(Georgia Institute of Technology, 2018-08-01)
Cavieres, Andres
The process of Building Design, as in many other forms of design, requires the effective integration of different types of knowledge. However, and in the specific context of Building Information Modeling, only structural knowledge is formally represented. Other types of necessary knowledge, such as those related to the functionality of design, and the set of causal behaviors from which such
functionality is delivered, remain tacit or indirectly referenced by using
structural properties as proxy representations (e.g. geometry). The lack of a more comprehensive and rigorous representational framework to formally describe various behavioral and functional aspects of buildings limits the scope of semantics required to support more effective interdisciplinary collaboration and
design integration. In particular, there is a lack of computational support to
describe cross-cutting behavioral interactions and side-effects that occur among different building sub-systems, which often play a role in the satisfaction of functional goals. To address this problem, the research proposes the development of a
representational framework for the functional and behavioral characterization of building systems and components based on the Functional Representation (FR)
schema developed by Chandrasekaran and Josephson (2000), and its recent
formalization following the DOLCE foundation ontology, by Borgo et al. (2009). A subset of FR axioms has been translated into Description Logic using the Web Ontology Language (OWL-DL) to explore query capabilities of the proposed
framework to support identification of behavioral interactions based on
inference capabilities of available OWL-DL reasoners. The dissertation provides a theoretical basis for the formulation of functional
modeling capabilities currently not available in Building Design. In particular, these capabilities are intended to support the incremental elucidation of behavioral interactions that emerge across different building sub-systems, based on the principle of co-participation of structural entities in a same behavioral phenomena (category of perdurants). The elucidation is expected to be supported by computational inference from structural relations asserted in BIM models by various stakeholders, and at different stages of the design process.