Event Trees Analysis – ETA
Once a top-event is assessed using Fault Trees Analysis (FTA) is required to assign a frequency to each different evolution of the top-event, this is done using Event Trees analysis.
To clarify the goal of this analysis consider the following example: we assigned a frequency to the top-event “Leakage of gasoline from a pipeline”, with ETA we will able to assign a frequency to the following events: pool-fire, flash-fire, VCE, etc.
In other words we need to assign a probability of ignition (immediate, delayed, late, etc.) given a release (the ETA needs to assign also the probability of failure of the protections but this easy to assess using Fault Trees Analysis (FTA).
The methods used to assign frequency to the every event consequent to a top-event are classified in two families:
- Probabilistic/Stochastic: usually based on a correlation between release flow rate and probability of ignition
- Quantitative: based on the presence of ignition sources. The presence has to be assessed for each scenario (the area with concentration in the plume greater than LFL will vary for each release) and requires onsite review for existing plant and plot plan review for future plant
The probabilistic method based only on flow rate has the following limits:
- Not related to the context (p.e. the methods gives the same results if the leakage is from a Xmas tree in a desert or in the furnaces area inside a chemical plant)
- Not related to the ignition energy of different substances (p.e. ammonia and hydrogen have the same probability of ignition if the flow rate is the same)
- Not related to the duration of the release (p.e. if a release last for 30 seconds or one day the probability of ignition are the same)
- Not related to ignition type (high and low energy sources)
- Not related to the leakage and plume height and shape (p.e. a leakage from a vent line at 50 meter height of methane has the same probability has LPG on the ground if the release flow rate is the same)
- Not able to suggest improvements on reduction and control of ignition source neither to evaluate the ignitions control procedure. If you recommend an improvement to reduce the ignition sources this doesn’t lead to a lower probability of ignition.
From what is written above the preferred Eidos way to deal with ignition sources and ETA is quantitative approach. Eidos developed a method, based on international technical documentation, that overtakes the aforementioned limitations.
The method is able to assign a probability of ignition for every ignition source listed in EN 1127-1 standard and was applied successfully to different plant/site (exiting or planned).
Another advantage of Eidos ETA method is that, in order to define the probability of VCE, assess all the congested on partially confined volumes defining for each of them blockage ratio, volume, fuel reactivity and flame expansion. These are the same inputs used by Baker-Strehlow-Tang model.