A Study of Multiple Feedback Loops to Optimize Parametric Evolution of Building Envelopes

Abstract

Parametric approaches to architectural design have the potential to integrate form generation with functional analysis, using multiple feedback loops for iterative synthesis and evaluation. This study explores multiple feedback loops to optimize the parametric evolution of building envelopes, integrating form generation with functional analysis. The study's first objective involves plotting the design process and activities in various parametric approaches to architecture, which are imitation, optimization, and form-finding based. This analysis provides a comprehensive understanding of the variations in the design processes and data flow inherent in these parametric design approaches. The second objective focuses on exploring, identifying, and organizing multiple feedback loops that inform the iterative evolution of building envelopes, ensuring that designs are functional and apt for the user. Finally, the third objective seeks to develop a structured decision-tree-based framework incorporating multiple feedback loops. This facilitates a rapid generation of efficient conceptual building envelope designs and optimization of parametric form evolution during the early design stages that can be further developed and detailed. This decision-tree framework, developed as an algorithm in the Rhino-Grasshopper environment, generates forms by combining user preferences and computational methods. Two design aids were developed to navigate the algorithm: a questionnaire for capturing qualitative user preferences and a correlation matrix for analyzing relationships between design parameters and performance metrics. These design aids facilitate informed decision-making through the algorithm. This algorithm also allows users to abstain from selecting values for the parameters, leaving them as variables for Multi-Objective Optimization (MOO) in the subsequent optimization stage. This decision-tree-based framework, supported by user preference inputs, narrows the design search space for effective MOO. The MOO process utilizes the NSGA II algorithm to generate Pareto optimal solutions. These optimal solutions are ranked based on the weighting of factors in the analytical hierarchy process (AHP) to get the desired conceptual form to develop further. This framework, validated by professionals in architecture and related fields, effectively integrates quantitative and qualitative feedback loops in parametric design. When tested with users, this algorithm showed a significant computational advantage (faster convergence) and a iv better precision (lower standard deviation) in the MOO-based parametric evolution of conceptual building envelope forms while maintaining user preferences. By achieving these objectives, this research contributes to a better understanding of feedback loops toward an optimized parametric evolution of building envelopes and creates a framework for quickly generating efficient and apt conceptual building envelope forms.