In this paper we lay the foundations for exchanging, adapting, and
interoperating engineering analysis models (EAMs). Our primary foundation
is based upon the concept that engineering analysis models are
knowledge-based abstractions of physical systems, and therefore knowledge
sharing is the key to exchanging, adapting, and interoperating EAMs within
or across organizations. To enable robust knowledge sharing, we propose a
formal set of ontologies for classifying analysis modeling knowledge. To
this end, the fundamental concepts that form the basis of all engineering
analysis models are identified, described, and typed for implementation
into a computational environment. This generic engineering analysis
modeling ontology is extended to include distinct analysis subclasses. We
discuss extension of the generic engineering analysis modeling class for
two common analysis subclasses: continuum-based finite element models and
lumped parameter or discrete analysis models. To illustrate how formal
ontologies of engineering analysis modeling knowledge might facilitate
knowledge exchange and improve reuse, adaptability, and interoperability
of analysis models, we have developed a prototype engineering analysis
modeling knowledge base, called ON-TEAM, based on our proposed ontologies.
An industrial application is used to instantiate the ON-TEAM knowledge
base and illustrate how such a system might improve the ability of
organizations to efficiently exchange, adapt, and interoperate analysis
models within a computer-based engineering environment. We have chosen
Java as our implementation language for ON-TEAM so that we can fully
exploit object-oriented technology, such as object inspection and the use
of metaclasses and metaobjects, to operate on the knowledge base to
perform a variety of tasks, such as knowledge inspection, editing,
maintenance, model diagnosis, customized report generation of analysis
models, model selection, automated customization of the knowledge
interface based on the user expertise level, and interoperability
assessment of distinct analysis models.