Pre-existing Joints Research Articles

Deep Fracturing of the Hard Rock Surrounding a Large Underground Cavern Subjected to High Geostress: In Situ Observation and Mechanism Analysis

Rocks that are far removed from caverns or tunnels peripheries and subjected to high geostress may undergo ‘deep fracturing’. Deep fracturing of hard rock can cause serious hazards that cause delays and increase the cost of construction of underground caverns with high sidewalls and large spans (especially when subjected to high geostress). To extensively investigate the mechanism responsible for deep fracturing, and the relationship between fracturing and the excavation & support of caverns, this paper presents a basic procedure for making in situ observations on the deep fracturing process in hard rock. The basic procedure involves predicting the stress concentration zones in the surrounding rocks of caverns induced by excavation using geomechanical techniques. Boreholes are then drilled through these stress concentration zones from pre-existing tunnels (such as auxiliary galleries) toward the caverns before its excavation. Continuous observations of the fracturing of the surrounding rocks are performed during excavation using a borehole camera in the boreholes in order to analyze the evolution of the fracturing process. The deep fracturing observed in a large underground cavern (high sidewalls and large span) in southwest China excavated in basalt under high geostress is also discussed. By continuously observing the hard rock surrounding the arch on the upstream side of the cavern during the excavation of the first three layers, it was observed that the fracturing developed into the surrounding rocks with downward excavation of the cavern. Fracturing was found at distances up to 8–9 m from the cavern periphery during the excavation of Layer III. Also, the cracks propagated along pre-existing joints or at the interfaces between quartz porphyry and the rock matrix. The relationship between deep fracturing of the surrounding rocks and the advance of the cavern working faces was analyzed during excavation of Layer Ib. The results indicate that the extent of the stress relief zone is about 7 m if footage of 3 m is adopted for the rate of advance of the cavern faces. An analysis of the effects of the initial geostress and evolving stress concentration on deep fracturing was also made. It could be concluded that the deep fracturing of the rocks in the upstream side of the cavern is caused by the combined effect of the high initial geostress, the transfer of the stress concentration zone toward the deep surrounding rocks, and the occurrence of discontinuities.

The Enigma of Large-Scale Permeability of Gas Shale: Pre-Existing or Frac-Induced?

The existing commercial programs for simulation of hydraulic fracturing (aka fracking, or frac) of gas (or oil) shale predict parallel vertical cracks to spread in vertical parallel planes, with no lateral branching. These cracks emanate from the perforation clusters on the horizontal wellbore casing, typically spaced 10 m apart or more. For such a large spacing, the rate of gas production observed at the wellhead can be explained only upon making the hypothesis that the large-scale (or regional) permeability of shale is (even at 3 km depth) about 10,000 times higher than the gas permeability of shale measured in the lab on drilled (nondried) shale cores under confining pressures corresponding to shale at the depth of about 3 km. This hypothesis has recently been rendered doubtful by a new three-phase medium theory that takes into account the body forces due to pressure gradients of pore water diffusing into the pores. This theory predicts the fracking to produce a dense system of branched vertical hydraulic cracks with the spacing of about 0.1 m. This value matches the crack spacing deduced from the gas production rate at wellhead based on the actual lab-measured permeability. It is calculated that, to boost the permeability 10,000 times, the width of the pre-existing open (unfilled) natural cracks or joints (whose ages are distributed from one to several hundred million years) would have to be about 2.8 μm (not counting possible calcite deposits in the cracks). But this width is improbably high because, over the geologic time span, the shale must exhibit significant primary and secondary creep or flow. It is shown that the creep must close all the cracks tightly (except for residual openings of the order of 10 nm) even if the cracks are propped open by surface asperities. The inevitability of secondary creep (or steady-state flow) is explained theoretically by activation of new creep sites at stress concentrations caused by prior creep deformation. The time of transition from primary to secondary creep is taken equal to the Maxwell time estimate from geology. The overall conclusion is that the 10,000-fold increase of large-scale permeability is most likely not pre-existing but frac-induced. Although this conclusion will make little difference for long-term forecasts, it would make a major difference for the understanding and control of the frac process.

