Laboratory assessment of rock fracturing state using infrared thermography

Abstract

<p> </p><p>The fracturing state of rocks is a fundamental control on their hydro-mechanical properties at all scales and provides a descriptor of the evolution of brittle deformation around faults, underground excavations, and slopes. Its quantitative assessment is thus key to several geological, engineering and geohazard applications.</p><p>Descriptors of rock fracturing are diverse depending on considered scale, fracture topology (traces, surfaces) and sampling dimension (linear, areal, volumetric). A complete representation of fracture distribution and abundance in a 3D space can be obtained in the laboratory by non-destructive imaging techniques (e.g. X-ray CT), in terms of volumetric fracture intensity (P<sub>32</sub>) and porosity (P<sub>33</sub>). Nevertheless, geophysical imaging is usually unable to resolve small objects in fractured media at field scale. Window and scanline sampling strategies are easily applied in the field to measure fracture intensity descriptors (e.g. P<sub>10</sub>, P<sub>21</sub>) or empirical rock mass quality indices (e.g. GSI), but are affected by scale and fracture orientation biases. Some authors suggested that rock mass fracturing states can be characterized by measuring their heating and cooling response through infrared thermography (IRT), but a physically-based, generalized approach to prediction is lacking.</p><p>In this perspective, we carried out an experimental study on the thermal response of rock samples with known fracturing state. We studied cylindrical samples of gneiss (7) and schist (8), pre-fractured in uniaxial compression that produced complex fracture patterns constrained by rock composition and fabrics.</p><p>Using MicroCT (voxel: 0.625 mm) we reconstructed the 3D fracture network and computed the P<sub>32</sub> and P<sub>33</sub> of each sample. Then, we set up cooling experiments in both laboratory and outdoor conditions. In laboratory experiments, samples were oven-heated at 80°C and let cool in a controlled environment. Sample surface temperature during cooling was imaged in time lapse using a FLIR<sup>TM</sup> T1020 IRT camera. In outdoor experiments, samples underwent natural solar forcing in a daily heating-cooling cycle.</p><p>The acquired multi-temporal thermal images were processed to extract: a) spatial temperature patterns corresponding to the response of individual features and fracture networks at different cooling steps; b) time-dependent cooling curves, described in terms of Cooling Rate Indices and a Curve Factor. These descriptors show statistically significant correlations with fracture abundance measures, stronger with P<sub>33 </sub>than with P<sub>32 </sub>and more robust for gneiss samples, characterized by more distributed fractures than schist. More fractured rocks cool at faster rates and the corresponding cooling curve shapes can be normalized to remove the effects of lithology and boundary conditions to obtain a predictive tool. Experimental results have been reproduced by 3D finite-element modeling of the cooling process in numerical samples including explicit fracture objects. Model results closely reproduce experimental data when fracture surfaces are included as convection surfaces, suggesting that overall sample cooling rates depend on the size of individual blocks forming the sample. Results of outdoor experiments show that differences in thermal response can be significantly detected even in natural conditions. Our results provide a starting point to develop an upscaled, quantitative methodology for the contactless <em>in situ</em> assessment of fracturing state of rock masses using thermal data.</p>