Abstract.Experimental work on shear localization in porous sandstone led to the observation of nonuniform compaction. By analogy with shear localization, the process is referred to as compaction localization. To gain insight into the process of compaction localization, acoustic emission locations were used to define and track the thicknesses of localized zones of compaction during axisymmetric compression experiments. Zones of acoustic emission, demarcating the boundaries between the uncompacted and compacted regions, developed and moved parallel to the sample axis at velocities an order of magnitude higher than the imposed specimen shortening rate. Thus tabular zones of compaction were found to grow (thicken) in the direction of maximum compressive stress. These structures may form due to tectonic stresses or as a result of local stresses induced during production of fluids from wells, resulting in barriers to fluid (oil, gas, water) movement in sandstone reservoirs.

When Palisades diabase and Westerly granite are made dilatant by cyclic differential stress not exceeding ≅ 85% of fracture strength, they behave as nonelastic, energy‐absorbing media. However, in some ways, their properties show reversibility more typical of elastic media. The quasi‐elastic behavior is characterized by the property ‘memory’. Memory in dilatant rock is manifested by the closure of material property hysteresis loops when the sample is returned to certain previous differential stress states. Simple examples of memory have been observed before in Westerly granite. The present work confirms the existence of memory for strain and compressional wave velocity in Westerly granite and Palisades diabase at confining pressures between 250 and 1500 bars. Memory of maxima or minima of the stress difference can be developed, with multiple points remembered simultaneously. A theory of memory based on reversible tensile Griffith cracks was developed which predicts a variety of phenomena, many of which have been observed. A property that is not predicted is the healing which occurs when unloading is stopped and the load is held constant. A time‐dependent increase in velocity and decrease in sample volume is observed, consistent with the healing of cracks. Relaxation ceases after approximately 2 min. After relaxation, a further decrease in load produces an initial elastic response followed by a return to the previous nonelastic behavior. Perfect memory is lost when microfracturing occurs as a result of either exceeding the previous absolute maximum stress or extensive cycling. Acoustic emissions are observed for conditions where memory is not perfect and are not observed when memory is perfect. Using travel time data, a new, low value was found for the dependence of dilatancy onset on pressure.

Localized compaction in porous rocks is a recently recognized phenomenon that has been shown to reduce permeability dramatically. Consequently, the phenomenon is relevant to a variety of technologies involving fluid injection or withdrawal. This article summarizes current understanding of localized compaction and impediments to further progress. The article is based on discussions at a small workshop on localized compaction sponsored by the Office of Science, U. S. Department of Energy.

[1] When subjected to nonhydrostatic, compressive stresses, some porous sandstones exhibit nonuniform compaction. The compaction occurs as a localization process, analogous to shear localization, but results in a thickening, tabular zone of compaction as opposed to culminating in a shear fracture. We report the results of several triaxial compression experiments done at a confining pressure of 45 MPa on Castlegate sandstone, measuring simultaneously, stress, strain, acoustic emission locations, and permeability. A major result is that compaction localization produces up to a 2 order-of-magnitude decrease in permeability. Correlation of local strain measurements and acoustic emission locations made on the same specimen show that the compaction process proceeds as a propagating front approximately 20 mm thick. A model of the compaction process was developed that incorporates the moving boundary between compacted, low-permeability regions and uncompacted, higher-permeability regions, and compaction-induced fluid injection at the boundaries. Because of the inhomogeneous nature of compaction produced by compaction localization, and its temporal evolution, a number of phenomena related to fluid flow are predicted by the model: locally increased pore pressures and spatial changes in the effective permeability. Experimental results are reported that show the evolution of effective permeability to be linear with respect to the distance the compaction fronts propagated as predicted by the model. Implications of the results for future experimentation and for reservoirs are briefly discussed; in particular, the interaction between compaction-induced fluid pressure and compaction localization should lead to a phenomenon analogous to dilatancy hardening, impeding the propagation of compaction bands.

No abstract