Open Cut » Geology
The project was undertaken to contribute to and improve the analysis and interpretation of data obtained from hyperspectral imagery for characterisation of the overburden material in open cut coal projects. In future geotechnical index parameters might be calculated with these in mind, along with strength indices developed from comminution tests on fragmented particles.
It is hypothesised that the mineralogy, with the focus on different clay types, controls the competence of the rock, with implications for geotechnical assessments and wall stability in open cut setting, including highwall, low wall and spoil piles. This work illustrates the benefits of the classification and characterisation of rocks using hyperspectral imagery on chip samples. It also demonstrates that a comminution device, such as a Short Impact Load Cell (SILC), can be used to obtain tensile strength on chip samples.
This project utilised hyperspectral imagery with associated high resolution digital photography and laser profiling data of the chip samples to estimate the variation in the lithology based properties, such as mineral composition and particle size distribution within the overburden intervals. Particle size of chip samples reflect the comminution response of the different lithologies and could be used as a geotechnical indicator. The hyperspectral imagery was collected by a commercial scanning instrument Corescan HCI-3.2/ 4.2 in Perth, WA, that works with the principles of reflectance spectroscopy in the VNIR-SWIR (Visible and Near-Infrared to Short Wavelength Infrared) wavelength ranges of the electromagnetic spectrum. The mineral classes derived from the hyperspectral data were first analysed manually to capture the relationship between stratigraphy and degree of weathering for igneous and sedimentary lithologies, and then by multivariate statistics using Principal Component Analysis (PCA). Further, PCA was applied to the raw hyperspectral data itself to develop an alternative unsupervised method to combine mineralogy and mineral assemblages with image segmentation results for characterising particles of various size fractions. As a result, proof of concept for unsupervised mineral domains was developed. Additionally, a comparison to geological identifications of fresh and weathered basalt, unconsolidated Quaternary sediments, Tertiary, Triassic Rewan and Permian strata was conducted. The SILC instrument was used to illustrate the tensile properties of select particles of various size fractions, while publicly available data sets on the geomechanical properties of various minerals were consulted for comparison.
A new functional workflow was developed, integrating the available hyperspectral imagery and geotechnical test results of key parameters for rock properties assessment. The main outcome is a demonstration of enhanced utilisation of RC chip samples for geotechnical characterisation. By knowing the mineralogy and/ or dominant mineral assemblages and the average size fraction of the particles, as determined from hyperspectral imagery and laser profiling of the chip samples, the tensile strength of the particles may be predicted. As a result, these techniques can contribute to prediction of highwall and low wall geotechnical assessment, and the overall understanding of rehabilitation behaviour of the generated coal mine spoil piles.