The SPE has split the former "Management & Information" technical discipline into two new technical discplines:
- Data Science & Engineering Analytics
The SPE has split the former "Management & Information" technical discipline into two new technical discplines:
|Theme||Visible||Selectable||Appearance||Zoom Range (now: 0)|
Khan, Mohammad Rasheed (SLB) | Kalam, Shams (King Fahd University of Petroleum & Minerals) | Asad, Abdul (SPRINT Oil & Gas services) | A. Abu-khamsin, Sidqi (King Fahd University of Petroleum & Minerals)
Abstract Unconventional reservoirs like shale oil/gas are expected to play a major role in many unexplored regions, globally. Shale resource evaluation involves the estimation of Total Organic Carbon (TOC) which correlates to the prospective capability of generating and containing hydrocarbons. Direct measurement of TOC through geochemical analysis is often not feasible, and hence researchers have focused on indirect methods to estimate TOC using analytical and statistical techniques. Accordingly, this work proposes the application of artificial intelligence (AI) techniques to leverage routinely available well logs for the prediction of TOC. Multiple algorithms are developed and compared to rank the most optimum solution based on efficiency analysis. Support Vector Regression (SVR), Random Forest (RF), and XGBoost algorithms are utilized to analyze the well-log data and develop intelligent models for shale TOC. A process-based approach is followed starting with systematic data analysis, which includes the selection of the most relevant input parameters, data cleaning, filtering, and data-dressing, to ensure optimized inputs into the AI models. The data utilized in this work is from major shale basins in Asia and North America. The AI models are then used to develop TOC predictor as a function of fundamental open-hole logs including sonic, gamma-ray, resistivity, and density. Furthermore, to strengthen AI input-output correlation mapping, a k-fold cross-validation methodology integrating with the exhaustive-grid search approach is adopted. This ensures the optimized hyperparameters of the intelligent algorithms developed in this work are selected. Finally, developed models are compared to geochemically derived TOC using a comprehensive error analysis schema. The proposed models are teted for veracity by applying them on blind dataset. An error metrics schema composed of root-mean-squared-error, and coefficient of determination, is developed. This analysis ranks the respective AI models based on the highest performance efficiency and lowest prediction error. Consequently, it is concluded that the XGBoost and SVR-based TOC predictions are inaccurate yielding high deviations from the actual measured values in predictive mode. On the other hand, Random Forest TOC predictor optimized using k-fold validation produces high R values of more than 0.85 and reasonably low errors when compared to true values. The RF method overpowers other models by mapping complex non-linear interactions between TOC and various well logs.
Abstract This paper presents the results of numerical simulations of hydraulic fracturing in the immediate vicinity of the wellbore. This research aims to identify the primary mechanisms underlying the complexities in both the fracture morphology and propagation of longitudinal fractures. The study shows that the perforation attributes and characteristics, the cement quality, and the reservoir heterogeneity have a significant impact on the resulting morphology and the trajectory of the propagating hydraulic fracture. The study is based on properties and conditions associated with a field study conducted in the Austin Chalk formation, and concludes that the pattern and the dimensions of the perforations are essential factors controlling the fracture initiation pressure and morphology. The results of the simulation studies provide insights into the principles and mechanisms controlling fracture branching and the initiation of longitudinal fractures in the near-wellbore region and can lead to improved operational designs for more effective fracturing treatments.
