Summary. Geologic discontinuities, such as joints, faults, and beddingplanes, can significantly affect the overall geometry of hydraulic planes, can significantly affect the overall geometry of hydraulic fractures. This can occur by arresting the growth of the fracture, increasing fluid leakoff, hindering proppant transport, and enhancingthe creation of multiple fractures. Results from mineback experimentsand laboratory tests and analyses of these data are integrated todescribe this complex fracture behavior. Introduction Hydraulic fracturing has become a valuable technique forthe stimulation of oil, gas, and geothermal reservoirs ina variety of reservoir rocks. In many applications, onlyshort fractures are required for economic production, andfor these, the constant-height, ideal-fracture models ofPerkins and Kern, Geertsma and de Klerk, and Nordgren Perkins and Kern, Geertsma and de Klerk, and Nordgren may quite adequately represent the fracturing process. In low-permeability gas reservoirs, however, long, penetrating fractures are generally needed; in penetrating fractures are generally needed; in this case, many assumptions about the fracturing processneed to be re-examined. In particular, the widely heldassumption that the hydraulic fracture is an ideal, planarfeature (usually of constant height) is probably untenablein many reservoirs because of geologic discontinuities. Geologic discontinuities such as joints, faults, beddingplanes, and stress contrasts are ubiquitous features whose planes, and stress contrasts are ubiquitous features whose effect on the hydraulic fracture depends on ancillarytreatment and such reservoir parameters as, the treatingpressure, in-situ stresses, orientations of the discontinuities, pressure, in-situ stresses, orientations of the discontinuities, and permeability. Previous analyses and laboratory andfield data have shown the effects of some of thesefeatures but only hinted at others. The effects of stresses, material properties, andunbonded bedding planes on fracture height are welldocumented and will not be discussed in detail in thispaper. Clearly, the in-situ stress distribution is the paper. Clearly, the in-situ stress distribution is the primary factor controlling containment, but when the stress primary factor controlling containment, but when the stress contrasts are small, material property variations may havesome effect. In addition, plasticity of the shale layersmay restrict fracture height. Cohesionless interfaces canprovide an excellent containment feature, but this is provide an excellent containment feature, but this is probably applicable mostly at shallow depths where the probably applicable mostly at shallow depths where the normal stress (in this case, the overburden) acting on theplane is small. plane is small. In a more general context, geologic discontinuities willinfluence the overall geometry and effectiveness of thehydraulic fracture byarresting vertical propagationas described above;arresting lateral propagation asat a fault or sand lens boundary where stresses mayincrease;reducing total length by fluid leakoff;reducing total length by facilitating the formation ofmultiple parallel-fracture systems;hindering proppanttransport and placement because of the nonplanarity ofthe fracture or fracture system; andinducing additionalfracture height growth from higher treating pressuresbecause of many of the above features (e.g., Items 2, 4, and 5). The result may range from negligible tocatastrophic, depending on the values of the ancillaryparameters. parameters. We show examples of several of these features that wereobserved in mineback experiments at the U.S. DOE NevadaTest Site. In addition, we present laboratory data andanalyses that give some guidelines as to when thesefeatures become important. Examples From Mineback Experiments The effects of many geologic discontinuities have beenobserved in mineback experiments conducted at DOE's Nevada Test Site. These facilities are ideal forhydraulic fracturing experiments because they provide anin-situ medium with the appropriate boundary conditions(in-situ stresses and no free surfaces) yet still allowfor detailed examination of the created fractures andgeological features through mineback (physical excavationof the rock to observe the fracture directly). A detailedphysical description can be obtained through photography physical description can be obtained through photography and mapping, and this can be correlated with measuredmaterial properties, in-situ stress distributions, geologicdiscontinuities, fluid behavior, and the operationalparameters of the test. parameters of the test. At tunnel level, there is approximately 1,400 ft [425 m]of overburden that provides a realistic in-situ stressdistribution. The experiments were conducted mainly inash-fall tuffs, which are soft, low-modulus, high-porosity, low-permeability tuffs that allow for easy excavation witha continuous-mining machine. Overlying the ash-fall tuffis an ash-flow tuff, which is much denser and has a highermodulus and lower porosity than the ash-fall tuff. Theash-flow tuff grades upward from an unwelded basal ash-flowtuff into a densely welded ash-flow tuff. Although thevarious volcanic tuffs in which these fractures are propagatedare not the sandstones and shales usually encountered ingas reservoirs, proper application of rock mechanicsprinciples allows the extrapolation of these results to principles allows the extrapolation of these results to gas well conditions. JPT P. 209