Abstract

Oilfield enterprises need to respond to the demand for high-quality development and severe external challenges, and the construction of intelligent oil fields has become an inevitable development direction. This article summarizes and analyzes the advanced practices of domestic and foreign oilfield enterprises in intelligent construction, and believes that intelligent oilfields should have four abilities: comprehensive perception, integrated collaboration, early warning and prediction, and analysis and optimization. On the basis of the comprehensive promotion of the Internet of Things construction, the Chinese oil and gas industry has relied on the gradually accumulated massive data resources, combined with intelligent technological means, to carry out a series of explorations and practices in improving the quality and efficiency of oil and gas field exploration and development, production and operation. It has achieved good application effects in business scenarios such as reservoir prediction, reservoir description, and condition diagnosis. In order to fully realize the digital transformation and intelligent development of oilfield enterprises, the next step is to deepen research and development in dynamic automatic monitoring and intelligent control technology on production sites, new generation oilfield industrial big data intelligent technology, intelligent optimization technology for reservoir development, digital twin and intelligent operation command technology for oilfields, etc., and form a leading series of intelligent oilfield basic key technologies such as intelligent collection, comprehensive perception, intelligent control, early warning and prediction, to support high-quality development of oilfields and ensure national energy security.

Affected by factors such as the global political and economic situation and market competition, the oil and gas energy industry is experiencing intensified volatility, and the pattern of the oil industry has undergone significant changes. The global energy trend is towards diversification and low carbonization. With the decreasing growth rate of oil and gas demand, the operating efficiency of oil fields is declining, and the pressure on operating costs is increasing year by year. The petroleum industry is facing severe challenges. At the same time, the rapid development and widespread application of the fourth global industrial revolution driven by new generation information technologies such as big data, artificial intelligence, 5G, and cloud computing have triggered disruptive changes in the entire society and industry chain, bringing new development opportunities to the petroleum industry. The digital and intelligent technological revolution is opening up new avenues for the sustainable development of the petroleum industry. Building intelligent oil fields is not only an effective support for reducing costs, improving quality, and increasing efficiency in the petroleum industry, but also an inevitable trend in the development of main business technologies such as oilfield exploration and development. It can solve the practical difficulties faced in the high-quality development of oil fields.

On the basis of a basic description of the concept and connotation of intelligent oil fields, this article conducts research and analysis on the specific practices of domestic and foreign oil companies in the construction of intelligent oil fields. The focus is on the research and key technologies formed in the intelligent exploration and development, intelligent production and operation of oil and gas fields in China. Combined with specific work practices, the research situation of Shengli Oilfield in related fields is analyzed, and the next development direction and key technologies that need to be tackled of intelligent oil fields are analyzed, predicted, and forecasted, in order to provide certain reference and guidance for the construction and early realization of intelligent oil fields.

1. The Concept and Construction Practice of Intelligent Oilfield

Intelligent oil fields are an advanced stage of development in digital oil fields. In recent years, after experiencing the construction of digital oil fields and having basic data collection and management foundations, domestic and foreign oil companies have actively explored the in-depth application of intelligent technology in oil fields around core businesses such as exploration and development. Intelligent technology has gradually been promoted from point to surface, and the concept and connotation of intelligent oil fields have gradually become clear and enriched. The construction of intelligent oil fields has achieved good results.

1.1 The Definition and Connotation of Intelligent Oil Fields

Many scholars at home and abroad have conducted analysis and discussion on intelligent oil fields, but there is currently no unified definition. Shi Chongdong and others proposed that intelligent oil fields comprehensively formulate and implement rational and efficient oil field development by analyzing different fields such as environment layer, data layer, model layer, and application layer. Zhang Kai and others believe that intelligent oil fields mean facing major production problems such as dynamic analysis, automatic historical fitting, optimization of development plans, and measures to improve oil recovery during the development process. Based on real-time big data, they can "perceive" problems in oil reservoir development, use advanced models to "analyze" existing problems, "think" about the best strategies and plans through intelligent optimization schemes, and ultimately assist oilfield engineers in "decision-making" on-site implementation. The construction of intelligent oil fields is the process of gradually replacing or to some extent replacing artificial mental labor with computers or intelligent devices. Shi Yujiang and others believe that digital oil fields have replaced repetitive statistical work of humans, which is a process of applying knowledge. Intelligent oil fields have replaced partial analysis and induction work of humans, which is a process of creating knowledge and sharing knowledge. It is an inevitable result of the intelligent collaborative development of various businesses such as exploration and development technology, mining supporting industries, oilfield production and decision-making, and modern information technology applications.

