Dragon Oil

The oil and gas industry is undergoing a digital transformation where artificial intelligence (AI) is becoming central to improving efficiency, safety, and sustainability. With drilling operations producing massive volumes of both structured and unstructured data from real-time sensor readings to daily drilling reports extracting timely, actionable insights remain a key challenge. This paper presents a Generative AI-based solution developed to bridge this gap by transforming how drilling data is analyzed and interpreted. Integrated within a broader AI platform, the solution leverages advanced Large Language Models (LLMs) to process and synthesize structured datasets alongside text-based operational documents. The solution focuses on addressing critical drilling challenges such as lost circulation, stuck pipe, mud and cement control, and non-productive time (NPT). By providing proactive insights and predictive guidance, the system empowers engineers and drilling teams to make faster, smarter, and more sustainable decisions across the well lifecycle. In addition to its predictive and diagnostic capabilities, the proposed system also plays a pivotal role in knowledge retention and continuous learning across drilling projects. By automatically capturing insights, lessons learned, and operational best practices from previous wells, it enables organizations to build a centralized, evolving knowledge base that supports decision-making in future operations. This not only reduces dependency on individual expertise but also ensures consistency in operational performance across teams and geographies. As AI continues to mature, such intelligent solutions will define a new era of cognitive drilling operations, where intelligent systems augment human expertise to achieve safer, faster, and more cost-effective wells.

Accurate reservoir characterization remains a critical challenge during the early stages of reservoir development. While traditional methods have dominated the oil and gas industry, recent advancements in artificial intelligence (AI) and machine learning (ML) techniques have offered transformative potential for addressing the challenges posed by reservoir complexity and the high dimensionality of geological, geophysical, and petrophysical data. This review examines the application of AI and ML in reservoir characterization, with a specific focus on inter-well connectivity and communication, as well as identifying optimal new drilling locations, also known as sweet spots. It discusses how integrating diverse data sources using AI/ML significantly enhances the accuracy of reservoir property analysis and behavior predictions. Additionally, the review discusses coupling methodologies that integrate AI/ML techniques with traditional reservoir characterization methods. It highlights the several benefits and drawbacks of such methods compared to the conventional approaches. Finally, the review discusses emerging AI/ML methods and potential future directions to further improve the accuracy and reliability of reservoir characterization.

The shallow water OBN survey area A, located in the Southern Caspian Sea, lies within a strike-slip fault zone. For the legacy data, the traditional time-domain root mean square (RMS) velocity was converted to depth-domain interval velocity for initial velocity model building, followed by reflection tomography to update the velocity model. However, this approach failed to meet the high-precision velocity model building and imaging requirements for this region. Considering the seismic data characteristics of the area, four innovative integrated techniques were applied to achieve high-precision velocity model building and depth migration imaging. These techniques include: (i)near-seabed velocity model building, (ii) structure-guided, well-controlled velocity model building, (iii) full waveform inversion, and (iv)tilted transverse isotropy (TTI)multi-azimuth grid tomography. The approach initially involved building a high-precision near-seabed velocity model by combining a first-break tomography velocity model with the corresponding high-velocity layer (HVL). A comprehensive velocity model for the mid-to-deep subsurface layers was developed by integrating well logs, geological horizons, and time-domain RMS velocity. The initial velocity model was then optimized using full waveform inversion (FWI). Finally, TTI multi-azimuth grid tomography was applied to address anisotropy in the subsurface media caused by the development of faults and fractures in the area. By combining these four advanced techniques, migration results of higher quality than the legacy data were achieved. The continuity and energy of events on the PSDM section improved significantly. The faults became clearer, and structural interpretation was more accurate. The migration result better satisfied the requirements for high-precision velocity model building and depth migration imaging in this region, providing higher-quality seismic imaging data for interpretation. The innovative integrated techniques proposed in this paper provide valuable references for seismic data processing in similar regions.

Abstract This study aims to evaluate the feasibility and potential value of CO2 injection for enhanced oil recovery (EOR) and carbon sequestration in an offshore clastic reservoir located in the Caspian Sea. By integrating geological, engineering, and environmental analyses, the research seeks to determine the effectiveness of CO2 injection in improving oil recovery rates while simultaneously reducing atmospheric CO2 levels through sequestration, beginning with numerical modeling. This feasibility study aspires to establish a comprehensive roadmap for Carbon Capture, Utilization, and Storage (CCUS) initiatives in the Caspian region, providing valuable insights and guiding future efforts in sustainable energy practices. The research includes an extensive review of Carbon Capture, Utilization, and Storage (CCUS) technologies and their current developments, analyzing technical and business challenges. A commercial numerical simulator was utilised performed this evaluation on a history-matched, upscaled dynamic model with geological properties analogous to the Caspian Sea reservoir. The model inputs included a calibrated compositional PVT model representing the reservoir fluids and a hypothetical CO2 fluid model. Multiple simulation scenarios were run to evaluate the benefits of CO2 injection for EOR and assess the potential for CO2 sequestration. The results demonstrate significant benefits from CO2 injection, notably an incremental recovery factor that extends the field's productive life. Preliminary injection prediction scenarios were presented, highlighting the potential to further optimise recovery by maximizing sweep efficiency. These findings provide a promising outlook for the use of CO2 injection in enhancing oil recovery and underscore the importance of continued optimization efforts. Key reservoir features in the Caspian Sea that they are heavily faulted, and require additional studies to address sealing and storage capacities under weakened rock conditions. This feasibility study aspires to establish a comprehensive roadmap for Carbon Capture, Utilization, and Storage (CCUS) initiatives in the Caspian region, providing valuable insights and guiding future efforts in sustainable energy practices.

The Early Paleocene to Middle Eocene Apollonia Formation in the BED-9 field has been particularly interesting for unconventional hydrocarbon exploration since its discovery in 2006. However, the multiscale compositional and diagenetic inconsistencies present challenges for its characterization. This study aims to evaluate the petrophysical properties of the Apollonia Formation to locate the sweet-spot intervals. Moreover, it seeks to investigate reservoir rock types (RRTs), depositional settings, and the impact of diagenesis on reservoir quality. The findings of this study are as follows. 1) The Apollonia A5 and C1 units are identified as "sweet-spot" intervals. Their effective porosity ranges from 18 % to 35 %, average permeability varies from 0.1 to 2.0 mD, and water saturation falls between 40 % and 50 %, indicating good reservoir quality. 2) High-order eustatic sea-level changes and repetitive climatic change cycles significantly influence the alternating carbonate productivity and dilution cycles. Five distinct RRTs are classified, denoting a gradational facies change from clean, argillaceous, and carbonaceous chalky limestone to marl and interbedded shale intervals. 3) Interpreting the electro-facies responses, collated with microfacies variations and faunal content, deepens our understanding of the depositional environment, which extends from the inner-to outer-shelf setting. 4) The diagenetic processes have a dual impact that enhances and diminishes the reservoir quality. Finally, the gap in evaluating the petrophysical characteristics of all the Apollonia members has been addressed based on integrating the petrophysical and facies analysis for A, B, and C members. The Apollonia Formation has unique characteristics as an unconventional hydrocarbon resource. (c) 2025 Petroleum Exploration and Production Research Institute Corporation, SINOPEC. Publishing services by Elsevier B.V. on behalf of KeAi Communications Co. Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).