Master ordered an “Abandon Ship” emergency. However, the general alarm system failed to activate for unknown reasons. Because the vessel lost stability within a very short period, an organised evacuation was impossible. Eight crew members escaped onto the surrounding ice, whilst one crew member—the boatswain—was unable to evacuate successfully.
• T4 – Capsize, sinking and rescue: Upon receiving the distress message, the icebreaker immediately reversed course and attempted to return to the casualty. Rescue operations commenced using the icebreaker’s crane after the vessel had stopped at a safe distance from the survivors. Continuous ice pressure eventually ruptured the engine room structure and damaged the superstructure, allowing seawater to flood the tugboat. At 19:56 LT, approximately one hour after the initial compression event, the Baikal finally sank. Eight survivors were recovered, whilst the boatswain remained missing despite search efforts.

5.5 Immediate causes

The official investigation concluded that the accident resulted from the interaction of severe environmental conditions and operational deficiencies rather than from a single initiating event. The immediate physical trigger was severe ice compression, which occurred when the navigation channel created by the icebreaker rapidly closed due to drifting fast ice. The compression exceeded the tugboat’s ability to maintain stability, eventually causing structural failure and flooding.

A second contributing factor was the failure to establish and maintain a safe escort distance. The investigation demonstrated that the separation between the vessels increased progressively due to their differing manoeuvring characteristics and speeds. Consequently, the tugboat entered the narrowing ice channel after it had already begun to close.

Communication deficiencies further aggravated the situation. Continuous VHF coordination between the two vessels was not maintained during the escort operation, and no agreed escort speed or following distance had been established before departure. Moreover, the icebreaker’s Master did not continuously direct the convoy as required by the NSR Icebreaker Escort Rules. The investigation also identified the absence of a dedicated work plan for towing operations under heavy ice conditions as an additional operational deficiency, reducing the crew’s preparedness for rapidly changing ice conditions.

5.6 Consequences

The accident resulted in the total loss of the tugboat Baikal after approximately one hour of continuous ice compression. Eight of the nine crew members survived after being evacuated onto the surrounding ice and rescued by the escorting icebreaker. One crew member, the boatswain, was reported missing and was presumed lost at sea after search operations were terminated. Beyond the immediate human and material losses, the investigation identified several systemic safety deficiencies affecting Arctic escort operations. These included inadequate escort planning, failure to define escort speed and separation distance, insufficient bridge-to-bridge communication, and the lack of continuous convoy leadership during icebreaker escort. Consequently, the investigation recommended strengthening voyage planning procedures, maintaining continuous VHF communication, establishing predefined escort distances and speeds, and improving situational awareness during escort operations in heavy ice conditions.

5.7 Swiss cheese model analysis of the case study: defence layers and their deficiencies

The primary physical mechanism of the accident was the sudden and severe ice compression that exceeded the operational and stability capabilities of the Arc 4-class tugboat. The escort operation was conducted under these hazardous environmental conditions, within which multiple safety barriers were expected to prevent the escalation of risk. In this study, five defence layers are identified: governance and regulatory safeguards, planning and preparedness, operational control, human and communication factors, and technical safeguards. Environmental and ice conditions are treated separately as external hazards and operational stressors rather than as defence layers. The accident is therefore interpreted as the result of an alignment between weaknesses in these defence layers and a hazardous environmental context. Within the SCM framework, the documented weaknesses can be organised as vulnerabilities across multiple defence layers operating under severe environmental stress. This configuration provides an interpretative representation of how the documented operational, organizational, human, and technical factors interacted during the accident sequence; it does not independently establish causal relationships beyond those identified in the official investigation report. In the process leading to the accident, the SCM was used as an interpretative framework to organise and examine documented weaknesses across the accident sequence. It was not used to establish independent causal relationships or to prove that the identified factors, individually or collectively, caused the accident. The analysis identified five defence layers: governance and regulatory safeguards, planning and preparedness, operational control, human and communication factors, and technical safeguards. Thus, the SCM analysis distinguishes between protective barriers, which may contain identifiable weaknesses or “holes”, and environmental conditions, which constitute the hazardous context against which these barriers must operate. Environmental and ice conditions were not treated as failed safety barriers. Instead, they were analysed as contextual stressors that shaped the operational environment within which the identified safety barriers and operational decisions functioned.

