COPYRIGHT 2017, NILOUFAR EMAMI, ALL RIGHTS RESERVED

Shade Wall

Presented & published at IASS 2014 Symposium

University of Michigan

2014 - 15

Description: Shade-Wall studies the performance-based design of computationally designed shading walls. The goal of the study is to look at the interaction of geometric variables and their effect on the daylighting and structural performance of the perforated reinforced-concrete walls. This research consists of two main phases. In the first phase, four possible configurations are opted for initial analysis, using Ansys and DIVA-for-Rhino. The second phase adopts ParaGen, which uses Genetic Algorithms (GA), for a more comprehensive search in the design pool. ParaGen uses a Non-Destructive Dynamic Population GA (NDDP GA) which is used to fill a database with solutions generated and analyzed using related software. ParaGen combines form generation and analysis steps which run in parallel on PC clients, with a web server that builds a searchable database and guides the search process. The ParaGen GA can be set to either generate random solutions or to breed solutions based on any combination of geometric or performance parameters. After the creation of an initial set of about 100 solutions, a fitness function is set to focus breeding on higher performing solutions. In this stage, pallets of possible design alternatives coupled with their relative performances values are generated, associated with the graphical representation of the alternatives for qualitative judgment. 

Credits: Niloufar Emami, Anahita Khodadadi, Peter von Buelow