Neural networks (NNs) are widely used to investigate the relationship among variables in complex multivariate problems. In cases of limited data, the network behavior strongly depends on factors such as the choice of network activation function and network initial weights. In this study, I investigated the use of neural networks for multivariate analysis in the case of limited data. The analysis shows that special attention should be paid when building and using NNs in cases of limited data. The linear activation function at the output nodes outperforms the sigmoidal and Gaussian functions. I found that combining network predictions gives less biased predictions and allows for the assessment of the prediction variability. The NN results, along with conventional statistical analysis, were used to examine the effects of folding, bed thickness, structural position, and lithology on the fracture properties distributions in the Lisburne Formation, folded and exposed in the northeastern Brooks Range of Alaska. Fracture data from five folds, representing different degrees of folding, were analyzed. In addition, I modeled the fracture system using the discrete fracture network approach and investigated the effects of fracture properties on the flow conductance of the system. For the Lisburne data, two major fracture sets striking north/south and east/west were studied. Results of the NNs analysis suggest that fracture spacing in both sets is similar and weakly affected by folding and that stratigraphic position and lithology have a strong effect on fracture spacing. Folding, however, has a significant effect on fracture length. In open folds, fracture lengths in both sets have similar averages and variances. As the folds tighten, both the east/west and north/south fracture lengths increase by a factor of 2 or 3 and become more variable. In tight folds, fracture length in the north/south direction is significantly larger than in the east/west direction. The difference in length between the two fracture sets creates a strong anisotropy in the reservoir. Given the same fracture density in both sets, the set with the greater length plays an important role for fluid flow, not only for flow along its principal direction but also in the orthogonal direction.

No description available.

Today everyone seems to agree that ultra-low permeability and shale reservoirs have become the potentials to transform North America's oil and gas industry to a new phase. Unfortunately, transient flow is of long duration (perhaps life of the well) in ultra-low permeability reservoirs, and traditional decline curve analysis (DCA) models can lead to significantly over-optimistic production forecasts without additional safeguards. Stretched Exponential decline model (SEDM) gives considerably more stabilized production forecast than traditional DCA models and in this work it is shown that it produces unchanging EUR forecasts after only two-three years of production data are available in selected reservoirs, notably the Barnett Shale. For an individual well, the SEDM model parameters, can be determined by the method of least squares in various ways, but the inherent nonlinear character of the least squares problem cannot be bypassed. To assure a unique solution to the parameter estimation problem, this work suggests a physics-based regularization approach, based on critical velocity concept. Applied to selected Barnett Shale gas wells, the suggested method leads to reliable and consistent EURs. To further understand the interaction of the different fracture properties on reservoir response and production decline curve behavior, a series of Discrete Fracture Network (DFN) simulations were performed. Results show that at least a 3-layer model is required to reproduce the decline behavior as captured in the published SEDM parameters for Barnett Shale. Further, DFN modeling implies a large number of parameters like fracture density and fracture length are in such a way that their effect can be compensated by the other one. The results of DFN modeling of several Barnett Shale horizontal wells, with numerous fracture stages, showed a very good agreement with the estimated SEDM model for the same wells. Estimation of P90 reserves that meet SEC criteria is required by law for all companies that raise capital in the United States. Estimation of P50 and P10 reserves that meet SPE/WPC/AAPG/SPEE Petroleum Resources Management System (PRMS) criteria is important for internal resource inventories for most companies. In this work a systematic methodology was developed to quantify the range of uncertainty in production forecast using SEDM. This methodology can be used as a probabilistic tool to quantify P90, P50, and P10 reserves and hence might provide one possible way to satisfy the various legal and technical-society-suggested criteria.

Numerical modeling of fluid flow and dissolved species transport in the subsurface is a challenging task, given variability and measurement uncertainty in the physical properties of the rock, the complexities of multi-fluid interaction, and limited computational resources. Nonetheless, this thesis seeks to expand our modeling capabilities in the context of contaminant hydrogeology. We describe the numerical simulator CompFlow Bio and use it to model invasion of a nonaqueous phase liquid (NAPL) contaminant through the vadose zone and below the water table in a fractured porous rock. CompFlow Bio is a three-phase, multicomponent, deterministic numerical model for fluid flow and dissolved species transport; it includes capillary pressure and equilibrium partitioning relationships. We have augmented the model to include randomly generated, axis-aligned, discrete fracture networks (DFNs). The DFN is coupled with the porous medium (PM) to form a single continuum. The domain is discretized using a finite-volume scheme in an unstructured mesh of rectilinear control volumes (CVs). Herein we present the governing equations, unstructured mesh creation scheme, algebraic development of fracture intersection CV elimination, and coupling of PM CVs over a fracture plane to permit asperity contact bridged flow. We include: small scale two-phase water-air and NAPL-water simulations to validate the practice of intersection CV elimination; small scale simulations with water-air, NAPL-water, and NAPL-water-air systems in a grid refinement exercise and to demonstrate the effect of asperity contact bridged flow; intermediate scale 3D simulations of NAPL invading the saturated zone, based on the Smithville, Ontario, site; intermediate scale 2D and 3D simulations of NAPL invading the vadose zone and saturated zone with transient recharge, based on the Santa Susana Field Laboratory site, California. Our findings indicate that: the formulation provides a practical and satisfactory way of modeling three-phase flow in discretely fractured porous rock; numerical error caused by spatial discretization manifests itself as several biases in physical flow processes; that asperity contact is important in establishing target water saturation conditions in the vadose zone; and simulation results are sensitive to relative permeability-saturation-capillary pressure relationships. We suggest a number of enhancements to CompFlow Bio to overcome certain computational limitations.