5.7.1 Governance and regulatory safeguards

The first layer of defence encompasses the governance and regulatory mechanisms governing Arctic escort operations. Investigation findings indicate that although the operation was conducted under the NSR Icebreaker Escort Rules, a detailed, operation-specific risk analysis assessing dynamic risks such as severe ice entrapment was not performed. Furthermore, the lack of clear procedures for dynamically determining safe following distances based on ice conditions is considered a significant weakness at the governance level. This can be considered the first “hole” in the organizational defence layer.

5.7.2 Planning and preparedness

One of the most significant shortcomings at the organizational level was the lack of a standardised operational plan for rescue and escort operations in severe ice conditions. The investigation report reveals that the escort arrangement, following distance, communication methods, and emergency procedures to be used during the tugboat’s rescue of the drifting barge were not planned in detail beforehand. Similarly, it appears that alternative scenarios were not developed before the operation, taking into account dynamic ice movements. Within the SCM framework, these documented shortcomings can be interpreted as latent weaknesses in the planning and preparedness layer.

5.7.3 Operational control

The SCM analysis identified several documented weaknesses within the operational-control layer. Although a general follow-up instruction was initially given between the icebreaker and the tugboat, a safe following distance could not be maintained throughout the operation. As the icebreaker increased its speed after exiting into open water, the distance between the two vessels continuously increased, and the escorted vessel entered the open ice channel with a delay. As a result, the channel was rapidly closed by moving ice, and the tugboat was subjected to intense ice pressure. The lack of continuous situational assessment and the failure to readjust the escort formation according to changing environmental conditions were key weaknesses in the operational defence layer.

5.7.4 Human and communication factors

From a human factors perspective, the investigation does not attribute the accident to a single personnel error. Within the SCM framework, the documented communication deficiencies and difficulties in maintaining appropriate separation can be interpreted as vulnerabilities within the human and communication layer. Situational awareness is treated here primarily as a preventive consideration, consistent with the recommendations of the investigation report, rather than as an independently established cause of the accident.

5.7.5 Technical safeguards

From a technical perspective, both the icebreaker and the tugboat were certified in accordance with classification rules, and there is no evidence that the accident was directly caused by a mechanical failure. However, the difference in power, acceleration, and manoeuvrability between the two vessels made it difficult to conduct the operation safely. In particular, the rapid departure of the icebreaker after entering open water prevented the escorting vessel from following the opened ice channel in time. Therefore, whilst technical factors were not the direct cause of the accident, they were considered a factor that increased operational risks.

5.7.6 Environmental and ice conditions as external hazards

Environmental and ice conditions were not treated as a defence layer in the present analysis. Instead, they were conceptualised as external hazards and contextual stressors that shaped the operational environment in which the five defence layers operated. The most decisive external hazard in the accident was severe ice pressure. Approximately 90% ice concentration, a compacted ice sheet 1.5–2.0 metres thick, and dynamic ice movements caused the opened channel to close rapidly. The tugboat was consequently subjected to intense horizontal ice loads, developed a large angle of heel within a short period, and lost stability.

These environmental conditions, however, are not conceptualised as a failed protective barrier. Rather, they represent the hazardous conditions against which the governance, planning, operational, human, and technical safeguards were expected to provide protection. Within the SCM framework, the accident can therefore be interpreted in terms of the interaction between an external environmental hazard and documented weaknesses across multiple defence layers. This distinction is important because the SCM is used here to identify failures or deficiencies in protective barriers, whereas environmental conditions represent the hazard context within which those barriers operate.

Figure 2 demonstrates a systems-based analysis of the Baikal–Vladivostok accident within the framework of the SCM. The model distinguishes between five defence layers—governance and regulatory safeguards, planning and preparedness, operational control, human and communication factors, and technical safeguards—and the external environmental hazard represented by severe ice conditions. The environmental conditions are not represented as a defence layer or as a failed barrier; rather, they constitute the hazardous operational context within which the defence layers were required to function. Figure 2 visually organises the documented factors according to the five SCM defence layers and the external environmental context. The alignment shown in the figure should be interpreted as an analytical representation of documented vulnerabilities rather than as proof that these factors necessarily or independently caused the accident. In SCM terms, the accumulation of weaknesses across multiple defence layers can reduce the system’s capacity to prevent escalation when exposed to severe external hazards.

