Enhanced Oil Field Data-Wrangling using Machine Learning

Within every aspect of the petroleum industry, time is equivalent to money. Using time wisely is critical to the success of a good operation, whether discovering new oil reserves, determining the best method for reworking a poorly producing site, or simply tracking and analyzing existing production over an entire field. Technology changes rapidly, and understanding the associated information must keep up. Data has become the new oil of this millennium, where advancements in artificial intelligence and machine learning have taken over almost every aspect. With the capability of new analytical methods came the drastic increase in the quantity of data that has been gathered. The oil and gas industry generates vast amounts of data daily, ranging from sensor readings and geospatial data to operational metrics collected from drilling sites, refineries, and pipelines. A new data paradigm, initially coined as “big data,” quickly grew in scope until previously acceptable data handling methods could no longer keep up. The time needed to process data can now require weeks, months, or even years in some cases when this processed data is usually required immediately. In an industry where precise and timely insights are crucial, data quality issues can result in costly mistakes, suboptimal resource allocation, and even safety risks. This research covers Data Quality and why proper data handling is necessary for correct judgment. It also quantifies ways to quickly improve data by automating anomaly detection and correcting missing data through imputation. While still not as skillful as a team tasked with sifting and cleaning data, automation of the data-cleaning process allows for quick and timely judgment with far more complete information than would usually be available. While the availability of oilfield data has improved over the years, the next step is to keep it properly “fit for use” by those who need it.