Based on the experience and understanding of carrying out intelligent oilfield construction planning in Shengli Oilfield, this article believes that the essence and goal of intelligent oilfields is to build four capabilities: comprehensive perception, integrated collaboration, early warning and prediction, analysis and optimization, based on digital oilfields, around core assets such as oil reservoirs, oil and water wells, pipeline networks, equipment and facilities, relying on information technology to comprehensively assist in asset management and efficiency optimization, to assist in high-quality exploration and efficiency development, and to achieve the maximization of oilfield asset value. From the perspective of construction characteristics and connotations, intelligent oil fields should have four characteristics: IoT, modeling, integration, and visualization. IoT is the foundation, which utilizes sensors and other devices to perceive and monitor all asset objects and management objects in the oilfield. At the same time, real-time production and management related data is collected and timely transmitted to the backend for processing. Modeling is the core, which is based on comprehensive modeling of oil reservoirs, oil and water wells, pipeline networks, key equipment and facilities, and integrates simulation through various related relationships such as data and topology, laying the foundation for the optimization of various plans and measures. Integration is the key, including promoting data integration based on data lake construction, implementing application integration into data integration through the industrial Internet platform, and implementing application integration visualization through the industrial Internet platform. It is to display the static and dynamic data of various business fields in an intuitive form anytime, anywhere and on demand with the help of graphical, three-dimensional, mobile and other technologies, including using digital twin technology to achieve transparent visualization of production processes and business collaboration, and support interactive processing.

1.2 Practice of Foreign Oil Companies in Intelligent Oilfield Construction

Foreign oil companies started early in the field of intelligent oilfields and implemented important strategic measures in different aspects such as oilfield perception, analysis, and optimization. Especially in the past two years, by strengthening cooperation with information technology companies such as Microsoft and Google, they have directly introduced mature technologies from information technology companies in big data and artificial intelligence, and achieved certain results in the intelligence of core business such as exploration and development. 

ExxonMobil has partnered with Microsoft to apply technologies such as data lake, machine learning, and cloud computing in the development of its Permian Basin oil fields for intelligent oil and gas field construction. Collect data from a wide range of sensor networks, such as pressure and flow from wellheads, and store it on cloud platforms. Scientists and analysts can seamlessly and in real-time access it from anywhere, using advanced digital technologies such as artificial intelligence and machine learning to deeply mine the value of data to support business decision optimization and workflow automation. It is expected that by 2025, intelligent technology will support a production growth of 50000bbl/d in the Permian Basin oilfield, and it is hoped that in the next decade, billions of dollars in net cash flow will be generated through improved analysis and increased asset operating efficiency, achieving cost reduction, increased production, and reduced methane emissions throughout the entire lifecycle of the oilfield. Shell is conducting a pilot construction of an intelligent oilfield in the SF30 oilfield off the Borneo coast of Malaysia. By utilizing production testing data and geological reservoir data, a reliable big data model is established to accurately predict production conditions and optimize oil well lifting efficiency in real time. Based on the predicted results, parameters such as lifting flow rate, temperature, and pressure can be adjusted more quickly, achieving adjustments every 1-5 minutes, greatly improving lifting efficiency. The underground pressure and temperature sensors and hydraulic unit control valve switches are simultaneously connected to the DCS system to monitor the underground flow in real-time; By remotely adjusting the control valves of each layer driven by hydraulic pressure, real-time optimization control of the flow rate of each layer underground is achieved, achieving optimized combination oil production of multiple layers in oil wells, and improving the recovery rate by 0.25%.

TotalEnergies has established an integrated collaborative research platform for oil and gas production, achieving simulation and optimization of the entire production system including oil and gas reservoirs-injection and production wells-surface gathering and transportation. It supports interdisciplinary comprehensive research, cross departmental collaborative work, multi model integration and sharing, visual management of oil and gas reservoirs, and management decision-making assistance. Integrated dynamic simulation of production is carried out for each link of oil and gas reservoirs, injection and production wells, surface pipeline networks, and equipment, closely connecting individual production links. Various development plans are compared and evaluated before production, and development effects are tracked and evaluated after production. The entire production and operation system is optimized to achieve highly unified technical research objectives, providing an integrated simulation model for intelligent management of oil and gas field development, improving the efficiency and economic benefits of oil and gas field development.

1.3 Practice of Domestic Oilfield Enterprises in Intelligent Oilfield Construction

Domestic oilfield enterprises attach great importance to the research and construction of intelligent oilfield technology. Since the 12th Five Year Plan, on the basis of digital and networked construction, intelligent oilfield construction has been regarded as a strategic transformation and upgrading development for enterprises. They have successively launched research and pilot applications of intelligent oilfield related technologies. China Petroleum's 13th Five Year Plan focuses on the development of intelligent oil fields, with "unified data lake for exploration and development, unified technology platform, and universal application environment" as the core, to build the Dream Cloud for exploration and development, achieve full business chain data interconnection, technology interoperability, and business collaboration among upstream enterprises, build a new ecosystem of information construction and application for co creation, co construction, sharing, and win-win, and support business digital transformation and intelligent development. In November 2019, Exploration and Development Dream Cloud 2.0 was put into operation, integrating new technologies such as artificial intelligence, big data, cloud computing, Internet of Things, and mobile applications. Through the promotion of data lakes and unified technology platforms, it broke through the bottleneck of "difficult data sharing and business collaboration" in the past. It provides intelligent application support for six major business areas, including oil and gas exploration, development and production, collaborative research, production operation, business management, and safety and environmental protection. It has also implemented application scenarios in risk exploration in the Sichuan Basin, Tarim Oilfield trap review, and oil and gas well production management.

In 2013, Sinopec initiated the planning work for the construction of intelligent oil fields and simultaneously carried out research on key technologies for intelligent oil fields, achieving significant results and showing a rapid development trend. Facing the development goal of digitalization, networking and intellectualization of the petroleum and petrochemical industry, Huawei has jointly developed and launched the petrochemical intelligent cloud industrial Internet platform, forming a new mode and new format of deep integration of traditional industries and information technology, and taking the lead in the application field of intelligent manufacturing in China's process industry. Since the 13th Five Year Plan period, Sinopec's intelligent oil and gas field construction has steadily focused on production operation, integrated collaboration, and demonstration of intelligent oilfield construction. It has developed and promoted the application of production command systems (PCS), exploration and development business system platforms (EPBP), exploration and development cloud platforms (EPCP), and other collaborative platforms. It has created three basic capabilities of comprehensive perception, integrated collaboration, and global optimization, achieved unified management of data and professional software and hardware, and achieved results in improving efficiency and reducing costs through leading demonstrations, laying the foundation for the digitalization, informatization, and intelligent development of oil fields.

In 2019, CNOOC carried out top-level design work for digital transformation, proposing a construction concept of "cloud+platformization+agile development and delivery+cloud edge collaboration". Based on the cloud architecture of "data+platform+application", information system construction was carried out, using the integrated development and operation collaboration (DevOps) system for system research and development. The intelligent application technology system of "data+computing power+algorithm" was adopted for system deployment, achieving integration, collaboration, and sharing. Targeting business scenarios such as exploration and development, we carry out intelligent oilfield function design, technical implementation, and functional research and development to provide robust technical support for intelligent oilfields. Facing the entire lifecycle of oil and gas fields, from comprehensive research, on-site operations, business management to strategic decision-making at four levels, focusing on four typical scenarios of "transparent oil reservoirs, unmanned operations, collaborative operations, and knowledge-based decision-making", utilizing advanced information technology means to build a new exploration and development model for oil and gas fields, achieving efficient operation and value enhancement of oil fields.

