Dataset of Wind Pressure Coefficients for Natural Ventilation Modeling in H-Shaped Buildings
Authors/Creators
Description
Overview
This dataset is the primary research data repository associated with the manuscript “Development of a Wind Pressure Database for Natural Ventilation Modeling in H-Shaped Buildings Using CFD Simulations.”
It provides wind pressure coefficient (Cp) data for 180 low-rise H-shaped building geometries obtained using computational fluid dynamics (CFD). The building geometries are parameterized using four dimensionless variables describing their proportions:
-
P1 = L/B: Plan aspect ratio
-
P2 = D/L: Relative cavity depth
-
P3 = W/B: Relative cavity width
-
P4 = H/B: Relative building height
A total of 1,260 steady-state Reynolds-averaged Navier–Stokes (RANS) CFD simulations were conducted, covering seven wind incidence angles from 0° to 90° in 15° increments for each geometry. Exploiting the geometric symmetry of the H-shaped buildings, the surface-level pressure statistics were reconstructed over the full angular domain, from 0° to 345° at 15° intervals.
Dataset contents
The repository contains two complementary files:
1. H-shaped_surface_Cp_statistics.csv
This file contains surface-level pressure statistics for each building geometry, wind incidence angle, and building surface, including:
-
Surface-averaged wind pressure coefficient (Cp)
-
Minimum and maximum local Cp values
-
Spatial standard deviation of the local Cp distribution
-
Identification of directly simulated and symmetry-reconstructed wind directions
The CSV file provides a compact representation of the pressure coefficients suitable for airflow network modeling, building performance simulations, and statistical analyses.
2. H-shaped_CFD_Cp_fields.zip
This archive contains spatially resolved CFD-derived pressure coefficient fields in Visualization Toolkit (VTK) format for the seven directly simulated wind incidence angles.
The VTK files preserve the building surface mesh geometry and local Cp values, enabling detailed visualization, post-processing, and spatial analysis of wind pressure distributions.
ANN surrogate models and web application
Complementary to the CFD dataset, surface-specific artificial neural network (ANN) surrogate models were developed to predict surface-averaged wind pressure coefficients (Cp) for intermediate H-shaped building geometries and wind incidence angles.
These surrogate models are accessible through the H-shape Cp web application:
https://hshapecp.streamlit.app/
The application provides Cp predictions for building configurations and wind directions within the investigated parameter ranges:
-
P1 (L/B): 0.50–2.00
-
P2 (D/L): 0.05–0.30
-
P3 (W/B): 0.30–0.50
-
P4 (H/B): 0.25–1.00
The application enables users to obtain surface-averaged Cp predictions without performing additional CFD simulations. The trained ANN models are accessed through the web application and are not included in the downloadable files of this repository.
Intended applications
This dataset is intended to support research and practical applications involving:
-
Wind-driven natural ventilation modeling
-
Airflow network (AFN) simulations
-
Building energy and performance simulations
-
Building aerodynamics and wind engineering
-
Analysis of surface pressure distributions on irregular building geometries
-
Development and validation of data-driven wind pressure prediction models
The VTK fields provide detailed spatial pressure information, whereas the CSV dataset offers surface-level statistics suitable for simplified airflow calculations. The complementary ANN application extends the database to intermediate building configurations and wind incidence angles.
Associated manuscript
Development of a Wind Pressure Database for Natural Ventilation Modeling in H-Shaped Buildings Using CFD Simulations.
Preprint DOI: https://doi.org/10.5281/zenodo.23757938
License
This dataset is distributed under the Creative Commons Attribution 4.0 International (CC BY 4.0) license. Reuse and redistribution are permitted with appropriate attribution to the original authors.
Files
H-shaped_CFD_Cp_fields.zip
Additional details
Related works
- Is supplement to
- Preprint: 10.5281/zenodo.23757938 (DOI)