Evaluation of bayesian networks for risk assessment of major hazards in oil and gas facilities

Abstract

Oil and gas industries handle highly inflammable and toxic fluids under pressure. They also have large inventories of the same. Therefore it is imperative that these fluids are processed under safe operating conditions and that any hazards posed during its operations are controlled and eliminated. Unless the risks are properly managed, hazards can escalate to accidents very rapidly. The industry relies on certain tools like Hazard And Operability Studies (HAZOP), Quantitative Risk Assessments (QRA), Safety Integrity Level (SIL) studies, etc. to understand, analyze and mitigate risks. Of the above, QRA involve quantitative understanding of the risk. QRA studies originated in the nuclear industry and have been successfully adopted by the process industry including oil and gas. Risk is a function of the hazard scenario, its likelihood of occurrence and its consequences and QRAs are the traditional method in the oil and gas industries to analyze risk quantitatively. QRA starts with identifying risk scenarios; mainly loss of containment (LOC) in a facility. Then the frequency or likelihood of occurrence of such scenarios is taken from published sources. (If site data is available the same is used). Consequences of LOC are computed by use of source term models and its impacts on personnel and property. These are combined to produce a measure of risk. The risk arrived at is compared with an established tolerable risk to see if it is acceptable and mitigation measures are taken up if it is not