Bayesian Hierarchical Modeling: Application Towards Production Results in the Eagle Ford Shale of South Texas

Abstract

Recently, the petroleum industry has faced the era of data explosion, and many oil and gas companies resort to data-driven approaches for unconventional field development planning. The objective of this paper is to analyze shale oil wells in a shale reservoir and develop a statistical model useful for upstream. Shale oil wells dataset comprises three aspects of information: oil production rate time series data; well completion data; and well location data. However, traditional decline curve analysis only utilizes the temporal trajectory of the production rates. Motivated by this, we propose a Bayesian hierarchical model that exploits the full aspects of the shale oil wells data. The proposed model provides the following three functionalities: first, estimations of a production decline curve at an individual well and entire reservoir levels; second, identification of significant completion predictors explaining a well productivity; and third, spatial predictions for the oil production rate trajectory of a new well provided completion predictors. As a fully Bayesian approach has been adopted, the functionalities are endowed with uncertainty quantification which is a crucial task in investigating unconventional reservoirs. The data for this study come from 360 shale oil wells completed in the Eagle Ford Shale of South Texas.