A system with n components each with k behavior
modes, can assume 
overall configurations,
however, in practical situations, only a small fraction of the
configurations are physically realized.
In previous work, researchers have defined a number of approaches, such
as transition functions (e.g., [8]), finite state automata
and switching bonds[2], to handle discrete changes in
physical system configuration and behavior. Most of these methods
assume that the range of system behaviors are pre-enumerated, but, in
general that can be a very difficult task.
Recent compositional modeling
approaches[1, 4] overcome
this problem and build system models dynamically by composing model
fragments. Our hybrid modeling scheme
adopts this methodology, and implements a dynamic
model switching methodology in the bond graph modeling framework.
Instead of identifying a global control structure and pre-enumerating bond graph models for each of the modes, the overall physical model is developed as one bond graph model that covers the energy flow relations within the system. Discontinuous mechanisms and components in the system are then identified, and each mechanism is modeled locally as a controlled junction which can assume one of two states - on and off. The local control mechanism for a junction is implemented as a finite state automaton and represented as a state transition graph or table.