2. Research Progress on Key Technologies of Intelligent Oil Fields

At present, both domestic and foreign oil fields have made significant progress in digitalization, especially major oil fields such as Daqing and Shengli in China have achieved real-time collection and processing of production site data, and have the conditions for real-time perception of ground production equipment and facilities, as well as wellbore and oil reservoir. At the same time, the accumulated massive data resources have also provided a good foundation for the construction of intelligent oil fields. China's major oil fields have carried out a series of explorations and practices on how to improve the quality and efficiency of oil and gas field exploration, development, and production operations, and have achieved certain results.

2.1 Intelligent Exploration Technology

In the field of oil and gas exploration, Chinese and foreign scholars have deepened their understanding of the six major elements of generation, storage, cap, transportation, circulation, and preservation for many years. In recent years, with the development and progress of relevant algorithms and computer software and hardware, the application of big data and artificial intelligence technology to solve the problems encountered in comprehensive exploration research has become a hot research direction. Among them, significant progress has been made in seismic data processing, evaluation of transport systems, seismic structures, and layer interpretation, forming a series of new technical means.

2.1.1 Intelligent Processing Technology for Seismic Data

Seismic data processing is a complex multi link system engineering, and it is also the foundation of reservoir prediction, reservoir description, and other work. The traditional working method relies on the experience of seismic data processing personnel, and there are many limitations in processing accuracy and efficiency. The application of big data and artificial intelligence technology to solve the problem of seismic data processing provides a new technical solution for accurate and efficient processing of seismic data.

At present, the main technical focus for intelligent processing of seismic data is on improving signal-to-noise ratio and reducing noise interference. For example, Gao Han aimed to address the issues of low-speed weathering layers that may exist on the refractive layer near the surface, as well as low signal-to-noise ratio and noise interference in seismic observation data. He used refractive wave travel time to calculate residual static correction values and constructed a dilated convolutional residual connection network that combines dilated convolution, ResNet residual units, and skip connections. This not only increases the perception field, but also accelerates training speed, improves the network's ability to learn details and stability, and achieves good denoising effects. Xiang Kui proposed an intelligent denoising method for seismic data based on convolutional neural networks. The model was trained using the deep learning framework Caffe, and the trained model was used for denoising seismic gathers. Six methods to accelerate the training speed of deep learning networks were discussed in depth. Guan Xizhu and others proposed a random noise suppression neural network based on residual convolutional neural network (ResNet) to address the problem of denoising in marine seismic data. By adding normalization layers and feedforward conduction processes in network layers, the gradient diffusion effect caused by deepening network layers can be effectively suppressed, which improves the training speed and denoising performance of the network model. 

In terms of seismic data processing, in addition to denoising research based on the data itself, reducing the process arrangement time and parameter experiments in each stage of the data processing process, promoting the sharing and inheritance of experience in seismic data processing, it is also an important research direction to improve the efficiency of seismic data processing. Therefore, Shengli Oilfield adopts clustering analysis technology in the fields of big data and artificial intelligence, and proposes an intelligent recommendation technology for seismic data processing flow. Using the collected processing module information to establish a sample set, determining the distribution pattern of the samples based on big data analysis results, and selecting the global K-means clustering analysis algorithm to cluster the sample set; To solve the problem of large sample size, the MapReduce computing framework is used to deploy the distributed K-means clustering algorithm. In the Map stage, the samples are segmented and clustered to form multiple cluster centers. In the re-production stage, the evaluation and selection of cluster centers are carried out to determine the optimal cluster center, ultimately achieving cluster analysis of all samples. Combining with the standard seismic data processing flow, search for the most similar module parameter combination from the clustering results, extract information to achieve intelligent recommendation of the processing flow. The intelligent recommendation technology for seismic data processing flow has been applied in surface consistency amplitude compensation, pre stack shot domain deconvolution, residual static correction and other processing steps. Static correction is a key technology that affects the signal-to-noise ratio and resolution of seismic profiles. By eliminating the residual error of baseline static correction, adjusting the stacking phase of common center point gathers, the goal of in-phase stacking is achieved (Figure 1). After processing, the cross-section is clear, and the data resolution and signal-to-noise ratio are effectively improved.