Advanced Horizontal Well Correlation Method for Dynamic Update of Subsurface Layers While Geosteering

Abstract

A new automated approach to well correlation is presented that utilizes real-time Logging While-Drilling (LWD) data and predicted well curve to dynamically update subsurface layers during geosteering operations. The automatically created predicted log and a dynamically updated structural framework provides the foundation of the process. The predicted log is created using vertical sections of the nearby wells, which provide high confidence for determining depth and stratigraphic position of the geosteered well. The results give a better understanding of thickness variation in the horizontal part of the reservoir and maximize the reservoir contact (Sung, 2008). A new advanced methodology introduced in this study involves the creation of a dynamic structural framework model, from which horizontal well correlation is performed using real-time well logs and predicted logs that are generated from adjacent wells. The predicted logs are correlated to the LWD logs using anchor points and an interactive stretching and squeezing process that honors true stratigraphic thickness. Each new anchor point results in the creation of an additional control point that is used to build a more precise structural framework model. This new approach enables more rapid well log interpretation, increased accuracy and the ability to dynamically update the subsurface model during drilling. It also enables more efficient steering of the wellbore into the most productive zones of the reservoir. This study demonstrates how wells with over 10,000 feet of horizontal reservoir contact can be correlated in a real-time geosteering environment in a dynamic, efficient and accurate manner. The proposed process dramatically helps reduce the cost of drilling and the time it takes to dynamically regenerate accurate updated maps of the subsurface. It represents a major improvement in the understanding and modeling of complex, heterogeneous reservoirs by fostering a multi-disciplinary environment of cross-domain experts that are able to collaborate seamlessly as asset-teams. Both accuracy and efficiency gains have been realized by incorporating this methodology in the characterization of multi-stacked reservoirs.