Datapages, Inc.Print this page

Perspectives on Uncertainty in Reservoir Models


W. Scott Meddaugh, Chevron Energy Technology Company, Houston, TX

The oil and gas industry regularly uses static and dynamic reservoir models to assess volumetrics as well as to “test” and ultimately help to choose or justify a development option for an asset at a particular time (e.g. primary, infill, IOR, EOR) in the asset’s history.   The models are routinely generated using sophisticated software – sophisticated both in terms of the user interface as well as the underlying algorithms.  Consequently, we can now easily generate very elegant geological models. Too often, models are generated without fully understanding the limitations or uncertainty of the available data or, for that matter, the limitations of the underlying stochastic algorithms.  Recent work, for example, has shown that areal grid size or the semivariogram range parameter selected to build a geological model may significantly impact the recovery factor obtained by dynamic modeling. 

In the past few years there has been increased recognition that incorporating uncertainty into static and dynamic reservoir modeling yields better business decisions often in less time than classic “deterministic” approaches.  Routine use of design of experiments (DoE) based approaches to incorporate and analyze the impact of volumetric and connectivity uncertainty is becoming widespread.  The industry appears to be moving away from an “in data alone we trust” paradigm to an “in data and uncertainty we trust” paradigm for model building.

Successful application of DoE-based approaches requires a robust assessment of uncertainty. Such assessments include the not only a list of the factors that contribute to the volumetric and connectivity uncertainty, but also an asset specific set of appropriate low-case and high-case values, and usually a “mid-case” value as well, for each of the uncertainty factors.  This is not a trivial task as uncertainty exists in what is known and what is unknown.  We can describe to a greater or lesser extent the uncertainty associated with what we know such as the measurement or interpretation uncertainty associated with well log and core data or with historical production data.  We can make assumptions about the uncertainty associated with what we don’t know by using secondary data or defining constraints based on analog information.  The assignment of the low-case and high-case values thus combines both science and art (also known as geological “experience”).  The results of an on-going, historical “look-back” study using data from the Humma Marrat reservoir in the Partitioned Neutral Zone (PNZ) between Saudi Arabia and Kuwait has provided significant insight into the magnitude of OOIP uncertainty as well as the relative contributions to the OOIP uncertainty from the various uncertainty factors as a function of time, quantity/quality of data, and modeling or interpretation workflows.


AAPG Search and Discovery Article #90098©2009 AAPG Education Department, Houston, Texas 9-11 September 2009