While mining-induced earthquakes in the deep gold mines of South Africa pose a risk to mines, mineworkers, and the public, they also provide an unusual opportunity to study the physics of earthquakes. The source zone of a M5.5 earthquake that occurred near Orkney, South Africa on 5 August 2014 was well-defined by tens of thousands of aftershocks recorded by instruments deployed as part of a Japanese - South African research project. The upper edge of the M5.5 rupture is located several hundred metres below the mining horizon. A proposal to drill into the M5.5 source zone, as well as several other active faults in other deep mines, was approved by the International Continental scientific Drilling Programme (ICDP) in August 2016. Here we describe the scope and objectives of the project, and the selection and preparation of the principal drilling sites. In addition to the studies of earthquake phenomena, some of the holes will be used by geomicrobiologists to investigate deep microbiological activity fuelled by hydrogen released by seismic rupture to address questions about early life on planet Earth. INTRODUCTION Gold-bearing quartz pebble conglomerates were discovered near present-day Johannesburg in 1886. As mining operations followed the orebody, the dipping ‘reefs’ were found to persist to great depths. Mining-induced seismicity and its hazardous manifestation, rockbursts, were first encountered in the early 1900s, when extensive stopes, supported solely by small reef pillars, reached depths of several hundred metres.

Classical rock support methods for underground engineering emphasize the importance of investigating and controlling the unloading deformation of surrounding rock, such as in the New Austrian Tunnelling Method (NATM) and Norwegian Method of Tunnelling (NMT). Current engineering practice in underground tunnels and caverns under high geostress conditions indicates that stress-induced collapse and rock bursts are the major engineering hazards. Prior to this kind of failure, crack initiation and propagation occurs. Thus, a new support concept, termed the crack-restraining method, has been developed based on the NATM and NMT for underground hard-rock engineering under high geostress conditions. This method includes three key technical points: The length of the rockbolts is slightly greater than the maximum depth of cracking; the optimal supporting time of the rockbolts is decided based on the crack evolution and deformation tendency; shotcrete with high tensile strength is used to ameliorate the stress state of the rock mass' surface. The aim of these supporting efforts is to restrain the subsequent cracking and enhance the stress-bearing capability of the rock mass. Application of this approach to the Baihetan underground cavern (the largest hydraulic cavern in China) and other deep underground tunnels indicated that this crack-restraining support method is appropriate for determining the rockbolt length and supporting time. INTRODUCTION Increasing attention is being given to deep-level excavation and mining due to the gradual exhaustion of shallow mineral resources and recent discoveries of deep mineral resources. Most South Africa gold mines have reached depths of more than 2000m, and the mining depth at Mponeng almost 4350m (Kwiatek, Plenkers, and Dresen, 2011); the Creighton mine in Canada has reached 2400m in depth (Snelling, Godin, and McKinnon, 2013); the Kolar gold mine has progressed to 2400m (Mishra and Panigrahi, 1999). Greater depths are also being attained in civil engineering projects–some traffic tunnels and hydraulic tunnels have been excavated at depths greater than 1000m (Zangerl et al ., 2008, 2010, 2014). All these engineering projects have shown that the unloading behaviour of hard rock under high geostress is very complex. The support scheme for hard rock under the conditions encountered in these projects obviously presents a great technical challenge.

Letšeng Diamond Mine (LDM) is the highest open-pit diamond mine in the world located at 3 100 masl. Two kimberlite pipes, the Main Pipe and the Satellite Pipe, are currently being excavated and are called the Main Pit and Satellite Pit respectively. Two failures of approximately 3 000 and 2 000 t of material occurred in August and September 2016 respectively in the southern portion of the Main Pit at Cut 3 West (MC3W) along a major shear zone that intersects the southern edge of both pits. An investigation was undertaken to determine the causes of failure in order to mitigate and prevent future failure occurrence. The investigation involved a detailed area inspection, geological mapping, rock mass characterisation, discontinuity survey, kinematic analysis and a back-analysis. The field inspection revealed that the shear zone is composed of sheared and friable kimberlite, calcite, pyrite and altered basalt. Rock mass classification confirmed that the shear zone is made up of a weak rock mass with low rock mass rating. It is evident from geological mapping and discontinuity survey analysis that the shear zone consists of a combination of faulting, kimberlite dykes and closely spaced sympathetic joints which are liable to toppling failure. Stereonet based kinematic analysis using Rocscience Inc. DIPS shows only one cluster, which has an average dip and dip direction of 85°/175°. The mode of failure was a combination of toppling failure on the major planes coupled with unravelling of friable material. Back-analysis using RocTopple suggests that failure may have been triggered by groundwater conditions as a result of infiltration of rain water into the relative weak rock mass. INTRODUCTION The stability of a rock slope is governed by the local geological and geotechnical characteristics of the rock mass, the hydrogeological, geomorphological conditions, and other environmental factors such as freeze and thaw cycles (Wyllie and Mah, 2004; Verma et al. , 2011). Structural geology plays a major role in the design and maintenance of an open pit mine as the presence of structural defects affects the mechanical and hydrogeological properties of the rock mass (Bye and Bell, 2001). Therefore, the assessment of geological structures forms an essential part of rock slope stability analysis and slope management.

Although the application of destress blasting as an active rockburst mitigation measure is not yet commonplace in hard coal longwall mining, the method is assuming heightened importance due to increases in mining depth and high horizontal stresses in the rock mass. The main goals of destress blasting are the softening of competent rock layers, the reduction of strain energy storage and rock mass stress release, which together contribute to minimizing rockburst occurrence and risk. One such region in which destress blasting in competent rock is applied is the Upper Silesian Coal Basin, mainly in its Czech part. Here a case study of destress blasting is presented and evaluated in terms of rockburst prevention, focusing on a thick coal seam (about 5 m) subject to longwall mining under very unfavourable geomechanical conditions (great depth – 1200 m, competent rigid rocks between coal seams, unfavourable stress field due to long-term mining). Destress blasting stages were carried out in a group of boreholes (4–8; diameter 95 mm) with the total explosive charge ranging from 2100 kg to 3750 kg, detonated regularly at a distance of 43 m to 114 m ahead of the advancing longwall face. Evaluation of destress blasting based on seismic monitoring revealed that the longwall blastings were very successful in terms of rock mass stress release and decreasing the rockburst risk, with only one rockburst occurring on a roadway 210 m from the longwall face. The following longwall section was mined safely without rockburst occurrence. INTRODUCTION In the hard coal reserve of the Upper Silesian Coal Basin (USCB), which is shared by the Czech Republic and Poland, longwall mining is the dominant underground mining method. The Czech part of the USCB, known as the Ostrava–Karvina Coalfield (OKR), lies in the northeastern part of the country (Figure 1). The exhaustion of the upper seams due to the continuation of coal mining activities for around 200 years has shifted activity to a greater depth (> 800 m). Under existing mining and geological conditions in the Karvina sub-basin of the USCB, underground extraction of coal is typically accompanied by rockbursts, the first of which occurred in 1912 (Pelnar 1938). A number of attempts have been made to address rockburst activity in both the Czech ( e.g . Straube et al ., 1972; Holecko et al ., 1999; Takla et al ., 2005; Holub, Rušajová, and Holecko, 2011) and Polish parts of the USCB ( e.g . Dubinski and Konopko, 2000; Drzewiecki and Kabiesz, 2008).

As projects advance to deeper levels, rockbursts occur more frequently. The rockburst failure mode is particularly problematic, as the rock mass fails abruptly, releasing large amounts of energy. Laboratory tests and observations in tunnels and mines with high overburden have shown that certain rock types have a high potential for storing elastic energy and are hence more prone to rockbursting. In this research we studied the influence of a rock's heterogeneity at grain scale on its rockburst propensity. Several artificial sample sets consisting of a very fine-grained fibreless ultra-high performance concrete (UHPC) and a constant volumetric fraction of different coarse rock grains as aggregate were produced and tested. The added rock grains were chosen based primarily on their differing stiffness characteristics. One sample set, consisting of pure UHPC, acted as reference for the matrix. The findings of this study confirm that the material's heterogeneity at grain scale has a major impact on the failure behaviour. With increasing stiffness heterogeneity between the ‘rock matrix’ and admixed grains the failure mode becomes more ductile and a lower uniaxial compressive strength is reached. Hence, the material is less prone to rockbursts as less elastic strain energy is developed prior to failure. The results agree well with various rockburst parameters, which were additionally evaluated and compared. The findings allow a better understanding of the underlying mechanism of rockbursts and demonstrate the usefulness of petrographic information in rockburst risk analysis. INTRODUCTION Predicting the behaviour of the rock mass is of utmost importance in tunnelling and mining. Only if the potential failure mechanisms of rock and rock mass are well understood can failure be foreseen and avoided and the consequences reduced. Rockbursts are particularly problematic, as the rock mass fails abruptly, rapidly releasing great amounts of energy. Laboratory tests and observations in tunnels and mines with deep overburden have shown that certain rock types have a high potential to store elastic energy. Hence, they are more prone to rockbursting. The amount of stored energy also determines the magnitude of the event.

Closure estimations by the timber pack assessment method were conducted at Mponeng gold mine at four different sites covering various rock formations and a depth range between 2382 m and 2747 m. The method assumes that the packs are built with timber units having standard thicknesses, allowing the estimation of original stoping width. The estimated closures were calculated by subtracting the current stoping widths from the estimated original stoping widths. An elastic numerical modelling program was used to evaluate the elastic convergences at the same positions where the closures were estimated. It was found that the estimated closure is 56% higher than the modelled convergence. It would be possible to estimate the underground closure from numerical modelling results by using this percentage difference. A limited number of underground observations could then be conducted to support the closure estimations further. INTRODUCTION To exploit the remaining gold reserves in the Witwatersrand Basin, innovative methods must be developed to address the challenges arising with increasing depth. The depths being planned for mining in future, if South Africa still desires to be a major role player in gold production, could not be imagined five decades ago. Mponeng gold mine currently holds the world record of being the deepest mine (AngloGold Ashanti, 2013). The longwall mining method without backfill was employed at moderate depths to 89 level at 2 700 m depth. The mining method was then changed to sequential grid with backfill from 89 level to the lowest level 120 at 3 400 m depth. As the depth increased, the seismic activity on the mine also increased.

The Changchun Shrine, located in the Taroko National Park in eastern Taiwan, was built in 1958 and has been damaged three times by rockfalls. An engineering project was launched to mitigate the influence of rockfalls and enhance the safety of tourists visiting the shrine. Airborne LiDAR and two types of ground-based LiDAR were utilized to acquire point clouds of topography and outcrops nearby the shrine. High-resolution point cloud data can be used not only to interpret the large-scale geological linearities, erosion gullies and scarps, but also the attitudes of ubiquitous discontinuity sets to find the potential rock wedge. Statistical hypothesis testing is used to validate the representativeness of the interpreted attitudes of discontinuity sets, which are used in stereographic projection analysis of possible rock wedge movement and related design for rockfall protection measures. INTRODUCTION The Changchun Shrine was built in 1958 to commemorate the workers who constructed No. 8 Highway and died in the line of duty. The shrine is a well-known tourist site within the Taroko National Park, eastern Taiwan, associated with beautiful cliffs and perennial waterfalls. A zigzag trail crosses steep slopes behind the Changchun Shrine. However, geological conditions in the vicinity of the shrine, formed in two geological units, are unfavourable for engineering due to the occurrence of several sets of joints, leading to frequent landslides on varying scales. The first serious disaster occurred in 1980 and damaged the first-generation Changchun Shrine. The second-generation Changchun Shrine was rebuilt after three years, and the second serious disaster occurred in 1987. The third-generation Changchun Shrine was opened to public access in 1997, with several rockfalls being observed in 2015. Figure 1 shows the three generations of the Changchun Shrine.

In most rock engineering projects, a fully grouted rockbolt system is the primary support for reinforcing unstable rock strata. A better understanding of the characteristics of the load transfer mechanism between rockbolts and rock mass could lead to a significant improvement in the performance of rockbolts. Previous studies have pointed out that the bond-slip relationship between rockbolts and the rock mass is nonlinear and hence the bond-slip model in Fast Lagrangian Analysis of Continua (FLAC2D) could not adequately represent the nonlinear behaviour of rockbolts. In this study we numerically modelled the behaviour of fully grouted rockbolts subjected to tensile loads by implementing a nonlinear bond-slip relationship for the bolt-grout interface into FLAC2D. The proposed numerical model was verified by pull-out tests on rockbolts. The findings show that the modified rockbolt model in FLAC2D is capable of capturing the nonlinear shear bond stress behaviour. INTRODUCTION In many engineering projects, especially underground excavations, a fully grouted rockbolt system is the primary support used to reinforce unstable rock strata. A better understanding of the characteristics of the load transfer mechanism between rockbolts and the rock mass could lead to a significant improvement in the performance of rockbolts. Various studies have been conducted to investigate the load transfer mechanism of rockbolts (Benmokrane, Chennouf, and Mitri, 1995; Aziz, 2004; Ren et al ., 2010; Ma, Nemcik, and Aziz, 2013; Nie et al ., 2014a, 2014b; He, An, and Zhao, 2014; Martin et al ., 2011a; Martin, Tijani, and Hadj-Hassen, 2011b; Ivanovic and Neilson, 2009). Benmokrane, Chennouf, and Mitri (1995) carried out a series of bolt pull-out tests and found that the relationship between the interfacial shear bond stress and the relative slip between the bolt and grout is not linear, as schematically shown in Figure 1. The term ‘bond-slip relationship’ is used to represent the relationship between the bond stress and the slip. The bond-slip relationship could significantly impact on the performance of fully grouted rockbolts under axial loading. Ren et al . (2010) and Martin, Tijani, and Hadj-Hassen (2011b) presented their respective analytical models in which a trilinear bond-slip relationship was considered. Ma et al . (2014) proposed an analytical bolt model based on a nonlinear bond-slip relationship. These analytical models are able to predict the mechanism of interfacial bond weakening of rockbolts.

Tunnelling in weak rock under high stresses often results in damage to the lining and failure of bolts. Large displacements can lead to violation of the clearance profile, requiring costly, dangerous, and time-consuming repairs. The main problem is that conventional supports have a rather limited strain tolerance. Shotcrete, for example, can tolerate a maximum of 0.6% to 0.8% strain without failure. Any imposed strains beyond this level will lead to destruction of the lining and consequently a more or less complete loss of support resistance, causing additional displacements and disintegration of the rock mass. Over the last two decades, ductile elements have been developed and successfully used on a considerable number of tunnels. The aim of these elements is optimal utilization of the capacity of the lining, while preventing damage of the lining by limiting the resistance. Although this sounds simple, the task of designing such a ductile lining system is not at all trivial. Many factors need to be considered, like development of the displacements in relation to face advance and time. The ground structure and the relative orientation between the tunnel axis and dominant geological features, like foliation or bedding, strongly influence the longitudinal displacement profile. The development of the displacements over time also depends on the advance rate, influencing the degree of utilization of the shotcrete, which changes its properties over time. Currently used ductile element systems are presented and their characteristics as well as pros and cons discussed. More or less sophisticated tools for a preliminary design of ductile linings exist. Examples of applications are presented. Another issue is the performance of grouted rock bolts under large strains. Installation of bolts at the time of the largest strain rates in the rock mass close to the face results in considerable shear strains in the grout while it is setting. This may lead to a loss of bond between bolt and ground, making the bolts more or less ineffective. Special rib designs of the bolts can reduce this effect.

In this study, rockbolts made of polypropylene random copolymer (PP-R) material were assessed as an alternative to steel split sets in a series of experimental studies that included both laboratory and field-scale tests. PP-R tubes and typical steel split sets were compared by means of static loading (pull and shear), impact loading, corrosion, diametric compression, insertion into drill-holes, and creep tests. Despite the advantages of PP-R; being a noncorrosive material, being shaped by the roughness of drill-hole surfaces, and having ideal support reactions because of high resistance to crack propagation, strain relaxation was found to be a considerable problem with the PP-R tubes inserted into the drill-holes. The results suggest that further work should be undertaken to develop new bolts using PP-R tubes instead of steel split sets. INTRODUCTION Split set rockbolts are popular, especially in the mining industry, as a result of their advantages such as practical application, ability to start carrying load without waiting for curing of the grout materials, and good load-bearing capacity despite high deformation. However, underground water can substantially affect steel materials and decrease their load-bearing capacity, especially with long contact times and acidic groundwater conditions (Hoek, 2006; Komurlu and Kesimal, 2015; Hassel and Villaescusa, 2005). The two main factors affecting the frictional load-bearing capacity of the split sets are the normal stress on the drill-hole surface and the coefficient of friction at the contact interface between the bolt and drillhole surfaces (Li, Stjern, and Myrvang, 2014; Heerden 2007; Qingliang et al ., 2013; Perras, Diederichs, and Besaw, 2014). Although a rigid rockbolt material is advantageous as the normal stress on the friction surface is supplied due to the decrease in the slit width and diameter of the bolt inserted into the drillhole, a rigid surface prevents the bolt from being shaped by the roughness of the drill-hole surface, which decreases the coefficient of friction. Rockbolt loading mechanisms resulting from rock mass deformation can be divided into axial and shear loading types. The rockbolts are exposed to both axial and shear loading as the rock mass deforms. The frictional load-bearing capacity plays an important role in changing the support reactions under both shear loading and the axial loading conditions (Pellet and Egger, 1996; Srivastava and Singh, 2014; Oreste and Cravero, 2008; Ranjbarnia, Fahimifar, and Oreste, 2016; Garga and Wang 1993).

Parameters needed for rock support design are discussed. Underground loading conditions and the methodology of rock support are covered first, followed by the parameters related to rock support design; that is, the factor of safety, the maximum allowable displacement, the ultimate displacement of support elements and the released strain energy, and their application in support design. The factor of safety needs to be calculated in different ways depending on whether the loading condition is gravitational, squeezing, or a dynamic rockburst. INTRODUCTION The major instability issue related to high in-situ stresses is either rockbursting in hard and strong rock or time-dependent squeezing in soft and weak rock. The ‘driving force’ behind both rockbursting and rock squeezing is the strain energy released from the rock mass after excavation. In the case of rockbursting, the energy released in a seismic event has to be absorbed by the rock support system in order to avoid rock collapse. Therefore, it requires a dynamic support system capable of absorbing a good amount of energy prior to failure. This concept was proposed as early as in the 1960s, and has been well acknowledged by others to date (Cook and Ortlepp, 1968; Ortlepp, 1969, 1992). A good number of energy-absorbing rockbolts have been invented (Ortlepp Bornman, and Erasmus, 2001; Charette and Plouffe, 2007; Varden , 2008; Li, 2010; Cai and Champaigne, 2010; Wu and Oldsen, 2010; Galler, Geschwandtner, and Doucet, 2011; He et al ., 2014; Darlington 2014). On the other hand, there exists a maximum allowable displacement from the point of view of engineering practice. Therefore, two parameters, the amount of energy to be absorbed and the maximum allowable displacement, need to be taken into account in dynamic rock support design. In addition, the displacement capacity of support elements, such as rockbolts, needs to be also taken into account. A satisfactory dynamic rock support system should be capable of absorbing a good amount of energy, but should also be capable of constraining the rock displacement within the maximum allowable limit. In principle, a strong support system is desired so that it does not need to displace much to absorb a given amount of released energy. In this paper, the methodology of rock support is discussed first and then the parameters used for rock support design, such as the factor of safety and the maximum allowable displacement, are presented for the following three types of rock conditions: rockfall under gravity, rock squeezing, and rockburst.

The slopes in open pit mines are typically excavated to the steepest feasible angle to maximize profits. However, there is a greater risk of slope failure associated with steeper slopes. An open pit slope represents a complex multivariate rock engineering system. Interactions between the factors affecting slope stability in open pit mines are therefore more complex and often difficult to define, impeding the use of conventional methods. To address the problem, the primary role of rock mass structure, in-situ stress, water flow, and construction have been extended into 18 key parameters. The stability status of slopes and importance of these parameters are investigated by means of computational intelligence tools such as artificial neural networks. An optimized back-propagation network is trained with an extensive database of 141 worldwide case histories of open pit mines. The inputs refer to the values of extended parameters, which include 18 parameters relating to open pit slope stability. The output is an estimated potential for instability. To minimize subjectivity, the method of partitioning the connection weights is applied in order to rate the significance of the parameters involved. The problem of slope stability is therefore modelled as a function approximation. A new open pit mine slope stability index is thus proposed to assess the potential status regime from a holistic point of view. These values are validated by computing the predicted values against the observed status of stability. The reliability of the predictive capability is computed as the mean squared error, and further validated through a receiver operating characteristic curve. Together with a mean squared error of 0.0001, and receiver operating characteristic curve of 98%, the application illustrates that the prediction of slope stability through artificial neural networks produces fast convergence giving reliable predictions, and thus constitutes a useful tool at the preliminary feasibility stage of study.

In open pit mining, a rock mass model can provide important input for the optimization of production in relation to stability challenges and a safe workplace. In this sense, rock mass classification systems are useful for defining rock mass domains that identify areas of poorer ground conditions that may require more detailed investigation or modifications to mine designs. These methods are widespread among geotechnical engineers as one practical way to assess quality of the rock mass. In open pit mine bench faces with no clear discontinuities, one single joint may dominate the pit slope plane failure mechanism. Analysis of low rock mass quality zones combined with the jointing orientation may provide an improvement in the assessment of instabilities. The aim of this paper is to develop a rock mass model of an open pit mine under active operation, based on extensive field mapping. Through this process, surface maps based on different rock mass classifications (RMR 89 , GSI, and Q-system) systems will be established for the mine. The mine was classified and assessed using RMR 89 . Finally, a three-dimensional rock mass model was developed based on the values obtained from surface and underground mapping. It was hoped that such a model allowed the identification of areas of the mine that are likely to have poorer ground conditions and may require more attention during the design and operational phases of the mine life. INTRODUCTION The planning of an open pit is a matter of determining the most profitable pit limit and the most economical mining sequence for a given orebody. In the exploitation phase, the excavation sequence implies the development of artificial slopes the stability of which is a crucial factor both for safety and economics of mining. Therefore, the assessment of rock slope stability is an essential requirement for the open pit mining industry, not exclusively during the engineering studies, but also throughout the operational life of a project. For this matter, rock mass classification systems are extensively used in order to quantitatively address the quality of the exposed rock mass in open pit slopes. A variety of rock mass classification systems have been proposed in the last 50 years, many of which were originally intended for tunnel design and subsequently adapted or modified for use in slopes. Four systems have gained broad acceptance in the mining industry: the RMR89 (Bieniawski, 1989), the Q-system (Barton, 1974), the MRMR (also known as Laubscher's RMR: Laubscher, 1990), and the GSI (Hoek, 1994). Flores and Karzulovic (2002) stated that the most widely used methods for rock mass classification in underground mines are Laubscher's RMR89, or MRMR, (53%), followed by Barton's Q (26%), and Bieniawski's RMR89 (15%). On the other hand, the most widely used method for rock mass classification in open pit mines is Hoek's GSI (39%), followed by RMR89 (26%) and MRMR (22%). This is shown in Figure 1.

In a previous study, the authors investigated the cable-strapping of pillars in room and pillar mines in order to solve local instability problems, by conducting laboratory tests. Based on experience in an old, small room-and-pillar haematite mine in Spain, the authors suggested a few steel cables placed around mesh-wrapped pillars as an effective means to improve pillar stability and control mine deformation. In the present study, and with the background of available laboratory uniaxial compressive strength tests on cabled samples and empirical knowledge of pillar behaviour, we present the results of numerical modelling of this type of pillar. Using FLAC and UDEC, it has been possible to develop models representing pillar response. These models add to the knowledge of pillar behaviour and improve our control of both standard and cable-strapped pillars, in terms of stress-strain evolution, plastic response, and the energy that the pillars are able to absorb. INTRODUCTION Standard practice, as well as published data on pillar design, showed that neither empirical formulae nor numerical modelling can fine-tune pillar design. However, either of these techniques (or better, both together) can be used as a preliminary approach to pillar design, provided monitoring is then carried out to control pillar performance (Malan and Napier, 2014; Esterhuizen, Dolinar, and Ellenberger , 2011). Rocks and rock masses tend to show variable behaviour and significant scale effects, so in a mine where a large number of pillars are generated in the same formation, a certain amount of variability in strength can always be expected. It is therefore not surprising that, even if by means of traditional approaches (tributary area and empirical formulae) and numerical models a design is reliable, in the mining practice (where pillars do not always have the prescribed dimensions and the rock has a variable nature) stability problems may arise – typically only in a few pillars, unless a very conservative approach was adopted.

Large-scale underground excavations usually pose a threat to the stability of the rock mass, and the damage is required to be modelled in order to plan the excavation schedule and support design. Various methods, such as displacement measurement, acoustic wave testing, and numerical calculation have been used to quantify the rock mass damage. In the present study, a microseismicity (MS) monitoring system was established to monitor the microcracks induced by excavation of the underground powerhouse at the Houziyan hydropower station in Sichuan Province, China and a damage-driven numerical model based on MS data was adopted to quantify the surrounding rock mass damage. The rock mass is simulated by a micromechanics-based constitutive model described by the fracture density, which is determined by the amount and size of the microcracks. The material property inputs for this model are Young's modulus, Poisson's ratio of the intact rock, and the MS information such as the location and the source size calculated from seismic source parameters. A typical profile of the underground powerhouse is selected for damage analysis. Combined with the excavation process, the related MS data is imported to the numerical model for damage calculation. The results show that the surface displacement of each bench excavation of the underground powerhouse using the damage-driven numerical model is approximately 30–40 mm larger than that using the original elastic model. Furthermore, the displacements of the rock mass considering the effect of MS damage are in good agreement with the field measurements.

The hydraulic characteristics of the rock mass are of great importance in many rock engineering projects. Various factors affect the characteristics, but in this study, the effect of roughness on hydraulic conductivity was investigated by a series of experiments. Based on the fractal theorem, and controlling roughness of a rock joint with 3D printing technology, we investigated the variation of hydraulic conductivity of a single rock joint. Two levels of joint roughness were tested under various water and normal pressures. As a result, a relationship between hydraulic conductivity and rock surface roughness is proposed. INTRODUCTION The hydraulic characteristics of the rock mass are of great importance in many civil and mining engineering projects such as grouting (Barton and de Quadros, 1997), oil/gas production (Wu, 2002), geothermal energy (Min, Rutqvist, and Elsworth, 2009) and tunnelling. The hydraulic characteristics of the rock mass normally interact with the mechanical conditions of rock discontinuity, therefore the coupling behaviour between them should be taken into account. However, in general, investigating this experimentally requires quite complex procedures, therefore the investigation variation in hydraulic properties according to various mechanical conditions has frequently been adopted in many previous studies (Brown, Caprihan, and Hardy, 1998; Lee and Cho, 2002; Singh, Singh, and Ranjith, 2015). In most rock engineering projects, fluid flow occurs mainly through rock discontinuities. Therefore, many experimental studies have focused on fluid flow in a single rock joint. Fluid flow through a joint can be affected by many factors, such as roughness, aperture, stiffness of joint surface, infilling material, and so on. Among these, aperture and roughness, which are closely related, are known to be highly influential. Roughness is usually controlled in the studies because of its relative simplicity. Nonetheless, controlling roughness requires excessive effort in terms of sample preparation because a joint surface having a desired roughness is very hard to obtain in ordinary procedures. Recent development in 3D printing technology can be used to overcome this cumbersome procedure since one can obtain any level of roughness using point cloud-type data.

Large underground unlined rock caverns for storage of crude oil or liquid hydrocarbons are very often designed to operate with a gaseous phase maintained to a pressure range above atmospheric pressure (>1 bar abs.). To maintain the pressure of the gaseous phase in a cavern above atmospheric pressure it is necessary to inject a compensation gas (gas in) during the cavern emptying phase. The reverse operation (gas out) must be carried out during product filling to avoid overpressurization in the cavern. The management of this compensation gas requires a dedicated industrial unit for production, recovery, and treatment. In case of large underground caverns with net capacity of several million tons, the construction and operation costs for such a unit become prohibitive. To limit the need for compensation gas, the permissible pressure range in operation must be extended. Whereas the maximum gas pressure is mostly limited by the tightness issue, the minimum pressure is mainly governed by stability aspects. In this paper, the possibility of lowering the minimum pressure in operation below the atmospheric pressure is investigated, focusing on the cavern stability aspects. During construction at atmospheric pressure, for the large cavern section and rock mass quality expected, long-term stability can be achieved by conventional spraying of shotcrete and rockbolt installation. During operation, minimum pressure in the storage cavern close to vacuum is investigated and its impact on reinforcement need is reviewed using results of two distinct approaches: homogeneous continuum as well as wedge stability. The particularities regarding the required reinforcement for caverns operated at atmospheric pressure are also investigated. INTRODUCTION Underground rock caverns excavated for the storage of crude oil have capacities of several millions of cubic metres. These caverns are typically excavated in hard rocks, allowing the excavation of large section supported with rock bolts and shotcrete. To reduce the construction cost of a vapour recovery unit and the associated cost of gas compensation during operation of underground rock caverns for storing liquid hydrocarbons, the minimum operating pressure anticipated in the cavern is extended below the atmospheric pressure. As the lower pressure is applied around the cavern wall during operation only, it is important to distinguish between the reinforcement needs during construction and during operation to optimize the construction schedule of the project. Indeed, only the support required to achieve stability during construction remains on the critical path, whereas the installation of the support needed to achieve stability in operation can be delayed and rejected from the critical path.

The crack initiation and propagation process of hydraulic fracturing were reproduced using a coupled particle flow code (PFC) computational fluid dynamics (CFD) method in mixed ground. The breakdown pressure of mixed ground was investigated based on orthogonal test principles and five influencing factors were analysed. The crack propagation process in mixed ground was investigated under different normal bond strength ratios between hard and soft grounds and stress ratio. The minimum principal stress has the most influence on breakdown pressure, followed by stress ratio. Stress ratio was the determining factor for crack initiation direction. The result can be used in the engineering design of hydraulic fracturing projects. INTRODUCTION Hydraulic fracturing is a process of initiation and propagation of cracks, which is used widely in the petroleum industry to stimulate oil and gas wells to increase their productivity (Adachi et al ., 2007). Various numerical methods have been developed to investigate hydraulic fracturing. The finite element method (FEM) and the boundary element method (BEM) have been introduced to simulate hydraulic fracturing in complex structures (Papanastasioou, 1997; Vychytil and Horii, 1998; Hossain and Rahman, 2008; Aghighi and Rahman, 2010). The discrete element method (DEM) has also been used to investigate hydraulic fracturing from the microscopic perspective (Al-Busaidi, Hazzard, and Young, 2005; Shimizu, 2010; Shimizu, Murata, and Ishida, 2011; Han, Damjanac, and Nagel, 2012; Wang et al ., 2014), and many significant conclusions have been obtained. Particle Flow Code (PFC), which is a DEM, has become an effective tool for simulating crack propagation, though it is not perfect (Potyondy and Cundall, 2004). However, there is little information in the literature concerning the hydraulic fracturing process in mixed ground based on DEM. In the current study, PFC2D (Itasca, 2010) was used to simulate the hydraulic fracturing process in mixed ground. The factors influencing breakdown for mixed ground were analysed based on orthogonal test principles. Crack propagation in mixed ground was investigated under different normal bond strength ratios between hard and soft ground ( R nb ) and stress ratios ( R σ ).

Understanding the evolution of micro-cracking induced during rock indentation by roller tricone drilling tools is a key aspect of maximizing drilling efficiency. Crack network patterns and interaction between neighbouring cracked zones induced by two adjacent tool bits are a challenge for drilling engineers. In this paper, a methodology based on the cavity expansion model is developed to analyse the indentation of rocks by a blunt spherical indenter. Rock elasto-plasticity behaviour is used to propose an analytical solution for the plastic zone radius and a description of stresses within and outside the plastic zone. A Maple program was written to derive the expressions of plastic zone radius and stresses when adopting either the Mohr-Coulomb criterion or the Hoek- Brown criterion. The two approaches were then compared. All analyses were conducted on two hard rock types: granite and limestone. The first is encountered in geothermal reservoirs and the second in petroleum drilling. The results were also compared with those obtained using an experimental procedure developed in a previous investigation, in which the cracked zone radius is estimated using image analysis to distinguish the indentation microcrack network in the vicinity of the tool bit. INTRODUCTION Rock indentation has been intensively studied for more than three decades, for two main reasons: the requirements of the underground space sectors (storage tunnels, mining activities) and the development of rock excavation technology. Drilling performance depends on the rate of penetration of the tool, which can be predicted by numerous methods, including those based on indentation tests on representative samples. Several phenomena occur during the indentation process. The first phase of deformation is elastic; however, the loading evolution leads inevitably to the development of irreversible deformation, which proceeds to the initiation and propagation of cracks due to the tensile stresses, and then to fragmentation. This behaviour of the rock under indentation has been widely studied (Lawn and Wilshaw, 1975; Kutter and Sanio, 1982; Sanio, 1985; Nelson, Ingraffea, and O'Rourke, 1985; Wijk, 1989; Fowell, 1993, Kou et al. , 1995; Tan, Kou, and Lindquist, 1996; Alhossein and Hood, 1996).

Karst caves are often encountered during tunnelling in China. When the caves are filled with groundwater with a high static head, drilling and blasting may induce failure or collapse in the cave wall, resulting in a water inrush or debris flow into the tunnel. The mechanism of damage to the cave wall must be elucidated in order to mitigate the hazards. A conceptual model was set up based on the Yuanliangshan Tunnel, where a cave wall collapse and debris flow were induced by drill-and-blast operations. Particle Flow Code was utilized to simulate the collapse of the cave wall and the evolution of the damage. INTRODUCTION Approximately one-quarter of the total land area of China is underlain by carbonate rocks. The carbonate rocks occur mainly in Guizhou, Yunnan, Guangxi, Guangdong, Sichuan, Hunan, and Hubei provinces, and in Chongqing municipality. The total area of the carbonate rocks, including sections covered by later formations, accounts for over 70% of the land area of China. The carbonate layers are generally 200 m to 300 m in thickness; the thickest layer is 1900 m (Lu, 1986). Water inrushes, mudbursts, tunnel collapses, loss of surface water and groundwater, and surface subsidence are typical geohazards resulting from tunnel construction in the deep karst zones and often result in the evolution of geohazard chains (Han, Wu, and Wang, 2007). Karst caves are often encountered during tunnelling. Occasionally the caves are filled with mud and groundwater with a high static head. As the surrounding rock is excavated, the cave wall becomes thinner and may collapse, inducing water inrushes and mudbursts, which can further decrease the stability of the surrounding rocks and result in tunnel collapses. The dewatering cone of tunnel drainage and excessive water inrushes can deplete the surface and upper groundwater systems, and this may be followed by extensive subsidence of the surface.