Paper Conference

Proceedings of ASim Conference 2012: 1st Asia conference of IBPSA-China, Japan, Korea


Agent-based geometry control optimized by genetic algorithm for daylighting

Yun Kyu Yi, Bo Rin Jung, John Sullivan

Abstract: This paper proposes a method for daylighting performance-driven building façade generation. The daylighting condition for the building is considered as a domain and is evaluated by advanced light simulation. Forms are generated by parametric CAD toolsand this form is optimized with genetic algorithm (GA), with the objective to find better indoor daylighting conditions. Using the proposed method, engineers and designers may optimize the building configuration to solve the complex combinatorial problems of designing illuminated spaces. GA optimization requires the computationally expensive process of generating a large number of simulation models automatically. This paper documents the implementation of an agent-based control system which decreases the number of daylight simulations required to reach an optimal configuration. The set of parameters that define the façade are controlled in a hierarchical system by a relatively small number of control points or “agent points.” This study demonstrates the scalability and feasibility, as well as reduces the cost in time and processing power of the method.
Pages: 17 - 24