Mimicking the process of manual sequence stratigraphy well correlation

Abstract

Stratigraphic correlation of well logs is based on interactive interpreter-based pattern recognition. A skilled interpreter identifies similar patterns (such as upward fining and coarsening) in user-defined well sections and links them using either a conscious or subconscious stratigraphic model. This manual stratigraphic correlation of numerous wells in mature fields can be time-consuming and error-prone. To expedite the process of stratigraphic correlation, we have performed semiautomatic stratigraphic correlation of wireline logs from multiple wells using improved dynamic time warping (IDTW). IDTW uses semblance, which compares the shape of the well logs, to replace the Euclidean distance in pairwise error computation. The resulting error matrix is compatible with the lateral nonstationary variation of well logs in the same formation. The workflow begins with interpreting stratigraphic well tops on user-defined well sections that is similar to the current process of stratigraphy analysis. The interpreted wells are then treated as reference wells to aid in interpreting well tops for other wells. Necessary manual interventions are incorporated during the semiautomatic stratigraphic correlation process. We applied our method to two experimental fields: a sand-rich reservoir and a mud-rich reservoir. The applications illustrate that our method performs well in aggradational strata and successfully predicts discontinuities with manual interventions.