Get Complete Project Material File(s) Now! »
Observation of Deformation and Crack Formation
 Correlation analysis and 3D image analysis were performed to determine the deformation of the sample before fracturing and the geometries of the developing cracks. The spatial fluctuations of the attenuation maps (see Figure 2b) served as markers when the digital correlation technique was used to compare successive images. To calculate the correlation matrix we used a correlation box of size 25 by 25 pixels (125 by 125 Mm2), which enabled us to measure the spatial distribution of micro-displacements.
 Between room temperature and 300°C, the shale dilated anisotropically in the vertical (perpendicular to the shale bedding) and horizontal (parallel to the shale bedding) directions, and the strain curves showed a quasi-linear increase with temperature, as expected for linear thermal expansion (Figure 3a). The coefficient for thermal expansion was determined to be 5.5 * 10 5/°C in the vertical direction and 2.5 * 10 5/°C in the horizontal direction, which clearly indicates the anisotropy of the shale, in agreement with other studies on the same shale [Grebowicz, 2008]. At 300°C, the vertical expansion started to deviate from linearity, which is likely related to the onset of organic degassing before crack formation. At a temperature of about 350°C the sample undergoes rapid localized deformation owing to fluid gen-eration and the onset of fracturing. When the sample breaks, black structures corresponding to the newly formed cracks appear in the images (see Figure 2b), which have no equiv-alent in the preceding ones, thus ruining the correlation technique. Moreover, after the sample fractures, global movement of the sample occurs (translation and rotation), with displacements from one data set to the next of ampli-tude greater than 12 pixels, which was estimated to be the maximum displacement that can be accurately measured using the correlation technique. Even though the correlation results during and after fracturing are not trustworthy, this technique accurately determines the sample deformation before fracturing occurs.
 With 3D tomography, we imaged 3 stages of fracture propagation. The first microfracture pattern was detected at a temperature of about 350°C. Figure 2a shows a 3D rendering of the most opened fractures at T = 391°C (third time step) and Figures 2b and 2c show a vertical slice through the tomography image. The general direction of crack propa-gation followed the shale bedding, and no perpendicular fractures were observed (Figure 2b). Pyrite grains, observ-able as bright spots in the tomography images (Figure 2c), affected crack growth by pinning the crack front, and con-trolling the out-of-plane fluctuations of the crack path.
Organic Decomposition Induces Fracturing in the Shale
 Thin sections were studied in order to compare petrographic characteristics of the shale before and after heating. Before heating, organic precursors, which were preferentially oriented parallel to the shale bedding, could be distinguished throughout the sample (see Figures 1a and 1b). After heating, an abundance of cracks, partially filled with residual organic material, was distributed parallel to the bedding (Figures 1c and 1d). Petrographic observations revealed that cracks propagated mainly in the finer grained layers where the highest concentrations of organic matter lenses were observed (Figures 1a and 1c). The coarser grained layers, where quartz grains and pyrite framboids were present in higher concentration, also displayed better cementation with larger calcite crystals. The preferential location of fracture propagation is ascribed to two main factors: (1) Higher amounts of organic matter (kerogen lenses), which decomposes leading to fluid formation, internal pressure build up and eventually fracture initiation and propagation. (2) Finer grained intervals are less cemented than the coarser grained ones and they fracture more easily.
 The link between hydrocarbon generation and fracturing was tested using thermogravimetry and gas chromatography. Aerobic and anaerobic (nitrogen) thermo-gravimetry analyses on 500 mg samples were performed to investigate the presence of organic and inorganic carbon (carbonates) using a ATG/SDTA 851 Mettler Toledo apparatus. We monitored mass loss of the sample during heating at 10°C/min in air or nitrogen between 20°C and 1000°C. The loss of mass during heating occurred in dis-tinct stages (Figure 3b), and the temperature range of each stage indicates the nature of the component that evaporates. We also used gas chromatography (GC 5890-MS 5973 Agilent) to analyze the gas that escaped (water, CO2, and organic volatiles) during heating at a rate of 5°C/min in air. The first step of mass loss (Figure 3b) in the temperature range 300–450°C corresponds to the release of various organic molecules (alkanes, alkenes, toluene, xylenes), water and CO2 (first peak on CO2 emission plot) and indi-cates decomposition of organic matter. The second step of mass loss around 600–800°C indicates decomposition of carbonates. A similar behavior was observed when nitrogen was used instead of air with two peaks of CO2 release located at the same temperatures.
 Figure 6 shows the correlations between strain evolu-tion, mass loss, CO2 emission and fracture surface area growth. Comparing the temperatures of degassing, mass loss and hydrocarbon release with the temperature of fracturing onset, we conclude that fracturing was induced by overpres-sure in the sample caused by organic matter decomposition.
Discussion and Conclusion
 Time-resolved high-resolution synchrotron X-rays tomography was performed during gradual heating (from 60 to 400°C) of organic-rich immature shales. At 350°C the nucleation of many small cracks was detected. With further temperature increase these microcracks propagated parallel to the shale bedding, coalesced and ultimately spanned the whole sample.
 The central point of our work is the observed corre-lation between hydrocarbon expulsion and fracturing within the sample. To do this, we combined thermogravimetry, gas chromatography, strain analysis and 3D observation of fracture formation and analyze the data in 3 steps:
 1. Analyzing tomograms, we found that fracturing begins at a temperature of about 350°C.
 2. Using thermogravimetry, we determined that the sample started to loose mass in the same temperature range (about 350°C). This alone does not provide us with the composition of the lost material.
 3. Using gas chromatography, we determine that the gases expelled at temperatures near 350°C were mainly oxidized hydrocarbons, and we conclude that they originate from kerogen decomposition.
 In the literature, some shale rocks are reported to contain horizontal (in the direction of shale bedding) mode I fractures as well as vertical (perpendicular to the shale bedding) fractures. The presence of vertical fractures indi-cates that the maximum stress is vertical [Olson, 1980; Smith and Chong, 1984]. However, at shallow depths, where the magnitudes of both the vertical and horizontal stresses are similar, horizontal hydro-fractures of significant lateral extent can be created due to anisotropy of shales [Thomas, 1972; Jensen, 1979]. Horizontal fractures are also observed to develop in clay-rich shales in response to high overpressures during maturation [Littke et al., 1988], even in regions where the vertical stress is larger than the hori-zontal stress.
Table of contents :
1 Hydrocarbon relevance
2 Organic-rich shales and primary migration
3.2 The stress tensor
3.3 Fracture criteria
3.4 Griffith cracks
3.5 Effect of pore fluid pressure
3.6 Hydraulic fracturing in nature
3.8 3D computed tomography
3.9 Image analysis of 3D data sets
4 Analogue experiment
5 Hydrofracturing by phase separation
6 Overview and contribution to the papers