Lithology prediction from well log data using machine learning techniques: A case study from Talcher coalfield, Eastern India

Kumar, Thinesh; Kumar, Seelam Naresh; Rao, G. Srinivasa

doi:10.1016/j.jappgeo.2022.104605

Cited by 50 publications

(13 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The resulting accuracy value is 93.55% through various stages, including hyperparameter tuning and feature selection [7]. Other studies that have been done previously are also making use of supervised machine learning to forecast lithology, such as support vector machines, decision trees, multi-layer perceptrons, random forests, and extreme gradient boosting, which show accuracy values above 80% [8]. From the previous experimental results, this study tries to evaluate the accuracy and performance of gradient boosting in the case of determining the sonic log value or DTCO.…”

Section: Methodsmentioning

confidence: 99%

Prediction of Compressional Slowness from Conventional Well Log Data using the Gradient Boosting Algorithm

Utama,

Komara,

Garini

et al. 2023

IOP Conf. Ser.: Earth Environ. Sci.

View full text Add to dashboard Cite

Compressional slowness (DTCO) is the most basic parameter in geophysics, petrophysics, and geomechanics. These parameters can be obtained through the sonic log tool. However, equipment constraints, relatively new technology, and high cost of measurement make the parameters generated by sonic logs unavailable in old wells or wells being developed. Therefore, it is essential to predict sonic logs, especially in the case of compressional slowness prediction. Using machine learning, predictions can be generated by studying data on existing log wells. One of the algorithms that can produce predictions on continuous data, such as log values, is gradient boosting. MAPE and RMSE were used as evaluation metrics. The inputs used are gamma ray log data (GR), density (RHOB), porosity (NPHI), and shear slowness (DTSM). MAPE results show an error value of 12.28% with an RMSE of 10.74, indicating that the predictive model obtained has good results and performance. Using hyperparameter tuning in machine learning can reduce the error rate by 2.29% with faster processing times. In addition, it was found that the quantity of training wells can affect the resulting error value. The existence of this research can help a petrophysicist, geologist, and geophysicist characterize a reservoir with limited data. The use of this method also has the potential to be an alternative solution when sonic log measurements are expensive.

show abstract

Section: Methodsmentioning

confidence: 99%

Prediction of Compressional Slowness from Conventional Well Log Data using the Gradient Boosting Algorithm

Utama,

Komara,

Garini

et al. 2023

IOP Conf. Ser.: Earth Environ. Sci.

View full text Add to dashboard Cite

show abstract

“…Liu et al, 20 Jiang et al, 21 and Sun et al 22 23 identified lithology using a gradient-boosted decision tree (GBDT) algorithm in 2021, demonstrating excellent classification performance. Kumar et al 24 in 2022 applied a range of supervised machine learning techniques to interpret horizons from geophysical logs, with all models showing over 88% accuracy. Mishra et al, 25 also in 2022, combined correlation tools with machine learning for lithology prediction, surpassing traditional methods, like knearest neighbors, support vector regression (SVR), and multiple regression.…”

Section: Introductionmentioning

confidence: 99%

New Deep Learning Network (Deep Residual Shrinkage Network) Is Applied for Lithology Identification To Search for the Reservoir of CO₂ Geological Storage

Sun,

Zhang,

Zhang

2024

Energy Fuels

View full text Add to dashboard Cite

Lithology identification plays a key role in CO 2 sequestration projects, helping to improve the feasibility, safety, and effectiveness of sequestration projects; therefore, accurate prediction of lithology is critical. The mapping relationship between the logging parameters and lithology complex and the logging response is multi-solution, resulting in inaccurate results of traditional logging lithology identification methods. In this study, we propose an enhanced predictive model for lithology, termed the deep residual shrinkage network (DRSN). This network incorporates a residual network, an attention mechanism, and a soft threshold strategy. The introduction of residual blocks in residual networks addresses the gradient vanishing problem inherent in deep neural networks by allowing the learning of residuals through skip connections. The attention mechanism enhances the focus of the model on crucial input elements, thereby improving its capacity to capture key information. Soft thresholding strategies are employed to eliminate noise from the inputs, enhancing the robustness of the model. Six logging parameters (photoelectric index, density, acoustic, gamma, spontaneous potential, and neutron) are chosen as inputs, with lithology serving as the output of the model. For comparison, we introduce the widely used residual network (ResNet), a classical lightweight network (SqueezeNet), and the convolutional neural network (CNN). Testing on three wells in China's Tarim Oilfield demonstrates that the DRSN model accurately locates reservoirs and identifies lithology more precisely. In lithology prediction for the three wells, DRSN achieved accuracy rates of 90.84, 85.51, and 93.70%, respectively. This research offers a novel approach to lithology prediction in the realm of carbon dioxide geological storage.

