The transition from conventional ships to Maritime Autonomous Surface Ships is often justified by the expectation that removing the onboard crew reduces human error and improves operational reliability. However, this assumption overlooks the crew’s positive contribution to detecting anomalies, interpreting uncertain situations, correcting automated decisions, reconfiguring systems and recovering from technical failures. This study proposes a human-centred reliability framework for comparing manned and unmanned vessel configurations by modelling the crew both as a potential source of failure and as an active recovery barrier. The model distinguishes the human failure contribution, represented through an equivalent human failure rate, from the human recovery contribution, represented through the probability of successful intervention before mission loss.
This study proposes a method for estimating an equivalent human failure rate using accident statistics published by the European Maritime Safety Agency. The number of marine accident attributed to human actions is related to the European fleet, calculated from the number of vessels and their estimated navigation hours.
The calculated human reliability parameter is integrated into a mission-based Reliability Block Diagram model of a Search and Rescue patrol vessel. Manned and unmanned configurations are compared under identical rescue and patrol missions. The analysis therefore addresses mission reliability rather than steady-state availability. In the unmanned configuration, crew functions are replaced by autonomous navigation, sensing, communication and control systems. Since software failures are mainly systematic and cannot generally be represented by a constant exponential failure rate, the corresponding software parameter is interpreted as an effective failure rate for comparative purposes.
The proposed approach provides a transparent method for representing human performance within ship reliability calculations and evaluating how the estimated human failure rate affects the reliability difference between conventional and unmanned vessels. It also highlights the sensitivity of the comparison to assumptions concerning accident attribution, fleet exposure and mission duration.