東榮一礦1.5Mta新井設(shè)計(jì)【含CAD圖紙+文檔】,含CAD圖紙+文檔,東榮一礦,mta,設(shè)計(jì),cad,圖紙,文檔 翻 譯 部 分 英文原文 Triggering of Seismicity Remote from Active Mining Excavations By S.D.McKinnon Department of Mining Engineering, Queen’s University, Kingston, Ontario, Canada Summary Observations of seismicity and ground control problems in the Sudbury mining camp have shown that late-stage (young) sub-vertical strike-slip faults are sensitive to small mining-induced stress changes. The strength-limited nature of stress measurements made in the region indicates that these structures are in a state of marginal stability. Numerical continuum models are developed to analyze the behavior of such structures. In the models, shear strain localizations (faults) evolve such that there is close interaction between the fault system, stresses, and boundary deformation. Fault slip activity in these systems is naturally sporadic and reproduces the commonly observed Gutenberg-Richter magnitude frequency relation. It is shown that a relatively minor disturbance to such a system can trigger significant seismicity remote from the source of the disturbance, a behavior which cannot be explained by conventional numerical stress analysis methodologies. The initially uniform orientation of the stress field in these systems evolves with increasing disorder, which explains much of the scatter commonly observed in data sets of stress measurements. Based on these results, implications for stress measurement programs and numerical stability analysis of faults in mines are discussed. Keywords: Triggering, microseismicity, stress analysis, rockbursts, fault stability. 1 Introduction The majority of seismic events around deep hard rock mines occur close to excavation boundaries. These events are related to mining-induced stress changes leading to damage involving fracturing of intact rock or slip along pre-existing discontinuities. Extraction layouts leading to highly stressed structures such as pillars and abutments are particularly prone to induced seismicity. With appropriate calibration of rock mass strength, numerical stress analysis can be used to estimate the extent of fracturing and therefore the extent of near-excavation seismicity (Beck et al., 1997; Potvin and Hudyma, 2001; Beck and Brady, 2002). Characteristics of near-excavation seismicity include swarms of events triggered by production blasts (which cause a rapid change in the stress field), followed by a gradual decay in event frequency to background levels over a period of hours or days. The regularity in the frequency and location of near-excavation events makes this type of seismicity a manageable mining problem. A certain amount of seismicity also occurs further away from mining excavations and appears to be uncorrelated in time and space with mining activities. Events have been recorded hundreds of meters away from active mining. On the basis of source locations, it has long been recognized that these events are the result of slip on preexisting structures such as faults, dykes or contacts (Smith et al., 1974; Gay et al., 1984). Although the number of events close to mining excavation boundaries vastly exceeds those further away, the latter are of great concern to mining since they tend to be of larger magnitude, increasing the risk of rockburst damage. Since neither their location nor magnitude can be predicted in advance, mines must consequently make more extensive use of heavier ground support to control potential rockburst damage than would be required if events were only located close to active mining excavations. Due to the complex geometry and geological environment of most mines and the availability of numerous commercially supported codes, numerical stress analysis is the tool of choice for the majority of mining stability analysis. However, only a limited amount of success has been obtained in using numerical modelling to understand seismic events on faults and other geological structures. In particular, numerical stress analysis has not been able to explain the occurrence of seismic events remote from mining. Due to the widespread use of numerical stress analysis in the mining industry, it would be desirable to develop a methodology that would enable modelling to be used to explain both types of seismic events. The objective of this paper is to present an investigation into the cause of seismic events remote from mining and the implications for applying numerical stress analysis to the problem. As would be expected, this type of seismicity is strongly influenced by the geological environment in which the mine is located. Motivation for the approach to modelling mining-remote seismicity is taken from stress measurements and known geological controls of seismicity in the Sudbury mining camp, which is a region of intensive mining activity in Ontario, Canada. 