Ostensen, R.W.; SPE; Sandia Natl. Laboratories Abstract A model of stress-dependent permeability was developed on the basis of flow through cracks. The compliance of the cracks is controlled by elastic deformation of a Gaussian distribution of surface asperities, indented into the opposing crack face. The theory was applied to data from tight gas sand cores, with the following results. 1. The theory explains the Jones and Owens permeability correlation and predicts a modification of it that yields a better fit. 2. The slope of the permeability/stress curve on a log-log plot is predicted to be about 1.0, which agrees with the data. 3. The Jones and Owens Klinkenberg correlation supports a slit-like flow model and indicates that flow occurs along almost every grain boundary. 4. The size of the surface roughness was computed from the Klinkenberg factor and the slope of the permeability/ stress curve. It is independent of stress, as predicted by the theory. 5. The porosity of tight sand cores decreases linearly with the log of net confining stress, in accordance with the Walsh-Grosenbaugh model of crack compressibility. This detailed agreement supports the theory and indicates that the permeability of tight gas sand cores is due to microcracks. permeability of tight gas sand cores is due to microcracks. Introduction The permeability of gas- and oil-bearing sandstone depends on the confining stress caused by the weight of the overburden. The effect, for most consolidated sandstones, is not large, being on the order of 10 to 50%. There are cases, however, where the permeability loss is much larger: Ferrell et al. for example, have reported a factor of 7 loss of permeability, caused by confining stress in an oil-bearing formation with permeability, caused by confining stress in an oil-bearing formation with average core permeability to air of 36 md. The fractional loss of permeabilities generally greater in lower-permeability sandstones in very permeabilities generally greater in lower-permeability sandstones in very tight formations, the loss of permeability is frequently as much as a factor of 10. This corresponds with the observation that flow rates in tight gas reservoirs are frequently much lower than predictions based on routine core-analysis data. Very large changes of permeability with confining stress are not explainable by a simple capillary tube model of permeability. A factor of 10 change of permeability would correspond to a 45% reduction in average capillary diameter. This is far beyond the elastic limit of the rock and is thus inconsistent with the repeatable nature of the permeability/stress curve. This objection is valid for other theories based on pore throats of approximately circular cross section. For example, if pore throats are defined by the minimum area between three adjacent grains in contact with each other, reduction of the effective hydraulic radius of that cross section by 45% would appear to require crushing of the grains or the cementation between them. A simple alternative model is based on flow through microcracks. The idea is an old one and was used successfully by Jones and Owens to correlate the permeability/stress curve in tight gas sandstone. They showed that the cube root of the permeability varies linearly with the log of confining stress. This is consistent with the idea that one should correlate confining stress to crack width, and that the permeability of the rock varies as the cube root of the mean crack width. SPEJ p. 919

The effect of stress-dependent permeability on gas production and well testing in tight gas sands was studied by use of a modified pseudopressure that included stress dependence. This work assumes that information on cores is applicable to reservoir behavior. Because microcracks are the primary path for flow through tight-gas-sand cores, average stress-dependent permeabilities were found as an average over all crack orientations in a nonuniform stress field, with a horizontal stress only 0.6 times the vertical one. Initial in-situ permeabilities are found to be two to five times the values obtained from uniformly restressed cores. Stress de-. pendence decreases initial gas production by at most 30 %, with the loss increasing with reservoir depletion.Buildup tests yield reliable results for the properties at average reservoir pressure. Drawdown tests are more sensitive to stress effects and are much less reliable.

No abstract

No abstract

No abstract