--> --> Innovations in Borehole Microresistivity Image Data Analysis

International Conference & Exhibition

Datapages, Inc.Print this page

Innovations in Borehole Microresistivity Image Data Analysis

Abstract

The automation of repetitive tasks in borehole image data analysis increases efficiency and consistency, and delivers time savings that can be invested in more effective integration and interpretation. Such automation has been enabled by the development of multiple innovative technologies, three of which are considered here: the data-driven deterministic reconstruction of missing data to address incomplete coverage from wireline logs (and gaps in LWD images), the calibration and characterization of microresistivity measurements to provide quantitative resistivity values for petrophysical evaluations, and dynamorphic processing for image enhancement, avoiding the artifacts introduced by the ubiquitous dynamic normalization. Data reconstruction is based on a modified morphological components analysis approach in which features in the image are represented sparsely in an appropriate domain such that inversion of the sparse representation recovers the missing data uniquely and completely to the extent that the information in the measured parts is representative of that in the gaps. Reconstructed images are almost indistinguishable from full-coverage images for coverage loss up to 30%, and the method performs well for coverage loss up to about 50%. Quantitative resistivity moves away from the previous practice of ad-hoc normalization. It imposes a model-derived calibration and correction for borehole effect. We have developed a high-fidelity numerical model of the whole measurement system; this shows that each measurement electrode has its own unique calibration coefficient, and that these vary in a systematic way across the button array. The response to environmental factors such as standoff is similarly unique for each button. Visualization addresses the challenge of rendering high dynamic range resistivity data with a necessarily limited color palette. Among the newly developed visualizations is the dynamorphic image which splits the data into sharp and less-sharp components, scales them independently then re-combines. The innovations have been used with new edge detection algorithms to enable the robust automated detection of planar boundaries. In a test involving 350m of image data from a fractured shale-sand interval, the new detection algorithm ran 30 times faster than a leading commercial auto-dip program, detected substantially more features, and false picks were reduced by more than an order of magnitude.