A Novel Sparsity Deploying Reinforcement Deep Learning Algorithm for Saturation Mapping of Oil and Gas Reservoirs
- PDF / 1,781,547 Bytes
- 7 Pages / 595.276 x 790.866 pts Page_size
- 62 Downloads / 198 Views
RESEARCH ARTICLE-PETROLEUM ENGINEERING
A Novel Sparsity Deploying Reinforcement Deep Learning Algorithm for Saturation Mapping of Oil and Gas Reservoirs Klemens Katterbauer1
· Alberto Marsala1
Received: 11 May 2020 / Accepted: 4 October 2020 © King Fahd University of Petroleum & Minerals 2020
Abstract The 4IR technology has assumed critical importance in the oil and gas industry, enabling automation at an unprecedented level. Advanced algorithms are deployed in enhancing production forecast and maximize sweep efficiency. A novel sparsitybased reinforcement learning algorithm, utilizing a surface response model approach, was developed for the estimation of hydrocarbon saturation in the interwell region. Application of the novel algorithms on a realistic reservoir box model exhibited strong performance in the estimation of the interwell saturation as well as the quantification of uncertainty. The results outline the broader application of the framework for interwell saturation mapping. Keywords Reinforcement learning · Uncertainty quantification · Saturation mapping · Surface response model
1 Introduction The Fourth Industrial Revolution (4IR) of technology has gained significant traction in the oil and gas industry, focusing on maximizing oil recovery and improving efficiency. From exploration to production of hydrocarbon reservoirs, 4IR technologies have led to significant advances in the industry. A major challenge in the industry is the sparsity of reservoir information that is available only close to the wellbores, from which extensive estimates of the hydrocarbon distribution inside the reservoir volumes between the wells are conducted. Although a broad number of interpolation and geostatistical techniques have been developed, these are mostly characterized by assumptions and a priori data interpretations. Reducing the uncertainties associated with the lack of direct geological, petrophysical and fluid distribution information in the interwell volumes of a field is a key opportunity to improve the understanding of a reservoir, assessing
Electronic supplementary material The online version of this article (https://doi.org/10.1007/s13369-020-05023-2) contains supplementary material, which is available to authorized users.
B
Klemens Katterbauer [email protected] Alberto Marsala [email protected]
1
Saudi Aramco, Dhahran, Saudi Arabia
its exploitation potential and maximizing the hydrocarbon recovery from a field. Data-driven AI approaches have found wider adoption in recent years for more accurately estimating reservoir properties far from wellbores. Ertekin et al. provided an extensive overview of the applications of artificial intelligence in reservoir engineering problems [1]. The authors present an overview about advanced machine learning approaches and more state-of-the-art artificial intelligence technologies for the reservoir engineering domain. A particular focus in the article is on advanced machine learning algorithms and their utilization for addressing challenges encount
Data Loading...
