HRSBallast: A collection of high-resolution, scanned railway ballast samples

Published: 22 October 2020| Version 2 | DOI: 10.17632/5txdfwdypb.2
Contributors:
André Broekman,
,

Description

HRSBallast is a collection of high-resolution, scanned railway ballast samples obtained from both a field installation and from a local quarry. The samples represent a variety of geometric properties, particularly the extend of rounding owing to in-service conditions experienced and as a function of the track performance. HRSBallast serves as a reference dataset for granular media (GM) simulations utilizing DEM (discrete element method), degradation or wear modelling, digital assets for the creation of synthetic datasets for deep learning applications, embedded railway instrumentation and video games requiring high resolution geometry. The 108 digitized samples are provided in an STL (stereolithography) file format. Both the original resolution files are provided (HighRes.zip) alongside a decimated collection consisting of 1% of the number of points of the high resolution collection (LowRes.zip). 45 of the 108 scans samples represent the ballast samples collected from the field installation (15 samples each from three location). Each of the locations provided either angular, semi-angular of rounded ballast samples. These files are denoted as FA0-FA9 (field samples, angular), FB0-FB9 (field sample, semi-angular) and FC0-FC9 (field sample, rounded). The remaining 63 samples were collected from the fresh ballast samples. Three experiments consisting of 10 samples each were carried out that produced these digital models: + fresh ballast (M00-M09) subjected to 100,000 load cycles in the small-scale box test, scanned again after the test (M10-M19), + fresh ballast placed in a concrete mixer for 15 minutes (M0A-M0J) and subjected to 100,000 load cycles in the small-scale box test, scanned again after the test (M1A-M1J), and + fresh ballast placed in a concrete mixer for 30 minutes (M0K-M0T) and subjected to 100,000 load cycles in the small-scale box test, scanned again after the test (M1K-M1T). Note: an additional 3 files were created from ballast samples that fractured during testing. The mass, volume, surface area and vertex count of each is provided in the spreadsheet / CSV file together with the samples identification number (Ballast_properties.xlsx). A high-resolution photograph of each sample is provided (SamplePhotos.zip). List of files + Ballast_Properties.csv - 6 kb + Ballast_Properties.xlsx - 22 kb + HighRes.zip - 6.8 Gb + LowRes.zip - 66 Mb + SamplePhotos.zip - 320 Mb + scanning_process.MOV - 19 Mb

Files

Steps to reproduce

Ballast samples were collected from both a railway test track near Bloubank in South Africa and from a local quarry for fresh ballast. The ballast collected from the test site already represents a large distribution with respect to the level of attrition experienced by the particles. The ballast samples from the quarry were rounded by placing the ballast in a concrete mixer for either 15 or 20 minutes; this represents a more controlled method of achieving a certain level of roundness. The ballast samples were washed and dried prior to marking each sample with a silver-based marker to uniquely identify each ballast sample. The reflective ink does not affect the scanning procedure. An EinScan Pro handheld 3D scanner (manufactured by Shining3D) was used together with the included turntable to digitize every individual ballast sample. The bundled software bundles with the scanner is used for both the scanning and post-processing of the scanned data. The aggregated point cloud is processed into a "watertight" (manifold) mesh which typically consists of 2 to 4 million points. Larger meshes are decimated to approximately 2.5 million points for easier post-processing. The final mesh is exported as an STL (steoreolithography) file. Using the fixed-mode scanning method with the turntable provides an accuracy of 40 micrometers and a point resolution of 240 micrometers (refer to website reference for technical specifications).

Institutions

University of Pretoria

Categories

Transport, Civil Engineering, Photogrammetry, Smart Transportation, Railway, Stereolithography

Licence