Fluid type identification in carbonate reservoir using advanced statistical analysis

Fluid type identification in the carbonate reservoirs is always a challenge work due to the heterogeneous composition. The normal resistivity well logging data cannot separate gas and water formation for such complex reservoirs has been reported in the former studies. Moreover, the traditional cross-plot also cannot find the low limit value. Thus in this paper, we proposed the new statistics evaluation methods to solve such problems based on multi-parameters identified method which chose 10 parameters as dataset and utilized decision tree to build the rules, and discriminant analysis which build 2 predictive models (A and B) to classify the gas and water. We indeed successfully applied our new methods on the true well logging data from a heterogeneous carbonate reservoir in China. Thus we concluded that our new methods could be a good way in the future well logging evaluation for complex carbonate reservoir.