Evaluation of historical production and performance prediction of shale oil wells are critical for operating companies to manage unconventional oil fields. Production from oil shale in North America is increasing significantly due to the recent development in hydraulic fracturing technology, favorable oil price to justify development and production, along with some policy support from the regulatory and local governments among other things. However, managing sustainable productions from such assets is always a challenge for the operating companies. One of the challenges is to analyze historical production and predict future performance for such oil wells. Over the past few years, many methods have been proposed to analyze shale well production data to address these challenges.

In this study, four well performance models are selected after an extensive literature review. These four models, the stretched exponential production decline (SEPD) model (Valko and Lee, 2010), the Duong model (Duong, 2011), the fracture flow model with a constant coefficient, and the traditional Arps models (Arps, 1945) are evaluated against well production data from eight oil wells in the Bakken and the Niobrara shale. The production history of these wells ranges from 7 to 10 years and the maximum monthly production rate ranges from 7,500 Bbl to 13,000 Bbl.

This study demonstrates that in certain situations, the performances of the mentioned four models are either not satisfactory to match actual production data or difficult to use. A new well performance prediction method - modified fracture flow model with variable coefficient (MFF) is proposed and evaluated with production data along with the other four models. Steps on how to use the MFF model are given. The results of this study can be used as guidelines for choosing appropriate models for horizontal shale oil wells.


Decline curve analysis is commonly used for the prediction of future production and reserves estimation of wells when production data are available. The commonly used decline curves are exponential and hyperbolic (Arps, 1945). In unconventional reservoirs, people often find that the estimation from exponential or hyperbolic decline curves can deviate significantly from the actual production trend. With the increasing development of unconventional reservoirs, forecasting production and estimating reserves accurately for these resources has become more urgent and more important.

URTeC 1582234

This content is only available via PDF.
You can access this article if you purchase or spend a download.