Abstract
Estimating ultimate recovery (EUR) in shale is a function of rock properties, well, and completion design parameters. The variation associated with these parameters are in source of uncertainty. In this paper the combined decline curve (CDC) approach is used to estimate the EUR of shale wells. CDC is a conservative approach that combines hyperbolic (in early time) and exponential (in later time) declines for production analysis. The major objective of this work is to condition the results of the CDC-EUR of shale wells to rock properties, well characteristics, and completion design parameters in a given shale asset. As the first step CDC-EUR is estimated. In the second step data-driven analytics using artificial neural networks is employed to condition the CDC-EUR to rock properties, well characteristics, and completion design parameters. Then, artificial Intelligence techniques are used in order to extract the nonlinear relationship between well productivity and reservoir characteristics and completion parameters. In this study 168 wells from Marcellus shale asset are examined.
The major rock properties that are studied are porosity, total organic content, net thickness, and water saturation. Moreover, the effect of several design parameters, such as well inclination and azimuth, stimulated lateral length, stage length, number of clusters per stage, and amount of fluid and proppant are studied. A data driven model is developed using AI techniques. Results show that the model is able to find the relationship between the input parameters and EUR. Thus, this model will help petroleum professionals to have a better understanding of the effect of rock properties and design parameters on the future production from shale wells.
