UC Berkeley

This dissertation describes a behavior simulation model which could reveal spatial usage pattern of small urban places in association with the tangible physical built environment and the intangible microclimatic conditions. The simulation model extends the existing environmental assessments which do not include human interactions and human behavior simulations which do not take into account microclimatic conditions. It addresses the problem of predicting and evaluating the impacts of microclimatic conditions on the spatial usage of its human inhabitants, which has been studied, but not successfully implemented in the past.

The concept is to carry out a virtual post occupancy evaluation which virtual users would exhibit their spatial usage pattern according to the environmental conditions and physical environment so experienced. Evaluators then could see how virtual users perform under certain conditions and assess if the design intention is achieved.

The simulation model is developed using Wei Yan's (Yan, 2005) urban simulation model as a base. In addition Yan's usability-based building model and agent-based behavior model, an environmental model is included. It manages environmental data from user input, database and other simulation programs. It records the shading and air speed profile onto the building model to affect individual user behavior, develops outdoor activity comfort maps which virtual users would use to develop activity option maps, and control collective usage pattern in terms of level of usage, activity type distribution and spatial distribution.

To ensure quality results, the environmental model handles only relatively simple computation, like generating outdoor activity comfort maps. It does not handle assessments which require specialized algorithms and processes like air flow simulation and comfort evaluation. Rather, these assessments are handled by certified programs outside the system. The environmental model processes the result and distributes the information to the building model and the agent model to control the behavior of the virtual users at both collective and individual levels.

At the collective level, the behavioral pattern is mainly governed by statistics obtained from a year round field study. At individual level, the behavior is based on well studied and defined behavioral rules which are derived from theoretical and practical environment-behavior studies.

The simulation generates both graphical and textual outputs. The two dimensional graphical output offers views of paths of virtual users; the three dimensional output offers shows spatial usage as a whole; and textual output provide information of usage and thermal states.

It is expected the result of this research to change how architects and environmental behavior experts will approach the design and evaluation of built environments.