Prediction of groundwater inflow into rock tunnels is one of the essential tasks of tunnel engineering. Currently, most of the methods used in the industry are typically based on continuum models, whether analytical, semi-empirical, or numerical. As a consequence, a regular flow along the tunnel is commonly predicted. There are also some discrete fracture network methods based on a discontinous model, which typically yield regular flow or random flow along the tunnel. However, it was observed that, in hard rock tunnels, flow usually concentrates in some areas, and much of the tunnel is dry. The reason is that, in hard rock, most of the water flows in rock fractures and fractures typically occur in a clustered pattern rather than in a regular or random pattern. A new method is developed in this work, which can model the fracture clustering and reproduce the flow concentration. After elaborate literature review, a new algorithm is developed to simulate fractures with clustering properties by using geostatistics. Then, a discrete fracture network is built and simplified. In order to solve the flow problem in the discrete fracture network, an existing analytical-numercial method is improved. Two case studies illustrate the procedure of fracture simulation. Several ideal tunnel cases and one real tunnel project are used to validate the flow analysis. It is found that fracture clustering can be modeled and flow concentration can be reproduced by using the proposed technique. / text

The transport and fate of contaminants in fractured sedimentary rock aquifers depends strongly on the nature and distribution of the fracture network. The current standard practice of using only vertical coreholes to characterize bedrock aquifers can result in significantly biased data that is insufficient for fracture orientation analysis. This study involves the addition of two inclined coreholes to supplement existing data from eleven vertical coreholes at a contaminated site in Guelph, Ontario to reduce the effects of this bias. A suite of high-resolution, depth discrete data collection methods including core logging, borehole geophysics, and hydraulic testing were conducted to determine fracture orientation and spacing as well as hydraulic aperture distributions. The results of the orientation analysis demonstrate that the inclined coreholes were more effective at sampling high-angled fractures than the vertical coreholes and were necessary to identify all three of the dominant fracture sets on the site. The fracture network properties from this study can be used as input parameters for static and dynamic discrete fracture network models to assess current and future risks to municipal supply wells.

Numerical modeling of fluid flow and dissolved species transport in the subsurface is a challenging task, given variability and measurement uncertainty in the physical properties of the rock, the complexities of multi-fluid interaction, and limited computational resources. Nonetheless, this thesis seeks to expand our modeling capabilities in the context of contaminant hydrogeology. We describe the numerical simulator CompFlow Bio and use it to model invasion of a nonaqueous phase liquid (NAPL) contaminant through the vadose zone and below the water table in a fractured porous rock. CompFlow Bio is a three-phase, multicomponent, deterministic numerical model for fluid flow and dissolved species transport; it includes capillary pressure and equilibrium partitioning relationships. We have augmented the model to include randomly generated, axis-aligned, discrete fracture networks (DFNs). The DFN is coupled with the porous medium (PM) to form a single continuum. The domain is discretized using a finite-volume scheme in an unstructured mesh of rectilinear control volumes (CVs). Herein we present the governing equations, unstructured mesh creation scheme, algebraic development of fracture intersection CV elimination, and coupling of PM CVs over a fracture plane to permit asperity contact bridged flow. We include: small scale two-phase water-air and NAPL-water simulations to validate the practice of intersection CV elimination; small scale simulations with water-air, NAPL-water, and NAPL-water-air systems in a grid refinement exercise and to demonstrate the effect of asperity contact bridged flow; intermediate scale 3D simulations of NAPL invading the saturated zone, based on the Smithville, Ontario, site; intermediate scale 2D and 3D simulations of NAPL invading the vadose zone and saturated zone with transient recharge, based on the Santa Susana Field Laboratory site, California. Our findings indicate that: the formulation provides a practical and satisfactory way of modeling three-phase flow in discretely fractured porous rock; numerical error caused by spatial discretization manifests itself as several biases in physical flow processes; that asperity contact is important in establishing target water saturation conditions in the vadose zone; and simulation results are sensitive to relative permeability-saturation-capillary pressure relationships. We suggest a number of enhancements to CompFlow Bio to overcome certain computational limitations.

