Ming Chen

Summary: Sealed wellbore pressure monitoring (SWPM) offers a low-cost fracture diagnostic method for identifying fracture-driven interaction and nonuniform propagation of multiple fractures. Although previous studies have revealed pressure responses in sealed wellbores (SWs) induced by hydraulic fractures, current understanding remains insufficient, and reliable fracture simulation together with wellbore deformation analysis is still required to interpret SW pressure signals. To address this, we use a planar 3D multifracture propagation simulator to calculate SW deformation and pressure evolution. Using the 3D displacement discontinuity method (DDM) to compute wellbore strain by treating the wellbore as an elastic line problem, we establish a forward model and computational method to simulate volume and pressure change in a SW induced by fracture propagation during horizontal well fracturing. Through numerical simulations, we investigate the evolution of SW volume and pressure during fracture growth under varying cluster number, injection rate, well spacing, and leakoff coefficient. Our results indicate that SW volume and pressure responses induced by both single and multiple fracture propagation exhibit similar characteristics. The time when a fracture hits the SW for a single fracture can be identified precisely from the peak in the pressure change rate, which also allows us to estimate the fracture propagation average velocity. After pumping shut-in, while the direction of SW volume and pressure change remains unchanged, their rate of change reverses, and the pressure change rate effectively reflects injection dynamics. Higher injection rate and lower leakoff coefficient lead to faster fracture propagation, larger fracture length and width, and stronger amplitudes and change rates in the SW volume and pressure responses. Larger well spacing results in weaker stress disturbance signals and lower SW volume and pressure response amplitudes. When the well vertical spacing exceeds the height range of fracture propagation, the SW response becomes too weak to effectively evaluate vertical fracture propagation and fracture hit. The peak in the SW pressure change rate may slightly lag behind the time when the fracture hits the offset well for multifracture growth, a delay attributed to the superposition of strain effects from nonuniform multifracture propagation. Analysis of post-shut-in SW pressure data using the G-function demonstrates that the dp∕dG and Gdp∕dG characteristic curves can accurately determine the closure time of fractures intersecting the offset well. This study provides technical insights for interpreting SWPM signals.

Summary: A carbon dioxide (CO2)-hybrid fracturing approach combines the benefits of CO2 and water-based fracturing techniques, where liquid CO2 is injected to initiates hydraulic fractures (HFs), and is then followed by the injection of proppant-laden water-based fluid to further extend and support the HFs. After being heated by the wellbore and reservoir, liquid CO2 generally transforms into its supercritical state. There is an interface between supercritical CO2 (Sc-CO2) and the water-based fluid, which advances inside the HF. Due to the strong contrast between the two fluid viscosities, the interface dynamics can affect fluid pressure and HF width distributions in two regions occupied by dissimilar fluids. For this study, we developed a fully coupled HF propagation model to deal with two-phase fluid flow in fractures by describing the time-dependent movement of interface with a volume of fluid (VOF) equation. First, we verified the model capability to track interface movement using an analytical solution to the fracture problem with constant inlet pressure. Then, we performed a parametric study to determine the mechanisms affecting HF propagation in layered reservoirs. Numerical results revealed that HF height is significantly smaller than its length at the end of Sc-CO2 fracturing, as the higher-stress layers strongly delay HF vertical growth due to higher-rate leakoff of low-viscosity Sc-CO2 into rock mass and bedding planes (BPs). At this stage, the Sc-CO2 front coincides with the horizontal HF tip, but as fracturing proceeds, the HF propagates vertically into the adjacent higher-stress layers, and horizontal growth is temporarily suppressed until the water-based fluid catches up with the HF tip. The duration of coexistence of two fluids in the HF depends primarily on the pumping rate of water-based fluid. The subsequent fracturing of water-based fluid promotes an increase in HF length and height. For the case studied, the leakoff distance of Sc-CO2 along BPs can reach approximately 56 m, which is about 13 times greater than that of high-viscosity linear gel. The large-scale infiltration of Sc-CO2 elevates the pressure within the BPs, limiting the leakage of linear gel in facilitating HF extension. Obviously, there is an optimization in Sc-CO2 leakoff and HF growth rate by adjusting the pumping rate and amount of liquid CO2, and the optimal operational parameters of CO2-hybrid fracturing play a role. Finally, we performed an optimization study for a vertical shale oil well, and the proposed model and findings offer practical guidance for designing CO2-hybrid fracturing treatments in multilayered reservoirs.

Summary: Distributed optical fiber sensing (DOFS) technology, with its high spatial resolution, long-distance monitoring capability, and resistance to electromagnetic interference, demonstrates significant application potential in hydraulic fracturing crack monitoring. However, due to the complexity of signals during crack propagation and their susceptibility to noise interference, accurately inverting crack geometric parameters remains a technical challenge. To address this issue, we propose an innovative forward-reverse interactive (FRI) long short-term memory (LSTM) (FRI-LSTM) model based on deep learning techniques. The model utilizes strain data from a forward model of crack propagation to invert crack parameters, including fracture length, inlet width, and average width. First, the study compares the performance of models such as recurrent neural network (RNN), gated recurrent unit (GRU), LSTM, and Transformer in inverting distributed optical fiber strain signals. The results show that traditional LSTM outperforms other models in capturing dynamic features and modeling nonlinear relationships. Building on this, the proposed FRI-LSTM model leverages forward and reverse feature interaction modeling and a joint optimization mechanism to significantly enhance the global modeling capability and prediction accuracy of the LSTM model. Compared with traditional LSTM, the proposed model achieves a 47% reduction in mean absolute error (MAE), a 47% reduction in root mean square error (RMSE), and an increase in R2 from 0.6179 to 0.7500. This study is the first to introduce neural networks into the field of distributed optical fiber strain data interpretation, providing an innovative solution for efficient crack parameter interpretation. The findings provide a new interpretation approach for hydraulic fracturing crack monitoring, while also offering a methodological reference for addressing complex time-series modeling challenges in broader applications such as pipeline surveillance and seismic signal analysis.

Summary: Low-frequency (LF) distributed acoustic sensing (DAS) (LF-DAS) facilitates the analysis of hydraulic fracture propagation by detecting strain variations in neighboring wells during the fracturing process. While earlier studies have established a theoretical foundation for interpreting simple fracture-induced optical fiber strains, field observations, particularly in geologically complex conditions, have exposed the intricacies of optical fiber strain patterns. The current comprehension of optical fiber strain evolution remains inadequate for interpreting fracture propagation dynamics. To address this, we use a 3D complex fracture propagation simulator to calculate optical fiber strain and strain rate using the 3D displacement discontinuity method. Through numerical simulations, we investigate the strain and strain rate evolution associated with various fracture types, including planar hydraulic, natural, and bedding-plane fractures. We also apply our findings to field data from shale gas fracturing in the Sichuan Basin, comparing the derived strain and strain rate patterns with LF-DAS monitoring results. Our findings indicate that, in horizontal wells, planar fractures with fixed and unrestricted heights produce similar strain and strain rate patterns, complicating the identification of fracture height. Natural fractures with nonperpendicular dip angles exhibit strain and strain rate patterns that align with their dip direction. The relative contributions of natural and hydraulic fractures to overall strain dictate the orientation of the strain pattern, which can be predominantly inclined or horizontal. When the strain contribution from natural fractures is minimal, the primary fracture characteristics are more discernible, overshadowing natural fracture features. In cases where 90° natural fractures extend, LF-DAS strain signals in horizontal wells exhibit symmetrical compression-tension patterns that correspond to the natural fracture length. The copropagation of hydraulic fractures and horizontal bedding in the absence of fracture-fiber interaction results in LF-DAS strain and strain rate signatures akin to those observed in vertical well monitoring. The field case study confirms the presence of natural fracture propagation strain patterns, corroborating the geological data of the region. This study lays a theoretical groundwork for interpreting complex strain signals and sets the stage for more precise fracture geometry inversion.

Summary: As the exploitation of oil and natural gas has progressed, hydraulic fracturing has become a primary method for increasing oilfield production. Simultaneous hydraulic fracturing of several perforation clusters, by employing limited-entry methods, has become standard in horizontal well stimulation. However, challenges such as unclear fracture identification and limited monitoring methods persist in hydraulic fracturing. This study integrates an optical frequency domain reflector with true triaxial fracturing of multilevel horizontal wells to develop a physical simulation system for monitoring fractures in a laboratory setting via distributed fiber optics. By employing fiber optics, dynamic monitoring of fractures during the fracturing process of multilevel horizontal wells can be achieved. The results indicate that monitoring with distributed fiber optics can clearly record data and accurately determine the initiation points of fractures. The strain data induced by the fractures on the fibers can be interpreted to deduce the fracture width. If a fracture deflects upon encountering a fiber, the fiber will exhibit tensile strain within an abnormal range. When using distributed fiber optics for monitoring fractures in multilevel horizontal wells, it is crucial to optimize the placement of fractures to ensure that the signals detected by the fibers are complete and avoid signal loss. In this paper, we demonstrate the feasibility of using distributed fiber optics for fracture monitoring in multilevel horizontal well fracturing experiments, overcoming the limitations of current single-method approaches to monitoring laboratory true triaxial hydraulic fractures.