2 Structural Geology and State of Stress in the Sudbury Structure Based on a variety of stress indicators, the stress field in mid-continental North America is characterized by a horizontal major principal stress with an orientation approximately ENE (Zoback and Zoback, 1980; Zoback, 1992b). Arjang (1991) found a similar orientation for the major principal stress based on overcoring stress measurements made in mines in the Eastern Canadian Shield. Using a more extensive database of stress measurements, Arjang and Herget (1997) noted that although there are reasonable consistencies in factors such as the orientation of the major principal stress and the ratios of horizontal to vertical stresses, there was a very large scatter in the trend of the sub-horizontal major and intermediate principal stresses. A significant portion of the stress measurements available in Eastern Canada were made in mines located in the Sudbury Structure. Poles of principal stresses measured in these mines are shown in the lower hemisphere projections of Fig. 1(a). Stress measurement data used in Fig. 1(a) were taken from Arjang (1998), plus unpublished data from INCO Ltd. (Galbraith, 2002). The minor principal stress is typically subvertical, which shows the least amount of scatter in orientation. The major and intermediate principal stresses generally dip sub-horizontally and have significantly higher variation in orientation compared to the minor principal stress. This is shown more clearly in Fig. 1(b), in which the data is restricted to those measurements with minor principal stress within ±20°of vertical. When displayed in strength space, Fig. 2, the stress data is strongly suggestive of a linear envelope (for the regression line, R2= 0.85). There is also a reasonably high degree of correlation (R2 = 0:70) between the magnitude of the two sub-horizontal principal stresses, Fig. 3. The implication is that the state of stress is strength limited, similar to the strength envelope formed by laboratory compression tests. If stresses were not in a state of limiting equilibrium, they would not necessarily define a failure envelope. Since large scale structural discontinuities form the weakest link in the fabric of the rock mass, this correlation reflects a limiting equilibrium relationship between the state of stress and the strength of some of the faults. Simple Andersonian ranking (Anderson, 1951) of the principal stresses (the commonly used geological method of determining style of faulting that would be expected for various orientations of principal stresses) suggests that thrust faults would be the most likely candidates for marginal stability. Fig.1.Lower hemisphere projection of principal stresses, a Sudbury region stress measurements, b subset of measurements with minor principal stress within ±20° of vertical This leads to an apparent discrepancy between the type of faulting most favored in an Andersonian sense, i.e. stress measurements indicating a thrust faulting regime, and the apparent marginal stability of the late-stage strike-slip faults and fractures. Cochrane (1989) addressed this problem by noting that hydrofracture stress measurements made in the Paleozoic cover of the mid-continental stress province, Haimson and Doe (1983), indicated that at depth the vertical stress is greater than the minimum principal horizontal stress, suggesting that strike slip faults originated at depth and that subsequent erosion has reduced the vertical stress to the minor principal stress. The relationship between structures and stress conditions in Eastern Canada has also been complicated by repeated episodes of glaciation post-dating the formation of most structures, which suggests that the neotectonic stress field is not completely consistent with the structural fabric. However, based on the strength-limited nature of the stress measurement data and the sensitivity of the late-stage faults and fractures to mining induced stress changes, it would appear that these structures, as opposed to any other system, are in a state of marginal stability with the regional tectonic stresses. Fig.2.Major and minor principal stress Fig.3.Sub-horizontal principal stress measurement data, Sudbury region. measurement data, Sudbury region. This geological setting provided the framework within which numerical stress analysis was used to investigate the occurrence of seismicity remote from mining excavations. The observations of Cochrane (1989) show that geological structures are not equally sensitive to mining-induced stress changes, and that in the Sudbury Structure the most important structures to account for are the late-stage sub-vertical strike-slip faults. While the importance of other faults in the system is unknown at this time, the focus on the behaviour of the sub-vertical faults enables the numerical representation to be simplified to two-dimensions, as will be described in more detail below. 3 Review of Numerical Stress Analysis Applied to Mining-induced Seismicity The standard approach to assessing slip on faults due to mining is well documented (CAMIRO, 1997). While the particular choice of numerical stress analysis code and method (continuum, discontinuum, boundary element etc.) may vary, the principles involved in assessing fault stability remain the same. The analysis sequence can be described by the following steps: the pre-mining state of stress is initialized, the new state of stress due to the mining geometry is computed, normal and shear stresses on the fault surfaces are resolved and compared to the assigned strength, and the potential for slip evaluated. Once the slip area has been computed, an estimate of maximum seismic moment and event magnitude can be made (CAMIRO, 1997). If elastic continuum codes are used, faults are generally not incorporated into the models and therefore have no effect on the stress field. Slip on these virtual faults is generally assessed using the Excess Shear Stress (ESS) criterion (Ryder, 1987). Examples of this approach can be found in the work of Board (1996); Urbancic and Trifu (1998); Hanekom (2001); Beck and Brady (2002). Using discontinuum codes, faults may be incorporated explicitly into models. However, since fault surface displacement is typically equilibrated with the initial stress field, no history of deformation is accounted for and there is generally little influence of faults on the pre-mining state of stress. Examples of this approach can be found in CAMIRO (1997). A more sophisticated approach has recently been proposed by Wiles et al. (2001). Using a boundary element method of analysis, observations of seismicity are used to prescribe a displacement on fault segments represented by displacement discontinuity elements. In this manner, a pre-strain is incorporated into seismically active fault segments, resulting in a local distortion of the stress field. This approach is promising, but is limited by the requirement that fault displacements can only be accumulated in the models based on estimates of slip computed from observations of seismicity. Significant seismic events resulting in rockburst damage can occur on faults with no prior history of activity. Also, this piecewise modification of the model is not well suited to the mine design process in which an assessment of risk is often required prior to extensive mining. A common limitation of these methods, and in particular the issue of explainin seismic events remote from mining, is related to fundamental assumptions about fault strength, the state of stress, and the degree of stability of fault segments. The initiation of a homogeneous state of stress prior to mining and the superposition of the computed shear and normal stresses on fault surfaces (virtual or explicit) does not capture the critical interaction between the structures and the local stress field. 4 Numerical Modelling of the Evolution of Fault Systems In order to numerically model a system of faults in a state of marginal stability, a number of approaches can be taken. It is tempting to make use of discontinuum codes, in which faults are modelled explicitly. However, there are problems related to the evolution of such systems, in particular the method of establishing compatibility of fault orientation, rock mass strength, and stress field orientation. The approach selected was to use a continuum representation starting from an initially intact material in which faults evolved as strain localizations in response to far-field (boundary) deformation. This ensured that faults would be correctly oriented relative to the evolving stress field, and that the stress field varied locally in response to changes in fault strength as a result of constitutive behavior such as slip weakening. The two dimensional finite difference code FLAC (Itasca Consulting Group Inc., 2002) was selected to carry out the stress analysis as it is well suited to modelling fractures and faulting in rock (Cundall, 1989; Hobbs and Ord, 1989; Cundall, 1990; McKinnon and Garrido de la Barra, 1998). To enable faults to form in an initially elastic material, a strain softening constitutive model was used in conjunction with a large strain finite difference formulation for computing gridpoint displacements. A strain softening constitutive model was not essential, but sharpened the shear bands. The memory of where faults occurred in the material was provided by permanent deformation of the grid due to the large strain formulation. The two-dimensional nature of the code resulted in the formation of linear strain localizations, corresponding to strike-slip faults, which are the focus of the investigation. 4.1 Model Construction A common problem with grid-based continuum models is the effect of grid characteristics on the formation of shear bands. Attempts to minimize these effects were made by various model construction strategies: l Circular boundary models were used in order to minimize formation of shear bands at model corners due to stress concentrations. l Relatively large numbers of zones were used in the models to facilitate the formation of shear bands, which typically span two or three zones. Models consisted of approximately 7000 zones. l The initially square gridpoint geometry was randomized to eliminate regular channels of constant size zones. This reduced the tendency of shear bands to follow the alignment of grid axes. l The Young’s modulus of each zone was varied randomly about a mean value. This effect represented the heterogeneous nature of material properties. Randomizing locations of displacement-induced stress concentration reduced grid triggering of shear band formation. l An annulus of softer material approximately 5 zones wide surrounded the central model material. This material represented a resilient compromise between rigid displacement controlled and soft stress controlled boundary conditions, and was thought to be a geologically more realistic alternative to conventional boundary conditions. A simplified example of the resulting model grid is shown in Fig. 4. Similar numerical experiments were also carried out with substantially larger models than those described here. In those models, the density of shear bands was higher and fault growth (generally initiated in the interior of the models) could be tracked in more detail. However, the behaviour of these larger models was essentially the same as those described, leading to the same overall conclusions. Fig.4.Example of gridpoint randomization used to reduce grid effects on shear band formation 4.2 Material Properties, Initial and Boundary Conditions Elastic properties for the models were chosen to represent a typical hard rock found in Ontario mines. These properties are listed in Table 1. It was found, however, that the main conclusions drawn from the models remained the same for other values of hard rock parameters. Table 1. Material properties Property Value Uniaxial compressive strength σc（intact） 150MPa RMR 69 Young’s modulus Erm 30 GPa Poisson’s ratio ν 0.2 Cohesion c 4.3MPa Friction angle φ 55° Tensile strength σt 0.5MPa The Young’s modulus of each zone was varied randomly about a mean value of 30GPa with specified variation chosen from a triangular distribution with a maximum deviation of ±20% of the mean value. The tensile strength was set to eliminate tensile failure, since the objective of the analysis was to promote the formation of shear bands representing strike-slip faults. The initial state of stress was set at the yield point of the material with a high mean stress of 25.0MPa, major and minor principal stresses of 47.9MPa and 2.1MPa, respectively (compressive stresses positive). This avoided unnecessary cycling of the model from an unstressed elastic state, and avoided tensile failure. Note that these yield stresses are based on the rock mass properties shown in Table 1. The rock mass strength implied by the stress measurements shown in Fig. 3 applies to a larger scale and is not related to the engineering scale rock mass strength. However, the general conclusions from the results of these models do not depend on these details. The prescription of appropriate boundary conditions was complex. In the crust, deformation is generally some combination of pure and simple shear. The proportion of each component of deformation is significant, as this determines the orientation of the principal stresses in the elastic regime. Using the definition adopted by McKinnon and Garrido de la Barra (1998), shown in Fig. 6, the orientation of the major principal stress for a particular combination of pure and simple shear is: where Φ is measured clockwise from the plane of simple shear displacement to the direction of the major principal stress (Fig. 5). Using an equal contribution of both pure and simple shear, i.e. α= 45°, the initial orientation of the major principal stress was computed asΦ= 76.7°. Boundary gridpoint velocity vectors reproducing the initial state of stress and deformation are shown in Fig. 6. Gridpoint velocities were continuously adjusted such that they did not exceed a value of 1×10-5 m/step. This velocity corresponds approximately to the maximum fraction of the gridpoint spacing that can occur in one time step. By experimentation, this velocity was found to result in a small and controllable unbalanced forces at gridpoints. Fig.5.Definition used for boundary displacement when combining pure and simple shear, and the resultingorientation of principal stresses (from McKinnon and Garrido de la Barra (1998)) Fig.6.Boundary displacement vectors used to reproduce equal proportions of pure and simple sheardeformation 5 Modelling Results Four aspects of the modelled systems were of interest: fault evolution, characteristics of seismicity, triggering, and characteristics of the stress field in between faults. 5.1 Fault Formation The fault pattern that developed in the modelled system, shown in Fig. 7(a), was similar in orientation to classical Riedel conjugate shears, i.e. close to the theoretical Mohr