|Theme||Visible||Selectable||Appearance||Zoom Range (now: 0)|
Choi, Seungbeom (Korea Institute of Geoscience and Mineral Resources) | Lee, Su-deuk (Seoul National University) | Jeong, Hoyoung (Seoul National University) | Jeon, Seokwon (Seoul National University)
Rock mass contains various discontinuities in terms of size and shape, such as fault, joint, and bedding plane. Among them, a joint is a planar discontinuity that has little strength and it shows smaller size but more frequency than a fault. In general, the joint exerts huge influence on mechanical behavior of rock mass since it acts as a weak plane. At the same time, the joint has several orders higher hydraulic conductivity that rock matrix so that the majority of fluid flow in rock mass occurs through the joint. Therefore, accurate understanding of characteristics of a rock joint is of great importance, as it affects both mechanical and hydraulic behavior of rock mass. Not only the mechanical conditions but also the geometric features affect the joint behavior. The geometric features can be explained by various properties, such as roughness, aperture, contact area, and so on. They exert complex influence on the hydromechanical characteristics of a joint, interacting with each other. Therefore, a series of laboratory experiments were conducted in order to investigate the hydraulic characteristics under various mechanical and geometric conditions. A fractal theory was used to generate coordinates of artificial rock joint surfaces so as to control the roughness of the joint. Then the coordinates were printed by a 3D printer and utilized to make cement mortar specimens. Four joint specimens, which had different levels of roughness and aperture, were prepared and tested. Before investigating the hydraulic characteristics, mechanical behavior of joint specimen was tested first. In order to consider various mechanical conditions, normal and shear stresses were applied and hydraulic tests were conducted with the same mechanical and geometric conditions. The results showed that flowrate per unit hydraulic head decreased with the increase of normal stress, increased with the shear displacement, and increased with the roughness of the joint. Also, comparison between hydraulic aperture, which was calculated based on the cubic law, and corresponding mechanical aperture was made and it showed more deviation from the cubic law when the roughness and stress were increased.
Lee, Kyungbook (Korea Institute of Geoscience and Mineral Resources) | Lim, Jungtek (SmartMind) | Yoon, Daeung (Hanyang University and Chonnam National University) | Jung, Hyungsik (Seoul National University)
Decline-curve analysis (DCA) is an easy and fast empirical regression method for predicting future well production. However, applying DCA to shale-gas wells is limited by long transient flow, a unique completion design, and high-density drilling. Recently, a long short-term-memory (LSTM) algorithm has been widely applied to the prediction of time-series data. Because shale-gas-production data are time-series data, the LSTM algorithm can be applied to predict future shale-gas production. After information for 332 shale-gas wells in Alberta, Canada, is obtained from a commercial database, the data are preprocessed in seven steps, including cutoffs for well list, data cleaning, feature extraction, train and test sets split, normalization, and sorting for input into the LSTM model. The LSTM model is trained in 405 seconds by two features of production data and a shut-in (SI) period from 300 wells. The two-feature case shows a better prediction accuracy than both the one-feature case (i.e., production data only) and the hyperbolic DCA, where the three methods are tested on unseen data from 15 wells. The two-feature case can predict future production rates according to the SI period and provide a stable result for available time-series data.
ABSTRACT: In this study, we present the results of numerical simulations of washout behavior of granular geomaterials. The washout of soil grains, which is driven by a fluid flow, is considered to be one of the most significant reasons of ground subsidence particularly in urban areas accompanies by underground excavations since the excavation may disturb and cause a groundwater gradient. For this, a computational fluid dynamics (CFD) module was implemented into particle flow code (PFC) so as to include a seepage force in the particle displacement calculation of the conventional PFC. Through this coupling, both particle dislocation created by fluid flow and porosity/permeability change induced by the particles redistribution are considered. To verify the validity of the current coupled PFC-CFD analysis, we compared our numerical estimation of washout rate to that of the previous laboratory experiments. They showed well comparable results for two comparative cases of stable and unstable conditions. We then performed a parameter sensitivity analysis to identify the influential parameters on internal stability against a washout behavior and showed it is the effect of particle size distribution, the area where fluid flows out and infiltration flow rate that affect a washout rate.
