Permeability Prediction of Carbonate Rocks Based on Digital Image Analysis and Rock Typing Using Random Forest Algorithm

Abstract

The diversity and multiscale characteristics of pore types in carbonate rocks usually result in extremely complex permeability–porosity relationships. Clarifying the main controlling factors of permeability and their response mechanisms is essential for improving permeability prediction. In this study, a digital image analysis (DIA) framework was developed to reveal the variation trend of permeability with pore structure parameters, and a rock-typing method consisting of flow zone indicator (FZI) and discrete rock type clustering technology (DRT) was applied to the carbonate samples to establish the binary permeability–porosity model. Results showed that permeability was approximately positively correlated with pore equivalent diameter, dominant pore size, and γ and negatively correlated with perimeter over area and fractal dimension. The commonly used multivariate linear regression (MLR) method failed to describe the response relationship between permeability and pore structure parameters. However, different discrete rock types exhibited special permeability–porosity relationships. Analysis showed that different discrete rock types were mainly controlled by the product of shape factor and the square of tortuosity. Further, three permeability prediction schemes were designed using the BP neural network and random forest algorithm based on the pore structure parameters derived from the thin sections. Compared with the direct prediction models trained by the optimized BP neural network and random forest algorithm, the indirect permeability prediction method based on FZI prediction showed better generalization ability with the highest R 2 of 0.936, demonstrating that the combination of digital image analysis, rock typing, and random forest algorithm is a robust and reliable method to realize the permeability prediction of carbonate rocks based on thin-section images.

Comparative Study on Hydraulic and Electrical Transport Properties of Carbonate Rocks Based on Rock Typing

“…Figure shows the FZI estimated by eq compared to that determined by eq for 64 samples from Verwer et al Obviously, eqs and overestimate FZI significantly, implying the difficulty of evaluating FZI using pore throat attributes. In view of this, in previous studies, , we established the logging parameters-based and pore structure parameters-based FZI prediction models using machine learning methods and achieved the permeability prediction of logging scale and thin-section scale, respectively.…”

“…The path indicated by the red arrows in Figure 15 has been implemented in our previous study, and better predictions were obtained than the conventional models. 26 In addition, to reduce the uncertainty of machine learning methods, researchers should understand the petrophysical response mechanism of different log series to ensure that the selected logs are closely related to the target prediction parameters, select as many samples as possible to capture all the complexity in the target data, and avoid overtraining the model.…”

“…Recent studies have shown that rock typing methods originated from traditional petrophysical models are very practical tools for hydraulic and electrical flow unit (HFU and EFU) identification, representative sample selection, permeability prediction, cementation exponent estimation, and fluid saturation determination. ,− Mirzaei-Paiaman et al and Kolah-Kaj et al gave a good summary of different rock typing indices. Different from the rock textural classifications of Dunham and the pore type classifications of Choquette and Pray, rock typing is to classify rocks according to special microstructure parameters or petrophysical properties.…”

“…However, whether the CZI method or EQI method, the key to applying it to the entire wellbore is to establish a quantitative response relationship between the rock typing index (e.g., CZI and EQI) and well logs. The studies by Zhang et al and Zhang and Cai revealed that the combination of flow zone indicator (FZI) proposed by Amaefule et al and discrete rock type (DRT) clustering technology could provide excellent classifications of carbonate permeability and porosity data. Their results showed that the permeability and porosity of the samples in each DRT group follow a good power-law relationship.…”

Study for the Carbon Emission Influencing Factors of China Based on Random Forest Model

“…The random forest algorithm is an algorithm based on the combination of multiple decision trees, using an integrated thinking method to solve classification problems and regression problems [8], proposed by Leo Breiman and Adele Cutler [9]. It mainly implements its core classification and regression functions through decision trees, each of which is a basic model consisting of a random vector and a training set.…”