Statistical, Data-Driven Approach to Forecasting Production from Liquid-Rich Shale Reservoirs

Abstract

The oil and gas industry needs fast and simple techniques of forecasting oil and gas production. Forecasting production from unconventional, low permeability reservoirs is particularly challenging. As a contribution to the continuing efforts of finding solutions to this problem, this paper studies the use of a statistical, data-driven method of forecasting production from liquid-rich shale (LRS) reservoirs called the Principal Components Methodology (PCM). In this study, production of five different highly volatile and near-critical oil wells was simulated for 30 years with the aid of a commercial compositional simulator. Principal Components Methodology (PCM) was applied to production data from the representative wells by using Singular Value Decomposition (SVD) to calculate the principal components (PCs). These principal components were then used to forecast oil and solution gas production from the near-critical oil wells with production histories ranging from 0.5 to 2 years, and the results were compared to simulated data and the Modified Arps’ decline model forecasts. Application of the PCM to field data is also included in this work. Various factors ranging from ultra-low permeability to multi-phase flow effects have plagued the mission of forecasting production from liquid rich shale reservoirs. Traditional decline curve analysis (DCA) methods have not been completely adequate for estimating production from shale reservoirs. The PCM method enables us to obtain the production decline structure that best captures the variance in the data from the representative wells considered. This technique eliminates the need for parameters like the hyperbolic decline exponents (b values) and the task of switching from one DCA model to another. Also, production forecasting can be done without necessarily using diagnostic plots. With PCM, production could be forecasted from liquid-rich shale reservoirs with reasonable certainty. This study presents an innovative and simple method of forecasting production from liquid-rich shale (LRS) reservoirs. It provides fresh insights into how estimating production can be done in a different way.