According to the Mine Health and Safety Administration (MSHA), between 2006 and 2016, the underground stone mining industry had the highest fatality rate in 4 out of 10 years, compared to any other type of mining in the United States. Additionally, the National Institute for Occupational Safety and Health (NIOSH) stated that structurally controlled instability is a predominant failure mechanism in underground limestone mines. This type of instability occurs when the different discontinuity sets intercept with each other forming rock blocks that displace inwards the tunnel as the excavation takes place, posing a great hazard for miners and overall mine planning. In recent years, Terrestrial laser scanning (TLS) has been used for mapping and characterizing fractures present in a rock mass. TLS is a technology that allows to generate a three-dimensional multimillion point cloud of a scanned area. In addition to this, the advances in computing power throughout the past years, have allowed numerical modeling codes to represent more realistically the behavior of a fractured rock masses. This work presents and implements a methodology that integrates laser scanning technology along with Discrete Element Modeling as tools for characterizing, preventing, and managing structurally controlled instability that may affect large-opening underground mines. The stability of an underground limestone mine that extracts a dipping ore body with a room and pillar (and eventual stoping) mining method is analyzed with this approach. While this methodology is proposed based on a specific case study that does not meet the requirements to be designed with current NIOSH published guidelines, this process proposes a general methodology that can be applied in any mine experiencing similar failure mechanisms, considering site-specific conditions. The aim of this study is to ensure the safety of mine workers and to reduce accidents that arise from ground control issues. The results obtained from this methodology allowed us to generate Probability Density Functions to estimate the probability of rock fall in the excavations. These models were also validated by comparing the numerical model results with those obtained from the laser scans. / M.S. / According to the Mine Health and Safety Administration (MSHA), between 2006 and 2016, the underground stone mining industry had the highest fatality rate in 4 out of 10 years, compared to any other type of mining in the United States. Additionally, the National Institute for Occupational Safety and Health (NIOSH) stated that structurally controlled instability is one of the main causes of rock falls in underground limestone mines. This type of instability occurs when the fractures present in the rock mass intercept each other forming rock blocks that displace into the tunnel as the excavation takes place and poses a great hazard for miners. In recent years, Terrestrial laser scanning (TLS) has been used for mapping and characterizing fractures present in a rock mass. TLS is a technology that allows to generate a three-dimensional multimillion point cloud of a scanned area. In addition to this, the advances in computing power throughout the past years, have allowed simulation softwares such as the Discrete Element Model (DEM) to represent more realistically the behavior of a fractured rock mass under excavation. The aim of this work was to develop and evaluate a methodology that could complement already exisiting design guidelines that may not apply to all kind of underground mines. The presented methodology evaluates rock failure due to presence of discontinuites, through the integration of TLS with DEM and considers site specific conditions. An area of a case study mine was assessed with this methodology, where several laser scans were performed. Information extracted from this laser scans was used to simulate the response of the rock mass under excavation by running Discrete Element Numerical Models. Results from these models allowed us to estimate the probability of rock failure in the analized areas. These, rock block failure probability estimations provide engineers a tool for characterizing, preventing, and managing structurally controlled instability, and ultimately improving workers safety.

