Laborversuche zur hydraulischen Risserzeugung in dreiaxial belasteten Granitquadern - Grundlagen, Versuchsentwicklung, -durchführung und Analyse

In hydraulic fracturing, cracks are generated, opened and propagated by high fluid pressure. Hydraulic fracturing as a technical process is used, among other applications, to install artificial geothermal systems - so-called engineered geothermal systems (EGS). In EGS, a fluid is pumped through the hydraulically generated cracks in the deep subsoil in order to promote geothermal energy. Since EGS are basically independent of water-bearing layers and temperature anomalies, a significant contribution of geothermal energy to Germanys energy supply is expected. Various international pilot projects have confirmed the feasibility of engineered geothermal systems. However, these projects were still inefficient in terms of profitability. For a successful realization of an EGS, it is necessary to improve, inter alia, the understanding and the numerical simulation of hydraulic fracturing in plutonic rock to make it more predictable. Against this background, the process of hydraulic fracturing in crystalline bed-rocks is investigated experimentally and numerically in a new research group at RWTH Aachen University under the leadership of the Institute of Applied Geophysics and Geothermal Energy. The projects approach is to use experimental results to examine and improve various numerical methods in order to develop a numerical tool for planning of hydraulic fracturing for future geothermal applications. Since field-scale experiments are very costly due to the need of deep drilling, a variation of some boundary condition is only possible by changing the location of the experiment and because of the poor reproducibility due to the natural heterogeneity of the subsoil, laboratory-scale experiments were carried out in this project. For this purpose a new experimental apparatus has been developed. In the presented test series, cuboidal rock samples (300 x 300 x 450 mm3) made from Tittlinger Feinkorn granite are loaded triaxially with flat-jacks to simulate the influence of the initial stress state on the fracturing process. Then, dyed glycerin is pressed into a delimited borehole section of the centrically drilled samples with a high-pressure precision-pump. When the fluid pressure reaches a critical value, a crack initiates at the loaded borehole section, and is propagated by the injection of further fluid. In order to decrease the speed of fracture growth and to keep the cracks within the specimens, a special injection method is used: in a so-called "pre-fracturing cycle", fluid is injected until the peak pressure is reached. Then the pressure is discharged abruptly. In the second injection cycle, the previously produced "flaw" is opened and propagated with a very low, constant injection rate of Q = 0.05 cm3 / min. Acoustic emissions of the fracturing process are recorded and subsequently localized to monitor the fracture propagation. In addition, the pressure in the injection string and the control volumes of the control-device connected to the flat-jacks are recorded. After the test, the sample is split in the crack plane and the "colored fracture surface" is scanned with a 3D scanner. The present work does not include a description of the numerous preliminary experiments which were necessary to develop the final experimental procedure. Instead, the six last test series are presented, each consisting of three individual tests with the same settings. In five series, cracks were generated from a circumferential notch on the wall of the borehole and propagated transvers to the borehole axis. The duration of injection, the injection process and the normal stress on the crack plane (z) were varied. In the sixth series, fractures parallel to the borehole-axis were produced. The tests show that the hydraulic fracturing in the developed test stand is reproducible with regard to the injection pressure level as well as the size and shape of the created fractures. In addition, the observations show that the hydraulic fracture propagation is dominated by different influences, varying with time: Directly after reaching the peak pressure, the volume flow, which is increased by the decompression of the fluid, causes the crack to propagate very rapid initially. With the reduction of the excess in elastic energy, the influence of fluid losses increases and the fracture propagation slows down significantly. In some experiments, even a crack stop is observed since the fluid losses exceed the injection rate in the meantime. The numerical difference between the control volume change of the loading apparatus (Vz), which correlates with the elongation of the sample, and the injected volume (Vp), shows that a high proportion of the injected glycerin migrates into the partially saturated rock without creation of new crack volume. By comparing the test results with a simple analytical model this conclusion could be confirmed. The preliminary assumption that the experimental rock is to be regarded as impermeable does not appear to be justified for the simulation of the presented experiments with its very low injection rate as well as the injection fluid and the type of rock used. In planning of the presented experiments, the question of scalability was deliberately ignored. For subsequent investigations it is recommended to derive the experimental setting from the field scale and to scale it into the laboratory scale. Corresponding scaling approaches were developed by "hydraulic fracturing research". Their transferability to the geothermal background has to be checked. Thus, in the future it could be excluded that in the future purely laboratory-specific phenomena will determine the experimental and numerical developments for hydraulic fracturing. To conclude, it can be stated that the newly developed experiment is a strong foundation for future investigations. Improvements in the described strain measurements by analyzing the volume changes in the flat-jacks and optimizations on the technique for the acoustic emission analysis promise additional insights in the interpretation of the experiments