Catastrophic ground subsidence, particularly at urban areas has been frequently reported these days. According to the report in Seoul, Korea, the subsidence is mainly involved with a damage of underground water pipes and reaches to 85% of its total occurrence (Seoul Metropolitan city, 2015). Any damages of subsurface water pipes including fracturing, break-up and dislocation may induce a groundwater flow and result in washout of soil grains and potential subsequent subsidence.
Mechanism of ground subsidence by groundwater flow can be explained as follows. First, cavity is created due to washout of grains around damaged water pipes. The cavity grows gradually along with groundwater flow by rainfall and at nearby underground excavation works where hydraulic gradient is induced. Finally, the cavity reaches to a ground surface then sudden collapse of the surface occurs.
Lee, Joo Yong (Korea Institute of Geoscience and Mineral Resources) | Lee, Jong-sub (Korea University) | Cho, Gye-Chun (Korea Advanced Institute of Science and Technology) | Kwon, Tae-Hyuk (Korea Advanced Institute of Science and Technology)
Gas hydrates are widespread, occurring in both permafrost and deep sea sediments. The large estimated areas of gas hydrate reservoirs suggest that the high potential of gas hydrates as an energy resource if economically viable production methods were developed. The production of natural gas from gas hydrate deposits poses challenges such as assessing hydrate recovery rates from physical properties and geological structure of the hydrate reservoir, securing the economic viability of produced gas from a particular resource, and keeping process safe from geomechanical impacts from hydrate dissociation. During the hydrate dissociation and the subsequent gas production from dissociated gas hydrate, geomechanical property changes due to the sediment deformation, the changes in hydrate saturations, and fine migrations. In this study, extensive laboratory studies have been conducted to quantify these issues and the implications of these changes to the gas production from gas hydrate deposits have been investigated.
Strength, stiffness, permeability changes due to gas hydrate saturations were examined in high-pressure oedometric system and tri-axial system. Fine migrations characteristics and the subsequent property changes were examined with many different experimental systems. The experimental system includes core-flooding system with X-ray CT monitoring, oedometric system, triaxial system, and one-dimensional fine migration experiment system. The sediment used in this study is synthesized gas hydrate-bearing sediments and the mean grain size of the sediments lies in fine sands. Hydrate saturation ranges from 10 to 50%. Fine fraction ranges also from 10 to 50%.
Sediment deformation from compressive stress concentration generally increases stiffness and decreases permeability. Hydrate saturation decrease induced from gas hydrate production generally decrease strength and stiffness and increase permeability. The property changes are not linearly related to gas hydrate saturations and the relations differ depending on the character of deposits. Fine migrations induced by gas hydrate production alter fine contents in producing intervals and also would change geomechanical properties. Moving particles generally concentrates near well-bore but the locus of concentration depends on the character of the producing interval, such as grain size distributions and flow rate. Even a small fraction of fine particles can induce significant changes in physical properties. In fine-concentrated zones, stiffness generally increases and permeability generally decreases.
The quantifications of these phenomena based on the systematic and extensive experimental studies are the essential steps before the development of THM numerical simulation code for gas hydrate production. For near future the quantitative relations in this study will be implemented to THM simulation code for gas hydrate production.
Sohn, Yongjune (Department of Civil and Environmental Engineering, Korea Advanced Institute of Science and Technology) | Lee, Joo Yong (Korea Institute of Geoscience and Mineral Resources) | Song, Ki-Il (Department of Civil Engineering, Inha University) | Kwon, Tae-Hyuk (Department of Civil and Environmental Engineering, Korea Advanced Institute of Science and Technology)
This study investigates how the variation in sediment layer geometry of hydrate-bearing sediments (HBS) affects geomechanical behaviors of HBS under depressurization. Two reservoir models with different layering structures but with the same hydrate quantity were constructed and the reservoir responses were numerically investigated during gradual depressurization process. To simulate thermo-hydro-mechanically coupled multiphysics processes occurring in HBS, a series of governing equations were discretized based on a finite volume concept, and coded into an explicit finite difference numerical simulator. An explicitly coupled, time-marching algorithm was used to couple thermo-hydro-mechanical responses associated with depressurization-driven hydrate dissociation. We herein modelded a hydrate deposit in Ulleung Basin, Korea for the sediment properties and geological setting. The simulation results clearly demonstrate that the "densely" layered HBS structure, composed of thin and interbedded clay-sand layers, is more prone to geomechanical instability though it led to more gas production. It is attributed to various mechanisms, including (i) the rapid water drainage from neighboring thin clay layers, (ii) the unique hydrate dissociation pattern in interbedded HBS, and (iii) the transfer of shear stress from hydrate-bearing, "stiff" sandy layers into adjacent thin "soft" clay layers. The layer geometry substantially affects not only the gas production but also the geomechanical stability of a hydrate reservoir. High-resolution sediment profiling appears to play an important role in numerical HBS simulations to reliably predict the feasibility of safe exploitation from layered HBS systems.