Summary: The interaction between natural fractures (NFs) and hydraulic fractures (HFs) in 3D space poses significant challenges to numerical simulation. The interaction behaviors between HFs and NFs in 3D space include crossing, offset, stopping, and bypassing. Many existing numerical simulators are 2D, which inherently limits their ability to account for the vertical growth of fractures. Consequently, they are unable to model the bypassing behavior effectively. In this paper, a mathematical model for the propagation of HFs and NFs in a naturally fractured reservoir is established. The displacement discontinuity method (DDM) is used to solve the rock deformation, while the finite difference method (FDM) is utilized to solve the fluid flow within fractures. The undirected graph structure is used to represent the complex fracture network, and a dynamic adjustment of grid connectivity method is implemented to describe the process by which HFs bypass NFs. By integrating with graphics processing unit (GPU) computing, an efficient 3D simulator for HFs and NFs propagation is developed. The accuracy is verified against analytical solutions and reference solutions. Subsequently, a series of numerical studies on the bypassing behavior are conducted. The primary findings are as follows: (1) The simulator can accurately capture the interactions between HFs and NFs in 3D space, including crossing, arresting, and bypassing behaviors. (2) The bypassing behavior is characterized by three distinct processes—intersection, height growth of HF, and bypassing. (3) During the intersection stage, both the injection pressure and maximum NF width increase. During the height growth stage, the pressure is relatively stable, while the maximum NF width continues to increase. These two stages together result in the “storage” phenomenon. Once the bypassing behavior occurs, both the pressure and the maximum NF width decrease sharply, leading to a “release” phenomenon. Additionally, the “storage” and “release” phenomena may impact proppant transport. (4) Given the presence of bypassing behavior, it is essential to consider the NFs in 3D fracture simulators. The simulator and its findings can provide valuable insights for field design.

Summary: Crosswell low-frequency distributed acoustic sensing (LF-DAS) strain rate measurements generate data useful for hydraulic fracture (HF) diagnostics. These data contain information about HF geometry; however, interpreting the data has relied on time-consuming and computationally expensive numerical interpretation methods. This study aims to evaluate the applicability of a rapid diagnostic tool, known as the “zero strain rate location method” (ZSRLM), for characterizing the geometry of multiple fractures propagating simultaneously. Simulations of multiple fracture propagation are conducted using a planar 3D multifracture simulator that incorporates 3D rock deformation, fluid flow in the wellbore, fluid leakoff, and multiscale fracture propagation regimes. A supertimestepping algorithm is used to solve the nonlinear parabolic equations governing fluid-driven fractures. The far-field strain profile induced by multiple propagating fractures along a crosswell fiber-optic cable is computed using the displacement discontinuity method (DMM). Various scenarios are simulated by adjusting key completion parameters, including the number of clusters, cluster spacing, in-situ stress states, and geomechanical properties. The ZSRLM is then applied to the computed strain field to estimate fracture extents and propagation rates. The accuracy of the ZSRLM is assessed by comparing the estimated fracture geometry parameters with the known simulated values. Our observations indicate that the ZSRLM is not effective when two or more fractures approach the observation well at similar times, as it becomes challenging to discern unique strain-rate converging patterns on waterfall plots for each fracture. However, when fracture arrivals at the monitor well are offset, distinct strain rate patterns emerge, enabling the application of the ZSRLM. Errors between simulated and estimated fracture extent propagation rates are quantified. Finally, the ZSRLM is applied to a field case involving multicluster fracture propagation, and the interpretation of the results is discussed in light of these findings. This research demonstrates the applicability of the ZSRLM for diagnosing the velocities and final extents of multiple fractures propagating simultaneously. The method offers a rapid means of characterizing stimulations using crosswell LF-DAS measurements, which can inform decisions regarding completion design, well spacing, and landing zone. The study contributes to the advancement of efficient fracture diagnostics in the industry, reducing the reliance on numerical interpretation methods.

Summary: The uneven propagation of multifractures is a key factor restricting production growth due to stress shadow and heterogeneity. To date, limited-entry fracturing techniques, nonuniform perforation, and in-stage diversion have been commonly used to promote even multifracture growth. In this study, a fully coupled multiple pseudo-3D (P3D) fracture simulator has been developed to examine the competitive propagation of multifractures during multicluster fracturing in a horizontal well. The present model considers stress interaction among multiple fractures, perforation erosion, fluid distribution among clusters, and in-stage diversion. The results of the model are validated against the reference data. Using the model, a series of numerical simulations are performed to investigate multifracture propagation with nonuniform perforation and in-stage diversion fracturing. We estimate the value of stress interaction for different fractures and time based on the approximate solution of Perkins-Kern-Nordgren (PKN) fracture in the viscosity-dominated regime and improve the dimensionless parameter that characterizes the competition between stress interaction and perforation friction. The fluid distributes evenly when the dimensionless parameter is less than unity (perforation friction is larger than stress interference). Based on this dimensionless parameter, a method to design nonuniform perforation and in-stage diversion is proposed. Results show that in the case of homogeneous in-stage stress, the perforation parameters should be selected under the condition that the dimensionless parameter is less than unity. In the case of heterogeneous in-stage stress and based on the perforation parameters selected under homogeneous stress conditions, the perforation holes in the high-stress cluster should be increased, making the reduction of perforation friction equal to the value of the in-stage stress heterogeneity. The stress heterogeneity can be balanced by decreasing the perforation friction of the high-stress clusters. In this way, nonuniform perforation under heterogeneous in-stage stress conditions can be designed quantitatively without numerical simulation. For in-stage diversion treatment, a method to design the number of ball sealers is proposed based on the results of nonuniform perforation, and only several or even zero groups of simulation are necessary to find the optimal number of ball sealers. A series of numerical simulations shows that the proposed design method is reliable and achieves a satisfactory result in an actual field case. The results can be helpful for nonuniform perforation and in-stage diversion design for multicluster fracturing in a horizontal well.