Evaluation of the performance of ANN in predicting of electrofacies (estimated by SOM, AHC, and MRGC models)

ABSTRACT

Log facies analysis is important for reservoir characterization, but is made particularly difficult by the problem of “dimensionality”: log space is not equivalent to geological space, and two points that are close to each other in log space may not always be similar geologically. Even with good visualization tools, performing classic method (two-step) manually in high-dimensional (>3) space is still difficult, slow, somewhat subjective, and requires a skill or expertise that is not always readily available. Recently, some novel methods are found such as multiregression graph-based clustering (MRGC), agglomerative hierarchical clustering (AHC), and self-organizing map (SOM). In comparison with the existing two-step tool, new models have been found to make the work much faster and easier, but they need porosity and permeability for training that requires skill and time. In this study a neural network-based electrofacies determination technique is presented and finally electrofacies that evaluated in new models were determined very fast by using some logs without any computing of porosity or shale volume.

KEYWORDS:

• Agglomerative hierarchical clustering and self-organizing map
• artificial neural network
• electrofacies
• multiregression graph-based clustering

Acknowledgments

The authors would like to thank the National Iranian Oil Company (NIOC) for supporting this study.