Jang, Youngho (Korea Institute of Geoscience and Mineral Resources) | Yoo, Hyunsang (Chonnam National University) | Sung, Wonmo (Hanyang University) | Lee, Jeonghwan (Chonnam National University) | Lee, Won Suk (Korea Institute of Geoscience and Mineral Resources)
Hydraulic fracturing in naturally fractured tight formations causes complex fracture geometry corresponding to the existence of natural fracture as well as geomechanical characteristics. It is extremely important when stimulating fluids such as water, brine, and CO2 are injected into tight carbonate reservoirs for increasing recovery. For fracture propagation modeling, several models have implemented multiple planar fracture with only opening mode, which have a problem in representing realistic fracture behavior.
In this regard, we proposed a hydraulic fracture propagation model implementing multiple planar approach with mixed mode including opening and sliding modes for being able to describe fracture propagation more realistically. When hydraulic fracture encounters natural fracture non-orthogonally, fracture tip slides first along the natural fracture face, and then propagates into rock mass. That is why sliding mode also needs to be considered. With the use of the model, this study analyzed the effects of the brittleness on fracture propagating behavior and gas recovery for tight formations with respect to Poisson's ratio and Young's modulus.
We investigated the modeling results for examining the importance of the sliding mode newly employed in the hydraulic fracture model. In the case of a formation having a high brittleness index, which presents greater deformation longitudinally rather than transversely, called the formation A, the effect of sliding mode was not critical on the fracture propagation. Meanwhile, in a formation having an intermediate brittleness index, named the formation B, the hydraulic fracture less easily crosses natural fracture because Young's modulus of this formation is lower comparing to the formation A, and consequently, implementation of the sliding mode is more dominant. In a formation representing higher Poisson's ratio and Young's modulus compared to formation A, which denotes a low brittleness index, since it shows larger deformation in transverse direction, hydraulic fracture hardly crosses the natural fracture, and thereafter, the propagating direction of the crossed fracture is highly deviated. The larger brittleness index, the greater in stimulated reservoir volumes by up to 17% because fracture crossing occurred easily.
Therefore, the model with the mixed mode proposed in this study was found to be extremely important in the analysis of fracture propagation behavior resulting the stimulated reservoir volumes and gas recovery differently in tight carbonate reservoirs.
A quick, simple and quantitative method for the estimation of surface subsidence susceptibility in mined areas with a lack of detailed geological and geometrical information in underground is presented in this paper. In the method only gangway depth from the surface and the attitude (dip and dip direction) of main geological features are used as input data based on the degree of availability and reliability. Underground gangways are represented as a series of points instead of closed polygons for easy calculation. The core assumption in this method is that the susceptibility to subsidence within a unit area increases both as the depth of the gangway from the surface decreases and as the number of gangways below the unit area increases. In spite of the simplicity of the proposed method, it gave satisfactory results when applied to a virtual excavation model and a closed coal mine where subsidence occurred actually.
Several methods for predicting ground subsidence due to mining excavation such as the profile method and the influence function method have been proposed (Whittaker and Reddish, 1989; Sheory, 2000). The National Coal Board (1975) has presented a basic technique to determine the surface area affected by coal mining based on the height and width of mined areas and the angle of inclination of coal seams. All these methods were developed and verified for conditions involving horizontal coal seams and long wall mining, which are the common mining conditions in Europe. However, coal-associated geological structures in Korea are very complicated, and coal seams have various widths and irregular dip angles. Consequently, the slant chute block caving method has been widely used in Korea, and sinkhole type subsidence is more common than trough type. As a result, the conventional prediction methods must be adapted to the Korean geology and mining conditions, or new subsidence estimation methods must be developed.
