Author | Gao Yuxian
The deepest oil well in China is the “Man Shen 10 Well” located in the Tarim Oilfield, with a drilling depth of 9,186 meters—over 300 meters deeper than the height of Mount Everest. It is easy to imagine that exploring oil and gas reserves in such ultra-deep strata, which pose significant drilling risks and construction challenges, is a “rare global and unique domestic” drilling and completion engineering problem. Continuous breakthroughs in technological innovation are necessary to elevate China’s ultra-deep oil and gas exploration and development levels to the forefront of the world.
Due to the unique nature of the oil operation environment, many high-risk or unsuitable tasks for manual completion—from drilling and extraction to transportation and processing—are gradually being replaced by machines. This has led to a higher overall level of industrial automation in the oil industry compared to other traditional manufacturing sectors, which is also why AIoT (Artificial Intelligence of Things) applications are relatively common in the oil industry.
For example, through “IoT + Big Data + Artificial Intelligence,” remote control and real-time monitoring of intelligent drilling rigs thousands of meters underground can be achieved, along with intelligent inspections and safety warnings for scattered operational area equipment, resource allocation and fault diagnosis during oil and gas production, and more precise predictions and positioning for the unfathomable underground exploration…
The implementation of AIoT is accelerating the realization of these basic smart oilfield scenarios—especially in the current unpredictable global environment, this is of great significance for a country like China, which does not have abundant oil resources.
A technical expert in the energy sector (referred to as “Sun Jie”) told InfoQ that China’s oil resources are relatively scarce, and the dispersed regions and complex geology make production costs remain high. “For example, 80% of oil in Middle Eastern countries is concentrated in shallow and medium-depth layers, where oil can be extracted at depths of over 2,000 meters. In China, each well must be drilled at least over 3,500 meters to reach oil, which means spending more money and incurring higher costs. Therefore, we must continuously use new technologies to increase oil production and reduce labor and cost inputs. The intelligent operations emphasized by AIoT can effectively meet the current demands of the oil industry for cost reduction and efficiency improvement.”
1 What Problems Can AIoT Solve in Oil Production?
According to Sun Jie, AIoT can mainly address several core issues in the oil production process: first, real-time monitoring; second, efficient management; third, intelligent analysis.
Firstly, based on AIoT-related technologies, data collection equipment can be placed at various key points, allowing for real-time monitoring of equipment through data transmission and analysis, thus achieving safety warnings. “For example, during the oil product delivery process at a gas station, leaks can lead to accidents. However, it is generally impossible for a person to monitor the pipeline valves in real-time. In this case, sensors can monitor the pressure in the pipeline and transmit it to the backend in real-time. If abnormal pressure values are detected, timely measures can be taken without waiting for problems to occur in the pipeline, while also reducing human operational errors.”
In addition to using sensors and other equipment for data transmission and monitoring to prevent gas leaks, another measure is to use drones for intelligent inspections. Sun Jie explained that during pipeline operations, if the sealing ring becomes loose or damaged, it can cause significant gas leaks, potentially leading to explosions and other safety incidents, jeopardizing safe transportation.
Drones can pre-plan flight paths and fly at altitudes of 50m to 80m above the pipeline, continuously monitoring methane concentrations on the surface through alternating flight patterns. Specifically, equipped with a three-axis stabilization gimbal, the monitoring laser beam can continuously and stably point to the measured area, with a minimum detectable concentration of 5ppm.m. Moreover, drones can monitor the measured area in real-time, displaying the data in slope maps, and with the H20t visible light camera, they can generate infrared thermal images. After the inspection task is completed, drones can automatically generate inspection reports, including alarm point locations, alarm concentrations, and on-site photos, etc. “Overall, drone inspections not only have better accuracy than human eyes but also higher efficiency, typically allowing for several inspections in a single day,” Sun Jie emphasized.
Secondly, another characteristic of oil production is the requirement for equipment to operate without interruption. Sun Jie provided an example to InfoQ: if a drilling rig stops operating during the drilling process, it not only affects the normal progress of drilling construction but also incurs significant economic losses. This means that equipment inspection and maintenance work must be proactive, rather than waiting for accidents to occur before taking action. In the past, this work was mainly performed by experienced workers. If a drill bit made unusual noises, whether due to bearing wear or jamming, they had to rely on their experience to judge the specific problem and decide whether to replace the drill bit.
However, handling such issues relies heavily on human experience and the long training cycle for skill development, which incurs high time costs. “Now, with AIoT technologies, we can transform the experience of skilled workers into algorithmic models, allowing machines to identify and diagnose equipment issues, perform predictive maintenance, and assist managers in making judgments through real-time data collection, transmission, and analysis of drilling environments and equipment, thereby shortening decision response times and improving management efficiency.”