show abstract

“…It implements machine learning algorithms under the Gradient Boosting framework and provides parallel tree boosting to solve many data science problems in a fast and accurate way. For example, recent application of XGBoost in lithology identification from geophysical logs has been achieved (e.g., Kumar et al., 2022; Zhang et al., 2022). ConvXGB method combines the performance of the CNN and XGBoost, two outstanding ML methods, and provides more precise output by integrating CNN as a trainable feature extractor to automatically operate feature learning with input, and XGBoost as a recognizer in the top level of the network to predict the labels, which has the premier combination of prediction performance and processing time compared with other algorithms (e.g., Pardede et al., 2019; Ren et al., 2017; Thongsuwan et al., 2021).…”

Section: Introductionmentioning

confidence: 99%

Mineralogical Characterization From Geophysical Well Logs Using a Machine Learning Approach: Case Study for the Horn River Basin, Canada

Hu,

Liu,

Chen

et al. 2023

Earth and Space Science

View full text Add to dashboard Cite

Accurate estimation of mineral composition is essential for refined reservoir characterization, thermal conductivity and mechanical determinations of sedimentary rocks, but is extremely challenging in shale units due to the mineralogical complexity, low porosity and ultra‐low permeability. Direct mineral measurements derived from laboratory X‐ray diffraction (XRD) analysis on core samples and borehole geochemical logging tool (GLT), and conventional geophysical logs from vertical wells penetrating sediments are widely available in some basins, which enables detailed mineralogical characterization of a well. A hybrid machine learning (ML) architecture that improves model training and validation by combining convolutional neural network (CNN) with XGBoost allows accurate description of the mineralogical compositions across a basin. We applied this ML approach to predict the mineral compositions using conventional well logs from the Horn River Basin, northeast British Columbia, Canada, where extensive drilling for shale‐gas and conventional hydrocarbon resources, complemented by high temperature geothermal energy potential is ideal for case testing. The predicted mineral compositions from the ML approach are consistent with the mineralogical readings from the GLT and are confirmed by the XRD mineral measurements. This allows basin‐wide mineral compositions mapping that reveals spatial trends of major mineral compositions and their relationship with the previously recognized geomechanical and geological features, which have important implications for thermal conductivity modeling, reservoir evaluation and extensive geological studies.

show abstract

Lithology prediction from well log data using machine learning techniques: A case study from Talcher coalfield, Eastern India

Cited by 50 publications

References 36 publications

Prediction of Compressional Slowness from Conventional Well Log Data using the Gradient Boosting Algorithm

Prediction of Compressional Slowness from Conventional Well Log Data using the Gradient Boosting Algorithm

New Deep Learning Network (Deep Residual Shrinkage Network) Is Applied for Lithology Identification To Search for the Reservoir of CO₂ Geological Storage

Mineralogical Characterization From Geophysical Well Logs Using a Machine Learning Approach: Case Study for the Horn River Basin, Canada

Contact Info

Product

Resources

About

Lithology prediction from well log data using machine learning techniques: A case study from Talcher coalfield, Eastern India

Cited by 50 publications

References 36 publications

Prediction of Compressional Slowness from Conventional Well Log Data using the Gradient Boosting Algorithm

Prediction of Compressional Slowness from Conventional Well Log Data using the Gradient Boosting Algorithm

New Deep Learning Network (Deep Residual Shrinkage Network) Is Applied for Lithology Identification To Search for the Reservoir of CO2 Geological Storage

Mineralogical Characterization From Geophysical Well Logs Using a Machine Learning Approach: Case Study for the Horn River Basin, Canada

Contact Info

Product

Resources

About

New Deep Learning Network (Deep Residual Shrinkage Network) Is Applied for Lithology Identification To Search for the Reservoir of CO₂ Geological Storage