Abstract: Pressure build-up caused by large-scale CO2 injection is one of the key concerns during a carbon sequestration project, for well-known reasons such as the risk to seal integrity, fault stability and induced microseismicity, among others. Furthermore, pressure build-up is directly related with storage capacity. In this work we study the geomechanical response to the CO2 injection in the Tubåen Fm at Snøhvit. During the first stage of the project CO2 was separated from the produced gas and stored underground in the Tubåen Fm. at approximately 2600 m depth. The Tubåen Fm. corresponds to a delta plain environment dominated by fluvial distributary channels and some marine-tidal influence. The area is extensively faulted, characterized by a dominant east-west-trending fault system, but with the presence of faults at high angles to this trend, leading to complex fault interactions. Injection into the Tubåen Formation was limited by reservoir heterogeneities and reservoir compartmentalization, leading to higher than expected pressure rise during operations. This lead to a well intervention operation designed to improve injectivity and a revised injection plan, with injection into a different unit. In the present work we perform a probabilistic assessment of the mechanical deformation caused by the CO2 injection and the potential for fault leakage and contamination of the producing interval in the adjacent block. For the majority of the cases, in the range of the evaluated parameters, we found that the increase in pressure due to CO2 injection does not pose risk for fault reactivation. However, observed variations in the orientation of the maximum horizontal stress have a high impact on the potential for reactivation of the studied faults.
Martín, L. Blanco (Lawrence Berkeley National Laboratory) | Rutqvist, J. (Lawrence Berkeley National Laboratory) | Birkholzer, J.T. (Lawrence Berkeley National Laboratory) | Houseworth, J.E. (Lawrence Berkeley National Laboratory)
Abstract: The coupled thermal-hydraulic-mechanical response of a hypothetical high-level nuclear waste repository in a saliniferous rock formation is explored in this study. Although only short-term results are presented in this paper, the ultimate aim of the ongoing research is to evaluate the mechanical and hydraulic barrier integrity of the repository over the long-term. The design adopted involves metallic waste packages horizontally disposed in drifts subsequently backfilled with crushed salt. Some of the THM interactions studied concern porosity and permeability changes within the backfill as consolidation occurs, and also permeability changes within the natural rock salt due to thermomechanically and hydraulically-induced damage processes. The simulations are conducted using an updated version of TOUGH2-FLAC3D, developed at Lawrence Berkeley National Laboratory. The mechanical behavior of the host rock is modeled using the Lux/Wolters constitutive model, developed at Clausthal University of Technology. Well-adapted relationships are used to couple non-isothermal multiphase flow and geomechanics. In order to ensure the grids consistency, the TOUGH2 mesh is updated as large strains take place. The comparison between TH and THM results shows the relevance of the latter to better simulate the long-term response of structures envisaged for nuclear waste disposal in saliniferous formations.
The results from two laboratory hydraulic fracture experiments are presented in this paper. Experiments were conducted in a tri-axial geophysical imaging cell equipped with 18 broadband acoustic emission transducers. Samples were loaded to a differential stress of 115 MPa and then distilled water (1 cP) was injected at a constant flow rate through a central cased borehole until failure. Two flow rates, 0. 2 mL/min and 0.5 mL/min were investigated and their failure sequences were analyzed in detail. A flow rate of 0.2 mL/min caused gradual failure with AE source locations aligning on two conjugate planes and AE activity lasting for ~120 ms. Using an injection rate of 0.5 mL/min initiated a two stage fracture process. AE hypocenters initially located around the borehole and then rapidly moved to the edge of the sample along a vertically oriented plane. The final burst of AE activity resembled a tremor-like-signal and lasted for approximately 22 ms. Frequency analysis of continuous acoustic emission waveforms show that the maximum energy is initially held at high frequency and gradually shift to lower frequencies as the fracturing nears completion. X-ray computed tomography scans revealed complex fracture geometry.
