TY  - BOOK
AU  - Gallart,Diego
AU  - Ortíz,Alberto Cesar
AU  - Bernhardt,Carolina
AU  - Hryb,Damián
AU  - Rodríguez Martino,Julio César
AU  - Horowitz,Gabriel
ED  - Congreso de Exploración y Desarrollo de Hidrocarburos (10mo. : 2018 nov. 5 - 9 : Mendoza)
ED  - Jornadas de Geotecnología (6tas. : 2018 nov. 5 - 9 : Mendoza)
TI  - Análisis multivariado para la predicción de fluidos a partir de la mineralogía de yacimientos no convencionales
U1  - 068.82 553.28 C62 2018 0015647 
PY  - 2018///
KW  - Ciencia de datos
KW  - Mineralogía
KW  - No convencionales
N2  - Multivariate analysis. Prediction of fluids from mineralogy variables on unconventional reservoirs.The characterization of the pore spaces and volumes of fluids from data of T1T2 NMR in an unconventional reservoir allows the understanding of the characteristics of it at nanoscale. For these signals to acquire meaning and relevance in the prediction of ";sweet spots" or places of hydrocarbon interest, it is necessary to analyze them in relation to the mineralogy and content of organic matter (kerogen) of the rock.The multivariate analysis techniques used have been oriented in this direction, which has allowed to consider the influence of each parameter on the others, revealing correlations that would be very difficult to discover and analyze using traditional methods.The present work was carried out in different stages beginning by studying the correlations between different mineralogical variables, organic matter, fluids and NMR T1T2 porosity in an unconventional reservoir well belonging to YPF. The degree of correlation between variables was studied, eliminating those redundant, and 4 machine learning models were used to predict the objective variables where the smallest error obtained in the prediction was around 10%.The relative importance of the variables in the prediction was analyzed, which in some cases coincide with the existing physical models and in others they lead to rethinking some assumptions about them, providing new knowledge to this type of analysis.The models created with the well data analyzed can be tested with data from other wells to measure their efficiency and thus become a predictive tool for poral and fluid characteristics of shale reservoirs
UR  - https://biblioteca.iapg.org.ar/ArchivosAdjuntos/Conexplor2018/SGeot/1632.pdf
ER  -