Fracture containment in soft sands by permeability or strength contrasts

Abstract

Abstract We have conducted laboratory tests to investigate the containment of hydraulic fractures in unconsolidated, layered formations. We varied both permeability and strength of the layers in the samples. The fracture evolution was observed with X-Ray CT scanning and by excavating the sample after the test. We found that permeability is the dominant factor in containment, if the stress contrast is small. With a factor of three to four between the permeability of the two layers there is a strong tendency for the fracture to propagate in the low permeability layer. Difference in strength is much less important for containment. A numerical fracture propagation model described the sand as an elasto-plastic material with parameters obtained from triaxial tests and used a simplified leak-off description of the cross-linked gel. The results agree qualitatively with the tests. Introduction Successful frac&pack treatments require covering the entire interval. This is obviously better for stimulation and it is also important for ensuring effective sand control. Absence of fracture extension can result in the formation of a hole in the gravel pack which causes failure of the screen where it is unprotected. Another method of sand control could be to initiate the fracture in a competent layer and then propagate the fracture into an unconsolidated layer. In that way the proppant pack would stop sand flow and provide sand control, without application of expensive screens. For these applications, it is necessary to model the fracture propagation into unconsolidated layers. Stress is known to have a strong effect on fracture arrest, but the influence of material properties is less well understood, especially in soft rock. Therefore, we performed tests on layered samples with a different strength of the layers. Since leak-off has a strong influence on fracture propagation in highly permeable formations, we tested also samples with different permeability of the layers. Unconsolidated sand is chosen as an extreme case of soft rocks, and the technique of X-ray CT scanning aids in obtaining real-time mapping of the fracture geometry during the experiments. The test results can serve as benchmarks for modeling the containment effect due to permeability differences. The model could then be used to estimate the effect of the fracture containment. Perhaps, it may be possible to design against unwanted containment. It would also be possible in some cases to enhance full coverage by using sand control screens that allow pumping multiple stages.