Substitutions (Delete Terms, Recovery Rules, Exchange Events) #154
Milestone
Comments
rakhimov
added a commit
that referenced
this issue
Nov 1, 2017
The name is modified to be required. The target is corrected to be a single event. Issue #154
rakhimov
added a commit
that referenced
this issue
Nov 1, 2017
rakhimov
added a commit
that referenced
this issue
Nov 1, 2017
This is missing from the Open-PSA MEF specification. Issue #154
rakhimov
added a commit
that referenced
this issue
Nov 1, 2017
Only basic validations for duplicates and missing events are implemented. Substitution specific validation must be implemented separately. Issue #154
rakhimov
added a commit
that referenced
this issue
Nov 1, 2017
- Only basic events are allowed. Issue #154
rakhimov
added a commit
that referenced
this issue
Nov 1, 2017
Delete Terms, Recovery Rules, and Exchange Event types are deduced from the Substitution setups. If the deduced type does not match the explicitly declared one, the input is considered invalid. Issue #154
rakhimov
added a commit
that referenced
this issue
Nov 1, 2017
This substitution case has no effect on analysis, and strictly, it cannot be classified as substitution. Issue #154
rakhimov
added a commit
that referenced
this issue
Nov 1, 2017
Non-declarative substitutions have to be implemented with set operations at post-processing. Complex Boolean formulas would require more expensive solutions (i.e., solving the general satisfiability). The declarative substitutions, in contrast, can simply use implication operator at pre-processing, so there's no need to restrict formulas for the declarative approaches. Issue #154
rakhimov
added a commit
that referenced
this issue
Nov 1, 2017
Target cannot appear in a source set of non-declarative Substitutions. Issue #154
rakhimov
added a commit
that referenced
this issue
Nov 1, 2017
No target or source events can be arguments of another hypothesis. Issue #154
rakhimov
added a commit
that referenced
this issue
Nov 9, 2017
Non-declarative substitutions with False target are the same as declarative substitutions (delete-terms) since the source set has no effect (irrelevant). Issue #154
rakhimov
added a commit
that referenced
this issue
Nov 9, 2017
rakhimov
added a commit
that referenced
this issue
Nov 9, 2017
Handling of CCF events with non-declarative substitutions requires major rework of the current CCF application approaches. Non-declarative substitutions need to be applied before CCF, and the analysis needs to rerun again. For example, this gets very hairy very quickly with application of truncations. Issue #154
rakhimov
added a commit
that referenced
this issue
Nov 9, 2017
Declarative substitutions (e.g., delete terms, recovery rules) are applied to the Boolean formula directly (=> operator) as described in the Open-PSA MEF. PDAG needs to be aware of all the events in the substitutions because they may not appear in the original fault tree for construction. The Model is passed directly to PDAG to gather substitution events. This is convenient but not the ideal approach. It would be more appropriate to pass only the relevant set of substitutions instead of the whole model. Issue #154
rakhimov
added a commit
that referenced
this issue
Nov 18, 2017
Non-declarative substitutions can only be analyzed with approximations. BDD and prime-implicants cannot apply non-declarative substitutions and guarantee the exact solutions. The analysis must run with minimal cut sets and probability approximations. Issue #154
rakhimov
added a commit
that referenced
this issue
Nov 18, 2017
These substitutions cannot be applied by PDAG at construction. The analysis must handle non-declarative substitutions with sets. Issue #154
rakhimov
added a commit
that referenced
this issue
Nov 18, 2017
The implementation is naive with O(N * M) complexity. It doesn't take advantage of ZBDD features. The reason for suboptimal implementation is that non-declarative substitutions should be avoided in the first place. This will be optimized if there is more need for it in future. In addition, more validation/verification tests are required before taking advantage of ZBDD features. The current solution for Substitution application is adhoc because all/most facilities assume declarative constructs, and non-declarative approaches are bolted-in. Issue #154
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
The construct is unique to the Open-PSA MEF covering the following extra-logical constructs:
The text was updated successfully, but these errors were encountered: