AI Building EnergyEvidence Atlas
Home / Topics
Research theme

AI-Based Building Design, Microclimate and Adaptive Reuse

These studies move AI upstream into environmental input generation and design decision-making. They include wind- and morphology-aware microclimate prediction, diffusion-based energy-aware residential layout generation, and neural-network assessment of adaptive-reuse expansion strategies using real building data.

Examples of residential floor plans generated by different AI layout-generation models.
Figure 12. Comparison of generated residential floor plans under fixed outer-contour input conditions.
Figure excerpt from Zeng et al. (2025), Building Simulation. DOI 10.1007/s12273-025-1337-4
Source-attributed paper figure
Workflow combining wind-driven building morphology maps, deep learning, and building-energy simulation.
Figure 1. Workflow for temporal microclimate prediction using wind-driven building morphology maps and subsequent building-energy simulation.
Figure excerpt from Li et al. (2025), Building Simulation. DOI 10.1007/s12273-025-1365-0
Source-attributed paper figure
Urban site-context model and parametric modeling interface for adaptive reuse analysis.
Figure 2. Urbano/Grasshopper workflow and the generated urban site-context model used in the adaptive-reuse analysis.
Figure excerpt from Garboushian et al. (2025), Building Simulation. DOI 10.1007/s12273-025-1310-2
Source-attributed paper figure

Questions addressed

  • How is AI used before building operation begins?
  • How can microclimate and morphology improve energy simulation?
  • How can generative models optimize energy-efficient layouts?
Source figures

Methods and applications represented in this theme

Examples of residential floor plans generated by different AI layout-generation models.
Figure 12. Comparison of generated residential floor plans under fixed outer-contour input conditions.
Figure excerpt from Zeng et al. (2025), Building Simulation. DOI 10.1007/s12273-025-1337-4
Source-attributed paper figure
Workflow combining wind-driven building morphology maps, deep learning, and building-energy simulation.
Figure 1. Workflow for temporal microclimate prediction using wind-driven building morphology maps and subsequent building-energy simulation.
Figure excerpt from Li et al. (2025), Building Simulation. DOI 10.1007/s12273-025-1365-0
Source-attributed paper figure
Urban site-context model and parametric modeling interface for adaptive reuse analysis.
Figure 2. Urbano/Grasshopper workflow and the generated urban site-context model used in the adaptive-reuse analysis.
Figure excerpt from Garboushian et al. (2025), Building Simulation. DOI 10.1007/s12273-025-1310-2
Source-attributed paper figure

Comparison of selected papers

PaperResearch problemMethodReported evidence
Li et al. (2025)Typical meteorological year files may not represent local urban microclimates well enough for accurate building energy simulation.A deep-learning model integrates seasonal wind dynamics with building-morphology maps encoded through frontal-area information.Campus validation reports lower temperature and humidity RMSE than TMY and benchmark models; when used in simulation for three buildings, cooling-energy prediction error is below 2%.
Zeng et al. (2025)Residential layout generators often satisfy spatial constraints without explicitly optimizing operational energy performance.The framework combines a deep-learning energy predictor trained on 71,125 RPLAN layouts, a diffusion-based floor-plan generator, and genetic-algorithm refinement.The predictor exceeds 99% accuracy for five monthly energy categories, and the optimized layouts reduce energy use by 17.5% relative to the best baseline under matched conditions.
Garboushian et al. (2025)The energy and daylight consequences of adaptive-reuse expansion strategies are not well understood across real urban buildings.A neural-network simulation framework uses multi-building data and variables including floor count, renewable generation, glazing, and expansion direction.Validation against actual building use yields a 7.88% average error; east-west expansion produces 41% greater energy-demand reduction than southward expansion in the reported study.

Paper pages