A particle mechanics approach for the dynamic strength model of the jointed rock mass considering the joint orientation

In nature, there exist several forms of anisotropy in rock masses due to the presence of bedding planes, joints, and weak layers. It is well understood that the anisotropic properties of jointed rock masses significantly affect the stability of surface and underground excavations. However, these critical anisotropic characteristics are often ignored in existing uniaxial dynamic failure criteria. This study investigates the effect of a pre-existing persistent joint on the rate-dependent mechanical behaviours of a rock mass using a particle mechanics approach, namely, bonded particle model (BPM), to realistically replicate the mechanical response of the rock mass. Firstly, in order to capture the rate-dependent response of the jointed rock mass, the BPM model is validated using published experimental data. Then, a dynamic strength model is proposed based on the Jaeger criterion and simulation results. To further investigate the dynamic behaviours, the dynamic uniaxial compressive strength (UCS) for anisotropic rock masses with various joint orientations is investigated by subjecting the BPM models to uniaxial compression numerical tests with various strain rate. The proposed dynamic strength model is validated based on numerical simulation results. Finally, the fragmentation characteristics of the jointed rock masses are analysed, which demonstrate that the failure mode affects the dynamic UCS. This is further confirmed by the analysis of the orientations of microscopic cracks generated by the compression loading.     

Error-controlled implicit time integration of elasto-visco-plastic constitutive models for rock salt

The mechanical behavior of rock salt is rate-dependent at different time scales. Using caverns in rock salt formations for renewable energy storage implies that the underground structures are subjected to both short-term and long-term loads, increasing robustness and flexibility requirements for numerical simulators used to assess the safety of such structures. So far, explicit time integration with model-specific heuristics for time-step size determination dominate in application studies. In this paper, the suitability of error-controlled adaptive time-integration schemes of the diagonally implicit Runge-Kutta type is investigated in comparison with the Backward Euler and Crank-Nicolson schemes when applied to the integration of typical elasto-visco-plastic constitutive models of rock salt. The comparison is made both for monotonic and for cyclic loads as well as taking account of thermo-mechanical coupling. Analyses of the time-integration errors and the time step-size evolution show the suitability of the integration scheme for these material models. The automatic adjustment of the time-step size was found to be robust across all material models and boundary conditions as well as for non-isothermal situations for a single algorithmic parameter set.     

Obtaining Constitutive Relationship for Rate-Dependent Rock in SHPB Tests

A large number of tests have recently been conducted with the Split Hopkinson pressure bar (SHPB) method to determine the characteristics of rock dynamics. However, it is still impossible to get test results at a perfect constant strain rate from this set-up owing to the rate dependency of rock materials. For instance in most cases, dynamic behavior of rock can only be described with an average strain rate. The results from these methods, including rich strain rate information, frequently tend to be inexplicable or self-contradictory. The obtained stress–strain curves can then never be directly treated as constitutive curves as in static tests. In this paper, the reasons behind the controversial stress–strain results with current methods are analyzed. In addition, the requirement for the rock specimen to deform at a constant strain rate is demonstrated after theoretical analysis of correlations among specimen, deforming stress, incident stress, reflected stress and transmitted stress. With test results from SHPB by pulse shaper and special shape striker methods, the requirement is verified. Finally, the method of 3D scattergram considering stress–strain–strain rate simultaneously is brought up to get constitutive relationships of rate-dependent rock. The new method gives reasonable predictions for constitutive relationships of rock at different strain rates. At the same time, the new method has fewer requirements and has a wider application scope for SHPB tests.     

Computational simulation of time-dependent behavior of soil–structure interaction by using a novel creep model: Application to a geosynthetic-reinforced soil physical model

In this paper, a model geosynthetic-reinforced soil retaining walls (GRS-RW) is tested by vertically loading it through a rough footing on the top near the retaining wall and the results are simulated by a sophisticated nonlinear Finite Element Method (FEM) having a novel rate dependent constitutive model for both the backfill material and the geosynthetic reinforcement. Usually, polymer geosynthetic reinforcement is known to exhibit more-or-less rate-dependent stress–strain or load–strain behavior due to their viscous properties. The geomaterials (i.e., clay, sand, gravel and soft rock) also exhibit viscous properties. The viscous behavior of geometrials are quite different from that of the polymer based geosynthetic-reinforcements. It has been revealed recently that viscous behavior of sand is a kind of temporary effect, which vanishes with time. So the rate-dependent deformation of backfill reinforced with polymer geosynthetic reinforcement becomes highly complicated due to interactions between the elasto-viscoplastic properties of backfill and reinforcement. In the present study, a scaled model geosynthetic-reinforced soil retaining wall is tested with a vertically loaded rough rigid footing. The results of the model test are simulated by using an appropriate elasto-viscoplastic constitutive model of both sand and geogrid embedded in a nonlinear plane strain FEM.     

