Comparison of Yet-To-Find Methods for the Determination of Recoverable Reserves From the Bakken: An Uncertainty Assessment Approach

Abstract

The assessment methods for estimating recoverable reserves from conventional versus unconventional systems are quite different. The classic yet-to-find method, described in Rose (2001), using analogs and field size distributions may not be applicable in continuous unconventional resources. Although, many unconventional plays show sweetspot and non-sweetspot areas, the petroleum accumulation is largely of continuous character and not confined to distinct trap structures. This study presents an approach for using uncertainty-based assessment methods for estimating the recoverable reserves from the Bakken and Three Forks reservoirs, investigating an area of twelve townships in Dunn County, North Dakota. A number of estimates for either oil in place or technically recoverable reserves have been published. Dow (1974) estimated that 10 billion bbls of oil were expelled from the Bakken shales, assuming an average organic carbon content of 3.8 %. After more data became available, Schmoker and Hester (1983), determined an average TOC of 11 to 12 % and calculated the expelled oil volume to amount to 132 billion bbls. The U.S. Geological Survey first conducted an assessment of technically recoverable reserves from the Middle Bakken reservoir and provided a mean of 3.65 billion bbls oil (Pollastro et al., 2008). Due to the immense drilling activity and expansion of the Bakken play into the Pronghorn and Three Forks units, a re-assessment was conducted by Gaswirth et al. (2013), raising the numbers to 7.4 billion bbls for all reservoirs combined. Not long afterwards, Harold Hamm, chairman of Continental Inc., announced that the recoverable reserves could be as much as 20 billion bbls, assuming continuous improvements in recovery technologies. The principle used in this Bakken case study is built on placing uncertainty on every single parameter influencing the volumetric calculation. Often, critical values such as initial hydrogen index and thus initial TOC are based on educated guessing. The kinetics of petroleum generation not very well understood and are roughly approximated by the Arrhenius equation. In fact, there many values which are not absolutely known. By placing a range of uncertainty on each of the variables, hundreds to thousands of simulations can be run, using either the Monte Carlo or Latin Hypercube sampling methods. The results are displayed in cumulative probability plots and tornado charts, as well as maps of the P10, P50, P90 values.

AAPG Datapages/Search and Discovery Article #90216 ©2015 AAPG Annual Convention and Exhibition, Denver, CO., May 31 - June 3, 2015