Joint-bounded crescentic scars formed by subglacial clast-bed contact forces: Implications for bedrock failure beneath glaciers

Glaciers and ice sheets are important agents of bedrock erosion, yet the precise processes of bedrock failure beneath glacier ice are incompletely known. Subglacially formed erosional crescentic markings (crescentic gouges, lunate fractures) on bedrock surfaces occur locally in glaciated areas and comprise a conchoidal fracture dipping down-ice and a steep fracture that faces up-ice. Here we report morphologically distinct crescentic scars that are closely associated with preexisting joints, termed here joint-bounded crescentic scars. These hitherto unreported features are ca. 50–200mm deep and involve considerably more rock removal than previously described crescentic markings. The joint-bounded crescentic scars were found on abraded rhyolite surfaces recently exposed (<20years) beneath a retreating glacier in Iceland, as well as on glacially sculpted Precambrian gneisses in NW Scotland and various Precambrian rocks in Ontario, glaciated during the Late Pleistocene. We suggest a common formation mechanism for these contemporary and relict features, whereby a boulder embedded in basal ice produces a continuously migrating clast-bed contact force as it is dragged over the hard (bedrock) bed. As the ice-embedded boulder approaches a preexisting joint in the bedrock, stress concentrations build up in the bed that exceed the intact rock strength, resulting in conchoidal fracturing and detachment of a crescentic wedge-shaped rock fragment. Subsequent removal of the rock fragment probably involves further fracturing or crushing (comminution) under high contact forces. Formation of joint-bounded crescentic scars is favoured by large boulders at the base of the ice, high basal melting rates, and the presence of preexisting subvertical joints in the bedrock bed. We infer that the relative scarcity of crescentic markings in general on deglaciated surfaces shows that fracturing of intact bedrock below ice is difficult, but that preexisting weaknesses such as joints greatly facilitate rock failure. This implies that models of glacial erosion need to take fracture patterns of bedrock into account.

Effect of normal load on the crack propagation from pre-existing joints using Particle Flow Code (PFC)

In this paper, the effect of normal load on the failure mechanism of echelon joint has been studied using PFC2D. In the first step, calibration of PFC was undertaken with respect to the data obtained from experimental laboratory tests. Then, six different models consisting various echelon joint were prepared and tested under two low and high normal loads. Furthermore, validation of the simulated models were cross checked with the results of direct shear tests performed on non-persistent jointed physical models. The simulations demonstrated that failure patterns were mostly influenced by normal loading, while the shear strength was linked to failure mechanism. When ligament angle is less than 90o, the stable crack growth length is increased by increasing the normal loading. In this condition, fish eyes failure pattern occur in rock bridge. With higher ligament angles, the rock bridge was broken under high normal loading. Applying higher normal loading increases the number of fracture sets while dilation angle and mean orientations of fracture sets with respect to ligament direction will be decreased.

A symmetric implementation of pressure-based fluid–structure interaction for nonlinear dynamic analysis of arch dams

Fluid–structure interaction formulated in the Lagrangian–Eulerian approach with pressure-based fluid elements, which inherently results in non-symmetric matrices, is commonly treated by either direct un-symmetric solvers or staggered solution methods. Developing an in-house finite element code called SNACS, we implement the pressure-based Eulerian formulation so that a usual symmetric solver can be employed for dam–reservoir interaction analyses. The equilibrium equations of motion for such a coupled system are directly integrated herein by the HHT time marching algorithm in an incremental–iterative solution procedure to tackle a highly nonlinear dynamic analysis of jointed arch dams. The HHT time integration scheme is able to damp out high-frequency noises resulting from sudden changes in the stiffness due to opening/closing of pre-existing joints during an earthquake. First, a Pseudo-Symmetric technique is proposed to store the total matrices and solve the coupled non-symmetric equations in a symmetric manner. Then, the interface element formulation and stress update procedures of two discrete crack constitutive models suitable for contraction and peripheral joints of arch dams are detailed. Afterwards, results of the analysis on a typical arch dam are compared with the case in which the interaction is approximately represented by added masses as a widely-used method in dam engineering practice to illustrate the effects of water compressibility and reservoir bottom absorption. Besides, the straightforward implementation of pressure-based fluid–structure interaction utilizing a symmetric solution strategy is validated herein to be efficient for a 3-D large scale practical usage in other finite element codes even the commercial ones to avoid using un-symmetric solvers. Employing this symmetric implementation for the dam–water interaction drastically reduces the computational effort, particularly when an iterative scheme is required to solve such a highly nonlinear system.

Evaluation of Slope Stability by the In Situ Monitoring Data Combined with the Finite-Discrete Element Method

We examine the stability of an ancient slope located at Jietai Temple in western Beijing of China, by using the remote monitoring system of sliding force combined with the combined finite-discrete element method (FDEM) as well as the field surveys. Potential pre-existing joints in the slope were firstly determined by the in situ study and rock mass mechanical properties were then estimated to be taken as input parameters in the numerical modelling, for understanding the instability mechanism of the slope. Results from this synthetic assessment together with the monitoring data suggest that current protection measures in the slope may not be sufficient so that the local authorities should define hazard zones and work out further development plans for the Jietai temple.

Numerical simulation of hydraulic fracturing in circular holes

For investigating the effect of the pre-existing joints on the initiation pattern of hydraulic fractures, the numerical simulation of circular holes under internal hydraulic pressure with a different pattern of the joint distributions are conducted by using a finite element code, FRANC2D. The pattern of hydraulic fracturing initiation are scrutinized with changing the values of the joint length, joint offset angle. The hydraulic pressures with 70% of the peak value of borehole wall breakout pressure are applied at the similar models. The simulation results suggest that the opening-mode fracture initiated from the joint tip and propagated toward the borehole for critical values of ligament angle and joint offset angle. At these critical values, the crack grow length is influenced by joint ligament length. When the ligament length is less than 3 times the borehole diameter the crack growth length increases monotonically with increasing joint length. The opening-mode fracture disappears at the joint tip as the ligament length increases.

Growth model for large branched three-dimensional hydraulic crack system in gas or oil shale.

Recent analysis of gas outflow histories at wellheads shows that the hydraulic crack spacing must be of the order of 0.1 m (rather than 1 m or 10 m). Consequently, the existing models, limited to one or several cracks, are unrealistic. The reality is 10(5)-10(6) almost vertical hydraulic cracks per fracking stage. Here, we study the growth of two intersecting near-orthogonal systems of parallel hydraulic cracks spaced at 0.1 m, preferably following pre-existing rock joints. One key idea is that, to model lateral cracks branching from a primary crack wall, crack pressurization, by viscous Poiseuille-type flow, of compressible (proppant-laden) frac water must be complemented with the pressurization of a sufficient volume of micropores and microcracks by Darcy-type water diffusion into the shale, to generate tension along existing crack walls, overcoming the strength limit of the cohesive-crack or crack-band model. A second key idea is that enforcing the equilibrium of stresses in cracks, pores and water, with the generation of tension in the solid phase, requires a new three-phase medium concept, which is transitional between Biot's two-phase medium and Terzaghi's effective stress and introduces the loading of the solid by pressure gradients of diffusing pore water. A computer program, combining finite elements for deformation and fracture with volume elements for water flow, is developed to validate the new model.This article is part of the themed issue 'Energy and the subsurface'.

Influence of mechanical stratigraphy on clastic injectite growth at Sheep Mountain anticline, Wyoming: A case study of natural hydraulic fracture containment

Sandstone injectites ranging from 1 km in outcrop length intrude the Cretaceous Mowry Formation in the vicinity of Sheep Mountain anticline (Bighorn Basin, Wyoming, USA). These injectites were sourced from the Peay Sandstone Member of the overlying Cretaceous Frontier Formation and represent a significant possible fluid pathway through impermeable shales. Sand injection occurred along dikes and sills, interacting with bedding discontinuities and preexisting joints in the Mowry Formation during the early folding of Sheep Mountain anticline. We argue that, in contrast to the passive sweeping of sediments into fissures characteristic of Neptunian dike formation, downward intrusion of the Peay sand was forceful and made possible by a highly stratified horizontal stress field resulting from the deposition, burial, and lithification history of the rock units in the area. The internal structure of the injectites is dominated by two sets of mutually offsetting deformation bands. The deformation bands have shear and compaction components, exhibiting significant porosity loss, as well as cataclasis and minor pressure solution. After formation of the deformation bands, subsequent faulting was localized along the margins of deformation bands, evidenced in the field by slickensided surfaces. A detailed kinematic analysis of slickenline lineations yields shortening and extension axes consistent with deformation band formation during early Laramide–oriented shortening, and continuing through the folding of Sheep Mountain anticline. Beyond the formation and deformation of these sandstone injectites, this study highlights the importance of mechanical stratigraphy in the containment of hydraulic fractures.

Structure and tectonic setting of the SE Sardinia mafic dyke swarm. Insights for the stress state during magma emplacement in the upper crust

The SE Sardinia mafic dyke swarm intruded in Permian times into the Late Variscan granitoids of the Sàrrabus massif. In order to characterize the structure of the dyke swarm and to assess the stress state and tectonic regime at the time of dyking, we have performed a multi-methodological study which encompasses paleostress analysis using the orientation distribution of dykes, estimation of crustal extension from dyke thickness/spacing ratios and true dilation direction from the matching geometry of offsets. The mean orientation of the dyke swarm is ∼N150°, with secondary main trends orientated ∼N77° and ∼N10°. The fracture network (mostly joints) exhibits multiple orientations grouped in four or more sets, being the orientation and density patterns in granitoid and dykes significantly different. Some of the pre-existing joints were exploited by the dykes, while others were reactivated as faults during and after dyke intrusion. The paleostress analysis yielded σ1 sub-vertical, indicating that an extensional tectonic regime predominated in the area during the emplacement of the dyke swarm. σ3 was sub-horizontal, sub-parallel to the mean NE–SW dilation direction. The inferred relative stress magnitudes are characterized by a σ1 much larger than the other two principal stresses. In this context, intrusion of the dyke swarm occurred under low magma pressure conditions and preferentially into a pre-existing joint nerwork during a regional extensional event. The SE Sardinian dyke swarm has been compared with other contemporary mafic dykes of the western Mediterranean region, by attempting to restore the effects of the Neogene and eo-Alpine rotational events. The reconstruction gives a complex crustal-scale pattern in terms of dyke orientation, a fact that could be partly attributed to differential rotations associated with localized strike-slip movements that might have operated coevally with regional extension or transtension during Permian times.

Dilatant normal faulting in jointed cohesive rocks: a physical model study

Abstract. Dilatant faults often form in rocks containing pre-existing joints, but the effects of joints on fault segment linkage and fracture connectivity are not well understood. We present an analogue modeling study using cohesive powder with pre-formed joint sets in the upper layer, varying the angle between joints and a rigid basement fault. We analyze interpreted map-view photographs at maximum displacement for damage zone width, number of connected joints, number of secondary fractures, degree of segmentation and area fraction of massively dilatant fractures. Particle imaging velocimetry provides insight into the deformation history of the experiments and illustrates the localization pattern of fault segments. Results show that with increasing angle between joint-set and basement-fault strike the number of secondary fractures and the number of connected joints increase, while the area fraction of massively dilatant fractures shows only a minor increase. Models without pre-existing joints show far lower area fractions of massively dilatant fractures while forming distinctly more secondary fractures.

Blast-induced damage and its impact on structural stability of underground excavations

Analysis of rock wedge stability plays important role in tunneling and hard rock mining. Today there are many procedures for prediction of wedge formation and their stability. These procedures consider preexisting joints in rock mass, and there is no procedure that considers blast-induced cracks and their impact on structural stability of excavations. Using main principles of new rock blasting theory it is possible to assess length and density of radial cracks around blastholes. Herein, one simple excavation situation in hard rock is presented, where only two joint sets were present. This made it impossible for rock wedges to form. After blast-induced crack with limited persistence was added to analysis as result unstable rock wedge was formed. in addition to presented problem it is suggested how mechanical properties of blast-induced cracks could be estimated using Barton-Bandis failure criterion. As final result it was clear that without blast- induced cracks no wedge stability analysis is complete.

Structural evolution of the Mokrsko-West, Mokrsko-East and Čelina gold deposits, Bohemian Massif, Czech Republic: Role of fluid overpressure

The studied Mokrsko-West (90–100t Au), Mokrsko-East (30t Au) and Čelina (11t Au) deposits represent three spatially and genetically interrelated deposits of supposed affiliation to the intrusion-related gold deposit type. The deposits differ in their dominant host rocks, which are represented by ca 354Ma old biotite tonalite (Mokrsko-West) and Neoproterozoic volcanic and volcanosedimentary rocks (Mokrsko-East, Čelina). Another difference lies in the style of veining — densely spaced networks of 0.1–5mm thin veins (Q2) within the tonalite, compared to thick (usually 5–20cm; Q1–2) and widely spaced veins within the Neoproterozoic rocks.Five generations of quartz veins, referred to as Q0 through Q4 were distinguished: Q0 veins are the oldest and ore-barren, Q1 veins mark the onset of the Au-ore formation, Q2 veins its culmination and Q3 veins its fading. Late quartz gangue (Q4) is associated with uneconomic Ag–Pb–Zn vein-type ores hosted by calcite–barite–(quartz) veins.Quartz vein thickness (~0.3 to ~300mm), spacing (~3mm to ~500mm), distribution, and related extensional strain (ca. 3–25%) evolve systematically across the studied ore district, reflecting both the major host rock and other tectonic factors. Detailed study of vein dimension parameters (thickness, length, width, aspect ratios) allowed estimation of the probable depth of the fluid source reservoir (~2km or ~4km) below the present surface. The depth to the fluid source seems to increase through time, being the shallowest for the Q0 veins and the deepest for the Q2 veins. Two independent methods of estimating fluid overpressure are discussed in the paper. Fluid overpressure during vein formation decreases from the Q0 through the Q2 veins, from 10 to 4MPa or from 26 to 10MPa, depending on the assumed tensile strength of the tonalite (5.5 and 15MPa, respectively).The origin of joints and veins is discussed in terms of the stress orientation and crack-seal and crack-jump mechanisms. Field relationships unambiguously indicate that the veins hosted by Neoproterozoic rocks originated by reopening of the pre-existing extension joints (J1) due to fluid overpressure. The origin of the densely-spaced thin veins (Q2) hosted by the tonalite at the Mokrsko-West deposit is, however, less certain. It is probable that the tonalite was already affected by microfracturing analogous to the J1 joints prior to the formation of quartz veins.The formation of the Q1–2 veins at the Mokrsko-East deposit was constrained by the Re–Os dating of molybdenite to 342.9±1.4Ma. The ore-bearing hydrothermal system is thus ca 12Ma younger than the tonalite that hosts the Mokrsko-West deposit. A similar ca 15–2Ma difference between the age of the host-intrusion and the age of the hydrothermal event was encountered in several other gold deposits in the vicinity of the Central Bohemian Plutonic Complex. Two hypotheses to explain this are discussed in the paper.

Rock avalanche scars in the geological record: an example from Little Loch Broom, NW Scotland

An embayment in the southern rock slopes of Little Loch Broom, NW Scotland, was surveyed using a combination of morphological mapping and engineering geology techniques. The evidence points to a large-scale rock slope failure (rock avalanche) mechanism of formation, occurring during glacial downwasting and retreat, perhaps initiated by debuttressing and stress release. Periglacial activity, such as macrogelivation or large-scale frost-wedging, may have exacerbated the internal joint network by expanding the aperture dimensions of pre-existing joints leading to an increase in stresses and a reduction in friction across joint surfaces. The rockwall cavity was completely evacuated by the rock debris, leaving the embayment with a mantle of sediments originating from the glacial diamicton that drapes the plateau above the backwall. The absence of any rock avalanche debris deposit has led to the conclusion that the debris run-out extended onto the surface of the glacier and was supraglacially transported and dispersed, which has implications for glacial dynamics and local glacial history for Little Loch Broom.

The interaction between pre-existing joints and normal faults in analogue models – implications for fault architecture and fracture connectivity

The interaction between pre-existing joints and normal faults in analogue models – implications for fault architecture and fracture connectivity

Failure process and modes of rockfall induced by underground mining: A case study of Kaiyang Phosphorite Mine rockfalls

The analysis of the failure processes and mechanisms of rockfalls that are associated with underground mining activities is presented. The study area is located in Kaiyang Phosphorite Mine in Southwest China, where the geological condition is dominated by anti-dip slopes with layers inclined backward into the natural slope with decreasing strength in the rock mass from the upper (dolomite-rich) to the lower (shale-rich) strata. The analysis is based on field investigation and laboratory experimental study using a gravitational simulation device. The results support the proposed failure mechanism by demonstrating the process of failure from the deformation of the roof and floor in the stopes, to the development of surface cracks near the top of slopes, and eventually to the formation of rockfalls as the surface cracks propagate along pre-existing joints. The rockfalls are classified into one of three failure modes: crack–toppling, crack–sliding, and crack–slumping, in which the failures are governed by the corresponding characteristics of the rock mass structure. The study of the failure process and their spatial and temporal correlation with the underground workings reveals warning signs or indicators of impending slope instability. Improved understanding of the failure process and indicators can aid in early identification and timely warning of geohazards in phosphorite mines.

Application of distinct element methods to simulation of hydraulic fracturing in naturally fractured reservoirs

The Distinct Element Method (DEM) represents a rock mass as an assembly of blocks (polygonal or polyhedral). Contacts between blocks correspond to discontinuities (i.e., fractures or joints) that can exhibit non-linear mechanical behavior, including slip and opening. If flow in rock fracture is approximated using the lubrication equation, coupled hydro-mechanical DEM models can be used for simulation of rock mass treatment by fluid injection. However, this approach has a limited capability for simulating fracture propagation. The synthetic rock mass (SRM) concept overcomes this limitation. In SRM, the bonded particle model (BPM), which is an assembly of circular or spherical particles bonded to each other, represents deformation and damage of intact rock. If pre-existing discontinuities are represented in the BPM, the resulting model, referred to as SRM, has the capability of simulating hydraulic fracturing in naturally fractured reservoirs. The model delivers a pattern of hydraulic fractures that evolves in response to both intact rock fracturing and sliding and opening of pre-existing joints.

Links between orogenic and placer gold on the Old Man Range, Central Otago, New Zealand

Gold-bearing normal faults of inferred Late Cretaceous age cut both Caples Terrane and Torlesse Terrane schists on the Old Man Range in Central Otago. Mineralised fault rocks are dominated by brittle-formed breccias and fault gouges, with minor hydrothermal silicification and quartz vein formation, and scattered pyrite and arsenopyrite. Gold occurs replacing silicates, encapsulated in sulphide minerals, and as free particles in vein quartz. The orientations of mineralised faults were controlled by pre-existing joints in the host rocks, with west strike in Torlesse schists and northwest strike in Caples schists. A major component of placer gold on the slopes of the Old Man Range has been derived from nearby mineralised faults, and detrital gold particles are equant and angular rather than flaky. Gold in a Late Pleistocene Clutha River channel is flaky and has been transported from outside the area, but there is little or no evidence of externally derived detrital gold along the course of a Middle Pleistocene Clutha River channel.

Rockslides on limestone cliffs with subhorizontal bedding in the southwestern calcareous area of China

Abstract. Calcareous mountainous areas are highly prone to geohazards, and rockslides play an important role in cliff retreat. This study presents three examples of failures of limestone cliffs with subhorizontal bedding in the southwestern calcareous area of China. Field observations and numerical modeling of Yudong Escarpment, Zengzi Cliff, and Wangxia Cliff showed that pre-existing vertical joints passing through thick limestone and the alternation of competent and incompetent layers are the most significant features for rockslides. A "hard-on-soft" cliff made of hard rocks superimposed on soft rocks is prone to rock slump, characterized by shearing through the underlying weak strata along a curved surface and backward tilting. When a slope contains weak interlayers rather than a soft basal, a rock collapse could occur from the compression fracture and tensile split of the rock mass near the interfaces. A rockslide might shear through a hard rock mass if no discontinuities are exposed in the cliff slope, and sliding may occur along a moderately inclined rupture plane. The "toe breakout" mechanism mainly depends on the strength characteristics of the rock mass.

Why Fracking Works

Although spectacular advances in hydraulic fracturing, also known as fracking, have taken place and many aspects are well understood by now, the topology, geometry, and evolution of the crack system remain an enigma and mechanicians wonder: Why fracking works? Fracture mechanics of individual fluid-pressurized cracks has been clarified but the vital problem of stability of interacting hydraulic cracks escaped attention. First, based on the known shale permeability, on the known percentage of gas extraction from shale stratum, and on two key features of the measured gas outflow which are (1) the time to peak flux and (2) the halftime of flux decay, it is shown that the crack spacing must be only about 0.1 m. Attainment of such a small crack spacing requires preventing localization in parallel crack systems. Therefore, attention is subsequently focused on the classical solutions of the critical states of localization instability in a system of cooling or shrinkage cracks. Formulated is a hydrothermal analogy which makes it possible to transfer these solutions to a system of hydraulic cracks. It is concluded that if the hydraulic pressure profile along the cracks can be made almost uniform, with a steep enough pressure drop at the front, the localization instability can be avoided. To achieve this kind of profile, which is essential for obtaining crack systems dense enough to allow gas escape from a significant portion of kerogen-filled nanopores, the pumping rate (corrected for the leak rate) must not be too high and must not be increased too fast. Furthermore, numerical solutions are presented to show that an idealized system of circular equidistant vertical cracks propagating from a horizontal borehole behaves similarly. It is pointed out that one useful role of the proppants, as well as the acids that promote creation of debris in the new cracks, is to partially help to limit crack closings and thus localization. To attain the crack spacing of only 0.1 m, one must imagine formation of hierarchical progressively refined crack systems. Compared to new cracks, the system of pre-existing uncemented natural cracks or joints is shown to be slightly more prone to localization and thus of little help in producing the fine crack spacing required. So, from fracture mechanics viewpoint, what makes fracking work?–the mitigation of fracture localization instabilities. This can also improve efficiency by fracturing more shale. Besides, it is environmentally beneficial, by reducing flowback per m3 of gas. So is the reduction of seismicity caused by dynamic fracture instabilities (which are more severe in underground CO2 sequestration).