Modeling of wave propagation induced by underground explosion

A piecewise linear Drucker–Prager strength criterion and an isotropic continuum damage model with the damage scalar depending on an equivalent tensile strain are suggested to model rock mass behavior under blast loading. A rate-dependent constitutive relation is employed to model the energy dissipation caused by two sources, namely irreversible degradation of damage and permanent deformation caused by plasticity. The suggested model is incorporated with a commercially available software AUTODYN through its user’s subroutine function. Coupling of Euler and Lagrange processors are used to include all the materials under consideration such as explosive, air and rock mass, in the calculation. Using AUTODYN and the suggested model, shock wave propagation in rock mass induced by an underground explosion is simulated. Numerical results obtained agree favorably well with those obtained from an independently conducted field test. It demonstrates that the suggested model can be used to predict the damage area, plastic zone and ground motions generated by underground explosions.     

Manuel Rocha Medal Recipient Simulating the Time-dependent Behaviour of Excavations in Hard Rock

Summary Although hard rock is not usually associated with large creep deformation, data collected from the tunnels and stopes of the deep South African gold mines illustrates significant time-dependent behaviour. Apart from application in mining, a better understanding of the time-dependent behaviour of crystalline rock is required to analyse the long term stability of nuclear waste repositories and to design better support for deep civil engineering tunnels in these rock types. To illustrate the subtle problems associated with using viscoelastic theory to simulate the time-dependent behaviour of hard rock, a viscoelastic convergence solution for the incremental enlargement of a tabular excavation is discussed. Data on the time-dependent deformation of a tunnel developed in hard rock further illustrates the limitations of the theory, as it is unable to simulate the fracture zone around these excavations. To simulate the rheology of the fracture zone, a continuum viscoplastic approach was developed and implemented in a finite difference code. This proved more successful in modelling the time-dependent closure of stopes and squeezing conditions in hard rock tunnels. A continuum approach, however, has limitations in areas where the squeezing behaviour is dominated by the time-dependent behaviour of prominent discontinuities such as bedding planes. To overcome this problem, a viscoplastic displacement discontinuity technique was developed. This, combined with a tessellation approach, leads to more realistic modelling of the time-dependent behaviour of the fracture zone around excavations. Received January 15, 2002; accepted June 3, 2002 Published online September 2, 2002     

Some recent advances in the modelling of soft rock joints in direct shear

This paper presents a review of recent developments made by the authors into the modelling of rock joints in direct shear. Careful observation of laboratory direct shear testing on concrete/rock joints containing two-dimensional roughness has allowed theoretical models of behaviour to be developed. The processes modelled include asperity sliding, asperity shearing, post-peak behaviour, asperity deformation and distribution of stresses on the joint interface. Model predictions compare extremely well with laboratory test results. These models were then applied to direct shear tests on rock/rock joints, and although behaviour in general was well predicted, the strength of rock/rock joints was over-predicted. Direct shear tests have also been carried out on samples containing both two- and three-dimensional roughness to test the accuracy of the two-dimensional approximation to roughness adopted in the theoretical models.     

Robust numerical implementation of a 3D rate-dependent model for reservoir geomechanical simulations

A 3D elasto-plastic rate-dependent model for rock mechanics is formulated and implemented into a Finite Element (FE) numerical code. The model is based on the approach proposed by Vermeer and Neher (A soft soil model that accounts for creep. In: Proceedings of the International Symposium “Beyond 2000 in Computational Geotechnics,” pages 249-261, 1999). An original strain-driven algorithm with an Inexact Newton iterative scheme is used to compute the state variables for a given strain increment.The model is validated against laboratory measurements, checked on a simplified test case, and used to simulate land subsidence due to groundwater and hydrocarbon production. The numerical results prove computationally effective and robust, thus allowing for the use of the model on real complex geological settings.     

On viscoplastic regularisation of strain-softening rocks and soils

Constitutive models for rocks and soils typically incorporate some form of strain softening. Moreover, many plasticity models for frictional materials use a nonassociated flow rule. Strain softening and nonassociated flow rules can cause loss of well-posedness of the initial-value problem, which can lead to a severe mesh dependence in simulations and poor convergence of the iterative solution procedure. The inclusion of viscosity, which is a common property of materials, seems a natural way to restore well-posedness, but the mathematical properties of a rate-dependent model, and therefore the effectiveness with respect to the removal of mesh dependence, can depend strongly on how the viscous element is incorporated. Herein, we show that rate-dependent models, which are commonly applied to problems in the Earth's lithosphere, such as plate tectonics, are very different from the approach typically adopted for more shallow geotechnical engineering problems. We analyse the properties of these models under dynamic loadings, using dispersion analyses and one-dimensional finite difference analyses, and complement them with two-dimensional simulations of a typical strain localisation problem under quasi-static loading conditions. Finally, we point out that a combined model, which features two viscous elements, may be the best way forward for modelling time-dependent failure processes in the deeper layers of the Earth, since it not only enables modelling of the creep characteristics typical of long-term behaviour but also regularises the initial/boundary-value problem.     

岩体结构面力学行为的尺寸效应研究

尺寸效应是岩体结构面力学行为的重要特征。本文列举了结构面力学行为中普遍存在的尺寸效应现象,并由实测统计资料的分析,论证了结构面力学行为的尺寸效应具有分形结构。通过结构面力学行为尺寸效应的机理研究,建立了结构面力学行为尺寸效应分维数和结构面粗糙度系数尺寸效应分维数之间的相关关系,从而简化了结构面力学行为尺寸效应规律的表述,为客观评价岩体结构面力学参数提供最为有效的手段。     

Micro- and macro-behaviour of fluid flow through rock fractures: an experimental study

Microscopic and macroscopic behaviour of fluid flow through rough-walled rock fractures was experimentally investigated. Advanced microfluidic technology was introduced to examine the microscopic viscous and inertial effects of water flow through rock fractures in the vicinity of voids under different flow velocities, while the macroscopic behaviour of fracture flow was investigated by carrying out triaxial flow tests through fractured sandstone under confining stresses ranging from 0.5 to 3.0 MPa. The flow tests show that the microscopic inertial forces increase with the flow velocity with significant effects on the local flow pattern near the voids. With the increase in flow velocity, the deviation of the flow trajectories is reduced but small eddies appear inside the cavities. The results of the macroscopic flow tests show that the linear Darcy flow occurs for mated rock fractures due to small aperture, while a nonlinear deviation of the flow occurs at relatively high Reynolds numbers in non-mated rock fracture (Re?>?32). The microscopic experiments suggest that the pressure loss consumed by the eddies inside cavities could contribute to the nonlinear fluid flow behaviour through rock joints. It is found that such nonlinear flow behaviour is best matched with the quadratic-termed Forchheimer equation.     

Geomechanical Properties of Shear Zones in the Eastern Aar Massif,Switzerland and their Implication on Tunnelling

Summary ¶Rock zones containing a high fracture density and/or soft, low cohesion materials can be highly problematic when encountered during tunnel excavation. For example in the eastern Aar massif of central Switzerland, experiences during the construction of the Gotthard highway tunnel showed that heavily fractured areas within shear zones were responsible for overbreaks in the form of chimneys several metres in height. To understand and estimate the impact of the shear zones on rock mass behaviour, knowledge concerning the rock mass strength and deformation characteristics is fundamental. A series of laboratory triaxial tests, performed on samples from granite- and gneiss-hosted shear zones revealed that with increasing degree of tectonic overprint, sample strength decreases and rock behaviour shows a transition from brittle to ductile deformation. These trends may be explained by increasing fracture densities, increasing foliation intensity, increasing thickness of fine-grained, low cohesion fracture infill, and increasing mica content associated with the increasing degree of tectonic overprint. As fracture density increases and the influence of discrete, persistent discontinuities on rock mass strength decreases, behaviour of the test samples becomes more and more representative of rock mass behaviour, i.e. that of a densely fractured continuum. For the purpose of numerical modeling calculations, the shear zones may be subdivided with respect to an increasing fracture density, foliation intensity and mica content into a strongly foliated zone, a fractured zone and a cohesionless zone, which in turn exhibit brittle, brittle-ductile and ductile rock mass constitutive behaviour, respectively.Received December 17, 2001; accepted January 9, 2003 Published online April 29, 2003     

Design of tunnel linings in a creeping rock

Summary The effect of long term rock deformation on lining pressure is considered using different concepts of rock behaviour. These include the conventional characteristic-line or convergence-confinement method, modified to allow for rock ageing, and lining-rock interaction methods using models of linear-viscoelastic, linear-elastic linear viscous, and linear elastic non-linear viscous rock behaviour. Calculations of lining pressures show that the former tends to underestimate, compared with the latter.This paper was presented at the 28th US Symposium on Rock Mechanics in Tucson, Arizona, July 1987.     

