--> ABSTRACT: End Member Geological Models Key to Full Range of Uncertainty Mapping for Development and Investment Decisions, by Shrivastava, Sanjay K.; #90155 (2012)

Datapages, Inc.Print this page

End Member Geological Models Key to Full Range of Uncertainty Mapping for Development and Investment Decisions

Shrivastava, Sanjay K.
E&P, Reliance Industries Limited, Navi Mumbai, India.

Applications of stochastic methods to produce multiple realizations are being widely used to showcase ability to capture uncertainty. However, the practical and feasible approach to simulate single model for performance prediction is widely abound. Obviously, most practical case studies therefore use stochastic methods to generate single model focused to manage best case scenario close to deterministically arrived number and supporting low and high cases managed conveniently around this best case. The so called best case of non-measurable confidence and unknown probabilities are then subjected to future performance and business decisions. This may lead to unpleasant surprises and affect the business. This can be reduced or avoided by generating complete distribution of resource to support conscious business decision. This can be achieved by creating only two stochastic models as end members to define the complete statistical distribution of outcomes. There exists larger uncertainty in all parameters of interest than it can be captured through hard/soft data. Conventional methods of generating multiple realizations are heavily guided by histogram of sampled properties, local conditioning probability and modeled spatial variability. Often various complex characterization and modeling parameters and its effect on "big picture" is beyond the grip of working geoscientists to generate models delinked to known data averages. It is proposed to incorporate independent variability in all parameters for geostatistical propagation between the end members aimed at generating geological possibilities of most pessimistic and most optimistic scenarios. The premise is that the multiple realizations as outcome of single input parameters (e.g. variogram, histogram, facies proportion etc.) fail to capture the much needed full range of uncertainty. The proposed method helps to capture all possibilities of the phenomenon which may causes uncertainty. Models at different uncertainty levels can be physically extracted between end members using simple improvisation of processes within the modeling tool. A large advantage of this method is that all possibilities and interrelationship between different types of data with uncertainty are automatically managed between two end members. Such geologically consistent range of possibilities of individual building blocks of a model shall help to ensure that the final reality fall almost always between the two end members.

 

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