Abstract

Probabilistic estimation of hydrocarbon resource volumes requires the knowledge (or at least, the assumption) of the likely probability distributions of key reservoir and pool parameters used to estimate oil and gas reserves. Lognormal distributions are commonly associated with recoverable reserves, although distributions other than lognormal could describe the individual reservoir and pool parameters.

This study was carried out in order to investigate the probability distributions of reservoir and pool parameters, to contribute to the better estimation of potential and actual reserves in the Cooper/Eromanga Basin, Australia.

The key reservoir and pool parameters considered in this study were: pool area, net pay, porosity, hydrocarbon saturation and oil and gas recovery factors. All of the available data were derived from approximately 220 oil wells, 650 gas wells and over 130 fields in the Hutton, Toolachee and Patchawarra Formations of the Cooper/Eromanga Basin, in South Australia and Queensland. The raw data sets compiled were reviewed and re-sampled where necessary. Data points which were considered unrepresentative of the population distribution were discarded in the analysis. This primarily involved applying minimum value constraints to each data set. Below these minimum values, data were observed to be poorly or randomly represented. The resultant data above the applied minima are considered to represent technically and economically significant data sets.

Frequency histograms were constructed for each of the key parameters and a range of distributions were fitted to each data set. The optimal or best-fitting distribution was computed for each parameter, in each formation. Results indicate that pool area, porosity, net pay and oil recovery factor are optimally or adequately modelled by lognormal distributions. Conversely, gas recovery factor and hydrocarbon saturation (which are both negatively skewed in the Cooper/Eromanga Basin), are better described by distributions other than lognormal.

Along with the distributional analysis of each parameter, correlations between various parameter pairs were investigated by regression analysis, in order to test for independence. Results were not clearly defined within the entire data sets. However, significant correlations were revealed when smaller subsets of data from isolated areas of the basin were observed. The two most distinguishable were the negative correlation between porosity and depth and the positive correlation between hydrocarbon Saturation and porosity. Although, over the entire Cooper/Eromanga Basin, independence between the key parameters was assumed. The assumption allows all the parameters to be combined without the mathematical influence of covariance.

Introduction

There is always a significant level of uncertainty associated with modelling potential and discovered hydrocarbon reserves. The amount of oil or gas recovered from a single accumulation depends on various reservoir and pool parameters including: pool area, hydrocarbon saturation, net pay, porosity, and recovery factor. Estimation of potential hydrocarbon reserves requires the knowledge, or at least the assumption, of the probability distributions of these parameters.

Any multiplicative process converges towards lognormality due to the central limit theorem. Furthermore, the product of a series of lognormal distributions is also lognormal (Capen). Potential reserves are calculated by the product of the above parameters and hence, are assumed to be lognormal. But what of the individual parameters? Are the parameters lognormally distributed themselves, or is it simply the multiplicative process that provides us with lognormal reserves? This paper addresses these issues.

P. 631

This content is only available via PDF.
You can access this article if you purchase or spend a download.