[Back]

A. Mahdavi:
"The middle way and the triple world: Toward co-existantial representations for aggregate design-performance models";
1995.

Structure and behavior - The multi-disciplinary nature of building design has implications for developmental strategies in computational design support. Research in this area has been frequently
conducted under the assumption that multiple behavioral modeling applications may be fed by a single structural model of the design artifact. The term building representation then has been mainly
associated with this latter model. Dynamic behavioral modeling tools have been conceived as customers of this single static building representation. However, this need not be. We may opt for treating behavioral models as integral to the representations of design artifacts. This distinction is not a matter of semantic finesse. It has important consequences for strategies in computational design
support. Keeping behavioral modeling outside the realm of design artifact representation inadvertently results in their reduction to black-box entities. The design artifact representation itself ends up as a static structure indifferent to and uninspired by the representational richness of domain applications. This circumstance may have had negative implications for integration efforts.
The problem of integration - Integration in computer-aided design denotes systematic incorporation of multiple domain applications within a unified computational design support environment. At one
end of the spectrum of common integration efforts, there is a top-down approach involving an all- encompassing maximal building representation. an the other end, there is a bottom-up approach
involving the ad hoc and as-needed production of translator and mediator routines to enable various existing applicatior.s to communicate with each other. W
However, neither of these approach es has been as successful as one would wish. One reason ~this
relative stagnation in integration research may be due, in part, to the "non-integrated" informational
context and short-termfoci of the professional communities involved. There are those whose primary
interest is in a broad view of building design. But they may be less familiar with domain-specific
knowledge in the behavioral modeling of buildings. Thus, they often treat technical applications (such
as building performance simulation routines) as isolated black-box type modules, without exploring
the characteristics of their internal representation of the building. Often, they reduce the integration
problem to interfacing efforts between view-independent all-encompassing product models and
existing domain applications (with their view-dependent building representations). Paradoxically, this
latter reductionist approach may have been reinforced by a second group of researchers who have been
primarily involved in the development of technical domain applications. They are typically interested
in fast solutions for the communication of design information to the application and have thus viewed
CAD systems mainly as service utilities (i.e. augmented drafting systems or glorified user interfaces)
for application modules. It is not difficult to see why this attitude does not naturally nurture a
systematic and explicit treatment of multiple building representations, and the extent to which such
representations may be brought together in a shared coherent scheme.
[...]

none - see english version