The inherent complexity of the flooding process on large passenger vessels has, in recent years, been recognised. Accounting for the fact that traditional emergency response is a highly time consuming and manual process, prone to human misjudgement or error, this might lead to severe consequences. This is also highly paradoxical given that the most important variables in an emergency are Time-to-Capsize and Time-to-Evacuate. The research culminating in this PhD dissertation has, therefore, been directed towards the development of a framework where sensor-technology and analytics are utilised to a higher degree, in an attempt to improve real-time information providing risk-informed situational awareness in flooding emergencies involving large cruise vessels, thus enabling optimised emergency response. A fully probabilistic prediction methodology has been developed and presented. The methodology takes advantage of available sensor readings as supportive evidence and utilises inference in the form of probabilistic multi-sensor data fusion techniques for manipulating conditional probability distributions. This targets the observed distribution, enabling reduction in uncertainty (probabilistic inference). The framework and its corresponding comprehensive probabilistic models are further seen to be highly suitable for implementation as a Life-Cycle flooding risk management framework.;A range of probabilistic models in the form of likelihood functions have been developed for a specific sample vessel using the state-of-the-art time-domain simulation code PROTEUS3. This enables the simulation of a damaged ship in a dynamic operational climate imposed by waves. Implementation and testing of the framework on a range of realistic test scenarios reveals that the method identifies the expected damaged region for all cases despite not having an exhaustive sensor array. This clearly provides improved survival assessment to enable the crew to implement emergency response in a more timely, targeted and efficient manner in accordance with the main aim set out in the thesis. Finally, having implemented the framework on an existing cruise vessel with the as-built sensor array indicates that the methodology may be implemented on a large cruise vessel without changes to the flooding detection system, as it is seen that the as-built sensor arrays allow for accurate predictions by relying entirely on the presented methodology. This allows for reliable estimation of realtime flooding risk in passenger ships through its life-cycles and most importantly, when it really matters, namely in emergencies. This is an innovation offering unique tangible benefits.
|Date of Award||1 Oct 2019|
- University Of Strathclyde
|Supervisor||Dracos Vassalos (Supervisor) & Jakub Cichowicz (Supervisor)|