It is worth emphasizing that the data collection, aggregation, and analysis processes are not completed manually but are automatically conducted according to fixed data standards. Sun Jie stressed that this is the third application value of AIoT—achieving intelligent analysis.
“In the past, manual aggregation and analysis took an average of two to three hours, and sometimes even a day to obtain analysis results. However, with intelligent prediction and analysis, all work is conducted in real-time and automatically.” Sun Jie provided an example: “If a manager wants to know the production of a single well or a specific operational area, they only need to input the relevant parameters into the corresponding model based on pre-set parameters, allowing the machine to automatically analyze and output results—usually, this can generate the required analysis report within minutes or even seconds, making it very convenient for managers to make timely decisions.”
2 What Infrastructure Is Needed for AIoT Implementation?
However, addressing these production issues involves more than just a simple combination of AI, IoT, and data. Although the overall automation level in the oil industry is higher than in other traditional manufacturing sectors, there are still many old difficulties and new challenges to face in the implementation of AIoT, requiring new infrastructure as a foundation.
The foremost challenge comes from data collection. The harsh conditions of oil field operations pose threats to both equipment and personnel. This means that sensors installed in equipment and environments for data collection must withstand these potential “threats”—for example, sensors placed in pipelines must be corrosion-resistant, those in offshore drilling must be waterproof, and those in high-temperature areas must also be humidity-resistant. “In addition to using special materials, we often design protective covers or insertable designs. Moreover, if they are in the sea, sensors must also comply with environmental protection standards and be regularly maintained and replaced to avoid polluting marine environments,” Sun Jie explained.
At the same time, during the data collection process, it is crucial to break down data barriers between different devices to achieve data visualization. Sun Jie stated that, like most traditional manufacturing industries, the equipment used in the oil industry often comes from different brands and models, meaning their data protocols and formats are not the same. The solution is to use data conversion boxes—similar to adapters for different brand phone chargers—that can parse different data protocols during data collection and automatically aggregate and present them according to a unified standard.
The second challenge comes from data transmission. Generally, manufacturing enterprises have relatively concentrated factory layouts, but oil production is limited by geographical conditions, with operational areas being very dispersed and mobile, often located in remote areas, at least dozens of kilometers away from city centers. Transmitting all data to a unified data center is not only costly but also inefficient. “In the past, we could only transmit this data through some collectors and dedicated cables, which was very slow. Additionally, because it is in the wild, the cables need to be very long and exposed, leading to rapid aging and instability in network transmission,” Sun Jie stated.
With the maturity of edge computing technology, this issue is being addressed. “For example, by deploying micro data centers in different well areas, we can process data from the operational site inside, achieving near real-time speeds,” Sun Jie further explained. “However, micro data centers have limited data capacity. Therefore, when the data volume exceeds a certain scale, it is necessary to first transmit the data to a nearby center and then aggregate it to the headquarters data center using dedicated lines.”
The third challenge comes from data computation. As intelligent operations progress and evolve, the scale of data that oil fields can collect and need to collect will continue to grow, and the types of data will also become more diverse. For example, AI can be used to improve the accuracy of oil exploration, but the exploration locations are in unexcavated underground areas, requiring three-dimensional modeling in conjunction with digital twins—resulting in a massive amount of data. For instance, constructing a high-precision three-dimensional digital model covering dozens of square kilometers can result in data volumes reaching TB or even PB levels.
Sun Jie stated that processing and analyzing such large data volumes requires high-performance computing to truly enhance computational accuracy. Storing such large amounts of data also incurs significant costs—therefore, considering economic and efficiency factors, data is typically managed in a classified and tiered manner.
“For different types of data, we will implement storage and computation separation,” Sun Jie provided an example. “For instance, data related to production that requires real-time computation and tracking can be transmitted to high-performance computing devices for real-time processing, while data used for preliminary geological exploration and mapping can be stored directly without real-time computation.”
3 What Pitfalls May Arise from the Introduction and Use of New Technologies?
Of course, the development of technology has its cycles, and the integration of technology with human capabilities also requires time, which is an objective law. From the agricultural era when humans used tools to meet their needs for food and clothing, to the industrial era when machinery helped humans achieve large-scale production, and to the information age when computers became important aids for human intellectual creation—every technology, when first introduced, goes through a “chaotic period.”
The intelligent era, centered on AI, IoT, and big data, is no exception. The two extremes are either blindly “worshiping” new technologies or resisting them. Sun Jie stated that they have experienced both of these “pits.”
“As a traditional industry, we initially thought that many technology companies had strong capabilities and believed they could help us utilize technologies like AI to improve production efficiency and operational levels. However, we gradually realized that relying solely on advanced technology is insufficient; their technology cannot deliver value if it is detached from practical application scenarios. It is akin to having a powerful hammer but no nails to drive; the hammer has no value,” Sun Jie emphasized. Technology must integrate with specific business scenarios to solve problems rather than merely showcase technical prowess.
Therefore, conducting business demand research before introducing technology is crucial. “In the early stages of a project, our technical personnel usually spend one to two months conducting in-depth research at the oil field operation site, gathering frontline feedback into a demand document; based on this, we conduct preliminary prototype technology development and provide it to frontline personnel for trial use. If issues arise during the trial process, we continuously improve it,” Sun Jie stated.
In this process, a core difficulty lies in how to translate business needs into technical language—for example, in the context of AIoT, it involves how to convert experts’ experiences into algorithmic models for optimizing drilling efficiency.
Taking oil drilling as an example: data comes from various stages of the drilling process, is diverse in type, and is vast in quantity; moreover, drilling parameters and underground environmental information are complex, diverse, and uncertain (e.g., underground environment, geological parameters), making drilling decisions rely more on expert experience rather than scientific decision-making.
“To enhance the scientific nature of decision-making, our drilling experts and technical personnel have conducted extensive optimization work on the drilling process, proposing numerous optimization models and improvement methods,” Sun Jie pointed out. “Especially as the data volume in the drilling industry continues to grow, traditional data processing models have limitations in handling unstructured data. Therefore, utilizing more advanced technologies and management methods to collect, integrate, and optimize various data related to oil drilling, transforming human experience into more optimized computational models, is of great significance for improving drilling efficiency.”
In Sun Jie’s view, communication and collaboration between technology and business are crucial in this process. Without input from frontline workers’ experiences, relying solely on technology to label and annotate application data may lead to numerous errors. However, another challenge is the gap in the language systems between technology and business, which can make this collaborative process less than smooth.
“For example, when oil workers mention drilling rigs, three-phase separators, oil products, logging, and measurement, these specialized business terms may be incomprehensible to technical personnel; conversely, when technical personnel discuss cloud computing, virtualization, and containers, they may also be difficult for workers to understand,” Sun Jie stated. “At this point, both sides need to sit down and listen to each other, with business explaining their workflows to technology and technology explaining how it solves problems. Sometimes, using simple analogies can help; for instance, when explaining the difference between public and private clouds to workers, we might say that buying materials to cook at home is like a private cloud, while going to a restaurant is like a public cloud. Through such relatable examples, they can understand more easily.”
However, Sun Jie also admitted that even so, there will still be some individuals who resist and reject new technologies. “In such cases, our approach is to let them directly experience the benefits of new technologies. Just like when smartphones first appeared, some people resisted them, but when they saw others not only using phones to make calls and send texts but also to book flights, send videos, watch movies, hail taxis, and order takeout, they gradually changed their minds.”
In addition to subtle influences, cultural reshaping, institutional norms, and skill training at the enterprise level are also important. “For example, we regularly organize relevant technical salons internally and offer online video learning courses to help workers better understand new technologies; additionally, we assess workers based on performance, such as using new technologies or more advanced solutions to achieve cost reductions and efficiency improvements, which can earn them innovation bonuses; furthermore, we regularly organize technical innovation competitions to set examples and inspire more people,” Sun Jie stated.
“In summary, the benefits that AIoT brings to the oil industry are tangible. In addition to reducing costs and improving efficiency in production operations, it can also balance supply and demand through intelligent operations, enhancing energy utilization efficiency. With the rapid development of technology, we believe that AIoT will accelerate the digital transformation and upgrading of oil and gas enterprises,” Sun Jie concluded. “Of course, as the industry continues to evolve with technology, we must adhere to the principle of collaborative innovation between business and technology, looking to the stars while keeping our feet on the ground, ensuring that technology is integrated into actual business scenarios to realize digital transformation effectively.”
Click at the bottomRead the original textVisit the InfoQ official website for more exciting content!
Today’s Recommended Articles
Is Java Making a Comeback? Don’t Joke, It’s Always Been Popular
Voices from Dozens of R&D Personnel: Measuring R&D Efficiency Shouldn’t Just Be a Numbers Game
“Only 28% of Huawei’s Employees Are Under 30” Trends on Hot Search; Tencent’s Q2 Net Profit Halved, Over 5,500 Employees Reduced; Baidu Cloud Responds to Manual Review of User Photos | Q News
The “Extra Second” That Drives Big Companies Crazy: Google, Microsoft, Meta, and Amazon Propose to Abandon It