This paper presents the numerical investigation results of failure process, failure modes and strength anisotropy of layered rock under a series of uniaxial compression and Brazilian tests. This paper employs 2-D Particle Flow Code (PFC2D) to simulate layered rock models for different inclination angles(θ) that varies from 0° (the layer is perpendicular to the loading direction) to 90° (the layer is parallel to the loading direction). Based on the numerical simulation results, the uniaxial compressive strength decreases with the increase of the inclination angle at 0° ≤θ ° then the uniaxial compressive strength slightly increases at 60° ≤θ °. Three kinds of failure modes under uniaxial compression test could be observed: (a) Sliding failure across layers mode, (b) Sliding failure along layer mode, and (c) Tensile split along layer mode. The inclination angle has a more significant effect on the tensile behavior of layered rock under Brazilian test. The tensile strength of layered rock decreases with the increase of the inclination angle. Through the numerical investigation of failure process during Brazilian test, the micro-cracks roughly initiate at 50% of ultimate stress near the contact boundary then propagate slowly until the peak is reached. After the ultimate stress is reached, the mico-cracks start to form macro-cracks and the failure of layered rock under Brazilian test can classified into four modes: (a) Split across layer mode: the cracks propagate across layers at 0° ≤θ °, (b) Sliding along layer mode: the cracks slide along the layer at 45° ≤θ °, (c) Mixed mode: both split across layer and sliding along layer mode could be observed at θ=75°, and (d) Split along layer mode: the cracks propagate along layers at θ=90°. Furthermore, the numerical simulation results compare with the experimental data from Cho et al. (2012). Although the failure modes under Brazilian test at 45° ≤θ ° are a little difference, the uniaxial compressive strength, failure modes under uniaxial compression test, and tensile strength agree well with the experimental data from Cho et al. (2012).
Abstract: Development of an inelastic zone ahead of a crack tip known as process zone is a common phenomenon observed in many quasi-brittle materials. Some experimental and numerical efforts have been conducted to scrutinize the parameters affecting the size of process zone. This study investigates the role of grain size on the process zone and size effect parameters by conducting a discrete element simulation of rock fracture. A softening contact bond model is used to study the development of the process zone around a notch tip in three-point bending tests. The numerical simulation is utilized to obtain nominal tensile strength, apparent fracture toughness and width of process zone. Bazant’s size effect law parameters were obtained using the change in nominal tensile strength with specimen size and it was found that by increasing the grain size, brittleness of the material decreases. It is also shown that apparent fracture toughness is in general a function of specimen and grain size and it increases with the increase in grain size .The change in process zone width with specimen size is investigated too. It is illustrated that for a less brittle material, the impact of grain size on the width of process zone is greater. Based on dimensional analysis, for sufficiently large specimens, a linear relationship between width of process zone and grain size is suggested.
: The paper deals with the geotechnical studies carried out to understand the possible reasons of 30m high coal rib failure and the adjoining dump failure. The slope design was also done to excavate the blocked coal in rib. The 30m high coal rib was under cut at its toe level due to fire. The unsupported bottom most toe portion of the standing 30m high coal rib was under the maximum stress but the overhanging coal rib slope, detached from its toe due to fire, was having least strength due to long time exposure and fire. The triggered failure caused horizontal shifting of coal rib up to the overhanging/ undercut burnt coal level along the bedding plane of the coal seam which was dipping towards the pit. The adjoining OB dump subsided behind the coal rib with formation of numerous cracks on the surface. The failed slope was designed to mine out the blocked coal in the rib. The remaining dump slope should be reformed with bench height and width of 5m and 6m respectively and 45° bench angle. The ultimate slope of southern coal rib should be mined with a maximum bench height of 10m and minimum bench width of 8m.
Abstract: World-wide mining operations utilize block caving as one of the most cost-effective techniques for ore extraction. Block caving has been addressed in the past by numerous discontinua methods to include Discrete Element Method (DEM), Discontinuous Deformation Analysis (DDA), Combined Finite-Discrete Element Method (FDEM), etc. However, most of these analyses were either limited to 2D or to elastic material representation. In this paper a representative 3D block caving problem is simulated using FDEM. Los Alamos National Laboratory’s Geophysics team conducted the modeling utilizing their in-house FDEM code, MUNROU. MUNROU is a fully parallel, 2D/3D FDEM code which utilizes material models that account for a number of plasticity effects, as well as has the capability to model explosive effects, irregular shapes and fracture initiation and propagation. Previously problems of this nature and size were un-tractable in 3D. However, the recent performance improvements seen in MUNROU through the implementation of state of the art parallelization algorithms (see Computational Mechanics of Discontinua, Wiley 2011), have prompted our team to begin intensive efforts to address real world problems, such as block caving. The code’s inherent capability to address fracture and fragmentation processes at laboratory scale level has been consistently proven in the past. In this paper the feasibility of extending MUNROU to large space scales in 3D is demonstrated. With this improved capability it is now expected that future analyses efforts can concentrate on 3D phenomenological considerations such as jointing, frictional fault behavior, etc.
