Small-scale laboratory experiments were performed to study the growth of hydraulically driven fractures in the vicinity of an unbonded interface in rocks. The purpose was to evaluate under which conditions the hydraulic fractures would cross the interface. The materials used in these studies were Nugget sandstone from Utah (3 to 6% porosity) and Indiana limestone (12 to 15% porosity). The fracturing fluid was oil (viscosity appx. 300 cp) injected into the rock through high-pressure steel tubing. Prismatic blocks of the rock materials to be studied were held adjacent to one another in a hydraulic press so that a normal stress was set up across their mutual interface. Lubricants and surface roughening were used to vary the frictional properties of the interfaces. It was found that as the interface surface friction coefficient was decreased, the normal stress had to be increased for a hydraulic fracture to cross the interface. The frictional shear stress that the interface can support without slippage appears to be critical in determining fracture growth across the interface. Additional experiments were performed to evaluate the coefficient of friction for the different interface surface preparations used. These experiments demonstrated that a variation in the frictional properties along an interfacial surface in the vicinity of hydraulic fracture growth can alter the path of the fracture. The experiments also demonstrated that cracks, which intersect the interface from the side opposite the approaching hydraulic fracture, can impede fracture growth across the interface. Introduction Hydraulic fracturing and a variant - massive hydraulic fracturing (MHF) - are primary candidates for stimulating production from the tight-gas reservoirs in the U.S. Hydraulic fracturing has been used widely as a well completion technique for about 30 years. MHF is a more recent application that differs from hydraulic fracturing in that larger quantities of fluid and proppant are pumped to create more extensive fractures in the reservoirs. Application of MHF to increase production from the tight reservoirs has provided mixed and, in many cases, disappointing results, especially in lenticular reservoirs. For MHF to be successful in enhancing the production of gas from tight reservoirs, it is important that the fractures be emplaced in productive reservoir rock providing large drainage surfaces in the low-permeability material and conductive channels back to the wellbore. We then are faced with the problem of containing fractures in a given formation.Under the U.S. DOE'S Unconventional Gas Recovery program, Lawrence Livermore Natl. Laboratory is conducting a research program to study the hydraulic fracture process. The general goal of this research is to determine if and to what extent the reservoir parameters control the geometry of the created fractures. These reservoir parameters include (1) the mechanical properties of the rock (i.e., elastic moduli, mechanical strength, etc.), (2) the physical state of the rock (i.e., presence of pre-existing cracks or faults, porosity, pore fluid, etc.), (3) presence of layering or interfaces between different rock strata, and (4) stress field on the rock. In addition to reservoir parameters, the growth of a hydraulically driven crack will be influenced by (1) the manner in which the driving fluid is injected into the rock, (2) the characteristics of the fracturing fluid (i.e., viscosity, presence of proppant, etc.), and (3) any chemical reaction between the fluid and rock. Previous work has shown that crack orientation is controlled primarily by the in-situ or applied stress field, with crack growth oriented perpendicular to the least principal stress. SPEJ P. 21^

Plane shock wave experiments have been conducted on two highly porous rocks, Mount Helen tuff and Indiana limestone, in both dry and water-saturated states up to stress levels of about 4 GPa. A light-gas gun was used to load the sample in uniaxial strain, and the subsequent wave motion was monitored with particle-velocity gages. All four materials studied show evidence of time-dependent behavior. The timedependent behavior in the highly porous dry rocks is associated with the closing of pores. The strong time dependence observed in these materials would seem to preclude the use of quasi-static data in constitutive models that are used to describe dynamic processes. In the water-saturated rocks the time dependence is associated with the water, which shows no indication of transformation to the high-pressure ice phases in the time frame of these shock wave experiments. This suggests the possibility of a metastable form of water existing under dynamic conditions.

We are conducting a U.S. DOE-funded research program aimed at understanding the hydraulic fracturing process, especially those phenomena and parameters that strongly affect or control fracture geometry. Our theoretical and experimental studies consistently confirm the well-known fact that in-situ stress has a primary effect on fracture geometry, and that fractures propagate perpendicular to the least principal stress. In addition, we find that frictional interfaces in reservoirs can affect fracturing. We also have quantified some effects on fracture geometry caused by frictional slippage along interfaces. We found that variation of friction along an interface can result in abrupt steps in the fracture path. These effects have been seen in the mineback of emplaced fractures and are demonstrated both theoretically and in the laboratory. Further experiments and calculations indicate possible control of fracture height by vertical change in horizontal stresses. Preliminary results from an analysis of fluid flow in small apertures are discussed also. Introduction Hydraulic fracturing and massive hydraulic fracturing (MHF) are the primary candidates for stimulating production from tight gas reservoirs. MHF can provide large drainage surfaces to produce gas from the low- permeability formation if the fracture surfaces remain in the productive parts of the reservoir. To determine whether it is possibleto contain these fractures in the productive formations andto design the treatment to accomplish this requires a much broader knowledge of the hydraulic fracturing process. Identification of the parameters controlling fracture geometry and the application of this information in designing and performing the hydraulic stimulation treatment is a principal technical problem. Additionally, current measurement technology may not be adequate to provide the required data. and new techniques may have to be devised. Lawrence Livermore Natl. Laboratory has been conducting a DOE-funded research program whose ultimate goal is to develop models that predict created hydraulic fracture geometry within the reservoir. Our approach has been to analyze the phenomenology of the fracturing process to son out and identify those parameters influencing hydraulic fracture geometry. Subsequent model development will incorporate this information. Current theoretical and stimulation design models are based primarily on conservation of mass and provide little insight into the fracturing process. Fracture geometry is implied in the application of these models. Additionally, pressure and flow initiation in the fractures and their interjection with the fracturing process is not predicted adequately with these models. We have reported previously on some rock-mechanics aspects of the fracturing process. For example, we have studied, theoretically and experimentally, pressurized fracture propagation in the neighborhood of material interfaces. Results of interface studies showed that natural fractures in the interfacial region negate any barrier effect when the fracture is propagating from a lower modulus material toward a higher modulus material. On the other hand, some fracture containment could occur when the fracture is propagating from a higher modulus into a lower modulus material. Effect of moduli changes on the in-situ stress field have to be taken into consideration to evaluate fracture containment by material interfaces. Some preliminary analyses have been performed to evaluate how stress changes when material properties change, but we have not evaluated this problem fully. SPEJ P. 321^

Single-crystal specimens (7-and 12-mm thick) of sodium chloride were impacted with flat-nosed, gasdriven projectiles, and the Hugoniot elastic limit (HEL) was determined by reducing quartz gauge measurements. The HEL for the [100J, [110J, and [111J crystal directions was 0.26, 0.77, and 7.4 kbar, respectively. Stress-time profiles for specimens shocked in the [100J and [l11J direction show evidence of stress relaxation behind the elastic precursor. This phenomenon is more pronounced in 12-mm-thick specimens. The ratio of the resolved shear stress on the active slip systems for uniaxial strain (shock loading) conditions to that for uniaxial stress (static loading) indicates a strain-rate effect. This ratio increases from 3.1 for loading in the [100J direction to 8 for the [110J direction and to 21 for the [111J direction. The anisotropy of the HEL with crystal direction is related to the resolved shear stress on the primary and secondary slip systems in single-crystal sodium chloride. The large HEL for shock loading in the [l11J direction is a consequence of the resolved shear stress on the primary slip systems being zero. Thus, for deformation by slip to occur, a secondary slip system (or systems) must be activated which will require a higher resolved shear stress. The experimental data for single crystals of copper and beryllium can also be explained in terms of the resolved shear stress.