As expected, when the high viscosity fluid is injected into the model, the crack initiation pressure becomes lower compared to the low viscosity fluid. More specifically, the particles start to separate at an earlier stage (9.2×1015 Pa) due
to the bond breakage. This means that the inter-particle forces become larger than the maximum bond strength resulting in bond breakage. Moreover, the fluid pressure which leads to failure (about 3000 micro-cracks) is measured to be about 2.59×1017 Pa, which is lower compared with the low viscous fluid
(about 3.5×1017 Pa). The fact that the crack initiation pressure, as well as the
pressure that leads to failure is lower for the high viscous fluid, is opposite compared with the observed behaviour in samples from other reserachers (Shimizu et al., 2011; Ishida et al., 2012), indicates the influence of the pre- existing fractures to the cracking mechanism. More specifically, in intact rocks
the breakdown pressures tend to be lower for low viscous fluids, compared with high viscous ones. Therefore, the presence of pre-existing fractures within the sample, allows more space for the fluid to travel, combined with the fact that the low viscous fluid can travel easier through the fractures, thus require additional pressure build up in order to reach the same amount of damage. Moreover, in both cases of the low and high viscosity fluids, tensile cracks (normal, with respect to the bond plane) are dominant, over the shear cracks, with their percentage being about 81% for the LIM1_fluid1 and about
89% for the LIM1_fluid2 sample, as shown in Fig.5.29.
Table 5-1 includes the measured values for the crack initiation and the failure pressure, as well as the total number of micro-cracks in the tensile and shear directions for both cases of the low and the high viscosity fluids, respectively.
Furthermore, it is observed that the cracking behaviour of the model is more aggressive; cracks propagate further under the influence of the high viscosity fluid and that the overall damage expands in several areas (Fig.5.30 (A), (B), (C)), creating secondary branches, rather than that of a propagation as a main volume. Fig. 5.30 illustrates the state of the damage of the model when it reaches about 3000 micro-cracks.
It can be observed that in the case of water (top picture) the cracks appear to expand as a unity, whereas in the case of the high viscosity fluid (bottom picture) the cracks appear to abandon the central volume of cracks and are reconfigured in individual groups that propagate further, thus covering the distance from the injection point up to the base of the fracture in the xz plane (red circles).
Figure 5.29 Microcracks in the normal and shear direction versus the
injection rate for the (a) LIM1_fluid1 and (b) LIM1_fluid2 samples,
respectively. (a)
Table 5-2 Crack initiation and failure pressure and the total number of micro-
cracks in the normal and shear directions for the low and high viscosity models.
Fluid pressure Micro-cracks
Crack initiation Pressure Pi (Pa) Failure pressure Pf (Pa) Normal direction Shear direction LIM1_fluid1(low) 3.0×1016 3.27×1018 2440 536 LIM1_fluid2(high) 9.2×1015 2.59×1017 2680 310
The objectives of Chapter 5 are the computational modelling of a hydraulic fracturing test for a naturally fractured limestone sample, the analysis of its mechanical behaviour and the interaction between the natural fractures and the new hydraulic fractures. A parametric study of (i) the angles of invidual induced fractures, as well as induced fracture network, (ii) the external stress regime, and (iii) the fluid viscosity, attempts to shed more light on how a fractured rock and the aforementioned parameters can influence and possible enhance the fracking process.
It analyses the mechanical response of the rock model due to fluid flow by using the fluid-couple DEM code in a number of hydraulic fracturing simulations. It involves detailed monitoring of the initiation/propagation of micro-cracks, analysis of the stresses in different regions within the rock’s
matrix and evaluation of the relation between the energy release and the development of cracks. Observations of the simulated fracking tests show that the angle of the fracture directly relates with the stress pattern within the model, thus affecting the direction and propagation of cracks.
Figure 5.30 Expansion of micro-cracks before the termination of the test for
the low viscosity LIM1_frac1 (top) and the high viscosity LIM1_frac2
(bottom) models.
It can be concluded that the cracking behaviour for angles below 45o is
followed by high stresses and expands mainly downwards as a group of cracks, whereas for angles above 45o the microcracking forms clusters that
stray from the main volume of cracks. In the single pre-cracked samples, the fracture is mainly observed towards the horizontal X axis, namely along the direction of the maximum compressive principal stress, and this is in agreement with the conventional theory, whereas in the case of the pre- carcked sample with multiple fractures the overall fracture is extended perpendicular to the maximum compressive stress. In addition, a relation between the important cracking events (large increases of micro-cracks) in
each model with the energy release within the models, has been observed. Finally, highest stresses have been observed in the upper part of the assembly, near the fracture tip, whereas the measurements taken from regions away from the fracture tip provide lower stresses.
Modelling of this nature, where natural fractured rocks are submitted into hydraulic fracturing and studied in the particle-scale, are in an early stage and therefore this study attempts to provide further insights.
The complete code that describes one of the simulations of the horizontal fluid injection to the pre-cracked limestone sample (LIM1_15o) has been included
in Appendix III. Furthermore, the codes that describe the rest of the simulations of Chapter 5 are similar with the aforementioned code and include fractures and measurement spheres at different locations.