Alterations of Breakdown Pressures in Rocks Exhibiting Stress- dependent Mechanical Properties

Breakdown pressures obtained from the classic, linear elastic breakdown model are compared with the corresponding pressures obtained using a nonlinear material model. Compression test results obtained on sandstone and siltstone are used for that purpose together with previously formulated nonlinear model which introduces elasticity functions to address nonlinear stress-strain behaviour of rocks exhibiting stress-dependent mechanical properties. Linear and nonlinear collapse pressures are also compared and it is shown that material nonlinearities have significant effect on both breakdown and collapse pressures and on tangential stresses which control breakdown pressure around a borehole. This means that the estimates of σH made using linear models give stress values which are different than the real values in the earth. Thus the importance of a more accurate analysis, such as provided by the nonlinear models, is emphasised. It is shown, however, that the linear elastic model does not necessarily over-predict borehole stresses and the opposite case can be true, depending on rock type and test interpretation. 1 Hydraulic fracturing and breakdown pressure Hydraulic fracturing is routinely performed in petroleum engineering to enhance the production of oil and gas from underground reservoirs. Hydrofracturing consists in initiating, then propagating, a fracture from a well using the pressure of a fluid as source of energy. Use of this technique in the petroleum industry began more then fifty years ago, Clark (1949). The hydraulic fracturing test procedure is described by Kim and Franklin (1987) and by Haimson (1978). Despite of well stimulations hydrofracturing has been also used for in-situ stress determination (Haimson and Fairhurst, 1969). It has been successfully applied for measuring stresses at great depths in deep and very deep boreholes, where the drillhole does not have to be assumed vertical and oriented perpendicular to principal in-situ stress components. Other methods such as wellbore breakouts or earthquake focal mechanisms principally indicate stress directions. Breakdown pressure Pb is an important parameter obtained during hydraulic fracturing. When a vertical fracture is induced, the maximum in-situ horizontal stress σH can be determined from the breakdown pressure if the minimum horizontal in-situ stress σh (obtained as the closure pressure after fracture extension and shut-in) and the properties of rocks, such as the tensile strength To or the fracture toughness, are known. Fracture is initiated for a pressure Pb then propagated at a lower pressure value Pp called the propagation pressure. Once a certain volume of fluid has been injected, pumping is stopped and the fracture, which is no longer supplied, begins to close. Breakdown is a complex process affected by many parameters such as the injection rate, the fracture fluid, the wellbore size, the state of stress, and the properties of rocks. As a result, many models and fracture simulators have been put forward to analyze breakdown pressures. Models include the classic linear elastic model by Hubbert and Willis (1957), Haimson's poroelastic model (1968), the model eliminating the tensile strength by Bredehoeft et al. (1976), Schmitt and Zoback's model (1989), models based on fracture mechanics (Abou-Sayed et al., 1978; Rummel, 1987; Detournay and Carbonel, 1994), and many others. The most popular one, which is the classic model, is summarized further in this paper. Hydraulic fracturing methods have not yet reached maturity and there is a far from universal consensus about which approaches, analyses and interpretations work best. Similarly, none of the existing breakdown models are generally accepted because they cannot explain all observed breakdown phenomena. Therefore, the estimation of σH is accorded a low level of confidence and the prediction or analysis of breakdown pressure is still an open question. At the same time, the characteristics and geometry of a hydraulic fracture at great depth are verifiable only at great expense. Due to limitations in test facilities and lack of the scale law, it is difficult to simulate the propagation of hydraulic fractures in a laboratory specimen. The reliability of a fracture model is therefore dependent on the soundness of its underlying mechanics. If the underlying mechanics in the simulator are correct, the prediction should not be far from reality. In this paper it will be shown that nonlinear stress-strain rock properties can lead to substantial uncertainties in