Full text loading...
-
Sensitivity Analysis with Correlated Inputs for Volumetric Analysis of Hydrocarbon Prospects
- Publisher: European Association of Geoscientists & Engineers
- Source: Conference Proceedings, ECMOR XIV - 14th European Conference on the Mathematics of Oil Recovery, Sep 2014, Volume 2014, p.1 - 15
Abstract
Assessing risk and volumetric estimates of hydrocarbon prospects in a probabilistic manner has become industry-standard in the last twenty years. This means that the G&G teams are asked to translate their expertise and their data in a number of probability distributions for volumetric, fluid and even economical parameters.
In recent years, it is becoming more and more common to introduce linear correlations and other more complex dependencies among some of these input parameters’ distributions, in order to get meaningful results representing the local geology of the prospect. For this reason, classical diagnostic tools focusing on the relative importance of the input parameters, such as tornado diagrams, are becoming too simplistic.
In the present work, we investigate a number of alternative approaches for performing a more accurate pre-drill diagnosis of importance factors. The first one is based on Global Sensitivity Analysis methodology and includes the computation of first, second and total order indices to quantify contribution from uncertain input parameters to uncertainty of the model prediction. The second one is based on an analysis of variance approach via a randomization over all the possible orderings of the input variables. Both these approaches allow a precise computation of the relative importance of the different factors, without losing the dependency structure existing among the variables.
We illustrate and contrast these approaches using a few simplified yet realistic case studies focusing on volumetric estimates of hydrocarbon prospects.