Wu, Bohong (Research Institute of Petroleum Exploration & Development, PetroChina) | Nie, Zhen (Research Institute of Petroleum Exploration & Development, PetroChina) | Li, Yong (Research Institute of Petroleum Exploration & Development, PetroChina) | Deng, Xili (Research Institute of Petroleum Exploration & Development, PetroChina) | Ma, Ruicheng (Research Institute of Petroleum Exploration & Development, PetroChina) | Xu, Jiacheng (Research Institute of Petroleum Exploration & Development, PetroChina)
Abstract Marginal reserves are an important play in future energy development. Based on the statistics of China National Petroleum Corporation (CNPC), the low permeability and unconventional reservoirs occupied 92% of newly found proven reserves in China. To overcome challenges such as poor reservoir conditions, weak natural energy, low displacement efficiency, and insufficient single well production, CNPC has conducted years of research and operation to cost-effectively develop China's marginal reserves. To develop the marginal fields economically, it is required to maximize single well production, recovery and reservoir sweep with minimum CAPEX and OPEX reasonably. The production enhancement is realized by 3 key technologies, namely, sweet spot identification, multi-layered 3D short spacing horizontal well pattern, and volumetric fracturing techniques. The cost reduction is achieved by the full life cycle practice of utilizing "large cluster, factory" well design and field operation, drilling prognosis optimization, integrated intelligent surface system, and unmanned operation. CNPC cost-effective development mode is practical and successful, marginal fields characterized with heterogeneous, multi-layered oil-bearing intervals with poor continuity are being economically developed in China. By comprehensive geological study, fit-for-purpose technologies application, and geoscience-to-engineering integration, the fracture control degree of horizontal wells increased from 60% to more than 90% based on micro-seismic events, stimulated reservoir volume (SRV) increased by 46.8%, average cumulative oil production per well is more than 100 times than original production in the field. Fast and early cash flow is realized by minimum production facilities. The average drilling cycle is shortened by 61%, the surface facility construction time is reduced by 65%, and the average single well investment is reduced by 42%.
Zhu, Jun (Vertechs Energy Group) | Zhang, Wei (Vertechs Energy Group) | Zeng, Qijun (Vertechs Energy Group) | Liu, Zhenxing (Vertechs Energy Group) | Liu, Jiayi (PetroChina Southwest Oil & Gas Field Company) | Liu, Junchen (PetroChina Southwest Oil & Gas Field Company) | Zhang, Fengxia (PetroChina Southwest Oil & Gas Field Company) | He, Yu (PetroChina Southwest Oil & Gas Field Company) | Xia, Ruochen (PetroChina Southwest Oil & Gas Field Company)
Abstract In the past decade, the operators and service companies are seeking an integration solution which combines engineering and geology. Since our drilling wells are becoming much more challenging than ever before, it requires the office engineer not only understanding well construction knowledge but also need learn more about geology to help them address the unexpected scenarios may happen to the wells. Then a novel solution should be provided to help engineers understanding their wells better and easier in engineering and geology aspects. The digital twin technology is used to generate a suppositional subsurface world which contains downhole schematic and nearby formation characteristics. This world is described in 3D modelling engineers could read all the information they need after dealt with a unique algorithm engine. In this digital twin subsurface world, the engineering information like well trajectory, casing program, BHA (bottom hole assembly) status, are combined with geology data like formation lithology, layer distribution and coring samples. Both drilling or completion engineers and geologist could get an intuitive awareness of current downhole scenarios and discuss in a more efficient way. The system has been deployed in a major operator in China this year and received lot of valuable feedback from end user. First of all, the system brings solid benefits to operator's supervisors and engineers to help them relate the engineering challenges with according geology information, in this way the judgement and decision are made more reliable and efficiently, also the solution or proposal could be provided more targeted and available. Beyond, the geology information from nearby wells in digital twin modelling could also provide an intuitional navigation or guidance to under-constructed wells avoid any possible tough layers via adjusting drilling parameters. This digital twin system breaks the barrier between well construction engineers and geologists, revealing a fictive downhole world which is based on the knowledge and insight of our industry, providing the engineers necessary information to support their judgement and assumption at very first time when they meet downhole problems. For example, drilling engineers would pay extra attention to control the ROP (rate of penetration) while drilling ahead to fault layer at the first time it is displayed in digital twin system, which prevent potential downhole accident and avoid related NPT (non-production time). The integration of engineering and geology is a must-do task for operators and service companies to improve their performance and reduce downhole risks. Also, it provides an interdisciplinary information to end user for their better awareness and understanding of their downhole asset. Not only help to avoid some possible downhole risks but also benefit on preventing damage reservoir by optimizing the well construction parameters.
Abstract Oil production via horizontal wells with multistage fracture stimulation treatment completions in the Bakken shale of North Dakota and Montana began in 2003. Since then, over 19,000 Bakken shale horizontal wells have been completed and placed into production. Oil production from horizontal Bakken shale oil wells peaked in November 2019 at 1.5 million barrels/day, and is at about 1.2 million barrels/day as of September, 2022 (EIA). There have been several shale oil EOR tests conducted over the last several years, involving the injection of water, CO2 and natural gas. This paper builds upon shale EOR modeling work described in a 2019 NETL report. In that report, a compositional simulation model of the Bakken was constructed, and a production history match on primary oil, gas and water production from a group of wells was obtained. The match model was then used to evaluate the enhanced oil recovery via cyclic injection of CO2, dry gas, and wet gas. This paper utilizes some data from that report to assess two novel, proprietary shale oil EOR processes in the Bakken, in the same area of the Williston Basin. The paper illustrates how these proprietary shale oil EOR processes may be implemented at lower BHP to mitigate interwell communication, while enabling greater oil recovery than via injection of water, CO2 or natural gas. Compositional reservoir simulation modeling of the two novel EOR processes in the modeled Bakken shale wells indicates potential incremental oil recoveries of 200% and 300% of primary EUR may be achieved. The two novel shale oil EOR methods utilize a triplex pump to inject a liquid solvent having a specific composition into the shale oil reservoir, and a method to recover the injectant at the surface, for storage and reinjection. One of the processes enables further enhanced oil recovery via cyclic fracture stimulation at the start of the EOR process. The processes are fully integrated with compositional reservoir simulation to optimize the recovery of residual oil during each injection and production cycle. The patent pending shale oil EOR processes have numerous advantages over cyclic gas injection - shorter injection time, longer production time, smaller, lower cost injection volumes, no gas containment issue - much lower risk of interwell communication, elimination of the need to buy and sell injectant during each cycle, much better economics, scalability, faster implementation, optimization via integration with compositional reservoir simulation modeling, and lower emissions. If implemented early in the well life, their application may preclude the need for artificial lift, to produce more oil sooner, resulting in a shallower decline rate and higher reserves.
Cornelio, Jodel (University of Southern California) | Mohd Razak, Syamil (University of Southern California) | Cho, Young (University of Southern California) | Liu, Hui-Hai (Aramco Americas) | Vaidya, Ravimadhav (Aramco Americas) | Jafarpour, Behnam (University of Southern California)
Abstract Given sufficiently extensive data, deep-learning models can effectively predict the behavior of unconventional reservoirs. However, current approaches in building the models do not directly reveal the causal effects of flow behavior, underlying physics, or well-specific correlations; especially when the models are trained using data from multiple wells of a large field. Field observations have indicated that a single reservoir does not have similar production behaviors. This makes pre-filtering the data to build local models that capture region specific correlations more pertinent than a single global model that will provide averaged-out predictions from different correlations. In this work, we investigate a sophisticated network architecture to expedite the clustering process by training the global model. We utilize attention-based (transformer) neural networks for the input data before mapping to the target variable to extract the attention scores between well properties and the production performance. We leverage the interpretability from these attention-based models to improve the prediction performance for data-centric models derived from clustered datasets. We show the benefits of building local models that are more accurate as they learn correlations that are more region/data specific. Specifically, by utilizing the attention mechanism, we can separate and curate data subsets to train local models, improving the prediction performance by reducing the variability in the entire field.
Mohd Razak, Syamil (University of Southern California) | Cornelio, Jodel (University of Southern California) | Cho, Young (University of Southern California) | Liu, Hui-Hai (Aramco Americas) | Vaidya, Ravimadhav (Aramco Americas) | Jafarpour, Behnam (University of Southern California)
Abstract Neural network predictive models are popular for production forecasting in unconventional reservoirs. They have the ability to learn complex input-output mapping between well properties and observed production responses from the large amount of data collected in the field. Additionally, the flow behavior in hydraulically fractured unconventional reservoirs is not well understood making such statistical models practical. Variants of neural networks have been proposed for production prediction in unconventional reservoirs, offering predictive capability of varying levels of granularity, accuracy and robustness against noisy and incomplete data. Neural network predictive models that incorporate physical understanding are especially useful for subsurface systems as they provide physically sound predictions. In this work, we propose a new Dynamic Physics-Guided Deep Learning (DPGDL) model that incorporates physical functions into neural networks and uses residual learning to compensate for the imperfect description of the physics. The new formulation allows for dynamic residual correction, avoids unintended bias due to less-than-ideal input data, and provides robust long-term predictions. The DPGDL model improves upon a static formulation by utilizing a masked loss function to enable learning from wells with varying production lengths and by improving the results when partially-observed timesteps are present. We also develop a new sequence-to-sequence residual model to correct additional biases in the long-term predictions from the physics-constrained neural networks. Several synthetic datasets with increasing complexity as well as a field dataset from Bakken are used to demonstrate the performance of the new DPGDL model.
Abstract Pore-scale dependent phase behavior describes a decrease in the hydrocarbon phase envelope as pore throat size decreases. This phenomenon is well documented in terms of confining effects on phase behavior with several analytical fluid models proposed that account for these effects. Results from a limited number of numerical reservoir models show the effects pore-scale phase behavior has on total production. However, fewer studies consider fluid transfer between different scale pore networks as a function of scale-dependent phase behavior. This work investigates fluid transfer between different scale pore networks related to scale-dependent phase behavior and the affects it has on production and fluid composition in the pore networks. A commercially available reservoir simulator is used with a dual porosity/permeability grid and scale-dependent fluid models to study the fluid transfer between pore networks. Fluid tracking is used to trace fluid phases and components that originate in both the nanoscale and macroscale pore networks. Fluid transfer between pore networks is considered at both the pore network scale and at the well stream scale by tracking the fluid components from nano-scale pores into macro-scale pores and ultimately to the well bore. The results from the model are used to quantify fluid transfer between pore networks. The results of the study show how the confining effects on fluid phase behavior affect fluid production rates and gas-oil ratios by linking the pore scale processes to the well stream scale production. For example, as fluid moves from the nanoscale pores, where the bubble point is suppressed and the fluid retains the initial solution gas-oil ratio (Rs), into the macro scale pores, the fluid in the macroscale pores is enriched by the nanoscale pore fluid. This work provides three main contributions to an improved understanding and characterization of unconventional plays. The first is demonstrating the ability to simulate the confining effects on fluid phase behavior using commercially available reservoir simulators. Second is the ability to capture some of the unique production trends observed for tight oil reservoirs, e.g., extended periods of stable GOR, when modeling these reservoirs. The third contribution is in tight oil EOR, providing insight into the composition of the fluid that remains in the pore networks following primary depletion or at the onset of an EOR process.
Deng, Xiao (Department of Petroleum Engineering, King Fahd University of Petroleum & Minerals, Dhahran, Saudi Arabia) | Kamal, Muhammad Shahzad (Center for Integrative Petroleum Research, King Fahd University of Petroleum & Minerals, Dhahran, Saudi Arabia) | Patil, Shirish (Department of Petroleum Engineering, King Fahd University of Petroleum & Minerals, Dhahran, Saudi Arabia) | Shakil, Syed Muhammad (Center for Integrative Petroleum Research, King Fahd University of Petroleum & Minerals, Dhahran, Saudi Arabia) | Al Shehri, Dhafer (Department of Petroleum Engineering, King Fahd University of Petroleum & Minerals, Dhahran, Saudi Arabia) | Zhou, Xianmin (Center for Integrative Petroleum Research, King Fahd University of Petroleum & Minerals, Dhahran, Saudi Arabia) | Mahmoud, Mohamed (Department of Petroleum Engineering, King Fahd University of Petroleum & Minerals, Dhahran, Saudi Arabia) | Al Shalabi, Emad Walid (Petroleum Engineering Department, Khalifa University, Abu Dhabi, United Arab Emirates)
Abstract Low permeability rock usually holds a large amount of residual oil after flooding. The two most important mechanisms for residual oil recovery are interfacial tension (IFT) reduction and wettability alteration (WA). There is confusion around the coupled effect between the two mechanisms. Permeability is found to be a critical factor on the coupled effect. In this study, the spontaneous imbibition oil recovery results from core plugs of different permeability by using two surfactants were compared. The comparison helps understand the impact of permeability on the coupled effect of IFT reduction and WA. Filtered crude oil (density 0.87 g/mL, viscosity 12.492 cP), Indiana limestone cores of different permeabilities, two locally synthesized cationic gemini surfactants, GS3 and GS6, were used in this study. The spinning drop method and static contact angle method were used to measure the oil/water IFT and the wettability. Spontaneous imbibition experiments using Amott cells were conducted at the ambient condition to relate IFT reduction and WA performance to the oil recovery contribution. Results showed that although the selected surfactants had comparable IFT reduction performance, GS3 is much stronger than GS6 in altering oil-wet carbonate rock to water-wet conditions. In core plugs with the same dimensions and comparable low permeabilities, the oil recovery values accorded with the WA performance. GS3 obtained faster and higher oil recovery (24%) than and GS6 (14%), indicating that enhancing WA alone contributes to oil recovery. The main difference between the selected surfactants was the spacer structure. It appeared that introducing unsaturation into the spacer group harmed the WA performance. Comparing different permeability conditions, GS6 obtained much higher oil recovery in a high permeability condition (922 mD) than in a low permeability condition (7.56 mD). Though permeability significantly impacted the whole imbibition process, it was more auspicious when IFT reduction became the main driving force. This study studied the WA mechanism alone by adopting surfactants with comparable oil/water IFT values. It also features the impact of permeability by comparing the recovery curve by the same surfactant under different permeability, showing that IFT reduction contributes more to oil recovery in high permeability rock.
Egbe, U. C. (University of Alaska Fairbanks) | Awoleke, O. O. (University of Alaska Fairbanks) | Olorode, O. M. (Louisiana State University (Corresponding author)) | Goddard, S. D. (University of Alaska Fairbanks)
Summary Several authors have worked on combining decline curve analysis (DCA) models and stochastic algorithms for probabilistic DCAs. However, there are no publications on the application of these probabilistic decline curve models to all the major shale basins in the United States. Also, several empirical and analytical decline curve models have been developed to fit historical production data better; there is no systematic investigation of the relevance of the efforts on new model development compared with the efforts to quantify the uncertainty associated with the “noise” in the historical data. This work compares the uncertainty associated with determining the best-fit model (epistemic uncertainty) with the uncertainty associated with the historical data (aleatoric uncertainty) and presents a procedure to find DCA-stochastic algorithm combinations that encompass the epistemic uncertainty. We investigated two Bayesian methods—the approximate Bayesian computation and the Gibbs sampler—and two frequentist methods—the conventional bootstrap (BS) and modified BS (MBS). These stochastic algorithms were combined with five empirical DCA models (Arps, Duong, power law, logistic growth, and stretched exponential decline) and the analytical Jacobi theta-2 model. We analyzed historical production data from 1,800 wells (300 wells from each of the six major shale basins studied) with historical data lengths ranging from 12 to 60 months. We show the errors associated with the assumption of a uniform distribution for the model parameters and present an approach for integrating informative prior (IP) probabilistic distributions instead of the noninformative prior (NIP) or uniform prior distributions. Our results indicate the superior performance of the Bayesian methods, especially at short hindcasts (12–24 months of production history). We observed that the duration of the historical production data was the most critical factor. Using long hindcasts (up to 60 months) leveled the performance of all probabilistic methods regardless of the decline curve model or statistical methodology used. Additionally, we showed that it is possible to find DCA-stochastic model combinations that reflect the epistemic uncertainty in most of the shale basins investigated. The novelty of this work lies in the development of IPs for the Bayesian methodologies and the development of a systematic approach to determine the combination of statistical methods and DCA models that encompasses the epistemic uncertainty. The proposed approach was implemented using open-source software packages to make our results reproducible and to facilitate its practical application in forecasting production in unconventional oil and gas reservoirs.