SCRAM uses XML input files to describe analysis. One model per run is assumed. See XML File Helper Tools for more convenient writing and reading of XML input files.
The input file format follows the Open-PSA Model Exchange Format ([MEF]) version 2.0d or later. The extensive description is given in the above format documentation by Open-PSA, and input files should be straightforward to create and understand. However, not all Open-PSA formatting is supported, and some additional assumptions/restrictions are made by SCRAM. See Supported Open-PSA MEF Features for the format description and current implementation with differences.
In addition to the XML format, the Aralia Input Format is supported indirectly.
Steps in XML Input Validation¶
Several input files to describe one model.
- The first file must define the name, label, and attributes of the model. This information about models in the other input files is ignored without a warning, which allows reuse of files with analysis constructs from other models.
The validation assumptions/requirements:
- Construct names and references are case-sensitive.
- Leading and trailing whitespace characters are trimmed.
- Names and references should conform to the MEF recommendations.
- Public names must be unique globally, and private names must be unique locally within containers.
- References are followed according to the public and private roles described in the Open-PSA MEF section IV.3.2. Solving Name Conflicts: Public versus Private Elements.
Additional validation of fault trees and values of parameters:
- Each gate or formula must have the correct number of arguments.
- A duplicate argument in Boolean formulas or gates is an error.
- Values of expressions and parameters must be valid, i.e., non-negative for probabilities.
- All events must be explicitly defined for probability calculations.
Error messages (with a file name, line numbers, types of errors):
- Report a cyclic tree (gates or event tree branches).
- Report a cyclic parameter with expressions.
- Report a cyclic rule with instructions.
- Report missing element descriptions.
- Report an error if an event or a construct is being redefined.
Warnings for potential errors:
Unused model constructs:
- Primary events
- Event trees
- Initiating events or event groups
- Functional events, branches, sequences, rules
Aralia Input Format¶
A more convenient format than the XML for writing simple fault trees
utilizes a shorter notation for gates, operators (
^), and events
to create a collection of Boolean formulas.
The Aralia format can be converted into the XML format with this script.
Input File Examples¶
Fault Tree Input File¶
<?xml version="1.0"?> <opsa-mef> <define-fault-tree name="TwoTrains"> <define-gate name="TopEvent"> <and> <event name="TrainOne"/> <event name="TrainTwo"/> </and> </define-gate> <define-gate name="TrainOne"> <or> <event name="ValveOne"/> <event name="PumpOne"/> </or> </define-gate> <define-gate name="TrainTwo"> <or> <event name="ValveTwo"/> <event name="PumpTwo"/> </or> </define-gate> <define-basic-event name="ValveOne"> <float value="0.5"/> </define-basic-event> <define-basic-event name="ValveTwo"> <float value="0.5"/> </define-basic-event> <define-basic-event name="PumpOne"> <float value="0.7"/> </define-basic-event> <define-basic-event name="PumpTwo"> <float value="0.7"/> </define-basic-event> </define-fault-tree> </opsa-mef>
TwoTrains TopEvent := (TrainOne & TrainTwo) TrainOne := (ValveOne | PumpOne) TrainTwo := (ValveTwo | PumpTwo) p(ValveOne) = 0.5 p(ValveTwo) = 0.5 p(PumpOne) = 0.7 p(PumpTwo) = 0.7