Expandable Connectors


As we saw in the Thermal Control example from this chapter, when building architectures it can be difficult to define an interface that covers all important configurations. One option, in this case, is to use expandable connectors.

An expandable connector uses exactly the same syntax as a normal connector definition (see previous discussion on Connector Definitions) with the exception that the expandable qualifier is added before the definition, e.g.,

expandable connector ConnectorName "Description of the connector"
  // Declarations for connector variables
end ConnectorName;

We can ignore the expandable qualifier and just treat this connector like a normal connector if we want. The assumption is that components that use such a connector as a normal connector will provide the correct number of equations such that any component models are Balanced Components.

If, however, a user connects to a named element that is not part of the connector definition, this causes a new element to be added to that particular instance of the connector. The type of this new element will be identical to the connector it is connected to. Ultimately, that element will be added to any other connector instances in the connection set as well.

In this way, an expandable connector definition can be “grown” to add additional signals not in the original definition. In this way, any interface definition with an expandable connector has the flexibility to exchange more information than just what was in the initial connector definition. As we saw in the Thermal Control example, this can be very useful when the interface can vary significantly depending on the choice of subsystems.

By far, the most common use case for expandable connectors is to add additional input and output signals to a connector. This frequently occurs because the information required when pairing controllers up with sensors and actuators can change depending on the implementation details of those subsystems. For example, an internal combustion engine with cam phasing will require a commanded cam angle from the controller. But an engine without this feature will not.

However, the expandable connector can also be used with acausal sub-connectors to add additional points of physical interaction between subsystems. For example, one subsystem may not include thermal effects while another one does. The one that does could include a thermal acausal connector to allow other subsystems to interact thermally with it.


Expandable connector definitions define a minimum set of variables that should appear on the connector. It is assumed that all component models will be balanced with respect to those variables in the connector definition. In addition, additional variables can be added to an expandable connector as long as the overall system model is still balanced. Expandable connectors are typically used in conjunction with interface definitions to define a minimum interface within an architecture that can be expanded by the choice of implementations.