A Pragmatic Theory of Occupants' Indoor-Environmental Control Behaviour

Abstract

Computational tools for building design and operation support entail fairly detailed representations of buildings' geometry, construction, and systems. Recent efforts aim at enhancing, in these tools, the relatively less developed models of building users. Thereby, one of the key challenges concerns the fit between the nature and level of needed support (e.g., performance queries) on the one hand and the required or appropriate resolution of the applied occupant model on the other hand. Some queries involving aggregate performance indicator may be sufficiently served by simple models of occupants' presence and actions in buildings. Detailed queries, however, may necessitate the implementation of high-resolution dynamic occupant representations. Methods to generate an occupant model may be fully data-driven, or they may be based on explicitly stated causal theories of human behaviour. However, there is not necessarily a sharp boundary between these approaches: Regularities harnessed by data-driven methods often reveal an implicit theoretical feature, as they are key to mapping processes from independent variables (model input) to dependent variables (manifest behaviour). Causal methods, on the other hand, need data to both develop and calibrate occupant methods. In this context, the present paper introduces the outline and main elements of a pragmatic theory of control-oriented human behaviour in buildings. The theory is suggested to inform the efforts towards construction of occupant models in computational applications related to building design, operation, and evaluation. Specifically, it can systematically guide the formulation of occupant-related ontologies and their instantiation in computational applications.