Hydraulic fracturing is essential for economic production in tight gas and shale gas reservoirs due to their low permeability nature. Slickwater fracturing has been successfully performed in shales and tight gas reservoirs using low viscosity fluid, usually water, with friction reducers. Slickwater fracturing has the advantage of reducing formation damage and generally being less expensive than conventional gel. Predicting which slickwater parameters are important to successful production and which are less important has been an important but unanswered question. This research uses multivariate statistical methods to discover whether production from the Jonah Field, Wyoming, and the Barnett Shale, Texas, can be predicted using slickwater parameters and whether these parameters can provide insight into the design and analysis parameters of the slickwater treatments.
Factor, cluster, and multiple regression analysis show that production data from the sampled Barnett Shale and Jonah Field databases group separately from the slickwater parameters. Multiple regression, used to predict the EUR and the cumulative produced water from the slickwater parameters in the Barnett Shale, yielded best adjusted-R²’s of 34.7% and 25.3%, respectively. Multiple regression was also used to predict the EUR from the slickwater parameters in Jonah Field resulting in the best R² of 22.9%. Multiple regression analysis established a relationship with an adjusted R² of 93.0% between the fluid pumped and the fluid recovered from the Barnett Shale treatments. Multiple regression analysis also established that the amount of proppant used for Jonah Field hydraulic fracturing operations was calculated from the total fluid pumped and the net pay.
This research provides a methodology to use multivariate statistics to analyze stimulation treatments. Additionally, it opens opportunities for the analysis of different fields with more data using multivariate statistics and can aid in improving designs in an operator's current field or in understanding previous designs in a newly acquired field. Finally, it also demonstrates how "no information" can be valuable in cutting costs on commodities purchased to stimulate a well when no benefit is seen from those additional purchases.