Agent-Based Semiology: Optimizing Office Occupation Patterns with Agent-Based Simulations

Abstract

Measuring the performativity of office space has been a long-standing topic of research. With the emergence of knowledge economy, the nature of work has changes considerably, foregrounding personal interaction and information interchange. Consequently, traditional tools of space evaluation, such as space syntax, have become increasingly difficult to apply. This research, therefore, uses a methodology based on agent-based simulation, focusing on agent behavior rather than on space morphology to assess the social performance of spaces. The research process is conducted in two phases: In a first research phase, simple social models are developed for the agent population in order to set up simulations that show differentiated agent behavior toward other agents and architectural frame dependency. A series of simplified yet plausible life-like office event scenarios with strategic changes to the furniture layout is used to evaluate and calibrate the simulation’s social performance. Based on these simulations and the social algorithms derived from them, in a second research phase an experimental setup that follows the generative logics of evolutionary design solving is devised to identify office layouts with the highest social performativity. Methodically searching the design space for performance peaks, algorithmic design is used to generate, simulate, and test an initially large number of random scenarios against a set of predefined success criteria in order to obtain a subset of the most successful configurations.