An Unsupervised Machine-Learning Workflow for Outlier Detection and Log Editing With Prediction Uncertainty

Abstract

Recent advances in data science and machine learning (ML) have brought the benefits of these technologies closer to the mainstream of petrophysics. ML systems, where decisions and self-checks are made by carefully designed algorithms, in addition to executing typical tasks such as classification and regression, offer efficient and liberating solutions to the modern petrophysicist. The outline of such a system and its application in the form of a multilevel workflow to a 59-well multifield study are presented in this paper. The main objective of the workflow is to identify outliers in bulk density and compressional slowness logs and to reconstruct them using data-driven predictive models. A secondary objective of the project is to predict shear slowness in zones where such data do not exist. The system is fully automated, designed to optimize the use of all available data, and provide uncertainty estimates. It integrates modern concepts for outlier detection, predictive classification, and regression, as well as multidimensional scaling based on inter-well similarity. Benchmarking of ML results against those created by experienced petrophysicists shows that the ML workflow can provide high-quality answers that compare favorably to those produced by human experts. A second validation exercise, that compares acoustic impedance logs computed from ML answers to actual seismic data, provides further evidence for the accuracy of the ML-generated results. The ML system supports the petrophysicist by easing the burden on repetitive and burdensome quality control tasks. The efficiency gains and time savings created can be used for enhanced effective cross-discipline integration, collaboration, and further innovation.