A nuclear magnetic resonance proxy model for predicting movable fluid of rocks based on adaptive ensemble learning

Abstract

The movable fluid percentage and movable fluid porosity of rocks are important parameters for evaluating the development potential of petroleum reservoirs, which are usually determined by expensive and time-consuming low-field nuclear magnetic resonance (NMR) experiments combined with centrifugation. In this study, an NMR proxy model based on adaptive ensemble learning was proposed to predict the rock movable fluid indexes efficiently and economically. We established adaptive ensemble learning via an opposite political optimizer (AEL-OPO), which adaptively combines 33 base learners through political optimization to increase the prediction accuracy of the NMR proxy model. To improve the generalization ability of the AEL-OPO, opposition-based learning was introduced to improve the global search speed and stability of the political optimizer. Accessible petrophysical parameters such as rock density, porosity, permeability, average throat radius, and maximum throat radius were used as a training set, a validation set, and a test set. The prediction results show that our new strategy outperforms the other 33 base learners, with R2 (coefficient of determination) values of 84.64% in movable fluid percentage and 74.09% in movable fluid porosity.

Effect of moisture on microwave ignition of bituminous coal

“…Although the permeability of the sample cannot be directly tested by NMR technology, the permeability can be estimated by the obtained pore distribution and porosity. The characterization of pore distribution is completed by measuring hydrogen nuclei in pore water in the coal [ 30 ]. Therefore, before NMR testing, coal samples needs to be saturated with water.…”

“…A commonly used classical model based on the T 2 spectra, the free fluid model (also known as the Coates Model), was used to calculate the permeability of the sample. The free fluid model was based on the Timur-Coates equation [ 5 , 31 ]: where K c is the estimated sample permeability and C is a constant (5.9 in this case) [ 30 ]. φ is the sample porosity (%), BVI and FFI are pore volume fractions occupied by bound (irreducible) and free (producible) water, respectively.…”

“…The premise of calculating BVI and FFI is to determine a relaxation time threshold, namely T 2 cutoff value ( T 2c ), which can divide T 2 spectra into two parts: bound water and free water [ 30 ]. For a typical bimodal T 2 spectrum, T 2c is determined as shown in Fig 3 , BVI and FFI are the peak areas in the spectrum [ 33 ].…”

“…where K c is the estimated sample permeability and C is a constant (5.9 in this case) [30]. φ is the sample porosity (%), BVI and FFI are pore volume fractions occupied by bound (irreducible) and free (producible) water, respectively.…”

“…where m s is the mass of water-saturated samples (g), m d is the mass of dry samples (g), ρ w is the water density (g/cm 3 ), and V c is the sample volume (cm 3 ). The premise of calculating BVI and FFI is to determine a relaxation time threshold, namely T 2 cutoff value (T 2c ), which can divide T 2 spectra into two parts: bound water and free water [30]. For a typical bimodal T 2 spectrum, T 2c is determined as shown in Fig 3, BVI and FFI are the peak areas in the spectrum [33].…”