Bionic Building Concept

Abstract

This paper describes a framework for creating a structured series of levels of building automation. It is designed to allow buildings to acquire intelligence about their own systems and equipment and to gradually obtain control of themselves. This allows planners, users, owners and other actors in the planning, construction and use to view and understand the building's operations and performance at many levels. The framework has five horizontal levels and three communication streams. Each level implies a degree of automation with the scale shifting from the mechanization at the bottom to intelligence at the top. The levels describe (bottom to top) the physical object, their representation as data, the history of the data, the processes to analyse and model the histories, and at the highest level, the ability to learn from this analysis to predict, model and plan future building behaviour. Information moves among these levels in an upstream path as well as in a downstream path. The upstream path describes how sensor information is curated to create logs that, using the IFC structure, create semantic histories. In the downstream path, the histories are measured against simulations and model-based predictions to create use-models and potential event sequences. The event sequences then become the instruction sets for the actuators and equipment in the building. Once these are carried out, the effects then feed the sensor data back upstream. In this way, a cycle of information both upstream and downstream feeds a system that can learn. The paper also describes the third stream of communication. This, at each level of the framework, shows how information given to people can be categorized in a scale of increasingly sentient perception. This denotes how the different levels allow users to perceive the building as a purely mechanistic process at the lowest level and as a sentient being at the highest level.