Shahin Parchekhari - Department of Petroleum Engineering, Kish International Campus, University of Tehran, Kish, Iran\Petroleum Engineering Department, National Iranian Southfield Oil Company (NISOC), Ahvaz, Iran
Ali Nakhaee - Department of Petroleum Engineering, Kish International Campus, University of Tehran, Kish, IranInstitute of Petroleum Engineering, School of Chemical Engineering, College of Engineering, University of Tehran, Tehran, Iran
Ali Kadkhodaie - Earth Sciences Department, University of Tabriz, Tabriz, Iran

Permeability is arguably the most critical property for evaluating flow in the reservoir. It is also one of the challenging parameters which must be measured in the field. Nuclear Magnetic Resonance (NMR) logging across the borehole is among the popular techniques, which it is utilized to determine permeability across the reservoir. However, available correlations in literature for estimating permeability from NMR data do not usually provide acceptable accuracy in the carbonate rocks. Therefore, a new model is proposed to estimate permeability by establishing a relationship between core derived permeability and extracted features from the T۲ distribution curve of NMR data with the ensemble LSBoost algorithm. The feature extraction process is performed using peak analysis on T۲ distribution curves which it leads to ۵ relevant parameters, including T۲lm, TCMR, prominence, peak amplitude and width. The proposed model is validated by comparing the proposed method’s correlation coefficient against Timur-Coates and SDR equation estimation accuracy. The results show that our model generally provides better prediction accuracies in comparison with the empirical equation-based derived permeabilities.

NMR, T۲ distribution, LSBoost, Timur-Coates, Permeability