A Computational Method for Quantitative Post Occupancy Evaluation of Occupants’ Spatial Behavior in Buildings

Abstract

This dissertation proposes a novel computational method for quantitatively describing the spatial behavior of building occupants to be used as a complement to qualitative techniques in post-occupancy evaluations.

The main elements of the proposed method are, first, a comprehensive computational assessment framework of behavioral variables and metrics that describe a wide variety of patterns of presence, movement, and spatial activity; and second, a computer vision algorithm that captures anonymous, high-resolution spatio-temporal data in a more efficient and accurate manner than comparable benchmarks. The proposed method is conceptually grounded in an architecture-oriented redefinition of the notion of spatial behavior, and in a thorough analysis of the limitations of current post-occupancy evaluation protocols.

The method was evaluated and validated in a series of case studies of post-occupancy evaluations of occupants’ spatial behavior in different university buildings. The results demonstrated the robustness, convenience, and applicability of the method, even in challenging situations of complex spatial configurations and privacy-sensitive environments. The method is simple to implement, flexible to adapt to multiple contexts, and cost-efficient, making it potentially competitive for scalable massive applications.

Ultimately, the study contributes to the use of computational methods to enhance the flow of information from occupation to design and the development of a data-driven culture in the building industry.