--> ABSTRACT: Subsurface Parameter Uncertainty: A Structured Approach, by Everts, Arnout J.; Alessio, Laurent; Friedinger, Peter; Rahmat, Faeez; #90155 (2012)

Datapages, Inc.Print this page

Subsurface Parameter Uncertainty: A Structured Approach

Everts, Arnout J.; Alessio, Laurent; Friedinger, Peter; Rahmat, Faeez
LEAP Energy, Kuala Lumpur, Malaysia.

Objective of this paper is to illustrate how a structured approach towards quantifying the expectation ranges of key subsurface parameters, differentiating between true uncertainty and mere variability and recognizing the possibility of biases in our subsurface data, can lead to significantly improved asset management, i.e., better reservoir models, improved accuracy of resource estimates, more objective assessment of appraisal value etc.

In the exploration/appraisal stage especially offshore, wells are scarce and therefore well-based property estimates are typically complemented by indirect evidence from seismic. On the other hand, developed fields especially onshore or resource plays may have a much higher well density but with a distribution clustered around interpreted reservoir sweet spots or chosen surface development sites. With this limited ‘ground truthing', our challenge is to make as accurate as possible assessments of the subsurface parameters in our fields, specifically:

-Determine the expectation ranges for the field- or block-wide average of key reservoir properties. For fields in the exploration or appraisal phase especially where wells have been drilled on seismic amplitude anomalies, data representativeness is an issue that may easily lead to over-optimism and underestimation of the uncertainty range. On the other hand, in fields with high well density such as resource plays, it is important to distinguish between reservoir variability which may be very profound from well to well, and genuine remaining uncertainty on the field property averages.

-Define the distribution model or models to be used for populating reservoir properties in our 3D reservoir models. Geostatistical property mapping typically uses distribution models fit to well data, in combination with transforms to reflect of spatial trends e.g., facies conditioning, depth trends or seismic inversion-based trends. However, where well data is scarce, clustered or sampling anomalous parts of the reservoir, distribution models based on wells alone may not be representative and give rise to inappropriate reservoir property mapping.

This paper will show examples of how to overcome these limitations and potential pitfalls in a structured manner, by relating back to the principles of sampling statistics and also, by introducing a holistic approach that addresses not only distribution uncertainty but also sampling biases and measurement uncertainty on the parameters themselves.

 

AAPG Search and Discovery Article #90155©2012 AAPG International Conference & Exhibition, Singapore, 16-19 September 2012