Lithofacies discrimination of the Ordovician unconventional gas-bearing tight sandstone reservoirs using a subtractive fuzzy clustering algorithm applied on the well log data: Illizi Basin, the Algerian Sahara

Abstract

The main objective of this paper is to prove the capability of the fuzzy clustering algorithm for discriminating between lithofacies that are derived from the borehole log data of Ordovician tight sandstone reservoirs in the Illizi Basin of the Algerian South-Eastern Sahara. The technique is based on applying an unsupervised machine learning fuzzy algorithm on the natural gamma-ray, the photoelectric factor, the sonic log, the neutron porosity, and the bulk density well log data as an input to discriminate lithology into four lithofacies types: clay, sand, clayey sand, and sandy clay. The primary advantage of the fuzzy algorithms is that they can handle uncertainties and errors associated with the inaccurate or incomplete data of any measurements. Furthermore, it has a broader scope and a higher level of generality than the binary logic, which means that the results and interpretations that are based on this approach can be more appropriate and probable than other possible interpretations. A comparison between our algorithm's results and the results from the Kohonen Self-organizing Maps (SOM) and the Multilayer Perceptron (MPL) has also been carried out which proved that our approach provides better results than the other commonly applied neural networks. The obtained lithofacies types are highly comparable with that obtained from the core description data with minimal estimated errors. Therefore, this fuzzy algorithm can be applied as an alternative, fast, and low-cost technique for adequate lithofacies discrimination in case of lacking the core data.