Danielson, J, Ellis, J, Schultz, C, Wilton, E, Schafer, K & Stilwell, I 2025, 'Automated televiewer data interpretation and validation: 
an applied case study at Kennecott copper', in JJ Potter & J Wesseloo (eds), SSIM 2025: Fourth International Slope Stability in Mining Conference, Australian Centre for Geomechanics, Perth, https://doi.org/10.36487/ACG_repo/2535_48
(https://papers.acg.uwa.edu.au/p/2535_48_Danielson/)
Abstract: Manual interpretation of borehole televiewer data for geological and geotechnical site characterisation is essential, but hampered by considerable time and cost investment; variation in interpretation (inconsistency); and critical orientation biases – including a "visual salience bias" against low-angle features often relevant to slope stability. This paper introduces Structura, a tool employing deep learning to automate the interpretation of acoustic and optical televiewer logs, identifying structures, breakouts and drilling-induced tensile fractures, estimating roughness, and assessing borehole image quality. We present results from a collaborative pilot study at Rio Tinto Kennecott Copper, where Structura was compared against manual interpretation, refined based on operational needs, and validated using blind test data and physical core comparisons. Validation showed good agreement on structural orientations with potential bias reduction, a positive correlation for automated roughness requiring resolution-based adjustment, and reliable identification of breakout features for stress analysis. Structura demonstrates the potential of deep learning-based tools to significantly improve the efficiency, consistency, and objectivity of televiewer data analysis; complementing traditional oriented data collection workflows.

Keywords: deep learning, data processing, automation, structural characterisation, televiewer data, machine learning, borehole image processing