[pt] Em uma mina a céu aberto, a estabilidade dos taludes rochosos é um dos maiores desafios na engenharia das rochas devido aos processos geodinâmicos que formaram o depósito de minério, fazendo de cada depósito complexo e único. Algumas das complexidades encontradas comumente são: a geologia nos arredores do depósito, a alta variabilidade das propriedades, os complexos defeitos estruturais, o grau de alteração das rochas, a informação geomecânica limitada, etc. Antes de avaliar a estabilidade de taludes devemos caracterizar o maciço rochoso. Para caracterizá-lo se têm construído os modelos geológico, estrutural e do maciço rochoso para formar o modelo geotécnico como recomenda o projeto Large Open Pit (LOP), um projeto de pesquisa internacional relacionado à estabilidade de taludes de rocha nas minas a céu aberto. Uma vez construídos os domínios geotécnicos, a estabilidade de taludes rochosos pode ser avaliada para cada domínio pelos métodos de equilíbrio limite ou numéricos como o método dos elementos finitos ou o método dos elementos discretos. O uso do método depende de diversos fatores, como a influência dos elementos estruturais, a importância da análise, a informação disponível, etc. Os métodos de equilíbrio limite como os tradicionais de Bishop e Janbu podem ser usados na avaliação de estabilidade de grandes taludes de rocha que são susceptíveis a falhas rotacionais do maciço rochoso. Já o método de elementos finitos se tem desenvolvido rapidamente e tem ganhado popularidade para a análise de estabilidade de taludes no caso em que o mecanismo de falha não esteja controlado por estruturas discretas geológicas. Os métodos de elementos finitos estão baseados em modelos constitutivos de tensão – deformação para rochas intactas e têm dificuldades em simular famílias com um número grande de descontinuidades dentro do maciço rochoso. O método dos elementos discretos permite simular um número grande de descontinuidades assim como também permite a simulação de grandes deformações. A presente dissertação usa o modelo SRM (Synthetic Rock Mass) para avaliar a estabilidade de taludes de uma mina a céu aberto no Peru. O SRM é uma nova técnica para simular o comportamento mecânico de maciços rochosos fraturados e permite simular a propagação de fraturas e os efeitos da anisotropia. Está técnica usa o modelo BPM (Bonded Particle Model) para representar a rocha intacta e o SJM (Smooth - Joint Contact Model) para representar as estruturas do maciço rochoso dentro do programa PFC. Para a modelagem estrutural se utilizou o método DFN (Discrete Fracture Network). Para a determinação dos modelos geológicos e estrutural se utilizou o programa Petrel e para a análise de estabilidade de taludes usando o modelo SRM se utilizou o programa PFC 4.0 na versão 2D. / [en] In an open pit mine, stability of rock slope is one of the most challenges in rock mechanics due to geodynamic processes that formed the ore deposit, making each deposit complex and unique. Some of the complexities commonly encountered are: the geology in the vicinity of the deposit, the high variability of properties, the complex structural defects, the rock alteration degree, limited geomechanical data, etc. Before evaluating the slope stability we should characterize the rock mass. To characterize it we have built the geological model, structural model and rock mass model to form the geotechnical model as it recommends the Large Open Pit project (LOP), an international research project related to stability of rock slope in open pit mines. Once constructed geotechnical domains, the stability of rock mass slope can be evaluated for each domain by using some known methods like limit equilibrium, the finite elements and discrete element methods. The use of the method depends of different factors like influence of structural elements (defects), importance of analysis, available information, etc. Limit equilibrium traditional methods like Bishop and Janbu can be used to evaluate the stability of large rock slopes that are susceptible to rotational failure of rock mass. Since the finite element method has developed rapidly and has gained popularity for the slope stability analysis in the case where failure mechanism is not controlled by discrete geological structure. Finite element method is based on constitutive models of stress-strain for intact rocks and has difficulties in simulating sets with a large number of discontinuities within the rock mass. The discrete element method allows to simulate a large number of discontinuities and also allows the simulation of large deformations. This dissertation uses the SRM (Synthetic Rock Mass) model to evaluate the stability of slopes in an open pit mine in Peru. The SRM model is a new technique that allows the simulation of the mechanical behavior of fractured rock mass taking into account propagation of fractures and anisotropic effects. This technique uses two well established techniques like BPM (Bonded Particle Model) for representation of intact rock and the SJM (Smooth-Joint Contact Model) to represent the structural fabric within the PFC program. For structural modeling it was used DFN method (Discrete-Fracture Network). To determine the geological and structural model it was used the Petrel program (Version 2010.1) and for slope stability analysis with the SRM model it was used the version 2D of the PFC 4.0 program.

Jointed rock masses are formed of intact rock and joints. There-fore, proper characterization of rock mass behavior has to consid-er the combined behavior of the intact rock blocks and that of the joints. This thesis presents the theoretical background of the Synthetic Rock Mass (SRM) modeling technique along with example applica-tions. The SRM technique is a new approach for simulating the mechanical behavior of jointed rock masses. The technique uses the Bonded Particle Model (BPM) for rock to represent intact ma-terial and the Smooth-Joint Contact Model (SJM) to represent the in situ joint network. In this manner, the macroscopic behaviour of an SRM sample depends on both the creation of new fractures through intact material, and slip/opening of pre-existing joints. SRM samples containing thousands of non-persistent joints can be submitted to standard laboratory tests (UCS, triaxial loading, and direct tension tests) or tested under a non-trivial stress path repre-sentative of the stresses induced during the engineering activity under study. Output from the SRM methodology includes pre-peak properties (modulus, damage threshold, peak strength) and post-peak proper-ties (brittleness, dilation angle, residual strength, fragmentation). Of particular interest is the ability to obtain predictions of rock mass scale effects, anisotropy and brittleness; properties that can-not be obtained using empirical methods of property estimation. Additionally, the nature of yielding and fracturing can be studied as the rock mass fails. This information can improve our understand-ing of rock mass failure mechanisms. / QC20100720