Analytical Modeling Enables Quantifying the Probabilities of the Outcome of Critical Complex Maritime Missions and Systems: Perspective and Examples
Suhir E* and Batrak Y
ABSTRACT
"Mathematical" and, particularly, probabilistic modeling enables shedding useful light on some critical complex safety-at-sea and offshore tasks and problems. The approach does that by predicting the most likely outcomes of planned missions including those of the "human-in-the-loop" type. In such missions the reliability of the instrumentation and equipment, both its hard- and software, and human performance contribute jointly to the mission's outcome. Two maritime-safety problems, in which the role of the human factor is taken into account, are addressed in this write-up as suitable examples: 1) predicted probability of failure of a planned maritime mission/ voyage and 2) probabilistic assessment of the possible roles of the more or less permanent human factors (such as age, experience, education, health, training, etc.) and the temporary state of the human's health (say, such as, e.g., cold, headache, runny nose) and/or mind (say, short-term loss of attention, drowsiness) that might affect the likelihood of the occurrence of what is known as human error. It is concluded that while some kind of predictive modeling should always be considered and conducted prior to and, whenever possible and appropriate, also during accelerated reliability testing of the navigation instrumentation or human performance, physically meaningful analytical ("mathematical"), preferably probabilistic predictive models should be developed and applied to complement the results of computer simulations: these two major modeling tools are based on different assumptions, employ different calculation techniques, and if the results obtained using these tools agree, then there is a good reason to believe that the obtained information is sufficiently accurate and, hence, trustworthy. Future work should address other suitable applications of the employed analytical modeling technique, as well as the development of practical analytical models for establishing the risks for critical complex systems and applications, considering both the predicted never zero probabilities and the consequences of the possible failures. These probabilities cannot be high, of course, but they should not be lower than necessary either: they should be adequate for the system and application. Systems that “never fail” are much more expensive than they could and should be.
