Common Cause Failure¶
Define groups of basic events that share common cause failure.
- Groups must be unique.
- Groups must not share members.
- All members of a CCF group have the same description for failure.
Choose a model for analysis: MGL, alpha-factor, beta-factor, and phi-factor.
- Factors must be provided by a user.
- Calculation must be performed only with the provided factors. The maximum number of events for grouping into new CCF events can be inferred from the number and levels of provided factors.
- There should be at least one and at most N factors for CCF models. N is the number of members of the CCF group. For the beta-factor model, only one factor is required.
- CCF grouping level numbers are optional, but it helps with input clarity and error-checking.
- Levels of the provided factors are inferred positionally and sequentially. If a factor for any level is omitted, it is not implicitly assumed to be 0. The factors must be set to 0 explicitly. The exception is the factors that are above the last input level. In other words, only the provided factors are used without requiring exactly maximum possible number of factors.
After construction of the tree, calculate CCF sub groups for each event.
- Validate groups.
- Calculate CCF model specific probabilities for new CCF events.
- Assign calculated probabilities to the newly created CCF sub-events.
Substitute CCF grouped primary events with OR gates with children as CCF-calculated sub groups. Give specific names to CCF sub groups that identify independent and common cause failures.
Perform usual analysis on the fault tree with CCF groups.
Report CCF-specific information with common cause events grouped in one element with special formatting.