To maximize the maintainability of the models, it is useful to graphically model down as far as possible, i.e., to postpone the transition from the graphical to equation-based models for as long as possible. Bond graphs are particularly well suited for the modeling of complex physical processes, because they represent the most primitive graphical modeling methodology that is still fully object-oiented. The transition from the bond graph layer to the equation layer is trivial and so generic that it can be programmed out once and for all times. Consequently, the user of a bond graph library should hardly ever face the need to model any physical phenomena using equations.
This library shall first be developed as an independent library that calls internally models of the BondLib library [2]. At a later time, the library shall then be integrated as a sub-library into the BondLib library.
In contrast to the already available library that forms part of the Dymola 6 distribution, the models of the new library shall also contain thermal ports that model the generation of heat. It should be made possible for the user to choose whether these thermal ports are to be externally connected, i.e., an approach should be taken that is similar to the one employed by the electrical sub-library of the BondLib library [3].
A series of test models shall be developed that document and illustrate the use of the new libraries.
The individual models shall be documented in similar ways as this has been done for the models of the BondLib library. In addition, machine readable software manuals shall be generated in HTML format that can, just as in the case of the BondLib library, be called upon from within the new libraries.
A recently developed hydraulics library may serve as an example.