Shell Exploration and Production Company
Well logs are key input data to construct and verify geomechanical models and unquestionably the most important data when it gets to rock property modeling. In addition, they usually show correlations, in some extent, with down-hole information such as fractured intervals, enlarged hole sections, oil/water/gas bearing zones etc. Although there are many petrophysical approaches to extract highest amount of information from logs, there is still a lot more valuable information buried in different frequency levels of log signals which cannot be pull out with the common petrophysical methods.
This presentation shares the results of an attempt to apply some helpful signal processing techniques such as wavelet de-composition, de-noising, Parzen classification, Bayesian algorithm, and data fusion to mine extra information out of log data.
In this study, the capability of signal processing techniques was examined for two main objectives: detecting borehole breakout intervals and identifying fractures zones. A new multi-variable workflow was proposed to identify zones of interest in correlation with basic well logs.
The workflow was applied to actual logs from a shale gas and a carbonate reservoir to investigate its accuracy and applicability. Results confirmed that the workflow is able to identify breakout and fractured zones with a significant accuracy.
AAPG Search and Discovery Article #90178©2013 AAPG Geosciences Technology Workshop, Baltimore, Maryland, July 16-17, 2013