Statistical analysis of categorical associations in LNG and LPG tanker accidents

The Global Maritime Conference (GMC’26), İzmir, Türkiye, 28 - 29 Eylül 2026, (Özet Bildiri)

Yayın Türü: Bildiri / Özet Bildiri
Basıldığı Şehir: İzmir
Basıldığı Ülke: Türkiye
Dokuz Eylül Üniversitesi Adresli: Evet

Özet

Background/Objectives: Liquefied Natural Gas (LNG) and Liquefied Petroleum Gas (LPG) tankers operate under stringent safety requirements due to the hazardous nature of their cargo and the potential consequences of marine accidents. Understanding the relationships among ship characteristics, operational conditions, accident subtypes, and accident significance may contribute to risk-based safety management. This study aimed to analyze the associations among selected categorical variables in LNG/LPG tanker accident records using Chi-square test and Cramér’s V.

Methods: In this study, LNG and LPG tanker accidents over a ten-year period were analyzed. The data used for this research was collected from maritime accident reports on the IHS Sea-Web database. Before the analysis, 391 maritime accident reports for LNG and LPG vessels were collected. After the data preprocessing process, a total number of 203 maritime accident reports were gathered for further analysis. Nine categorical variables were examined, including cargo load status, ship type, ship age, ship length, ship flag, season, environment, event (accident) subtype, and accident significance. Associations between categorical variables were assessed using Pearson’s chi-square test and bias-corrected Cramér’s V. In the crosstab analyses, comparisons that violated the fundamental Chi-square assumption and that did not show a statistically significant association (p>0.05) were excluded from the findings. Cramér’s V values were used to interpret the strength of association, while statistical significance was evaluated at p<0.05.

Results: The statistical analysis revealed seven significant relationships (p<0.05) among the examined variables. The strongest association was observed between ship type and ship length group, indicating a strong association (V=0.66, p<0.001). A moderate association was found between event subtype and accident significance (V=0.30, p<0.001). Cargo load status was moderately associated with operating environment (V=0.28, p<0.001) and event subtype (V=0.23, p<0.001). Operating environment was moderately associated with event sub type (V=0.26, p<0.001). Weaker but statistically significant associations were observed between ship length group and event subtype (V=0.14, p<0.05) and between season and event subtype (V=0.13, p<0.05). Overall, the results suggest that both structural ship characteristics and operational conditions are associated with the distribution of event subtype.

Conclusions: The findings indicate that ship type, ship length, accident significance, cargo load status, operating environment, event sub type and season are associated with specific accident patterns among LNG/LPG tankers. The strongest relationship was between ship type and ship length group, while event subtype showed meaningful associations with accident significance, operating environment, cargo load status, ship length and season variables. These results may support the development of targeted risk assessment strategies for LNG/LPG tanker operations. Further studies using larger datasets and multivariable models are recommended to validate these associations and to explore predictive risk factors for maritime accidents.