Shale gas as one of the most important source of energy in US requires long-term development plans and reservoir management. Better understanding of the inherent complexities of shale system and identifying the key contributing parameters are essential for optimum operation of such assets. Different assessment techniques such analytical, numerical and statistical analyses have been applied to large multi-variable data set from shale assets with different degrees of success.
In this paper, advanced Data-Driven Analytics that incorporates pattern recognition capabilities of Artificial Intelligence, Machine Learning and Data Mining is used to understand and rank the contribution and influence of rock characteristics and design parameters during the production from shale. Advanced Data-Driven Analytics provides much needed insight into hydraulic fracturing practices in Shale. Unlike analytical and numerical techniques that are based on “Soft Data,” Advanced Data-Driven Analytics incorporates “Hard Data”. “Hard Data” refers to field measurements (facts) such as drilling information, well logs (Gamma ray, density, sonic, etc.), hydraulic fracturing fluid type and amount, proppant type, amount and concentration, ISIP, breakdown and closure pressures, and rates, while “Soft Data” refers to variables that are interpreted, estimated or guessed (and never measured), such as hydraulic fracture half length, height, width and conductivity or the extent of the Stimulated Reservoir Volume (SRV).
In this study, a spatio-temporal database including well locations and trajectories, reservoir characteristics, completion, hydraulic fracturing, and production parameters for a large number of horizontal wells in Marcellus Shale is generated. The impact of parameters such as up-dip versus down dip deviation of wells, TOC, porosity, stimulated lateral length and cluster spacing, etc. on production from wells in a shale asset is analyzed and the understandable trends and patterns in the database are identified. The analyses are performed using production from multiple time intervals (i.e. different production indicators) throughout the life of wells.