Login
- Title:
AirfoilMNIST: A Large-Scale Dataset based on Two-Dimensional RANS Simulations of Airfoils
- Document type:
- Forschungsdaten
- Publication date:
- 05.08.2024
- Responsible:
- Schlichter, Philipp
- Authors:
- Schlichter, Philipp; Reck, Michaela; Widmann, Sebastian; Indinger, Thomas
- Author affiliation:
- TUM
- Publisher:
- TUM
- Identifier:
- doi:10.14459/2024mp1745500
- End date of data production:
- 31.05.2024
- Subject area:
- DAT Datenverarbeitung, Informatik; MAS Maschinenbau; MTA Technische Mechanik, Technische Thermodynamik, Technische Akustik; TEC Technik, Ingenieurwissenschaften (allgemein)
- Resource type:
- Simulationen / simulations
- Data type:
- Bilder / images
- Description:
- AirfoilMNIST is a dataset of two-dimensional, RANS-based flow fields for NACA 4- and 5-series airfoils. The dataset consists of around 148,000 simulations for Mach numbers up to 0.6 and angles of attack ranging from −5 ≤ α ≤ 15.
The dataset is divided into five subsets:
- AirfoilMNIST-raw: Raw CFD output from the NacaFOAM pipeline. Size: 3.5TB.
- AirfoilMNIST: Preprocessed version of AirfoilMNIST-raw subset. Size: 500GB.
- AirfoilMNIST-incompressible: Subset of AirfoilMNIST with samples with M < 0.3. Total samples: 78,010. Size: 50GB.
- BluffBodyExtension-raw: Raw CFD output from the BluffFOAM pipeline with 11,760 samples. Size: 125GB.
- BluffBodyExtension: Preprocessed version of BluffBodyExtension-raw subset. Size: 12GB
Preprocessed subsets have been interpolated onto a 256x256 uniform grid. For every combination of geometry and angle of attack, one random velocity is put in the test split. - Method of data assessment:
- The data for the AirfoilMNIST dataset was acquired using open-source tools, specifically Python and OpenFOAM. No proprietary software was used in the data acquisition process. The CFD data was generated using the server capacities from the chair of aerodynamics and fluid dynamics, which includes 15 server modules, each with an average of 12 cores and a RAM size of 128GB. Post-processing of the simulations was done using four NVIDIA GTX1080Ti GPUs.
- Links:
- Details and the sourcecode for the data generation are available via our homepage: https://gitlab.lrz.de/2DAeroML
- Key words:
- AirfoilMNIST; dataset; CFD; airfoil; RANS; compressible; incompressible; neural networks; aerodynamics
- Technical remarks:
- View (3,6 TB total, 312004 Files)
The data server also offers downloads with FTP
The data server also offers downloads with rsync (password m1745500):
rsync rsync://m1745500@dataserv.ub.tum.de/m1745500/
Note: folder "airfoilMNIST-raw" is downloadable only via rsync - Language:
- en
- Rights:
- by-sa, http://creativecommons.org/licenses/by-sa/4.0
BibTeX
Occurrences: