外文翻译-长臂开采工作面回采巷道顶板支护的设计方法

英文原文A method for the design of longwall gateroad roof supportW.Lawrence Geowork Engineering,Emerald,QLD,AustraliaAbstract：A longwall gateroad roof support design method for roadway development and panel extraction is demonstrated. It is a hybrid numerical and empirical method called gateroad roof support model（GRSM）, where specification of roof support comes from charts or equations. GRSM defines suggested roof support densities by linking a rock-mass classification with an index of mining-induced stress, using a large empirical database of Bowen Basin mining experience. Inherent in the development of GRSM is a rock-mass classification scheme applicable to coal measure strata. Coal mine roof rating（CMRR）is an established and robust coal industry standard, while the geological strength index（GSI）may also be used to determine rock-mass geomechanical properties.An elastic three-dimensional numerical model was established to calculate an index of mining induced stress, for both roadway development and longwall retreat. Equations to calculate stress index derived from the numerical modelling have been developed. An industry standard method of quantifying roof support is adopted as a base template（GRSUP）.The statistical analyses indicated that an improved quantification of installed support can be gained by simple modifications to the standard formulation of GRSUP. The position of the mathematically determined stable/failed boundary in the design charts can be changed depending on design criteria and specified risk.Keywords: Coal mine；Roof control；Support Design1. IntroductionLongwall gateroad strata stability is essential to ensure uninterrupted production. In Central Queenslands Bowen Basin, immediate gateroad roof lithology varies from coal to weak interlaminated material, to strong almost massive sandstone, with localised areas of weak fault affected strata. It is usual for roof conditions within any one mine to vary significantly. Typically, longwall mines in the Bowen Basin have specified gateroad roof support based on past practice. Modifications to gateroad support are generally reactive, due to encountered difficult strata conditions, and less proactive. Current gateroad support design approaches have limitations, which have restricted their applicability and adoption as mine site design tools.A prototype for an improved gateroad support design methodology has been developed that is integrated and systematic, based on rock engineering principals, but requires engineering judgement and experience 1. There were several broad objectives for the design methodology. A consistent and unambiguous definition of strata conditions and behaviour was required. Gateroad roof support needed to be assessed and specified. The method had to provide design calculations and justification for compliance and statutory purposes, and could serve as a frame work for a mine strata management system. Mine site support designers must be able to readily use the method to manage uncertainty and risk. The method must be able to be reviewed, modified and expanded.2. Current roof support design methods for longwall gateroadsNumerous roof support design methods have been proposed over the years, but none have gained widespread acceptance by the coal mining industry 2. There are empirical databases, some proprietary, based on industry practice, which specify gateroad primary and secondary support densities, using a statistical approach 3,4. Analytical methods are not appropriate when rock-mass yield due to high mining induced stresses occurs, but may be applicable and adapted to low stress environments 5. The application of complex post-yield numerical modelling in the design process for excavation support is valid although contentious, and requires a more comprehensive justification and better industry understanding of its strength and limitations 6. The complete mathematical representation of rock-mass properties and behaviour is a complex issue, which is still outside the capability of current numerical modelling code 6.Engineers and mathematicians do not have the current capability to fully define rock-mass geomechanical properties and their mathematical representation. Elasticplastic numerical modelling is a useful tool if used appropriately. It is not exclusively correct or unique, or always superior to other available and accepted design techniques. These aspects have been recognised during recent collaborative Australian Coal Association Research Program research on longwall microseismics 7, where it was considered that current 3D numerical models lack sufficient validated constitutive relationships, and are forced to make compromises when dealing with complex rock-mass behaviour.Simplified elastic numerical methods 8,9 have merit and are certainly applicable for more massive sedimentary rock-masses 5. An assessment of their applicability to weaker, laminated clastic rock-masses is required. Hybrid numerical and empirical methods have been developed for the geotechnical design of undercut and production level drifts of block caving mines 10.3. Geotechnical roof classification of longwall gateroadsTwo classification schemes were considered appropriate. Firstly, the coal mine roof rating （CMRR） 11, which is an established coal industry standard. Secondly, the Geological Strength Index, GSI with strength parameters included 12. A recent publication 13 has contended that GSI estimates of rock-mass strength should not be used for coal mine roof problems, where the geometrical scale of the problem is similar to discontinuity spacing. A distinction needs to be made between the GSI classification and the related HoekBrown failure criterion. This scale effect and situations where the failure criterion should not be used have been discussed 14 However, this does not mean that a classification of the rock-mass cannot be made. Indeed, this scale issue is a problem inherent in any rock- mass classification scheme, not just GSI, and for any failure criterion. For example, some mines appropriately use unconfined compressive strength （UCS） as an index or failure criterion, but UCS is also scale dependent and has the same limitations.Within the support design methodology, the rock-mass classification schemes will link mining-induced stresses （or stress index） and required installed roof support. Therefore, the classifications should be independent of environmental and geometrical factors, such as mining induced stresses and excavation orientation and size. A rock-mass classification scheme must also provide rock-mass geomechanical properties to enable the calculation of mining induced stresses.It is anticipated that CMRR will be the principal classification scheme used. However, the single rock-mass classification scheme that is best suited is the GSI derived global rock-mass strength. For numerical or analytical models, HoekBrown failure criterion parameters, modulus of deformation and rock-mass strength can be estimated from GSI 15,16.Direct utilisation of either CMRR or GSI is included within the design methodology.4. An index of mining induced stressAn index of mining induced stress in the gateroad roof at a location of interest is required. The three-dimensional （3D） stress distribution about a longwall panel including goaf reconsolidation, and the continuous stress redistribution that occurs during panel retreat, is a complex and difficult phenomena to quantify. One approach would be to construct a full elasticplastic, 3D numerical model. This approach would have limitations to a verified, unique and readily achieved calculation of stress, for several reasons. Generalised model roadway and goaf geometry may not always match the actual geometry. Generalised model roof lithology may not always match the actual lithology and variations. The roof/seam/floor interaction is a complex system and is difficult to model accurately. Rock-mass geomechanical properties, in particular post-yield cannot be fully defined. The geomechanical properties of the goaf, extent and behaviour of strata fracturing and caving, and goaf stress reconsolidation are largely unknown. The model may take many days to complete just a single scenario.While calculated mining induced stress from a detailed elasticplastic, 3D numerical model may be an appropriate parameter, there is little justification to improved accuracy compared to other methods. An alternative approach is to calculate mining induced stress from elastic 3D numerical models. Calculated mining induced stress in the immediate gateroad roof just outbye of the face-line may not be accurate if rock-mass yield occurs, but as an index of stress, it may be appropriate. An important criterion of its suitability would be how reasonable its relative variation is with changes in input parameters. A significant advantage is that it could be readily calculated for variable scenarios and would be within the range of capability of more geotechnical engineers.Maximum elastic tangential stress in the roof of a modelled gateroad could be considered a better indicator of rock-mass failure than the residual post-yield stress. Undoubtedly, significant rock-mass failure and subsequent stress redistribution do occur, which are not reflected in an elastic model. In the immediate roof of the gateroad, these failures are initiated at a critical mining induced stress. The stress index is a reasonable and appropriate measure of this critical stress, even if it may not agree in absolute magnitude after stress redistribution occurs. For mining induced stresses from an elastic 3D numerical model to be a reasonable representation, several issues influencing the stress distribution must be considered, which include strata fracturing and caving and goaf reconsolidation.For bulking-controlled caving, empirical relationships are used to predict the height of caving （goaf） and fracturing 17: （m） （1） （m） （2）where Hc is the caving（goaf） height above top fextracted horizon, Hf is the thickness of the fractured zone above top of caving zone, h is extraction thickness,and c1, c2, c3, c4, c5 and c6 are coefficients depending on lithology （Table1）.Table 1 Coefficients for average height of caving zone 17LithologyCompressive strength/MPaCoefficients /mC1C2C3C4C5C6Strong and hard402.1162.51.22.08.9Medium strong20-404.7192.21.63.65.6Soft and weak206.2321.53.15.04.0Weathered-7.0631.25.08.03.0Goaf stressstrain behaviour can be been defined 18 （Eq. （3）, based on earlier work 19, as follows: （MPa） （3） where, and are the vertical goaf strain and stress,respectively, E0 is the initial tangent modulus,and m is the maximum possible strain of the bulked goaf material.The initial bulking factor, BF, defines m as follows: （4） The initial tangent modulus, E0, can be defined as a function of the compressive strength of rock pieces, c, and the bulking factor, BF18,20: （MPa） （5）The FLAC3D double-yield constitutive model is used to simulate a strain-stiffening material with irreversible compaction,i.e.volumetric yield,in addition to shear and tensile failure.Upper-bound tangential bulk and shear moduli are specified21, with the incremental tangent and shear moduli evolving as plastic volumetric strain takes place.In addition to the shear and tensile strength criteria,a volumetric yield surface or cap has to be defined. The cap surface,defined by the cap pressure, pc, is related to the plastic volume strain, pv. The cap pressure, pc, is not the goaf vertical stress, v. The relationship between cap pressure and plastic volume strain is derived from an iterative FLAC2D compression test model,using a one element,1m1m,grid. Loading was simulated by applying a velocity to the top of the element, which has confined sides and base.The constitutive equation was derived from the iterative results by a Microsoft Excel Solver regression analysis,assuming a linear function.Goaf deformation and material strength parameters are defined as follows（Table2）. Table 2 FLAC3D goaf reconsolidation parameters1Upper bound tangent modulus230 MPa2Poissons ratio0.303Density1.7 gm/cc4Cohesion0.001 MPa5Friction angle256Dilation27Tensile strength0 MPa8Table 3 FLAC3D numerical model geometrical, geomechanical and geotechnical para meters1ParameterRangeUnitage2Roadway height2-3.4m3Roadway width4.8-6.5m4Longwall panel width200-300m5Pillar width15-45m6Depth60-330m7Immediate roof USC8-62MPa8Ratio of in situ horizontal to vertical stressRange from 1.2 to 2-9Rock-mass stiffness-10Rock-mass poissonratio0.25 for stone ,0.3 for coal-There are many theories on goaf reconsolidation, based on sound principles. Results from the various formulations do vary significantly. Which, if any, are correct is unknown, as goaf stresses have not been measured 18,22. For no other reasons than it is well described, and includes more of the parameters perceived to be important, the goaf stressstrain behaviour as defined is utilised in the calculation of a stress index 18. The elastic FLAC3D numerical model simulates a single two-heading longwall. Roof and floor strata are composite, uniform continuum. Strong contact is assumed between the coal seam and roof and floor. No discontinuities were modelled. Pillars will always be stable, which means that the actual pillar design must be appropriate and pillars adequately sized for the strata conditions .A range of geometrical,geological and geotechnical parameters must be specified,with the database distribution of some parameters listed in Table3. Some rock-mass geomechanical properties may be derived from thegeological strength index 15. Model 3D geometry may be visualised in Figs.1 and Figs.2 In these figures, scale may be judged from the seam thickness（3m）and thickness of immediate roof and floor. Axes of geometric symmetry are used, e.g. only half of the total goaf width is shown. Fig. 1. Typical 3D model geometryentire modelFig. 2. Typical 3D model geometryhorizontal section taken from the top of seamStress measurements are typically taken in discrete, more competent, and stiffer strata. Defining in situ horizontal stresses in all strata units of different stiffness is a difficult issue. There are problems associated with stress measurements in less competent strata and coal. The approach taken was simply to define in situ horizontal stresses in terms of rock-mass competency or classification. This does assume a correlation between the rock-mass parameter and elastic modulus. Quite rightly there are limitations with this approach. Model output （stress index） is simply the elastic mining-induced major principal stress in the immediate maingate roof just outbye of the longwall face-line . Example of FLAC 3D model output shows a longwall retreat situation, where the plane shown is horizontal, in the immediate roof. Note that the orientation of the major principal stress may be near-horizontal （immediate roof of gateroad） or near-vertical （pillar or face）. In this particular case, the stress index used in GRSM is 31 MPa. In the context of this design methodology and the development of a stress index, it is not critical that mining induced stress magnitudes agree. It is important is that relative changes in magnitude are reasonable and occur appropriately as parameters change. The effect of the intermediate （or minor） in situ horizontal principal stress must also be considered when assessing this model. This stress component will superimpose its own mining induced stress. 5. Characterisation of installed roof support A standard measure of the intensity of installed support, widely used within the Industry is GRSUP （ground support rating）, given by 4 （kN/m） （6） where Lb is the thickness of the bolted horizon defined by roof- bolts （m）, Nb is the average number of roof-bolts in each bolt row, Cb is the ultimate tensile strength of roof-bolts （kN）, Sb is the spacing between roof-bolt rows （m）,Nt is the average number of cables in each cable row, Ct is the ultimate tensile strength of cables （kN）, St is the spacing between cable rows（m）, w is the roadway width （m）, and 14.6 is a constant that is needed to convert from the original NIOSH equation, which was in Imperial units, to SI units; this will allow for compatibility with all USA data using the standard NIOSH equation.6. Database Data points have been collated from underground mines throughout the Bowen Basin coalfields. Nonstable, or failed data points are not restricted to situations where roof falls have occurred. Any situation where installed support was not sufficient was classified as a nonstable data point. Such situations include roof falls, where supplementary support was required for strata stabilisation, and where the area was mapped or observed to have experienced excessive deformation and deterioration. There are currently 280 defined data points, of which 106 are non-stable, and cover a range of strata conditions. Depth of cover ranges from 60 to 330 m. Roof conditions vary from weak interbedded to strong sandstone, also thick coal. There is high magnitude in situ horizontal stress in stone roof and relatively low magnitude stress in coal roof. Bolting densities range from a minimal four-bolt primary support in strong roof conditions, to intensive bolt plus cable support in weak and fault affected roof conditions. Longwall gateroad development contributes 166 data points （60 fa