5.7.7 Failure chain

Within the framework of the SCM, the Baikal–Vladivostok accident can be evaluated as a chain of failures consisting of successive organizational and operational failures. The development process of the accident can be summarised as follows:

• Latent planning deficiency: Lack of a standardised escort plan and emergency scenarios specific to severe ice conditions.
• Inadequate risk assessment: Failure to conduct a detailed risk assessment regarding dynamic ice movements, safe following distance, and escort formation before the operation.
• Unsafe escort configuration: Failure to manage the speed and following arrangement between the icebreaker and the escorted tug in a manner appropriate to the environmental conditions.
• Reduced situational awareness and increasing separation: Failure to ensure continuous coordination between the bridge crews during the operation, and the distance between the two vessels gradually increasing.
• Hazard escalation: ice-channel closure and severe ice compression: The rapid closure of the opened ice channel, resulting in the tug being subjected to intense ice pressure. Loss of stability and capsize: The accident occurred when the ship lost its stability because of heavy ice loads, capsized, and resulted in loss of life.
• Physical consequence: loss of stability and capsize

This chain demonstrates that the accident was not the result of a single operator error, but emerged through the interaction of weaknesses across governance, planning, operational control, human and communication processes, and technical safeguards under severe environmental and ice conditions. This finding is consistent with the fundamental assumption of the SCM and reveals that safety in Arctic escort operations depends not only on ship performance or ice conditions, but also on the effectiveness of multi-layered defence systems (Reason, 2000; Fukuoka and Furusho, 2016; Shabani et al., 2024).

6 Results and prevention strategies

The findings of the SCM analysis demonstrate that the Baikal–Vladivostok accident was not solely a result of severe ice conditions, but also stemmed from deficiencies in risk assessment, escort planning, operational coordination, and safety governance. Therefore, preventing similar accidents is not possible simply by building stronger ships or more advanced icebreakers. The real need is to strengthen the multi-layered safety barriers related to the planning and execution of escort operations. This section proposes risk-based prevention strategies for Arctic escort operations based on the findings from the case analysis.

6.1 Risk-based escort planning

The Baikal–Vladivostok accident demonstrates that current escort procedures do not adequately consider the dynamic risks that may arise in severe ice conditions. Although a general navigation plan was prepared before the operation, a systematic risk assessment including ice entrapment, escort formation, minimum following distance, and alternative manoeuvring options was not carried out.

Therefore, it is recommended to develop a mandatory “Escort Risk Assessment Matrix” for Arctic escort operations. This matrix should assess the following variables together before the operation begins:

• ice concentration,
• ice thickness,
• ice entrapment risk,
• visibility,
• wind and current conditions,
• ice class of the escorted vessel,
• navigability and manoeuvrability,
• purpose of the operation (transit, rescue, towing, etc.).

By assigning a risk score to each variable, the total risk level of the operation can be determined, and if acceptable risk limits are exceeded, the escort operation can be postponed, or additional safety measures can be implemented. In addition, it is suggested that operational threshold values, which could be called “ice severity thresholds”, be established. Above certain levels of ice entrapment or ice movement, lower speeds, shorter following distances, additional icebreaker support, or alternative route options should be applied instead of standard escort procedures. Thus, operational decisions would be based not only on the captain’s experience but also on predefined objective risk criteria.

6.2 Real-time decision support

The development of the accident demonstrates that environmental conditions can change very rapidly during escort operations. Therefore, planning before the operation alone is insufficient. The use of real-time decision support systems during the operation is of paramount importance. In this context, it is proposed to develop a decision support system that integrates AIS data with high-resolution satellite-based ice maps, meteorological forecasts, and ice movement models. This would allow the escort leader to assess not only the position of the ships but also the likelihood of the ice channel closing in the coming hours.

Furthermore, “minimum separation alert” systems should be used in escort operations. These systems should automatically alert both bridges if the distance between the two ships falls below or exceeds a predetermined safe limit. Similarly, providing visual and audible warnings when the speed difference reaches a critical level can prevent the escort formation from breaking down. In the long term, integrating AI-powered ice prediction models into these systems will contribute to more accurate prediction of dynamic ice movements and more reliable operational decisions.

6.3 Human and organizational measures

The SCM analysis indicates that the documented accident factors extend beyond environmental conditions and include human, operational, and organizational dimensions. On this basis, joint escort training programmes may represent an appropriate preventive measure. This training should focus on:

• escort formation in heavy ice conditions,
• bridge coordination,
• emergency communication,
• ice jamming scenarios,
• joint decision-making processes and should be practiced practically.

