AUTHOR=Ahmed Syed Adnan , MonaLisa  , Hussain Muyyassar , Khan Zahid Ullah 

TITLE=Supervised machine learning for predicting shear sonic log (DTS) and volumes of petrophysical and elastic attributes, Kadanwari Gas Field, Pakistan

JOURNAL=Frontiers in Earth Science

VOLUME=Volume 10 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/earth-science/articles/10.3389/feart.2022.919130

DOI=10.3389/feart.2022.919130

ISSN=2296-6463

ABSTRACT=The availability of shear sonic log (DTS) is essential in the development and production phases of hydrocarbon reservoirs. It is employed in a variety of procedures, such as rock physics analysis, seismic inversion to constrain the results of producing facies, reservoir characterization, and seismic amplitude anomalies assessment i.e., amplitude versus offset. The lack of technology in older wells and high operating expenses, are the reasons for the missing or acquiring the DTS logs in a specific portion of the reservoir. To forecast the missing DTS curve, techniques such as empirical relations, rock physics, and others are applied; however, these approaches make assumptions or need multiple prerequisites, all of which have an impact on the computations. In recent years, intelligent predictors based on machine learning (ML) have emerged as robust, optimized and quick techniques that require fewer data sets as input. To achieve maximum accuracy in predicting DTS log, three separate supervised ML algorithms, i.e., random forest (RF), decision tree regression (DTR), and support vector regression (SVR) are applied in this research. The best algorithm is labeled based on the maximum determination of correlation (R2) and minimum mean absolute percentage error (MAPE). The RF algorithm predicted the DTS curve most accurately and is further employed to generate elastic attributes, such as P-impedance (P-imp), S-impedance (S-imp), Lambd-rho (λρ), Mu-rho (μρ), and petrophysical attributes, such as effective porosity (PHIE) and clay volumetrics (Vcl). The stratigraphic slices at the reservoir level revealed a feasible gas sands zone based on the elastic and petrophysical properties.