• Energy Research

SummaryWeak bedding planes (BPs) that exist in many tight oil formations and shale–gas formations might strongly affect fracture–height growth during hydraulic–fracturing treatment. Few of the hydraulic–fracture–propagation models developed for unconventional reservoirs are capable of quantitatively estimating the fracture–height containment or predicting the fracture geometry under the influence of multiple BPs. In this paper, we introduce a coupled 3D hydraulic–fracture–propagation model considering the effects of BPs. In this model, a fully 3D displacement–discontinuity method (3D DDM) is used to model the rock deformation. The advantage of this approach is that it addresses both the mechanical interaction between hydraulic fractures and weak BPs in 3D space and the physical mechanism of slippage along weak BPs. Fluid flow governed by a finite–difference methodology considers the flow in both vertical fractures and opening BPs. An iterative algorithm is used to couple fluid flow and rock deformation. Comparison between the developed model and the Perkins–Kern–Nordgren (PKN) model showed good agreement.I–shaped fracture geometry and crossing–shaped fracture geometry were analyzed in this paper. From numerical investigations, we found that BPs cannot be opened if the difference between overburden stress and minimum horizontal stress is large and only shear displacements exist along the BPs, which damage the planes and thus greatly amplify their hydraulic conductivity. Moreover, sensitivity studies investigate the impact on fracture propagation of parameters such as pumping rate (PR), fluid viscosity, and Young's modulus (YM). We investigated the fracture width near the junction between a vertical fracture and the BPs, the latter including the tensile opening of BPs and shear–displacement discontinuities (SDDs) along them. SDDs along BPs increase at the beginning and then decrease at a distance from the junction. The width near the junctions, the opening of BPs, and SDDs along the planes are directly proportional to PR. Because viscosity increases, the width at a junction increases as do the SDDs. YM greatly influences the opening of BPs at a junction and the SDDs along the BPs. This model estimates the fracture–width distribution and the SDDs along the BPs near junctions between the fracture tip and BPs and enables the assessment of the PR required to ensure that the fracture width at junctions and along intersected BPs is sufficient for proppant transport.

The wettability of reservoir rocks is important for oil recovery and reserve calculations. However, current methods for evaluating the wettability of rocks are time-consuming and expensive. Previous work has shown that low-field nuclear magnetic resonance (NMR) is a potentially useful and non-invasive technique for rock wettability determination. However, for rocks with strong internal magnetic field gradients, the current method is less efficient. In this study, the bipolar pulsed field gradient (PFG)-Carr-Purcell-Meiboom-Gill (CPMG) pulse sequence was applied to the study of rock wettability. This method can suppress the effect of the internal magnetic field gradient in rocks and accurately extract wettability information. The diffusion-transverse relaxation time (D-T2) method was employed to quantitatively estimate the wettability of rocks. Results of Amott wettability tests and NMR T1-T2 maps were combined to provide a more complete wettability characterization of tight sand. The results demonstrate the feasibility of the new method for characterizing wettability. The proposed method and workflow is of significance to the development of oil fields.

Abstract As the fiber Bragg grating (FBG) sensors, with a high credibility, high temperature resistance, corrosion-resistant, and anti-electromagnetic interference, are suitable for working in a harsh environment of oil and gas wells, we develop a FBG wireless sensor network to monitor the temperature and pressure of the reservoir formation. A data acquisition module is set up at wellhead to demodulate the analog signals into digital signals. Another data transmission module installed at wellhead can send data from the data acquisiton module through a RS-232 interface to dadabase by a GPRS wireless mobile communication network. The user can browse the real-time published data through internet. We build an experimental apparatus to simulate high temperature and high pressure of the downhole environment. We put a FBG sensor into the apparatus, increase the temperature and pressure gradually, and then reduce them back. The data acquisition module and data transmission module succeeded in their roles. In addition, we determined the extremes of the FBG sensor on temperature and pressure. Through repeating the above operation a couple of times, we obtained a satisfactory match between the input values and measured values. Our system can measure the deferent depth temperature and pressure of the formation in real time. It has many properties: responsivity, accuracy, a high speed transmission rate, and a low bit error rate. In addition, it can work for 24 hours and 7 days a week in all weather. To real-time monitor the temperature and pressure of the formation, the system can provide more reliable bases to engineers to predict and solve production problems. It has important practical significance particularly for outlying remote areas and offshore oil production. The application of this technology will effectively reduce the production of human errors and labor costs. Moreover, it will benefit the statistical analysis of massive data that require a unified management and sharing.

Summary Knowing the location of sweet spots benefits the horizontal well drilling and the selection of perforation clusters. Generally, geoscientists determine sweet spots from the well-logging interpretation. In this paper, a group of prevalent classifiers [extreme gradient boosting (XGBoost), unbiased boosting with categorical features (CatBoost), and light gradient boosting machine (LightGBM)] based on gradient-boosting decision trees (GBDTs) are introduced to automatically determine sweet spots based on well-log data sets. Compared with linear support vector machines (SVMs), these robust algorithms can deal with comparative scales of features and learn nonlinear decision boundaries. Moreover, they are less influenced by the presence of outliers. Another prevailing approach, named generative adversarial networks (GANs), is implemented to augment the training data set by using a small number of training samples. An extensive application has been built for the field cases in a certain oilfield. We randomly select 73 horizontal wells for training, and 13 features are chosen from well-log data sets. Compared with conventional SVMs, the agreement rates of interpretation by XGBoost and CatBoost are significantly improved. Without special preprocessing of the input data sets and conditional tabular GAN (CTGAN) model fine tuning, the fake data set could still bring a relatively low agreement rate for all detections. Finally, we propose an ensemble-learning framework concatenating multilevels of classifiers and improve agreement rate. In this paper, we illustrate a new tool for categorizing the reservoir quality by using GBDTs and ensemble models, which further helps search and identify sweet spots automatically. This tool enables us to integrate experts’ knowledge to the developed model, identify logging curves more efficiently, and cover more sweet spots during the drilling and completion treatment, which immensely decrease the cost of log interpretation.

AbstractPorosity sensitivity is an important index for determining the spacing of the neutron porosity tool, and optimized spacing can improve formation porosity resolution. Based on the neutron flux formula obtained through a double-group approach, we study the variations of fast-neutron slowing-down length and thermal-neutron diffusion length and their derivatives with formation porosities; calculate porosity sensitivity for different porosity values; and derive relationships for how the distance between the source and the near detector or the distance between the two detectors depends on porosity sensitivity. Both the theoretical analysis and the calculations show that the relationship between porosity sensitivity and the short spacing is not monotonic. For a fixed spacing interval, sensitivity increases with increasing spacing in lower porosity formations, whereas the opposite occurs in relatively higher porosity formations. The spacing has little influence on the sensitivity. For a fixed short spacing, the sensitivity is a monotonically increasing function of the spacing interval in the whole range of formation porosity.

Abstract Organic matter in shale is very important for shale evaluation. However, current methods for evaluating the structure and porosity of organic matter are time consuming and expensive. Previous low-field NMR work using magic-echo methods provided relationships between magic echo T 1 − T 2 data and total organic carbon (TOC) of shales using partial least-squares regression (PLSR), but uncertainty in estimation of organic pores indicated further study was required. The magic echo and solid echo measurements produce additional signal in the oil shale samples compared to the standard methodologies. The difference between these signals is mainly caused by the homonuclear dipolar coupling in the organic matter; therefore, it implies the content of the organic matter. Compared to the spin echo method, the solid echo and magic echo measurements are able to access additional information of the quantity and porosity of the organic matter in shale samples. Based on this, we suggest a new method using the amplitude of magic echo, solid echo and spin echo to obtain information about quantity and porosity of the organic matter in oil shale samples. The results demonstrate the feasibility of the new method for organic matter estimation.

Abstract Heavy oil has been playing a critical role in today's world energy supply. The total amount of heavy oil in place is five to ten times greater than that of the current proven conventional crude. One of the recovery methods, which produces both oil and unconsolidated sands, is known as Cold Heavy Oil Production with Sand (CHOPS). The advantages of CHOPS lie in its commercial success as an inexpensive start up application for heavy oil reservoirs as well as its considerable recovery rates. The general reservoir characteristics associated with successful applications of CHOPS have been established, particularly highlighted in thin reservoirs with non-active edge and bottom water. Heavy oil researchers have accumulated local knowledge for the CHOPS fields; particularly, research groups in Alberta have taken integrated approaches to the questions posed by the field success of cold production. CHOPS gives high early production rates and becomes very efficient in the thin reservoirs where some thermal methods have been economically unsuccessful. Aggressive sand production was encountered in California prior to the First World War. Two key mechanisms lead to the success of cold production in laboratory and field studies: foamy oil flow and wormhole network growth. A variety of numerical models are presented and compared in this paper. Such models can be mainly divided into two broad categories: preliminary model and comprehensive model. With a large number of variables still in limited recognition for the complex mechanisms, several models lack capability in fully simulating CHOPS processes, while progress was achieved in modeling the reservoir heterogeneity with the integration of seismic attributes at specific fields. A detailed discussion of the strengths and weaknesses of cold production models is proposed. The paper ends with the future work of modeling proposed on cold production.

Abstract The augmenting necessity of energy saving and environmental protection has led to the increasing technical requirements about on-line monitoring of key parameters and elements in coal. In this study, terahertz spectroscopy combined with PCA (principal components analysis) was employed to analyze nine kinds of coal materials. Due to the strong absorption of the organics with the relatively high C/H ratio, such as aromatic compounds, coals with lower hydrogen content show higher absorption effects in terahertz range. Based on PC1 score calculated by PCA, the anthracite and bituminous coal can be clearly classified, and the clean as well as meagre coals were also distinguished by opposite trends. All the critical points in PCA system are in agreement with those (volatile matter and ash, respectively) classified in international standard of coal classification. In addition, significant elements, including carbon, hydrogen, nitrogen and sulfur, can be directly characterized using PC1 score with linear and non-linear models. This research indicates that the terahertz spectral analysis of key parameters and elements of coal is a promising tool for improving the detection method and advancing the technical innovation in coal processing industry.