Capturing the complete stress–strain behaviour of jointed rock using a numerical approach

This paper presents the results of a series of numerical experiments using the synthetic rock mass (SRM) approach to quantify the behaviour of jointed rock masses. Field data from a massive sulphide rock mass, at the Brunswick mine, were used to develop a discrete fracture network (DFN). The constructed DFN model was subsequently subjected to random sampling whereby 40 cubic samples, of height to width ratio of two, and of varying widths (0.05 to 10 m) were isolated. The discrete fracture samples were linked to 3D bonded particle models to generate representative SRM models for each sample size. This approach simulated the jointed rock mass as an assembly of fractures embedded into the rock matrix. The SRM samples were submitted to uniaxial loading, and the complete stress–strain behaviour of each specimen was recorded. This approach provided a way to determine the complex constitutive behaviour of large‐scale rock mass samples. This is often difficult or not possible to achieve in the laboratory. The numerical experiments suggested that higher post‐peak modulus values were obtained for smaller samples and lower values for larger sample sizes. Furthermore, the observed deviation of the recorded post‐peak modulus values decreased with sample size. The ratio of residual strength of rock mass samples per uniaxial compressive strength intact increases moderately with sample size. Consequently, for the investigated massive sulphide rock mass, the pre‐peak and post‐peak representative elemental volume size was found to be the same (7 × 7 × 14 m). Copyright © 2015 John Wiley & Sons, Ltd.     

Numerical model for rock bolts with consideration of rock joint movements

Summary The stability of any underground structure during and after excavation is the most important question for designers, because any kind of collapse may destroy large parts of a finished tunnel, causing major repairs and time loss. Preliminary calculations are therefore of great importance. A calculation is only useful, however, when the underlying numerical model correctly describes natural behaviour. The rock bolts used in tunnel excavations are mostly untensioned grouted bolts, and this type of bolt is the main focus of this work. From the model of the grouted bolt, other types of rock bolts can also be modelled by the theory presented herein. Bolt behaviour in intact rock mass is so different from behaviour when a bolt intersects a joint, that a model with two different elements is suggested for a numerical calculation; one element for the bolt in the rock mass and one as a bolt intersecting with a joint.The model for both elements is verified by the experimental results. The numerical results correspond favourably with the experimental work. A variation of the parameters important for the behaviour of the bolt in intersection with the joint is shown. As an implementation of the bolt model, the numerical simulation of excavation and stabilisation of one road tunnel is presented.     

Modelling of intact and jointed mudstone samples under uniaxial and triaxial compression

Rock mass is a highly complex entity where the strength and deformation behaviour can be significantly affected by its secondary structures such as joints, fissures and bedding planes. Whilst many research works have been conducted to study the behaviour of a specific rock mass, a thorough understanding of its strength and deformation behaviour incorporating different joint sets has not been established. In this study, a comprehensive numerical modelling using a three-dimensional distinct element code, 3DEC, was undertaken to study the strength and deformation behaviour of a mudstone, locally found in Melbourne, in unconfined and confined states. The initial unconfined model established for intact mudstone was calibrated against the well-established laboratory-based empirical strength relationships and subsequently compared with some strength test data available for field samples. The intact unconfined model was then extended to study the strength behaviour in the confined state. The results obtained from this confined intact model were compared with existing strength criteria and were found in good agreement. The confined model was extended further to investigate the effects of joint sets and dip angles on the rock mass strength and deformation behaviour by incorporating two different joint configurations (one-joint and two-joint) with varying dip angles (0°–90°). This study found that the rock mass strength in a confined state varied significantly between the two joint configurations.     

Analysis of the Dynamic Performance of an Underground Excavation in Jointed Rock under Repeated Seismic Loading

Results from field observations of dynamic behaviour of an underground excavation have been compared with numerical studies of the rock deformation history. The field behaviour shows progressive accumulation of rock displacement and excavation deformation under successive episodes of dynamic loading. It is possible to reproduce the modes of rock response quite well using a Distinct Element model of the rock mass, but the way displacements develop is dependent on the joint model used in the analysis. It is suggested that, in rock masses subject to repeated dynamic loading, excavation design may need to take account of the prospect of repeated episodes of transient loading at the excavation site.