Estimation of Loss Of Containment Frequencies
When developing the Hazard Identification phase, BC identifies and classifies the Loss Of Containment (LOC) scenarios into two categories: generic and specific.
The identification of all type of pieces of process equipment susceptible to release hazardous materials is the main purpose when generic LOCs are intended to be identified.
The identification of process deviations, human errors during the design, operation and maintenance of the process is the main purpose when specific LOCs are intended to be identified.
The Frequency Analysis is focused on estimating the frequencies of occurrence of LOCs. These frequencies are difficult to be predicted, and criteria is established and followed from historical data, experiments, expert opinion, specific plant data (if available), using worldwide recognized references for generic process equipment failure rates, and/or developing detailed fault trees by applying the Fault Tree Analysis (FTA) methodology.
When addressing generic LOCs, BC uses worldwide recognized references for generic process equipment failure rates. Examples of process equipment types considered are the following: valves, flanged joints, pressure vessels, pipes, pumps, heat exchangers, compressors, filters, hoses, arms, etc. Some of the most used and recognized release databases are developed entities/organizations such as:
American Petroleum Institute (API)
UK Health & Safety Executive (HSE)
European Gas Pipeline Incident Data Group (EGIG)
“Purple Book” – CPR 18E (VROM)
When addressing specific LOCs, BC uses the FTA methodology, which is a recognized and deductive methodology that uses a graphical representation of the combination of faults leading to a predefined undesired event; i.e., Top Event. The methodology uses Boolean logic gates (e.g., AND, OR) to describe how equipment failures and human errors are combined to cause a main system failure.
The FTA not only is valuable for estimating the frequency of occurrence of specific LOCs, but also is capable of identifying the unique combinations of failures that lead to the Top Event. In other words, which are the key events necessary to cause the Top Event; i.e., Minimal Cut Set analysis.
Estimation of Final Outcomes Frequencies
Loss Of Containment (LOC) scenarios identified during the Hazard Identification phase are related with the location of enabling conditions/events:
Immediate/delayed ignition sources
Unconfined congested regions or confined regions
Active/passive protection layers.
During the Source Term Modeling development, the source term physical and thermodynamic properties are estimated, and LOCs are classified accordingly; e.g., two-phase release, gas release.
This classification allows identifying all potential outcomes that the LOC may cause; procedure that accounts for the enabling conditions/events.
1. Developing the Event Tree:
Given a LOC, BC uses the Event Tree Analysis (ETA) for identifying all the final outcomes, inductive technique that graphically explores all the different responses (i.e., outcomes) through a single initiating event (i.e., LOC) and lays a path for assessing the success/failure of enabling conditions, defining as a result all possible effects from a hazardous release. Accordingly:
The Event Tree is built-up systematically starting from a LOC and working through ignition sources to the calculation of the effects of cloud fires, jet fires, explosions, and toxic exposures for unignited releases. Escalation to other parts of the system/process via vessel failure through overpressure and/or flame engulfment is considered.
2. Quantifying the Event Tree:
The Event Tree is quantified systematically starting from the LOC frequency of occurrence and estimating the probabilities of all enabling conditions/events:
BC uses several recognized ignition probability models which are available in the process safety field.
Probabilities of other enabling conditions/events are estimated by considering specific information from the system/process being analyzed.
The quantitative solution of the Event Tree provides the frequencies of the final outcomes.