A case study of petrophysical rock typing and permeability prediction using machine learning in a heterogenous carbonate reservoir in Iran

Abstract

Petrophysical rock typing (PRT) and permeability prediction are of great significance for various disciplines of oil and gas industry. This study offers a novel, explainable data-driven approach to enhance the accuracy of petrophysical rock typing via a combination of supervised and unsupervised machine learning methods. 128 core data, including porosity, permeability, connate water saturation (Swc), and radius of pore throats at 35% mercury injection (R35) were obtained from a heterogeneous carbonate reservoir in Iran and used to train a supervised machine learning algorithm called Extreme Gradient Boosting (XGB). The algorithm output was a modified formation zone index (FZIM*), which was used to accurately estimate permeability (R2 = 0.97) and R35 (R2 = 0.95). Moreover, FZIM* was combined with an unsupervised machine learning algorithm (K-means clustering) to find the optimum number of PRTs. 4 petrophysical rock types (PRTs) were identified via this method, and the range of their properties was discussed. Lastly, shapely values and parameter importance analysis were conducted to explain the correlation between each input parameter and the output and the contribution of each parameter on the value of FZIM*. Permeability and R35 were found to be most influential parameters, where Swc had the lowest impact on FZIM*.

Modeling Permeability Using Advanced White-Box Machine Learning Technique: Application to a Heterogeneous Carbonate Reservoir

“…The reservoir of interest is highly heterogeneous; hence, no clear correlation between porosity and permeability can be observed. The Lorenz coefficient of 0.66 indicates high heterogeneity in the zone of interest . We used the K-nearest neighbor (KNN) algorithm to find an optimum number of clusters based on the reservoir data.…”

“…Although some differences exist between petrophysical dynamic rock types (PDRTs) and petrophysical static rock types (PSRTs), these two have often been considered the same. In the literature, the terms petrophysical rock types (PRTs), flow unit, and hydraulic flow unit (HFU) have interchangeably been used. ,, …”

“…In the literature, the terms petrophysical rock types (PRTs), flow unit, and hydraulic flow unit (HFU) have interchangeably been used. 1,2,17 In the literature, the core-based (RCAL and SCAL) petrophysical rock typing methods can be classified into three separate categories: 2 • Methods that utilize permeability, porosity, and connate water saturation to some extent. Excluding the cutoffbased methods, 18 other ones are either empirical such as Winland r35, 5 Pittman equations, 19 and r 35 19 or theoretical: FZI by 10 Amaefule et al and FZI* (FZI-Star) by 13 Mirzaei-Paiaman et al…”

Determining the electrical facies utilizing multi-resolution graph-based and artificial neural network clustering methods in an Oilfield, SW Iran

“…The dimensions of the reservoir at the contact surface of water and oil (2272 m below sea level) are 30 km long and 3.5 km wide, stretching northwest-southeast 2,9,23,[64][65][66][67][68][69][70][71] . In addition to the Asmari reservoir and sandstone section of Ahwaz, the Bangestan reservoir (Ilam and Sarvak Formations) are also present in this field 39,62,65,66 .…”

“…Facies data are used in cases such as separating reservoir zones from non-reservoir and in field-scale and large-scale structural adaptation. This data's importance is called the virtual core 8,24,62,77 . In this study, facies with common geological/reservoir properties were classified into different categories using readings of gamma, neutron, density, acoustic, and resistivity diagrams.…”