In addition, existing Bridge Resource Management (BRM) training should be expanded to cover the specific requirements of Arctic escort operations. Escort leadership, task sharing, shared situational awareness, effective communication, and dynamic risk assessment should become an integral part of standard BRM training. At the organizational level, it is recommended that structured After Action Review (AAR) meetings be held after each escort operation. This allows for a systematic assessment of risks encountered during the operation and the transfer of experience to future operations.

6.4 Regulatory implications

The Baikal–Vladivostok accident highlights several aspects of the current NSR escort arrangements that require improvement. In particular, the lack of standard procedures that detail operational practices can leave safe escort operations solely to the captain’s experience. In this context, the following regulatory improvements are proposed (Table 2).

Table 2. Evidence-informed preventive implications derived from the Baikal–Vladivostok accident analysis.

Implementing these recommendations will not only enhance the safety of escort operations on the NSR but will also contribute to more effective management of operational risks that may arise in the future in parallel with the growth of Arctic maritime traffic. Especially considering the expansion of the Project 22220 fleet and the maintenance of year-round navigation targets, establishing risk-based governance mechanisms accompanying the development of technical capacity will be a fundamental element of sustainable and safe Arctic shipping.

The preventive implications presented in Table 2 are derived from the documented safety weaknesses identified in the Baikal–Vladivostok accident analysis. They are classified according to their evidentiary basis to distinguish investigation-supported findings from author-derived preventive implications. The proposed measures should not be interpreted as validated interventions or prescriptive operational requirements. In particular, parameters such as escort separation, communication frequency and decision thresholds should remain condition-dependent, as their effectiveness may vary with ice conditions, vessel characteristics, convoy configuration and operational circumstances. Measures identified as author-derived implications require further assessment through operational data, expert elicitation, simulation and/or additional accident cases.

7 Conclusions

The commissioning of the Yamal LNG project in 2017 and the subsequent increase in Arctic energy transportation have significantly expanded vessel traffic along the NSR. This growth has increased the operational importance of icebreaker escort services and intensified concerns regarding navigation safety under challenging Arctic conditions.

In conclusion, Arctic escort operations are no longer merely an icebreaking engineering problem; they are complex socio-technical systems in which human factors, organizational preparedness, and environmental stressors are intertwined. Preventing future catastrophic accidents depends on a holistic analysis of this layered structure.

The Baikal–Vladivostok accident demonstrates that catastrophic escort accidents cannot be explained solely by extreme ice conditions. The SCM analysis revealed that the accident emerged through the interaction of multiple latent organizational deficiencies, inadequate operational planning, ineffective escort management, communication shortcomings, and severe environmental conditions. Rather than representing an isolated human error or unavoidable environmental event, the accident illustrates how failures across multiple defence layers can align to produce a catastrophic outcome.

The findings further suggest that the primary safety challenge for future Arctic escort operations is gradually shifting from purely environmental hazards towards organizational resilience, operational decision-making, and safety governance. As Arctic shipping continues to expand and Russia further develops its icebreaker fleet and year-round navigation capability, strengthening risk-based escort planning, real-time decision support systems, standardised communication procedures, and escort-specific bridge resource management practices will become increasingly important.

From a policy perspective, this study proposes that Arctic escort operations should be managed through a proactive, system-based safety framework rather than relying primarily on vessel capability or navigator experience. The introduction of mandatory escort risk assessment matrices, dynamic ice severity thresholds, standardised escort planning procedures, and integrated real-time ice information systems could substantially reduce the likelihood of future catastrophic accidents. Future catastrophic Arctic escort accidents are likely to arise not from ice conditions alone, but from the interaction of latent organizational deficiencies, inadequate operational planning, ineffective escort management, communication shortcomings, and, finally, severe environmental and ice conditions as external hazards rather than failed defence layers.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. Generative AI tools were used in a limited capacity for language editing and proofreading to improve English grammar, clarity, and readability. AI assistance was also used solely for the graphical preparation and visual presentation of Figure 2. Generative AI was not used to identify, select, code, or interpret the accident evidence, nor was it used to define the analytical layers of the Swiss Cheese Model framework. The analytical structure, classification of evidence, interpretation of findings, and conclusions were developed by the author.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmars.2026.1965827/full#supplementary-material.