Abstract: Thousands of shale wells are being drilled and fracked each year. These extremely tight formations are impossible to produce without fracking, which in turn requires knowledge of mechanical properties of shale. The chemical and mechanical instability of shales limits the recovery of full length cores and plugs necessary for conventional mechanical testing. Recovering cores is expensive. Nanoindentation provides a means to obtain Young’s modulus and hardness of rocks from drill cuttings, fragments and sidewall cores which are millimeters in size. The measurements obtained are reliable and agree well with other standard measurements. Additionally, measurement over the whole pay zone through drill cuttings can help in improving hydraulic fracturing design. This paper will focus on the applicability of nanoindentation to shales. It will also present a correlation to estimate rock mechanical properties based on the measurements of principal rock components affecting mechanical behavior: mineralogy, porosity and organic content. The precise measurement capability of the nanoindenter also enables the study of rock frame and organic content separately. Nanoindentation measurement provides a measure of anisotropy when bedding directions are known.
Abstract: Determination of water saturation is one of the most important items in reservoir characterization and calculation of hydrocarbon volume in place. The most common approach in estimation of water saturation is based on petrophysical models. The existing models have many operational restrictions which may lead to error in estimation of water saturation. The current paper is made an attempt to develop the Support Vector Machine (SVM) method for estimating reservoir water saturation (Sw) from Petrophysical Logs (PLs) related to four wells of an Iranian carbonate reservoir. In order to show the superiority of the proposed method, Sw was predicted by a Back-Propagation Artificial Neural Network (BP-ANN) model as well as three conventional equations (i.e. Archie, Hossin and Dual-water). The results were compared based on determination coefficient (R2), Root Mean Square Error (RMSE), and Mean Absolut Error (MAE) indexes. The measured R2 of predicted Sw in the test data, using SVM was 0.88, whereas it was 0.81 by both ANN model and Dual-water equation as best fitted model. The RMSE and MAE were 0.02 and 0.82 by SVM, whereas they were 0.12 and 1.04 by ANN and 0.04 and 0.96 by Dual-water.
Abstract: Geomaterials that are assumed to have symmetry about a single preferred direction have five independent transversely isotropic elastic constants. These elastic constants can be determined from data obtained through a series of macroscale calibration experiments, but only a subset of these five constants can be found directly from axial and lateral stressstrain measurements on a single cylindrical sample of material. Substructural axisymmetric inhomogeneities present in the material and decoupling methods used in modeling can imply constraints on transversely isotropic elastic constants, potentially reducing the number of macroscale experiments needed to characterize a geomaterial model. Morphology of substructural heterogeneities, such as distributions of microscale inclusions, cracks, pores and fibers, lead to homogenization or distribution parameters that affect the fourth-order elastic stiffness of the material. Constitutive models that decouple the elastic stiffness often neglect interaction components, which impose constraints on the transversely isotropic elastic constants. We consider the mathematically motivated decoupling of tensorially linear and non-linear functions of a structural or fabric tensor. Neglecting the non-linear components, as often done for rock models, imposes a constraint that the lateral shear modulus depends on the remaining elastic constants. We also consider the mathematically motivated decoupling of purely-volumetric and purely-deviatoric components, often used in the field of biomechanics. When the mixed volumetric-deviatoric components are neglected, the axial and lateral Poisson’s ratios are constrained to become dependent on the two tensile moduli and a new independent bulk modulus type parameter. The described constraints reduce the number of independent elastic constants from five to four. In biomechanics, the accuracy of the approximation from the constraints can be verified through knowledge that the substructural source of anisotropy is fibers embedded in a mostly incompressible water matrix. The potential for using similar techniques to investigate approximations that use constraints on elastic constants is discussed for geomaterials, specifically non-interacting cracked solids, with the goal of reducing the number of macroscale experiments needed to characterize a geomaterial model.