A Radical Shift to Probabilistic Thinking: Integrating Energy Analysis in Performance-Driven Building Morphology Generation

Wenjing Li,Wanting Hong,Xinhui Xu,Mehdi Makvandi,Binghan Zhang,Qingchang Chen,Philip F. Yuan
DOI: https://doi.org/10.1016/j.buildenv.2024.112443
IF: 7.093
2024-12-16
Building and Environment
Abstract:In the design phase, analyses of thermal and wind characteristics inside and outside buildings are subject to uncertainty, which affects the precision of energy requirements for building clusters. Moreover, the complex characteristics of urban morphology and microenvironmental data at the street-block scale impact the speed of quantitative calculations of the energy demands of buildings. Intelligent spatial design workflows, driven by artificial intelligence tools and evolutionary algorithm-based self-feedback mechanisms, have been developed to address these issues. In constructing performance-driven simulation workflows for space generation, architectural simulations are influenced by model accuracy, boundary conditions, and application limitations. These problems have different solutions according to the study scope or focus; hence, the proposed solutions are specific and limited. Currently, numerical simulations in the design phase are based on detailed physical models such as EnergyPlus. However, existing algorithms are not tailored for data-driven generative design, which requires the quantification of limited information and has high computational costs, limiting the ability to balance simulation accuracy and computational costs. In this paper, bibliometric analysis is applied to systematically investigate the application of technical methods in energy consumption analyses in the early stage of architectural design. The applicability of forward- and reverse-simulation workflows in performance analyses of environmental systems is analyzed considering differences in the data flow direction. Additionally, considering the data limitations in the design phase, the accuracies of detailed physical models and simplified engineering models in performance-driven architectural space generation workflows are compared. Furthermore, optimization methods, common algorithms, and plugins and their advantages and disadvantages are summarized.
engineering, environmental,construction & building technology, civil
What problem does this paper attempt to address?