Predicting Fracture Attributes in the Travis Peak Formation using
Quantitative Mechanical Modeling and Structural Diagenesis
Margaretha (Peggy) Rijken
Department of Petroleum & Geosystems Engineering The University of Texas at Austin

Fluid flow prediction in naturally fractured reservoirs is challenging, since fracture pattern attributes such as aperture, filling, spacing, length and connectivity are difficult to obtain from core samples and logs. We combine two methods that circumvent this sampling problem to predict fracture characteristics within the Travis Peak Formation. Firstly, a geomechanical approach is employed to estimate aperture, spacing, length and connectivity of the fracture pattern. This model uses subcritical fracture index, a rock property, and geological boundary conditions as input. Secondly, micro-fractures are used to estimate conductivity and orientation of macro-fractures.

The relationship between micro-fractures and macro-fractures has been established in previous studies. Likewise, subcritical fracture index has been shown to influence natural fracture pattern characteristics such as length, spacing and connectivity. However, the dependence of subcritical index on petrologic features is still largely unknown. In this study, suites of subcritical fracture growth experiments were performed on the Travis Peak Formation to investigate systematic variations in subcritical fracture index. Observations show that subcritical fracture index increases with decreasing grain size. Preliminary testing also suggests that subcritical index varies with carbonate cement. Additional tests were carried out in which artificial cement was added to samples. The added cement resulted in increased strength, but decreased subcritical index. We postulate that this kind of artificial cementing of the sample can be equated to secondary carbonate cementation. Using these techniques and understanding the depositional and diagenetic history of a reservoir provides for natural fracture pattern prediction without extensive sampling.