ESTIMATING UNCERTAINTY IN WELL LOG ANALYSIS BY MONTE CARLO SIMULATION

Abstract

Quantitative interpretation of well logs is always associated with some amount of uncertainty. The source of this uncertainty lies in errors inherent in the values of input parameters, uncertainties in tools response and errors in laboratory measurements. The above sources of errors are good reasons for the need for properly assessing the uncertainty in reservoir parameters, e.g., porosity (CD), water resistivity (R), formation resistivity (Ri) and water saturation (S) obtained from well logs. In addition, errors in the measurements and the parameters used in the empirical relationships are carried forward and may not be accounted adequately in the final solution. A method to estimate this uncertainty by Monte Carlo simulation procedure has been illustrated. The method consists in mathematically simulating an experiment to determine the probability distribution of a variable obtained from a mathematical or empirical relation which involves one or more input parameters, each of which has its associated uncertainty. The estimates of uncertainty are expected to be better when the distribution of input parameters are determined from log responses.