Association rule mining for identifying factors in dynamic positioning incidents and accidents

Abstract

Dynamic Positioning (DP) systems play a pivotal role in the offshore industry, automating vessel position and heading control. This study investigates thoroughly into the causes of DP incidents using association rule mining, specifically the Apriori algorithm within the Weka program, developed by the University of Waikato. The dataset, drawn from the International Maritime Contractors' Association (IMCA) database spanning 2004 to 2021, encompasses 691 DP incidents, providing a robust foundation for analysis. Dynamic Positioning is hailed as a technological boon in the offshore industry, ensuring vessels maintain specific positions and headings automatically. Despite its advantages, the need for vigilant monitoring persists, managed by Dynamic Positioning Operators (DPOs). This study aims to reveal the causes of DP incidents through association rule mining, offering a comprehension into the factors contributing to these incidents. The DP control system relies on data from gyro compasses and position reference systems (PRS) to maintain a vessel's heading and position. During DP operations, the setpoints for the ship's heading and position are defined by the DPO and then processed by the dynamic positioning control system, which provides control signals for both the thrusters and main propulsion systems of the vessel. The DP control system consistently aims to distribute the most efficient thrust to each of its propulsion units that are currently operational. The heading or position measurements are compared to the predicted data generated by the mathematical model. Subsequently, the variations are applied to modify the model. DP incidents, categorized into loss of redundancy and loss of position, pose significant risks, and understanding their causes is imperative for the industry's safety. Loss of position incidents can also be divided into two different types: drift-off and drive-off. While numerous studies investigate the causes of marine accidents, a noticeable gap exists in understanding the factors contributing specifically to DP incidents. This research aims to address this gap by employing association rule mining to identify patterns and relationships within a dataset comprising 1352 incidents from the IMCA database. The Apriori algorithm, integrated into the Weka program, was employed for association rule mining. Weka, a machine-learning workbench, provided a user-friendly interface for dataset analysis. The initial dataset was filtered to 691 incidents for in-depth exploration. The dataset's origin from the IMCA, a reputable source in the maritime offshore industry, enhances the study's credibility. Analysis reveals primary causes of DP incidents, with thrusters and propulsion failures (24%), computer failures (15%), power-related failures (14%), PRS-related failures (13%), and human factor-related failures (9%) dominating. Secondary causes, notably human factors (54.5%) and electrical problems (13.5%), exhibited an upward trend from 2017 to 2021. In-depth association rule analysis provides insights into the triggers and outcomes of DP incidents. Accidentally pressing buttons (55.2%) and pressing wrong (standby) buttons (13.1%) emerged as significant triggers. Environmental factors, particularly squalls, swells, and waves, were strongly associated with drift-off incidents (100%). Specific causes, such as power-related failures, thruster or propulsion failures, and PRS-related failures, exhibited distinct outcomes, contributing to a nuanced understanding of DP incidents. Based on findings, recommendations focus on enhancing DPO training and industry practices. Suggestions include extending the duration of DP simulator courses, revising sea time requirements, and introducing specialized certifications for each vessel types. Simulator-based training should encompass specific actions during emergencies, blackout recovery, and responses to environmental changes. Extending DP Simulator Courses: Recognizing the significance of practical knowledge, it is recommended to increase the duration of DP simulator courses. This extended training period allows trainee DPOs to gain hands-on experience in a simulated environment, comparing real-life and simulator incidents. Revising Seatime Requirements: Current sea time requirements post DP simulator course completion may be insufficient for gaining the necessary practical experience. Policymakers are encouraged to consider extending the DP sea time duration (currently set at 2 hours for each DP sea time day) or extending the mandated overall sea time, currently set at 60 days, providing prospective DPOs with more hands-on opportunities. Moreover, it is not recommended to transfer any surplus active sea time days acquired prior to the DP simulator course (Phase 3) to subsequent phase (Phase 4). Specialized Certification: A specialized certification program is proposed, wherein DPOs gain certification based on their experience with specific vessel types. This approach ensures that DPOs possess vessel-specific knowledge before assuming responsibilities, enhancing overall competency. Industry stakeholders are encouraged to develop expert specialization programs, allowing DPOs with extensive experience on specific types of vessels to obtain specialized certificates, ensuring they possess the requisite knowledge for their roles. Simulator-Based Emergency Training: DPOs are recommended to undergo simulator-based emergency training covering various scenarios, such as blackout recovery, response to environmental changes, and actions during emergencies. This ensures DPOs are well-prepared for critical situations. Meteorology Training: Given the association of environmental factors with drift-off incidents, meteorology training combined with DP operations is recommended. This training should include actions during not only emergencies, but also extreme weather conditions such as squalls, solitons, unpredictable changes in current, etc. Additionally, maintaining continuous and vigilant watchkeeping is of significant importance for ensuring the safety of operations. Power-related Training: Considering the significant impact of power-related failures, blackout recovery training is highlighted as crucial. DPOs, EROs, ETOs, should be well-versed in blackout recovery processes and vessel-specific characteristics. Additionally, it is recommended to provide comprehensive training for DP key personnel before they embark on their vessel assignment. This training should encompass ship-specific thruster/propulsion training, as well as power management training, covering the concept of different propulsion systems, their limitations, and blackout prevention and recovery processes. Home Training Modules: DP vessel managers and operators should implement home training modules for personnel engaged in DP operations. These modules, covering significant documents related to the vessel and its operations, enhance familiarity and competence. These documents encompass a variety of materials, including, but not limited to, the most recent FMEA and DP annual trial reports of the joining vessel, the DP operation manual, DP incident or failure reports (if any), the latest DP audit report, WSOG, TAM, CAMO, etc. Human Factor Analysis: A more detailed analysis of human factors and situational awareness is recommended for future research. Understanding the intricacies of human error and awareness can contribute to minimizing risks associated with DP incidents. Competency Factor Analysis: For future studies, competency factors of DPO for different types of DP vessels can also be identified in more detail. By means of the work, the competency factors can be achieved by taking a specific vessel type or incident model and employing advanced approaches such as machine learning or Bayesian network analysis, among others. Situational Awareness: Further studies should also explore human factors and situational awareness in greater detail. Situational awareness is vital for DPOs during emergencies, and detailed analysis can contribute to better decision-making and risk mitigation. This condensed summary provides a detailed overview of the original thesis, retaining the essence of the research while presenting key findings and recommendations. Understanding the causes of DP incidents is crucial for enhancing safety in the offshore industry. The comprehensive recommendations aim to guide industry stakeholders, DP vessel managers and operators, training centers, and policymakers in implementing effective measures to reduce the occurrence of DP incidents and improve overall safety standards. The implications of this research extend beyond the academic realm, addressing the substantial financial stakes in the offshore industry. By identifying causes and providing targeted recommendations, this study offers a practical guide for industry practitioners. Potential directions for future research include detailed competency factors, advanced analytical approaches, and in-depth analyses of human factors and situational awareness. In conclusion, this study illuminates the complexities of DP incidents, providing a comprehensive understanding of causes and recommending measures to enhance safety. Given the substantial financial implications of incidents in the offshore industry, this research contributes to a proactive approach, allowing stakeholders to anticipate and address risks effectively.