The goal of this study is to develop a simple, general, quantitative and reliable method for identifying subsidence susceptibility of the closed or abandoned coal mines, which is proper to be employed in geologically complicated areas. The proposed method in this paper considers only gangway depths and attitude of geological features like dip and dip direction is an optional parameter, because these data are relatively easy to acquire and generally reliable.
2. Estimation of subsidence susceptibility
2.1 Basic assumption
The depth of gangways is selected as an input data of this study after surveying the availability and effectiveness of data because it is reliable and can be easily acquired. In fact, several researchers revealed that the magnitude (volume) and depth of excavation are the principal factors influencing on the subsidence (Whittaker and Reddish, 1989; Singh and Dhar, 1997; MIRECO, 2008).
The method proposed in this study is based on the fact that the excavation volume and shape (or distribution of coal seams) are closely related to the gangway distribution. Two basic assumptions considered in the method are that the susceptibility to subsidence within a unit surface area increases as the depth of a gangway from the surface decreases and the number of gangways below the unit area increases. The first assumption is based on the bulking of failed rock mass which can fill the excavation and prohibit the propagation of roof failure. The second assumption comes from the fact that the rock mass around the excavation is damaged due to blasting and induced stresses.
The susceptibility related to the depth of a gangway is quantified using a negative exponential equation based on the results of numerical analyses (Park et al., 2005) and statistical data of subsidence occurrences in Korean coal mines as shown in Fig. 1 (MIRECO, 2008). Park et al. investigated the influence of the depth and width of excavation and of the spacing and dip of discontinuity on ground subsidence using PFC2D capable of modeling the bulking effect and showed that the overburden remains undamaged as the mining depth increases. Fig. 1 shows that most of subsidence occurred within a depth of 100 m from the surface. The number of subsidence events decreases exponentially as the gangway depth increases.
Cheon, Dae-Sung (Korea Institute of Geoscience and Mineral Resources) | Jin, Kwangmin (Korea Institute of Geoscience and Mineral Resources) | Jung, Yong-Bok (Korea Institute of Geoscience and Mineral Resources)
Microseismicity is an generated elastic wave when a crack is generated due to deformation or damage of a material, and it tends to increase sharply before macro-failure of the material. It can be used to monitor the safety of the rock mass structure such as mine and tunnel etc., and also used to determine the locations of cracks or macro-failures. In order to analyze the source location of cracks, it is important to consider the elastic wave propagation velocity, arrival picking, source location analysis algorithm, and sensor array. However, the location of the sensor may be restricted due to site conditions and economic problems, which may result in inability to interpret the source location or decrease reliability of MS monitoring. In this study, to improve the accuracy of source location analysis, we analyzed source locations according to various arrival picking method and source location algorithm. Among the methods, AIC and Generic algorithm for source location were found to be superior to other methods.
Microseismicity(or Microseismic event) can be defined as a very small earthquake caused by natural(wave, wind etc.) or artificial (hydraulic fracturing, blasting etc.) causes. Generally, it is a small size (< M_w 2.0}) and high frequency (> 50Hz) compared to earthquakes. Earthquakes are primarily caused by nature, but microseismic event is often caused by induced earthquakes. Figure 1 is a brief summary of the frequency domain and the audible domain for earthquakes, microseismic event, and acoustic emissions. Microseismicity is an generated elastic wave when a crack is generated due to deformation or damage of a material, and it tends to increase sharply before macro-failure of the material. Microseismic monitoring can be traced back to 1938 when the U.S. Bureau of Mines attempted to relate seismic wave velocity with pillar load. It is used for geotechnical safety monitoring based on the characteristics of the increase in the number of events before major failures. Thesedays in situ microseismic monitoring of the rock mass fracturing process has been widely used in rock mechanics tests and rock engineering projects throughout the world
Park, Jung-Wook (Korea Institute of Geoscience and Mineral Resources) | Kim, Taehyun (Korea Institute of Geoscience and Mineral Resources) | Park, Eui-Seob (Korea Institute of Geoscience and Mineral Resources)
As part of the DECOVALEX-2019 project Task B-Fault slip modeling, we are developing a numerical model for simulating fault activations induced by water injection. The work of Task B is scheduled to be conducted until 2019 in three research phases. The topic of the first step is developing a numerical method for a benchmark model to simulate the injection test in a single fault zone. We present a numerical model to reproduce the coupled hydro-mechanical process of fault activation using the TOUGH-FLAC simulator. The mechanical behavior of a single fault is represented by the zero-thickness interface element of FLAC3D upon which a slip and/or separation is allowed. The fluid flow along a fault is represented using finite thickness elements in TOUGH2 on the basis of Darcy’s law with the cubic law. The hydro-mechanical coupling between the fracture hydraulic transmissivity and the slip-induced displacement was established for two different fault models (FM1 and FM2). A coupling module was developed in the TOUGH-FLAC simulator to continuously update the changes in geometrical features, as well as hydrological properties induced by mechanical deformation. Then, the transient responses to stepwise pressurization of the fault and host rock were examined during the simulation. The hydro-mechanical behavior, including the injection flowrate, pressure distribution around the borehole, stress conditions, and displacements in normal and shear directions induced by water injection were monitored along the fault and/or surrounding rock. The results of benchmark calculations suggest that the developed model can reasonably represent the hydro-mechanical behavior of a fault and the surrounding rock, including the progressive evolutions of the pathway and fault slip zone. This study will be extended and enhanced through continuing collaboration and interaction with other research teams of Task B.
