Machine learning-assisted well log data quality control and preprocessing lab

Log and core data have been collected for decades in mature super-large fields in the Middle East, but the quality and age of the data varies. These data can be affected by artifacts created by borehole conditions, various logging tools, water or oil-based mud systems, and various processing parameters. Data often lacks the traceability to borehole/experimental conditions necessary to apply appropriate and consistent corrections for further modeling workflows. As a result, petrophysicists must perform extensive logging, core quality control (QC), and editing before interpreting logs or using core data. The purpose of this paper is to describe a machine learning (ML) application grouped into a solution for automatically QC and processing core and log data from hundreds of wells. The goal was to detect bad data and outliers and apply corrections that mimic human-driven interpretation to at least 80-90% of the processed data.

We have successfully used ML for various aspects of log editing in several datasets and cases. (Akkurt et al. 2018; Liang et al. 2019; Mawlod et al. 2019; Singh et al. 2020). However, the 2020 survey found many of the lessons highlighted in this study.

• Geological and geospatial information can be as valuable as data and cannot be overlooked by ML algorithms processing well data at field scale.
• ML algorithms are more efficient than human-driven data processing and can produce more accurate and consistent results. However, parameter selection and QC of results require expert evaluation. Therefore, ML-based applications should not be “black boxes” and should include a user interaction toolbox for efficient workflow control.
• To provide more efficient solutions in terms of performance for large datasets, we need to take full advantage of the power of cloud-based technologies to enable parallelization of operations while building and analyzing multiple solution scenarios simultaneously.

In our current work, we develop ML applications to QC core and well log data to perform log editing (artifact correction) and missing log prediction using other available data.

The ML application is guided by petrophysics domain expertise and advanced data-driven algorithms to perform complex data homogenization and predictions across hundreds of oil wells using a cloud-based environment. The application can communicate with well platform projects and read data from separate files in Log ASCII Standard (LAS) and Digital Log Interchange Standard (DLIS) formats. The methodology is as follows.

• Data release from well platform projects
• Integrate geospatial and geological data with available core and log data into a single data frame
• Outlier detection in petrophysics logs and core data
• Selecting functions for log editing
• Clustering, predicting, and validating the training data set using a variable number of clusters to select the best model for predicting the target variable.
• Results are transferred to the well platform project

The proposed ML application integrates geological and geospatial information to provide high-quality homogenized data sets used for rock typing, permeability, and saturation modeling. ML applications save a lot of time and effort and eliminate repetitive human tasks. Combining cloud-based implementations with existing petrophysics platforms allows for the best performance and data exchange between different software platforms. Interactive and user-friendly dashboards give geoscientists full control over each step of their ML data-driven workflows.