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Footnotes

Edited by: Serdar Yildiz, Sharjah Maritime Academy, United Arab Emirates

Reviewed by: Müge Büber, Dokuz Eylül University, Türkiye

Loay M. Aboud, Arab Academy for Science, Technology and Maritime Transport (AASTMT), Egypt

Footnotes

In the liquefied natural gas (LNG) industry, a "train" is the technical term for each large, independently operating liquefaction and refining unit/plant used to liquefy crude natural gas. The Yamal LNG project was constructed in three main production trains (phases), each with a nominal annual capacity of 5.5 Mt. Train 1 was commissioned in December 2017; Train 2 was commissioned in July 2018, and Train 3 was commissioned in November 2018. Finally, a smaller fourth train with a capacity of 0.9 Mt, produced using domestic technology, was later added to the project.

The Official Russian definition for the NSR: “The water area of the Northern Sea Route shall be considered as the water area adjacent to the Northern coast of the RF, comprising the internal sea waters, the territorial sea, the adjacent zone and the exclusive economic zone of the RF and confined in the East with the Line of Maritime Demarcation with the USA and Cape Dezhnev parallel in Bering Strait, with the meridian of Cape Mys Zhelania to the Novaya Zemlya Archipelago in the West, with the eastern coastline of the Novaya Zemlya Archipelago and the western borders of Matochkin Strait, Kara Strait and Yugorski Shar.” (The Ministry of Foreign Affairs of the RF, 2025). However, RF strategic documents are beginning to adopt a new term, the “Greater NSR”, extending from Saint Petersburg to Vladivostok (Bellona, 2025). For this study, the NSR is understood as in Figure 1.

As of mid-2026, Russia has 42 icebreaking vessels in operation, including 8 nuclear-powered icebreakers and 34 diesel-powered icebreakers, which assist vessels with no ice class or low ice class in navigating the NSR. Currently, two types of icebreakers are planned for construction: Project 22220 and Project 10510. Project 22220 icebreakers are expected to replace the Soviet-built nuclear-powered icebreakers Vaigach, Taymyr, Yamal, and 50 Let Pobedy, which are nearing the end of their operational lifespan. Project 10510 is an icebreaker with an increased power of 120 MW, and these icebreakers are planned for use in year-round navigation along the eastern part of the NSR. Five Project 22220 nuclear-powered icebreakers have been delivered or entered service, whilst the construction programme continues despite geopolitical and economic challenges. In parallel with the Project 22220 programme, Russia has continued the development of the Project 10510 Lider-class nuclear icebreaker, designed to provide year-round escort capability for large-capacity vessels operating along the NSR. Unlike the Project 22220 icebreakers, which are intended for operation in a wide range of Arctic waters, the Lider class is specifically designed to maintain navigation through ice exceeding four metres in thickness and to create wider navigation channels for next-generation LNG carriers and large commercial vessels. However, the programme has experienced repeated schedule revisions, rising construction costs, and procurement challenges associated with international sanctions and limited availability of specialised equipment. Despite these delays, the Russian government continues to regard the Project 10510 programme as a strategic component of its long-term Arctic development policy (Bellona, 2025).

Particularly difficult ice conditions are found between the Kara and Laptev Seas, around the Severnaya Zemlya archipelago, and in the western East Siberian Sea, where ice often persists even through the summer (Bellona, 2025).

POLARIS is a risk assessment methodology developed under the leadership of the International Maritime Organization (IMO) and the International Association of Classification Societies (IACS).

Yamal LNG produced 19.88 million tons of LNG in 2023. Almost the entire volume was exported to Europe and Asia, accounting for over half of the total NSR cargo traffic, which reached 37.9 Mt in 2024. The growth of cargo traffic along the NSR began to slow after 2019, when the two main freight suppliers (the Novoportovskoye oil field and Yamal LNG) reached their projected capacity. In 2022, because of international sanctions imposed on Russian industries in response to their invasion of Ukraine on 24 February 2022, the cargo traffic along the NSR declined for the first time in many years. Although it once again reached a historic high in 2024, it had increased by only 20% since 2019—from 31.5 to 37.9 Mt—which is significantly below the targets set in state strategies. In 2024, LNG accounted for 57.69% of the cargo flow, crude oil for 21.37%, petroleum products for 2.34%, gas condensate for 3.5%, and bulk cargoes, including coal, for 2.85% (Bellona, 2025).