Skip to main content
SIGraDi 2022 | Critical Appropriations

Full Program »

177: Performance predictive metamodel for dynamic facade shading
Original title: Metamodelo preditivo de desempenho para sombreamento por fachada dinâmica

Research in PORTUGUÊS

Dynamic shading compositions present a challenge for computer simulation of illuminance performance still in the design phase of the project. Such architectural elements have a conflicting function, shading without blocking natural light. The evaluations of these dynamic elements present an infinity of parameters and several combinations, resulting in compositional complexity and making it difficult to read and understand the performance. The objective of this research is to identify the optimized positioning of dynamic facade shading elements. The dependent variable to be analyzed is the annual average illuminance. The algorithm that creates the shape of the Shoebox and manipulates the independent variables is the same that performs the annual daylight assessment. Simulation results are the metamodel input data, Machine Learning is used for the optimization. The contribution of this research is to test the set of results of the independent variables, training an algorithm capable of replacing the simulation.
Machine Learning, Metamodel, Dynamic Facade, Performance Evaluation.

Cassio Daher
c209606@dac.unicamp.br
Universidade Estadual de Campinas
Brazil

Regina Ruschel
ruschel@unicamp.br
Universidade Estadual de Campinas
Brazil

 


Powered by OpenConf®
Copyright ©2002-2021 Zakon Group LLC