A Novel Shale Well Production Forecast Model Achieves >95% Accuracy Using Only 1.5 Years of Production Data

Abstract
Objective: Reliable production forecasting for shale wells is crucial for investment decisions, optimized drilling rate, energy security policies, and informed green transition scenarios. The industry has struggled with inaccurate production estimates from decline curve analysis (DCA) and from a long production history requirement for data-driven models. We have developed a state-of-the-art, physics-guided, data-driven model for accurate production forecast of unconventional wells for up to 10 years into the future. With an error of less than 5%, our hybrid model requires only 1.5 years of production data. The method facilitates long-term production diagnostics, well survival probability estimates, and profitable economic decisions. Method: The hybrid state-of-the-art production forecast method combines our τ-M physical scaling model with the higher-order derivatives of the production rate. For a set of 4000 wells, the first 1.5 years of production data were used to develop a universal hybrid model to estimate the pressure interference time, τ, for each well. The estimated τ is used to calculate the stimulated mass, M, of individual wells using the physical scaling curve. Finally, the data-driven estimate of τ, and physics-driven estimates of M are used to forecast future well production and well survival probability with time. Results: The robustness of the hybrid model has been tested on 6000 new wells in the Barnett, Haynesville, Eagle Ford, and Marcellus shale plays. Using the initial 1.5 years of production data and a single hybrid model, the predicted pressure interference time, τ, for 6000 wells has an R2 of 0.98. The maximum error in the predicted cumulative production of 2000 Barnett wells for any given year between the 2nd year of production to the 15th year of production is only 2%. Similarly, the maximum error in the predicted cumulative production for Marcellus (500 wells), Haynesville, (1500 wells) and Eagle Ford (200 wells), is 2%, 5%, and 3%, respectively. The achieved outstanding accuracy is further used to calculate the well survival probability with time and optimize the future drilling rate required to sustain a given energy demand. Novelty: We have developed a new, robust state-of-the-art hybrid model for unconventional well production forecasting. The model achieves an outstanding accuracy of > 95% and uses only the initial 1.5 years of production data. Early and accurate estimation of future production governs future investment decisions, re-fracking strategy, and improved energy security strategy.

Publisher
SPE

Conference/Event Name
2023 SPE Annual Technical Conference and Exhibition, ATCE 2023

DOI
10.2118/215091-MS