Abstract
Using environmentally friendly and low-cost chemical ingredients that are also effective in terms of oil and/or gas production is the ultimate goal of formulating fracturing fluids. The relatively transparent production data and fracturing fluid ingredients through FracFocus provide an opportunity to mine large amount of data for insights on the efficacy of fracturing fluid ingredients, which were not possible through traditional means of individual lab tests and statistical analysis of individual ingredients and small clusters of wells.
Here we showcase a big data analytics framework where through data preprocessing and feature engineering, algorithms and models can be built to quantify the efficacy factors of fracturing fluid ingredients for gas production. Special attention is paid to geological properties so that they are similar or identical for comparison.
As an example, we demonstrate the big data analytics process on a dataset of more than 3100 horizontal Marcellus shale gas wells in Pennsylvania, and the data-based insights not only corroborate the empirical wisdoms known in the industry, but also established quantitative measures such that ingredients with higher efficacy factors should be used more confidently, while caution should be exercised on those ingredients that have negative efficacy factors for gas production.
This work also shows the potential value of the FracFocus database in improving the efficiency and efficacy of hydraulic fracturing, in addition to its intended environmental cause. Big data and data sharing should be encouraged not only for public safety grounds, but also for economic reasons for the industry.