The DECOVALEX project, which began in 1992, is an international research and model comparison collaboration for thermo-hydro-mechanical-chemical processes in geological systems. Task B of the current DECOVALEX-2019 phase, running from 2016 to 2019, addresses the potential creation of permeable flow paths for contaminant transport in low-permeability host rocks. The objective of the task is to develop numerical models for coupled hydro-mechanical processes of fault activation. The work is planned to be conducted until 2019, through the following three steps of progressively increasing complexity: 1) The benchmark calculation of a simplified single fault plane, 2) the interpretive modeling of an observed activation in a minor fault, and 3) the interpretive modeling of an observed activation in a major fault. The model developed in the benchmark calculation will be modified and verified using the field data from fault activation experiments recently performed at the Mont Terri underground research laboratory in Switzerland.
Kim, Gvan Dek (Seoul National University) | Choi, Jaeho (Seoul National University) | Lee, Kyungbook (Korea Institute of Geoscience and Mineral Resources) | Shin, Hyundon (Inha University) | Choe, Jonggeun (Seoul National University)
In this paper, the selective use of measurement data using Ensemble Smoother is suggested in order to improve its performances by reducing the possibility of misuse of observed data. Key idea is that observed data are selected out on the basis of water breakthrough for better reservoir characterization. We use oil production rates before water breakthrough and water cut rates after water breakthrough for each well because ES cannot interpret the physical characteristic of water breakthrough properly. The consequence is that the proposed method gives us the best reservoir characterization of results with clear channel patterns and connectivity.
Reservoir characterization is one of the most important things for decision making in petroleum engineering. The way to make reliable and proper reservoir models is using static and dynamic data. Prior reservoir models made by using static data only have high geological uncertainties. In order to reduce these uncertainties, history matching is applied to integrate dynamic data, but the uncertainty range might be still high due to modelling error, limited data, or measurement error. Therefore, uncertainty quantification is vital for future performance. To improve performance estimation, many ensemble members are very often utilized to various reservoir characterization methods. The process is called ensemble-based reservoir characterization.
Evensen (1994) offered Ensemble Kalman filter to ocean dynamics. In case of reservoir engineering, EnKF was introduced by Naevdal et al. (2002). EnKF has many advantages such as uncertainty quantification in predicted productions, real-time updating of observed data, easy coupling with any forward simulator, and flexibility for types of model parameters and observed data.
However, EnKF has two critical limitations: overshooting and filter divergence (Aanonsen et al., 2009; Jeong et al., 2010; Oliver and Chen, 2011). These problems occur when model parameters do not follow Gaussian distribution or initial ensembles are not reliable and quite different from the true model. The importance of overcoming EnKF demerits was described by many researchers. However, most of the proposed methods still take high simulation time and have the restart option of a forward simulator due to recursive update of EnKF. Van Leeuwen and Evensen (1996) applied Ensemble Smoother (ES) for meteorology and compared EnKF with ES for history matching. Skjervheim et al. (2011) first proposed ES to reservoir characterization. They suggested that ES showed quite reliable results compare to EnKF. ES is very fast and simple because it assimilates all dynamic data at once, simultaneously. Also it is easier than EnKF for coupling with any reservoir simulator since it doesn't need restart option. However, it is still unstable and exposed to the possible overshooting and filter divergence.