Synergizing Crosswell Seismic and Electromagnetic Techniques for Enhancing Reservoir Characterization

Abstract

SummaryIncreasing complexity of hydrocarbon projects and the request for higher recovery rates have driven the oil-and-gas industry to look for a more-detailed understanding of the subsurface formation to optimize recovery of oil and profitability. Despite the significant successes of geophysical techniques in determining changes within the reservoir, the benefits from individually mapping the information are limited. Although seismic techniques have been the main approach for imaging the subsurface, the weak density contrast between water and oil has made electromagnetic (EM) technology an attractive complement to improve fluid distinction, especially for high-saline water. This crosswell technology assumes greater importance for obtaining higher-resolution images of the interwell regions to more accurately characterize the reservoir and track fluid-front developments. In this study, an ensemble-Kalman-based history-matching framework is proposed for directly incorporating crosswell time-lapse seismic and EM data into the history-matching process. The direct incorporation of the time-lapse seismic and EM data into the history-matching process exploits the complementarity of these data to enhance subsurface characterization, to incorporate interwell information, and to avoid biases that may be incurred from separate inversions of the geophysical data for attributes. An extensive analysis with 2D and realistic 3D reservoirs illustrates the robustness and enhanced forecastability of critical reservoir variables. The 2D reservoir provides a better understanding of the connection between fluid discrimination and enhanced history matches, and the 3D reservoir demonstrates its applicability to a realistic reservoir. Historymatching enhancements (in terms of reduction in the historymatching error) when incorporating both seismic and EM data averaged approximately 50% for the 2D case, and approximately 30% for the 3D case, and permeability estimates were approximately 25% better compared with a standard history matching with only production data.

SAMA: Spatially-Aware Model-Agnostic Machine Learning Framework for Geophysical Data

“…To overcome some challenges of seismic data such as low resolution, crosswell electro magnetic surveys are proposed to better understand the fluid types and saturation in the inter-well region of a reservoir [32]. A group of studies is deticated to mapping water saturation from these surveys using machine learning models [10], [33] [23], [11].…”

Recent Developments in Closed-Loop Approaches for Real-Time Mining and Petroleum Extraction

“…By repeating seismic surveys after a period of production it is often possible to identify the fronts between water or gas displacing oil, and this information can help to drastically increase the quality of the update; see e.g., Skjervheim (2007), Trani et al (2012), Jin et al (2012), Emerick and Reynolds (2013) and Hadavand and Deutsch (2016). Also, the increasing use of different geophysical measurements such as gravity or electromagnetic surveys can be of help to obtain a better picture of the subsurface fluid distribution, and, through inversion, of the reservoir model parameters; see e.g., Glegola et al (2012) and Katterbauer et al (2016).…”

Feature-Oriented Joint Time-Lapse Seismic and Electromagnetic History Matching Using Ensemble Methods

“…In general, no matter which level is selected, obtaining a consistent comparison between measured (including interpreted or inverted geophysical attributes) and simulated data is not trivial. There is a growing investigation of incorporating different forms of geophysical data and studying their performances under the ensemble-based history-matching framework [51,15,30,27,60]. Lastly, geophysical data involve large uncertainty that stems from the whole acquisition-processing-interpretation process and is usually difficult to quantify.…”