Unsupervised machine learning technique for classifying production zones in unconventional reservoirs

Abstract

Significant volumes of data sets are fast expanding in quantity as a result of the rapid development of unconventional tight and shale reservoirs. Productive zones in unconventional reservoirs are determined by the geomechanical and petrophysical features of the wellbore rock. Machine learning techniques have the potential to greatly enhance procedures for identifying sweet spots in such complicated reservoirs. The objective of this paper is to classify productive zones inside unconventional reservoirs by using clustering unsupervised algorithms. The application made use of approximately ten different variables of rock characteristics including zonal depth, effective porosity, permeability, shale volume, water saturation, total organic carbon, young's modulus, poisson's ratio, brittleness index, and pore size. Unsupervised machine learning clustering algorithms consisting of k-means and hierarchical were used to classify the sweet spots by applying them to the input variables; a support vector machine as a supervised machine learning algorithm is then used to evaluate the classification accuracy of the unsupervised algorithms based on the clustering labels. In this study, we evaluated two unsupervised machine learning algorithms for their capacity to pick clusters and identify productive zones. A supervised machine learning classifier was then used to assess the classification accuracy based on projected clusters. The findings show that both clustering techniques identified the sweet areas with high accuracy and were less time-consuming. This study aims to enhance the process of looking for productive zones utilizing unsupervised and supervised machine learning scenarios.

More information

Identifiers

ORCID for Mohamed Hassan: orcid.org/0000-0003-3729-4543

Catalogue record

Export record

Altmetrics