Application of Machine Learning for Estimating Petrophysical Properties of Carbonates Using NMR Core Measurements

Abstract

Evaluation of petrophysical properties such as porosity, permeability, and irreducible water saturation is crucial for reservoir characterization to determine the hydrocarbon initially in place and further optimize predictions for hydrocarbon production. However, the determination of these parameters is challenging for carbonate rocks due to their heterogeneity. One of the commonly used methods to determine petrophysical properties is the application of nuclear magnetic resonance (NMR). The NMR method is capable of finding formation porosity directly from T2 distribution. Also, several models have been developed to estimate formation permeability, such as free fluid and Schlumberger Doll Research (SDR) models. In addition, irreducible water saturation is calculated by using NMR parameters such as free fluid index and porosity. Although these models show accurate results in sandstone reservoirs, there is still a lack of accuracy when they are applied and compared to laboratory measurements in carbonate rocks. Therefore, there is a need for calibration of NMR models for carbonate rocks and even proposing new models based on efficient and chapter methods.

The main objective of this research is to develop an empirical correlation for predicting porosity, permeability, and irreducible water saturation in carbonate rocks in the Middle East by comparing both NMR- and laboratory-data. Furthermore, the machine learning (ML) technique was applied to predict these three petrophysical parameters utilizing NMR data. Two approaches were utilized in ML: direct T2-spectrum application and features extraction. Different ML algorithms were trained, validated, and tested to estimate these petrophysical properties of carbonate rocks, such as tree-based and neural networks. The obtained results from ML algorithms were further compared with core measurements to ensure their accuracy. The findings are useful in better characterizing carbonate reservoirs in the Middle East through accurate estimations of hydrocarbon resources and related reserves.

Date of Award	Dec 2022
Original language	American English
Supervisor	Emad Al Shalabi (Supervisor)