Reliability Block Diagrams (RBD) provide an intuitive modeling of the reliability structure of a system. Ansys medini supports the RBD method as a dedicated editor that can connect to a SysML design and provides the reliability view on top of the modelled components. If no system model is available, RBDs can be used standalone.
A graphical editor supports the editing of an RBD that can span multiple hierarchical design levels. You can create such an RBD Model as follows:
Select a package that is tagged with "FTA Models" (see Property View Contents field)
Right click to open the context menu and select New -> Reliability Block Diagram...
Provide a name in the upcoming dialog and confirm with OK
After creation, the graphical editor opens with a diagram that is initialized with an Input node to Output node connection. Now you are ready to model the RBD structure by using block nodes from the palette or dragging and dropping SysML design from the Model Browser. For more details see RBD Editor Concepts.
Important: Unlike other graphical editors the Delete from Diagram is not available for RBDs. That means there is only one view of a block on a diagram (and the model element cannot be shown elsewhere).
The RBD editor features a number of modeling concepts for parallel, serial, n-out-of-m, and hierarchical modeling. The different RBD nodes are explained in the following:
Input Node ("I" node) and Output Node ("O" node) are the fixed points of any RBD graph. They are created automatically for you and cannot be deleted. Only block nodes connected consistently with a dependency directly or indirectly will be considered for RBD calculation.
Basic Block nodes represent a component failure and will appear in the cut sets after evaluation. A basic block has a name and an autocounter ID and can represent a SysML element or failure. It can have a probability model in the same way than any FTA event, see Event and Gate Properties for more details.
Complex Block nodes provide a means to structure an RBD hierarchically. A complex block is an own RBD graph with Input to Output that is logically inserted at the position of the complex block. Note that you can either create a complex block directly (from the palette) or convert a basic block at a later stage into a complex block by selecting "Turn into complex block" from the context menu
Transfer node allows to reference any fault tree event in the same project. Semantically, this means that a failure event at the transfer gate position is defined by the cut sets of the referenced tree. That means, the cut sets evaluation of the RBD will include all basic events that come from the fault tree.
Voting node allows to specify an n-out-of-m combination for a parallel configuration. The threshold number n specifies how many parallel connections must work for the system to function. Note that this voting node operates in the success space while the corresponding voting gate in FTA operates in the failure space (i.e. it specifies how many failures must occur to let the component/system fail).
Dependencies are the connecting lines between nodes of an RBD graph.
The blocks display a name, probability/failure rate label, as well as frequency. The settings can be adjusted in the Project Settings under FTA/RBD.
The editor features convenient editing capabilities such as drag and drop to insert blocks/nodes. You can select a block/node from the palette or an element from the Model Browser (e.g. SysML part or failure mode) and drop it either onto:
a dependency line to insert the block in series configuration
another block to create a parallel configuration
If you need to create a multiple occurring block, you can drag an existing block on a diagram holding the Alt-key down. Dragging a SysML element or failure mode twice or more often into an RBD will create automatically another occurrence of the block. Note that the "Turn into complex block" feature is not available for multiple occurring blocks.
Beside these fast editing features you can of course freely create, delete, or redirect any dependency lines to build the RBD desired graph that matches your system best. Make sure that eventually all block nodes are connected properly to "I" and "O" nodes.
By default, the RBD editing operations are always re-layouting the graph automatically. If you want your custom positioning or in case you are about to restructure the RBD, you can (temporarily) disable the automatic layout by toggling the Enable/Disable auto layout option in the button bar.
The reliability R(t) can be calculated by evaluating the RBD. The evaluation is triggered as follows:
Open the context menu of an RBD or complex node in the Model Browser or right click onto the diagram in an opened RBD editor
Choose Evaluate RBD... from the context menu
In the upcoming dialog, you can configure the following options (same as described for Probabilistic Fault Tree Evaluation):
Restrict cut set order to n: specifies the order cutoff for the minimal cut set evaluation. This option has no effect on the reliability figures.
Mission time: enter the mission time T (in hours) for which to calculate R(T)
Compute unreliability, PFD/PFH/PMHF, CFI average: this option must be ON. Otherwise the evaluation will not have any effect
Calculate importance measures: optional calculation of Birnbaum, Fussel-Veseley, and Criticality importance measures for basic blocks/events.
Confirm with OK. After successful calculation, the analysis result editor opens that is the same as described for Analysis of Fault Trees. Note that the path highlighting features are not available for RBDs.
In addition to the analysis, the RBD diagram that was selected in step 1 is updated and shows now two labels on the Output Node ("O"):
R: the calculated reliability R(T) in the range [0..1].
T: the mission time for the above R(t)
If you have linked fault trees via transfer nodes, the minimal cut sets will be a mixture of basic blocks and fault tree events.
Sometimes it is required to connect fault trees to an RBD. While the inclusion of a fault tree in an RBD is seamless via the transfer gate, the reverse is currently only supported by transforming the RBD into a fault tree. Thereby the original RBD will not be changed, but a semantically equivalent fault tree will be created:
Select the RBD to be converted into a fault tree in the Model Browser (or RBD diagram)
Open the context menu and select Derive | FTA from RBD
A dialog is coming up that confirms completion. You will find a new FTA model in the same package as the RBD is residing in.
Note that this transformation will only work if the RBD is well-formed. That means all blocks are connected consistently to "I" and "O" nodes.
The derived FTA does not come with a diagram by default. You can create a new diagram, drag the top node to it and choose Restore subtree to quickly get the full fault tree.