Quantifying the influence of continuous and discrete design decisions using sensitivities
Early-stage design decisions have a large impact on the performance of buildings. Especially the case of mixed (continuous and discrete) design variables can complicate human and computer understandings of how individual choices contribute to performance, in both direction and intensity. We propose a new approach upon conditional variational autoencoders and sensitivity analysis for assessing the influence of mixed variables on performance and test in on a gridshell design problem.
In response, we propose a new approach for assessing the influence of mixed variables on performance based on variable sensitivities, with a focus on architectural applications… Both local and global insights are attainable: 1) the method computes the influence of design decisions at a design instance of interest, and 2) local results can be aggregated to summarize variable influence in the global design space. The method is demonstrated on gridshell structures parameterized with continuous and discrete variables and assessed in terms of embodied carbon.
Related publications:
D. Fang, S. V. Kuhn, W. Kaufmann, M. A. Kraus, and C. Mueller, “Quantifying the influence of continuous and discrete design decisions using sensitivities,” in Advances in Architectural Geometry 2023, Stuttgart, Germany, Oct. 2023. https://doi.org/10.1515/9783111162683-031.
