Design Optimization of Non-Uniform Perforated Solar Screens for Enhanced Daylighting in a Tropical Gymnasium
Description
This dataset contains the complete preliminary analysis and optimization results supporting the research article "Design Optimization of Non-Uniform Perforated Solar Screens for Enhanced Daylighting in a Tropical Gymnasium." The study investigates the performance of non-uniform perforated solar screens (PSS) in a wide-span gymnasium located in Bangkok, Thailand. The dataset covers simulations for both uniform and non-uniform PSS configurations to optimize for Spatial Daylight Autonomy (sDA), Useful Daylight Illumination (UDI), Spatial Disturbing Glare (sDG), Annual Average Uniformity (U_avg), and Temporal Uniformity (U_t). The data is provided in two formats: XML (Archive) and CSV (for Visualization). 1. CSV Files (for Visualization) These files are formatted specifically for use with Design Explorer (https://tt-acm.github.io/DesignExplorer/). They are separated by different sections described in the paper. 2.XML File (Archive) This file contains the raw, aggregated data from all simulation runs. It is intended for archival purposes. Variable Dictionary (Data Attributes) The datasets contain the following design variables and performance metrics: Type: Classification of the screen (Uniform / Non-uniform). G (E/W/S): Gradient Contrast for East, West, and South façades (Controls minimum perforation size at the bottom). F (E/W/S): Perforation Sizing Factor for East, West, and South façades (Controls maximum perforation size at the top). PP (E/W/S): Resulting Perforation Percentage for the respective façade. PP (avg): Average Perforation Percentage across all façades. sDA: Spatial Daylight Autonomy UDI: Useful Daylight Illuminance (100-3000 lux). sDG: Spatial Disturbing Glare. E_avg: Average Illuminance Level (lux). U_avg: Annual Average Uniformity (Average of Minimum Illuminance / Average Illuminance for all analyzed daylit hours). U_t: Temporal Uniformity (% of Hours that meets Uniformity Ratio of 0.40 and Spatial Useful Daylight Illumination of 75%)
Files
Steps to reproduce
Software and Environment Setup 3D Modeling: Rhinoceros 3D (Version 8) Parametric Design: Grasshopper (bundled with Rhino). Simulation Engine: ClimateStudio (v2.0) for daylighting and glare analysis Weather File: Bangkok, Thailand (TMYx) https://climate.onebuilding.org/WMO_Region_2_Asia/THA_Thailand/CRG_Central/THA_CRG_Bangkok.Metropolis.484550_TMYx.zip Utilizes a 0.914 m sensor height for daylight (ANSI/IES RP-6-24) and 1.5 m for glare (BS EN 12193) with a 1.0 m spacing. Simulation Parameters (Climate Studio) Samples per sensor 16,384 Ambient bounce 6 Weight limit 0.01 Material Properties (Input Parameters) Walls Reflectance = 84.4% Floor Reflectance = 29.9% Ceiling Reflectance = 64.8% Perforated Solar Screen (PSS): Reflectance = 67.6% Translucent Skylight: Reflectance = 77.0%, Transmittance = 6.0% To overcome sampling bias in a large design space (287,496 combinations), a two-stage hybrid sampling strategy was implemented. Phase I: Discrete Grid Search Objective: To establish an unbiased dataset to analyze broad performance trends across all design categories without the clustering bias of optimization algorithms. Method: Simultion of discrete variables using the Colibri or Fly aggregator in Grasshopper. Sampling Resolution: Uniform PSS (G=0): Variable: Sizing Factor (F). Domain: {0.4, 0.5, 0.6, 0.7, 0.8, 0.9}. Combinations: All permutations across East, West, and South façades (6x6x6 = 216 iterations). Non-Uniform PSS (G > 0): Variables: Gradient Contrast (G) and Sizing Factor (F). Domain G: {0.0, 0.5, 1.0} (Low, Medium, High). Domain F: {0.5, 0.8} (Low, High). Combinations: All permutations across East, West, and South (6x6x6 = 216 iterations). Total Samples: 432 simulations. Phase II: Optimization Objective: To find the global maximum fitness score Method: Model-based optimization using the RBFOpt algorithm (Radial Basis Function) via the Opossum plugin in Grasshopper. Search Space (Continuous): Uniform PSS: G (Fixed): 0.0 F (Variable): Continuous range [0.40 - 0.90] with 0.01 step. Stopping Criteria: 400 iterations. Non-Uniform PSS: G (Variable): Continuous range [0.0 - 1.0]. F (Variable): Continuous range [0.4 - 0.9]. Stopping Criteria: 1,000 iterations. Total Samples: 1,400 simulations. Combined Samples: 432+1,400 = 1832 simulations