Carnegie Mellon University
Browse
- No file added yet -

Automated bond graph modeling and simplification to support design

Download (245.5 kB)
journal contribution
posted on 1991-01-01, 00:00 authored by J Rinderle, L Balasubramaniam, Carnegie Mellon University.Engineering Design Research Center.
Abstract: "Designs evolve from a preliminary concept by a process of iterative evaluation and refinement. Every cycle of this process necessarily consists of assessing the extent to which functional specifications have been satisfied. It is hypothesized that providing a tool for automatically modeling and analyzing devices and relating individual component characteristics to device behavior would aid the conceptualdesigner by facilitating the consideration of more varied alternatives. In this paper we discuss issues pertaining to automated modeling tosupport design including issues of modeling relevance and model simplification.We propose a framework for automatic modeling based on the modular aggregation of bond graph fragments corresponding to components and connections. Although a component based modeling paradigm is convenient for the designer, it leads to bond graph models which are difficult to analyze and difficult to comprehend due to the presence of redundant, irrelevant and circuitous structure and due to a high degree of dependency among energy storage elements. Bond graph simplification methods have been developed that mitigate these problems. These methods identify and eliminate inert elements and redundant and constraining junction structures and replace dependent energy storage elements with functionally equivalent modifications to the remaining storage elements.The application of these methods makes it possible to apply standard equation formulation techniques and results in simple bond graph modelswhich can more easily be interpreted by the designer."

History

Publisher Statement

All Rights Reserved

Date

1991-01-01

Usage metrics

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC