EUVIMEDEUVIMED entdecken
Crossref

Real-Time Data Assimilation Using ML-Augmented Ensemble Kalman Filters for Dynamic Reservoir Management

Oghenekevwe Godspower Ovbije, Akindele Michael Okedoye

International Journal of Scientific Research and Modern Technology · 25.06.2026 · 152

DOI: 10.38124/ijsrmt.v5i6.1508 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Scientific Research and Modern Technology, online erschienen am 25.06.2026. Verfasst von Oghenekevwe Godspower Ovbije und 1 weiteren Autor:innen. Die Arbeit nennt 25 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Dynamic reservoir management is critical for optimizing hydrocarbon recovery and ensuring efficient resource utilization. Effective decision-making in such environments requires rapid integration of real-time data to adjust operational strategies. However, traditional ensemble-based data assimilation methods, such as the Ensemble Kalman Filter (EnKF), face computational bottlenecks when applied to high-dimensional reservoir systems. These limitations introduce latency in decision-making, making it challenging to adapt to evolving subsurface conditions efficiently. This article explores the integration of machine learning (ML) with EnKF to enhance real-time data assimilation in reservoir management. The proposed hybrid ML-EnKF workflow leverages neural networks to accelerate covariance estimation and reduce ensemble size requirements, significantly improving computational efficiency. By training a surrogate model on historical ensemble data, the framework bypasses iterative forecast steps, cutting data assimilation time by 50% while maintaining accuracy. A case study on a synthetic CO₂ storage reservoir demonstrates the effectiveness of this approach in adaptive well-control optimization. Real-time updates of pressure and saturation fields enable rapid adjustments to injection rates, mitigating overpressure risks. Results indicate a 22% improvement in historical production data matching and a 40% reduction in forecast errors compared to conventional EnKF methods. This ML-augmented EnKF approach bridges the gap between reservoir simulation and real-time analytics, offering a scalable solution for dynamic reservoir management, including carbon capture and storage (CCS) and hydrocarbon recovery.

Zitieren

APA

Ovbije, O. G., & Okedoye, A. M. (2026). Real-Time Data Assimilation Using ML-Augmented Ensemble Kalman Filters for Dynamic Reservoir Management. International Journal of Scientific Research and Modern Technology, 152. https://doi.org/10.38124/ijsrmt.v5i6.1508

Vancouver

Ovbije OG, Okedoye AM. Real-Time Data Assimilation Using ML-Augmented Ensemble Kalman Filters for Dynamic Reservoir Management. International Journal of Scientific Research and Modern Technology. 2026:152. doi:10.38124/ijsrmt.v5i6.1508

BibTeX
@article{Ovbije2026,
  author = {Ovbije, Oghenekevwe Godspower and Okedoye, Akindele Michael},
  title = {Real-Time Data Assimilation Using ML-Augmented Ensemble Kalman Filters for Dynamic Reservoir Management},
  journal = {International Journal of Scientific Research and Modern Technology},
  year = {2026},
  pages = {152},
  doi = {10.38124/ijsrmt.v5i6.1508}
}
Kostenlos starten →