162
SULAIMAN et al.
limitations especially for complex system and their low frequency, high consequence nature. Therefore creative procedures are required to develop compensation for data relation- ships. The model could use probabilistic, stochastic, simulation and expert judgments couple existing deterministic and historical method for a reliable system analysis of de- sired design (M. Kok, H.L. Stipdonk, W.A. de Vries, 1995; Stiehl, G.L., 1977). When insufficient local data is available,
world wide data from other areas may be referred to (e.g., Europe, south and North America). However, ones need to make as- sumptions about the similarity of operations in the concerned area or elsewhere. This is to ensure how behaviour in one aspect of opera- tional (e.g., company management quality) parameter (e.g., loss of crew time) correlates with another area (e.g., operations safety). The data from other areas can be used as long as major parameters and environmen- tal factors are compared and well matched. Care is required with the use of worldwide data as much of those data are influenced by locations or local environmental condi- tions (Skjong, R., 2002). Electronic access to worldwide casualty data such as the Paris MOU, U.K., Marine Accident Investigation Board (MAIB) and IMO Port State detention databases makes possible access to worldwide casualty statistics. Diligence should also be observed about the large number of small scale, localized incidents that occur that are not tracked by marine safety authorities, e.g small craft (not always registered or being able to be detected by VTS, AIS) accidents in waterways. American Bureau of shipping (ABS) has begun an effort to identify precur- sors or leading indicators of safety in marine transportation. Human Factors modelling should be considered for distributive, large
scale systems with limited physical oversight. Assessing the role of human and organiza- tional performance on levels of risk in the system is important, such error is often cited as a primary contributor to accident, which end up leaving system with many more un- known. Expert judgments and visual real- ity simulation can be used to fill such un- certainty gaps and others like weather data. Even when attempts are made to minimize errors from expert judgments, the data are inherently subject to distortion and bias. With an extensive list of required data, there are limits that available data can place on the ac- curacy, completeness and uncertainty in the risk assessment results. Expert judgments give prediction about the likelihood that failures that would occur in specific situations can be used to quantify human reliability input in risk process. Uncertainty is always part of system behav-
iour. Two common uncertainties are: aleatory uncertainty (the randomness of the system itself) and epistemic uncertainty (the lack of knowledge about the system). Aleatory un- certainty is represented by probability mod- els while epistemic uncertainty is represented by lack of knowledge concerning the param- eters of the model (Pate Cornell, M.E., 1996). Aleatory uncertainty is critical, it can be ad- dressed through probabilistic risk analysis while epistemic uncertainty is critical to al- low meaningful decision making. Simulation offers one of the best options to cover extreme case uncertainty besides probability. Evalua- tion and comparison of baseline scenario to a set of scenarios of interest (tug escort) and operational circumstance including timelines and roles can also be incorporated. Response Scenarios can also be analysed for things that can not be imagined or modeled to be accounted for in the simulator (especially real
Journal of Marine Environmental Engineering
145-173 pp MB_37(Kader)
2.indd 162
2/19/2010 11:04:25 AM
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29