Decarbonisation Technology May 2026 Issue

highlighting potential system failures, human errors, and emergency conditions. HAZOP analyses further enrich this by detailing operational deviations and their potential impacts on safety and operational reliability. By combining these qualitative findings from HAZID and HAZOP with quantitative results from CFD, the QRA employs probabilistic modelling techniques, such as fault trees and event trees, to systematically assess potential initiating events and subsequent escalation pathways identified through workshops and CFD simulations. These techniques quantify both the likelihood and severity of identified incidents, allowing safety planners to prioritise risks effectively and make informed decisions regarding risk control measures. Dynamic human factor modelling through agent-based simulation However, traditional QRAs use static risk equations, which, while valuable, typically assume fixed scenarios and conditions, limiting their effectiveness in capturing dynamic aspects of maritime emergency situations, particularly human behaviour. In reality, crew members continuously move throughout the ship, following distinct daily patterns influenced by shift rotations, operational requirements, and personal routines. For instance, during daytime hours, crew activities might concentrate heavily on operational areas such as the bridge or engine room, increasing the potential risks associated with those locations. Conversely, nighttime typically finds crew members dispersed in accommodation areas, significantly altering risk profiles and evacuation routes in the event of an incident. In addition, real-life emergencies rarely follow a predictable script. Human decisions under stress – shaped by fatigue, situational awareness, training, or panic – can lead to unpredictable outcomes. These responses affect how quickly alarms are raised, how evacuation routes are chosen, and how effectively hazards are mitigated. Considering factors such as moving speeds, decision-making delays, miscommunication, and hesitation becomes critical in building a more realistic understanding of how emergencies unfold.

Incorporating such dynamic and behavioural considerations into safety assessments can allow for a more accurate evaluation of risks. It reveals how seemingly minor differences in human behaviour can lead to significant variations in evacuation timelines, potential bottlenecks, or unintended outcomes. These insights are particularly valuable in developing emergency response plans that are more adaptable, practical, and aligned with the real- world operational environment on board ships. Agent-based ERP evaluation While traditional emergency response plans, especially those designed for oil spills, focus primarily on environmental contamination and fire risks, ammonia-related incidents present immediate and severe toxic exposure threats that demand a fundamentally different set of response priorities. Recognising this, ABS has developed an advanced agent-based probabilistic risk analysis approach that accounts for the unique uncertainties involved in ammonia emergencies, including both the “ Considering factors such as moving speeds, decision-making delays, miscommunication, and hesitation becomes critical in building a more realistic understanding of how emergencies unfold ” complex dispersion behaviour of ammonia plumes and the human factors involved in emergency situations (see Figure 3 ). This methodology integrates key inputs from CFD simulations, insights from HAZID and HAZOP workshops, and realistic leakage scenarios to simulate crew responses under a wide range of hazard conditions. At the core of this approach is agent-based modelling, a sophisticated computational technique used to simulate the behaviour of complex systems by modelling individual entities, or ‘agents’, each with distinct decision-making abilities, physical properties, and adaptive behaviour rules. In the maritime context, agents represent individual crew members, each programmed with realistic movement patterns and decision- making logic that reflects daily operational