Shale-Gas Reserves Estimation: Multiple Decline-Curve-Analysis Models

Abstract

This article, written by Senior Technology Editor Dennis Denney, contains highlights of paper SPE 161092, ’A New Approach to Reserves Estimation in Shale-Gas Reservoirs Using Multiple Decline-Curve-Analysis Models,’ by Srikanta Mishra, SPE, Battelle Memorial Institute, prepared for the 2012 SPE Eastern Regional Meeting, Lexington, Kentucky, 3-5 October. The paper has not been peer reviewed. Recent interest in the Marcellus and Utica plays has renewed attention to the problem of reliably estimating recoverable reserves from low-permeability shale-gas formations. Over-optimistic results obtained from the commonly used Arps’ hyperbolic model have led to development of alternative decline-curve-analysis models that are based on empirical considerations (e.g., the Duong power-law model) or mechanistic considerations [e.g., the Valko stretched-exponential-decline model (SEDM)]. This work addressed discriminating between such models (including a new mechanistic model proposed for decline-curve analysis that is based on the Weibull growth curve) with only limited production data. A new approach to aggregating estimated-ultimate-recovery (EUR) forecasts from multiple plausible models is presented. Introduction The use of horizontal drilling, supplemented by multistage hydraulic fracturing, spurred the production of shale gas, including the Barnett, Eagle Ford, Haynesville, Marcellus, and Utica plays. The result is challenges in production forecasting and reserves estimation. Projecting production-decline curves is, perhaps, the single most widely used method for forecasting production from tight gas and shale-gas wells. Future-production potential at any given time and an estimated ultimate recovery (EUR) are assessed by fitting an empirical model of the well’s production-decline trend and by projecting this trend to the well’s economic limit or a common cutoff time (e.g., 30 years). The most commonly used production-decline-curve model is Arps’ hyperbolic model. However, forcing the model to fit production data from shale-gas wells has been found to result in over-optimistic results of EUR, stemming from physically unrealistically high values of the decline exponent to force the fit. Several alternatives have been proposed for analyzing decline curves for tight gas wells. One approach involves constraining the late-time decline rate to a more-realistic value on the basis of experience or analogs. Another approach involves searching for empirical decline-curve models that impose physically meaningful parameter definitions and finite EUR values on model predictions. A key issue associated with the use of multiple models is how to discriminate between them with limited production periods, and how to combine the model results to yield an assessment of uncertainty in reserves estimates.