Prediction of Oil Spill Occurrence Probabilities in the Alaskan OCS

Probabilistic estimates of oil spill occurrences are required for the development of environment impact assessments for possible future developments in the US Beaufort Sea. Due to the embryonic state of offshore oil development in this region, it was not possible to base these oil spill probability estimates on empirical data. Rather, statistically significant non-Arctic empirical data, together with their variance, was used as a starting point. Next, both the frequency distributions and spill causal distributions were modified to reflect specific effects of the Arctic setting and the resultant fault tree model was evaluated using Monte Carlo simulation to adequately characterize the combinations of probability distribution inputs to the fault tree. This paper summarizes the methodology and gives results of its application to the prediction of oil spill probabilities and their characteristics for the Beaufort Sea region for typical future offshore development scenarios.